EPA/600/R-15/262
September 2015
Science Supporting Numeric
Nutrient Criteria for Lakes and
Their Watersheds: A Synopsis of
Research Completed for the US
Environmental Protection
Agency
-------
EPA/600/R-15/262
Science Supporting Numeric Nutrient
Criteria for Lakes and Their Watersheds: A
Synopsis of Research Completed for the US
Environmental Protection Agency
James D. Hagy III
Research Ecologist
U.S. Environmental Protection Agency
Office of Research and Development
National Health and Environmental Effects Research Laboratory,
Gulf Ecology Division
U.S. Environmental Protection Agency
Office of Research and Development
National Health and Environmental Effects Research Laboratory,
Gulf Ecology Division
1 Sabine Island Drive, Gulf Breeze, FL 32561
-------
Disclaimer
This document has been reviewed in accordance with U.S. Environmental Protection Agency
policy and approved for publication. The views expressed are those of the author. The results
presented are a synopsis of research completed by Tetra Tech, Inc, under Task Order 0005 under
Contract EP-C-11-037 and Work Assignment 02-12 under Contract EP-C-12-060. This report is
intended to be distributed with the Final Reports resulting from those research projects. The U.S.
EPA made comments and suggestions on the Final Reports intended to improve the scientific
analysis and technical accuracy of the documents. However the author did not contribute
otherwise to the production of the cited Final Reports. The Final Reports and views expressed
are those of Tetra Tech, Inc. or its employees.
Citation:
Hagy, J. D. III. 2015. Developing Numeric Nutrient Criteria for Lakes and Their Watersheds: A
Synopsis of Research Completed for the US Environmental Protection Agency. US
Environmental Protection Agency, Office of Research and Development, National Health and
Environmental Effects Laboratory, Research Triangle Park, NC. EPA/600/R-15/262.
To be distributed with:
Paul, M., J. Butcher, D. Allen, L. Zheng, and T. Zi (2015). Methods for Computing
Downstrewam Use Protection Criteria for Lakes and Reservoirs. Prepared for US
Environmental Protection Agency by Tetra Tech, Inc, Research Triangle Park, NC and
Tetra Tech, Inc, Center for Ecological Sciences, Research Triangle Park, NC.: 144 pp.
Paul, M., A. Herlihy, D. Bressler, L. Zheng and A. Roseberry-Lincoln (2014). Methodologies for
Development of Numeric Nutrient Criteria for Freshwaters. Prepared by Tetra Tech,
Inc., Research Triangle Park, NC and Tetra Tech, Inc, Center for Ecological Science,
Research Triangle Park, NC.: 345 pp. (includes Appendices 1-25)
Cover Photo
Interfalls Lake as viewed from Pattison State Park, Wisconsin. Pattison State Park is located on
the Black River and contains Big Manitou Falls, the highest waterfall in Wisconsin at 165 feet.
Photo by Jessica Aukamp.
-------
List of Figures
Figure 1. Relationship between Log Average total nitrogen (TN) in lakes from the
North Temperate Lakes Long-Term Ecosystem Research (LTER) site and a 6-taxa
metric based on fish species with high nutrient tolerance values. Reprinted from
Figure 15 in Paul, Herlihy et al. (2014).
Figure 2. A frequency histogram and cumulative distribution function for total phosphorus
concentrations in the French Broad River, TN computed on the base of an inverted
LOADEST model adjusted to ensure attainment of the downstream TP loading target.
Across the long term distribution of flow levels, this TP distribution results in attainment of
a flow-weighted average concentration of 0.39 mg/L at discharge to Douglas Reservoir.
Reprinted from Paul, Butcher et al. (2015).
-------
Abstract
Nutrient pollution remains one of the most prevalent causes of water quality impairment
in the United States. The U.S. Environmental Protection Agency's (EPA) approach to
addressing the challenge of managing nutrient pollution has included supporting development of
numeric nutrient criteria for the Nation's waters. To create scientific information that could
assist the Agency and its partners work toward this goal, EPA's Office of Research and
Development funded a two-year extramural research project focused on criteria development for
lakes and reservoirs, and in particular the challenge of relating criteria for streams to protection
of downstream lakes and reservoirs. Research focused on two areas of the US with abundant
water resources, namely the upper Midwest and the Southeast. Study areas in Wisconsin and
Tennessee were selected based on the availability of extensive long-term data sets quantifying
both water quality and aquatic life variables. The resulting research was documented in
considerable detail in two final project reports (Paul, M., A. Herlihy, D. Bressler, L. Zheng and
A. Roseberry-Lincoln (2014). Methodologies for Development of Numeric Nutrient Criteria for
Freshwaters. Prepared by Tetra Tech, Inc., Research Triangle Park, NC and Tetra Tech, Inc,
Center for Ecological Science, Research Triangle Park, NC.: 345 pp.; Paul, M., J. Butcher, D.
Allen, L. Zheng, and T. Zi (2015). Methods for Computing Downstrewam Use Protection
Criteria for Lakes and Reservoirs. Prepared for US Environmental Protection Agency by Tetra
Tech, Inc, Research Triangle Park, NC and Tetra Tech, Inc, Center for Ecological Sciences,
Research Triangle Park, NC: 144 pp) This report provides an accessible overview of the
research with interpretation of its possible significance from a scientific perspective. The
research illustrates the fact that relating aquatic life condition to nutrients and water quality is
challenging but tractable and that the application of optima and tolerance models is a useful
approach. Exploration of theoretical considerations related to developing downstream use
protection criteria identified several separate, yet related challenges. Novel methods that were
developed and applied illustrated that empirical approaches of intermediate complexity may
offer a viable way to improve explicit consideration of downstream use protection in water
quality management. These methods are, nonetheless challenging to understand. Further
development, application and explanation will be needed to make widespread application a
greater possibility.
Introduction
Excess loading of nitrogen (N) and phosphorus (P) is one of the most prevalent causes of
water quality impairment in the United States, affecting nearly 7,000 surface water bodies for
nutrients and nearly as many for organic enrichment or oxygen depletion (2010 CWA Sec.
303(d) List). Excess N and P in aquatic systems comes from many point and nonpoint sources,
including urban and suburban storm-water runoff, municipal and industrial waste water
discharges, fertilizer use, livestock production, atmospheric deposition, and legacy groundwater
nutrient pollution. Land use alterations in watersheds across the U.S. increase the delivery of N
and P applied to the landscape into surface and groundwater, impacting aquatic life uses, human
health and economic prosperity (Compton, Harrison et al. 2011). Because of the complexity of
the problem, the large number and diversity of stakeholders, and the inherent integration with
1
-------
social and economic systems, many recognize that nutrient pollution is a "wicked" problem, and
that solutions will not necessarily be simple or permanent. Such problems are "at best re-solved
- over and over again" (Rittel and Webber 1973).
One way to advance the prospects for long-term success is to identify quantifiable
environmental goals or targets. These could include numeric nutrient criteria (NNC), loading
limits, biocriteria, or other objectives based on environmental outcomes (e.g., recovery of
seagrass habitat, fish populations, etc.). Clear goals help define the current status relative to the
goal, and inform efforts needed to restore and protect the environment, even when policy allows
an extended period of time to fully attain the goal (e.g., Montana Department of Environmental
Quality 2014). Absent clear, quantifiable goals that could serve as a fixed reference point, the
"shifting baselines syndrome" (Pauly and Christensen 1995, Papworth, Rist et al. 2009) can
erode public awareness of degraded environmental conditions. As one approach to addressing
these challenges in the context of the nutrient pollution, the US Environmental Protection
Agency (EPA) in 1998 called for "accelerating development of scientific information concerning
the levels of nutrients that cause water quality problems", and working with states and tribes to
adopt nutrient criteria as part of enforceable state water quality standards (Environmental
Protection Agency 1998). EPA (2011) reaffirmed the Agency's focus on partnering with states
to develop nutrient criteria and reduce nutrient loading.
Development of NNC for freshwaters continues to present technical and policy
challenges despite the fact that freshwater ecology has developed over a significant period of
time and has addressed the effects of nutrient pollution in a variety of ways (Environmental
Protection Agency 2014). A major complicating factor is that many stressors that co-occur with
nutrient pollution also impact biotic condition. For example, conversion of natural land uses to
agriculture or developed land uses usually increases nutrient concentrations in streams (Beaulac
and Reckhow 1982), but can also change hydrology, stream channel morphology, temperature
regimes, and sediment loading, among other effects that can impact biotic condition. Whereas a
decline in biotic condition may be readily observed in relation to developed landscapes,
condition may not relate strongly to nutrient concentrations specifically. Karr (1981) noted the
difficulty of characterizing stream condition via water quality proxies such as nutrients and
chlorophyll-a, and instead proposed biotic indices as a more direct approach to condition
assessment. While this approach is viable, it leaves open the question of causes and possible
solutions. A lack of relationship between a stream condition index (i.e., a biotic index) and
nutrient concentrations in Florida streams led EPA and the State to develop NNC via a reference
stream approach, rather than via the explicit linkage between nutrients and biotic condition
preferred by stakeholders (Environmental Protection Agency 2010a).
Another technical challenge associated with development of NNC is protection of
downstream waters. The strategy EPA outlined in 1998 recognized that "the finally developed
criteria must limit not only the unacceptable enrichment of a given water body or watercourse,
but also must factor in the effects of that enrichment on downstream receiving waters"
(Environmental Protection Agency 1998). However, the technical guidance and recommended
criteria that EPA published during the next several years (e.g., Environmental Protection Agency
2000) utilized a reference approach that does not explicitly consider protection of downstream
waters. Downstream use protection received much greater attention with EPA's 2010 proposed
-------
nutrient criteria for the State of Florida, which explicitly addressed stream water quality required
to protect downstream lakes and estuaries (Environmental Protection Agency 201 Ob).
Development of these criteria proved both complex and controversial, as the proposed total
nitrogen (TN) concentration limits required to protect downstream estuaries were often more
stringent than required to protect the streams. By later in 2010 when EPA's proposed rule was
finalized, the lake DPVs had been revised in response to public comment and the proposed DPVs
for estuaries were removed and slated for consideration in a later rulemaking (Environmental
Protection Agency 2010c). When the State of Florida ultimately adopted NNC, the State noted
that limits for nutrients in streams to protect downstream estuaries was to be addressed via the
narrative nutrient criterion, but provided no indication of how this was to be done (Florida
Department of Environmental Protection 2013). These developments suggest that additional
scientific research addressing key areas of uncertainty are needed to encourage implementation
of NNC that explicitly address downstream protection.
Given the ongoing need for science to support development of NNC in freshwaters, and
in particular science to support development of numeric criteria to protect downstream uses,
EPA's Office of Research and Development funded an extramural research project which was
conducted in two phases by researchers at Tetra Tech, Inc. The overarching research concept
was that nutrient thresholds for streams and could be explored principally by considering
protection of downstream uses, particularly lakes and reservoirs. Moreover, with possibly fewer
mitigating factors, quantifying nutrient limits for protection of nutrient-sensitive aquatic life in
lakes and reservoirs may be more tractable than for streams. Phase I, which concluded in 2014,
focused on identifying nutrient-sensitive aquatic life uses in freshwater lakes and reservoirs,
specifically lakes in the upper Midwest and reservoirs in the Southeast US. Phase II, which was
conducted during 2014 and 2015 examined how NNC could be developed to support attainment
of the identified nutrient-sensitive aquatic life uses within the same regional lakes and reservoirs.
Phase II went on to explore how, given NNC for a lake or reservoir, one could approach
development of numeric criteria for the associated network of streams in the contributing
watershed that would ensure attainment of the downstream NNC. These criteria have been
referred to as "downstream protection values" or "DPVs." Phase II aimed to develop solutions
that were intermediate in complexity and data requirements, such that they might be applied
broadly to support nutrient management. This report examines the results of these extramural
research projects, which are presented in considerable detail in a Final Report for each phase
(Paul, Herlihy et al. 2014, Paul, Butcher et al. 2015) and associated supplemental information.
These Final Reports are intended to be distributed as attachments to this report. Key findings are
evaluated to identify potential applications in nutrient management as well as questions in need
of further investigation.
Research Approaches, Data and Study Sites
Study sites within the broad target regions were identified based on both the availability
of suitable data. For the upper Midwest region, the study used a data set from the North
TemperateJ^^ a National Science
Foundation funded research project focused on the ecology of lakes and sustained since 1981.
Data selected for use in the study include water quality and nutrient concentrations,
phytoplankton community composition, macrobenthic invertebrates, zooplankton, and fish
-------
communities. Depending on the dataset, the period covered is a subset of the overall period of
record, which is 1981 through 2013. For the Southeast region, data collected by the Tennessee
Valley Authority covered as many as 32 lakes and included water quality data collected from
1960 to 2006 and fish community data collected from 1993 to 2012. Research addressing on
quantifying DPVs focused on two watersheds, one in each of the target regions. For the upper
Midwest, the study focused on Holcombe Flowage, Wisconsin, located on the Chippewa River.
Although the intent for the upper Midwest region was to study natural lakes, this reservoir site
was considered a good site for developing analysis for downstream protection because of the
availability of data in both the target waterbody and its watershed. For the Southeast, analysis
was focused on Douglas Reservoir, located in eastern Tennessee on the French Broad River.
During research Phase I, relationships between water quality variables or abundances of
biota and nutrients were explored using several variations on indicator value approaches
characterizing either optimal nutrient concentrations, nutrient tolerance limits, or both (Yuan
2006). In Phase II, relationships among biotic and water quality variables were explored using
scatterplots, locally-weighted scatterplot smoothing (LOESS) regressions, and change-point
analysis, the objective being to identify water quality conditions at which nutrient-sensitive
biotic endpoints may be impaired (i.e., such as identified in Phase I). Given nutrient targets for
Holcombe Flowage and Douglas Reservoir (either hypothetical or based on state water quality
standards), research in Phase II explored methods that could be used to calculate DPVs. The
analysis examined key theoretical concerns associated with development of DPV criteria, an
important task that has not received sufficient attention, and also reviewed prior work in the State
of Florida that relates to calculation of DPVs. Phase II sought specifically to identify approaches
of intermediate complexity; that is, methods that lie between the simplest possible approaches,
which may result in erroneous or perhaps unnecessarily restrictive criteria, and detailed
mechanistic modeling of both watershed and receiving water. The latter may be sufficient, but
generally comes at a high cost and may not actually offer the expected increase in accuracy that
might be assumed on the basis of complexity (Reckhow 1994). Several alternative approaches
were considered. One approach involved empirical analysis of data from a stream network using
statistical methods designed to account for spatial correlations among monitoring sites
distributed in a stream network. Analysis of the network topology was accomplished using the
Spatial Tools for the Analysis of River Systems (STARS) and the Functional Linkage of
Waterbasins and Streams (FLoWS) toolsets, which generate data objects that can be analyzed
using spatial linear models using the SSN package in R (Ver Hoef, Peterson et al. 2014). The
resulting estimates of the spatial structure of water quality in the stream network were used to
generate complete realizations of water quality data in the stream network. These were then
evaluated to quantify how sample data similar to a real assessment could be interpreted relative
to the downstream or "pour point" requirement associated with attainment of water quality and
aquatic life goals in the receiving water lake.
Overview of Results
Phase I
Research in Phase I included an evaluation of biotic metrics that have been used in
nutrient management by states in the Midwest and Southeast. A number of states were found to
have collected data on aspects of biotic condition in their lakes and reservoirs and several were
-------
developing new indicators and methods involving macrophytes and fish metrics. Regardless,
most states continue to use water quality measures as indirect indicators of aquatic life use
attainment, rather than direct measures of biotic condition. Florida was the only state found to
use biotic measures directly in regulatory assessments. A number of states had noteworthy
approaches, however, which are examined state by state in the Final Reports.
For the upper Midwest region, nutrient optima and tolerance analysis showed that
inference models performed best (in decreasing order) for fish, zooplankton, benthic
macroinvertebrates, and phytoplankton. This result was counterintuitive, as one might expect
strongest inference among ecosystem components where causation was most direct (e.g.,
nutrients to phytoplankton). The generality of these observations is unclear, as it its dependence
on unique aspects of the datasets involved. For example, the investigators noted that
macroinvertebrate inference models may have been poor because the gradient in nutrient
concentrations was relatively small within the dataset that included macroinvertebrate data. The
relatively weak relationships between phytoplankton species and nutrient data could also reflect
a mismatch between observation scale and scales of variability for these potentially very
dynamic ecosystem components. Because the strongest results were obtained for fish optima
models, use of multispecies metrics considering either nutrient-tolerant or nutrient-sensitive fish
species could be worthy of further consideration (e.g., Fig 1). Based on the analysis of upper
Midwest vs. Southeast region lakes and reservoirs, it is clear that regional models are needed to
determine regionally applicable thresholds based on fish metrics. An additional benefit of an
approach based on fish-indicators is that there is a relatively direct conceptual linkage between
fish abundance and most definitions of aquatic life use attainment.
Phase II
Phase II research explored linkages between nutrient inputs to Midwest and Southeast
region lakes and reservoirs and support for nutrient-sensitive aquatic life uses such as those
identified in Phase I. The objective was to establish useful new approaches that could be applied
or adapted by those tasked with determining criteria, rather than to develop or recommend
criteria values. Thus, applications were developed even if a state had already adopted regulatory
criteria based on another approach. Where applicable, any state-adopted values were also noted
and considered in later analyses. This was the case for lakes in the State of Wisconsin
(Wisconsin Administrative Code, NR 102.06, accessed via
http://docs.legis.wisconsin.gov/code/).
In the conceptual model applied, an important effects pathway was identified in which nutrients
were related to phytoplankton chlorophyll-a and planktonic chlorophyll-a was related to
hypolimnetic dissolved oxygen (DO). Neither chlorophyll-a nor hypolimnetic DO were closely
related to most biotic measures in the lakes, however, and none provided a relationship sufficient
to quantify a target nutrient concentration. As a result, nutrient thresholds were derived based on
existing DO thresholds (e.g., 2.0 mg/L and 5.0 mg/L, based on Wisconsin statute) and
relationships between nutrients, chlorophyll-a, and DO. Specifically, the proportion of DO
observations greater than 2 and 5 mg/L decreased in relation to growing season average
chlorophyll-a (p<0.01 for both).
-------
6-taxa High Nutrient Metric vs. Average TN
IZU
-». i rin
CTl I UU
OJ
X
DJ
I 80
mi
1 60
*-fr
~ 40
,—(.
^' 20
X1
O
0
on
-20
-1
O
.2 -1
O
O
0
.0 -0
Q*wrt
.8 -0
0
o
°,»
>3tf V*
mm
.6 -0
8*V
.4 -0
Q
O
°°o
014 |ig/L and, via a nutrient-chlorophyll-a
regression relationships, TP<40 |ig/L and TN<0.71 mg/L. These values were used in the
subsequent analysis of downstream use protection. As in the Wisconsin lakes, the empirical
relationship between DO and biotic condition in Tennessee reservoirs was difficult to establish.
Among individual taxa, statistically significant relationships showed that higher proportions of
low DO were associated with a slightly higher relative abundance of Bluegill (Lepomis
macrochirus) but lower relative abundance of largemouth bass (Micropterus salmoides).
However, absolute abundances of both fish taxa increased with increasing chlorophyll-a, perhaps
indicating a larger positive effect associated with increasing productivity than any negative effect
attributable to decreased habitat quality. These results may outline tradeoffs regarding
conflicting uses of reservoirs in the Southeast, wherein increasing trophic status may support
increased fish production and benefit to fisherman, but at the cost of decreased aesthetic value
for other recreational users, such as swimmers (Keeler, Polasky et al. 2012).
-------
Downstream Use Protection
The ultimate research objective in Phase II was to explore approaches for quantifying
stream nutrient concentrations that would protect the identified nutrient-sensitive aquatic life
uses in downstream lakes and reservoirs (from Phase I), as indicated by protective nutrient
thresholds (also explored during Phase II). As noted in the Final Report from Phase II (Paul,
Butcher et al. 2015), states have frequently acknowledged the need to ensure protection of
downstream waters in the development of numeric criteria, but no state other than Florida was
found to have proposed any detailed procedure for doing so. Moreover, Florida's approach does
not address all cases where downstream use protection may be needed. For example, protection
of downstream estuaries in Florida continues to be addressed via narrative criteria. The noted
lack of procedural detail in state treatment of downstream use protection may reflect in part the
technical complexity associated with rigorous treatment of the issue.
Development of Theory. Phase II research began by exploring some of the key technical
issues related to downstream use protection. One issue is translation from either causal or
response variable criteria in the receiving water to loading thresholds. Approaches that were
examined include a "Vollenweider approach" which refers to the frequently cited 1976 paper
relating TP loads to chlorophyll-a in lakes (Vollenweider 1976), and the US Army Corps of
Engineers' BATHTUB model, which was used by EPA to develop lake DPV approaches for
Florida (Environmental Protection Agency 2010a). Another challenge is relating a loading
threshold to an assessable threshold for nutrient concentrations at the point that a stream or river
discharges into the receiving water. Allocation of loading to the receiving water among
watershed sources, as well as contributions other than from defined stream channels (e.g.,
groundwater discharge, air-deposition or diffuse run-off) must also be considered. Another issue
involves translating loading thresholds to observable concentrations. Complexity is increased by
the reality that average nutrient concentration is not the same as nutrient loading divided by
average stream flow. The latter describes a flow-weighted average concentration, which is
quantitatively the same as average concentration only under the unrealistic scenario that flow is
entirely invariant. Not only does flow generally vary, but nutrient concentrations often depend
on stream flow (Hirsch, Moyer et al. 2010). EPA's proposed DPV criteria for estuaries
(Environmental Protection Agency 201 Ob), while novel in approach and intent, neglected this
difference. A final issue relates to translation of target concentrations at the point of discharge to
the receiving water (i.e., the pour point) into similar thresholds for other points throughout the
watershed. As has been argued previously, translation of nutrient thresholds into watersheds is
likely to be useful for understanding, managing and ultimately limiting the risk of downstream
impairments posed by accumulation of otherwise acceptable nutrient sources within watersheds.
For example, downstream use protection criteria could be useful for evaluation of NPDES
permits, TMDLs, or other management actions in a watershed (Environmental Protection
Agency 201 Ob). Minimally, an appreciation of the data, theory, and calculations that could be
used to quantify and evaluate the risk of downstream impacts will be invaluable for considering
protection of watersheds and downstream receiving waters from nutrient impacts over the long
term. The Phase II research evaluated several approaches to addressing the technical challenges
described above, but realistically requires more analysis to be done before the methods are fully
developed, evaluated, and can be explained to stakeholders.
-------
Relating Load to Concentration. One of approaches that was developed addresses the
task of translating a loading threshold to limits on the observed concentration at the watershed
pour point. This approach adapts simple formulations of the USGS LOADEST model to
quantify correlations between flow and concentration. The simplest version involves a linear
regression of loading (Li) on flow (Qi) in log-log space (i.e., Lt = exp(a0 + ax In Qi)H ), where
H is a bias-correction factor. A simple elaboration of this approach involves a "broken-line
regression" that changes slope at the flow level that optimizes goodness-of-fit. Adjustment of
model coefficients such that the average load attains a loading target, while re-arranging the
equation to compute concentration conditional on flow, provides an approach to computing a
distribution of nutrient concentrations that would be expected if the downstream loading
threshold were being attained. In application to both Holcombe Flowage and Douglas Reservoir
(Fig. 2), the stratified LOADEST model provided the best fit, leading to an estimated TP
concentration distribution that would be consistent with attainment of the critical downstream
loading rate. A key advantage of this approach is its strong reliance on local data. A similar
scaling approach can be applied using a fitted SPARROW model, but it should be recognized
that SPARROW models are generally most accurate at relatively large spatial scales. In
application to Holcombe Flowage, the SPARROW estimates were considerably different from
estimates based on LOADEST.
T7TT
100%
• 90%
- 80%
• 70%
• 60%
50%
• 40%
- 30%
20%
10%
0%
_
01
CL
ooooooooooo
Total P (mg/l)
Figure 2. A frequency histogram and cumulative distribution function for total
phosphorus concentrations in the French Broad River, TN computed on the base of an
inverted LOADEST model adjusted to ensure attainment of the downstream TP loading
target. Across the long term distribution of flow levels, this TP distribution results in
attainment of a flow-weighted average concentration of 0.39 mg/L at discharge to Douglas
Reservoir. Reprinted from Paul, Butcher et al. (2015).
Spatially-Correlated Models. Development of spatially-correlated models of nutrient
concentrations in stream networks represents perhaps the most interesting, and yet most
challenging new approach to quantifying downstream use protection criteria that was developed
-------
during Phase II of the research. At its core is the recognition that the distribution of water
quality observations in a stream network may be understood via an empirical model that
considers both suspected drivers of nutrient concentration, such as land use, and expected spatial
correlation of deviations from mean predictions based on the drivers. Implementation of the
approach involves estimating the spatial model and implementing Monte Carlo simulations to
quantify the relationship between upstream exceedances and downstream water quality. Monte
Carlo simulations were implemented in two stages, wherein the first stage generates complete
sets of simulated water quality and the second Monte Carlo resampling generates samples of
water quality similar to what might be obtained from a realistic monitoring program. Within the
subsamples, the distribution of TP concentrations across sampled locations can be compared to
the distribution of associated downstream concentration, providing insight into the relationship
between water quality throughout the stream network and attainment of the target "downstream
concentration" at the downstream pour point.
As applied to the study sites, the statistical analysis considered the accumulated upstream
area (total), and the accumulated area of water, agriculture, grazing, urban land, and upstream
distance. Spatial covariance models included "exponential tail-up" (i.e., correlation with
upstream observations), "exponential tail-down" (spatial correlation with downstream
observations) covariance, as well as Euclidean correlations (spatial correlation with nearby
observations, whether on the stream network or not). For Holcombe Flowage, none of the
landscape factors were significant predictors of TP, so the final spatial regression considered
only mean TP and estimated spatial correlations. In contrast, accumulated upstream agriculture
area was a significant predictor of TP concentration in the Douglas Reservoir watershed.
Analysis of Simulated Distributions. Analysis of simulated water quality scenarios within
the stream networks provided insight into the relationship between the TP concentrations at
attainment of the downstream threshold at the pour point. This could be approached via the
relationship between TP means in the watershed and the downstream concentration, or the
number of sampled locations in the watershed exceeding a threshold and the probability that the
downstream TP concentration exceeds the threshold for protection of the reservoir. Because of
the log-normal distribution of water quality, an interesting result was that simulated mean TP in
the watershed varied to a relatively small amount even as the associated downstream
concentration was increased. The number of observations within the watershed that exceed the
downstream concentration threshold, which reflects a shift in the right tail of the log-normal
distribution, may be a better indicator of the probability that the pour point concentration will
meet the threshold.
Discussion and Future Work
Empirical analysis of relationships between biota and water quality variables, even in
locations with remarkably good long-term data sets, illustrates the challenges associated with
quantifying such relationships. Accordingly, it is not entirely surprising that many states rely on
reference distributions and indirect water quality indicators of the status of lake ecosystems.
Nonetheless, analysis of optima and tolerance values shows promise for evaluating the responses
of biotic assemblages to nutrients. Continued monitoring of biotic condition in association with
contemporaneous monitoring of nutrients and other water quality conditions will foster
-------
increasing understanding of the relationships. Development of viable approaches for developing
NNC may depend on sufficient interpretation of state use designations and/or narrative criteria to
better define the questions that could be addressed via scientific analysis.
Consideration and enumeration of steps needed to develop downstream protection criteria
illustrates that several distinct analytical challenges are involved, rather than just one. For
example, water quality thresholds for a lake are generally expressed as seasonal lake-wide means
or similar metrics. Attainment of these values, however, will most likely to be linked to loading
from watershed sources, rather than concentrations, and will also depend on sources not
discharged from stream networks. A sufficient quantification of nutrient sources across the
landscape is therefore a key data requirement.
Evaluation of nutrient distributions in several real watersheds, however, suggests that not
every case presents an impossibly complex challenge. For example, internal losses of TP in the
Holcombe Flowage watershed are likely to be small, and water quality within the stream network
could be modeled as spatially-correlated fluctuations about the mean, simplifying analysis of
downstream protection. In the Douglas Reservoir watershed, spatial correlations were less
important, but gradients in water quality could be predicted based on a key landscape driver,
namely the distribution of agricultural land use. In both watersheds, it was possible to predict the
likelihood of attaining a downstream water quality target given the frequency of observing
higher values in a sample population, as was demonstrated in the Final Reports. These
demonstrations suggest that analytical methods of intermediate complexity could be used to
enable some systematic treatment of downstream use protection without resorting to the most
complex, expensive and data intensive modeling approaches. That said, review of the Final
Reports suggests that further development and clarification of the methods is needed before they
are likely to be used more broadly.
An interesting possible application of the methods explored in this study, though one
requiring further development, is use of spatial correlations in stream networks to generate more
efficient and cost effective monitoring and assessment strategies. The presence of along-network
correlations in water quality, which one would be expected a priori in stream networks, and
especially the existence of established computer-based methods to quantify and analyze it,
suggests that this should be pursued. In addition, use of water quality monitoring designs that
consider such methods could lead to an improved ability to evaluate the potential for downstream
impacts. Relative to random or stratified random stream monitoring and associated evaluation of
stream condition at a large scale using simple statistical metrics, the potential for improved
assessment using these methods, and the associated scientific and policy benefits of reducing
uncertainty, may be significant.
References
Beaulac, M. N. and K. H. Reckhow (1982). "An Examination of Land-Use - Nutrient Export
Relationships." Water Resources Bulletin 18(6): 1013-1024.
Compton, J., J. Harrison, R. Dennis, T. Greaver, B. Hill, S. Jordan, H. Walker and H. Campbell
(2011). "Ecosystem services altered by human changes in the nitrogen cycle: a new
perspective for US decision making." Ecology Letters 14: 804-815.
10
-------
Environmental Protection Agency (1998). National Strategy for the Development of Regional
Nutrient Criteria. Washington, DC, United States Environmental Protection Agency: 53.
Environmental Protection Agency (2000). Nutrient Criteria Technical Guidance Manual, River
and Streams. Washington, DC, US Environmental Protection Agency: 253 pp.
Environmental Protection Agency (2010a). Technical Support Document for U.S. E.P.A.'s Final
Rule for Numeric Criteria for Nitrogen/Phosphorus Pollution in Florida's Inland Surface
Fresh Waters. Washington, DC, Environmental Protection Agency: 156 pp.
Environmental Protection Agency (201 Ob). "Water Quality Standards for the State of Florida's
Lakes and Flowing Waters." Federal Register 75(16): 4174-4226.
Environmental Protection Agency (2010c). Water Quality Standards for the State of Florida's
Lakes and Flowing Waters (Final Rule). EPA-HQ-OW-2009-0596 Washington, DC,
Environmental Protection Agency, Office of Water: 168 pp.
Environmental Protection Agency (2011). Memo from Nancy K. Stoner, Acting Assistant
Administrator to Regional Administrators, Regions 1-10. Working in Partnership with
States to Address Phosphorus and Nitrogen Poluution through Use of a Framework for
State Nutrient Reductions. Washington, DC, Environmental Protection Agency, Office of
Water: 6 pp.
Environmental Protection Agency (2014). U.S. EPA Expert Workshop: Nutrient Enrichment
Indicators in Streams. Proceedings April 16-18, 2013. EPA-822-R-14-004. Washington,
DC, US Environmental Protection Agency: 61 pp.
Florida Department of Environmental Protection (2013). Report to the Governor and Legislature:
Status of Efforts to Establish Numeric Interpretations of the Narrative Nutrient Criteria
for Florida Estuaries and Current Nutrient Conditions of Impaired Waters. Tallahassee,
FL, Division of Envireonmental Assessment and Restoration, Florida Department of
Environmental Protection: 68 pp.
Hirsch, R. M., D. L. Moyer and S. A. Archfield (2010). "Weighted Regression on Time,
Discharge and Seaon (WRTDS), with An Application to Chesapeake Bay River Inputs."
Journal of the American Water Resources Association 46(5): 857-880.
Karr, J. R. (1981). "Assessment of Biotic Integrity Using Fish Communities." Fisheries 6(6): 21-
27.
Keeler, B. L., S. Polasky, K. A. Brauman, K. A. Johnson, J. C. Finlay, A. O'Neill, K. Kovacs and
B. Dalzell (2012). "Linking water quality and well-being for improved assessment and
valuation of ecosystem services." Proc Natl Acad Sci U S A 109(45): 18619-18624.
Montana Department of Environmental Quality (2014). Base Numeric Nutrient Standards
Implementation Guidance. Version 1.0. Helena, MT, Montana Department of
Environmental Quality: 58 pp.
11
-------
Papworth, S. K., J. Rist, L. Coad and E. J. Milner-Gulland (2009). "Evidence for shifting
baseline syndrome in conservation." Conservation Letters 2(2): 93-100.
Paul, M., J. Butcher, D. Allen, L. Zheng, and T. Zi (2015). Methods for Computing
Downstrewam Use Protection Criteria for Lakes and Reservoirs. Prepared for US
Environmental Protection Agency by Tetra Tech, Inc, Research Triangle Park, NC and
Tetra Tech, Inc, Center for Ecological Sciences, Research Triangle Park, NC.: 144 pp.
Paul, M., A. Herlihy, D. Bressler, L. Zheng and A. Roseberry-Lincoln (2014). Methodologies for
Development of Numeric Nutrient Criteria for Freshwaters. Prepared by Tetra Tech,
Inc., Research Triangle Park, NC and Tetra Tech, Inc, Center for Ecological Science,
Research Triangle Park, NC.: 345 pp. . (includes Appendices 1-25)
Pauly, D. and V. Christensen (1995). "Primary production required to sustain global fisheries."
Nature 374: 255-257.
Reckhow, K. H. (1994). "Water-Quality Simulation Modeling and Uncertainty Analysis for Risk
Assessment and Decision-Making." Ecological Modelling 72(1-2): 1-20.
Rittel, H. W. J. and M. M. Webber (1973). "Dilemmas in a General Theory of Planning." Policy
Sciences 4: 155-169.
Ver Hoef, J. M., E. E. Peterson, D. Clifford and R. Shag (2014). "SSN: An R Package for Spatial
Statistical modeling on Stream Networks." Journal of Statistical Software 56(3): 1-48.
Vollenweider, R. A. (1976). "Advances in defining critical loading levels for phosphorus in lake
eutrophication." Memorie dell'Instituto Italiana di Idrobiologia 33: 53-83.
Yuan, L. L. (2006). Estimation and Application of Macroinvertebrate Tolerance Values.
Washington, DC, National Center for Environmental Assessment, Office of Research and
Development, U.S. Environmental Protection Agency: 89 pp.
12
-------
Methodologies for Development of Numeric
Nutrient Criteria for Freshwaters
Final Report
Prepared for
US EPA National Health and Environmental Effects Research Laboratory
Gulf Ecology Division
1 Sabine Island Drive
Gulf Breeze, FL 32561
Prepared by
Michael J. Paul
Terra Tech, Inc., Center for Ecological Sciences
1 Park Drive, Suite 200
PO Box 14409
Research Triangle Park, NC 27709
Alan Herlihy
Department of Fisheries & Wildlife, Oregon State University
c/o USEPA Western Ecology Division
200 SW 35th St.
Corvallis, OR 97333
David Bressler, Lei Zheng, and Ann Roseberry-Lincoln
Terra Tech, Inc., Center for Ecological Sciences
400 Red Brook Boulevard, Suite 200
Owings Mills, MD 21117-5172
July 30, 2014
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Nutrient pollution remains a vexing national pollution problem. Numeric nutrient criteria are one
dimension of a national nutrient reduction strategy. The United States Environmental Protection Agency
(USEPA) nutrient criteria guidance recommends the use of multiple lines of evidence (reference, stressor-
response, mechanistic modeling, and scientific literature) in developing nutrient criteria. This same
guidance also encouraged development of other scientifically defensible approaches, consistent with
water quality standards (WQS) regulations [40 CFR §131.10(b)(2)]. Therefore, refining existing
approaches and developing scientifically defensible alternative numeric nutrient criteria analytical
approaches remain technical challenges and, therefore, ripe research opportunities. This project
highlights two needs along this research horizon: 1) directly linking water quality conditions to nutrient
sensitive aquatic life uses (ALU), and 2) increasing ecological specificity in developing nutrient criteria.
The concept of using a receiving lake/reservoir approach for setting watershed criteria for influent streams
is a viable, creative, and relatively unexplored approach for numeric criteria setting for streams. The
originally proposed Florida stream criteria1 used a conceptually similar approach to derive downstream
protective values (DPV) for lakes and estuaries, which introduced the concept of DPVs. Developing
receiving water based approaches relies, first, on defining sensitive aquatic life uses (ALU) in receiving
lakes/reservoirs. In-lake targets are then developed to protect nutrient sensitive aquatic life. Identifying
nutrient sensitive aquatic life is non-trivial as the definition of lentic ALUs is not always clear and
sometimes conflicts with other uses. Therefore, defining sensitive ALUs is an important part of
developing DPVs and Part 1 of this report is a review of existing nutrient sensitive aquatic life use
measures in the US midwest (natural lakes) and southeast (reservoirs); and an exploration of potential
nutrient sensitive indicators for midwest lakes and southeast reservoirs based on a taxonomic nutrient
optima approach for comparison across assemblages (phytoplankton, zooplankton, macroinvertebrates,
and fish) in each of the two regions.
Part 2 of this report addresses development of reference numeric nutrient criteria for Oregon using
Oregon Hydrologic Landscape Regions (OHLRs). The United States Environmental Protection Agency
(USEPA) published recommended regional numeric nutrient criteria that relied strongly on reference
based approaches (e.g., USEPA 2001), which is a recommended approach for criteria derivation (USEPA
2000). However, USEPA strongly encouraged states to refine the recommended regional reference based
criteria using regionally specific reference sites and more resolved classification. Few states have
pursued this approach, in part due to a lack of technical examples. Part 2 of this report addresses the
development of refined regional reference criteria for Oregon (OR) based on OHRLs, providing not only
value in furthering OR nutrient criteria development, but also providing an example of the technical
process by which reference criteria can be developed.
federal Register, Vol. 75, No. 16, 4174, January 26, 2010. Water Quality Standards for the State of
Florida's Lakes and Flowing Waters; Proposed Rule.
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
This project was funded by the US Environmental Protection Agency through Task Order 0005 to Terra
Tech, Inc. under Contract EP-C-11-037. The Technical Representative and Program Manager was Dr.
Sandip Chattopadhay of Tetra Tech EM Incorporated in Cincinnati, OH. The EPA Task Order Manager
was Dr. James Hagy III of the US EPA National Health and Environmental Effects Research Laboratory,
Gulf Ecology Division, in Gulf Breeze, FL.
The authors wish to thank the National Science Foundation funded North Temperate Lakes Long Term
Ecological Research program and the Tennessee Valley Authority for support in using their data for
analysis. This included both facilitating data retrieval as well as answering many questions about the
data. Their patience and help are greatly appreciated.
The primary authors of this document were Michael Paul (Tetra Tech, Inc., Research Triangle Park, NC ),
Dr. Alan Herlihy (Oregon State University), David Bressler, Ann Roseberry-Lincoln, and Dr. Lei Zheng
(Tetra Tech, Inc., Owings Mills, MD).
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Executive Summary ES-1
1 Nutrient Sensitive Aquatic Life Uses 1
1.1 Overview 1
1.2 Literature Review: Aquatic Life Use Applications in the Upper Mississippi Basin and
Southeastern States 2
1.2.1 Upper Mississippi 2
1.2.2 South Eastern States 4
1.2.3 Summary 7
1.3 Sensitive Taxa Analysis 11
1.3.1 Datasets 11
1.3.1.1 North Temperate Lakes Long Term Ecological Research Program Data 11
1.3.1.2 Tennessee Valley Authority Long-Term Monitoring Data 16
1.3.2 Analysis 17
1.3.2.1 North Temperate Lakes Long Term Ecological Research Program Data 18
1.3.2.1.1 Phytoplankton dataset 18
1.3.2.1.2 Zooplankton dataset 22
1.3.2.1.3 Littoral Macroinvertebrate dataset 38
1.3.2.1.4 Fish Dataset 45
1.3.2.2 Tennessee Valley Authority Long-Term Monitoring Data 54
1.3.3 Summary 59
2 Nitrogen and Phosphorus in Oregon Hydrologic Landscape Regions (OHLR) 60
2.1 Overview 60
2.2 Data Sufficiency 60
2.2.1 Analysis 61
2.2.1.1 Stream Data 62
2.2.1.1.1 OHLR coverage 64
2.2.1.1.2 SOIL C/N coverage 69
2.2.1.1.3 SPARROW coverage 70
2.2.1.2 Stream Watershed Data 71
2.2.1.3 Lake Data 72
2.2.1.3.1 Lake OHLR Coverage 76
2.2.1.3.2 Lake Soil C/N coverage 83
2.2.2 Summary 83
2.2.2.1 Streams 83
2.2.2.2 Lakes 84
2.3 Data Analysis 85
IV
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
2.3.1 Data Preparation 85
2.3.1.1 OHLRData 85
2.3.1.2 Soil CandN data 86
2.3.1.3 SPARROW data 87
2.3.1.4 Reference Screening 87
2.3.2 OHLR Class Statistics 88
2.3.2.1 Streams 88
2.3.2.2 Lakes 101
2.3.2.3 Analysis of OHLR class differences 113
2.3.3 Surface water Nutrients versus Soil Chemistry 116
2.3.4 Streamwater nutrients versus SPARROW modeled estimates 121
2.4 Summary 124
2.4.1 OHLR Classification 124
2.4.2 Soil C, N GIS data 124
2.4.3 Comparison to SPARROW Model 124
3 References 125
Table 1 Summary of state lake and reservoir aquatic life use information 8
Table 2 North Temperate Lakes Long Term Ecological Research program data reviewed for use
in this analysis. Shaded cells indicate data selected for use 12
Table 3 Asynchronous benthic macroinvertebrate and water chemistry data 15
Table 4 Tennessee Valley Authority program data used in this analysis 16
Table 5 Summary statistics of water chemistry in the phytoplankton dataset 19
Table 6 Taxa indicator value rankings (family and higher on top, genus and lower on bottom) for
phytoplankton taxa. Indicator values for TP, TN, and Secchi depth were rank scaled
with 2 indicating nutrient sensitive taxa (low nutrient optima) and 5 indicating nutrient
tolerant taxa (high nutrient optima). Also shown are the sample sizes (N) for each
estimate 19
Table 7 Descriptive statistics for the TP, TN, Secchi Depth, and Chlorophyll for data set
(summer, all sites except the two bogs) 25
Table 8 Sample count by lake for optima calculations - each observation indicates that
zooplankton and nutrient/indicator data were available (e.g., for Allequash Lake there
were 26 samples for which zooplankton and nutrient/indicator data were available
from this lake) 26
Table 9 Taxa indicator value rankings (family followed by genus/species) for zooplankton taxa.
Indicator values for TP, TN, Chi a, and Secchi depth were ranked with 2 indicating
nutrient sensitive taxa (low nutrient optima) and 5 indicating nutrient tolerant taxa
(high nutrient optima). Also shown are the sample sizes (N) for each estimate 28
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Table 10 Descriptive statistics for family level nutrient optima across 17 taxa from Table 9 30
Table 11 Descriptive statistics for species/genus level nutrient optima from Table 9 30
Table 12 Metrics developed by scaling and averaging three taxa for each nutrient (TN, TP) or
indicator (Secchi, Chla) that showed clear relationship (either increased or decreased
abundance) with nutrient/indicator levels 33
Table 13 Summary statistics 38
Table 14 Macroinvertebrate Indicator values (IV) to total nitrogen (TN) concentrations from
different datasets. An IV of 2 is an indication of sensitivity to high TN concentration
and 5 is an indicators of tolerance to high TN concentrations. Also shown are the
sample sizes (N) for each estimate 40
Table 15 Descriptive statistics for nutrients (NH3, NO23, TN, TP) and nutrient indicators
(Secchi depth and Chlorophyll-a) 45
Table 16 Sample count by lake for calculations - each observation indicates that fish and
nutrient/indicator data were available 46
Table 17 Fish Indicator values (IV) to total nitrogen (TN) concentrations from different datasets.
An IV of 2 is indicative of sensitivity to high TN concentration and 5 is indicative of
tolerance to high TN concentrations. Also shown are the sample sizes (N) for each
estimate 49
Table 18 High and low nutrient tolerance 6-taxa metric optima 52
Table 19 Descriptive statistics for TVA lake dataset TN, TP and Chlorophyll-a 54
Table 20 Sample count by lake for optima calculations - each observation indicates that fish
and nutrient/indicator data were available 54
Table 21 Fish Indicator values (IV) for the TVA dataset. An IV of 2 is an indication of
sensitivity to high concentration and 5 is indicative of tolerance to high
concentrations. Also shown are the sample sizes (N) for each estimate 55
Table 22 Descriptive statistics for TVA nutrient optima 57
Table 23 Comparison of optima and optima categories for taxa in common between the WI and
TVA datasets 58
Table 24 Number of samples by stream size category 63
Table 25 Sample size by Level III Ecoregion 63
Table 26 Sample size by stream nutrient ecoregion 63
Table 27 Stream samples by OHLR aquifer permeability subclass 65
Table 28 Stream samples by soil permeability subclass 65
Table 29 Stream samples by climate subclass 66
Table 30 Stream samples by seasonality subclass 66
Table 31 Stream samples by terrain subclass 66
Table 32 Stream samples by OHLR class 66
Table 3 3 Distribution of sample stream watersheds by soil permeability class 71
Table 34 Distribution of lake samples by lake area class 75
VI
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Table 35 Distribution of lake samples by ecoregion 75
Table 3 6 Distribution of lake samples by nutrient ecoregion 76
Table 37 Distribution of lake samples by aquifer permeability subclass 78
Table 38 Distribution of lake samples by soil permeability subclass 79
Table 39 Distribution of lake samples by climate subclass 79
Table 40 Distribution of lake samples by seasonality subclass 79
Table 41 Distribution of lake samples by terrain subclass 80
Table 42 Distribution of lake samples by OHLR subclass 80
Table 43 Pearson correlations between mapped soil Total Nitrogen concentrations in the 4 soil
depth (cm) categories in the 684 stream watersheds 86
Table 44 Pearson correlations between mapped soil Total Carbon concentrations in the 4 soil
depth (cm) categories in the 684 stream watersheds 86
Table 45 Descriptive statistics for TP (ug/L) and TN (ug/L) by OHLR subclasses in Reference
Streams in Oregon 91
Table 46 Descriptive statistics for Soil N (kg/m2) concentrations by OHLR subclasses in
Reference Streams in Oregon 92
Table 47 Descriptive statistics for TP (ug/L) and TN (ug/L) by OHLR subclasses in Reference
Lakes in Oregon 104
Table 48 Descriptive statistics for Soil N (kg/m2) concentrations by OHLR subclasses in
Reference Lakes in Oregon 105
Table 49 Summary of F-statistics for one-way ANOVA testing for OHLR subclass effect. Lake
and streamwater chemistry data were loglO transformed before analysis 116
Table 50 Pearson correlation coefficients between soil nutrient GIS data and observed lake and
stream water nutrient chemistry in chemistry/habitat screened reference sites 118
Table 51 Summary statistics for the observed and SPARROW modeled nutrient concentrations
in the 413 Oregon stream and river sites that had SPARROW model output 121
Figure 1 Relationship between probability of occurrence and TP for Caenis. Solid line is the
mean relationship between probability of occurrence and TP determined by GAM
model. Dotted lines are estimated 90% confidence limits for the model fit. The dots
are the relative abundance of Caenis at a particular location. The vertical lines show
the central tendency (median cumulative probability) and limit (95% cumulative
probability) based on the GAM model 18
Figure 2 Examples of algal species response (capture probability) along the TP gradient,
showing TP preferred species Aphcmizomenon flos-aquae and less TP dependent
genus Schroederia. For all taxon response curves for phytoplankton, see Appendices
1-3 and 13-15 21
VII
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Figure 3 Performance of phytoplankton WA models with nutrient variables (LogTN and LogTP
in mg/L) and Secchi depth (m). Phytoplankton inferred environmental variable values
are the sum product of relative abundance and the optima at a sample/site 22
Figure 4 Examples of zooplankton taxa response curves (capture probability) along the TP
gradient. For all zooplankton response curves, see the Appendices 4-7 and 16-19 31
Figure 5 Performance of zooplankton WA inference models for nutrient variables (LogNH3, TN
and LogTP in mg/L), dissolved oxygen (DO in mg/L), Secchi depth (m), and
Chlorophyll a (LogChl a in ug/L). Inferred environmental variable values are the sum
product of relative abundance and nutrient optima across taxa at a sample/site 32
Figure 6 Three-taxa metric values representing high TN-tolerant (metric A) and intolerant taxa
(metric E, see Table 12) versus TN (loglO mg/L) 34
Figure 7 Three-taxa metric values representing TP-tolerant (metric B) and intolerant taxa (metric
F, see Table above) versus TP (loglOmg/L) 35
Figure 8 Three taxa metric values representing taxa found in waters of different clarity (Secchi
depth)(see metrics D andF in Table above) versus Secchi depth (m) 36
Figure 9 Three taxa metric values representing taxa found in waters of high (metric D) and low
Chlorophyll-a densities (metric F, see Table above) versus Chlorophyll-a (mg/L) 37
Figure 10 Principal component analysis of environmental variables in the invertebrate dataset.
The first axis is associated with nutrient (TP, TN, and Chi a) concentrations, while the
second axis is associated with Secchi Depth andpH 39
Figure 11 Examples of macroinvertebrate response curves (capture probability) along the TN
gradient. For all TN curves for macroinvertebrates, see Appendix 8 and 20 43
Figure 12 Performance of invertebrate WA inference models for four nutrient variables. (TN and
LogTP in mg/L, Log Chi a in ug/L). Invertebrate inferred environmental variable
values are the sum product of relative abundance and nutrient optima across taxa at a
sample/site. However, more than one third of the invertebrate data were qualitative
(recorded as 0 or blank in the original dataset), these values were coded as 0.1 to
separate from the actual quantitative values (range from 0.17 to 2490). As a result,
weighted averages based on abundance data were not really helpful. Instead, the final
indicator values most heavily rely on present/absence and tolerance values. (Spearman
r in the graphs are 0.2, 0.53, 0.19, 0.26 respectively). TN has the best performance
even with the qualitative dataset 44
Figure 13 Examples offish response curves (capture probability) along the TP gradient. For all
nutrient curves for fish, see Appendices 9-11 and 21-23 51
Figure 14 Performance offish WA inference models for nutrient variables (LogNH3, LogTN
and Log TP in mg/L), dissolved oxygen (DO in mg/L), Secchi depth (m), and
Chlorophyll a (Log Chi a in ug/L). Inferred environmental variable values are the
sum product of relative abundance and nutrient optima across taxa at a sample/site 52
Figure 15 Example plot of 6-taxa nutrient tolerance metric versus nutrients (High TN Nutrient
Metric vs. TN) 53
Figure 16 Example nutrient optima plot for TVA fish data (Bluegill relative abundance vs. TN).
For complete set of plots see Appendix 12 57
Figure 17 Oregon State map showing the location of the lake (Red dots) and stream (Yellow
dots) sample sites in the compiled ALUNC data. Shaded polygons show the Oregon
Hydrologic Landscape classes 62
viii
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Figure 18 Cumulative distribution functions of soil Total C and N for the compiled Stream
sample data 69
Figure 19 Scatterplot of distance from sample point to nearest SPARROW reach for each of the
796 sites in the compiled stream database versus site watershed area 70
Figure 20 Box and whisker plot of stream watershed area by aquifer permeability subclass
classified based on watershed dominance (>90% of watershed area in subclass).
Mixed sites are those with no dominant subclass 72
Figure 21 Cumulative distribution functions of soil Total C and N for the compiled lake sample
data 83
Figure 22 Five attributes and their subclasses that make up the OHLR classification 86
Figure 23 Stream data box and whisker plots of streamwater Total Phosphorus (TP) by OHLR
subclasses for chemistry/habitat screened reference sites (red) and all data (blue).
Boxes represent the 25th and 75th percentiles, the line in the box the median and the
whiskers the range 95
Figure 24 Stream data box and whisker plots of streamwater Total Nitrogen (TN) by OHLR
subclasses for chemistry/habitat screened reference sites (red) and all data (blue).
Boxes represent the 25th and 75th percentiles, the line in the box the median and the
whiskers the range 98
Figure 25 Stream data box and whisker plots of 0-200 cm depth Soil Nitrogen concentration by
OHLR subclasses for chemistry/habitat screened reference sites (red) and all data
(blue). Boxes represent the 25th and 75th percentiles, the line in the box the median
and the whiskers the range 101
Figure 26 Lake data box and whisker plots of lakewater Total Phosphorus (TP) by OHLR
subclasses for chemistry/habitat screened reference sites (red) and all data (blue).
Boxes represent the 25th and 75th percentiles, the line in the box the median and the
whiskers the range 108
Figure 27 Lake data box and whisker plots of lakewater Total Nitrogen (TN) by OHLR
subclasses for chemistry/habitat screened reference sites (red) and all data (blue).
Boxes represent the 25th and 75th percentiles, the line in the box the median and the
whiskers the range 110
Figure 28 Lake data box and whisker plots of 0-200 cm depth soil nitrogen concentration by
OHLR subclasses for chemistry/habitat screened reference sites (red) and all data
(blue). Boxes represent the 25th and 75th percentiles, the line in the box the median
and the whiskers the range 113
Figure 29 Scatterplot of surface water nutrients versus 0-200 cm depth soil nitrogen for both
lakes and streams 119
Figure 30 Scatterplot of surface water nutrients versus 0-200 cm depth soil carbon for both lakes
and streams 120
Figure 31 Scatterplots of SPARROW model nutrient concentrations versus observed nutrient
concentrations for beatable rivers in Oregon. The black line is the 1:1 line 122
Figure 32 Scatterplots of SPARROW model nutrient concentrations versus observed nutrient
concentrations for wadeable streams in Oregon. The black line is the 1:1 line 123
IX
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Executive Summary
Numeric nutrient criteria are being developed to protect the designated uses, including aquatic life uses,
of waterbodies nationwide. Developing numeric nutrient criteria that protect aquatic life is a scientific
challenge because nutrients are natural components of healthy ecosystems in natural levels, the effects of
nutrient pollution vary depending on many covariables and confounding stressors, and the measures that
states apply to protect aquatic life vary state by state.
Traditional analytical approaches for developing numeric criteria include reference distributions, stressor-
response relationship modeling, and mechanistic modeling, with scientific literature frequently used to
support these approaches. Derivation of stream criteria using reference approaches has been challenged
on the grounds of over-protection while derivation using stressor-response modeling has been challenging
due to the wide range of co-variates that affect stream responses (e.g., flow, substrate, light, grazing) and
the panoply of co-occurring stressors (e.g., sediment). For this reason, developing stream numeric criteria
has been difficult.
Water quality standards regulations require that criteria "ensure... .attainment and maintenance of the
water quality standards of downstream waters" [40 CFR 131.10(b)]. This requires that any stream criteria
in addition to protecting in-stream uses also be required to protect receiving water uses downstream.
Stressor-response models in such receiving waters, such as lakes, are frequently more precise because
there are fewer confounding variables affecting nutrient response in open water systems with longer
residence times. As a result, the central research question of this research is identifying concentrations
that need to be met in tributary streams in order to assure protection of sensitive aquatic life uses in
receiving waters. Such research is not without precedent, and the development of downstream protective
values was pursued in the United States
Environmental Protection Agency (USEPA)
proposed Florida numeric criteria effort
(Federal Register, Vol. 75, No. 16, 4174,
January 26, 2010). This research effort is a
multi-step process: 1) develop sensitive
aquatic life use indicators for receiving waters;
2) develop numeric nutrient criteria for the
receiving water to protect sensitive aquatic
life; 3) calculate watershed loads and stream
concentrations necessary to protect the lake.
The first part of this report describes step 1 in
this process and attempts to define sensitive
aquatic life uses for natural lakes in the
midwest United States (US) and for reservoirs
in the southeastern US. It consists of a
literature review of existing aquatic life use
measures and an analysis of biological
assemblage changes along nutrient gradients in the two regions.
Streams
Central Research Question:
What concentrations need to be met
here...
...to protect uses here?
The second part of this report addresses shortcomings in reference distribution approaches for streams and
lakes; namely, the lack of regional specificity. The original draft regional reference criteria were
developed at coarse scales and states were encouraged to refined these distributions using more highly
resolved data from their own state or nutrient regions (USEPA 2000). Part 2 of this report describes
analysis of reference stream and lake nutrient distributions using Oregon Hydrologic Landscape Regions,
a more refined regionalization than previously used.
ES-1
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Executive Summary
Nutrient Sensitive Aquatic Life Uses - Literature Review
A review of state water quality standards and consolidated assessment and listing methodologies was
conducted to characterize the measures currently being used to assess aquatic life use in lakes in Midwest
(WI, MN, SD, IA, IL, MO) and Southeast (KY, TN, NC, SC, MS, AL, GA, FL) states. The results of this
review were surprising. Most of these states evaluate aquatic life use in lakes and reservoirs using indirect
measures (chlorophyll a, Secchi depth clarity, and dissolved oxygen) rather than direct biological
measures, as frequently occurs in assessing stream aquatic life uses. Of these states, only Florida has a
direct use measure (macrophytes). However, several states either supplement water chemical information
with biological information (e.g., IA uses macrophyte information in their assessment, IL uses percent
macrophyte cover as part of an aquatic life use index along with chlorophyll a, clarity, and sediment) or
they are developing direct biological indices (e.g., WI is developing a macrophyte index. MN is
collaborating with the National Lakes Assessment to pursue a lake fish index). The USEPA National
Lakes Assessment, in contrast, used direct measures of phytoplankton, sediment diatoms, zooplankton,
and macroinvertebrates in their first round survey.
Nutrient Sensitive Aquatic Life Uses - Analyses to Identify Nutrient
Sensitive Aquatic Life
Due to the lack of state direct aquatic life use measures, it was necessary to go outside state databases to
identify sufficient data to comparing the response of biological assemblages in lakes to nutrient gradients.
The US National Science Foundation Long-Term Ecological Research (LTER) program, Northern
Temperate Lakes (NIL) site in Wisconsin proved invaluable in housing a wealth of biological
assemblage data for a wide range of lakes in Wisconsin that varied in nutrient concentrations. The core
long-term monitoring data and the more than sixteen associated research studies provided data on
phytoplankton, zooplankton, macroinvertebrates, and fish for a variety of lakes in Wisconsin. The data
are available online and LTER staff facilitated its
assembly into a useful database of water chemical and
biological response data.
Optimum
Caenis
Similarly, the Tennessee Valley Authority (TVA)
manages more than 30 reservoirs in the southeastern US
and has been monitoring water chemistry and fish
assemblages in their lakes for decades. These data were
also made available for use in comparing responses in
Midwestern lakes to those in southeastern reservoirs.
An indicator value approach (Yuan 2006) was used to
identify sensitive aquatic life along gradients in nutrient
conditions across lakes and over time. Nutrient optima
were calculated for different taxa within each assemblage
group and used to evaluate the range in sensitivities to
nutrients. Some taxa combination models were also explored for their utility in deriving nutrient sensitive
indicators. Optima were converted into ranks from 2 (nutrient sensitive) to 5 (nutrient insensitive).
0 005 001 0 02
Total Phosphorus (mg/L)
ES-2
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Executive Summary
Diacyclops.thomasi
ts-
0001 0003 0,01 003
Total Phosphorus (mg/L)
/Wesocyc/ops edax
i-s-
CL
1
§
3
S 8
0001 0.003 0.01 003
Total Phosphorus (mg/L)
Two Wl zooplankton
taxa exhibiting
different nutrient
optima.
In the Wisconsin data, phytoplankton taxa were relatively poor
predictors of nutrient concentration and exhibited fairly uncertain
optima in response to nutrient gradients. This was a surprise as these
taxa, and diatoms in particular, almost universally exhibit strong
sensitivity to water chemistry (including nutrients) and are frequently
used to infer water quality conditions. The fairly narrow gradient over
which there was sufficient data and the coarse level of the existing
taxonomic information that was available to use likely explain the lack
of more convincing phytoplankton responses.
In contrast, Wisconsin zooplankton data produced better nutrient
inference models and a range of nutrient sensitivities. There were more
than 53 zooplankton taxa available and more than 12 showed
consistent responses on both ends of the sensitivity spectrum.
Combination metrics were constructed from based on these
sensitivities to produce low nutrient and high nutrient sensitive multi-
taxa indicators that showed significant responses to nutrient gradients.
Macroinvertebrate data from Wisconsin was intermediate in terms of
its sensitivity to nutrient gradients in lakes, but there was also limited
range in nutrient conditions for the lakes with macroinvertebrate data.
Models were better than phytoplankton, but not as precise as
zooplankton.
Fish produced surprisingly precise nutrient inference models from the
Wisconsin data and exhibited a wide range in nutrient sensitivities.
This was somewhat surprising given the distance offish taxa
from nutrient inputs but may reflect a pronounced influence of -vo -
nutrient loading on oxygen concentrations in lakes. The fish
indicator values were also used to produce multi-taxa indicators
that were responsive to nutrient gradients.
a. -1.5 -
TVA fish data also produced robust models and also generated a
variety of nutrient sensitivities among fish taxa.
Indicator values and ranks were compared for taxa common to
the WI and TVA datasets and, not surprisingly, produced
positive but imprecise correlations across the various nutrient
measures (0.23 to 0.38) between the two regions. It is likely that
evolutionary as well as waterbody management factors
influenced the nature of taxonomic sensitivities across the two
disparate regions. At least 8 taxa did show surprising
consistency in rank sensitivity for rock bass and walleye and
insensitivity for largemouth bass and golden shiners.
1
; : -
; - -
-25
-20 -1.5
Measured TF
-• :
TP inference model
using Wl fish data
In summary, nutrient sensitivity appears consistently across taxonomic assemblages. Zooplankton and
fish produced more precise inference models than phytoplankton or macroinvertebrates, but a lot of the
basis for this is likely methodological. Nutrient sensitive aquatic life use measures can be derived from
single or multiple assemblages from the indicator value/optima analysis easily. The USEPA National
Lake Assessment has shown this is possible with phyto- and zooplankton. States should be encouraged to
ES-3
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Executive Summary
develop more direct measures of sensitive aquatic life uses in lakes, which would be useful for a variety
of water quality program applications including nutrient criteria development.
One remaining issue that was unresolved was translating aquatic life uses narrative with assemblage
information. Reservoirs, in particular, provide a quandary. These are constructed lake-like systems that
intrinsically do not have aquatic life use reference conditions. Identifying the proper aquatic life
assemblage expectation is difficult for such systems. Moreover, protection of diversity per se as is often
presumed for natural systems, may not be the most valued aquatic life use for such an artificial system.
Defining such management goals will be important to define aquatic life use targets and, therefore,
numeric nutrient criteria needed to protect them.
The next step in this research is translating these sensitive aquatic life use indicators into lake numeric
nutrient targets to assure their protection and then translating in-lake targets into tributary targets. These
are the next steps in this ongoing research.
and in
- Data
USEPA nutrient criteria guidance recommends the use of multiple lines of evidence (reference, stressor-
response, mechanistic modeling, and scientific literature) in developing nutrient criteria. However, EPA
guidance focused strongly on reference based approaches for criteria development and the USEPA
recommended regional criteria were developed using this approach, as were USEPA's final criteria for
streams in Florida. USEPA strongly encouraged states to refine the reference analyses used in the
recommended regional criteria using more refined reference sites and more highly resolved classification.
However, few states have taken advantage of this, in part due to a lack of technical examples. The second
part of this report is focused on refining regional reference criteria for Oregon which provides not only a
product of value to the state in furthering nutrient criteria for this line of evidence, but also an example of
how refined regional reference criteria can be developed. This part is split into two sections, the first
describes the data and its sufficiency for estimating reference criteria by OHLR and the second section the
analysis to calculate the criteria.
The focus of this work was assembling data on streams and lakes across these regions, describing nutrient
distributions in entire and reference populations, and describing the sufficiency of spatial data coverage
across those regions. Existing data from previous work and new data from state and federal monitoring
programs were compiled. All the data were compiled into one comprehensive dataset. The sufficiency of
data for spatial analysis in Oregon is reported by OHLR unit based on the spatial coverage of all these
parameters across three geographic layers: OHLRs, soil carbon and nitrogen classes (C/N), and
SPARROW model units. The latter was only applicable for stream point coverage. The OHLR class is a
CsATS ^ combination of 5 attributes;
aquifer permeability (3
subclasses), Soil
Permeability (3 subclasses),
Climate (5 subclasses),
Seasonality (3 subclasses),
and Terrain (3 subclasses).
Aquifer
Climate Class Seasonally Sub-Class Permeability Class Terrain Class
V - Very wet
W-Wet
M - Moist
D- Dry
S - Semiarid
A - Arid
w- Fall or winter
s - Spring
u - Summer
H - High
M - Moderate
L-Low
M - Mountain
T - Transitional
F - Flat
Soil
Permeability Ctass
H - High
M - Moderate
L - Low
OHLR Attribute Classes and Subclasses
ES-4
-------
Methodologies for Development of Executive Summary
Numeric Nutrient Criteria for Freshwaters
Stream data sufficiency summary:
• There were 796 unique stream sample sites: 66% from small streams; 20% from intermediate size
streams; 14% from beatable rivers.
• All Level III ecoregions were represented. The Coast Range had the most samples (N=217);
• All nutrient ecoregions were represented with a minimum N of 44;
• Only 100 or so actually OHLR combination classes exist. Only 8 OHLR classes have a sample
size greater than 20 sites. The five OHLR attribute classes all had sufficient sample sizes;
• Many watersheds were dominated by one OHLR category, supporting analysis of individual
OHLR attribute classes;
• Soil C/N deciles ranges were 10-40 kg/m2 Total C and 0.5-2.0 kg/m2 Total N. Different soild
depths were similar and were combined;
• SPARROW data are weighted towards larger systems but should be linkable to entire dataset;
Lake data sufficiency summary:
• There were 366 lake sample sites, well distributed among area classes from 500ha;
• Level III ecoregions were well represented except for Columbia Plateau, Klamath Mountains,
Snake River Plain, and Willamette Valley, which all had fewer than 13 samples;
• The Western Forested Mountains nutrient region dominated the sample distribution (89%) and
the Willamette Valley nutrient ecoregion had a small sample size (N=13);
• Only 4 OHLR classes contained more than 10 sample sites. The five OHLR attribute classes all
had sufficient sample sizes;
• Soil results were similar to streams.
Nitrogen and Phosphorus in Oregon Hydrologic Landscape Regions
(OHLR) - Data Analysis
Additional adjustments to the datasets were made to resolve inequities in sample sizes for particular
OHLR classes and data were organized accordingly. Streams and lakes with large watershed areas often
had multiple OHLR classes in their watersheds and were not dominated by any one class. Therefore, the
analysis of the OHLR class data was restricted to streams that had watershed area < 1000 km2 (which
dropped 85 sites from the analysis) and lakes larger than 250 ha (which eliminated all river
impoundments).
A set of least disturbed reference sites were identified for calculating reference distribution statistics.
Stream reference sites were identified using site measurements of non-nutrient water chemistry and
physical habitat, which varied by ecoregion. There were 152 streams considered to be in least-disturbed
reference condition. Stream reference condition was also defined based on biological condition using
EPT richness (number of different mayfly, stonefly, and caddisfly taxa) thresholds by region. There were
322 sites considered to be reference by biological condition.
A semi-quantitative aerial photo screening method was used to identify least-disturbed lakes. Lakes were
scored on an integer scale of 0 (undisturbed) to 3 based on degree of disturbance in each of seven
categories (agricultural, residential, recreation, logging, roads, mining, and commercial) and thresholds
ES-5
-------
Methodologies for Development of Executive Summary
Numeric Nutrient Criteria for Freshwaters
varied by ecoregion. Overall, 113 of the 264 lakes in the final analysis set were considered least disturbed
reference.
Descriptive statistics were then calculated for stream and lake nutrient chemistry as well as soil N.
Comparisons were made between surface water and soil chemistry and stream chemistry and SPARROW
model estimated concentrations.
Summary of analysis:
• There was a consistent pattern of increasing reference TP concentrations as climate progressed
from very wet, to wet, to moist, to dry in streams and lakes. There was also a strong pattern of
increasing TP with terrain (Transitional > Mountain) and seasonality (Winter > Spring) in lakes;
• Lake TN had highly significant relationships with climate (increasing wet to dry) and terrain
(Transitional > Mountain) and to a lesser extent aquifer permeability (L > M=H). Stream TN was
weakly related to soil permeability (L < M=H);
• Soil N was strongly related to both climate (Dry < Wet) and aquifer permeability in the stream
dataset;
• Overall, the most consistent patterns were with climate and seasonality. In general, stream water
nutrient concentrations increased and soil N concentrations decreased as climate changed from
wet to dry;
• Correlations between soil N and C and surface water TP and TN were very weak or non-existent;
• There were no relationships between observed stream water concentrations and SPARROW
model results for either TP or TN. However, observed values in this database represent summer
base flow conditions and SPARROW models mean annual concentration which likely explains
the discrepancies.
ES-6
-------
Methodologies for Development of Part 1: Nutrient Sensitive Aquatic Life Uses
Numeric Nutrient Criteria for Freshwaters
1 Nutrient Sensitive Aquatic Life Uses
LI
The United States Environmental Protection Agency (USEPA) nutrient criteria guidance (USEPA 2000)
recommends the use of multiple lines of evidence (e.g., reference, stressor-response, mechanistic
modeling, and scientific literature) in developing nutrient criteria. Beyond these standard approaches,
USEPA guidance, including the lakes guidance (USEPA 2000), offered additional approaches and
encouraged development of other scientifically defensible approaches, consistent with water quality
standards (WQS) regulations [40 CFR §131.10(b)(2)]. Therefore, developing scientifically defensible
numeric nutrient criteria analytical approaches remained a technical challenge and, therefore, a research
opportunity. Part 1 of this project addressed one need along this research horizon: directly linking water
quality conditions to nutrient sensitive aquatic life uses (ALU).
For streams, nutrient criteria development challenges are particularly acute, largely due to the variety of
factors (e.g., flow, temperature, light, scour, and grazing) that confound nutrient-response relationships,
and the variety of co-occurring pollutant stressors that impact sensitive aquatic life (e.g., sediment). The
concept of using a receiving lake/reservoir approach for setting watershed criteria for influent streams is
viable, creative, and relatively unexplored. It has the added benefit of meeting two criteria requirements -
protecting uses and downstream receiving waters. The stream criteria proposed by USEPA for Florida2
were developed using conceptually similar analyses to derive downstream protective values, which
introduced the concept; this research furthers that approach.
Developing receiving water based approaches relies, first, on defining sensitive ALUs in receiving
lakes/reservoirs. This is non-trivial as the definition of lentic ALUs is not always clear and sometimes
conflicts with other uses. Another problem is that development of lake bioassessment tools lags behind
streams (USEPA 1998, USEPA 2009); therefore, defining discrete, direct targets has been difficult and
has relied, principally, on the trophic state concept. Also, the management of many lakes and reservoirs
focuses on productive sport fisheries, resulting in active fertilization programs (e.g., VWRRC 2005) and
algal targets that pose a trade-off between aquatic life uses and competing uses like fishing (low clarity,
high production) and swimming (low production, high clarity). But, criteria must protect the most
sensitive use [40 CFR 131.11(a)(l)]. Therefore, defining sensitive ALUs is an important part of this
research. Once sensitive uses are defined, quantifying nutrient concentrations needed to protect them
provides nutrient criteria that can be projected up to streams across the watershed to support ALU
attainment in downstream lakes and reservoirs. Part 1 of this research is focused on the problem of
identifying and modeling sensitive ALUs in lakes/reservoirs.
This document includes a summary of existing lake aquatic life use characterization efforts among states
followed by an analysis of the sensitivity of different aquatic life assemblages to nutrient gradients. The
review was conducted to evaluate the state of applied practice related to aquatic life use protection from
nutrient enrichment. The analyses were conducted to identify which assemblages present sensitive
responses and, therefore, potential applicability for use in deriving in-lake criteria to protect aquatic life.
These analyses used available sample data from the Upper Mississippi Basin lakes and were
compared/contrasted with data from southeastern US reservoirs. These two lentic system types (Upper
Mississippi natural vs. southeaster US constructed) differ not only in climate and limnological drivers, but
also in the values placed on different aquatic life and will ultimately be compared/contrasted in follow on
work. This was reflected in the literature review.
2 Federal Register, Vol. 75, No. 16, 4174, January 26, 2010. Water Quality Standards for the State of
Florida's Lakes and Flowing Waters; Proposed Rule.
-------
Methodologies for Development of Part 1: Nutrient Sensitive Aquatic Life Uses
Numeric Nutrient Criteria for Freshwaters
The Clean Water Act requires water quality standards to be protective of human health and aquatic
organisms. Specifically water quality must support the "protection and propagation offish, shellfish, and
wildlife, and provide for recreation in and on the water" (40 CFR 131.100). Protection of water bodies is
achieved in part by defining designated uses which provide more specifically manageable goals. Effective
criteria protect and support designated uses. For example, broad classes of common designated uses
include the protection of aquatic life (e.g., fish, birds, and shellfish), recreation (e.g., swimming, boating,
fishing), water supply, and livestock or irrigation uses. Some categories of uses including the protection
and propagation of fish, shellfish, and wildlife are subcategorized by States to ensure protection for a
particular group of ecologically or economically important species. For example, designated aquatic life
uses (ALU) may include protection of coldwater fish, warmwater fish, or aquatic wildlife. Some states
designate separate use classifications for stocked/recreational fisheries and natural, self-supporting
populations. Aquatic life uses may also vary based on water body type (i.e. lakes/reservoirs versus rivers).
Water quality criteria are adopted to protect designated uses and, for aquatic life in lakes, this includes a
variety of potential aquatic life endpoints (USEPA 1998) that vary in nutrient sensitivity. Many game
fish species which are highly valued aquatic life, benefit from enrichment (e.g., Ney 1996, Maceina and
Bayne 2001). This conflicts with the numerous, well documented negative effects of nutrients on lakes
(e.g., Edmondson 1970, Schindler 1974, OECD 1982, Carpenter et al. 1998) including responses of
aquatic life at all trophic levels. Algae species composition is altered by competitive differences for
nutrients (e.g., Tilman 1982), as are heterotrophic microbial assemblage structure and activity (Pace and
Cole 1996, Dodds and Cole 2007). Aquatic macrophytes also respond to nutrients, with shifts in species
composition and coverage (Sand-Jensen and Borum 1991). There are a variety of invertebrates impacted
by nutrient enrichment including zooplankton species and zoobenthos (e.g., Schindler 1990). Lastly,
while production of certain fish species may benefit from enrichment, this is not true of all species and
trade-offs in species response are frequently observed, depending on the nature and extent of enrichment
(Leach et al. 1977, Jeppesen et al. 2002, Mehner et al. 2005).
This review of nutrient sensitive ALU measures in the Upper Mississippi and southeastern United States
focused on current state practices, with specific emphasis on application of the multiple aquatic life uses
described above.
Two aquatic life uses are designated in Illinois: protection of general aquatic life and protection of
indigenous aquatic life (IEPA 2012). The state uses water quality and the Aquatic Life Use Index (ALI)
as the primary tool for assessing aquatic life uses in inland lakes (not including Lake Michigan, Table 1).
The ALI score is calculated using the Trophic State Index (TSI, Carlson 1977), the peak percent surface
coverage of macrophytes (June-August) and concentration of nonvolatile suspended solids. The TSI is
calculated using median total phosphorus samples (epilimnetic), chlorophyll a concentrations, and/or
Secchi disk transparency values. The TSI values range between <30 (clear, nutrient poor lakes) to >70
(highly productive, nutrient rich). The ALI uses the resulting scores to indicate a degree of support for
aquatic communities (good/fair/poor), with higher scores suggesting greater impairment.
Data are collected at least five times per year between April and October, and samples are collected from
one or more site on a lake.
-------
Methodologies for Development of Part 1: Nutrient Sensitive Aquatic Life Uses
Numeric Nutrient Criteria for Freshwaters
In their 2012 assessment process, Iowa describes two lake classes for assessing protection of aquatic life
uses - one for lakes and reservoirs and a second for shallow lakes. Both rely principally on the TSI using
chlorophyll and Secchi depth, as well as traditional water chemical measures. In addition, the Iowa
Department of Natural Resources (IDNR) conducts fishery surveys in lakes and maintains a database of
fish survey data. These include data for macrophyte type and coverage and also include length, weight,
age class and growth data for sport fishes collected by electrofishing (IDNR, 2013a). Typical fish species
reported include crappie, sunfish, bass, walleye, and catfish. Information from IDNR field staff on aquatic
macrophytes and macroinvertebrates is used to supplement the lake water quality assessments (Table 1).
Currently Minnesota has one general aquatic life use, but the State is also investigating tiered aquatic life
uses to include: 1) exceptional uses for high quality resource waters; 2) general uses; 3) modified uses for
waters with anthropologically modified habitat (ie, channelized streams); and 4) limited uses for water
with limited habitat (ie, ephemeral streams, concrete channels, etc) (MN PCA 2013). It is unclear whether
any of these categories will be applied to lakes, though tiered aquatic life uses in other states generally
apply to streams and rivers.
Assessment for support of aquatic life uses in lakes is based on general chemical and physical
measurements including dissolved oxygen, pH, temperature, ammonia, chlorine, and metals. Management
of lake eutrophication is considered part of the protection of aquatic recreation designated use (which
includes fishing uses). Measurements of total phosphorus, chlorophyll a, and Secchi depth are collected
over at least 8 dates between June and September (MN PCA, 2012, Table 1).
Bioassessment procedures are not used to assess lake health, but such methods are currently under
development. In 2009, the state collaborated with US EPA during the National Lake Assessment.
Minnesota used the data from the national survey work to start a fish Index of Biotic Integrity (IBI) for
lakes, but the tool has not yet been developed to the point where scores can be categorized to measure the
health of a system (good/fair/poor, MN PCA 2009). Aquatic vegetation, which was also surveyed as part
of the 2009 study, was closely correlated with the fish IBI, but additional research is needed to determine
which metrics work best in lakes larger than 2000 acres.
Missouri has three classes of aquatic life uses that apply to lakes. These include coldwater fishery,
protection of aquatic life in a general warm water fishery), and protection of warm water aquatic life and
human health for fish consumption. Typical water quality measurements to determine if uses are
supported include chlorophyll a, nutrients, dissolved oxygen, and temperature (Table 1).
South Dakota has six designated uses that incorporate fish propagation, combined with coldwater or
warmwater and permanent versus marginal habitats. These uses include the following: coldwater
marginal fish life propagation; coldwater permanent fish life propagation; fish and wildlife propagation,
recreation, and stock watering; warmwater marginal fish life propagation waters, warmwater permanent
fish life propagation waters, and warmwater semi-permanent fish life propagation waters.
Previously, lakes were assessed using the Trophic State Index to classify eutrophication. However, the
TSI was dismissed from the 2010 listing methodology after years of modifications to address
uncertainties. Many lakes could not meet the assigned numeric TSI targets due to unattainable watershed
phosphorus reductions required to meet the targets. Lakes were often fully supporting based on numeric
-------
Methodologies for Development of Part 1: Nutrient Sensitive Aquatic Life Uses
Numeric Nutrient Criteria for Freshwaters
water quality standards though considered impaired for TSI. A clear linkage was not often evident to
suggest that the fishery was impaired due to exceeding the TSI target (SD DENR 2010).
The Fish and Aquatic Life (FAL) use is designated for streams, rivers, and lakes in Wisconsin. Lakes and
reservoirs are classified on the basis of size, stratification characteristics, and hydrology. These
characteristics are often used to determine the expected biological communities (WI DNR 2013). The ten
categories used to classify Wisconsin's lakes and reservoirs are
• Lakes/Reservoirs < 10 acres
• Lakes/Reservoirs > 10 acres
o Shallow seepage
o Shallow headwater
o Shallow lowland
o Deep seepage
o Deep headwater
o Deep lowland
• Other classifications for lakes of any size
o Spring ponds
o Two-story fishery lakes
o Impounded flowing waters
The three "other" classifications listed above are specifically designed to represent unique natural
fisheries communities. Spring ponds generally consist of shallow headwater lakes and support a coldwater
fishery, that include trout and other cold water species. Two story fishery lakes also support a coldwater
fishery but are generally deeper (>50ft) and larger (>5 acres) than spring ponds and are thermally
stratified during the summer. The coldwater species are supported in the hypolimnion. Impounded
flowing waters are streams or rivers that have been impounded in some other manner, but are not
classified as reservoirs. These systems are considered lotic and aquatic uses are evaluated using the river
and stream criteria generally applied to the body of water entering the impoundment. Use designations
that apply to streams and rivers under Wisconsin's FAL include Coldwater Community, Warmwater
Sport Fish Community, Warmwater Forage Fish Community, Limited Forage Fish Community, and
Limited Aquatic Life Community.
Support of designated uses in lakes is determined by water quality metrics that are shown to correlate
strongly with communities offish, macroinvertebrates, and aquatic plants. Wisconsin uses the TSI. It is
calculated using chlorophyll a concentration or Secchi depth values and also considers measured
concentrations of phosphorus.
Wisconsin is exploring the development of an aquatic macrophyte index (AMCI) to supplement
traditional trophic state measures for assessing aquatic life use in lakes. The AMCI metrics are still under
development and WDNR is not using them in assessments yet. Wisconsin is still trying to determine
which metrics are best suited to judge impairment and signify degraded habitat or eutrophication (WI
DNR 2013).
-------
Methodologies for Development of Part 1: Nutrient Sensitive Aquatic Life Uses
Numeric Nutrient Criteria for Freshwaters
Alabama lakes include two uses related to aquatic life use: fishing and propagation offish and wildlife.
The TSI based on chlorophyll a is used to determine trophic status of lakes in Alabama (Table 1). Score
ranges used to predict trophic status are as follows: hypereutrophic (>70), eutrophic (50-69), mesotrophic
(40-49), oligotrophic (<40) (ADEM 2010).
Aquatic life uses in Georgia are described for general aquatic life use to protect propagation of fish,
shellfish, game, and other aquatic life. Lake evaluations use traditional chemical measurements including
chlorophyll a, TN, and TP in some lakes (Table 1). No biological indicators are evaluated for regulatory
purposes.
The Georgia Department of Natural Resources (GA DNR) manages sport fishing populations in
reservoirs and large rivers in Georgia and has collected fisheries data since 1987. These surveys have
been used to estimate the impact of pollution and changes to physical habitat, track sport fish population
levels, and predict fishing trends (GA DNR 2013).
Florida has one general designated use to protect freshwater fish and wildlife propagation. It is the only
state that uses a method of biological measurement in addition to typical water chemistry parameters for
regulatory evaluation of lakes and reservoirs (Table 1). Methods evaluating the macroinvertebrate
community and vegetative community have both been investigated for use in Florida lakes. Previously,
the State developed a bioassessment procedure incorporating macroinvertebrates to assess the biological
condition of lakes (Gerritsen et al 2000). However a more recent bioassessment method incorporating
vegetation is currently utilized because it appears to be more closely associated with human disturbance
parameters. The Lake Vegetation Index (LVI) is used to evaluate the health and balance of aquatic plant
communities (Fore et al 2007). The LVI is a multimetric index that is comprised of four individual
metrics including % native taxa, % Florida Exotic Pest Plant Council [FLEPPC] Category 1 invasive
exotic taxa, coefficient of conservatism of the dominant taxon or co-dominant taxa, and % sensitive taxa
(FL DEP, 2011). These metrics are not specific to nutrients, but measure a range of potential effects.
Sampling for the LVI is conducted April 1-November 30 in a defined southern Florida region, and from
May 1-October 31 in the northern region.
In Kentucky, aquatic life uses apply to over 218,000 acres (98%) of lakes and reservoirs in the state (KY
DEP, 2010). At least two categories of waters are defined for lakes including coldwater aquatic habitats
and warmwater aquatic habitats. Coldwater habitats are those waters that are supportive year round of
self-sustaining or reproducing trout populations. Warmwater habitats are those lakes capable of
supporting indigenous warm water aquatic species.
Lakes are monitored at least three times during the "growing season" (spring, summer, fall). In 2010,
122,247 acres (59%) of lakes supported their designated aquatic life uses. Most lakes that are impaired are
listed due to nutrients, pH, or dissolved oxygen. Evaluation of lake condition is based on the Carlson
Trophic State Index (TSI) using chlorophyll a, and lakes are ranked based on trophic state as oligotrophic,
mesotrophic, eutrophic, and hyper-eutrophic (KY DEP, 2010). Other water quality variable measured
include the following: unionized ammonia, nitrite-nitrate, total phosphorus, TKN, total soluble
phosphorus, soluble reactive orthophosphate, total organic carbon, total suspended solids, chlorides,
sulfates, alkalinity, hardness, chlorophyll a, DO, pH, temperature, and specific conductance (Table 1).
In addition to State efforts to monitor lakes in a regulatory capacity, long term lake monitoring occurs at
Kentucky Lake as part of Murray State University's Hancock Biological Station Center for Reservoir
-------
Methodologies for Development of Part 1: Nutrient Sensitive Aquatic Life Uses
Numeric Nutrient Criteria for Freshwaters
Research (Murray State University, 2013). Since 2005, water quality parameters have been recorded
multiple times per day at a surface and bottom site. Available parameters include water temperature,
specific conductance, dissolved oxygen, pH, oxidation-reduction potential, turbidity, and chlorophyll a
(surface measurement only).
While the lake evaluation in the regulatory context does not involve fish or macroinvertebrate data,
several programs routinely collect some biological information for fisheries management purposes. The
Kentucky Department of Fish and Wildlife Resources collects fisheries data used to evaluate the health
and viability of the state's lakes. This includes general sportfish surveys for lakes and tailwaters (riverine
segments immediately below dams) and targeted surveys evaluating largemouth bass populations
(Kentucky Department of Fish and Wildlife, 2013).
In 2010, 78 lakes and reservoirs were sampled to evaluate general sportfish populations in Kentucky lakes
and tailwaters (Kentucky Department of Fish and Wildlife, 2011)3. Measured parameters offish include
catch rates, mortality, recruitment, length/weight, age/growth. Additionally, water quality measurements
including temperature and dissolved oxygen were recorded along with physical habitat data. Reported
species data usually include bass, sunfish, crappie, catfish, and other fishes valued as sportfishes.
Largemouth bass assessments are specifically conducted for the following lakes: Barren River Lake,
Buckhorn Lake, Carr Creek Lake, Cave Run Lake, Cedar Creek Lake, Dewey Lake, Fishtrap Lake,
Grayson Lake, Green River Lake, Greenbo Lake, Guist Lake, Herrington Lake, Kentucky Lake, Lake
Barkley, Lake Beshear, Lake Carnico, Lake Cumberland, Lake Reba, Laurel River lake, Martins Fork
Lake, Nolin River Lake, Paintsville Lake, Rough River Lake, Taylorsville Lake, Wilgreen Lake, and
Yatesville Lake4. The Spring electrofishing surveys have been conducted since 2003 and are designed to
predict the well-being of the bass population and estimate fishing trends for particular lakes. Lakes are
divided into categories of > 1000 acres and <1000 acres, and are rated as poor to excellent, depending on
condition of the fishery. For each lake, parameters are recorded for number of bass in 4 size/age
categories (less than 1 year, 12-14.9 inches, >15 inches, >20 inches) and bass length at age 3.
Mississippi
Mississippi has a generic aquatic life use that applies to all surface waters including lakes and involves
propagation offish and wildlife. Measures to assess aquatic life use attainment in lentic waters include
those typically used for lake monitoring and this review did not uncover additional novel measures (Table
1).
N ' "
North Carolina designates an aquatic life propagation and survival use as applicable to lakes and
reservoirs. Measures included those typically used for lake monitoring and this review did not uncover
additional novel measures (Table 1).
South Carolina specifies a general aquatic life support designated use to lakes and reservoirs. Typical
water quality indicators are used to evaluate lake health for compliance with designated uses. Core
indicators used to evaluate aquatic life use in lakes include temperature, dissolved oxygen, pH, turbidity,
ammonia nitrogen, metals, chlorophyll a, total nitrogen (nitrate/nitrite, total Kjeldahl nitrogen), and total
phosphorus, Secchi depth (SCDHEC, 2013, Table 1). It does not appear that any biological indicators are
used to evaluate the condition of lakes for regulatory purposes.
3Yearly survey data available since at least 2004 at http://fw.ky.gov/navigation.aspx?cid=881&navpath=C552
4Lake summary results available at http://fw.ky.gov/navigation.aspx?cid=582&navpath=C552C579
-------
Methodologies for Development of Part 1: Nutrient Sensitive Aquatic Life Uses
Numeric Nutrient Criteria for Freshwaters
While not used for compliance with aquatic life uses, the South Carolina Department of Natural
Resources (SC DNR) conducts fisheries research to monitor biological aspects of lakes and streams of the
state (SC DNR 2011). The SC DNR manages and evaluates sport fisheries and has also undertaken
specific projects to examine forage requirements for brown trout in Lake Jocassee, water quality in Lake
Greenwood, and striped bass movements in the Santee-Cooper drainage (SC DNR, 2013).
Tennessee applies a fish and aquatic life use to lakes for the protection of aquatic life. Typical water
quality measurements are used to evaluate whether uses are supported (TDEC 2012).
Using state Consolidated Assessment and Listing Methodology (CALM) and Water Quality Standards
(WQS), aquatic life uses (ALUs) in lakes and reservoirs were investigated. At this time, according to the
literature available for review, all of the states that were evaluated use traditional trophic state
measurements of chlorophyll a (Chl-a) and Secchi depth along with other traditional water quality criteria
(e.g., dissolved oxygen (DO), temperature, pH, metals, etc.) to measure and assess aquatic life attainment
in lakes and reservoirs (Table 1). None of the states use direct biological measures for assessing aquatic
life uses in lakes/reservoirs, but some are developing them.
Some noteworthy observations include:
1. Florida is the only state in Region 4 using metrics other than the traditional lake water chemistry
measurements. These include the Lake Vegetation Index (LVI) which considers:
a. % native taxa
b. % invasive/exotic taxa as determined by FL EPPC (Florida Exotic Pest Plant Council)
c. Coefficient of Conservatism (C of C) of Dominant or Co-dominant taxa
d. % sensitive taxa
2. Wisconsin is developing metrics for an Aquatic Macrophyte Community Index (AMCI) for use in
lakes. While WDNR is not using plant metrics as primary impairment indicators for the 2012
assessment cycle, it is expecting to develop guidance that directs biologists to incorporate plant
data as supporting information for listing decisions.
3. Iowa uses a "shallow lakes" category that includes chlorophyll a (using the TSI), submerged
aquatic vegetation (SAV), and total suspended solids (TSS). Information on aquatic macrophytes
and macroinvertebrates will be used to supplement the lake water quality assessments. IDNR also
collects fish data, but there is no fish index being used to measure aquatic life use.
4. Illinois has an Aquatic Life Use Index (ALI) which it uses as the primary tool for assessing
aquatic life use in inland lakes, except Lake Michigan. A higher ALI score indicates increased
impairment. The index provides scores to measure degree of support (good/fair/poor). Metrics
include:
a. TSI
b. % surface area macrophytes coverage during peak growing season (June through August)
c. Median concentration of non-volatile solids (NVSS)
5. Minnesota - In 2009, the state collaborated with US EPA during the National Lake Assessment.
Minnesota used the data from the national survey work to start a fish Index of Biotic Integrity
(IBI) for lakes, but the tool has not yet been developed to the point where scores can be
categorized to measure the health of a system (good/fair/poor). Aquatic vegetation, which was
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
also surveyed as part of the 2009 study, was closely correlated with the fish IBI, but additional
research is needed to determine which metrics work best in lakes larger than 2000 acres.
Table 1 Summary of state lake and reservoir aquatic life use information.
State
Document
Section
Page(s)
Indicators
Florida Development of
Aquatic Life Use
Support Attainment
Thresholds for
Florida's SCI and
LVI(FLDEP2011b)
24-25
Traditional
Transparency
TN and TP
Chl-a
LVI
Comments
Alabama
Alabama Department
Of Environmental
Management
Water Division -
Water Quality
Program
Volume I
Division 33S-65
335-6-10-.il: Water 306-309;
Quality Criteria 314-322
Applicable to
Specific Lakes
• Traditional6
• TSI
• Chl-a
LVI developed
using biological
condition gradient
approach with a
human disturbance
gradient.
Georgia 391-3-6-.03 Water
Use Classifications
and Water Quality
Standards7
Specific Criteria for
Classified Water
Usage: Fishing:
Propagation of Fish,
Shellfish, Game and
Other Aquatic Life
14-15:46-51
Traditional
Additional criteria
applied to specific
lakes:
• Chl-a, TN, TP
5 http://www.adem.state.al.us/alEnviroRegLaws/files/Division6Voll.pdf (Accessed June 2013)
6 Dissolved oxygen, pH, metals, and other 304(a) aquatic life use water quality criteria.
7 http://www.gaepd.org/FilesJW/techguide/wpb/WQS/EPA_Approved_WQS_March2012.pdf (Accessed June
2013)
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
State
Illinois
Iowa
Kentucky
Minnesota
Document Section Page(s)
ILLINOIS Aquatic Life - 38-47
INTEGRATED Inland Lakes
WATER QUALITY
REPORT
AND SECTION
303(d) LIST, 2012
Clean Water Act
Sections 303(d),
305(b)and314
Water Resource
Assessment
Information
and List of Impaired
Waters
Volume I: Surface
Water (IEPA 2012)
Methodology for Lakes and Shallow 75 & 122
Iowa's 2012 Water Lakes (attachment
Quality Assessment, 4)
Listing, and
Reporting Pursuant
to
Sections 305(b) and
303(d) of the Federal
Clean Water Act
(Draft)(IDNR2013b)
401 KAR 10:031. Section 4 - Aquatic 2-4
Surface Water Life
Standards8
Guidance Manual for V. Protection of 15-35
Assessing the Quality Aquatic Life
of Minnesota Surface VII. Pollutants w/
Waters for Wildlife-Based
Determination of WQS
Impairment: 305(b) VIII. Protection of
Report and 303(d) Aquatic Recreation
List (MN PCA 2013)
Minnesota National 18
Lakes Assessment
Project:
Fish-based Index of
Biotic Integrity (IBI)
for Minnesota Lakes
(MN PCA 2009)
Indicators
Aquatic Life Use Index
• TSI
• % surface area
macrophyte
coverage during
peak growing
season (June - Aug)
• Median
concentration of
nonvolatile solids
(NVSS)
Measure:
• Secchi depth
• TP
• Chl-a
• NVSS
• % surface area
macrophyte
coverage
• Traditional
• TSS
• TSLChl-aand
Secchi depth
• Traditional
• NandP
• TSI:Chl-a
• Traditional
Lake Eutrophication
• TP
• Chl-a
• Secchi depth
• Fish IBI
Comments
Can use index to
assess degree of use
support
(good/fair/poor)
Most rivers,
streams, lakes, &
flood control
reservoirs are
grouped together,
but there is a
separate "shallow
lakes" category.
For the most part,
lakes are not
separated out from
streams. The
exception for lakes
and/or reservoirs
that support trout.
Sampling design
and assessment may
be different for
certain reservoirs or
other special
situations (e.g.,
lakes w/ distinct
bays) than what is
usually utilized in
lakes.
As of 2009, the fish
IBI had not been
developed enough
to use in WQ
assessments. MPCA
was working on
additional metrics
for lakes >2000
acres, as well as
correlating aquatic
! http://www.lrc.ky.gov/kar/401/010/031 .htm (Accessed June 2013)
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
State
Document
Section Page(s)
Indicators
Comments
vegetation scores to
the fish IBI.
Mississippi
Missouri
North
Carolina
South
Carolina
South
Dakota
Tennessee
State of Mississippi
Water Quality
Assessment
2012 Section 305 (b)
Report (MDEQ 2012)
Methodology for the
Development of the
2012 Section 303(d)
List in Missouri
(MDNR2010)
2014
303(d) Listing
Methodology (NC
DENR2013)
State Of South
Carolina
Monitoring Strategy
For Calendar Year
2013 (SCDHEC
2913)
The 2010 South
Dakota
Integrated Report
For
Surface Water
Quality
Assessment
(SDDENR2010)
2012 305(b) Report
The Status of Water
Quality in Tennessee
(TDEC 2012)
Lakes: Statewide 23-28
Assessment
Summary
How Water Quality 15-21
Data is Evaluated to
Determine Whether
or Not Waters are
Impaired
for 303(d) Listing
Purposes:
Physical, Chemical,
Biological and
Toxicity Data
5-10
Core and 11-12
Supplemental Water
Quality Indicators
(Measurements)
III. Surface Water 27-33
Quality Assessment
Lake Listing
Criteria
E. Water Quality 27-30
Assessment
Methodology (1)
Application
Methodology for
Specific Criteria
• Traditional
• Secchi depth
• TP
• Chl-a
• Traditional
Nutrients in Lakes:
• TN and TP
• Chl-a
• Traditional
Lakes:
• DO
• Chl-a
• Turbidity
• Traditional
Core (lakes):
• TN and TP
• Chl-a
• Secchi depth
(supplemental)
• Traditional
• Secchi depth
• TN and TP
• Chl-a
• Traditional
No lakes being
impaired based on
this. Simply
describing trophic
state
There is a Lake
Assessment section,
but these
Application
Methods aren't
specific to lakes.
They seem more for
all waterbodies.
Wisconsin
Wisconsin 2014
Consolidated
Assessment and
Listing Methodology
(WisCALM) for
Clean Water Act
Section 305(b), 314,
and 303(d)
Integrated Reporting
(WI DNR 2013)
Lake Impairment
Assessment: Fish &
Aquatic Life (FAL)
Uses
27-30
Traditional
TP
Chl-a
Aquatic Macrophyte
Community Index
(AMCI)
10
-------
Methodologies for Development of Part 1: Nutrient Sensitive Aquatic Life Uses
Numeric Nutrient Criteria for Freshwaters
Even though the current landscape of aquatic life use tools is focused principally on trophic state
endpoints (algal biomass estimate as chlorophyll a and Secchi depth) as well as dissolved oxygen, the
potential exists for biological assemblage data to be used to assess aquatic life use condition, as is
currently done in streams (USEPA 1998). Such assemblages include algae, invertebrates, and vertebrates.
While few states have developed assemblage based aquatic life use measures, federal programs have.
The USEPA National Aquatic Resource Survey National Lakes Assessment, for example, collected
phytoplankton, sediment diatom, macrophyte, zooplankton, and littoral macroinvertebrate information
from lakes across the nation as part of their assessment (USEPA 2009). The results of that program
suggest that assemblage based information is a viable and valuable component of ecological
characterization beyond that provided by trophic state measures alone. This effort builds off that
experience.
The approach developed for this analysis focused on identifying datasets where the broadest range of
aquatic life, including fish was available to explore for nutrient sensitivity. Requiring a broad range of
biological assemblage information along with water chemistry information for lakes greatly restricted the
landscape of available data. Fortunately, the National Science Foundation Long Term Ecological
Research Program (NSF LTER) has been funding long-term collection of this type of information at the
Northern Temperate Lakes (NTL) LTER site in Wisconsin for several decades. The core monitoring
program along with associated research projects that have built off this program provided an invaluable
resource for this application. A comparable program for Southeastern US reservoirs could not be found.
However, the Tennessee Valley Authority (TVA) has had a long commitment to biological assemblage
collection for reservoirs in this region and had a large fish and nutrient dataset for use. Without these two
programs, especially the NSF LTER NTL, the project would not have been possible or would have been
greatly diminished. These datasets are described in detail following this data analysis description.
Having identified appropriate data, the analytical approach chosen was to use a taxon based weighted
averaging or nutrient optima approach (Yuan 2006). This approach used taxon specific relative
abundance information along nutrient gradients to identify differences in taxa preferences or optima.
Specific optima across taxa as well as the range of these optima within different assemblages were then
used to infer the sensitivity of different aquatic life to nutrient gradients. This information will be used in
subsequent work to inform the application of biological responses for identifying protective nutrient
criteria. The specific analyses used are described below in more detail.
A comprehensive lakes data set was acquired from the North Temperate Lakes Long Term Ecological
Research Program (NTL LTER, lter.limnology.wisc.edu, downloaded July 2013). The data set consisted
of 150 data files representing both data collected as part of the baseline monitoring for the NTL LTER
and data collected for 16 additional projects that were also available from the NTL LTER program
website (Table 2).
11
-------
Numeric Nutrient Criteria for Freshwaters
Methodologies for Development of
Part 1: Nutrient Sensitive Aquatic Life Uses
Table 2 North Temperate Lakes Long Term Ecological Research program data reviewed for use in this analysis. Shaded cells indicate
data selected for use.
Project
Biocomplexity at North Temperate
Lakes Long-Term Ecological Research
Cascade Project
Cross Lake Comparison
Crustacean Zooplankton Species
Richness in 66 North American Lakes
EPA Eastern Lake Survey, Upper
Midwest Region
EPA Environmental Research Lab
Duluth Lake Survey
Fluxes project at North Temperate
Lakes LTER: Random Lake Survey
Lake Mendota Phosphorus Entrainment
2005
Lake Metabolism
Lake Wingra
Landscape Position Project
Little Rock Lake
Madison Lakes Zooplankton
# Lakes
61
8
19
11
254
435
7,588
1
31
1
32
1
2
Water
^ 1-4. T^ * Nutrient Data Fish Data Invertebrate Data
Quality Data
200 1 - 2004 200 1 - 2004 200 1 - 2004
1984-2007 1991-1999 1984-2003
2006 2006 20032
19923
1984 1984
1979-1982 1979-1982
2004 2004
2005 2005
2000 2000
1996-2013 1996-2013
1998-2000 1998-2000 1998-1999 1998-1999
1983-2000 1996-2000 2002-20042
Phytoplankton Data Zooplankton Data
2001 -2004 '
1984-1995 1984-2007
2006
1992
1983-2007
1976-1994
North Temperate Lakes LTER (Base
Program)
North Temperate Lakes LTER (High
Frequency Data)
1981-2012
1989-2012
1981-2012
1981-2012
1981-2004
1984-2012
1981-2012
12
-------
Numeric Nutrient Criteria for Freshwaters
Methodologies for Development of
Part 1: Nutrient Sensitive Aquatic Life Uses
Project
# Lakes
Water
Quality Data
Nutrient Data Fish Data
Invertebrate Data
Phytoplankton Data Zooplankton Data
Primary Production and Species
Richness in Lake Communities4
Wisconsin Historical Lakes Data
Yahara Lakes Fisheries
13,093
4
1925-2009
1925-2009
1997-20005
1987-1998°
1997-2000
1997-2000
Presence/Absence data only
Paired with Biocomplexity water quality and nutrient data
Conductivity data only
Richness data only; taxonomic abundance data are not available.
Paired with NIL LTER (base program) water quality and nutrient data
Paired with Crustacean Zooplankton Species Richness water quality and nutrient data
13
-------
Methodologies for Development of Part 1: Nutrient Sensitive Aquatic Life Uses
Numeric Nutrient Criteria for Freshwaters
To prepare the most complete data sets possible for the analysis, lake identifying features (name or other
identifier, geographic coordinates, lake size, and county) were extracted from each file containing that
information. Source filenames were maintained along with the lake information for tracking purposes.
Lake information was compared in order to identify lakes that were the same across projects. In many
cases, it was not possible to determine if similarly named lakes were, in fact, the same lake. Geographic
coordinates were not available for lakes in the Cascade, Cross-Lake Comparison, Lake Metabolism, or
Wisconsin Lake Plants Historical Data projects, nor were they available for approximately 10% of the
lakes in the Wisconsin Historical Lakes Data.
Next, water quality and nutrient data across all projects for which they were available were combined.
Different projects contained data for one or more of the following parameters (parenthetical is number of
studies with relevant data):
• Conductivity (7/18) • Total nitrogen (TN uf)(7/l 8)
• Dissolved oxygen (8/18) • Total dissolved nitrogen (TN f)(4/18)
• pH (10/18) • Dissolved inorganic phosphate (PO4
• Water temperature (9/18) f)(6/18)
„. , , . , /™v-\/-in/io\ * Total phosphorus (TPuf)( 12/18)
• Dissolved organic carbon (DOC)( 10/18) r r
~ . , • , /Tr\mno\ * Total dissolved phosphorus (TPf)(5/18)
• Total organic carbon (TOC)(6/18) r r
rv i j • r\Tu\/r/io\ • Secchi depth (9/18)
• Dissolved ammonia (NH3)(6/18) F
TV i A •«,* i •.-.nvrn-L. ' Chlorophyll a (4/18)
• Dissolved nitrate plus nitrite (NO2 +
NO3)(4/18) • Phaeophytin(4/18)
• Total Kjeldahl nitrogen (TKN)(4/18)
Prior to combining the data, all collection, laboratory analysis, and reporting methods and units were
reviewed to ensure that only comparable data were combined. All nitrogen analytes were reported on an
"as Nitrogen" basis, and all phosphorus analytes were reported on an "as Phosphorus" basis. Due to lack
of sufficient data, total nitrogen or total phosphorus was not calculated for samples lacking those analytes.
Where necessary, data were converted to ensure that reported units were identical.
Water quality and nutrient data were reviewed to identify flagged data, outliers, and likely erroneous
values (negative or zero values considered to be erroneous). All data that were flagged for quality reasons
were removed from the data set as were negative or zero values for analytes having a method detection
limit. Dissolved oxygen values greater than 25 mg/L and water temperature values greater than 35 °C
were also removed.
All projects contained chlorophyll data that were corrected for phaeophytin, but projects variously applied
fluorometric methods, spectrophotometric methods, or both. Spectrophotometric data were selected over
fluorometric data when both were available, since those data were more prevalent across the different
projects. Likewise, spectrophotometric data for phaeophytin were chosen over fluorometric data when
both were available.
Nutrient data (DOC, TOC, NH3, NO2 + NO3, TKN, TN, PO4, and TP) were Iog10-transformed prior to
calculating site averages. Water quality data, Secchi depth, chlorophyll, and phaeophytin were not
transformed. Site-specific water quality and nutrient data were then averaged over replicates (samples
obtained on the same day from the same location using the same methods that represent either field
duplicates or laboratory duplicates). After averaging replicate data, site-specific averages over depths and
time were calculated as follows:
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
• Water data to be combined with benthic macroinvertebrate data were first averaged over all
sampled depths (since samples were in the littoral), then annually.
• Water data to be combined with fish data were first averaged over sampled depths ranging
between 0 and 2 m (to estimate photic zone conditions), then annually.
• Water data to be combined with zooplankton data were first averaged over sampled depths
ranging between 0 and 2 m (to estimate photic zone conditions), then (separately) annually and
seasonally.
• Water data to be combined with phytoplankton data were not averaged over depths or time.
Biological data were similarly combined. Fish data were segregated by method: separate data files of
combined project data were prepared for electrofishing, gill net, and seine data. Similarly, benthic
macroinvertebrate data were segregated by sampling method: separate data files of combined project data
were prepared for Hester-Dendy (passive sampler), coarse woody habitat, and SCUBA vacuum data. All
zooplankton data were collected using the same methods, as were phytoplankton data. Therefore, all
project data were combined for these two data categories into one file for each category.
Lastly, water quality and nutrient data were merged with biological data. Specifically, for all but the
phytoplankton data, water data for each lake/site were joined with biological data for that lake/site where
sampling occurred during the same time frame as the water data averaging (e.g., same year, same season)
and for the same project. When a biological sample did not match a water sample from the same project,
an attempt was made to identify appropriate samples from a different project. A water sample was
considered an appropriate match if it was sampled in the same time frame as the biological sample, or in
the case of benthic macroinvertebrates and fish, if the water sample was obtained during the preceding 2
years (using the closest year's data). Temporally shifted data for 1 lake that was sampled for fish (West
Long Lake, Cascade project, fish data from 2001 and 2002, water data from 2000) and for 8 lakes
sampled for benthic macroinvertebrates were used (Table 3).
Table 3 Asynchronous benthic macroinvertebrate and water chemistry data.
Lake Name
Pallete Lake
Vandercook Lake
Jute Lake
Found Lake
Towanda Lake
Camp Lake
Moon Lake
Little Rock Lake
Invertebrate Data Source
Cross-Lake Comparison Project
Cross-Lake Comparison Project
Cross-Lake Comparison Project
Cross-Lake Comparison Project
Cross-Lake Comparison Project
Cross-Lake Comparison Project
Cross-Lake Comparison Project
Little Rock Lake Experiment
Invertebrate
Data Year
2003
2003
2003
2003
2003
2003
2003
2004
Water Data Source
Biocomplexity Project
Biocomplexity Project
Biocomplexity Project
Biocomplexity Project
Biocomplexity Project
Biocomplexity Project
Biocomplexity Project
Biocomplexity Project
Water Data
Year
2001
2002
2002
2002
2002
2002
2002
2003
Phytoplankton data were joined to water data collected within 30 days (before or after) of the
phytoplankton data collection, as part of the same project.
15
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
1.3.1.2 Tennessee Valley Authority Long-Term Monitoring Data
A second data set was acquired from the Tennessee Valley Authority (TVA, Table 4). Water quality and
nutrient data collected prior to 2000 from EPA's Legacy STORET database were downloaded
(www.epa.gov/storpubl/legacy/gateway.htm. downloaded September 30 - October 2, 2013). TVA
personnel provided water quality and nutrient data collected since 2000 (TVA, personal communications
with Tyler Baker, September - November 2013) and with fish community data. TVA has been
monitoring water quality since 1960, and has collected fish community data since 1993.
Table 4 Tennessee Valley Authority program data used in this analysis.
Data Source
# Lakes Water Quality Data Nutrient Data Fish Data
EPA's Legacy STORET Database 32 1960-1997 1961-1997
TVA Water Chemistry Data 31 2000-2006 2000-2006
TVA Fish Community Data 31 1993-2012
The TVA data sets included the following water quality and nutrient parameters:
Conductivity
Dissolved oxygen
pH
Water temperature
Turbidity
Total dissolved solids (TDS)
Total suspended solids (TSS)
Total organic carbon (TOC)
Ammonium, unfiltered (NH4)
Nitrite, unfiltered (NO2)
Nitrate plus nitrite, unfiltered (NO2 +
NO3)
Nitrate, unfiltered (NO3)
Organic nitrogen, filtered (DON)
Organic nitrogen, unfiltered (TON)
Total kjeldahl nitrogen (TKN)
Total nitrogen, unfiltered (TN uf)
Phosphate, filtered (PO4 f)
Total organic phosphorus, unfiltered
(TOP)
Total phosphorus, unfiltered (TP uf)
Total phosphorus, filtered (TP uf)
Secchi depth
Chlorophyll a
Phaeophytin
TVA data were prepared following the quality control requirements as for the NTL LTER data. Unlike
the NTL LTER data, the TVA data reported Method Detection Limits (MDLs). For each reported value of
less than the MDL for any analyte, final result value was set equal to one-half the value of the MDL. Site
replicates were next averaged, and site-specific average calculated for samples obtained between 0 and
2m deep. Because TVA collected fish community data multiple times between August and November at
each site, both monthly and seasonal averages were prepared for the water quality and nutrient data.
Young-of-year fish were excluded from the fish community data, because they typically are under
sampled and are not representative of the fish community. Next, catch per unit effort (CPUE) was
16
-------
Methodologies for Development of Part 1: Nutrient Sensitive Aquatic Life Uses
Numeric Nutrient Criteria for Freshwaters
calculated from the abundance and effort data. Lastly, the fish community data were separated by
sampling method (electrofishing or gill net). Fish data were then joined with the monthly and seasonal
water quality and nutrient data to create the final data sets used in the analysis.
An indicator value approach was used for identifying sensitive aquatic life use that is based on the taxon
optima or preference approach also applied for deriving tolerance values (Yuan 2006). Three statistical
approaches for developing indicator values of biological assemblages to various environmental stressors
include: (1) central tendency, (2) environmental limits, and (3) optima (Yuan 2006). Tolerance values
expressed in terms of central tendencies attempt to describe the average environmental conditions under
which a taxon is likely to occur; indicator values expressed in terms of environmental limits attempt to
capture the maximum or the minimum level of an environmental variable under which a taxon can persist;
and indicator values expressed in terms of optima define the environmental conditions that are most
preferred by a given taxon. These three types of indicator values are expressed in terms of locations on a
continuous numerical scale that represents the environmental stressor gradient of interest. Both abundance
based and presence/absence based models could be built using these three statistical approaches.
A variety of approaches were tested to characterize the taxon-environmental relationship (Figure 1).
Weighted averaging (WA) was used to estimate the central tendency of taxa along nutrients gradients; it
computes the mean of the product of species abundance and nutrient concentrations. Both abundance and
present/absence data can be used and, indeed, had to be used given the constraints of the data. Such WA
optima values are often referred as tolerance values. This is one of the more commonly used approaches
for characterizing species distributions along environmental gradients (Yuan 2006).
When using weighted averages, a normal distribution across the environmental gradient is assumed. The
width of the bell shape is often called tolerance which can also be used to characterize the environmental
niche for taxa along the environmental gradient. This statistical tolerance is also used as an indicator
value. Such environmental limits can be estimated by computing cumulative percentiles (CPs) from field
data. The 95th percentile range which a taxon can tolerate (tolerance value) was calculated. When using
CPs, problems arise due to non-uniform sample distributions. This problem can be solved by weighting
samples within equal width bins and then using the binned data to compute the CPs. These are referred to
as weighted cumulative distribution functions (weighted cdf, e.g., Yuan 2006).
Regression estimates of taxon-environment relationships can also be estimated using linear (LRM),
quadratic (QLRM) logistic regression models, or generalized additive models (GAM) to model nutrient-
response relationships. It is commonly done with presence/absence data using a binomial distribution.
After models are established, the 95th percentile cumulative probability (area under the curve of models)
can be estimated as the environmental limits a taxon can tolerate. Note that central tendency can also be
estimated from CPs and regression models using either the median value of the cumulative distribution
function or the median cumulative probability of the regression models.
In this analysis, indicator values were developed from the above approaches and can be considered as
either optima (central tendency, WA or 50th percentile) or tolerance (limits, 95th percentile). While this
project focused on the optima, the tolerances are informative and considered. All these methods have
limitations, but indicator values developed from these statistical methods were generally correlated or
similar to each other, as reported elsewhere (Yuan 2006). Variations due to statistical approaches were
minimized by averaging or selecting the most consistent results.
17
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
Caenis
-f S
!o
oj
.Q
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
Table 5 Summary statistics of water chemistry in the phytoplankton dataset.
Variables
Depth (m)
DO(mg/L)
pH
WTemp_C
DOC (mg/L)
TOC (mg/L)
NH3-N (mg/L)
NO2+3-N (mg/L)
TN
PO4-P (mg/L)
TP_uf(mg/L)
SecchiDepth (m)
Chi a_(ug/L)
Number
of
Samples
508
432
56
445
47
48
41
28
46
43
65
450
15
Min
0
0.1
5.25
0.1
1.58
1.73
0.014
0.01
0.032
0.007
0.005
0.6
0.1
1st
quartile
5
2.65
7.7675
8.9
4.98
4.95
0.172
0.227
0.28
0.064
0.078
2
1.3
median
10
8.7
8.2
12.1
5.74
5.62
0.374
0.296
0.62
0.105
0.111
3
2.1
3rd
quartile
16
10.9
8.56
15.7
6.20
6.11
0.554
0.406
1.24
0.191
0.191
5.2
5.15
Max
25
19.6
9.36
27.1
38.23
38.91
2.14
1.29
2.86
0.68
0.678
13.4
13.8
Though hundreds of phytoplankton samples had been collected through time, only very few water
chemistry samples were also taken in the 30 day windows (this was not improved even when the window
was expanded to 60 days or larger)(Table 5). Only water temperature, Secchi depth, and DO were
measured with most of the samples. Other nutrient variables were found in a limited number of samples.
Despite the limited number of samples of environmental variables, modeling was still performed to
develop indicators for algal taxa at species, genus, and division levels (Figure 2, Table 6, Appendices 1-
13 and 13-15). As seen from Table 6, planktonic diatoms (Bacillariophyta) preferred high TP
concentrations, while Chrysophytes in general dominated low TP environments. Many blue green algal
genera, e.g.,Ancibaena and Aphanizomenon, dominated when TP concentrations were high, while other
genera, e.g., Synechocystis, Schroederia, and Cryptomonas, were less nutrient dependent and were often
found in low nutrient environments. DO was also modeled but the results were less clear. Many blue
green algae exhibited optima at lower DO.
Table 6 Taxa indicator value rankings (family and higher on top, genus and lower on bottom) for
phytoplankton taxa. Indicator values for TP, TN, and Secchi depth were rank scaled with 2
indicating nutrient sensitive taxa (low nutrient optima) and 5 indicating nutrient tolerant taxa (high
nutrient optima). Also shown are the sample sizes (N) for each estimate.
Taxon
Bacillariophyta
Chlorophyta
Chrysophyta
Cryptophyta
Cyanophyta
Anabaena
Level
Division
Division
Division
Division
Division
Genus
TP_N
51
63
36
64
65
36
TPJV
5
3
2
3
3
5
TN_N
32
45
25
46
46
28
TNJV
3
3
2
3
3
5
Seechi N
357
442
282
442
445
214
SecchiJV
o
5
o
5
3
o
J
4
5
19
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
Aphanizomenon
Aphanocapsa
Aphanothece
Chlamydomonas
Coelosphaerium
Cryptomonas
Erkenia
Microcystis
Monoraphidium
Oocystis
Pseudanabaena
Rhodomonas
Schroederia
Synechococcus
Synechocystis
Aphanizomenon flos-aquae
Aphanocapsa delicatissima
Aphanothece nidulans
Coelosphaerium naegelianum
Cryptomonas erosa
Erkenia subaequiciliata
Microcystis aeruginosa
Pseudanabaena limnetica
Pseudanabaena mucicola
Rhodomonas minuta
Schroederia judayi
Genus
Genus
Genus
Genus
Genus
Genus
Genus
Genus
Genus
Genus
Genus
Genus
Genus
Genus
Genus
Species
Species
Species
Species
Species
Species
Species
Species
Species
Species
Species
56
39
36
36
37
50
25
47
25
25
46
64
47
50
30
56
38
34
37
46
25
47
37
32
64
47
5
5
2
2
4
2
5
4
5
2
4
3
2
3
2
5
5
2
4
2
5
4
3
4
3
2
40
27
28
28
27
35
18
34
17
19
33
46
33
42
27
40
27
27
27
34
18
34
24
26
46
33
4
4
5
2
4
2
5
5
5
5
5
3
2
2
3
4
4
5
4
2
5
5
2
5
3
2
360
220
256
243
198
384
213
284
157
147
285
432
282
308
232
358
192
207
198
368
213
283
220
171
432
280
4
5
4
o
J
5
o
3
3
5
5
4
5
3
-\
3
4
2
4
5
4
5
3
o
J
5
5
5
o
J
o
J
20
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
Aphantzomenon flos-aquae
Schroedena
TO
.Q <0
O b
Q.
0) •».
1°
(0 rv»
O <=•
•^
d
0
re
8
*- <1>
o Q:
-
d
o 0)
5 £
. = 1
Q 0
O >
^
3 ^55
o OL
0.01 0.03 0.1 0.3 °01 °03 01 °-3
Total Phosphorus (mg/L) Total Phosphorus (mg/L)
Figure 2 Examples of algal species response (capture probability) along the TP gradient, showing
TP preferred species Aphanizomenon flos-aquae and less TP dependent genus Schroederia. For
all taxon response curves for phytoplankton, see Appendices 1-3 and 13-15.
Inference models were created using the taxon abundance data and resulting optima (Figure 3).
Surprisingly, inferred TP and TN models using phytoplankton had low accuracy and precision (r2<0.30);
lower than Secchi depth inference models (r2=0.43). The phytoplankton is usually a good predictor of TP
and TN concentration and is frequently used for such applications (e.g., Dixit et al. 1999). It is unclear
why these data did not perform better with this dataset, but the hypothesis is it was likely due to the
limited data and coarse available taxonomy. Further evaluation would be needed to clarify the issues.
21
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
3 -
2 -
0.0 -
-o.e -
-O.B -
—T~
1C
Measured Seech i Depth (m)
-1.5 -1.0 -0.5
Measured TN
0.0
= 0.43
—I—
12
R*= 0.14
O.E
-0.6 -
-o.s -
-1.0 -
-1.4 -
-2.C
-1.5 -1.0
Measured TP
R2= 0.29
-0.5
Figure 3 Performance of phytoplankton WA models with nutrient variables (LogTN and LogTP in
mg/L) and Secchi depth (m). Phytoplankton inferred environmental variable values are the sum
product of relative abundance and the optima at a sample/site.
1.3.2.1.2 Zooplankton dataset
As detailed above, zooplankton data (density [number of individuals/L]) were downloaded from the North
Temperate Lakes Long Term Ecological Research (NTL LTER) website, which contained data generated
by the NTL LTER program, as well as data developed in association with other state level research
22
-------
Methodologies for Development of Part 1: Nutrient Sensitive Aquatic Life Uses
Numeric Nutrient Criteria for Freshwaters
efforts. Downloaded data originated from four different projects: NIL LTER, Cascade Project, Cross-
Lake Comparison, and Little Rock Lake Experiment. Zooplankton biomass data were also downloaded;
however, these data were only available for Cascade and Cross-lake projects, which were a minor
component of the total data set so this measure of zooplankton was not investigated. Zooplankton data
were available for spring and summer from stations located throughout each lake, however specific
locations of these sites were not available. Samples were identified generally to the genus and species
levels and for analyses these data were consolidated to the family level.
Nutrient and nutrient-related data downloaded and investigated in the analyses included NH3, NO23,
Total Nitrogen (unfiltered), Total Phosphorus (unfiltered), Secchi depth, and Chlorophyll-a. After
assessing correlations with zooplankton taxa, NH3 and NO23 were removed from the analysis because
there were not enough data points associated with zooplankton data to warrant confidence in the
relationships that were observed and also because TN and TP are the general focus of nutrient criteria
development. Descriptive statistics for the water chemistry data are presented (Table 7).
For all four project datasets zooplankton were collected from the deepest locations on each lake via
vertical tows with a Wisconsin Net (80um mesh), and a series of Schindler Patalas (53um mesh) samples
spanning the water column. The NTL LTER data were collected over approximately 30 years (1981-
present) from seven lakes in north central Wisconsin, throughout the NTL study region (Table 8). These
zooplankton samples were collected at night but employed the same gear and processing techniques used
in the other studies. Two of the NTL LTER lakes (Crystal Lake and Sparkling Lake) were manipulated:
Crystal Lake had thermal removal of rainbow smelt in 2011 and Sparkling Lake had thermal removal of
rainbow smelt, rusty crayfish trapping and removal, and walleye stocking, all in 2001-2002. The Cross-
Lake Comparison project was conducted in 2006 and collected zooplankton data, DO, temperature and
secchi depth but no nutrient or nutrient-related data other than Secchi depth; therefore, these data were of
minimal use in investigating relationships of zooplankton to nutrient levels. There was spring-collected
nutrient data available for Cross-Lake comparison sites and these were investigated but ultimately not
used in the final analysis, which was based on summer data. The data in the Cascade Project were from
six experimental lakes (two of which served as reference sites) also located in the NTL study region.
Documentation of Cascade project sampling indicated that lake manipulation was conducted throughout
the 1990s, however, there was little documentation regarding specific experiments done on these lakes.
Total Nitrogen was not available for Cascade Project lakes. Minimal documentation along with
knowledge that these were manipulated lakes suggested that using these data to assess
zooplankton/nutrient relationships should be done with caution. The Little Rock Lake Experiment project
data, which were from Little Rock Lake exclusively, was an acidification study until 1990 when
acidification experiments were ended and the lake was allowed to recover naturally. There was also a
coarse woody debris removal conducted in 2002 on Little Rock Lake.
Because of the potential for ecological patterns to be distorted by lake manipulations, data from
manipulated lakes were investigated to assess whether lake manipulations appeared to affect the ability to
accurately assess specific nutrient/biota relationships. Data plots for individual manipulated lakes were
generated in which the x axis was the sample year and y axis was individual zooplankton taxon. Many of
the changes in taxa abundances observed through time occurred across sites, manipulated and un-
manipulated alike (i.e., changes that occurred at certain points in time at manipulated lakes also occurred
in un-manipulated lakes). The relationship of these taxa abundance changes to nutrient levels was then
investigated and there was no correlation. These observations served as evidence toward using these data
from manipulated lakes in the final analysis. To further assess the influence of manipulated lake data on
broad-scale relationships, plots were generated in which lakes from the experimental projects, as well as
specific manipulated lakes, were removed. The general trends in taxa relationships to nutrients persisted
throughout this process. The analyses on manipulated lake data showed that the trends for the dataset
23
-------
Methodologies for Development of Part 1: Nutrient Sensitive Aquatic Life Uses
Numeric Nutrient Criteria for Freshwaters
were consistent in spite of manipulation and suggested that the data were suitable for use in the final
analysis.
All combinations of data were investigated to assess whether relationships of zooplankton to nutrients
varied depending on dataset composition or season. Relationships of summer and spring zooplankton
data were investigated with summer, spring, and annual water quality data for datasets composed of all
project data and variations thereof with regard to exclusion of manipulated lake data and naturally distinct
data. The issue of natural variability and data usability applied primarily to two bogs in the NIL LTER
dataset (Crystal and Trout Bogs). These systems had some similarities in zooplankton communities to the
other lakes, however, pH levels and nutrient levels were substantially different (low pH~5, higher
nutrients). Furthermore, overall patterns in relationships for the entire dataset were substantially shifted
by the influence of the data points from these two lakes; therefore, these lakes were removed from the
final analysis due to the apparent distinct natural differences. After investigation of various combinations
of seasonal data, manipulated lake data, and natural variability the final dataset for development of optima
was chosen - it consisted of summer zooplankton and water quality samples and contained all site data
with the exception of the two acidic, nutrient rich bogs.
24
-------
Methodologies for Development of Part 1: Nutrient Sensitive Aquatic Life Uses
Numeric Nutrient Criteria for Freshwaters
Table 7 Descriptive statistics for the TP, TN, Secchi Depth, and Chlorophyll for data set (summer, all sites except the two bogs)
Nutrient/Indicator
TN (unfiltered)(mg/L)
TP (unfiltered)(mg/L)
Secchi Depth (m)
Chlorophyll-a (mg/L)
Valid
N
138
160
243
248
Mean
0.224
0.007
5.135
3.653
Mean
(loglO)
-0.650
-2.158
Median
0.220
0.006
4.971
2.468
median
(loglO)
-0.658
-2.196
Minimum
0.017
0.001
1.123
0.350
Minimum
(loglO)
-1.776
-3.000
Lower
Quartile
0.180
0.004
3.859
1.507
Lower
Quartile
(loglO)
-0.744
-2.349
Upper
Quartile
0.304
0.011
6.286
4.266
Upper
Quartile
(loglO)
-0.517
-1.945
Maximum
0.439
0.047
10.150
52.174
Maximum
(loglO )
-0.358
-1.327
25
-------
Methodologies for Development of Part 1: Nutrient Sensitive Aquatic Life Uses
Numeric Nutrient Criteria for Freshwaters
Table 8 Sample count by lake for optima calculations - each observation indicates that
zooplankton and nutrient/indicator data were available (e.g., for Allequash Lake there were 26
samples for which zooplankton and nutrient/indicator data were available from this lake).
Lake Name
Allequash Lake
Anvil Lake
Big Muskellunge Lake
Big Portage Lake
Camp Lake
Crab Lake
Crampton Lake
Crystal Lake
East Long Lake
Indian Lake
Lac du Lune
Lake Laura
Little Crawling Stone Lake
Little Rock Lake
Lynx Lake
Paul Lake
Peter Lake
Sparkling Lake
Star Lake
Stormy Lake
Trout Lake
Tuesday Lake
West Long Lake
Total
TN TP
(uf) (uf)
26 26
25 26
25 26
5
10 10
5
5
26 27
26 25
5
138 160
Secchi Depth
31
1
32
1
1
1
2
33
6
1
1
1
1
36
1
9
10
33
1
1
33
1
6
243
Chlorophyll-a
30
31
2
31
7
1
36
16
17
31
31
8
7
248
Individual species/genus and family level optima were developed (Figure 4) as described above and the
resulting indicator values presented for genera and family/higher taxonomic levels (Table 9, Appendices
4-7 and 16-19).
Because of the low number of sample lakes in the data set (Table 8), the range of values was a limiting
factor in distinguishing between nutrient tolerant and nutrient sensitive taxa. Nonetheless, there were taxa
that consistently exhibited trends that suggested nutrient sensitivity (i.e., low rankings, e.g., 2-3) and that
suggested nutrient tolerance (i.e., high rankings 4-5). Some taxa had substantial differences in TP, TN,
Secchi and Chi a rankings; TN rankings in particular were less closely related to the other rankings (Table
9). The low degree of correlation between TN and other nutrient/indicators may be due in part to
limitations in the dataset given the fact that the majority of samples were from five lakes (Table 8) and
such individual lake data could potentially influence the overall pattern. There were no TN samples from
any of the Cascade Project lakes and these had some of the higher nutrient levels (as represented by high
26
-------
Methodologies for Development of Part 1: Nutrient Sensitive Aquatic Life Uses
Numeric Nutrient Criteria for Freshwaters
TP, high Chl-a, and low Secchi) in the data set and were potentially influential on broad patterns.
Although breadth of the dataset may be a limiting factor in describing actual zooplankton/nutrient
relationships, it appears likely that the differences between TN optima and the rest of the optima may also
have been due in part to actual differences in response of taxa to TN and TP; rankings between TP and
Secchi and Chl-a were much more consistent than those for TN. This may be an indication that TP levels
in these lakes were a greater influence on general productivity as represented by zooplankton taxa
abundances, Secchi depth, and Chl-a levels. Average optima by family and genus level resolutions were
also shown for comparison (Table 10, Table 11).
27
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
Table 9 Taxa indicator value rankings (family followed by genus/species) for zooplankton taxa.
Indicator values for TP, TN, Chi a, and Secchi depth were ranked with 2 indicating nutrient
sensitive taxa (low nutrient optima) and 5 indicating nutrient tolerant taxa (high nutrient optima).
Also shown are the sample sizes (N) for each estimate.
Taxon
Asplanchnidae
Bosminidae
Brachionidae
Calanoid
Calanoida
Chydoridae
Collothecidae
Conochilidae
Cyclopidae
Cyclopoida
Daphniidae
Diaptomidae
Gastropodidae
Holopediidae
Lecanidae
Lepadellidae
Rotifera
Sididae
Synchaetidae
Temoridae
Testudinellidae
Trichocercidae
Trochosphaeridae
Acanthocyclops. vernalis
Alona
Ascomorpha. ecaudis
Ascomorpha. ovalis
Asplanchna
Bosmina. longirostris
Ceriodaphnia
Chydorus
Collotheca
Collotheca. mutabilis
Colonial, conochilus
Conochiloides
Conochiloides. dossuarius
TP_N
124
150
160
72
72
95
92
157
160
90
152
146
155
95
91
11
38
132
160
34
10
137
70
23
12
38
102
124
19
26
85
63
29
18
36
10
TP_IV
5
4
4
2
2
3
3
4
4
5
4
2
4
3
4
3
2
3
4
3
5
5
5
4
5
5
3
5
5
5
3
2
5
5
2
5
TNJM
104
130
138
70
70
76
92
136
138
88
132
135
135
85
78
10
38
117
138
27
10
116
51
23
NA
27
103
104
NA
24
76
63
29
NA
35
NA
TNJV
4
3
3
4
4
2
4
3
3
5
3
3
3
4
3
2
4
3
3
2
5
4
2
4
NA
3
4
4
NA
5
2
4
4
NA
2
NA
ChlaJM
182
225
235
92
92
132
117
231
248
124
240
216
228
167
137
14
52
201
235
65
15
206
101
32
23
59
129
182
33
37
114
71
46
32
64
30
ChlaJV
4
5
o
J
2
2
5
2
o
J
4
4
4
3
4
3
o
5
3
2
o
J
4
o
5
5
4
5
4
5
5
3
4
5
5
3
2
4
5
2
5
Seechi N
169
229
217
97
97
139
121
214
243
129
235
215
210
158
140
14
53
202
217
65
15
190
88
34
17
41
134
169
20
42
127
75
46
19
67
12
SecchiJV
5
4
4
2
2
4
4
4
4
5
4
3
4
2
4
5
4
o
J
4
2
5
4
5
5
5
2
4
5
5
5
3
2
5
5
o
5
5
28
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
Taxon
Conochilus
Cyclops, varicans. rubellus
Daphnia
Daphnia.dubia
Daphnia. longiremis
Daphnia. mendotae
Daphnia.parvula
Daphnia.pulex
Daphnia.pulicaria
Daphnia. retrocurva
Daphnia.rosea
Diacyclops. thomasi
Diaphanosoma
Diaphanosoma. birgei
Epischura. lacustris
Filinia
Filinia. longispina
Filinia. terminalis
Gastropus
Gastropus. hyptopus
Gastropus. sty lifer
Holopedium. gibberum
Kellicottia. bostoniensis
Kellicottia. longispina
Keratella. cochlearis
Keratella. crassa
Keratella. earlinae
Keratella. hiemalis
Keratella. quadrata
Keratella. taurocephala
Keratella. testudo
Lecane
Lecane.inermis
Leptodiaptomus. minutus
Leptodiaptomus. sicilis
Mesocy clops, edax
Monostyla
Monostyla. lunaris
Notholca.foliacea
Orthocyclops. modestus
Ploesoma
Ploesoma. lenticulare
TP_N
136
10
54
33
62
75
15
17
51
40
18
120
10
129
34
17
10
47
14
70
129
94
78
149
160
126
108
36
89
97
24
40
35
102
15
102
18
29
18
17
74
13
TP_IV
2
5
2
3
2
2
5
5
4
o
3
5
2
2
o
5
3
5
5
2
2
5
4
o
3
5
2
4
-\
5
-\
5
2
2
5
4
5
2
2
2
5
2
5
2
5
5
2
TNJM
135
NA
53
30
62
73
15
NA
50
37
NA
113
NA
116
27
NA
NA
47
11
54
121
84
60
133
138
125
108
34
87
76
NA
26
32
101
15
84
16
29
19
NA
61
12
TNJV
3
NA
3
2
2
2
3
NA
5
5
NA
2
NA
3
2
NA
NA
2
2
5
4
4
2
3
3
4
4
2
4
2
NA
2
2
2
2
5
5
5
2
NA
5
5
ChlaJM
188
28
81
65
73
96
42
42
79
52
40
168
24
187
58
27
27
58
31
105
193
152
128
214
235
168
124
56
106
150
54
78
42
147
17
157
39
34
19
59
99
17
ChlaJV
2
5
o
3
3
2
2
o
3
5
2
o
3
5
2
5
o
3
2
5
5
2
2
5
3
o
3
5
o
3
3
2
2
2
2
4
5
5
2
2
2
4
3
o
3
2
5
4
2
Seechi_N
190
11
96
60
77
100
41
19
88
56
20
182
24
188
58
27
12
60
31
89
176
143
109
199
217
173
128
56
109
143
32
78
42
151
17
150
39
35
19
36
96
17
SecchiJV
3
5
-\
5
2
2
2
5
5
2
4
5
2
5
o
3
2
5
5
2
2
5
4
o
3
5
2
4
4
o
3
2
2
4
5
5
2
2
3
5
4
4
4
2
3
2
29
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
Taxon
Polyarthra. dolichoptera
Polyarthra. major
Polyarthra. remata
Polyarthra. vulgaris
Polyphemidae
Pompholyx
Skistodiaptomus. oregonensis
Synchaeta
Trichocerca
Trichocerca. birostris
Trichocerca. cylindrica
Trichocerca. multicrinis
Trichocerca.porcellus
Trichocerca. rousseleti
Trichocerca. similis
Tropocyclops.prasinus.mexicanus
Table 10 Descriptive statistics
„ , Valid
Optima N
TN(uf) (mg/L) 17
TP(uf) (mg/L) 17
Secchi Depth (m) 17
Chlorophyll-a (ug/L) 17
Table 11 Descriptive statistics
Optima Valid N
TN(uf) (mg/L) 34
TP(uf) (mg/L) 34
Secchi Depth (m) 34
Chlorophyll-a (ug/L) 34
TPJM
68
76
137
159
12
10
93
119
11
61
79
92
15
35
30
119
TP_IV
2
4
2
4
5
5
3
5
2
o
5
5
5
2
5
5
4
for family level
Mean
0.233
0.008
4.850
5.226
Median
0.234
0.007
4.843
3.582
TNJM TNJV
67
71
138
137
NA
10
90
100
NA
61
64
78
16
34
30
113
2
5
3
4
NA
5
5
5
NA
4
5
4
2
5
5
5
nutrient optima
ChlaJM
92
102
184
233
21
15
107
180
37
86
135
124
17
48
37
183
ChlaJV SeechiJM SecchiJV
2
-\
5
2
4
4
5
o
5
4
2
2
5
5
2
4
5
3
across 17 taxa from
Minimum Maximum
0.200
0.005
3.807
2.262
0.271
0.014
5.873
12.075
for species/genus level nutrient optima
Mean
0.238
0.008
4.939
3.675
Median
0.242
0.007
4.828
3.237
Min
0.178
0.004
3.807
1.610
Max
0.284
0.014
6.403
10.776
Lower
Quartile
0.218
0.006
4.444
3.064
from Table
95
102
189
215
19
15
111
162
37
88
117
117
17
49
39
167
Tables.
Upper
Quartile
0.245
0.009
5.240
5.695
9.
Lower Quartile Upper Quartile
0.214
0.006
4.408
2.868
0.262
0.009
5.397
3.841
2
4
2
4
4
5
o
5
5
3
5
5
5
5
5
5
3
Std.Dev
0.022
0.002
0.508
3.234
Std.Dev.
0.027
0.002
0.662
1.952
30
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
Diacyclops. thomasi
>» CO
= e>
ID
O o
£
o> •»
°
o .
o
(» », fc
§ I
o C
CD
o Q)
I 7 ! 1 I T ) I 7 T I
0001 0003 001 003
Total Phosphorus (mg/L)
Mesocyclops edax
= o
!D
2 *>
Q.
S? S
3
Q.
rc CM
O o
s S
6 C
_ T3
-s|
_ <
- O Q)
C3 >
M
0001 0003 001 003
Total Phosphorus (mg/L)
Figure 4 Examples of zooplankton taxa response curves (capture probability) along the TP
gradient. For all zooplankton response curves, see the Appendices 4-7 and 16-19.
WA inference models were constructed based on zooplankton data as for phytoplankton. In contrast, the
performance of these models for Secchi depth, TN, TP, and Chi a was better than for phytoplankton, and
was especially good for TN and TP (Figure 5).
31
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
.§
s.
I 4
FP=O.J4
4 6 8 10 0.0 0.1 0.2 0.3 0.4
Measured Secchi Depth {m} Measu red TN
-3.;
-2.5 -2.0
Vss = „•=•: T=
0.0 -
-2.0 -
<
I-
R-=0.22
o 6° 3°Ji$S$^&
10 -
R-= 0.4
-O.J
0.0 0.5 1.0
Measured Chi a
-2.5
-2.0 -1.5
sss.'K NH-
89
SSLra CO
Figure 5 Performance of zooplankton WA inference models for nutrient variables (LogNHS, TN and
LogTP in mg/L), dissolved oxygen (DO in mg/L), Secchi depth (m), and Chlorophyll a (LogChl a in
H9/L). Inferred environmental variable values are the sum product of relative abundance and
nutrient optima across taxa at a sample/site.
For species level analysis those taxa that were most responsive to each nutrient/indicator and which had
trends that were due to data from multiple lakes (i.e., samples from a single lake were not creating the
entire pattern) were scaled and averaged to arrive at a 3-taxa metric for high and low levels of
nutrients/indicators (Table 12). For consistency, attempts were made to specifically choose taxa that were
responsive across the four nutrient/indicators. The taxa that were found in high TN environments were
substantially different from those that were found at high levels of the three other nutrient/indicators;
therefore a separate metric was developed to represent high TN environments. For high TN, Synchaeta,
Tricocerca cylindrica, and Tropocyclopsprasinus mexicanus (high categories [4-5] in Table 9) were used
as the high nutrient indicator. For high TP, low Secchi Depth, and high Chlorophyll-a the taxa
Asplanchna, Trichocerca cylindrica, wdMesocyclops edax (high categories [4-5] in Table 9) were used
in assembling the metric. For the low nutrient/indicator metric Diacyclops thomasi, Leptodiaptomus
minutus, and Kellicotia longispina (low categories [2-3] in Table 9) were used for all four
nutrients/indicators (low TN, low TP, high Secchi, low Chl-a). The eight metrics developed showed
consistently predictable responses to nutrient concentrations (Figure 6-9). From such plots,
regression/quantile regression could be used to identify P thresholds associated with specified metric
targets, if such were developed for assessing/managing aquatic life.
32
-------
Numeric Nutrient Criteria for Freshwaters
Methodologies for Development of
Part 1: Nutrient Sensitive Aquatic Life Uses
Table 12 Metrics developed by scaling and averaging three taxa for each nutrient (TN, TP) or indicator (Secchi, Chla) that showed clear
relationship (either increased or decreased abundance) with nutrient/indicator levels.
A.
B.
C.
D.
E.
F.
G.
H.
Metrics
TN HiTN (Synchaeta, Tricocerca cylindrica, and Tropocyclops prasinus mexicanus)
TP HiTP,Chla,loSecc (Asplanchna, Trichocerca cylindrica, and Mesocyclops edax )
Secc HiTP,Chla,loSecc (Asplanchna, Trichocerca cylindrica, and Mesocyclops edax )
Chla HiTP,Chla,loSecc (Asplanchna, Trichocerca cylindrica, and Mesocyclops edax )
TN LoTP,Chla,TN,hiSecc (Diacyclops thomasi, Leptodiaptomus minutus, and Kellicotia longispina )
TP LoTP,Chla,TN,hiSecc (Diacyclops thomasi, Leptodiaptomus minutus, and Kellicotia longispina )
Secc LoTP,Chla,TN,hiSecc (Diacyclops thomasi, Leptodiaptomus minutus, and Kellicotia longispina )
Chla LoTP,Chla,TN,hiSecc (Diacyclops thomasi, Leptodiaptomus minutus, and Kellicotia longispina )
Optima
0.266
0.010
4.251
5.470
0.194
0.005
5.848
2.410
Unit
mg/L
mg/L
m
mg/L
mg/L
mg/L
m
mg/L
Optima
Calculated
-0.574
-2.003
6.487
5.470
-0.713
-2.293
5.848
2.410
Unit
loglO mg/L
loglO mg/L
m
mg/L
loglO mg/L
loglO mg/L
m
mg/L
Sum
count
124
137
198
208
137
154
223
225
33
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
so
40
30
s-
10
0
-10
-1
90
30
r- 70
^ 6°
0 50
M"
3"
0) 20
(0
0 10
0
00
o
0
0
8 o o°
a n °
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
HiTP, Chla. loSecc metric vs. Average Unfiltered TP
T)
O
ff
I
o
70
50
JO
30
20
10
in
0
o
o
0 *
o
> 0
o
o
0
o
0 . °
0 %° °0
Oo °g ° o00*o
0° "o^oo10
vfr0
°°0 0
3
0
0
0
0 °
0° 0
-28 -26 -24 -22 -20 -13 -16 -14
Log Average Unfiltered TP (mg/L)
-12
LoTP,Chla,TN,hiSecc metric vs. Average Unfiltered TP
80
o
Tl
O 50
5>" 40
H
Z30
-3
1
2 -3
)
0 -2
c
o
0°,!
8 -2
f
0
0
0 <
«te° '
O0<
0 <
6 -2
' »
0
°o <
J
b°fc~
^*°nj
-I -2
J
1
f
.
^°
%t^
2 -2
o
O
0 -1
0
0*
8 -1
o
6 -1
* -1
Log Average Unfiltered TP (mg/L)
Figure 7 Three-taxa metric values representing TP-tolerant (metric B) and intolerant taxa (metric F,
see Table above) versus TP (log 10 mg/L).
35
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
HiTP, Chla, loSecc metric vs. Average Secchi Depth
60
3 50
•
O 40
B
30
O
sr -
o
10
0
.in
o
o
d> 0
0
0 0
O ' ° 0
00 ~ *
°oS
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
80
70
60
q so
5T 30
o
8?
HITP, Chla, loSecc metric vs. Average Chllorophyll a
p o p oo>
CJ i. '.n :r.
o b b
O O 00000
Average Chlorophyll a (ug/L)
LoTP,Chla,TN,hiSecc metric vs. Average Chlorophyll a
O O 00000
Average Chlorophyll a (ug/L)
Figure 9 Three taxa metric values representing taxa found in waters of high (metric D) and low
Chlorophyll-a densities (metric F, see Table above) versus Chlorophyll-a (mg/L).
37
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
1.3.2.1.3 Littoral Macroinvertebrate dataset
There were a total of 981 invertebrate samples collected from 43 WI lakes from 1981 to 2010. The
majority of samples were collected in Little Rock Lake (280, by Little Rock Lake Experiment project),
Trout Lake (240, LTER), Sparkling Lake (231, LTER), and Crystal Lake (42). Six different collection
methods were used, including benthic samples (vacuum, 140), benthic sample on site without wood
(vacuum, BTH, 54), benthic sample on site with wood (vacuum, CBTH, 52), coarse woody habitat
(vacuum, CWH, 193), vacuum (151) method, and Hester-Dendy substrate (391) method. Due to the
relatively small sample sizes, all the vacuum samples from natural habitat were combined to perform data
analysis. The artificial substrate samples were analyzed separately but were also combined with natural
habitat samples to increase the sample size and environmental gradients. The results for the three different
datasets are presented separately.
224 unique taxa were identified, many of them occurring in less than 25 samples; OTU were developed
for different levels of taxonomy, but in the end, species, genus, and family levels were used for the
analysis. Average chemical conditions across the samples were calculated (Table 13) and PCA axes
derived from these variables (Figure 10). TP and TN gradients were small compared to other assemblages
and other regions, likely due to the lack of invertebrate data for the southern WI dataset.
Table 13 Summary statistics
Variable
Cond (uS/cm)
DO (mg/L)
pH
WTemp (°C)
DOC (mg/L)
TOC (mg/L)
NH3_N (mg/L)
N023_N(mg_L)
TN_f(mg/L)
TN_uf(mg/L)
TP_f(mg/L)
TP_uf(mg/L)
SecchiDepth (m)
Ch a (ug/L)
Number of
All Samples
152
913
658
913
612
415
498
509
392
545
509
556
535
535
Number of
Natural
Samples
124
555
275
555
350
151
151
151
151
275
151
275
151
151
Number of
Artificial
Substrate
28
358
383
358
262
264
347
358
241
270
358
281
384
384
Minimum
12
6.01
5.69
7.25
1.478
1.517
0.011
0.003
0.136
0.099
0.003
0.004
1.00
1.58
1st
quartile
16.88
8.26
7.39
8.92
2.905
2.816
0.021
0.008
0.205
0.261
0.005
0.008
4.36
3.02
Median
19
8.60
7.48
10.57
3.293
3.049
0.030
0.011
0.257
0.314
0.005
0.010
4.74
3.66
3rd
quartile
48
9.10
7.53
12.19
4.255
3.278
0.042
0.031
0.305
0.357
0.007
0.013
5.47
4.58
Maximum
191.67
10.49
8.33
21.52
8.240
10.600
0.080
0.041
0.403
0.880
0.011
0.045
7.69
106.75
38
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
-30
I
-20
-10
0
I
10
I
20
30
o
ci
m
p
ci
CM O
O o
CL o
m
p
ci
o
ci
pH
L_N023_N_f_mg_L
DO_mg_
WTemp_C
Secchi Depth_m
\
-0.10
T
-0.05
o
CO
o
CM
o
- o
o
CM
o
CO
0.00
0.05
0.10
PC1
Figure 10 Principal component analysis of environmental variables in the invertebrate dataset.
The first axis is associated with nutrient (TP, TN, and Chi a) concentrations, while the second axis
is associated with Secchi Depth and pH.
Due to the absence of various environmental variables, PCA axes were not used to regress against
invertebrate compositions. Instead, the relationships between TN, TP, Chi a and Secchi depth and
invertebrate compositions and individual taxa were examined. Macroinvertebrate indicator values (IV) to
these four nutrient variables were developed from the methods described earlier for three different
datasets (all data, natural habitat data, and artificial habitat data). These indicator values were developed
for the highest resolution taxonomy first, most likely at species level; then taxa were combined into
genus, and family level to develop indicator values. Of these four variables, TP was affected by detection
limits, Chi a concentrations only span for a narrow range and its peak is not easy to capture, while Secchi
39
-------
Methodologies for Development of Part 1: Nutrient Sensitive Aquatic Life Uses
Numeric Nutrient Criteria for Freshwaters
depth was also strongly associated with lake depth (a natural variable). As a result, only TN indicator
values were examined further.
Table 14 Macroinvertebrate Indicator values (IV) to total nitrogen (TN) concentrations from
different datasets. An IV of 2 is an indication of sensitivity to high TN concentration and 5 is an
indicators of tolerance to high TN concentrations. Also shown are the sample sizes (N) for each
estimate.
All data
Taxon
Dorylaimidae
Thomia
Halacaridae
Elmidae
Dubiraphia
Dubiraphia minima
Dubiraphia quadrinotata
Dubiraphia robusta
Ectopria
Stenelmis
Gyrinidae
Gyrinus
Psephenidae
Ceratopogonidae
Bezzia/Palpomyia
Dasyhelea
Probezzia
Chaoboridae
Chaoborus
Chironomidae
Phaenopsectra/Tribelos
Tanytarsini
Baetidae
Callibaetis
Centroptilum
Procloeon
Caenidae
Caenis
Ephemerellidae
Eurylophella
Ephemeridae
Ephemera
Number of
Samples
77
77
45
194
108
74
74
74
79
75
77
74
79
245
74
140
75
73
73
534
99
96
122
74
74
74
303
283
164
109
119
81
IV
2
2
4
5
2
3
3
3
4
5
2
3
4
5
3
5
2
3
3
5
4
4
5
3
3
2
5
5
2
2
5
4
Hester-Dendy
TN_N
50
50
45
107
78
49
49
49
49
17
50
49
49
106
49
49
50
49
49
264
49
49
90
49
49
49
190
170
137
84
49
49
TNJV
2
2
5
5
2
3
3
3
3
2
5
3
3
5
3
3
2
3
3
5
3
3
5
3
3
3
5
2
2
2
3
3
Natural
TN_N
27
27
47
87
30
25
25
25
30
58
27
25
30
139
25
91
25
24
24
270
50
47
32
25
25
25
113
113
27
25
70
32
Habitat
TNJV
2
2
5
5
4
3
o
3
o
3
4
5
4
o
3
4
5
o
3
4
o
5
2
2
5
2
2
4
o
3
3
o
3
5
5
5
3
5
4
40
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
All data
Taxon
Hexagenia
Heptageniidae
Stenacron
Stenacron interpunctatum
Stenonema
Stenonemafemoratum
Leptohyphidae
Tricorythodes
Leptophlebiidae
Leptophlebia
Paraleptophlebia
Sialidae
Sialis
Aeshnidae
Basiaeschna
Basiaeschna Janata
Coenagrionidae
Enallagma
Enallagma ebrium/hageni
Corduliidae
Epitheca
Macromia
Gomphidae
Dromogomphus
Dromogomphus spinosus
Gomphus
Hagenius
Hagenius brevistylus
Libellulidae
Celithemis
Leucorrhinia
Sympetrum
Hydroptilidae
Hydroptila
Oxyethira
Leptoceridae
Mystacides
Oecetis
Triaenodes
Phryganeidae
Number of
Samples
113
267
75
74
210
87
74
74
170
75
51
119
119
124
84
74
232
191
74
121
92
95
217
128
99
130
93
76
108
76
74
75
142
82
97
307
116
212
89
78
IV
5
5
4
3
2
5
3
3
5
4
4
5
5
5
3
3
5
4
3
4
4
4
5
5
5
4
5
3
5
3
3
3
5
2
2
5
4
5
2
2
Hester-Dendy
TN_N
49
220
49
49
163
49
49
49
144
49
51
50
50
98
58
49
179
151
49
60
55
51
70
51
49
49
55
49
65
49
49
49
87
57
65
158
59
86
61
52
TNJV
3
5
3
3
3
3
3
3
5
3
5
2
2
5
5
3
5
5
3
5
5
4
5
4
3
3
5
3
5
3
3
3
5
2
2
5
3
4
3
2
Natural
TN_N
64
26
25
47
38
25
25
26
26
69
69
26
26
25
53
40
25
61
37
44
147
77
50
81
38
27
43
27
25
26
55
25
32
149
57
126
28
26
Habitat
TNJV
5
4
3
5
5
3
o
3
4
4
4
4
3
-\
5
3
5
5
3
4
4
4
5
5
5
4
5
3
5
2
3
o
3
5
-\
5
3
5
5
5
2
2
41
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
All data
Taxon
Polycentropodidae
Polycentropus
Hyalellidae
Hyalella
Hyalella azteca
Cambaridae
Orconectes
Asellidae
Caecidotea
Pisidiidae
Pisidium
Ancylidae
Hydrobiidae
Lymnaeidae
Physidae
Physa
Planorbidae
Gyraulus
Planorbella
Erpobdellidae
Erpobdella
Hirudinidae
Haemopis
Haemopis grandis
Planariidae
Planaria
Enchytraeidae
Lumbriculidae
Naididae
Chaetogaster
Ripistes
Ripistes parasita
Slavina
Slavina appendiculata
Stylaria
Stylaria lacustris
Tubificidae
Number of
Samples
219
146
340
340
174
97
97
114
97
212
199
74
167
100
124
96
169
91
86
74
74
74
74
74
83
83
74
83
74
74
74
74
74
74
74
74
95
IV
5
2
2
2
4
2
2
2
4
5
5
3
4
4
4
4
4
2
4
3
3
3
3
3
2
2
3
4
3
3
2
2
3
3
3
3
4
Hester-Dendy
TN_N
165
114
221
221
55
65
65
66
49
52
52
49
83
49
81
53
85
54
49
49
49
49
49
49
83
83
49
49
49
49
49
49
49
49
49
49
49
TNJV
5
2
2
2
5
2
2
2
3
5
5
3
2
3
5
5
2
2
3
3
3
3
3
3
2
2
3
3
3
3
3
3
3
3
3
3
3
Natural
TN_N
54
32
119
119
119
32
32
48
48
160
147
25
84
51
43
43
84
37
37
25
25
25
25
25
25
34
25
25
25
25
25
25
25
25
46
Habitat
TNJV
5
2
4
4
4
4
4
4
4
5
5
3
4
2
4
4
4
2
2
3
o
3
o
3
3
o
3
3
2
o
3
3
o
3
3
o
3
3
o
3
3
2
42
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
All data
Taxon
Tetrastemmatidae
Prostoma
Hydridae
Hydra
Number of
Samples
95
95
150
ISO
IV
4
4
2
2
Hester-Dendy
TN_N
70
70
125
125
TNJV
4
4
2
2
Natural Habitat
TN_N
25
25
25
25
TNJV
3
3
3
3
As seen from Table 14 above, the IVs developed (e.g., Figure 11 and see Appendices 8 and 20) from
different datasets are fairly consistent for the majority of taxa, as long as the number of samples was large
(>40). However, several taxa may have to be examined further to determine the sensitivity to nutrient
concentrations since the rank in different datasets were not consistent, e.g., Stenonema. The most
sensitive taxa, on average, across all methods and datasets were Dorylaimidae, Thornia, Phryganeidae,
and Polycentropus, while Pisidiidae (Sphaeriidae), Pisidium, Hydroptilidae, Leptoceridae,
Polycentropodidae, Coenagrionidae, Gomphidae, Hagenius, Libellulidae, Caenidae, Heptageniidae,
Ceratopogonidae, Chironomidae, and Elmidae were consistently tolerant. See Appendix 20 for the full
table of IVs.
Thornia
-Q ^P
O o
Q.
TO (v
(J o
o
o
02
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
As for phyto- and zooplankton, WA inference models were developed with macroinvertebrate data and
were generally poor (Figure 12). Again, the abbreviated nutrient gradient due to a lack of southern WI
lakes was likely a factor.
Q.
Q
T3
(D
5.4
5.2 -\
5.0
4.8 -
4.6 -
4.4 -
\ I I I I I
1 23456
Measured Secchi Depth (m)
8>
0.40 -
0.35 -I
I
<
0.30 -
0.25 -
ooS
o
o
0.2
o o Oo
I I
0.4 0.6
Measured TN
0.8
-1.90 -
H-1.95 -|
T3
|
^-2.00 H
-2.05 -
I I I I I
-2.2 -2.0 -1.8 -1.6 -1.4
Measured TP
0.70 -
ro
6
(D
OJ
0.65
0.60 -I
0.55 -
0.50 -
0.45 -I
\
\
\
0.5 1.0 1.5
Measured Chi a
T
2.0
Figure 12 Performance of invertebrate WA inference models for four nutrient variables. (TN and
LogTP in mg/L, Log Chi a in ug/L). Invertebrate inferred environmental variable values are the
sum product of relative abundance and nutrient optima across taxa at a sample/site. However,
more than one third of the invertebrate data were qualitative (recorded as 0 or blank in the original
dataset), these values were coded as 0.1 to separate from the actual quantitative values (range
from 0.17 to 2490). As a result, weighted averages based on abundance data were not really
helpful. Instead, the final indicator values most heavily rely on present/absence and tolerance
values. (Spearman r in the graphs are 0.2, 0.53, 0.19, 0.26 respectively). TN has the best
performance even with the qualitative dataset.
44
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
1.3.2.1.4 Fish Dataset
Fish data were downloaded from the North Temperate Lakes Long Term Ecological Research (NTL
LTER) website, which contained data generated by the NTL LTER program, as well as data developed in
association with other state level research efforts. Downloaded data originated from three different
projects: NTL LTER, Biocomplexity, and Landscape Position (See data description above). Fish were
collected by electrofishing, beach seines, and gill nets (http://her, limnology.wise. edu/dataset/north-
temperate-lakes-lter-fish-abundance-1981-current; http://lter.limnologv.wisc.edu/protocol/landscape-
postition-project-fish). After downloading to an Access database the data were standardized on a catch
per unit effort basis; for electrofishing and gill net data CPUE was expressed as a time-based level of
effort; seine data were expressed as number of hauls. Sites were located primarily in the northern part of
the state; however, there were four sites (Lake Mendota, Lake Monona, Lake Wingra, and Fish Lake) that
were located in the southern half of the state near Madison (figure in zooplankton text).
Each of the three collection method datasets was analyzed and optima were developed. Electrofishing
and seine data showed similar patterns to each other but electrofishing data were more comprehensive and
so were used for the final analysis. Gill net data were minimal and the number of taxa caught was
substantially lower than through electrofishing and seine methods.
Nutrient and nutrient-related data (NH3, NO23, TN, TP, Secchi depth, and Chl-a) were downloaded and
logio transformed for analysis (excluding Secchi and Chl-a). PO4 data were also downloaded but were
only available for three lakes so these data were not used in the final analysis. For some of these
parameters measurements were available for multiple depths throughout the water column. For these, the
measurements within one meter of the water surface were averaged and used as a single sample for each
site. Descriptive statistics for the nutrients (NH3, NO23, TN, TP) and nutrient indicators (Secchi depth,
Chlorophyll-a) are presented in Table 15. Table 16 presents the number offish and nutrient/indicator
samples available for each lake.
Table 15 Descriptive statistics for nutrients (NH3, NO23, TN, TP) and nutrient indicators (Secchi
depth and Chlorophyll-a).
Nutrient/Indicator Valid N Mean Median Minimum Maximum Lower Quartile Upper Quartile Std.Dev.
NIC (mg/L)
NO23 (mg/L)
TN (unfiltered, mg/L)
TP (unfiltered, mg/L)
Secchi Depth (m)
Chlorophyll-a (ug/L)
182
186
243
271
241
155
0.034
0.020
0.369
0.014
3.839
4.217
0.026
0.013
0.339
0.012
3.750
3.293
0.005
0.002
0.096
0.003
0.553
1.393
0.315
0.761
1.457
0.108
7.031
15.131
0.017
0.007
0.259
0.007
2.700
2.457
0.055
0.028
0.491
0.020
5.000
5.372
2.773
4.193
1.703
2.293
1.567
2.472
45
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
Table 16 Sample count by lake for calculations - each observation indicates that fish and
nutrient/indicator data were available.
Lake Name
Allequash Lake
Anvil Lake
Arrowhead Lake
Bass Lake
Big Lake
Big Muskellunge Lake
Big Portage Lake
Big Saint Germain Lake
Birch Lake
Black Oak Lake
Boulder Lake
Brandy Lake
Carpenter Lake
Circle Lily Lake
Crab Lake
Crampton Lake
Day Lake
Diamond Lake
Eagle Lake
Erickson Lake
Escanaba Lake
Fish Lake
Flora Lake
Found Lake
Indian Lake
Island Lake
Jag Lake
Johnson Lake
NH3 NO23 TN
29 30 26
1
3
1
4
29 30 25
1
1
1
1
2
3
1
1
1
1
2
1
1
1
16 16 16
2
1
1
1
1
3
TP Secchi Depth
27 31
1
3 2
1
4 3
26 31
1
1
1
1
2 2
3 2
1
1
1
1
2 2
1
1
1
16 17
2 2
1
1
1 1
1
3 2
Chlorophyll-a
30
2
3
30
2
2
2
2
2
46
-------
Methodologies for Development of Part 1: Nutrient Sensitive Aquatic Life Uses
Numeric Nutrient Criteria for Freshwaters
Jute Lake
Katinka Lake
Lac du Lune
Lake Laura
Lake Mendota
Lake Monona
Lake Wingra
Little Arbor Vitae Lake
Little Crawling Stone Lake
Little Crooked Lake
Little John Lake
Little Muskie Lake
Little Saint Germain Lake
Little Spider Lake
Little Star Lake
Little Sugarbush Lake
Little Trout Lake
Lost Lake
Lynx Lake
McCullough Lake
Moon Lake
Morton Lake
Muskellunge Lake
Muskesin Lake
Nebish Lake
Nelson Lake
Otter Lake
Oxbow Lake
Pallette Lake
Palmer Lake
Papoose Lake
1
3
1
1
17 17 4
17 17 4
16 16 16
1
1
3
1
2
1
1
1
2
2
1
3
1
1
1
1
2
1
1
1
1
1
1
1
1
32 2
1
1
17 18
17 18
16 17
1
1
32 2
1
22 2
1
1
1
22 2
22 2
1
3 2
1
1
1
1
22 2
1
1
1
1
1
1
1
47
-------
Methodologies for Development of Part 1: Nutrient Sensitive Aquatic Life Uses
Numeric Nutrient Criteria for Freshwaters
Pioneer Lake
Round Lake
Sparkling Lake
Star Lake
Statenaker Lake
Stearns Lake
Stormy Lake
Tenderfoot Lake
Towanda Lake
Trout Lake
Twin Lakes
Upper Buckatabon Lake
Upper Kaubashine Lake
Vandercook Lake
White Birch Lake
White Sand Lake
Wild Rice Lake
Wildcat Lake
Total
1
3
29 30 26
1
2
2
1
1
1
29 30 25
1
1
2
1
2
-\
5
3
2
182 186 243
1
32 2
26 31 30
1
2 2
22 2
1
1
26 31 30
1
1
2 2
1
22 2
3 2
3 3
22 2
271 241 155
There were variations in pH levels between north and south sites - average pHs were substantially lower
in the north than the south; however there were lakes in the north that had pH levels similar to those in the
south indicating consistency in this naturally variable parameter. There were higher nutrient levels in the
south; however this is also a more highly developed area (near Madison) and these levels were likely due
to this anthropogenic influence, as opposed to entirely natural variation.
Weighted averages were used to describe the point along each nutrient/indicator gradient at which
abundance of each taxon was maximized (Figure 13). Optima and indicator values were calculated for
fish taxa that were found in at least 20 samples (Table 17, and see Appendices 9-11 and 21-23). Optima
for each nutrient/indicator were placed into categories according to nutrient tolerance and based on 25th,
50th, 75th percentiles with 2 being least tolerant and 5 being most tolerant (i.e., 2=<25th, 3=25th-median,
4=median-75th, 5=>75th percentile). Generally, there was consistency across the nutrient/indicators with
regard to the specific taxa that were tolerant or intolerant to nutrient enrichment (e.g., all Bluegill were
category 5; all Mimic Shiner were category 2).
WA inference models were developed for the fish dataset as for the other assemblages and exhibited the
best performance of any assemblage (Figure 14). The inference models were best for TP, followed by
Secchi depth, NH3, TN, and Chl-a respectively.
48
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
Two metrics, one representing nutrient tolerant taxa and one representing nutrient sensitive taxa, were
developed for each of the nutrient indicators. Six taxa were chosen based on indicator values and average
rankings for TN and TP, and abundances to represent high and low nutrient tolerance (six taxa in each).
Taxa included in the nutrient tolerant metric were White Bass, Freshwater Drum, Common Carp, Bowfm,
Green Sunfish, and Bluegill; taxa included in the nutrient sensitive metric were Mimic Shiner, Mottled
Sculpin, Smallmouth Bass, Rock Bass, Johnny Darter, and Walleye (Table 18). Abundance values for
these taxa were standardized on a 100 point scale using the 1st and 99th percentiles of the value ranges and
averaged to arrive at final metric scores. Indicator scores for each of the nutrient/indicators were then
developed based on values of the two metrics. Four of the six taxa included in the high nutrient tolerance
metric were only found at the southern group of sites. Green Sunfish and Bluegill were the two taxa
found in both the north and the south and only Bluegill was prominent throughout (Green sunfish were
only found at two sites in the north). Three of the six nutrient sensitive taxa were only found in the north
- Mimic Shiner, Mottled Sculpin, Johnny Darter; Walleye Smallmouth Bass, and Rock Bass were found
throughout the north and the south. Plots of the metric values versus nutrient/indicator values are
presented in Figure 15. The southern sites often in these plots separate out from the main group of sites;
however the trend appears consistent with regard to general fish relationship to nutrient/indicator levels.
Additional combinations could clearly be considered to derive additional plots and regression/quantile
regression used to identify P thresholds associated with specified metric targets, if such were developed
for assessing/managing aquatic life.
Table 17 Fish Indicator values (IV) to total nitrogen (TN) concentrations from different datasets. An
IV of 2 is indicative of sensitivity to high TN concentration and 5 is indicative of tolerance to high
TN concentrations. Also shown are the sample sizes (N) for each estimate.
Taxon
TP N TP IV TN N TN IV Chla N Chla IV Seechi N Secchi IV
Bigmouth.buffalo
Black.bullhead
Black.crappie
Blackchin.shiner
Blacknose. shiner
Bluegill
Bluegill .x.pumpkinseed.hybrid
Bluntnose. minnow
Bowfm
Brook.silverside
Brown.bullhead
Burbot
Common.carp
Common.shiner
Freshwater.drum
Golden. shiner
15
13
137
48
36
204
33
178
19
50
17
14
61
62
35
96
5
5
5
4
2
4
3
3
5
5
5
2
5
2
5
4
NA
NA
64
23
14
96
15
94
NA
24
NA
NA
35
38
NA
52
NA
NA
5
5
2
4
5
3
NA
5
NA
NA
5
2
NA
5
NA
NA
48
18
19
97
NA
90
NA
NA
NA
11
NA
49
NA
37
NA
NA
5
5
2
4
NA
2
NA
NA
NA
2
NA
4
NA
5
15
10
114
35
22
174
18
139
20
53
17
14
65
54
37
81
4
3
5
5
2
4
4
4
4
5
4
2
5
4
3
5
49
-------
Methodologies for Development of Part 1: Nutrient Sensitive Aquatic Life Uses
Numeric Nutrient Criteria for Freshwaters
Taxon
Green, sunfish
Green. sunfish.x.pumpkinseed. hybrid
lowa.darter
Johnny, darter
Largemouth.bass
Logperch
Longnose.gar
Mimic. shiner
Mottled. sculpin
Mudminnow
Muskellunge
Northern, pike
Pumpkinseed
Rock.bass
Rosyface. shiner
Shorthead.redhorse
Smallmouth.bass
Troutperch
Walleye
White.bass
White.crappie
White, sucker
Yellow.bullhead
Yellow.perch
Yellow.bass
TP_N
45
28
35
62
173
111
13
113
34
25
76
79
194
212
15
24
188
20
212
26
17
169
80
264
NA
TPJV
5
5
2
2
4
3
5
2
2
2
3
4
4
2
2
4
2
4
3
5
5
3
5
3
NA
TN_N
30
22
18
38
81
62
NA
67
18
NA
42
45
100
103
NA
11
98
14
110
NA
14
89
54
136
NA
TNJV
5
5
o
3
3
4
2
NA
2
2
NA
o
3
3
4
2
NA
2
2
2
o
3
NA
5
o
3
5
4
NA
ChlaJM
NA
NA
21
44
73
74
NA
72
24
13
58
47
96
141
NA
18
126
19
138
NA
NA
120
33
148
NA
ChlaJV
NA
NA
4
2
5
3
NA
2
3
3
4
5
4
2
NA
5
2
5
3
NA
NA
3
4
2
NA
SeechiJN
48
29
27
48
147
102
14
84
29
16
69
73
169
187
NA
21
167
19
192
27
18
159
85
233
10
SecchiJV
5
5
4
2
4
2
4
2
2
4
4
4
4
2
NA
5
2
5
2
3
5
3
5
3
3
50
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
Mimic.shiner
0003 001 003 0.1
Total Phosphorus (mg/L)
Brook, silverside
f- O
o C
«
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
: -
4 -
= 3
I
0.2 -
0.0 -
. -°'2 "
j -0.4 -
• -0.6 -
-0.8 -
R*=0.46
. -1.5 -
-2.5 -
R-= 0.67
2 3 4 6 £
Measured Secchi Depth {m}
-1.0 -0.8 -0.6 -0.4 -3.2 0.0 0.2
'.'53 = .. •« TN
-2.0 -1.
V*3 = -•=:: T=
1
= 0.42
-2.0 -
FP= 0.51
10 -
0.2 0.4 0.6 0.8 1.0
Measured CM a
-2.:
-0.5
.'K NH-
j CO
Figure 14 Performance of fish WA inference models for nutrient variables (LogNHS, LogTN and
Log TP in mg/L), dissolved oxygen (DO in mg/L), Secchi depth (m), and Chlorophyll a (Log Chi a in
Hg/L). Inferred environmental variable values are the sum product of relative abundance and
nutrient optima across taxa at a sample/site
Table 18 High and low nutrient tolerance 6-taxa metric optima.
Metric
TN 6-taxa High Nutrient Metric
TN 6-taxa Low Nutrient Metric
TP 6-taxa High Nutrient Metric
TP 6-taxa Low Nutrient Metric
Secchi Depth 6-taxa High Nutrient Metric
Secchi Depth 6-taxa Low Nutrient Metric
Chlorophyll-a 6-taxa High Nutrient Metric
Chlorophyll-a 6-taxa Low Nutrient Metric
Optima
0.660
0.311
0.030
0.011
2.657
4.307
6.035
4.284
Unit
mg/L
mg/L
mg/L
mg/L
m
m
ug/L
ug/L
Calculated
-0.180
-0.508
-1.529
-1.968
2.657
4.307
6.035
4.284
Unit
log 10 mg/L
log 10 mg/L
log 10 mg/L
log 10 mg/L
m
m
ug/L
ug/L
Sum count
174
213
201
240
169
211
91
151
52
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
6-taxa High Nutrient Metric vs. Average TN
IZU
1
ft-
QJ
X
QJ
1 oU
dq'
— ^
fu'
fD
— ^
0/1
n
O
— *
n> n
u
-ZU
-1
.2 -1
O
0 <
.0 -0
c
OcPSS^I
.8 -0
O
O
o
>sJP%Lcf
&BM
cv^^v
.6 -0
> o
_ O a
w ^
.4 -0
O
O
°
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
As for the Wisconsin datasets, optima and indicator values were developed for fish taxa occurring in 20 or
more TVA samples using weighted average methods as above (e.g., Figure 16). Summary nutrient
condition distributions for the TVA lakes were calculated (Table 19) and sample count data for optima are
also shown (Table 20). From these datasets, optima were calculated and optima indicator values category
assignments (based on 25th, 50th, and 75th percentiles of optima distributions across all taxa found in 20 or
more samples; 2=<25th, 3=25th-median, 4=median-75th, 5=>75th, Table 21 and Appendix 12). Averages
across taxa were also calculated (Table 22).
Table 19 Descriptive statistics for TVA lake dataset TN, TP and Chlorophyll-a.
Nutrient/Indicator
NH4 (mg/L)
NO23 (mg/L)
TN (unfiltered, mg/L)
TP (unfiltered, mg/L)
Chlorophyll-a (ug/L)
Valid N
274
274
274
272
274
Mean
0.020
0.129
0.380
0.029
9.039
Median
0.010
0.094
0.373
0.020
6.500
Minimum
0.005
0.005
0.065
0.001
0.500
Maximum
0.145
0.815
1.400
0.235
55.000
Lower Quartile
0.005
0.015
0.218
0.008
4.000
Upper Quartile
0.030
0.195
0.495
0.040
11.000
Std.Dev.
0.022
0.139
0.199
0.030
8.034
Table 20 Sample count by lake for optima calculations - each observation indicates that fish and
nutrient/indicator data were available.
Lake Name
Apalachia
Bear Creek
Beech
Blue Ridge
Boone
Cedar Creek
Chatuge
Cherokee
Chickamauga
Douglas
Fontana
Fort Loudoun
Ft Pat Henry
Guntersville
Hiwassee
Kentucky
Little Bear Creek
Melton Hill
Nickajack
Normandy
NH4
5
2
5
4
11
5
16
5
21
4
12
15
4
10
10
9
4
12
4
4
NO23
5
2
5
4
11
5
16
5
21
4
12
15
4
10
10
9
4
12
4
4
TNuf
5
2
5
4
11
5
16
5
21
4
12
15
4
10
10
9
4
12
4
4
TPuf
5
2
5
4
11
5
16
5
21
4
12
15
4
10
10
9
4
12
2
4
Chlorophyll-a
5
2
5
4
11
5
16
5
21
4
12
15
4
10
10
9
4
12
4
4
54
-------
Methodologies for Development of Part 1: Nutrient Sensitive Aquatic Life Uses
Numeric Nutrient Criteria for Freshwaters
Lake Name
Norris
Nottely
Ocoee#l
Pickwick
South Holston
Tellico
Tims Ford
Watauga
Watts Bar
Wheeler
Wilson
Total
NH4
10
10
4
14
10
9
10
10
13
18
4
274
NO23
10
10
4
14
10
9
10
10
13
18
4
274
TNuf
10
10
4
14
10
9
10
10
13
18
4
274
TPuf Chlorophyll-a
10
10
4
14
10
9
10
10
13
18
4
272
10
10
4
14
10
9
10
10
13
18
4
274
Table 21 Fish Indicator values (IV) for the TVA dataset. An IV of 2 is an indication of sensitivity to
high concentration and 5 is indicative of tolerance to high concentrations. Also shown are the
sample sizes (N) for each estimate.
Fish Taxon
Black buffalo
Black crappie
Black redhorse
Bluegill
Bluntnose minnow
Brook silverside
Bullhead minnow
Channel catfish
Chestnut lamprey
Common carp
Emerald shiner
Flathead catfish
Freshwater drum
Gizzard shad
Golden redhorse
Golden shiner
Green sunfish
Hybrid sunfish
Largemouth bass
Logperch
Longear sunfish
Mississippi silverside
TNJV
5
o
J
3
o
3
2
3
5
4
5
3
4
2
5
5
4
4
2
2
4
o
J
o
J
4
TN_N
56
129
71
274
70
96
28
179
20
195
81
165
139
230
88
67
221
65
272
133
116
68
TPJV
4
o
5
3
-\
^
2
3
4
4
5
3
5
o
5
5
4
3
5
2
2
4
4
4
5
TP_N
55
128
71
272
68
95
28
177
20
193
79
163
137
228
88
65
219
65
270
133
115
66
ChlaJV
4
4
5
3
2
2
5
3
5
3
2
2
5
4
5
2
3
3
4
3
4
4
ChlaJM
56
129
71
274
70
96
28
179
20
195
81
165
139
230
88
67
221
65
272
133
116
68
55
-------
Methodologies for Development of Part 1: Nutrient Sensitive Aquatic Life Uses
Numeric Nutrient Criteria for Freshwaters
Fish Taxon
Northern hog sucker
Redbreast sunfish
Redear sunfish
River redhorse
Rock bass
S auger
Silver redhorse
Skipjack herring
Smallmouth bass
Smallmouth buffalo
Spotfm shiner
Spotted bass
Spotted gar
Spotted sucker
Striped bass
Threadfm shad
Walleye
Warmouth
White bass
White crappie
Whitetail shiner
Yellow bass
Yellow perch
TNJV
2
2
4
2
2
4
5
5
•-y
5
4
-\
5
2
5
5
4
5
2
3
4
5
2
5
•-y
J
TNJM
103
127
165
39
37
33
20
32
216
86
163
194
60
131
25
113
38
168
73
59
34
63
85
TPJV
2
2
4
2
2
4
o
5
5
o
J
5
2
2
5
5
5
5
2
o
3
o
3
4
2
5
4
TP_N
103
125
163
39
37
oo
33
20
30
215
86
162
192
58
131
23
111
38
166
73
59
34
62
83
ChlaJV
4
2
3
4
2
4
5
5
3
5
4
2
3
5
3
5
3
2
4
5
2
5
2
ChlaJM
103
127
165
39
37
33
20
32
216
86
163
194
60
131
25
113
38
168
73
59
34
63
85
56
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 1: Nutrient Sensitive Aquatic Life Uses
0.08
-0.01
Bluegill Catch Per Unit Effort vs. Average TN_uf (mg/L)
0.2
0.4
0.6 0.8 1.0
Log Average TN_uf (mg/L)
1.2
1.4
1.6
Figure 16 Example nutrient optima plot for TVA fish data (Bluegill relative abundance vs. TN). For
complete set of plots see Appendix 12.
Table 22 Descriptive statistics for TVA nutrient optima
Nutrient/Indicator
TNuf optima (mg/L)
TPuf optima (mg/L)
Chl-a optima (ug/L)
Valid N
45
45
45
Mean
0.401
0.031
10.972
Median
0.415
0.029
10.522
Minimum
0.137
0.006
4.058
Maximum
0.558
0.064
27.594
Lower Quartile
0.332
0.019
8.901
Upper Quartile
0.471
0.042
12.565
Std.Dev.
0.094
0.015
4.073
There was overlap in taxa between the two datasets. Some TVA taxa were substantially different from
the assignments in the WI dataset and Spearman rank correlations of the two datasets were low (0.23 for
TP optima to 0.39 for Chi optima). For example, Green Sunfish, which were ranked as 4 or 5 for all
nutrient/indicator optima in the WI dataset were ranked consistently in the TVA dataset as 2. The actual
optima for the two datasets were fairly different as well for some taxa such as Green Sunfish: TN optima
for TVA and WI were 0.257 and 0.904 mg/L, respectively; TP optima were 0.013 and 0.030 mg/L. These
results suggest that a national inference model for fish would be difficult to use based on these data and
analytical approach.
57
-------
Methodologies for Development of Part 1: Nutrient Sensitive Aquatic Life Uses
Numeric Nutrient Criteria for Freshwaters
Table 23 Comparison of optima and optima categories fortaxa in common between the Wl and TVA datasets.
Fish Taxon (in TN
common between Optima
WI and TVA) (mg/L)-WI
Black crappie
Bluegill
Bluntnose minnow
Common carp
Freshwater drum
Golden shiner
Green sunfish
Largemouth bass
Logperch
Rock bass
Smallmouth bass
Walleye
White bass
Yellow perch
0.51
0.66
0.51
0.94
na
0.53
0.90
0.65
0.32
0.30
0.31
0.33
na
0.35
TN
Optima
(mg/L)-TVA
0.40
0.34
0.32
0.40
0.47
0.47
0.26
0.42
0.39
0.27
0.34
0.28
0.41
0.41
TN
Optima
category-
WI
4
5
4
5
na
4
5
5
2
2
2
3
na
4
TN
Optima
category-
TVA
3
3
2
3
5
4
2
4
3
2
3
2
4
3
TP
Optima
(mg/L)-
WI
0.02
0.03
0.02
0.04
0.07
0.02
0.03
0.02
0.01
0.01
0.01
0.01
0.07
0.01
TP
Optima
(mg/L)-
TVA
0.03
0.02
0.01
0.02
0.04
0.04
0.01
0.03
0.03
0.01
0.02
0.01
0.03
0.03
TP
Optima
Category-
WI
4
5
4
5
5
4
5
5
3
2
2
2
5
3
TP
Optima
Category-
TVA
3
3
2
3
5
5
2
4
4
2
3
2
3
4
Chi
Optima
(mg/L)-
WI
7.46
6.02
5.07
na
na
6.26
na
5.34
4.92
4.48
3.78
4.41
na
5.07
Chi
Optima
(mg/L)-
TVA
11.27
10.36
6.97
9.28
13.63
7.34
8.90
11.60
10.22
5.47
10.21
10.05
11.15
7.12
Chi
Optima
Category-
WI
5
5
-\
5
na
na
5
na
4
3
2
2
2
na
3
Chi
Optima
Category-
TVA
4
3
2
3
5
2
3
4
3
2
-\
5
3
4
2
58
-------
Methodologies for Development of Part 1: Nutrient Sensitive Aquatic Life Uses
Numeric Nutrient Criteria for Freshwaters
Nutrient sensitive taxa appeared across all assemblages with all four assemblage groups providing a range
of indicator taxa from nutrient sensitive to nutrient tolerant. While inference models suggest robust
response across the different assemblage groups, zooplankton and fish appeared to produce the most
precise models in the WI datasets (there was only one assemblage in the TVA dataset). It was surprising
that phytoplankton were not more responsive, given the consistent demonstration of phytoplanktonic
responsiveness to nutrients elsewhere (reviewed in USEPA 2001, 2009). It is hypothesized that part of
the reason for this was a lack of sufficient data for this assemblage across the gradient.
Moving forward, it appears that nutrient sensitive aquatic life use measures could be constructed from any
of the assemblages or a combination of assemblages, both of which have also been recognized by the
national survey datasets (USEPA 2009). The weakness of some relationships may be a criticism for use
of these assemblage tools in management for nutrients. However, when put in the context of existing
experimental nutrient-response evidence at whole lake and mesocosm scales, the data are compelling.
They provide additional evidence that assemblage responses to nutrient gradients exist and can be used as
valuable tools for setting and assessing against aquatic life use management goals with respect to
nutrients. Some assemblage responses evaluated here (e.g., phytoplankton) clearly suggest this particular
assemblage may not be applicable for developing aquatic life use assessment tools with this dataset, but
this assemblage has consistently proven its sensitivity to nutrients (e.g., USEPA 2000, Wetzel 2001,
USEPA 2009) and would likely do so with improved taxonomic resolution. Uncertainty exists in any
relationship and, rather than provide a simple threshold for use or non-use of stressor-response
relationships, is better used to guide evaluation of risk in making specific decisions. To this end, the
uncertainty in the relationships should be communicated and factored into the decision-making process,
as detailed in guidance on using stressor-response data for nutrient criteria development (USEPA 2010).
These results suggest that states should be encouraged to use biological assemblage responses for
characterizing aquatic life use in lakes, especially zooplankton and fish, but also theoretically
phytoplankton and macroinvertebrates. All of these groups contain sufficient breadth of sensitivity to
nutrient enrichment to provide a useful response indicator for this stressor. Macrophytes were not
valuated in this analysis due to insufficient data, but the literature review suggests this assemblage is also
a valuable potential indicator for use in characterizing aquatic life sensitivity to nutrients.
During the next phase of this research, select assemblage groups should be pursued along with the
traditional trophic measures used by states in these regions and identified in the literature review above, to
characterize nutrient sensitive aquatic life use and develop nutrient targets for use in receiving water
modeling.
59
-------
Methodologies for Development of Part 2: N and P in Oregon Hydrologic Landscape Regions
Numeric Nutrient Criteria for Freshwaters
2 Nitrogen and Phosphorus in Oregon Hydrologic
Landscape Regions (OHLR)
Z T
Nutrient pollution remains a vexing national pollution problem. Numeric nutrient criteria are one
dimension of the pursuit of nutrient reduction nationally. The United States Environmental Protection
Agency (USEPA) nutrient criteria guidance recommends the use of multiple lines of evidence (reference,
stressor-response, mechanistic modeling, and scientific literature) in developing nutrient criteria.
However, EPA guidance focused strongly on reference based approaches for criteria development (e.g.,
USEPA 2000) and the USEPA recommended regional criteria were developed using this approach (e.g.,
USEPA 2001), as were USEPA's final criteria for streams in Florida (USEPA 2010a). USEPA strongly
encouraged states to refine the reference analyses used in the recommended regional criteria using more
refined reference sites and more highly resolved classification. However, few states have taken advantage
of this, in part due to a lack of technical examples. This part of the report is focused on refining regional
reference criteria for Oregon which provides not only a product of value to the state in furthering nutrient
criteria for this line of evidence, but also an example of how refined regional reference criteria can be
developed. Part 2 of this document is split into two sections, the first describes the data and its
sufficiency for estimating reference criteria by OHLR and the second section the analysis to calculate the
criteria.
; . ' ,- ;
This section describes the sufficiency of data available to complete analyses of reference based nutrient
criteria for the Oregon Hydrologic Landscape Regions (Task 3). The focus of this work was assembling
data on streams and lakes across these regions, describing nutrient distributions in entire and reference
populations, and describing the sufficiency of spatial data coverage across those regions.
A database of streamwater nutrient chemistry was compiled from the large number of stream surveys that
have been conducted in the Pacific Northwest since 1994. As part of the nutrient criteria analysis
reported in Herlihy and Sifneos (2008), a database of streamwater nutrient chemistry and related data for
720 sites in Oregon sampled prior to 2006 had already been compiled. Since that time, other survey
activities have produced data that were used to enhance the pool of available data to help complete this
task. In particular, the EPA National Rivers and Stream Assessment (NRSA) sampled 55 streams and
rivers in Oregon in 2008-2009. Other agencies in Oregon were also contacted to look for recent data that
could be useful for this project. For these datasets, most sample sites were selected using a systematic,
randomized sample providing a sound statistical basis for assuming that the data were representative of all
the streams in the study region (Herlihy et al., 2000). These data were all collected using field protocols
developed for regional EMAP surveys summarized in Peck et al. (2006). There was also a large amount
of available data on lakewater nutrient chemistry in the state of Oregon. Estimates from EPA's National
Lake Assessment (NLA) showed 694 lakes in Oregon with surface areas > 4 ha and maximum depths > 1
m. EPA's initial nutrient criteria analyses in the 1990s were compiled from available STORET data and
other datasets (e.g. Western Lake Survey and Oregon Lake Atlas) and had data on 350 lakes in Oregon
(Vaga and Herlihy, 2004). In addition to these lakes, recent water chemistry data were available for
60
-------
Methodologies for Development of Part 2: N and P in Oregon Hydrologic Landscape Regions
Numeric Nutrient Criteria for Freshwaters
Oregon from 35 lakes sampled in the NLA, 21 reservoirs (Vaga et al., 2006), and 35 lakes in the Coast
Range (Vaga et al., 2005). Other available lake data were searched for and compiled into one working
dataset. This required checking for duplicate lakes across all these data. All the data were compiled into
one comprehensive dataset with sources and extent described as above. The sufficiency of data for spatial
analysis in Oregon is reported by OHLR unit based on the spatial coverage of all these parameters.
As described, lake and stream data for Oregon were compiled from existing nutrient criteria databases and
augmented by recent NARS (National Lakes Assessment and National Rivers and Streams Assessment)
data. All told there were 366 unique lake sample sites and 796 unique stream/river sample sites (referred
to as streams). Geographic distribution of the sample sites are shown in Figure 17. Southeast Oregon is
very arid and has far fewer lakes and streams than the rest of the state which is reflected in the sample site
distribution. Lakes are concentrated in the Cascade Mountains and to a lesser extent the Blue Mountains
of Northeast Oregon and on the Pacific Coast. Streams and rivers are more evenly distributed across the
non-arid part of the state.
Three GIS spatial coverage layers were assessed to determine the sufficiency of data distributed across
Oregon for this project: Oregon Hydrologic Landscape Regions (OHLR), Soil C/N classes, and
SPARROW modeled reach nutrient concentrations. Data layers were examined for point coverage
(location of the sample point) for both lakes and streams. Stream watersheds were digitized and data
were evaluated by watershed composition. Lake watershed delineations were not available and, therefore,
watershed composition was not assessed. SPARROW results were only applicable to streams at point
locations. Thus, this assessment looked at sample coverage across the following spatial coverages:
Stream Point Coverage: OHLR, Soil C/N, SPARROW
Stream Watershed Coverage: OHLR, Soil C/N
Lake Point Coverage: OHLR, Soil C/N
61
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
Figure 17 Oregon State map showing the location of the lake (Red dots) and stream (Yellow dots)
sample sites in the compiled ALUNC data. Shaded polygons show the Oregon
Hydrologic Landscape classes.
2.2.1.1 Stream Data
There were 796 lotic sites in the data, 66% of them were small streams (defined as watershed area < 50
km2, Table 24). Small streams tended (interquartile range) to have mean wetted widths of 2-5 m and mean
thalweg depths of 13-33 cm. Beatable rivers made up 14% of the samples and they tended to have
watershed areas >500-1000 km2 with widths of 30-65 m and thalweg depths of 103-243 cm. Large
streams were intermediate between these two categories. All humid area level III ecoregions were well
represented by stream sample sites. Arid area ecoregions were more sparse (n=19 for Basin & Range,
n=25 for Columbia Plateau); a sample size that is very close to insufficient for statistical analysis, but
should be defensible (Table 25). All three Omernik nutrient ecoregions were well represented (Table 26).
62
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
Table 24 Number of samples by stream size category.
Stream Size Category
BOATABLE RIVER
LARGE STREAM
SMALL STREAM
Frequency
112
159
525
Percent
14.07
19.97
65.95
Cumulative
Frequency
112
271
796
Cumulative
Percent
14.07
34.05
100.00
Table 25 Sample size by Level III Ecoregion.
Level III Ecoregions
Level III Ecoregion
Blue Mountains
Cascades
Coast Range
Columbia Plateau
Eastern Cascades
Klamath Mts.
N. Basin & Range
Willamette Valley
Frequency
159
154
217
25
40
73
19
109
Percent
19.97
19.35
27.26
3.14
5.03
9.17
2.39
13.69
Cumulative
Frequency
159
313
530
555
595
668
687
796
Cumulative
Percent
19.97
39.32
66.58
69.72
74.75
83.92
86.31
100.00
Table 26 Sample size by stream nutrient ecoregion.
Stream Nutrient Ecoregions
Stream Nutrient
Ecoregion
W. Forested Mt.
Willamette Val.
Xeric West
Frequency
643
109
44
Percent
80.78
13.69
5.53
Cumulative
Frequency
643
752
796
Cumulative
Percent
80.78
94.47
100.00
The following three sections review stream coverage by OHLR, Soil C/N class, and SPARROW class.
63
-------
Methodologies for Development of Part 2: N and P in Oregon Hydrologic Landscape Regions
Numeric Nutrient Criteria for Freshwaters
2.2.1.1.1 OHLR coverage
Nine of the sample streams could not be assigned an OHLR class due to gaps in the GIS coverage. The
OHLR class is a combination of 5 attributes; aquifer permeability (3 subclasses, Table 27), Soil
Permeability (3 subclasses, Table 28), Climate (5 subclasses,
64
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
Table 29), Seasonality (3 subclasses, Table 30), and Terrain (3 subclasses, Table 31). Thus there were
405 potential OHLR classes. Stream sites were located in 68 of them (Table 32). However most of the
classes were represented by very few streams. Only 19 classes had more than 10 streams with VwLMM
and VwLML being the most common with 121 and 101 streams. Thus looking at specific OHLR classes
for streams will be limited to a few classes with sufficient sample size. However, each of the five
attributes that went into the OHLR has subclasses that were well represented by a sufficient number of
streams so the data can easily be analyzed in that regard. Within attributes, seasonality was well
represented by Spring and Winter subclasses but the Summer subclass was rare (n=l). All other
subclasses for the other attributes were well represented by sample streams.
Table 27 Stream samples by OHLR aquifer permeability subclass.
Stream OHLR Aquifer Permeability Subclass (Hi/Med/Lo)
H
L
M
Frequency
152
505
130
Percent
19.31
64.17
16.52
Cumulative
Frequency
152
657
787
Cumulative
Percent
19.31
83.48
100.00
Table 28 Stream samples by soil permeability subclass.
Stream OHLR Soil Permeability Subclass (Hi/Med/Lo)
H
L
M
Frequency
41
393
353
Percent
5.21
49.94
44.85
Cumulative
Frequency
41
434
787
Cumulative
Percent
5.21
55.15
100.00
65
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
Table 29 Stream samples by climate subclass
Stream OHLR Climate Subclass
D)ry
M)oist
S)emi-arid
V)ery Wet
W)et
Frequency
89
83
52
265
298
Percent
11.31
10.55
6.61
33.67
37.87
Cumulative
Frequency
89
172
224
489
787
Cumulative
Percent
11.31
21.86
28.46
62.13
100.00
** there was an Arid subclass but no sites were located in it.
Table 30 Stream samples by seasonality subclass
Stream OHLR Seasonality Subclass (Winter/Spring/sUmmer)
s
u
w
Frequency
107
1
679
Percent
13.60
0.13
86.28
Cumulative
Frequency
107
108
787
Cumulative
Percent
13.60
13.72
100.00
Table 31 Stream samples by terrain subclass.
Stream OHLR Terrain Subclass (Ml/Trans/Flat)
TERRAIN
F
M
T
Frequency
28
681
78
Percent
3.56
86.53
9.91
Cumulative
Frequency
28
709
787
Cumulative
Percent
3.56
90.09
100.00
Table 32 Stream samples by OHLR class.
OHLR Class
1
2
OHLR
VwLMM
VwLML
COUNT
121
101
PERCENT
15.3748
12.8335
66
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
OHLR Class
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
OHLR
WwLML
WwLMM
SwMML
WwHTL
DwMML
WwHFL
DwLMM
DwMMM
WwLTL
MsMMM
MsLMM
MwLMM
SwLML
VsHMH
VsHMM
DwLML
VwHMM
WsMML
MsLML
MwHMH
WsHMH
WsLML
DwHML
MwMMM
WsMMM
WwHFM
WwMTL
SwHTL
COUNT
89
80
26
26
21
21
18
18
17
15
14
14
12
12
11
10
10
9
9
7
7
7
7
6
6
6
6
6
5
PERCENT
11.3088
10.1652
3.3037
3.3037
2.6684
2.6684
2.2872
2.2872
2.1601
1.9060
1.7789
1.7789
1.5248
1.5248
1.3977
1.2706
1.2706
1.1436
0.8895
0.8895
0.8895
0.8895
0.7624
0.7624
0.7624
0.7624
0.7624
0.6353
67
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
OHLR Class
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
OHLR
WwMMM
DwHMM
DwHTL
MsHMH
MsMML
WwHML
WwMML
DwHTM
MwHTH
VsLML
VwMMM
WsHMM
DwHMH
MwHMM
MwLML
MwMML
SwHMM
SwHTM
SwMTL
VsLMM
WwLTM
DwHTH
DwLTL
DwMTM
MsHTH
MwHTM
MwMMH
SwHFL
SwHML
COUNT
5
4
4
4
4
4
4
3
3
3
3
3
2
2
2
2
2
2
2
2
2
1
1
1
1
1
1
1
1
PERCENT
0.6353
0.5083
0.5083
0.5083
0.5083
0.5083
0.5083
0.3812
0.3812
0.3812
0.3812
0.3812
0.2541
0.2541
0.2541
0.2541
0.2541
0.2541
0.2541
0.2541
0.2541
0.1271
0.1271
0.1271
0.1271
0.1271
0.1271
0.1271
0.1271
68
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
OHLR Class
61
62
63
64
65
66
67
68
69
OHLR
SwLMM
VwLTL
VwMTM
WsHTH
WsLMM
WuLML
WwHMH
WwHMM
WwLMH
COUNT
1
1
1
1
1
1
1
1
1
PERCENT
0.1271
0.1271
0.1271
0.1271
0.1271
0.1271
0.1271
0.1271
0.1271
2.2.1.1.2 SOIL C/N coverage
Five stream sites were outside the soil GIS layer and are missing data. Soil nutrient GIS data report total
C and N concentrations in 4 soil depth categories (0-50 cm, 0-100 cm, 0-150 cm, and 0-200 cm). Sample
data analysis boundaries (10th-90th percentiles) for streams are between -10-40 kg/m2 for total C and -0.5-
2.0 kg/m2 for total N (Figure 18). Data for the 0-150 cm and 0-200 cm class are virtually identical,
therefore these classes were combined for subsequent analysis.
ALUNC Stream Data
ALUNC Stream Data
1.0
0.8
a
9
Q_
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
2.2.1.1.3 SPARROW coverage
SPARROW model results were reach scale based on the 1:500,000 scale Enhanced River Reach File 2
(ERF 1-2) stream frame. ERF 1-2 is a somewhat coarse frame and did not contain all streams, especially
small ones. Thus many of the sample streams were not modeled by SPARROW. A GIS analysis was
conducted associating all 796 of the stream points to the closest SPARROW reach. For each site there is
a name of the closest SPARROW reach and the distance to that reach. Many of the small streams clipped
to larger river names had distances > 500 m. An initial impression was that sites < 500 m away were
likely to be "matches" in that the selected site was referring to the same reach as modeled by SPARROW.
Large streams/rivers (watershed areas > 500 km2) tended to be within 500 m and were likely matches
(Figure 19). Many of the small streams however were more than 1000 m away and unlikely to be
matches. It should be feasible to identify strong matches for the entire dataset, which will require manual
matching. However, the analysis of matches will be heavily weighted towards larger systems as those
systems were what were modeled by SPARROW.
3000(7
1.00 10.00 100.00 1000.00 10000.00 100000.0 1000000
Watershed Area (km2)
Figure 19 Scatterplot of distance from sample point to nearest SPARROW reach for each of the
796 sites in the compiled stream database versus site watershed area.
70
-------
Methodologies for Development of Part 2: N and P in Oregon Hydrologic Landscape Regions
Numeric Nutrient Criteria for Freshwaters
Watershed areas were also calculated for each of the sample streams for each of the OHLR classes and
soil C/N nutrient classes for comparison of size. Of the 796 streams, 793 had watershed delineations and
were used in this analysis. The results indicated that, for the soil nutrient data, calculating a watershed
area weighted mean soil nutrient concentrations from this coverage would be possible. For small lake
watersheds there should be little if any difference. For larger watersheds, 1:1 plots can be evaluated and it
can be decided if there's a significant enough difference to warrant separate analyses.
For categorical data like OHLR, areal weighted measures were more complex. The data can be evaluated
and analyses conducted using only sites that have watersheds dominated by single classes or single
attribute classes. For example, dominance by >90% of the watershed by a single class could be used.
Using this criterion for the Stream Aquifer Permeability attribute of the OHLR, the sample site
breakdown is shown in Table 33:
Table 33 Distribution of sample stream watersheds by soil permeability class.
Stream Aquifer Permeability Subclass # of Sites Percentage
Watershed dominated by High ... -0/
Permeability
Watershed dominated by Medium _ . -0,
Permeability
Watershed dominated by Low .„,. ,„„,
Permeability
Watershed with no Dominance (Mixed) 91 11%
Figure 20, below, shows clearly that sites with no Dominance (Mixed) had orders of magnitude larger
watershed areas than sites dominated by a single aquifer permeability subclass. Sites with no dominance
using the 90% criterion typically had watersheds > 1,000 km2.
Many of the site watersheds were dominated by one class because so much of the data consisted of small
streams. These resulted indicate that the resultant analysis will work much better for the individual
OHLR attribute classes than for the overall OHLR class. Looking at sites with only >90% of the
watershed in one OHLR class will just reduce the number of sites in the point allocation table.
71
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
1000000
100000
10000
1000
100
10
HIGH-clom LOW-dom MED-dom MIXED
Stream Aquifer Permeability Class - >90% Watershed Area
Figure 20 Box and whisker plot of stream watershed area by aquifer permeability subclass
classified based on watershed dominance (>90% of watershed area in subclass). Mixed sites are
those with no dominant subclass.
As for streams, lake samples were also evaluated by size and for distributions among OHLR and Soil C/N
class. The 366 sample lakes were fairly evenly distributed across lake area size classes (
72
-------
Methodologies for Development of Part 2: N and P in Oregon Hydrologic Landscape Regions
Numeric Nutrient Criteria for Freshwaters
Table 34). A decision will have to be made about including the very small and very large lakes in the
analysis. Lakes < 1 ha were not part of the NARS target population and were often unrepresented on the
NHD hydrologic GIS data layer. Lakes < 4 ha were not considered targets for the original nutrient
criteria work. The large lakes > 500 ha were mostly run of river reservoirs (e.g. Columbia River and other
large river impoundments). They represent such large watersheds that it would be impossible to assign
their condition to any particular OHLR or SOIL C/N class.
Half of the lakes (51%) were in the Cascade mountain ecoregion (Table 35). Another 25% of the lakes
were in either the Blue Mountains or Coast Range. These three ecoregions were all considered to be in
the Western Forested Mountains Omernik nutrient ecoregion (
73
-------
Methodologies for Development of Part 2: N and P in Oregon Hydrologic Landscape Regions
Numeric Nutrient Criteria for Freshwaters
Table 36). All told 89% of the lakes were in this nutrient ecoregion. It would be very difficult to assess
condition in the Willamette Valley nutrient ecoregion (n=13) and to a lesser extent the Xeric West
nutrient ecoregion (n=28).
74
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
Table 34 Distribution of lake samples by lake area class.
Lake Area Class (ha)
<1
A: 1-4
B:4-10
C: 10-50
D:50-100
E: 100-500
F:500+
Frequency
25
61
48
87
34
60
51
Percent
6.83
16.67
13.11
23.77
9.29
16.39
13.93
Cumulative
Frequency
25
86
134
221
255
315
366
Cumulative
Percent
6.83
23.50
36.61
60.38
69.67
86.07
100.00
Table 35 Distribution of lake samples by ecoregion.
Lake Level III Ecoregion Name
Blue Mountains
Cascades
Coast Range
Columbia Plateau
Eastern Cascades
Klamath Mountains
Northern Basin and Range
Snake River Plain
Willamette Valley
Frequency
49
188
45
5
34
9
22
1
13
Percent
13.39
51.37
12.30
1.37
9.29
2.46
6.01
0.27
3.55
Cumulative
Frequency
49
237
282
287
321
330
352
353
366
Cumulative
Percent
13.39
64.75
77.05
78.42
87.70
90.16
96.17
96.45
100.00
75
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
Table 36 Distribution of lake samples by nutrient ecoregion.
Lake Omernik Nutrient Ecoregion Name
W. Forested Mt.
Willamette Val.
Xeric West
Frequency
325
13
28
Percent
88.80
3.55
7.65
Cumulative
Frequency
325
338
366
Cumulative
Percent
88.80
92.35
100.00
As for streams, the following two sections review lake coverage by OHLR and Soil C/N class.
2.2.1.3.1 Lake OHLR Coverage
Sixteen of the sample lakes could not be assigned an OHLR class due to being located outside
the GIS coverage. They were mostly lakes adjacent to the Pacific Ocean. As stated earlier, the
OHLR class was a combination of 5 attributes; aquifer permeability (3 subclasses,
Table 37), Soil Permeability (3 subclasses,
76
-------
Methodologies for Development of Part 2: N and P in Oregon Hydrologic Landscape Regions
Numeric Nutrient Criteria for Freshwaters
Table 38), Climate (5 subclasses, Table 39), Seasonality (3 subclasses, Table 40), and Terrain (3
subclasses,
77
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
Table 41). Thus there are 405 potential OHLR classes. Lake sites were located in 74 of them (Table 42).
However the vast majority of the classes were represented by very few lakes. Only 4 classes had more
than 10 lakes with VsHMH being the most common with 63 lakes (VsHMM, VwLMM, and WsHMH
being the other 3 classes). Thus looking at specific OHLR classes for lakes is not going to be practical
due to sample size issues. However, each of the five attributes that went into the OHLR had subclasses
that were well represented by a sufficient number of lakes so the data can easily be analyzed in that
regard. Within attributes, seasonality was well represented by Spring and Winter subclasses but the
Summer subclass was rare (n=7). Similarly, for the Terrain attribute, Mountainous and Transitional
subclasses were well represented but the Flat subclass was relatively rare (n=l 1). All other subclasses for
the other attributes were fairly well represented by sample lakes.
Table 37 Distribution of lake samples by aquifer permeability subclass.
Lake OHLR Aquifer Permeability Subclass (Hi/Med/Lo)
H
L
M
Frequency
214
107
29
Percent
61.14
30.57
8.29
Cumulative
Frequency
214
321
350
Cumulative
Percent
61.14
91.71
100.00
78
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
Table 38 Distribution of lake samples by soil permeability subclass.
Lake OHLR Soil Permeability Subclass (Hi/Med/Lo)
H
L
M
Frequency
114
99
137
Percent
32.57
28.29
39.14
Cumulative
Frequency
114
213
350
Cumulative
Percent
32.57
60.86
100.00
Table 39 Distribution of lake samples by climate subclass.
Lake OHLR Climate Subclass
D)ry
M)oist
S)emi-arid
V)ery wet
W)et
Frequency
28
23
37
179
83
Percent
8.00
6.57
10.57
51.14
23.71
Cumulative
Frequency
28
51
88
267
350
Cumulative
Percent
8.00
14.57
25.14
76.29
100.00
** there was an Arid subclass but no sites were located in it.
Table 40 Distribution of lake samples by seasonality subclass.
Lake OFILR Seasonality Subclass (Winter/Spring/sUmmer)
s
u
w
Frequency
175
7
168
Percent
50.00
2.00
48.00
Cumulative
Frequency
175
182
350
Cumulative
Percent
50.00
52.00
100.00
79
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
Table 41 Distribution of lake samples by terrain subclass.
Lake OHLR Terrain Subclass (Mt/Trans/Flat)
F
M
T
Frequency
11
252
87
Percent
3.14
72.00
24.86
Cumulative
Frequency
11
263
350
Cumulative
Percent
3.14
75.14
100.00
Table 42 Distribution of lake samples by OHLR subclass.
OHLR Subclass
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
OHLR
VsHMH
VsHMM
VwLMM
WsHMH
VwLTH
SwHTL
SwHTM
WwLML
WwLMM
VsLMM
VwLTL
WsHTH
WwLTL
DwHTM
VuLML
VwHMM
VwLML
WsHMM
DwMML
COUNT
63
37
24
16
16
9
8
8
8
8
7
7
7
7
6
6
5
5
5
4
PERCENT
18.0000
10.5714
6.8571
4.5714
2.5714
2.2857
2.2857
2.2857
2.2857
2.0000
2.0000
2.0000
2.0000
1.7143
1.7143
1.4286
1.4286
1.4286
1.1429
80
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
OHLR Subclass
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
OHLR
MwLMM
SwHML
VsHTH
VwHTH
WsHTL
WsLML
WwHTL
DwHMM
DwHTL
MwHMM
SwHFL
SwHTH
SwMML
VsLML
VsMMM
WsMML
WsMMM
WwHTM
DsMMM
DwHML
DwLMM
DwMMM
MsLMM
MsMML
MwHFH
SwLML
SwMTL
WsLMM
WwHML
COUNT
4
4
4
4
4
4
4
3
3
3
3
3
3
3
3
3
3
3
2
2
2
2
2
2
2
2
2
2
2
PERCENT
1.1429
1.1429
1.1429
1.1429
1.1429
1.1429
1.1429
0.8571
0.8571
0.8571
0.8571
0.8571
0.8571
0.8571
0.8571
0.8571
0.8571
0.8571
0.5714
0.5714
0.5714
0.5714
0.5714
0.5714
0.5714
0.5714
0.5714
0.5714
0.5714
81
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
OHLR Subclass
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
OHLR
DwHMH
DwLML
DwLTL
DwMTL
MsHMH
MsHMM
MsLML
MsLTL
MsMMM
MwHFL
MwHTH
MwHTM
MwLML
MwMMM
SwHFH
SwHFM
SwLTL
SwMMM
VsHFH
VsHML
WsHML
WuLML
WwHFL
WwHFM
WwHMM
WwHTH
WwMTL
COUNT
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
PERCENT
0.2857
0.2857
0.2857
0.2857
0.2857
0.2857
0.2857
0.2857
0.2857
0.2857
0.2857
0.2857
0.2857
0.2857
0.2857
0.2857
0.2857
0.2857
0.2857
0.2857
0.2857
0.2857
0.2857
0.2857
0.2857
0.2857
0.2857
82
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
2.2.1.3.2 Lake Soil C/N coverage
Forty-one (41) of the sample lakes were outside the soil nutrient GIS layer, mostly due to the fact that
large lakes were blank in the GIS coverage. Soil nutrient GIS data report total C and N concentrations in
4 soil depth categories (0-50 cm, 0-100 cm, 0-150 cm, and 0-200 cm). Sample data analysis boundaries
(10th-90th percentiles) for lakes were between 10-40 kg/m2 for total C and 0.5-2.5 kg/m2 for total N
(Figure 21). Data for the 0-150 cm and 0-200 cm class were virtually identical and there is, therefore, no
need to analyze them separately in the future.
ALUNC Lake Data
1.0
0.0
Soil Depth
0-050 cm
— 0-100 cm
— 0-150 cm
— 0-200 cm
o
0.0 10.0 20.0 30.0 40.0
Soil Total C (kg/m2)
50.0
ALUNC Lake Data
0-050 cm
— 0-100 cm
0-150 cm
0-200 cm
1.0 1.5 2.0
Soil Total N (kg/m2)
Figure 21 Cumulative distribution functions of soil Total C and N for the compiled lake sample
data.
2.2.2 Summary
In summary, with regards to the spatial coverage of stream and lake samples within different categories,
the following was observed:
2.2.2.1 Streams
• There were 796 unique stream sample sites;
• Stream samples were mostly from small streams, although beatable rivers made up 14% of the
samples and intermediate size streams 20% of samples;
• All Omernik Level III ecoregions were represented by at least 19 samples and most had from 25-
217. The Coast Range was the most numerous (N=217);
• All nutrient ecoregions were represented by samples, with a minimum N of 44;
• The OHLR was derived from combining attributes of 5 component subclasses and has 405
possible classes. Of those, probably only 100 or so actually occur. Most OHLR classes have too
83
-------
Methodologies for Development of Part 2: N and P in Oregon Hydrologic Landscape Regions
Numeric Nutrient Criteria for Freshwaters
few sites to estimate percentiles or analyze statistically by themselves. Only 8 OHLR classes
have a sample size greater than 20 sites. The five component subclasses (aquifer permeability,
soil permeability, climate, seasonality, and terrain), however, were all adequately represented by
sufficient sample sizes with the exception of the summer seasonality subclass;
• Cumulative sample proportions in different Soil C/N types indicate deciles between -10-40 kg/m2
for total C and -0.5-2.0 kg/m2 for total N, covering a wide range. Data for the 0-150 cm and 0-
200 cm class are virtually identical and there is no need to analyze them separately in the future;
• Many stream sample sites were not modeled by SPARROW, since it is weighted towards larger
systems. It should, however, be feasible to identify strong matches for the entire dataset, which
will require manual matching;
• Many of the site watersheds were dominated by one OHLR category class because so much of the
data consisted of small streams, supporting analysis of individual OHLR attribute classes rather
than the overall OHLR class. Larger watersheds are generally made up of a mixture of multiple
OHLR classes and it may be useful to restrict the analyses to smaller watersheds (< 100-5 00 km2
watershed area);
Lakes
There were 366 lake sample sites;
Lakes samples were well distributed among area classes from 500ha;
Omernik Level III ecoregions were well represented except for Columbia Plateau, Klamath
Mountains, Snake River Plain, and Willamette Valley, which all had fewer than 13 samples, but
these ecoregions contain relatively few lakes;
The Willamette Valley was the only nutrient ecoregion with a small sample size (N=13), the
other two had sufficient sample size. The Western Forested Mountains dominated the sample
distribution (89%);
As with streams, most OHLR classes contained too few sites to estimate percentiles or
statistically analyze. Only 4 OHLR classes contained more than 10 sample sites. The five
component subclasses (aquifer permeability, soil permeability, climate, seasonality, and terrain),
however, were all adequately represented by sufficient sample sizes with the exception of the
summer seasonality subclass and, perhaps, the flat terrain subclass;
As with streams, cumulative sample proportions in different Soil C/N types indicate deciles
between -10-40 kg/m2 for total C and -0.5-2.5 kg/m2 for total N, covering a wide range. Data for
the 0-150 cm and 0-200 cm class are virtually identical and were combined.
84
-------
Methodologies for Development of Part 2: N and P in Oregon Hydrologic Landscape Regions
Numeric Nutrient Criteria for Freshwaters
This section describes the preparation and analysis of data available to complete analyses of reference
based nutrient criteria for the Oregon Hydrologic Landscape Regions (OHLR). The focus of this section
was assembling data on streams and lakes across these regions and describing nutrient distributions in
entire and reference populations. This involved data preparation and analysis of the OHLR regions with
regard to stream and lake nutrient data, soil GIS data, and SPARROW stream data across those regions.
Lake and stream data were combined as described in Section 2.2.2. Further investigation showed that 24
of the stream sites were on the same stream within 1 km of another site. It was decided that those sites
were likely duplicates and the site with the older data was dropped. Similarly, 3 duplicate lake sites were
found and dropped. As a reminder, geographic distribution of the sample sites are shown in Figure 17.
Three GIS spatial coverage layers were analyzed across Oregon for this project: Oregon Hydrologic
Landscape Regions (OHLR), Soil C/N classes, and SPARROW modeled reach nutrient concentrations.
GIS coverages were obtained from EPA personnel at the Corvallis EPA Laboratory. Digitized stream
watersheds were available for all sites and OHLR and soil C/N data were evaluated by watershed
composition as described below. As stream watersheds get bigger and bigger, they start encompassing
many different OHLR classes which defeated the purpose of an analysis by OHLR class. Streams with
watershed areas < 1000 km2 were mostly dominated by one OHLR class (>90% of watershed area was in
one class). Streams with larger areas often had multiple OHLR classes in their watersheds and were not
dominated by any one class. Therefore, the analysis of the OHLR class data was restricted to streams that
had watershed area < 1000 km2 which dropped 85 sites from the analysis. In addition, 21 smaller stream
sites that were not dominated (>50%) by any one OHLR class and three sites that were missing OHLR
coverage were dropped. The remaining 663 stream sites were each classified into one OHLR class based
on the dominant (by areal proportion) OHLR class in their watershed. For the soil C/N data, watershed
area weighted mean soil C and N were calculated for each site. SPARROW results were only applicable
to stream reaches at sample point locations.
Lake watershed delineations were not available so watershed composition could not be assessed. OHLR
and soil C/N classes were determined by the point location of the lake. As with streams, it doesn't make
sense to apply the OHLR class analysis to large lakes that represent water flowing through many OHLR
classes. An extreme case would be the impoundments on the Columbia and Snake Rivers which are in
the lake database. So, lakes from the analysis that were > 250 ha in lake area were eliminated. That
removed all the reservoirs known to be an impoundment of a major river. Twenty-five lakes < 1 ha in
lake area, which was below the target size used in the EPA national lake surveys, were also found; so,
those were dropped as well. The final lake analysis dataset had 264 lakes. SPARROW results were only
applicable to streams and not relevant to lakes.
The OHLR categorization is based on a combination of 5 classes (Figure 22); aquifer permeability (3
subclasses), Soil Permeability (3 subclasses), Climate (5 subclasses), Seasonality (3 subclasses), and
Terrain (3 subclasses). Thus there were 405 potential OHLR classes although many combinations
actually did not exist. Stream sites were located in 56 different OHLR classes and lakes were located in
62 of them. However most of the classes were represented by very few sample sites. Only four classes
had more than 10 stream reference sites (VwLMM, VwLML, WwLMM, WwLML) and only two classes
had more than 10 lake reference sites (VsHMM, VsHMH). Thus analyzing results by specific OHLR
classes was not done. However, each of the five attributes that went into the OHLR has classes that were
85
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
well represented by a sufficient number of stream and lake samples so the data could easily be analyzed in
that regard. Within attributes, seasonality was well represented by Spring and Winter classes but the
Summer class was very rare. Similarly, within the terrain class, there were few flat reference sites and
within the climate class, there were no arid sites. All other classes for the other attributes had reasonable
representation by sample lakes and streams. Maps of the distribution of each class across Oregon are
given in Appendix 24. Three stream sites and 12 lake sites were missing OHLR data, mostly right on the
Coast where the OHLR was not defined.
CsATS
Climate Class Seasonality Sub-Class Permeability Class Terrain Class
V - Very wet
W-Wet
M - Moist
D-Dry
S - Semiarid
A-Arid
w - Fall or winter
s - Spring
u - Summer
H - High
M - Moderate
L- Low
M - Mountain
T - Transitional
F - Flat
Soil
Permeability Class
H - High
M - Moderate
L- Low
Figure 22 Five attributes and their subclasses that make up the OHLR classification
Four stream and 13 lake locations were missing on the soil GIS layer and were thus missing data. Soil
nutrient GIS data reported total C and N concentrations in 4 soil depth categories (0-50 cm, 0-100 cm, 0-
150 cm, and 0-200 cm). Sample data analysis boundaries (10th-90th percentiles) for streams were between
-10-40 kg/m2 for total C and -0.5-2.0 kg/m2 for total N. Concentrations in the four depth categories were
highly correlated with each other (Table 43 and Table 44). For example, TN concentrations in the 0-150
cm and 0-200 cm categories were virtually identical (r=0.997) for streams. Thus only the mapped soil
data for the 0-50 cm and 0-200 cm depth categories were used as they were the least correlated. Maps of
the total C and N data across Oregon are provided in Appendix 24.
Table 43 Pearson correlations between mapped soil Total Nitrogen concentrations in the 4 soil
depth (cm) categories in the 684 stream watersheds.
Soil Depth
0-50
0-100
0-150
0-200
0-50
0-100
0.946
0-150
0.840
0.949
0-200
0.824
0.938
0.997
Table 44 Pearson correlations between mapped soil Total Carbon concentrations in the 4 soil
depth (cm) categories in the 684 stream watersheds.
86
-------
Methodologies for Development of Part 2: N and P in Oregon Hydrologic Landscape Regions
Numeric Nutrient Criteria for Freshwaters
Soil Depth 0-50 0-100 0-150 0-200
0-50
0-100 0.938
0-150 0.807 0.943
0-200 0.791 0.930 0.998
SPARROW model results were generated at the reach scale based on the 1:500,000 scale Enhanced River
Reach File 2 (ERF1-2) stream frame. ERF1-2 is a somewhat coarse frame and does not contain all
streams, especially small ones. Thus many of the sample streams in the compiled data were not modeled
by SPARROW. A GIS analysis was conducted associating all 772 of the stream points to the closest
SPARROW reach. For each site there was a name of the closest SPARROW reach and the distance to
that reach. Sites were matched based on name, stream size, and distance apart. Sites with the same
stream name and within 500 m were assumed to be on the same reach. Streams that were small with the
nearest reach being a large river were assumed to be not on the same reach (e.g., Tributaries not on ERF1-
2). The remaining sites were examined individually to see whether they were from the same reach or not.
Any site over 1 km from a SPARROW reach was assumed to be not matched regardless of whether it was
on the same stream or not. In the end, about half of the sample sites (n=413) were matched to modeled
SPARROW reaches. The median of these matched sites was 276 m from the SPARROW reach
(IQR= 147-452 m).
In order to minimize the influence of human nutrient additions on the analyses, a set of least disturbed
reference sites were identified for this analysis. For streams, reference sites were identified using site
measurements of non-nutrient water chemistry and physical habitat to filter out impacted sites as
described in detail in Herlihy and Sifneos (2008). Screening filter variables included measures of sulfate,
chloride, pH, riparian disturbance, bank canopy density, and % fine sediment. Disturbance thresholds
varied by ecoregion. Overall, there were 152 streams considered to be in reference or least-disturbed
condition. Stream reference condition was also defined based on biological condition using
macroinvertebrate data. Stream EPT richness (number of different mayfly, stonefly, and caddisfly taxa)
was used as the screening metric as described in Herlihy and Sifneos (2008). Sites with EPT richness >
10 in the Willamette Valley nutrient ecoregion were considered reference by biology. Similarly, in the
Xeric nutrient ecoregion EPT richness > 13 was considered reference and EPT richness > 17 was
considered reference in the Western Forested Mountains. Overall, there were 322 sites considered to be
reference by biological condition. Note that the biological reference sites were not used for the
SPARROW analysis.
Lake data generally lacked the associated measurements that existed for streams, so could not be screened
in the same way. Instead, a semi-quantitative aerial photo screening method was used based on current
Google Earth images of each sample lake. Lakes were scored on an integer scale of 0 (undisturbed) to 3
based on degree of disturbance in each of seven categories (agricultural, residential, recreation, logging,
roads, mining, and commercial) as described in Herlihy et al. (2013). The seven 0-3 disturbance scores
87~
-------
Methodologies for Development of Part 2: N and P in Oregon Hydrologic Landscape Regions
Numeric Nutrient Criteria for Freshwaters
were summed into one overall disturbance index score that could range from 0-21. The definition of
"least-disturbed" in Oregon did vary by ecoregion. It was very hard to find undisturbed lakes in some
ecoregions. Lakes in the Coast Range were considered reference with an overall score of 0, 1 or 2. Lakes
in the Cascades and Blue Mountain ecoregions, however, were considered reference only if they had an
overall disturbance index of zero. In all other ecoregions, an overall score of 0 or 1 was considered
reference. Overall, 113 of the 264 lakes in the final analysis set were considered reference.
Descriptive statistics showing the mean, SD and range in nutrient concentration by all the OHLR classes
for chemistry/habitat screened reference streams were generated (
88
-------
Methodologies for Development of Part 2: N and P in Oregon Hydrologic Landscape Regions
Numeric Nutrient Criteria for Freshwaters
Table 45). The same information for soil N data was also generated (
89
-------
Methodologies for Development of Part 2: N and P in Oregon Hydrologic Landscape Regions
Numeric Nutrient Criteria for Freshwaters
Table 46). The information was also generated graphically as box and whisker plots for TP (Figure 23),
TN (Figure 24), and Soil N in the 0-200 cm depth layer (Figure 25). The box and whisker plots show
results for all sites in blue and only reference sites in red.
90
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
Table 45 Descriptive statistics forTP (ug/L) and TN (ug/L) by OHLR subclasses in Reference
Streams in Oregon.
OHLR Subclass
Climate
Very Wet
Wet
Moist
Dry
Semiarid
Seasonality
Spring
Summer
Winter
Aquifer Perm
High
Medium
Low
Terrain
Flat
Transitional
Mountain
Soil Perm
High
Medium
Low
TP(N)
79
47
12
7
1
26
1
119
21
19
106
1
4
141
7
81
58
TP Mean TP SD
16.3 15.0
27.8 37.5
35.5 30.6
42.3 32.3
52
25.7 24.6
2
22.7 28.1
34.1 53.4
34.5 24.0
18.8 18.1
16
85.3 115
21.3 19.5
18.0 14.7
23.9 20.1
22.5 36.2
TP Range
2-100
1-256
5-117
7-86
—
2-117
—
1-256
2-256
5-86
1-117
—
13-256
3-117
2-40
3-117
1-256
TN(N)
79
49
12
7
1
28
1
119
22
20
106
1
4
143
8
82
58
TNMean
170
191
134
178
334
115
29
191
222
159
169
94
782
159
129
135
239
TNSD TN Range
141 20-750
296 20-2080
60.8 55-250
152 47-500
—
81.0 20-310
—
219 20-2080
425 20-2080
130 33-500
133 20-750
—
900 47-2080
121 20-750
96.4 20-310
92.4 20-520
292 20-2080
91
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
Table 46 Descriptive statistics for Soil N (kg/m2) concentrations by OHLR subclasses in Reference
Streams in Oregon.
OHLR Class
Climate
Very Wet
Wet
Moist
Dry
Semiarid
Seasonality
Spring
Summer
Winter
Aquifer Perm
High
Medium
Low
Terrain
Flat
Transitional
Mountain
Soil Perm
High
Medium
Low
SoilN
0-50 cm
N
78
49
12
7
1
27
1
119
22
20
105
1
4
142
8
82
57
Soil N Soil N
0-50 cm 0-50
Mean cm SD
1.00 0.233
0.783 0.258
0.496 0.111
0.436 0.088
0.479
0.660 0.248
0.437
0.905 0.281
0.798 0.292
0.518 0.135
0.934 0.265
0.800
0.803 0.463
0.859 0.289
0.762 0.320
0.836 0.279
0.901 0.304
SoilN
0-50 cm
Range
0.364-1.46
0.254-1.12
0.254-0.635
0.303-0.540
—
0.364-1.07
—
0.254-1.46
0.303-1.28
0.254-0.895
0.254-1.46
—
0.479-1.46
0.254-1.46
0.364-1.03
0.254-1.28
0.303-1.46
SoilN
0-200 cm
N
78
49
12
7
1
27
1
119
22
20
105
1
4
142
8
82
57
Soil N Soil N
0-200 cm 0-200 cm
Mean SD
1.57 0.411
1.40 0.516
0.894 0.276
0.700 0.228
0.708
1.15 0.486
0.730
1.47 0.480
1.39 0.515
0.886 0.309
1.51 0.461
1.69
1.39 0.813
1.41 0.491
1.43 0.654
1.31 0.413
1.56 0.555
SoilN
0-200 cm
Range
0.645-2.86
0.374-2.86
0.374-1.50
0.468-0.980
—
0.597-2.02
—
0.374-2.86
0.480-2.02
0.374-1.77
0.374-2.86
—
0.708-2.33
0.374-2.86
0.645-2.02
0.374-2.02
0.468-2.86
92
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
Oregon Stream Data
10000
1000
i
^ 100
Q_
10
1
All Data
-Reference Sites
1:High 2:Medium
OHLR Aquifer Permeability
3: Low
Oregon Stream Data
10000
1000
S
3, 100
10
All Data
Reference Sites
1:High 2:Medium
OHLR Soil Permeability
3:Low
93
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
Oregon Stream Data
10000
1000
i
^ 100
Q_
10
1
All Data
-Reference Sites
1:Very Wet 2:Wet 3:Moist 4:Dry 5:Semiarid
OHLR Climate
Oregon Stream Data
0000
1000
100
10
1
All Data
-Reference Sites
1
T
1:Mt 2:Trans 3:Fla
OHLR Terrain
94
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
Oregon Stream Data
10000
1000
i
^ 100
Q_
10
All Data
-Reference Sites
Spring Summer
OHLR Seasonally
Winter
Figure 23 Stream data box and whisker plots of streamwater Total Phosphorus (TP) by OHLR
subclasses for chemistry/habitat screened reference sites (red) and all data (blue). Boxes
represent the 25th and 75th percentiles, the line in the box the median and the whiskers the range.
95
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
Oregon Stream Data
100000
10000
1000
100
10
All Data
-Reference Sites
1:High 2:Medium 3:Low
OHLR Aquifer Permeability
Oregon Stream Data
100000
10000
ol
•3 1000
100
10
1:High
All Data
Reference Sites
2: Medium
OHLR Soil Permeability
3:Low
96
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
Oregon Stream Data
100000
10000
a 1000
100
10
6
a
i
: g
B
All Data
— Reference Sites
E
3
1
T
1:Very Wet 2:Wet 3:Moist 4:Dry 5:Semiarid
OHLR Climate
Oregon Stream Data
100000
10000
"a
3, 1000
100
10
B
fl f
All Data
— Reference Sites
—
--
1:Mt 2:Trans 3:Flal
OHLR Terrain
97
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
Oregon Stream Data
100000
10000
1000
100
10
All Data
Reference Sites
Spring
Summer
OHLR Seasonally
Winter
Figure 24 Stream data box and whisker plots of streamwater Total Nitrogen (TN) by OHLR
subclasses for chemistry/habitat screened reference sites (red) and all data (blue). Boxes
represent the 25th and 75th percentiles, the line in the box the median and the whiskers the range.
98
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
„
I3
o
I 1
o
I3
o
Oregon Stream Data
All Data
-Reference Sites
1:High 2:Medium
OHLR Aquifer Permeability
Oregon Stream Data
3: Low
All Data
Reference Sites
High 2:Medium
OHLR Soil Permeability
3: Low
99
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
„
I3
o
I 1
o
Oregon Stream Data
All Data
-Reference Sites
1:Very Wet 2:Wet 3:Moist 4:Dry 5:Semiarid
OHLR Climate
CM
E 3
o
I 1
o
1:Mt
Oregon Stream Data
All Data
Reference Sites
2:Trans
OHLR Terrain
3: Flat
100
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
Oregon Stream Data
„
I3
o
I 1
o
All Data
-Reference Sites
Spring
Summer
OHLR Seasonally
Winter
Figure 25 Stream data box and whisker plots of 0-200 cm depth Soil Nitrogen concentration by
OHLR subclasses for chemistry/habitat screened reference sites (red) and all data (blue). Boxes
represent the 25th and 75th percentiles, the line in the box the median and the whiskers the range.
2.3.2.2 Lakes
Descriptive statistics showing the mean, SD and range in nutrient concentration by all the OHLR classes
for chemistry/habitat screened reference lakes were generated (
101
-------
Methodologies for Development of Part 2: N and P in Oregon Hydrologic Landscape Regions
Numeric Nutrient Criteria for Freshwaters
Table 47). The same information for soil N data was also generated (
102
-------
Methodologies for Development of Part 2: N and P in Oregon Hydrologic Landscape Regions
Numeric Nutrient Criteria for Freshwaters
Table 48). The information was generated graphically as box and whisker plots for TP (Figure 26), TN
(Figure 27), and Soil N in the 0-200 cm depth layer (Figure 28). As in the stream box and whisker plots,
results for all sites are shown in blue and only reference sites are shown in red.
103
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
Table 47 Descriptive statistics forTP (ug/L) and TN (ug/L) by OHLR subclasses in Reference
Lakes in Oregon.
OHLR Subclass
Climate
Very Wet
Wet
Moist
Dry
Semiarid
Seasonality
Spring
Summer
Winter
Aquifer Perm
High
Medium
Low
Terrain
Flat
Transitional
Mountain
Soil Perm
High
Medium
Low
TP(N)
76
23
o
3
6
3
77
7
27
72
7
32
0
22
89
40
45
26
TP Mean
12.3
13.4
37.3
201
417
12.7
8.86
103
42.9
28.2
16.5
—
117
13.8
14.4
51.0
36.4
TPSD
19.5
21.9
34.0
359
193
20.8
6.90
216
139
27.7
24.6
—
237
21.9
21.3
165
82.4
TP Range
0-105
0-103
4-72
23-933
271-636
0-105
2.4-18
0-933
0-933
5-72
0-105
—
1-933
0-105
0-95
0-933
1-344
TN(N)
19
12
3
6
3
22
4
17
27
2
14
0
15
28
16
13
14
TNMean
267
276
331
1230
1190
291
207
772
573
872
225
—
843
275
309
716
435
TNSD
129
128
157
721
705
127
93.5
688
571
634
108
—
701
127
150
731
438
TN Range
68-730
55-557
150-423
540-2230
385-1670
55-730
68-270
110-2230
228-2230
423-1320
55-421
—
110-2230
55-730
110-730
140-2230
55-1530
104
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
Table 48 Descriptive statistics for Soil N (kg/m2) concentrations by OHLR subclasses in Reference
Lakes in Oregon.
OHLR Class
Climate
Very Wet
Wet
Moist
Dry
Semiarid
Seasonality
Spring
Summer
Winter
Aquifer Perm
High
Medium
Low
Terrain
Flat
Transitional
Mountain
Soil Perm
High
Medium
Low
SoilN
0-50 cm
N
68
19
3
6
3
69
3
27
68
6
25
0
22
77
37
45
17
SoilN
0-50 cm
Mean
0.801
0.814
0.427
0.446
0.559
0.744
0.437
0.850
0.753
0.574
0.837
—
0.836
0.743
0.820
0.753
0.668
SoilN
0-50
cmSD
0.270
0.285
0.061
0.119
0.313
0.219
0
0.394
0.247
0.251
0.362
—
0.415
0.232
0.302
0.248
0.313
SoilN
0-50 cm
Range
0.364-1.59
0.437-1.59
0.362-0.482
0.247-0.602
0.378-0.920
0.364-1.07
—
0.247-1.59
0.247-1.59
0.407-1.07
0.362-1.59
—
0.247-1.59
0.362-1.12
0.364-1.59
0.247-1.12
0.378-1.46
SoilN
0-200 cm
N
68
19
3
6
3
69
3
27
68
6
25
0
22
77
37
45
17
SoilN
0-200 cm
Mean
1.29
1.33
0.735
0.871
1.15
1.20
0.730
1.44
1.23
1.04
1.36
—
1.41
1.21
1.26
1.27
1.19
SoilN
0-200 cm
SD
0.512
0.563
0.213
0.288
0.608
0.447
0
0.652
0.483
0.346
0.634
—
0.694
0.454
0.529
0.500
0.601
SoilN
0-200 cm
Range
0.645-2.57
0.730-2.57
0.525-0.950
0.418-1.30
0.801-1.85
0.645-2.52
—
0.418-2.57
0.418-1.30
0.730-2.57
0.525-2.57
—
0.418-2.57
0.525-2.02
0.645-2.57
0.418-2.02
0.730-2.52
105
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
Oregon Lake Data
10000
1000
i
^ 100
Q_
10
1
All Data
-Reference Sites
1:High 2:Medium
OHLR Aquifer Permeability
3: Low
Oregon Lake Data
10000
1000
S
3, 100
10
1
All Data
Reference Sites
1:Very Wet 2:Wet 3:Moist 4:Dry 5:Semiarid
OHLR Climate
106
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
Oregon Lake Data
10000
1000
i
^ 100
Q_
10
1
1:
All Data
-Reference Sites
High 2:Medium 3:Lov
OHLR Soil Permeability
Oregon Lake Data
10000
1000
s
3, 100
10
1-
All Data
-Reference Sites
J 1 — 1
1:Mt 2:Trans 3:Fla
OHLR Terrain
107
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
Oregon Lake Data
10000
1000
i
^ 100
Q_
10
All Data
-Reference Sites
Spring Summer
OHLR Seasonally
Winter
Figure 26 Lake data box and whisker plots of lakewater Total Phosphorus (TP) by OHLR
subclasses for chemistry/habitat screened reference sites (red) and all data (blue). Boxes
represent the 25th and 75th percentiles, the line in the box the median and the whiskers the range.
Oregon Lake Data
10000
1000
100
10
All Data
-Reference Sites
1:High 2:Medium
OHLR Aquifer Permeability
3: Low
108
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
Oregon Lake Data
0000
1000
100
10
1
r.
LI
71
LJ a
All Data
-Reference Sites
1
iJ
High 2:Medium 3:Lov
10000
1000
100
OHLR Soil Permeability
Oregon Lake Data
All Data
Reference Sites
10
1:Very Wet 2:Wet 3:Moist 4:Dry 5:Semiarid
OHLR Climate
109
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
Oregon Lake Data
10000
1000
100
10
All Data
-Reference Sites
1:Mt 2:Trans
OHLR Terrain
Oregon Lake Data
3:Flat
10000
1000
100
10
All Data
Reference Sites
Spring Summer
OHLR Seasonally
Winter
Figure 27 Lake data box and whisker plots of lakewater Total Nitrogen (TN) by OHLR subclasses
for chemistry/habitat screened reference sites (red) and all data (blue). Boxes represent the 25
th
,th
and 75 percentiles, the line in the box the median and the whiskers the range.
110
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
„
I3
o
I 1
o
CM _
I3
O
I 1
o
Oregon Lake Data
All Data
-Reference Sites
1:High 2:Medium
OHLR Aquifer Permeability
Oregon Lake Data
3: Low
All Data
Reference Sites
1:High 2:Medium
OHLR Soil Permeability
3: Low
111
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
„
I3
o
I 1
o
Oregon Lake Data
All Data
-Reference Sites
1:Very Wet 2:Wet 3:Moist 4:Dry 5:Semiarid
OHLR Climate
CM _
I3
O
I 1
O
1:Mt
Oregon Lake Data
All Data
Reference Sites
2 Trans
OHLR Terrain
3:Flat
112
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
„
I3
o
I 1
o
Oregon Lake Data
All Data
-Reference Sites
Spring
Summer
OHLR Seasonally
Winter
Figure 28 Lake data box and whisker plots of 0-200 cm depth soil nitrogen concentration by OHLR
subclasses for chemistry/habitat screened reference sites (red) and all data (blue). Boxes
represent the 25th and 75th percentiles, the line in the box the median and the whiskers the range.
2.3.2.3 Analysis of OHLR class differences
Each of the stream and lake OHLR classes were compared using a one-way analysis of variance
(ANOVA) to test for differences in means within classes using the reference site data (
113
-------
Methodologies for Development of Part 2: N and P in Oregon Hydrologic Landscape Regions
Numeric Nutrient Criteria for Freshwaters
Table 49). For TP, there was a consistent pattern of increasing concentrations as climate progressed from
very wet, to wet, to moist, to dry. It was evident in both lakes and streams although it was stronger in
lakes (Figure 23 and Figure 26). There was also a strong pattern in lakes but not streams of increasing TP
with terrain (Transitional > Mountain) and seasonality (Winter > Spring). In streams, there was a weak
relationship with aquifer permeability but it was not consistent across all three classes (L
-------
Methodologies for Development of Part 2: N and P in Oregon Hydrologic Landscape Regions
Numeric Nutrient Criteria for Freshwaters
Table 49). Other than that, patterns with OHLR classes were not consistent between lakes and streams
for TN. Lake TN (but not stream TN) had highly significant relationships with climate (increasing wet to
dry) and terrain (Transitional > Mountain) and to a lesser extent aquifer permeability (L > M=H). On the
other hand, stream TP but not lake TP was weakly related to soil permeability (L < M=H).
For soil N, there were no significant relationships with any of the OHLR classes in lakes. In streams, soil
N was strongly related to both climate (Dry < Wet) and aquifer permeability. Although there was no
consistent trend in aquifer permeability (M < H=L). There was also a weak relationship of soil N to
seasonality (spring < winter).
115
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
Table 49 Summary of F-statistics for one-way ANOVA testing for OHLR subclass effect. Lake and
streamwater chemistry data were log 10 transformed before analysis.
Dependent Variable
Streamwater TP
Lakewater TP
Streamwater TN
Lakewater TN
Stream data Soil N
0-200 cm
Lake data Soil N
0-200 cm
Aquifer
Permeability
6.11*
1.04
0.33
6.87*
15.7**
1.14
Soil Permeability
2.48
1.30
5.63*
1.69
4.66
0.14
Climate
5.15*
16.3**
0.44
12.2**
11.5**
1.85
Terrain
2.75
15.6**
4.43
16.1**
0.16
2.34
Seasonality
2.79
12.6**
7.19*
5.75*
5.69*
3.84
*p<0.01, **p<0.001
Scatterplots were constructed to analyze the relationship between observed surface water nutrient
concentrations and mapped soil nitrogen and carbon data (Figure 29 and Figure 30). In general,
correlations between soil N and C and surface water TP and TN were very weak or non-existent (
116
-------
Methodologies for Development of Part 2: N and P in Oregon Hydrologic Landscape Regions
Numeric Nutrient Criteria for Freshwaters
Table 50). The highest correlations (-0.3) were between lake water TN and soil N and C. Even that
correlation explained less than 10% of the variance between the variables.
117
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
Table 50 Pearson correlation coefficients between soil nutrient GIS data and observed lake and
stream water nutrient chemistry in chemistry/habitat screened reference sites.
Soil Variable
TP
TN
Lakes and Streams
0-50 cm depth total nitrogen
0-200 cm depth total nitrogen
0-50 cm depth total carbon
0-200 cm depth total carbon
Sample Size
-0.184*
-0.130
-0.192*
-0.105
250
-0.111
-0.067
-0.075
0.166
191
Streams Only
0-50 cm depth total nitrogen
0-200 cm depth total nitrogen
0-50 cm depth total carbon
0-200 cm depth total carbon
Sample Size
-0.101
-0.059
-0.099
-0.090
149
0.130
0.120
0.137
0.110
151
Lakes Only
0-50 cm depth total nitrogen
0-200 cm depth total nitrogen
0-50 cm depth total carbon
0-200 cm depth total carbon
Sample Size
* p<0.01
-0.249*
-0.171
-0.266*
-0.162
101
-0.322
-0.244
-0.312
0.093
40
118
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
10000
i
c 1000
o
D)
100
I
CO
o
I
10
1000
100
10
Lake
" Stream
0 1 2
0-200 cm Soil N (g/m2)
00 Lake
*** Stream
o *
on
o
0 1 2
0-200 cm Soil N (g/m2)
Figure 29 Scatterplot of surface water nutrients versus 0-200 cm depth soil nitrogen for both lakes
and streams.
119
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
10000
i
o
I
CO
o
I
1000
100
10
Lake
Stream
3
0 10 20 30 40 50 60
0-200 cm Soil Carbon (g/m2)
1000
100
10
1
00 Lake
*** Stream
o
0 10 20 30 40 50 60
0-200 cm Soil Carbon (g/m2)
Figure 30 Scatterplot of surface water nutrients versus 0-200 cm depth soil carbon for both lakes
and streams.
120
-------
Methodologies for Development of Part 2: N and P in Oregon Hydrologic Landscape Regions
Numeric Nutrient Criteria for Freshwaters
There were 413 stream and river sites in our data that could be matched to SPARROW modeled reaches.
The summary statistics for these sites showing observed and SPARROW modeled TP and TN are shown
in Table 51. Also constructed were 1:1 scatterplots showing the relationship between observed and
SPARROW model results. The data were broken out into two size classes for reporting, rivers and
streams based on field sampling protocol. Streams were defined as sites that could be safely waded down
the thalweg (generally < 1 m depth). Rivers were sites too deep to be safely waded and were sampled
with a raft.
In both streams and rivers, there was no meaningful relationship between observed streamwater
concentration and SPARROW model results along the 1:1 line for either TP or TN (Figure 31 and Figure
32). SPARROW model results are confined to a much tighter range than observed values. Maximum
SPARROW values are two orders of magnitude lower than maximum observed values. It should be
noted that observed values in the database represent summer baseflow conditions. SPARROW models
annual loads and flows to calculate a mean annual concentration so the two variables measure different
things.
Table 51 Summary statistics for the observed and SPARROW modeled nutrient concentrations in
the 413 Oregon stream and river sites that had SPARROW model output.
Observed TP SPARROW TP Observed TN SPARROW TN
(ug/L) (ug/L) MatlStl° (u/L) (ug/L)
22.0
10.0-50.0
6.0-83.4
6720
407
17.2
14.4-19.7
3.5-24.7
48.2
413
Median
Interquartile Range
10th- 90th Percentiles
Maximum
Sample Size
174
110-290
56-484
38,100
413
166
102-190
18.6-203
232
413
121
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
Beatable Rivers
1000.0
§ 100.0
O
|
I
cc
10.0
1.0
0.1
10 100
Observed TP (ug/L)
1000
Beatable Rivers
•33
8
tr
I
CO
10000
1000
100
10
10
100 1000
Observed TN (ug/L)
10000
Figure 31 Scatterplots of SPARROW model nutrient concentrations versus observed nutrient
concentrations for boatable rivers in Oregon. The black line is the 1:1 line.
122
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Part 2: N and P in Oregon Hydrologic Landscape Regions
Wadeable Streams
10000.0
g 1000.0
Q_
•75 100.0
l_l_
I
CO
10.0
1.0
0.1
100000
"Si
3
•33
8
tr
CO
10000
1000
100
10
1 10 100 1000 10000
Observed TP (ug/L)
Wadeable Streams
10 100 1000 10000
Observed TN (ug/L)
100000
Figure 32 Scatterplots of SPARROW model nutrient concentrations versus observed nutrient
concentrations for wadeable streams in Oregon. The black line is the 1:1 line.
123
-------
Methodologies for Development of Part 2: N and P in Oregon Hydrologic Landscape Regions
Numeric Nutrient Criteria for Freshwaters
There were too many (~60) individual concatenated OHLR classes to analyze individually.
However, each of the 5 OHLR attribute classes that make up the overall OHLR class (climate,
terrain, seasonality, aquifer permeability, and soil permeability) could be analyzed.
For TP, there was a consistent pattern of increasing concentrations as climate progressed from
very wet, to wet, to moist, to dry. It was evident in both lakes and streams although it was
stronger in lakes (Figure 23 and Figure 26). There was also a strong pattern in lakes but not
streams of increasing TP with terrain (Transitional > Mountain) and seasonality (Winter >
Spring). In streams, there was a weak relationship with aquifer permeability but it was not
consistent across all three classes (L Mountain) and to a
lesser extent aquifer permeability (L > M=H). On the other hand, stream TP but not lake TP was
weakly related to soil permeability (L < M=H).
For soil N, there were no significant relationships with any of the OHLR classes in lakes. In
streams, soil N was strongly related to both climate (Dry < Wet) and aquifer permeability.
Although there was no consistent trend with aquifer permeability subclasses (M < H=L). There
was also a weak relationship of soil N to seasonality (spring < winter).
Overall, the most consistent patterns were with climate and seasonality. In general, stream water
nutrient concentrations increased and soil N concentrations decreased as climate changed from
wet to dry. Both stream water nutrients and soil N tended to be higher in areas with maximum
flows in winter than in those areas with maximum flows in spring (summer maximum flow areas
are very rare).
• In general, correlations between soil N and C and surface water TP and TN were very weak or
non-existent. The highest correlations (r = -0.3) were between lake water TN and both soil N and
C and were not significant at p<0.01. Even that correlation explained less than 10% of the
variance between the variables.
• In both streams and rivers, there were no relationships between observed stream water
concentrations and SPARROW model results for either TP or TN. SPARROW model results are
confined to a much tighter range than observed values. Maximum SPARROW values were two
orders of magnitude lower than maximum observed values. It should be noted that observed
values in this database represent summer baseflow conditions. SPARROW models annual loads
and flows to calculate a mean annual concentration so the two variables measure different things.
124
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
3 References
Alabama Department of Environmental Management (ADEM). 2010. 2010 Integrated Water Quality
Monitoring and Assessment Report: Water Quality in Alabama 2008-2009. Alabama Department of
Environmental Management, Montgomery, AL. Accessed via the internet June 2013 at
http://www.adem.state.al.us/programs/water/waterforms/2010AL-IWQMAR.pdf
Carlson R.E. 1977. A trophic state index for lakes. Limnology and Oceanography 22:361-369.
Carpenter, S.R., J.J. Cole, J.F. Kitchell, and M.L. Pace. 1998. Impact of dissolved organic carbon,
phosphorus, and grazing on phytoplankton biomass and production in experimental lakes. Limnology
and Oceanography 43: 73-80.
Dixit, S.S., J.P. Smol, D.F. Charles, RM. Hughes, S.G. Paulsen, and G.B. Collins. 1999. Assessing water
quality changes in the lakes of the northeastern United States using sediment diatoms. Canadian
Journal of Fisheries and Aquatic Science 56: 131-152.
Dodds, W. K. and J. J. Cole. 2007. Expanding the concept of trophic state in aquatic ecosystems: It's not
just the autotrophs. Aquatic Sciences. 69:427-439
Edmondson, W.T. 1970. Phosphorus, nitrogen, and algae in Lake Washington after diversion of sewage.
Science 169: 690-691.
Fore LS, Frydenborg R, Wellendorf N, Espy J, Frick T, Whiting D, Jackson J, and Patronis J. 2007.
Assessing the Biological Condition of Florida Lakes: Development of the Lake Vegetation Index
(LVI). Florida Department of Environmental Protection, Nonpoint Source Bioassessment Program.
Accessed via the inter net June 2013 at
http://publicfiles.dep.state.fl.us/dear/sas/sopdoc/lvi_fmal07.pdf
Florida Department of Environmental Protection (FL DEP). 201 la. Sampling and Use of the Lake
Vegetation Index (LVI) for Assessing Lake Plant Communities in Florida: A Primer. Florida
Department of Environmental Protection, Bureau of Assessment and Restoration Support, Standards
and Assessment Section, Tallahassee, FL. DEP-SAS-002/11. Accessed via the internet June 2013 at
http://www.dep.state.fl.us/water/sas/training/docs/lvij3rimer.pdf
Florida Department of Environmental Protection (FL DEP). 201 Ib. Development of Aquatic Life Use
Support Attainment Thresholds for Florida's Stream Condition Index and Lake Vegetation Index.
Florida Department of Environmental Protection, Bureau of Assessment and Restoration Support,
Standards and Assessment Section, Tallahassee, FL. DEP-SAS-003/11. Accessed via the internet
June 2013 at http://www.dep.state.fl.us/water/bioassess/docs/attainment-thresholds-sci-and-lvi.pdf
Georgia Department of Natural Resources (GA DNR). 2013. Fisheries Management in Public Waters.
Georgia Department of Natural Resources, Wildlife Resources Division, Atlanta, GA. Accessed via
the internet June 2013 at http://www.georgiawildlife.com/node/933
Gerritsen J, Jessup B, Leppo EW, and White J. 2000. Development of Lake Conditions Indexes (LCI) for
Florida. Prepared by Tetra Tech for Florida Department of Environmental Protection. Accessed via
the internet June 2013 at http ://publicfiles. dep. state, fl.us/dear/sas/sopdoc/lci fmal.pdf
Herlihy, A.T., D.P. Larsen, S.G. Paulsen, N.S. Urquhart, and B.J. Rosenbaum. 2000. Designing a
spatially balanced, randomized site selection process for regional stream surveys: the EMAP mid-
Atlantic pilot study. Environmental Monitoring and Assessment 63:95-113.
12?
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Herlihy, A.T., and J.C. Sifneos. 2008. Developing nutrient criteria and classification schemes for
wadeable streams in the conterminous USA. Journal of the North American Benthological Society
27:932-948.
Herlihy, A.T., J.B. Sobota, T.C. McDonnell, T.J. Sullivan, S. Lehmann, and E. Tarquinio. 2013. An a
priori process for selecting candidate reference lakes for a national survey. Freshwater Science 32:
385-396. doi: 10.1899/11-081.1.
Illinois Environmental Protection Agency (IEPA). 2012. Illinois Integrated Water Quality Report and
Section 303(d) List, 2012. Clean Water Act Sections 303(d), 305(b) and 314. Water Resource
Assessment Information and List of Impaired Waters. Volume I: Surface Water. Illinois
Environmental Protection Agency, Springfield, IL. Accessed via the internet June 2013 at
http://www.epa.state.il.us/water/tmdl/303-appendix/2012/iwq-report-surface-water.pdf
Iowa Department of Natural Resources (IDNR). 2013a. Iowa Fish Survey Data. Iowa Department of
Natural Resources. Fishery Division, Des Moines, IA. Accessed via the internet June 2013 at
http://limnoweb.eeob.iastate.edu/fishpub/default.aspx
Iowa Department of Natural Resources (IDNR). 2013b. Methodology for Iowa's 2012 Water Quality
Assessment, Listing, and Reporting Pursuant to Sections 305(b) and 303(d) of the Federal Clean
Water Act. Iowa Department of Natural Resources, Des Moines, IA. Accessed via the internet June
2013 at
http://www.iowadnr.gov/Portals/idnr/uploads/watermonitoring/impairedwaters/2012/Iowa%20fmal%
202012%20methodology.Ddf
Jeppesen, E., J.P. Jensen, and M. S0ndergaard. 2002. Response of phytoplankton, zooplankton and fish to
re-oligotrophication: An 11-year study of 23 Danish lakes. Aquatic Ecosystem Health and
Managements: 31^43.
Kentucky Department for Environmental Protection (KY DEP). 2010. 2010 Integrated Water Quality
Report. Kentucky Department for Environmental Protection, Frankfurt, KY. Accessed via the internet
June 2013 at http://water.ky. gov/waterqualitv/Pages/IntegratedReport.aspx
Kentucky Department of Fish and Wildlife Resources. 2011. Annual Performance Report: District
Fisheries Management, Part I. Project 1: Lake and Tailwater Fishery Surveys. Kentucky Department
of Fish and Wildlife Resources, Fisheries Division, Frankfurt, KY. Accessed via the internet June
2013 at http://fw.kv.gov/pdf/201 Olake&tailwatersurvevs.pdf
Leach, J.H., M.G. Johnson, J.RM. Kelso, J. Hartman, W. Numann, and B. Entz. 1977. Responses of
percid fishes and their habitats to eutrophication. Journal of the Fisheries Research Board of Canada
34: 1959-1963
Maceina, M.J., and D.R Bayne. 2001. Changes in black bass community and fishery with
oligotrophication in West Point Reservoir, Georgia. North American Journal of Fisheries
Management. 21:745-755.
Mehner, T., M. Diekmann, U. Bramick, and R. Lemcke. 2005. Composition offish communities in
German lakes as related to lake morphology, trophic state, shore structure and human-use intensity.
Freshwater Biology 50: 70-85.
Minnesota Pollution Control Agency (MN PCA). 2009. Minnesota National Lakes Assessment Project:
Fish-based Index of Biotic Integrity (IBI) for Minnesota Lakes. Minnesota Pollution Control Agency,
126
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Saint Paul, MN. Accessed via the internet June 2013 at http://www.pca.state.mn.us/index.php/view-
document.html?gid=6882
MN PCA. 2012. Tiered aquatic life use (TALU) framework. Minnesota Pollution Control Agency, Saint
Paul, MN. Accessed via the internet June 2013 at http://www.pca.state.mn.us/index.php/water/water-
permits-and-rules/water-rulemaking/tiered-aquatic-life-use-talu-framework.html
MN PCA. 2013. Guidance Manual for Assessing the Quality of Minnesota Surface Waters for
Determination of Impairment: 305(b) Report and 303(d) List: 2012 Assessment Cycle. Minnesota
Pollution Control Agency, Saint Paul, MN. Accessed via the internet June 2013 at
http://www.pca. state.mn.us/index.php/view-document.html?gid=16988
Mississippi Department of Environmental Quality (MDEQ). 2012. State of Mississippi Water Quality
Assessment 2012 Section 305 (b) Report. Mississippi Department of Environmental Quality, Jackson,
MS. Accessed via the internet June 2013 at
http://www.deq.state.ms.us/MDEQ.nsf/pdf/FS MS 12 Section 305b WQA report/SFile/2012 305
b_report.pdf? OpenElement
Missouri Department of Natural Resources (MDNR), 2010. Methodology for the Development of the
2012 Section 303(d) List in Missouri. Missouri Department of Natural Resources, Jefferson City,
MO. Accessed via the internet June 2013 at http://dnr.mo.gov/env/wpp/waterquality/303d/2012-lmd-
approved090810.pdf
Murray State University. 2013. Hancock Biological Station on Kentucky Lake: Kentucky Lake Water
Quality Monitoring. Accessed via the internet June 2013 at
http://www.murraystate.edu/qacd/cos/hbs/WO.cfm
Ney, J.J. 1996. Oligotrophication and its discontents: effects of reduced nutrient loading on reservoir
fisheries. American Fisheries Society Symposium. 16:285-295.
North Carolina Division of Environment and Natural Resources (NC DENR). 2013. 2014 North Carolina
303(d) Listing Methodology. North Carolina Division of Environment and Natural Resources,
Raleigh, North Carolina. Accessed via the internet June 2013 at
http://portal.ncdenr.org/c/document_library/get_file?uuid=lfld590f-a096-4eba-9853-
c5dab2c5c431&groupld=38364
Organization for Economic Cooperation and Development (OECD). 1982. Eutrophication of waters.
Monitoring, assessment and control. Final report, OECD cooperative programme on monitoring of
inland waters (eutrophication control), Environment Directorate, OECD, Paris.
Pace, M.L, and J.J. Cole. 1996. Regulation of bacteria by resources and predation tested in whole lake
experiments. Limnology and Oceanography 41: 1448-1460.
Peck, D.V., A.T. Herlihy, B.H. Hill, RM. Hughes, P.R Kaufmann, D.J. Klemm, J.M. Lazorchak, F.H.
McCormick, S.A. Peterson, P.L. Ringold, T. Magee, and M.R. Cappaert. 2006. Environmental
Monitoring and Assessment Program - Surface Waters Western Pilot Study: Field Operations Manual
for Wadeable Streams.EPA 620/R-06/003, U.S. Environmental Protection Agency, Office of
Research and Development, Washington, DC.
Sand-Jensen, K. & Borum, J. 1991. Interactions among phytoplankton, periphyton, and macrophytes in
temperate freshwaters and estuaries. Aquatic Botany 41:137-75.
127
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Schindler, D.W. 1974. Eutrophication and recovery in experimental lakes: Implications for lake
management. Science 184: 897-899
Schindler, D.W., 1990. Experimental perturbations of whole lakes as tests of hypotheses concerning
ecosystem structure and function. Oikos 57, 25-41.
South Carolina Department of Natural Resources (SC DNR). 2011. Statewide Research—Freshwater
Fisheries: Annual Progress Report January 1, 2011-December 31, 2011. South Carolina Department
of Natural Resources. Division of Wildlife and Freshwater Fisheries, Columbia, SC. Accessed via the
internet June 2013 athttp://www.dnr.sc.gov/fish/fwfi/files/201 l_annual_report.pdf
South Carolina Department of Health and Environmental Control (SCDEHEC). 2013. State of South
Carolina Monitoring Strategy for Calendar Year 2013. South Carolina Department of Health and
Environmental Control, Bureau of Water, Columbia, SC. Technical Report No. 0109-13. Accessed
via the internet June 2013 at https://www.scdhec.gov/environment/water/docs/strategv.pdf.
South Dakota Department of Environment and Natural Resources (SDDENR). 2010. The 2010 South
Dakota Integrated Report for Surface Water Assessment. South Dakota Department of Environment
and Natural Resources, Pierre, SD. Accessed via the internet June 2013 at
http: //denr. sd. gov/documents/1 Oirfmal. pdf
Tennessee Department of Environment and Conservation (TDEC). 2012. 2012 305(b) Report. The Status
of Water Quality in Tennessee. Tennessee Department of Environment and Conservation, Nashville,
TN. Accessed via the internet June 2013 at
http://www.tn.gov/environment/water/docs/wpc/2012 305b.pdf
Tilman, D. 1982. Resource competition and community structure. Monographs in Population Biology,
Princeton University Press, Princeton, New Jersey, USA.
United States Environmental Protection Agency (USEPA). 1998. Lake and reservoir bioassessment and
biocriteria. Technical guidance document. United States Environmental Protection Agency, Office of
Water, Washington, DC. EPA 841-B-98-007.
USEPA. 2000. Nutrient Criteria Technical Guidance Manual, Lakes and Reservoirs. United States
Environmental Protection Agency, Office of Water, Washington, DC. EPA-822-BOO-001.
USEPA. 2001. Ambient Water Quality Criteria Recommendations. Information Supporting the
Development of State and Tribal Nutrient Criteria. Lakes and Reservoirs in Nutrient Ecoregion III.
United States Environmental Protection Agency, Office of Water, Washington, DC. EPA-822-B01-
008.
USEPA. 2009. National Lakes Assessment: A Collaborative Survey of the Nation's Lakes. U.S.
Environmental Protection Agency, Office of Water and Office of Research and Development,
Washington, D.C. EPA 841-R-09-001.
USEPA. 2010. Technical Support Document for U.S. EPA's Final Rule for Numeric Criteria for
Nitrogen/Phosphorus Pollution in Florida's Inland Surface Fresh Waters. United States
Environmental Protection Agency, Office of Water, Office of Science and Technology, Washington,
DC.
Vaga, R.M., and A.T. Herlihy. 2004. A GIS inventory of Pacific Northwest lakes and reservoirs and
analysis of historical water quality data. EPA 910-R-04-009, U.S. Environmental Protection Agency,
Region 10, Seattle, WA.
128"
-------
Methodologies for Development of
Numeric Nutrient Criteria for Freshwaters
Vaga, R.M., R.R. Petersen, M.M. Sytsma, M. Rosenkrantz, and A.T. Herlihy. 2005. A Classification of
Lakes in the Coast Range Ecoregion with Respect to Nutrient Processing. EPA 910-R-05-002. U.S.
Environmental Protection Agency, Region 10, Seattle, WA.
Vaga, R.M., A.T. Herlihy, R Miller, and M.M. Sytsma. 2006. A Classification of Pacific Northwest
Reservoirs with Respect to Nutrient Processing. EPA 910-R-06-003. U.S. Environmental Protection
Virginia Water Resources Research Center (VWRRC). 2005. Issues Related to Freshwater Nutrient
Criteria for Lakes and Reservoir in Virginia. Virginia Water Resources Research Center, Blacksburg,
VA. VWRRC Special Report SR27-2005.
Wetzel, R. G. 2001. Limnology: Lake and Reservoir Ecosystems. Third edition. Academic Press, New
York.
Wisconsin Department of Natural Resources (WI DNR). 2013. Wisconsin 2014 Consolidated Assessment
and Listing Methodology (WisCALM) for Clean Water Act Section 305(b), 314, and 303(d)
Integrated Reporting. Wisconsin Department of Natural Resources, Bureau of Water Quality,
Madison, WI. EGAD#3200-2013-01. Accessed via the internet June 2013 at
http://dnr.wi.gov/news/input/documents/guidance/wiscalmguidance.pdf
Yuan, L. 2006. Estimation and Application of Macroinvertebrate Tolerance Values. United States
Environmental Protection Agency, National Center for Environmental Assessment, Office of
Research and Development, U.S. Environmental Protection Agency, Washington, D.C. EPA/600/P-
04/116F
129
-------
Appendix 1 - Wl TP GAM Models Phytoplankton
-------
Capture Probability of Phytoplankton Taxon Along TP Gradient
Anabaena circinalis
Anabaena flos-aquae
Aphanizomenon flos-aquae
o
"-
^ 00
:= d
o
CD
o to
O d
CL
0 •* _
3 °
-i—*
Q.
CD CN
O o ~
0
d
.
. •
* * * • • "
- T- P _
d "~
28 &<°
d C i= °
CD .Q
T3 CD
-- c .Q co
d 3 o d
< Q-
-00 CD "* _
d > ^ °
'-I—' -1—'
CD Q.
-80 $ g -
- 0 °- -
.
m
.
• •
• .' J"
•M»._ •>»•.•
- CN P -
d "~
00 g >,m
d C 1= °
CD ^2
"O CD
C o to
o ^ O d
< Ol
-80 0) "* _
d > ^ °
_ "CD "o.
.00 CD
'•4— '
CD
. | 0
~ O
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
Aphanizomenon issatschenkoi
Aphanocapsa delicatissima
Aphanothece nidulans
p _
^~
^ 00
:= °
CD
O CD
0 d
Q.
0 •* _
3 °
Q.
CD CN
O o ~
0
0
.
•
• •
•
" ~"~" " ™
. 0 P _
0 "~
.08 £»-
d C :^ °
CD ^2
. 8 1 & ^ °
CD Q.
.50 CO CN _
OQ: o °
o
° d
.
.
..... .J.'.in..^ j* .. •
;! p _
0 "~
.88 ^» _
d C :^ °
CD .Q
. T3 CD
. 0 C .Q CD
d ^ O o
< ^
-00 ^ °
CD Q.
.80 CD CN _
OQ: O °
o
° d
.
•
.
•
.
i
- C*3
0
1 1
0.22 0.3
>undance
. £ 0
0 >
K_ CD
- fe 0
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
Page 1 of 11
-------
Capture Probability of Phytoplankton Taxon Along TP Gradient
Aulacoseira granulata
Chlamydomonas globosa
Coelosphaerium naegelianum
o
•<-
,>> 00
™ d
o
CD
O CD
O d
CL
0 •* _
is °
-i—*
Q.
CD CN
O c i -
0
d
•
V.-JL
. i^ P _
d "~
. fe 8 & °° -
d C 1= °
CD .Q
CO "° tO
. ^ C ^2 CD _
° * 2 °
< Q-
- CN 0 CD ^ _
° > 3 °
-I—* -1—'
_^ CO Q_
• 5 ® o 2 -
„ o
o d -
•
I ^^ .
- 0 P _
d "~
.88 £>» -
d C 1= °
CD .Q
-* "° ro
. § C ^2 CD _
d =3 00
< Ol
-80 0) * _
° > ^ °
-1—' -1—'
CD Q.
- 5 0 ro ^ _
OQ: o °
._. o
o d -
.
_
• . • •" 4 • • «bMhv • off »^ f • * •
CN
d
CD CD
- - O
d C
CD
"U
C
~ . ^
^2
<
00
-00
d >
'•4— '
- to
- 0 0
o o:
~ O
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
Cryptomonas erosa
Erkenla subaequlclllata
Lyngbya limnetica
p _
"<~
^> 00 _
:= d
o
CD
O CD
2 °
CL
0 •* _
3 °
-1—*
Q.
CD CN
O °
0
d
t
*
'
'
,
,
. •
*
• . . :C*' '-
00 O _
d ^ ~
. s 8 ^» _
d C = °
CD .Q
"O CO
T? C O CD
o ^ O o
< Q-
- co 0 0) "* _
d > ^ °
'•4— ' -I— '
CD Q.
- • 0 O d ~
._. o
0 o
t
. *
•
.
• idi-.' i:
. 0 P _
d ^~
.88 ^» _
d C :^ °
CD ^2
_,. T3 CD
. S C ^2 cp
o ^ O o
< Q-
-80 ^ °
'•4-* -4-1
CD Q.
.50 tO CN _
OQ: o °
._. o
0 o
^
;
.
i --
d
^r 0
- - O
d C
CD
T3
- C
o 3
<
- O 0
d >
'•*-*
CD
. 8 0
o o:
~ O
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
Page 2 of 11
-------
Capture Probability of Phytoplankton Taxon Along TP Gradient
Merismopedia tenuissima
Microcystis aeruginosa
Monomastix astigmata
o
"<~
,>> 00
— d
O CD
O d
CL
0 •* _
-1—*
Q.
CD CN
O o ~
0
d
'
•
: .. . —. • .'• .
1 1 1 1 1 | 1 1 1 1 1 1 1 1 | 1 1 1 1 1 1
- 0 P _
d "~
. g 8 ^» _
• f— -^ Q
O C 7=
CO _Q
~o tn
ro ^
. § C .£2 co _
° 2 e °
< Q-
-00 0 "* _
d > ^ °
'-I—' -1—'
CD Q.
- S 0 CO CN _
OQ: o o
o_
• _ „ -
--t''^^'
^^^ ,
.S^ j,-~
^^^ ''
X
X • *
X
X „
X
X
x a •
* "*«"."
* Si J* _. . . •
- 0) P _
d "~
. fs 8 3?» -
^ f— -^ Q
O C 7=
CO _Q
Ti m
I-. ^^ VU
. C5 C ^2 5 °
-f-* -f-*
CD Q.
. 5 "0 CD CN _
d £ O o
o_
•
••
• •••••«•*• «Mift**oa MM> •• ••
1 1 1 1 1 | 1 1 1 1 1 1 1 1 | 1 1 1 1 1 1
- 0
d
r^ 0
- p O
d C
CD
CD f£
- O *—
0 E
^r *^
-00
d >
'•4— '
CD
- S 0
o o:
- 0
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
Monoraphidium capricornutum
Nltzschla palea
p _
"~
^> 00 _
:= d
o
CD
O CD
0 d
Ql
0 •* _
^
"o.
CD CN
O o ~
o
d
.
.
.
- o
0
- o O
CD
T3
c
3
^
o 0
._
"CD
-00
a:
- o
p _
"~
>, CO _
r^ ^
o
CD
ol
0 ^r
>- d
^
"o.
o
< ,
" '
• • o o « « a o B^I^^B* o^ «a*a a a •
. 0
d
^ 0
- o O
o C
CD
T3
- ° §
^
o 0
._
"CD
-00
a:
- o
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
Oocystis parva
CD
O CD
O d
0
-I—*
Q.
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
8
d
co 0
o O
d C
CD
8 I
o 3
o 0
d >
'•4—'
JD
o 0
o a:
Page 3 of 11
-------
Capture Probability of Phytoplankton Taxon Along TP Gradient
Pseudanabaena limnetica
& oo
:= d
CD
.a
o
CL
0
-I—*
Q.
CD
O
{.*.*£ ,'m »•
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
O 0
d >
si
Pseudanabaena mucicola
r^
d
d C = o
CD .£2
T3 CD
-- c .Q cq
o'3 O d
0)
Q.
CD
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
8
CN CD
o O
d C
CD
CN "2
o £
_ <
q 0
J5
o 0
o a:
Quadrigula lacustris
:= o
!a
0 "*
"o.
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
8
d
^ 0
o O
d C
CD
o 0
d >
'•4—'
JD
o Q)
a:
Rhodomonas minuta
Schroederia judayi
Sphaerocystis schroeteri
p _
^
-1— »
:= oq _
o d
CD
.a
0
n~ co _
d
0
3
"o
0 °"
CN _
d
.
•
m
•"..
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
-In P -
d ^~
„ 8 ^«.
. " c = °
d tO ^2
T3 CD
C ^2 CD _
13 00
- ™ 3 °
co CD Q.
- T "m CD CN
0 & Oo-
^~ O
- ^ d ~
o
i
.
^
.
• * * j*" * •
i i 1 1 1 1 i i i i i 1 1 1 1 i i i i i i
T- P _
d "~
CD g >,„
d C j= °
CD ^2
"O CO
_ 2 C ^2 3 °
- "to H
. S "0 tD CN _
OQ: o °
_ 0
— o • ~~
.
0
<
•
*
•
" • • •
B •«•*.. fhl«>. B OB B ««0 .• . D
1 1 1 1 1 | 1 1 1 1 1 1 1 1 | 1 1 1 1 1 1
"J-l
d
in CD
- - O
d C
CD
T3
^ C
" ""I
- o 0
d >
"Jg
. S 0
o a:
- O
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
Page 4 of 11
-------
Capture Probability of Phytoplankton Taxon Along TP Gradient
Stephanodiscus parvus
Synechococcus elongatus
Anabaena
o
•<-
_^> CO
:= d
CD
O CD
O d
CL
0 •* _
Q.
CD CM
O o ~
0
0
.
•
. *
- 0 P
d "~
.§8 £>» -
d C 1= °
CD .a
^ T3 CD
r~i C~ O (O
d ^ O o i ~
< G-
-80 CD •* _
d > 3 °
CD a.
-50 ro undance
- ^ 0
0 >
.si
o Q:
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
Aphanizomenon
Aphanocapsa
Aphanothece
p _
^~
^ CO
:= °
CD
O CD
0 d
CL
0 •* _
3 °
Q.
CD CM
O o ~
0
0
.
'
*. • "•
•
• * •
. 5> P _
0 ^~
CD 0 >, („
d C •= °
CD £)
. T3 CD
. &5 c ^ CD _
d ^ O o
- co 0 ^ °
CD Q.
- ™ 0 ^ °
CD Q.
.80 ro undance
. £ 0
0 >
_8 J
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
Page 5 of 11
-------
Capture Probability of Phytoplankton Taxon Along TP Gradient
Aulacoseira
& 00
:= d
CD
O CO
O d
CL
0 •*
13 °
-i—*
Q.
CD CN
O o
q
d
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
0
0
"-^
JD
0
o:
Chlamydomonas
r- CD >, _
in o ••-• <°.
d C 1= °
CD .a
^ C ^2 co
d 5 P o
0
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
r- 0
o O
d C
CD
in ^
o b:
-00
:*=
CD
0
a:
Chroococcus
:= o
!a
CD
0
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
u s
'•4—'
JD
o 0
a:
Coelosphaerium
Cryptomonas
Cyclotella
p _
•<-
^* 00
:= °
o
CD
O CO
0 d
CL
0 •* _
3 °
-i— »
Q.
CD CN
O o ~
0
s
,
.
'•'„••'
a .0 a 4 a • " «• »"• *•
1 1 1 1 1 | 1 1 1 1 1 1 1 1 | 1 1 1 1 1 1
OM O _
d T^ ~
CO 0 >,
_ ^ — o *^ ^ —
d C •= °
CD .a
-. T3 CD
0x1 C ^2 co
" o =3 0 d
^ °
'•4— ' -I— '
CD Q.
- o 0 TO CN _
da: o °
_ 0
— o • ~~
^
0
• .
• .
. *
• '
• o | a • ** Jfc * *• • J
1 1 1 1 1 | 1 1 1 1 1 1 1 1 | 1 1 1 1 1 1
.00 P _
d "~
0 >,
r^ f\ j^* oo
° C •= °
co !Q
"O CO
[o C O CD
o ^ O o
f^ ^
^ °
'•4— ' -I— '
CD Q.
" d Ct O 2 ~
_ 0
— o • ~~
s
•
. •
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
VJ^
d
CO 0
. ;- o
d C
CD
™ "^
- o £=
d =5
^2
<
CO
- O 0
d >
'•4— '
CD
. 8 0
- O
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
Page 6 of 11
-------
Capture Probability of Phytoplankton Taxon Along TP Gradient
Erkenia
& 00
:= d
CD
O (D
O d
0 •*
13 °
-i—*
Q.
CD c\i
O o
q
d
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
£
in CD
o O
d C
CD
_ T3
CO
O 0
d >
'•4—'
JD
o 0
o Q:
Gymnodinium
:= o
!a
CD
o to
O d
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
r- 0
o o
d C
CD
•« c
o b:
80
sl
o o:
Lyngbya
:= o
!a
Q.
0
P M «• • * 11 I • MR %• •
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
^r 0
^ O
d C
CD
T3
^ C
d 3
^2
o 0
d >
si
o a:
Merismopedia
P _
^~
^ 00
:= °
o
CD
o to
0 d
Ql
0 •* _
5 °
"o.
CD OM
O o ~
0
d
B
•
.
8
; J1 , ^_. \
. 8
d
oo CD
- o O
d C
CD
- 1 §
^
- o 0
"CD
. 8 0
o a:
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
Microcystis
Q.
CD
Monomastix
B
.
•
»•
•
• • '• •• '
• :*'".'
. 5i
d
to CD
- r^ o
d C
CD
-s 1
00
- CO 0
d >
'"CD
. 2 0
o o:
- 0
P _
^
-1—' . —
1— O
!Q
CD
o to _
O o
0 •* _
3 °
-1—*
Q.
CD OM
O o ~
P
.
•„
•
• • ••• -• — "•— '—
. 8
d
r^ 0
- o O
d C
CD
-ij
^r *^
- o 0
d >
.si
o a:
- 0
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
Page 7 of 11
-------
Capture Probability of Phytoplankton Taxon Along TP Gradient
Monoraphldlum Nltzschla Oocystls
o
•<-
& °°
™ d
o
CD
O CD
O d
CL
^ 00
- o O <-• • -
^- "^ o
CD !5
T3 CD
_ C o co
" ° 13 0 d ~
o ^_
,
-00 -^ °° -
d C 1= °
CD .Q
^ T3 CD
- O ^ -" . —
f^ t-_
< °-
-50 0) •* _
o > — ^ o
•i= i2
CD Q.
- ° £ o 2 -
o_
•
•
.'• ••
. • ... ;. . -JL- -..'.-.
- 0
d
- o o
d C
CD
« C
- P §
_Q
CM <
CN
-00
d >
'-I—'
CD
-00
d Q:
- 0
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
Pseudanabaena
Quadrigula
Rhodomonas
O _
T~
^ 00
:= °
o
CD
O CD
0 d
CL
0 •* _
3 °
"o.
CD CN
O o ~
0
d
•
. • .' i ' ' *
1 1 1 1 1 | 1 1 1 1 1 1 1
P"
. •
•• *
•• , m
',-*: .v ' o t-
L*
1 '
T- P
d "~
d C •= °
CO &
~O CO
-- C £) to
" d D Od
_Q *-
-00 CD ^ _
d .> ^ °
"co "o.
- 8 0 co CN _
OQ: o o
p.
.
•*'
.". . _Jn?... ...00°. o.
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
. 0 P _
d "~
T- 0 *i^
- o O •<-?
d C = op _
CD .Q o
^ T3 CD
_ o ^ -Q
d =5 0
^ < Q: d -
-00 3
"CD "o. ^.
- o
-------
Capture Probability of Phytoplankton Taxon Along TP Gradient
Scenedesmus Schroederia Sphaerocystis
CD
O d
tt „
0 d
^
"o. d
CD
o -
o
p
d
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
p
00 -^
O ~
CD
T3
5 g
0
CD
0
o a:
.a
CD
O CD
O d
0 •*
3 d
"o.
CD CN
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
in CD
*- O
d C
CD
T3
^1 C
-00
d >
'•4—'
_ CD
o 0
o a:
Stephanodiscus
Synechococcus
Synechocystis
p _
"<~
^> CO _
:= d
o
CD
O CD
2 °
CL
^ co
- CD O *J °° _
d C :^ o
CD ^2
"O CD
^ C ^2 CD
o ^ O d "
^2 >-
^ °
"CD "o.
- - 0 /O CN _
OQ: o o
_ 0
— o • ~~
t
•
0
r
• ° B t * "
• " " 2&ML/ • "* °
"' ••»• • . . •
- -T- P _
d ^~
T- 0 >. 00
d C •= °
CD .a
. § "c 5 CD _
o ^ O o
f} ^
^ °
"CD "o.
. 8 "0 to CN _
do: O o
_ 0
— o • ~~
u
. •
B *
B B
• . _. B.Bt, _ . ^. ..
•»-
d
oo 0)
- o O
d C
CD
- 8§
f^
^
- o 0
o .>
"CD
. 8 0
o a:
- O
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
Page 9 of 11
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
-------
Capture Probability of Phytoplankton Taxon Along TP Gradient
Uroglena
& 00
:= ci
CD
.a
o
CL
0
-i—*
a.
CD
O
q
ci
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
in CD
in o
o C
CD
T3
oo
CN
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
r- 0
r- O
d C
CD
oo "a
"> §
0 E
O)
CO 0
d >
"CD
- 0
o a:
Chlorophyta
>> 00
:= d
CD
O CO
O d
Ql
0 ^
3 °
O
q
d
IV
•i ;•. ••«,:*.•...• o
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
£
d
CD 0
r*~ o
d C
CD
^ "O
oo <
CO 0
d >
'•4—'
CD
2 0
Chrysophyta
Cryptophyta
Cyanophyta
q _
•<~
^ 00
:= °
o
CD
O CD
0 d
CL
0 •* _
3 °
"o.
CD CN
O d -
0
d
^
„
• a a •" • B *H llJIB 1 % Vha •. a*
00 O _
d ^ ~
f 0 "S".
- CD O ••-•
d C = oq _
CD ^2O
"O CD
02 (^ Q
" °| 2 (D_
- CO 0 0
° .> 3
CD "o. ^.
- T- 0 to o -
- 0 CN
t
•
q
a
s
_
'
t
_
B>
• . • i .... •O ...* •.'
. &5 ° -
d ^~
. S C = § ~
d tO .Q
T3 CD
C .Q
"CD
-ss
- 0
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
Page 10 of 11
-------
Capture Probability of Phytoplankton Taxon Along TP Gradient
Pyrrhophyta
_g> 00
:= o
CD
O CO
O o
CL
0 •*
Q.
CD CN
O o
0
0
0.01 0.03 0.1 0.3
Total Phosphorus (mg/L)
r^
ci
co CD
- O
d C
CD
T3
^ C
ci 3
-00
d >
si
o a:
Page 11 of 11
-------
Appendix 2 - Wl TN GAM Models Phytoplankton
-------
Capture Probability of Phytoplankton Taxon Along TN Gradient
Anabaena circinalis
Anabaena flos-aquae
Aphanizomenon flos-aquae
o
•<-
_^> 00
:= d
o
CD
o to
O d
Q.
^ °
"-I— • -1—'
CD Q.
-80 $ g -
- 0 °- -
. .
.
..»M . •. «•••«»•
- T- P _
d "~
28 ^ra
d C i= °
CD .Q
T3 CD
T- c o to
-03 Od-
— ^ ^
'i= ii
CD Q.
. 8 "0 CO CM _
OQ: o °
- 0 °- -
.
• • .
.
• • *
•
0
m
•
•
D * •'
"• " • •
- 0)
d
CD CD
- r^ O
d C
CD
^ T3
" S 1
m <
00
- CO 0
d >
'•4— '
CD
. | 0
~ O
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
Aphanocapsa dellcatlsslma
Aphanothece nidulans
Chlamydomonas globosa
p _
T~
^> oq _
:= d
o
CD
o to
0 d
CL
0 •* _
3 °
"o.
CD OM
O o ~
0
d
.
.
.-.....:/ . •
ji p _
d "~
.88 ^» _
d C := °
CD ^2
. T3 CD
- 0 C .Q to
(— j 3 O o
f^ t—
^ °
"CD "o.
. 8 0 co tN _
do: o °
p_
.
.
•
•
-co P _
d ^~
in , m
- CM O 4? °° _
d C := °
CD ^2
T3 CD
f~ Q CO
• 5 =3 0 d "
o s—
"CD
.50
- 0
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
Page 1 of 9
-------
Capture Probability of Phytoplankton Taxon Along TN Gradient
Coelosphaerium naegelianum
& oo
:= d
CD
.Q
O
CL
0
-I—*
Q.
CD
O
0
ci
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
CN
d
d C
CD
CN "a
oo
O 0
d >
'•4—'
•* ^
o 0
Cryptomonas erosa
:= o
CD
O CO
O d
o
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
Erkenla subaequlclllata
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
£
0
O
CD
• ^
.Q
CO
O 0
d >
'•4—'
JD
p 0
d (V
Merismopedia tenuissima
Microcystis aeruginosa
Monoraphidium Capricornutum
p _
"~
_^> 00 _
:= d
o
CD
O CO
0 d
CL
0 •* _
3 °
"o.
CD CN
O o ~
0
d
.
,
. _ „ .f. •„.».. , .
. 0 P _
d "~
. s 8 ^0° _
d C = °
CD .Q
,o ~& ro
. 8 £= ^2 tp _
dl 1°
-00 0) ^ _
° -^ 5
"CD "o.
. 8 0 ro CN _
OQ: o °
-_, 0
. o d -
.
.
•
*
^
,
'. •' • .•
.00 P _
d ^~
P 8 £»»
d C = °
CD ^2
^. T3 CD
S C ^2 co
' ° E 2 ° "
< °-
- c8 0 0) "* _
d .> ^ °
"CD "o.
2 "0 ro CN
' d Qi 0 °
-_, 0
. o d -
.
.
.
. . . __ -. rf. ......
- o
0
- o O
C
CD
T3
c
^2
-00
^
- o QJ
o:
- O
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
Page 2 of 9
-------
Capture Probability of Phytoplankton Taxon Along TN Gradient
Nltzschla palea
Oocystis parva
Pseudanabaena limnetica
o
•<-
,>> 00
™ d
o
CD
o to
O ci
CL
0 •* _
^ °
-i—*
Q.
CD CN
O o ~
0
d
•
• **», *•»*«• •.«••••
- 0 P _
d "~
.08 ^» -
d C = °
CD .Q
T3 CD
_ o c -Q ^ -
0 3 O o
-° £-
< CL
-00 0) •* _
> 3 °
-1—' -1—'
CD Q.
- o 0 ro ^ -
o: o o
0 o
o d -
.
B
•
*. •
• *«*
- 0 P _
d "~
.88 £>» -
d C 1= °
CD .Q
rsi "^ "5
S C ^2 (D
' o E 2 ° "
< °-
-00 0 "* _
° > 5 °
-1—' -1—'
CD Q.
- 5 0 ro ^ _
OQ: o °
._. o
. o d -
.
*
*
^
d
0) 0
- o o
d C
CD
^ T3
- o ^
0 E
„ <
to
-00
d >
'-I—'
CD
- S 0
o Q:
~ O
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
Pseudanabaena mucicola
Rhodomonas minuta
Schroederia judayi
p _
^ 00
:= °
CD
o to
0 d
CL
0 •* _
3 °
Q.
CD CN
O o ~
0
d
0
•
o*
.. .n.. :.'v.--V.- ..
03 0.1 0.3 1 3
Total Nitrogen (mg/L)
. 0 P _
d "~
.88 £» _
d C :^ °
CD ^2
. 8 1 & ^ °
CD Q.
.50 CO . (JO
d C •= °
CD .Q
- 8 c S undance
00 <
- O 0
. s 1
o Q:
- 0
Page 3 of 9
-------
Capture Probability of Phytoplankton Taxon Along TN Gradient
Sphaerocystis schroeteri
Stephanodiscus parvus
Synechococcus elongatus
o
•<-
,>> 00
™ d
o
CD
o to
O d
CL
0 •* _
^ °
-i—*
Q.
CD CN
O d -
0
d
B
^ ,
'
*
' •" ' • 1 .1
0 •*••*••
. -?- P _
d "~
.58 3?» _
d C 1= °
CD .Q
T3 CD
^1 c .Q —
< °-
-00 0 "* _
° > 5 °
-1—' -1—'
_ CD Q.
-00 ,^ ^ °
'-I—' -1—'
CD Q.
- 5 0 ro ^ _
do: o o
._. o
- 0 d -
•
•
"**"" """** *** *"
d
^r 0
- o o
d C
CD
"^
O ^
~ • ^
<£
-00
O >
'-I—'
CD
-00
o Q:
~ O
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
Anabaena
Aphanizomenon
Aphanocapsa
p _
^
^* 00
:= d
o
CD
O (D
0 d
CL
0 •* _
3 °
"o.
CD OM
O d -
0
d
.
,
•
.
• , •
'
. • .
•' '.'*..'*
*" " •" • '
-ft P _
d ^~
^ ® >, m
- -
^ -
d .> 3 °
"CD "o.
. 8 0 ro c. _
OQ: o °
o
° o ~
.
f
• • .
• ^
B
.
• •
*
•
•
• • ' ••
• ....
. & p _
d "~
(D 0 >, („
d C •= °
CD .Q
. T3 CD
. u5 C ^2 tp _
d ^ O o
O I—
^ °
"CD "o.
.20 22-
o
° o ~
.
"
•
. ^ ,•
. . ..<._ .• .. ... .d> .
. ^
d
0) 0
-00
d C
CD
. fel
D
0 .Q
^. <
- o 0
d >
-ll
- O
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
Page 4 of 9
-------
Capture Probability of Phytoplankton Taxon Along TN Gradient
Aphanothece
& 00
:= ci
CD
o
O
CL
0
Q.
CD CN
O o
q
ci
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
CO
d
in 0
c\i O
d C.
CD
m T3
T- 0
o >
§1
Aulacoseira
& 00
:= o
CD
O CD
O d
Q.
0 •*
=i °
-I—*
Q.
CD CM
q
ci
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
Chlamydomonas
CM 0
co O
d C
CD
* c
^ §
0 E
CD
T- 0
O >
_ "CD Q.
§ 0 CO CN
OQ: o o
q
ci
,>> 00
:= ci
!O
CD
O (D
O O
Ql
0 ^
=i °
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
r^ 0
q O
d C
CD
•« c
q iz
d 5
t *^
o 0
d >
si
o a:
Coelosphaerium
Cryptomonas
Cyclotella
q _
^> 00
:= °
CD
O CD
0 d
Q-
0 •* _
Q.
CD CM
O o i -
0
d
•
•
•
. . .1... .. .V. •.. .»• . •
CM O _
d T^ ~
0
d C •= °
CD .Q
^ 1 ^ CD
" o = 00-
^ °
_ CD Q.
2 "0 CO CM
"do: o o
. o g -
•
•
.. — •. '_.._ • ..
d
CO 0
- - O
d C
CD
'I §
^
CD
- O 0
d >
.sl
o Q:
- 0
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
Page 5 of 9
-------
Capture Probability of Phytoplankton Taxon Along TN Gradient
Erkenia
Gymnodinium
Lyngbya
o
•<-
_^> 00
:= d
o
CD
o to
O d
CL
0 •* _
3 °
-i—*
Q.
CD CN
O o ~
0
d
.
m *
* * " •.*•*&. ••
"*
_ 0 P _
d "~
.§8 ^» _
d C 1= °
CD .a
^ T3 CD
O C O CD
d ^ O d
< G-
-80 CD "* _
d > 3 °
'•4— ' -1— '
CD a.
- o 0 tO CN _
da: o °
o
o d ~
•
•
• a
- 0 P _
d "~
.08 ^» -
d C 1= °
CD .a
in "° ro
_ o ^~ ^^ • —
f^ ^
< G-
-80 a) •* _
d > ^ °
"CD "o.
.00 CD CN _
OQ: o °
- 0 °- -
-«*.%"
""*"" ••••••• •*••
1 —
d
^r 0
- T- o
d C
CD
T3
^ C
d 3
.a
<
-00
d >
"CD
. 8 0
o a:
~ O
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
Merismopedia
Microcystis
Monoraphidium
P _
"~
^> 00 _
:= d
o
CD
O CD
0 d
ol
0 •* _
3 °
"o.
CD CN
O o ~
0
d
.
,
.
" " ™ •*• ••••— • •
. 0 P _
d "~
. s 8 £» _
d C :^ °
CD ^2
"O CO
. 8 £= ^2 ^ °
"-i—' -i—'
CD a.
. 8 0 ro CN _
da: o °
o
° d ~
.
.
•
^
,
'. •' • •'
.00 P _
d "~
_ ^ ^ >,co _
d C •= °
CD ^2
•* "° ro
^ C O CD
d ^ O o
f^ t_
^ °
'•4— ' -I— '
CD a.
-20 ^ g -
o
- o d —
.
.
.
.
- o
0
- o O
C
CD
T3
- ° §
.a
-00
>
'•4— '
JO
- o Q)
a:
- O
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
Page 6 of 9
-------
Capture Probability of Phytoplankton Taxon Along TN Gradient
Nitzschia
Oocystis
Pseudanabaena
o
•<-
_^> 00
:= d
CD
o to
O d
Q.
0) •* _
Q.
CD CN
O o ~
0
0
•
• •
• , t
' * ••••• •••q»» •oa«»* ••
- 0 P
d "~
T- CD ^>.
-00 -^ °° -
d C — °
CD .Q
T3 CD
. 5 c .£ cq _
d ^ O o
^ °
CD Q.
- o » -
d C 1= °
CD ^2
. 8 C ^
.sl
o o:
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
Rhodomonas
Schroederia
Sphaerocystis
P _
^~
^* 00
:= d
o
CD
O CD
0 d
CL
0 •* _
3 °
"o.
CD CN
O o ~
0
d
B
•
.
*
.
• •
*" • . .'
ot • "••*•** •* • •• fa
- -T- P _
d "~
^ G> >, m
- ^ O 4? °° _
d C := °
CD ^2
"O CO
_ 8 £= ^2 ^ °
_ "CD "o.
- 8 0 /O CN _
do: o °
p_
t
. '. .ij''' ...••» d. 'V.. I ..
T- P _
d ^~
CD 0) >,
- ^— O -*— ' • —
d C = °
CD .Q
"O CO
_ 2 C ^2 CD _
d =3 00
o s—
^ °
- "to H
. S "0 tD CN _
do: o °
P_
^
0
B
'
"
"
•:: . •_•._ ^ ... . ..
"j-i
d
in 0)
- - O
d C
CD
T3
^ C
• ^
f^
<
- o 0
d >
. s -i
- 0
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
Page 7 of 9
-------
Capture Probability of Phytoplankton Taxon Along TN Gradient
Stephanodlscus Synechococcus Synechocystls
p _
T~
_^> 00
:= d
o
CD
o to
O d
CL
0) •* _
13 °
"o.
CD c\i
O o -
0
d
0
B
.
*
•
..\ ..•..- ...«'».• . .
. $
d
in 0)
- to O
d d
CD
•Q
O) *—
d -^
^
- CO 0
d >
"CD
. ^ 0
o a:
- 0
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
o
-si'.
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
^t
d
^ 0
- O
d d
CD
oo "a
o £
in
q 0
sl
0
"o.
o
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
0
o
CD
T3
o 0
d >
JD
0
Bacillariophyta
Chlorophyta
Chrysophyta
p _
^~
^* 00
:= d
o
CD
O (D
0 d
ol
0 •* _
3 °
"o.
CD OM
O d -
O
d
f
•
^
^
0 • •
-00 P _
d "~
r- , m
-(DO •<-? m- -
d C := °
CD ^2
T3 CD
in C .Q (D
o 3 O o
.Q >-
^ °
"CD "o.
- - 0 , 03
d C := °
CD .Q
. T3 CD
. fe C ^2 (D _
d ^ O o
f^ S-_
^ °
"CD "o.
-20 Q g -
o -
t
o
• • .
. .A . •"•*- ••••"— "
00
d
^r 0
-(DO
d C
CD
"U
^F ^
~ • ^
f^
^
- CO 0
d >
"CD
.20
o o:
- 0
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
Page 8 of 9
-------
Capture Probability of Phytoplankton Taxon Along TN Gradient
Cryptophyta
Cyanophyta
o
•<-
_^> CO
:= d
.Q
CD
O CD
O d
Q_
0 •* _
Q.
CD CN
O o -
0
0
,,
- 00
d
is. ®
- o H
CD
« C
- "•! §
5
in
- CO 0
d >
'Si
°
Pyrrhophyta
>-.r
.
* *
" "
.
•
•
00
d
CD
o
^r
d
CN
0
~ O
0
o
CD
T3
f~
^
^2
<
•i=
CD
0
>,
-i—*
• —
o
CD
.a
o
0
^
"o.
CD
u
o
•<-
CO
o
CD
d
^r
d
CN
d
0
d
m
^
_
• '•
V N
d
O)
- O
d
Q
~ •
ID
- 0
d
CN
- o
d
0
o
c
CD
"^
f~
^
o
0
"-^
CD
0
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
0.03 0.1 0.3 1 3
Total Nitrogen (mg/L)
Page 9 of 9
-------
Appendix 3 - Wl Secchi GAM Models Phytoplankton
-------
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
Achnanthes minutissima Actinastrum hantzschii Anabaena aphanizomenoides
CO
^ °
-i—*
• — ^Q _
o o
CD
O d
Ql „
0 d
^
"o. d ~
CD
o - _
0
0
d
•
,
•
*
-
'» .
0 >, ™ -
_ 0 ~ °
- o C 1=
d CD .Q
T3 CD CD _
13 0 °
-Q "r
- o < °- ^
00 CD Q -
> 3
CD Q.
"0 CD ™ _
- o Q: O °
o
•
^- in
- ro 0 ->, Js ~
do -&1 "
Q :^
CD !Q CN -
^ T3 CD d
0 =3 0 in
,_ < Q- d
" ^ 0 CD o
•^ -I— ' o
JD Q.
0 o: OP-
o
- o p -
1— i
•
•
.
.!
•^
:••'.
CN
~ ri CD
(_)
C
- ^ CD
0 T3
C
_ o o
d <
in CD
- q >
o •4~*
CD
CN 0
d 0-
- 0
2 4 6 8 10 12
Secchi Depth (m)
Anabaena augstumalis
2 4 6 8 10 12
Secchi Depth (m)
Anabaena circinalis
2 4 6 8 10 12
Secchi Depth (m)
Anabaena flos-aquae
d ~
,^
— m. _
o o
CD
.a
0
Q- d ~
0
-1—*
a. T-
CD d
O
0
d
•
*
a
•
.
*
• *
*
/•"•_/! *.'
1 1 1 1 1 1
"J-l
d
0 d -
O ^~t
- CN C =
d CD .Q
T3 CD <^> _
C .Q o
_ ^ E 2
d *^ CN
.1 § °"
^ "CD "o.
" ° ^ Od"
o
O
•
•
.
' .
, _
[ '
•?/•••
«•*?:.'.,• . . ••
i i i i i i
o
in
O (_) _i_i
C =
CD .Q f
. ^ T3 CD d
o ^ -^
2 CO
co < Q- d ~
" d CD CD
> 3^-
•4-t 4-1 O
^r CD Q.
" § $ o - -
Ql **- ' o
o
o
•
*
*,•
. '. . •
* • • • •
•*£.' *.
1 1 1 1 1 1
- 1^
d
-
-------
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
Anabaena macrospora
in
ci
CD
O CO
O ci
CL
0 ™
3 d
-I—*
Q.
CD T-
O o
q
d
u s
2 4 6 8 10 12
Secchi Depth (m)
o O
d C
CD
« C
p iz
d 3
C\l
q 0
JD
q 0
d (V
Anabaena mendotae
o
co -
.Q
2 S
§ o
-I—* T—
Q- d
O
o
o -
2 4 6 8 10 12
Secchi Depth (m)
0
CO O
in C
d CD
T3
.Q
s <
d 0
>
J5
-r- ^ °
'-I—' -1—'
_ CD Q.
-30 /O CH _
do: o °
._. o
° 0
."'•
'.•.'*; .*••/•
* r . -: ' ' .
, \ * • • "
" • •
° *• " •
• «*0* •
J« "• • •
•.•:'•' '.-'•'•''. '
'' % " ' * .
s!$- •:'/• :
•iMjiJ'''! -'-_'" '-il •
- 0) |^
0 ci ~
s S >»S -
r^ O -i— • o
d C =
CD ^2 m
m T3 CD o
- 8 C -Q
dl Is-
O) „. CO
- co 0 0) d -
° .> U
ro 0.2 -
.20 co °
do: o -
0
0 o
° 0
.
.
• \m
.iL '•
1 1
CO
d
0
O
C
- co CD
d u
C
- |
- o J
"CD
0
-------
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
Aphanocapsa dellcatlsslma
Aphanocapsa elachlsta
& 00
:= d
CD
.a
o
CL
0
-I—*
a.
CD
O
o
d
u s
d
2 4 6 8 10 12
Secchi Depth (m)
CD
O -^
c = «>
CD o o
T3 CD
° E
0
. CD
00.,
d Q: o 5
§
O d
Q. „
0 o
"o. d
2 4 6 8 10 12
Secchi Depth (m)
0
r- O
d CD
T3
C
o 0
>
^
O) OJ
P C£
d
Aphanothece nidulans
p _
CD
O
-. to
. 2 0
o o:
- 0
2 4 6 8 10 12
Secchi Depth (m)
Arthrodesmus incus
Arthrodesmus octocornis
Arthrodesmus subulatus
in _
d —
,^ ^
•.= o
o
CD
O CO
0 d
CL
0 CN
3 °
-1—*
a.
CD T-
O d -
0
d
•
•
. •
•
.
• "• .•
i
- "~ d
0
^0 >, ^ -
^— O !•* o
d C =
CD £) -*r
8 1 S ° ; "
" ° ^ n2 " -
< Q- o
-80 ^ 3 °
° "CD Q.
- "55 to - _
- O QX O O
O
0 o
° 0
•
q
•
1 --
d
^r 0
- ^ O
d C
CD
T3
d 3
.a
<
- O 0
d .>
'•4— '
CD
. 8 0
o a:
~ O
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Page 3 of 33
-------
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
.Q
CD
.Q
O
CL
0
-I—*
Q.
CD
O
Asterionella formosa
to
0
±± o
q
o
2 4 6 8 10 ^2
Secchi Depth (m)
°> 0
5 g
CD
T3
C
*- .£2
5 <
0
"-^
JD
•<*• 0
2 o:
.Q
CD
.Q
2
Q.
§
-I—*
Q.
CD
O
in
CN
o
o
CN
o
in
o
o
o
8
o
o
q
o
Aulacoseira ambigua
2 4 6 8 10 12
Secchi Depth (m)
- CN CD
:j=
D
Aulacoseira granulata
in
d
>, -3-
i= d
CD
O CO
O o
CL
0 ™
° £ O °
q
o
2 4 6 8 10 12
Secchi Depth (m)
1^ 0
in O
d d
CD
^ "O
o, <
OM 0
O >
6 a:
Bitrichia phaseolus
Botryococcus braunii
>,
-1— »
• —
o
CD
.a
0
CL
0
3
"a.
CD
O
O _
T~
00
d
CD
d
•* _
o
CN _
0
d
.
^
.
'
B
"
• •
.
. CD
d
CO
- O
d
- 9
- O
d
- 9
o
0
O
c
CD
-o
c.
13
^
0
>
"CD
0
>,
'•^
o
CD
o
o
0
3
"Q.
CD
CN _
d
o
CN _
d
^2
Q ~
° _
o
§ _
d
o
o _
r~ i
.
•
• •
*
•
*-
- f
d
CO
- CO
d
CN
Q
r^
d
. S
d
~ O
0
o
c
CD
"^
f~
D
<
0
>
'ro
0
a:
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Chlamydomonas globosa
CD _
in
0
. —
^
d ~
CN
d
o
0
d
t
•
•
o. •
• *
. " .'.
1 ii*' t.J:^
. •
• '
-
. J 0
o 0
CD
-o
c.
- § 1
d <
0
>
"-^
CD
CO 0
- q fy
d
~ o
2 4 6 8 10 12
Secchi Depth (m)
Page 4 of 33
-------
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
Chlamydomonas incerta
Chlamydomonas inepta
Chlamydomonas platystigma
CN
d
.^
±± in
O o
CD
o
£_ o
d
0
-•— ' lf\
Q- § -
8°
O
O —
t— i
•
•
* *. *
•„ ''
. '•
- 0
T— _
0 >, °
in O ±i
- o C :=
d CD .Q
~C CD 0
f~ O
CO ^ 2 °
" o Z! in
rxi "*""* •*-* O —
. g CD Q- d
d 0 ro
o: o
o
- o o —
l—i
•
.
in
6
- o
d 0 >,
O *~* • —
CD !5
T3 CD
. o ^ ^ m. -
O l—
< °-
0 2g-
O -^ -i— '
o CD Q.
S. o5-
o
o o ~
•
•
•
%
• ,
• .
. •"*• •.• •
1
- 0
d
CO 0
- o O
d C
CD
CO "°
C*J ^
- O *—
. ^
_Q
CN <
-00
d >
'•4— '
CD
-00
d Q:
~ O
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Chroococcus limneticus
o
CO
E 8
2 °
Q-
0
CD
O
o
d
kr C
d ro
0
"-^
CD
2 4 6 8 10 12
Secchi Depth (m)
Chroococcus minimus
Chroococcus minutus
^_(
^^
o
5
0
s_
2
"o.
CD
O
in
CN _
Q
"CD
o 0
~ O
-i_«
^^
o
CD
^2
0
0
3
"o.
0
O _
•^
°P _
o
(D
d -
^r
0
C\l
d
0
d
f
•
;
*
• .
**"• •
'-t .
o
d
^.
- O
d
- q
OM
- O
d
- q
~ O
0
o
c
CD
T3
3
^2
0
>
"CD
0
a:
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Page 5 of 33
-------
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
Chrysococcus minutus
Chrysolykos planctonicus
Closterium moniliferum
o
"-
^ 00
:= d
o
ro
o to
O d
CL
0 •* _
13 °
•4-'
Q.
ro CN
O o -
0
d
•
•
*
. . •
1 . •• *
^ CD
_ O • —
d °
d C 1=
ro .Q — .
~o ro ._; -
. 5 £= .£2
d ^ O
O !— ro
< 0. 2 -
-50 CD
d > ^ CN
'•4- * •4-' d
ro Q.
- o , 0
d O ,+i
ro !Q
o "o ro
C o y
3 O d
• ° < ol
0 2
- o -^ 5 2 _
JD Q. d
- o £ O
o
- o p -
1— i
•
.
.*.r*t. j •
-
0
. 2 c
d ro
T3
c
3
. g <
0 0
_ "-^
ro
. s£
0
- 0
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Cocconeis placentula
Coelosphaerium naegelianum
Cryptomonas erosa
CD
d
•4- »
'— "} _
.0 ^
ro
0 d ~
Ql oo _
0 d
-3 ^
Q. d
ro
o - _
o
0
d
t
.
•
.J'»> ' *' '
p _
T—
!s§ tS-
d ro !Q
T3 ro
C ^2 CD
13 0 d
. 8 < £
d 0 0) ^ _
> 3 °
"ro "o.
.si o2-
d
„ 0
- 0 d -
t
'
.
.. •!
»&y> •
. K ° -
d *~
3 8 3?»
d C = °
ro ^2
,%, "^ ro
. c^ £= ^2 D
"ro "o.
. ^ -0 ro CN _
d Q: o o
-_, 0
. o d -
B
•
• •
* *
• •
• a
t •
0 • •
: " ".
8
1 *•••"•"• '
* • *
" • , b •
' ..' ,'.»•' • '. • '
J ••*§ . " I.'?"' t .' • '
••JEj^KJiSSf *•• ' *jr'm l ' '
" 1 "*" * l" "
00
d
^r 0
- CD O
d C
ro
•Q
^F ^
~ • ^
<
- CO 0
0 ^
ro
- 1 ^
- o
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Page 6 of 33
-------
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
Cryptomonas gracilis Cryptomonas lucens Cryptomonas ovata
:= cq
:5 o
CD
.a
o
2
0
-I—*
a.
CD
o
q
d
2 4 6 8 10 12
Secchi Depth (m)
83 P
d "~
«, 8 *•>.
CN C = °
d CD .Q
T3 CD
-. ^ O d
0
o I
d 0
a:
-i—*
Q.
CD
2 4 6 8 10 12
Secchi Depth (m)
o 0
J5
o 0
a:
q
d
o O -&
d C 1=
CD .Q
T3 CD
5 £ .Q
Q. o
CD
O -
o
2 4 6 8 10 12
Secchi Depth (m)
0
a> O
co d
d CD
T3
C
.a
8 <
d 0
JO
§ a:
o
o
Cryptomonas rostratiformis
CD
d
CD
^2 •*
O o
CO
0 d
^
"o. d
CD
o -
o
q
d
o
o
2 4 6 8 10 12
Secchi Depth (m)
Cyanocatena planctonica
Cyanogranis ferruginea
CD
o
CD
T3
f~
^2
0
>
^
C£
^^
^_»
^^
o
CO
o
2
Q.
2
"o.
^
^.
d ~
CO
d
C\l
d
^ -
0
d
t
*
,
B_° ^ ^_
CO
~ ^7
Q
,_
o
r^
- o
o
CO
_ o
~ o
CD
^)
f~
CD
T3
C
^2
<
0
>
J5
CC
IT) _
0
^^
-I— ' . —
r^ o
o
CD
O CO
0 d
0 0! _
3 °
"o.
to ^ _
O °
0
d
|.
•
:
•
• •
*• d •••
o
d
^
- O
d
_ o
d
- O
d
- o
— o
0
O
c
CD
T3
f~
^2
0
>
"CD
0
a:
2 4 6 8 10 ^2
Secchi Depth (m)
Page 7 of 33
2 4 6 8 10 12
Secchi Depth (m)
-------
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
Cyclostephanos invisitatus
Cyclotella meneghiniana
Cyclotella ocellata
CD
d
-i—*
™ ID
'o d
CO
O o
£ co
0 d
^
Q. d ~
CD
O ^ _
0
0
d
. „
,
*
,
f*-
• _ - '
^ •»
_ — - •" "
_ ; ! "*
' -•
LO
CN -
O -i— •
- ^ C ~~
0 co !Q CM -
"D CO d
1 1 5 _
3 < °- 0
00 0)
> 3 - -
_ -i—' -i—' O
CD Q.
•* 0 ^ 8
- g o: OP-
O
- o o —
e- \
m
p _
^~
0)
- CD „
H 0 >, _
o O £? °9 _
C • — o
- in CD !Q
d "O CD
C ^2 3 °
o •|~' ••— '
CD Q.
-* "0 CO CN _
d '•'-
- o P -
.
.
••"
. 6
d
co CD
- o O
d C
CD
CN "2
- o c
°E
CN <
CN
-00
o .>
CD
-00
o Q£
~ O
2 4 6 8 10 ^2
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 ^2
Secchi Depth (m)
Cylindrospermopsis raciborskii
Deasonia gigantica
Dictyosphaerium pulchellum
00
d
..^
^-i
'•^
o ^P —
CD °
^2
0
Ql ^ _
0 °
3
"o. CN
O
0
d
•
0
•
-.
•
r.
*^*
>i • • •
u^
. ^ o -
d
CD >, o
O *J CN _
^ C = d
- co CD .Q
d T3 CD ^
1 s s -
- CN < Q-
° 0 2 ° -
> 3 o
^ "CD "o.
- d 0 CD § _
Q; O d
o
- o o —
(— i
•
.
•
,
.
•° t« .
•* *" " :
•?- P _
d ^~
co CD >,
^— O -*— ' • —
d C •= °
CD ^2
T3 CD
•<- C JS CD
- o _g 00-
,„ < °-
-80 g * _
o .> 3 °
"CD "o.
. o "0 CD CN _
OQ: O o
0 o
0 o
•
• '
q
0
"
•" . 1 • . . s
1 1
V N
d
0
- ^ O
o C
CD
. T3
- o £=
3
0 .Q
„ <
in
- o 0
o .>
"CD
. 8 0
o a:
~ O
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Page 8 of 33
-------
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
Dinobryon bavaricum
Dinobryon divergens
Dinobryon seratularia
r^
o
CO
.i_i
'—
'o . _
CD °
O x -
L_ O
Q_
CO
2 d ~
-t-1 CN1
0- d ~
CD
O -
d
0
d
^°
' »
.
'•!-' . _
— ^~ o
0
LO
O -i— •
^ C 1=
- ^ CD .£2 ^r
o T3 CD d ~
(~ O
- ^ O
O I— CO
^ <; CL d
° > § ™ -
<= <-• d
JO Q.
, ^.
o o £? ^ _
C = °
12 CD .Q
^ "O CD
0 C jS co
^ 3 O d
^ ^2 £-
d < Q-
-8.1 §S-
CD Q.
-* "0 TO - _
- 0 QX O 0
0
o_
0
.
•.
'•
" ' •
•
. 0
d
CM CD
-00
d C
CD
^ T3
- o ^
• ^
.Q
-00
d >
'•4— '
CD
- o CD
o:
- 0
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Erkenia subaequiciliata
Fragilaria capucina
Fragilaria construens
p _
^~
^^
£? °P _
:= °
o
CD
O CO
0 d
Q-
0 •* _
5 °
"o.
CD CN
O o ~
0
d
.
_
. • "
• •
0 •
• *
• • • , „
•^A...- :.v. ' .
\- 1
° 0
CO -
OM CD >, °
_ ^r O &
d C =
CD .Q
OM f. 5o
_ co c -Q c\i _
d _g 0 o
>
- 8 2 = u,
d CD ^2 ^- -
T3 CD d
C ^2
-§ 2 o
U3 << Q_ T- _
o 0 0) °
.^ 3
CD Q. |fi
"m CD • ~
. o c£ O °
d
o
- o o —
(— i
.
• •
.• "^ •
0)
- 00
d
0
O
- co CD
d T3
^
_ -^- <<
0 0
.>
CM ^D
- ^H ~*r:
° Q^
~ O
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Page 9 of 33
-------
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
Fragilaria crotonensis
Gomphosphaerla lacustrls
Kephyrion gracile
in
CN -
.i_i
•= o
.Q CN -
CD d
O "5
CL d ~
f o
-^ •<- —
<-• d
Q.
CD in
0 P-
o
O —
t— i
•
•
,0
•
•
•
. •
" *"" •"••• "
° -1
d
. T- in
d CD >* ^ —
O -*— ' ^
^- ^ i^
- ^ CD !Q ^
0 T3 CD ,-;
C ^2
' sl Is-
r- CD P o
- P .> D ° -
O -i—' -i—' (—i
CD Q. °
ro m CO m
- P ry O ° -
o LH ^o
o
i— i
i
r
1
t
t
1
t
t
l
\ *
V
\ \ ^
\'\ ^
* *? \ x
'''* v^'--
JpV" * ^~~~>_J1~ ~
OM
d
OM
- P o> >, „
O Q J? ' • -
C. ~^
r^ CD O
" d ~° ro
C ^2
^3 O OM
- P -9 rT d ~
0 < Ll-
0 0)
- q .> ^
CD Q. H "
"fl5 CO
- 0 Q, 0
o
O
0
.
•
*
•
•' >
*- 0
- O Q
d c
CD
T3
C
T— ^
d ^
0
'•^
- o CO
0
o:
- O
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Kephyrion planctonicum
Lyngbya circumcreta
Lyngbya lagerheimia
o
(D
>> O
-I—*
5 "•! -
CD °
S d -
0 d -
^
-i-' CN
Q- 2 -
CO
O -
0
0
0
.'.
•
:
d CN _
d
0 -^
£= = in
- o CD o T- _
"O CD d
C .0
- o 3 O
- ° 0 0) °
.> 3
4-^ ^-^
" ° CO o in
~0) /O g -
- o o: °
o
- o o —
r~i
a
•
'
•;
;
.-. ;
^
o -2? °
f— -^
- o co !Q
d "O CD <*} _
C ^20
§| 1
d 0 CD d "
> ^
'•4— ' -I— '
CN co a.
- P -0 co g -
0
o
•".
••
V.
' 1 8
c
CO
- o "O
0 «=
13
- o
d CD
•4—t
CD
-00
a:
~ O
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Page 10 of 33
-------
CD
CD
CD
.Q
0
-I—*
Q.
CD
2
q
CD
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
Lyngbya limnetica
2 4 6 8 10 ^2
Secchi Depth (m)
8§
Si
CN .Q
0
"-^
JD
P rv
CD
Mallomonas akrokomas
:= o
.Q
CD
-Q co
P CD
U °
-I—*
Q.
CD ^
O CD
q
CD
2 4 6 8 10 12
Secchi Depth (m)
00
q
o
8 0
° £
sl
CD CC
Merismopedia minima
fl> >,
O ^
CD !Q
T3
CD
O CN
CL °
Q.
CD
O
q
CD
2 4 6 8 10 12
Secchi Depth (m)
CD 0
P o
° c
CD
T3
s i
0
J5
0
a:
Merismopedia tenuissima
Merismopedia warmingiana
Microcystis aeruginosa
>* *°
•*— ' O ~
'-^
CD
0 •* _
CL °
0
^
Q_ • —
O
0
CD
t
s
^ •
IT)
fD 0 >, °
Q O .-ti
CD !Q o -
T3 CD
- ^ _§ 'O co
0 < it °
ill CN
O) '^ +_i °
- P CD Q.
0 0 /O _
o: o CD -
„ o
° 0
a
•
.
"
t
tV ••
q _
00 ^~
- o
° ® >, m
0 ^ °P -
CD C — °
- P CD ^2
0 T3 CD
C .Q ^ °
0 '^3 -£i
CD Q.
OM "m CD CN
O fT*, f*\ Q
o ^~
o
o
B
• u
• .
.".
. ,
»• . .
,1
• -r*^ •
*•' .
t •• *
." " . .
" " B .
^f'?:*;i ._
sSa^Sjfeji' .•••' . .
O)
CD
0) 0
- h- O
CD C
CD
- S §
o
™ <
O)
- CO 0
CD >
' ~
CD
-2£
~ o
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Page 11 of 33
-------
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
Monomastix astigmata
& 00
:= ci
CD
^2
O
CL
0
-—
Q.
CD CN
O o
q
o
O)
q
o
CD
0
2 4 6 8 10 12
Secchi Depth (m)
Monomastix minuta
Monoraphidium arcuatum
>>
'—
o
CD
.Q
O
s_
Q.
£
"o.
CD
IT) _
O
•*
o
CO
CD
•^ _
o
s -
0
0
.
• • •
. /
- 0
CD
- O
— O
- 0
- o
— o
0
o
C
CD
T3
f~
3
o
0
"CD
0
a:
-i—*
I —
O
CD
O
CL
0
3
"o.
CD
O
S _
0
_
o
CN _
CD
-
°.
CD
-
O
O —
l—i
.
If&S^nt. : '•'
. 8
0
_
CN
- O
CD
-
- O
CD
- O
— O
0
O
C
CO
T3
C
^
<£
0
>
"CD
0
a:
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Monoraphidium capricornutum
Navicula cryptocephala
Nitzschia acicularis
q _
^~
^> CO
:= °
o
CD
O CD
0 CD
ol
0 •* _
3 °
"o.
CD CN
O o ~
0
CD
dlnli*
» " •» • •
_ ,-
CD
CD _
O) 0 •>. °
- O O -&"
CD C = ^ _
CD ^2 °
. T3 CD
- 0 C .Q •*
CD ^ O O
f^ S-_
< ^- « _
-00 0 CD
CD > ^
'*3 J2 CN
CD Q. CD
-12 o^_
o
„ o
° 0
.
t
.
_
•
0* 1
q _
^~
0 •>.
O i^ °^ —
O C "~ ^
o CD !Q
T3 CD
C .Q -
. 8 < °-
00 0 ^ -
•^ 3 °
"CD "o.
"0 CD CN
- 5 Q^ O o
O
o
O
•
.
.
•• *
• •
! • •
*" • •
.
•
. 0
0
T- 0
- o O
o C
CO
~u
_ o ^
f^
<
-50
0 >
"CD
-00
o:
— o
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Page 12 of 33
-------
>,
to
- 0. 0
- 0
2 4 6 8 10 12
Secchi Depth (m)
Oocystis parva
Oocystis pusilla
Oscillatoria agardhii
p _
"<~
^> 00 _
:= d
o
CD
O (D
0 d
CL
0 •* _
3 °
"a.
CD OM
O ci -
0
d
.
•
„
• . '
• • • • •
JjjrX;. ,•! . '••.'•
1
V M
d ID
d
. 8 8 .&
o C — ^ _
CD !Q d
. T3 CD
o £= ^2
' o =5 2 o ~
< CL °
-90 0 CM
"CD "o.
-°£ o ^ -
O fr ^— ' O
o
° o ~
.
•
•"•« * t
(D
- 0 °'
d
0 ^. u^
^ O *-• d ~
. o C =
d CO ^2
T3 CD 2 -
- o 3 O
° < £ s -
0 CD
" ° .> D ^
"5g "Q_ °
" ° f^ O ^ -
o
o
.
.
0
•
.
.
:••
00
d
0
O
c
- LO "O
d C
d
-------
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
Peridinium cinctum
in
d
CD
.a
o
CL
0
-I—*
Q.
CD
O
0
0
2 4 6 8 10 12
Secchi Depth (m)
o
0
co O
d CD
T3
CN C
Peridinium umbonatum
*•>.-
CD
.£! CD
O o
° £ 3
JO Q.
- tD /O
o a: o
o
2 4 6 8 10 12
Secchi Depth (m)
CD
q 0
8l
Phacotus lendneri
s 8 *«?
d C 1= °
CD .a
§1 5«
0
"o.
o
2 4 6 8 10 12
Secchi Depth (m)
in 0
o O
d C
CD
_ T3
o 0
d >
'•4—'
JO
o 0
o a:
Pseudanabaena limnetica
Pseudanabaena mucicola
Pyramichlamys dissecta
p _
T~
_^> 00 _
:= d
o
CD
O CD
0 d
Q-
0 •* _
3 °
"o.
CD CN
O o -
0
d
.
a
0
•
*
m
'
. " ".
• •' •.
**'"•' '.""•.*.
""fc'^f**! ° *
•ttvii :il' rt"v,.i •.•.-.'. . . _____
- T- P _
d "~
28 -^ °°
d C = °
CD ^2
T3 CD
" d 13 0 d ~
^2 <—
3 °
"CD "o.
- 8 0 /O CN _
da: o °
._. o
° 0
€
•
•
•.
• *
> • .
. v«-
"• • a
' '{!<
o %v.y ",
talti.Lg.:. ;..'
d ^ -
o
CN 0 >,
- o O -t-t —
d C =
CD o
.. T3 CD CN _
_ 8 C ^2 d
d _g 0 °
<£ Q- -
-50 ^ o
"CD "o. d
- o 0 to
d a: o
o
- o o —
(— i
.
•
B
^
" t •
*\ 1 "
r! .
~
^r 0
- P 0
o c
CD
T3
C
^
CN ^2
- P <
5
"CD
^ 0
- P a:
o
— o
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Page 14 of 33
-------
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
Quadrigula lacustris
Rhabdoderma lineare
Rhodomonas minuta
o
•<-
_^> 00
:= d
!a
CD
J3 <° _
O o
CL
0 •* _
13 °
^— '
Q.
CD c\i
O o -
0
d
-
^
~r-^:^^
•
K » _ J
'. " ~ •
- CN CO -
d d
. - 8 ^ -
0 c =
CO -Q o
"D CO ^ -
C o o
d ^ O
-Q /t-
< Q- -
- ^ 0 2 _
0 > ^ °. _
^-^ ^-J •
CD Q. °
00 to
- g £ o
O
r-i O
— LJ * — ; ~~
r-i
.
*
.
* a a
• |
_ o P _
d "~
8 £•«
c i= °
T- CD ^2
- P T3 CD
o c o oo
3 O d
0 0) ^ -
- o .> 3 °
° ro Q.
"0 CD to
- o £ O o
IO
_ o d -
;
• . " "
.....'• '
ftii&'iVii iri
•
.
*
•
* *
*
rft*c:.f &•.
_ 0)
d
n 0)
- r^ o
d C
CD
•Q
U3 ^
~ • ^
^2
r^
- CO 0
d >
•+—I
CO
_20
~ o
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Scenedesmus abundans
Scenedesmus bijuga
Scenedesmus quadricauda
o
CO —
-1—*
• — _
o
CO
o Q
2 d
ol
0
3 0
-1— ' T— _
CD
O
o
O —
r~i
.
•
•
.
^ • "B
C\l _
d
- o
CD >, o
O *J OM _
- o C := d
CD .Q
"O CD
C o
" ° =3 0 d ~
_Q !_ "
° 0 2 ° -
.> 3 °
- r~i •|~' ••— '
0 to Q- ^
0 ,to § _
- o Ql O d
o
r~ i O
— u — ; ^
(—1
.
•
i_3nC£ -f 1
.6 cxi _
d d
co 0) >,
- o O -t-t y-j
d C = T- _
CD .Q d
-. T3 CD
. o £= -Q
f^ ^ O
.800°
° > ^
"CD Q. § _
.50 to d
o
r~ i O
— u — ; ^
(—1
.
• J .• •
. s
d
0
CD O
_ o ^"
d CD
T3
C
^
^" O
0 <
0
. 8 1
d 0
a:
~ O
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Page 15 of 33
-------
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
Scenedesmus serratus
Schizochlamys compacta
Schroederia judayi
d
^ CM -
: — d
.Q
0 o
Q_
0 ° _
3 d
£S
0 d "
o
•
•.
d °
d c i= ° ~
CD .a
,_ "O CD
_ o C .Q co
d ^ O d
^ °
CD Q.
_ fl) CD •>-
a: o °
o
o
X
/
X
^'
' ''''^-~^-~~~~~~~
_______ — '--•_ — "\
d "~
. •* o ^ °p _
o c =o
CD la
T3 CD
co c o to
d 3 O d
< ^-
- 2 | § 5 -
CD Q.
T- "m CD CM
- d £ O o i -
._. o
0 d
•
"
••s ••' •! •
d
CO
^
- ^ 0
o >
§ 1
o a:
2 4 6 8 10 ^2
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
.a
CD
.a o
o T
Ql °
0
^ in
Q. °
CD °
O
Selenastrum minutum
Sphaerocystis schroeteri
Staurastrum paradoxum
o
c
CD
T3
C
^
^2
0
>
"Jg
"0
a:
>,
-i—*
• —
o
CD
^2
O
!_
Q-
2
"o.
CD
O
O _
T~
00
d
(D
d
^r
0
CM _
0
d
.
.
•
.
• •* a
• •""*"•* •
(•• £t» %-« J |f o
ttfcZ**- *«• M' jiff • *
i ,
. r^
d
CM
- to
d
$2
•
- m
d
- 5
O
0
o
c
CD
"^
f~
^
•Q
0
>
'ro
0
a:
-i—*
^ —
o
CD
O
CL
0
3
"o.
CD
O
O _
T~
00
d
(D
d
^r
d
CM _
0
d
.
.
.
•
• *
g
o *
0 o .
. K
d
. 5
d
CO
~ •
CM
_ CM
d
d
- o
0
o
c
CD
~u
f~
3
^2
0
>
"CD
0
a:
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Page 16 of 33
-------
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
Stephanodiscus hantzschii
in
d
CD
O ". -
CL °
Q.
CD ^
O d
q
d
2 4 6 8 10 12
Secchi Depth (m)
§
d
CO
Stephanodiscus niagarae
O £>
™ C 1=
o CD .Q
T3 CD
0 . Z! 0
o
o
0
O
CD
T3
C
.Q
<
0
"CD
- - a:
2 4 6 8 10 12
Secchi Depth (m)
Stephanodiscus parvus
:= o
!a
CD
O CD
O d
0 "*
3 d
"o.
CD CN
/ .
u
2 4 6 8 10 12
Secchi Depth (m)
£
00 0
r- O
d C
CD
^ "O
O)
CO 0
d >
'•4—'
CD
2 0
o o:
Stichogloea olivacea
Synechococcus elongatus
Synedra cyclopum
CD
d
•4- »
'— "? _
o o
CD
^2 •* _
0 o
0 d ~
^
"o. d
CD
o - _
0
0
d
t
.
.
•
. •
. • "* • .
p _
•<-
;s§ is-
d CD !Q
T3 CD
C ^2 CD
M 1
o 0 0 ^ _
> 3 °
"CD "o.
"0 CD CN
. 8 & 0 ° -
o
p_
t
•
• .• •
.. r.va .•; .-..»', ^ .-. ..
. 0 ^ _
d °
in 0) >, ^.
- O O -!-•.—
d C = °
CD ^2
_,. T3 CD
. S C ^2 CO _
d| !°
.00 0 0! _
°' £ 3 °
"CD "o.
. 5 0 ro - _
OQ: o °
p_
B
.
«-l
- CD
d
^r 0
- in o
d C
CD
T3
. ¥ £=
° E
<
- CN 0
d >
"Jg
d a:
- 0
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Page 17 of 33
-------
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
Synedra filiformis
Tabellaria flocculosa
Tetrastrum staurogeniaeforme
o
"<~
_^> 00
:= d
!a
CD
.Q CO _
O o
Q_
0 •* _
3 °
Q.
CD CM
O o ~
0
0
" X
• t
S
• ^
|J%A|5| « a i ! * •
1
- ^ d
o
(D 0 >,
- •<- O -•-• CD
d C := d ~
CO _Q
. - c 5
° D 2 ^r
^
CD Q. ™ _
-00 tO °
0
d
*
•
•
* *
d
_ undance
- 0 <
0
CD
0
- o (V
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Woronichinia naegeliana
Achnanthes
Actinastrum
d
o
-I— ' CM —
:= d
tO ,n
^2 !£ _
2 °
Q-
2 ° -
"o.
to §
O d
o
O —
(— i
•
*"
a *
m m
' •
"
d
(D _
0 d
- o o -&"
d C = ^ _
CD ^2 °
co "° ro
d ^ O o
=5
'*3 J2 CM
CD Q. d
-IS o -_
°
._. o
° 0
•
_
1
0 ^ °°. _
m 0 -^ °
- r^ C :=
d CD .Q
T3 CD tp
C _Q o ~
E 2
5 < CL ^
00 £ C) ~
"CD "o.
•0 CD ^ _
. £ £ 0 °
d
o
d
•
B.
- 00 0
d O
C
CD
rn "^
" . f~
^
^
- 5 |
"CD
-sg
~ O
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Page 18 of 33
-------
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
Anabaena
Ankistrodesmus
Aphanizomenon
00
d
>,
-i—*
• —
!Q co
CD d
.a
o
£ •* _
0 o
3
"o. (M
5s-
0
d
q _
*^
fO 00
:= 0
o
CD
O CO
0 d
CL
0 •* _
13 °
-i—*
Q.
CD CN
O d -
0
d
a.
\
"A"
vV
V \
°" ^V
. v
*v\\
V\ V
^ \ ^
\ \
.• . • A;,
. • . x\ ^
• \ X v
i :••'•''. • vs^.
.•l^a.iK..: '.^J. V^^L:
•^ ' *" ~"»»«
1 1 1 1 1 1
2 4 6 8 10 12
Secchi Depth (m)
Aphanocapsa
-
t
.
°
t
. » '
«t.%. • . •
q _
O) *~
- CO
° 8 ^oo _
C 1=0
CN (Q Q
~ i cj CD
C o co
3 O d
O !—
. S < Q-
00 0 •* _
> 3 °
1^ CD Q.
- 5 0 /O ^ -
0 a: o o
„ o
o 0-
r\i
. S P -
d "~
CD ">k
LO f\ j^* 00
" o C := o
CD !Q
. T3 CD
co C ^2 ^ °
_ "CD "o.
-00 ,^ CN _
d C£ O o
._. o
o 0-
OO
. CO
d
in
^-0 >, d
. in o ••-•
d C =
CD ^2 ^ _
T3 CD o
. ¥ C ^2
d =3 0 „
< CL o ~
- CN 0 0
d > ^ ^ _
'*3 J2 O
CD Q.
3 0 CO ,_
o a: o d ~
o
d
•"" " • .
8. :•'' '• '' •
!•, • •
• • • •
"* °«a ,\a " ,
•• "« ^ ° •
•*
'•i
' •'« f!i .'••'•'
• !»*' ' "• .' •
!^*-:<-: ;i /
«tiia-"* -ii- 'J
"* ^ ••••
1 1 1 1 1 1
2 4 6 8 10 12
Secchi Depth (m)
Arthrodesmus
•
"
*
* B
*
q
'.
• a
0
. . J ' * 1 '
•4 1'
- 0)
d
00 0
- r- O
d C
CD
00 ^
d ^
1-1 o
O)
- CO 0
o .>
ID
. 2 0
o a:
~ O
CO
" H 0
O
. 2 CD
d "O
C
^_ 3
° <
co
d -*-1
CD
CO 0
- o -^.
d LL-
- o
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Page 19 of 33
-------
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
Asterionella
Aulacoseira
Bitrichia
to
0
±± o
.Q
CD
.Q
O
CL
0
-I—*
Q.
CD
O
q
o
2 4 6 8 10 ^2
Secchi Depth (m)
2 0
O £-
CD
T3
3
o <
0
"-^
JD
•* 0
2 o:
IT)
o
CD
O CO _
O d
Q_
0 ™
3 °
-I—*
Q.
CD T-
O o
q
o
2 4 6 8 10 12
Secchi Depth (m)
r^ 0
in O
d d
CD
° E
o, <
CM 0
d >
& 00
:= o
CD
O (D
O d
CL
0 •*
CD Q.
3 "0 to CM
o
q
d
U 8
2 4 6 8 10 12
Secchi Depth (m)
ro 0
o O
d C
CD
q
0
O 0
d >
'•4— '
JD
o 0
Botryococcus
Chlamydomonas
Chromulina
d
0
-I— ' CM —
:= d
f in
-Q !£ _
0 d
CL
0 2 _
^ d
"o.
f o
O d
o
o _
r~ i
s
•
• •
V-
. y p _
d ^~
CO 0 >, m
- co O 4? °° _
d C = °
CD .Q
,n ~& TO
. CM C ^2 D °
"-i—' -i—*
oo JO Q.
- o 0 tO CM _
OQ: o °
0
d
s
*,
•
.
•
':•'.
.. .-. \ ..•;"• • •
•ki}». ji.:Lj':^-! ."• '
CO O _
d T^ ~
•* tD >, „,
. CM O 4? °° _
d C = °
CD .Q
T3 CD
f~ Q CD
' 5 =3 0 d ~
o s—
^ °
'•4— ' -I— '
CD Q.
.80 tO CM _
OQ: o °
0
d
g
*
,
• .
B
t
•
. °.
*
. • . 4. ' .
'l.-i ..'*.• \ ' *
i
CM
d
to 0
- - O
d C
CD
"^
^- ^
~ • ^
f^
m <
00
- O 0
d >
. s -i
o Q:
~ O
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Page 20 of 33
-------
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
Chroococcus
Chrysococcus
Chrysolykos
o
•<-
_^> CO
:= o
!Q
CD
J3 c° _
O o
CL
0 •* _
3 °
-1—*
Q.
CD CM
01— ; ~
0
0
0
__--""
" a- . /. -7Tr:T:rr:rrrrr:
-CO P _
0 "~
- 2 § & 2 -
CD !5
^ T3 CD
2 3 O o i ~
< '-L-
- CM 0 CD ^ _
0 > ^ °
'•4— ' -1— '
CO Q_
2 "0 <0 ^
O Q^ O O
o
o
0
"
0 •
V. .. • .
i_^ CD
— O • —
0 °
CM CD >, ^ -
- o O •& o
o C 1=
CD .Q — .
„_ T3 CD 0' -
_ o ^ -^
o =3 0
_Q >— co
< CL 2 -
-50 CD
0 > ^ CM
'-i-> -t-t O
CD Q.
- o
CO
0
- o fy
~ o
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Closterium
Cocconeis
Coelastrum
in _
o ~
^ * _
:= o
o
CD
O CO
0 d
CL
0 CM
3 °
"o.
CD T-
O d -
0
ci
t
:
A:-.' ' 1 .- •••
. CM
ci
CD
i-O) >, °
- ^ 0 -^ .
d C = "> _
CD ^2 °
"O CD
C O ^"
" o 13 0 d
" _Q !_
^
'*3 J2 CM
_ CD Q. d
.30 ^ _
d
._. o
0 o
^
.
•
.j*''.» % ' '
CM _
d
CD >, o
U, 0 -^ CM _
- T- C := d
o CD ^2
"° ro in
=3 0 d ~
- 8 < ^ o
00 0 ^~ -
~ ^ °
^ "^ in
o
o
- o o —
(— i
t
•
.
•
,
•••
1 . . •
o
d
CM CD
- o O
d C
CD
^ T3
_ o ^
<
-50
d .>
J5
- o Q)
o:
~ O
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Page 21 of 33
-------
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
Coelosphaerium
& 00
:= ci
CD
.a
o
CL
0
-i—*
a.
CD
O
q
ci
2 4 6 8 10 12
Secchi Depth (m)
CO
in
CD
- c\| 0
'•i^
JO
^ 0
in
d
CD
O CO
O d
0
Q.
CD T-
O o
q
d
Cosmarium
oo 0
co O
d C
CD
oo ^
d 5
o, <
*- 0
o >
"CD
U 8 0
2 4 6 8 10 12
Secchi Depth (m)
Crucigenia
in
c\i
O d
Q.
0 2
o
q
ci
2 4 6 8 10 12
Secchi Depth (m)
r- 0
o o
d C
CD
•« c
q §
0 E
CO <
O 0
d >
si
o a:
Cryptomonas
Cyanocatena
Cyanogranis
p _
^~
£> oo
:= °
o
CD
o to
0 d
CL
0 •* _
-1—*
Q_
CO CM
O o ~
0
d
B ^
H
'
• „ q
0 *
" • t
H •
.'••• »:'* •! '•
• '" ••*•..*'. i '
•: '•':/' ./:" .•• . '
t^^ttKraJni^fl*1""1 J*/ * '•
."
«-i
- 00
d
r- 0
° C
CD
CO ^
- IO ^-
0 <
- CO 0
'•^
CD
.20
O' rv
LL.
~ o
^r
^, °
:=
-Q co
CD d
0
s-_
0 o
-i—*
Q.
«d-
0
d
a
.
,
.
•
B " ..
2
° 0
c
^ CD
- d?
^
.0.3
d 0
co CD
- §|
~ O
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
in _
-i—* . _
CD
O CO
0 d
0 o| _
3 °
Q.
CD T-
O o ~
0
0
•
• •
,
L " "
2 4 6 8 10 12
Secchi Depth (m)
^ 0
o O
d C
CD
T3
9 §
0 .Q
o 0
d >
'•4—'
JD
o 0
Page 22 of 33
-------
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
Cyclostephanos
Cyclotella
^
-1—*
.a
CD
o
Q_
0
—r
.|_l
U.
CD
CD
°
in
o
^
o
CO
o
CN _
O
d
0
0
. .
.
•
,
•
• ^ •
. f f
• f "
, ^ «• "*
0
. - C
o CO
T3
C
13
. 3 <
o 0
>
_ -i— •
CD
d
°
^
±±
o
CD
.a
o
Q.
0
-i—*
Q_
CD
p _
d
CD
o
^ _
o
CN _
0
0
.
^
„ -
- " ~
, -
- * _—
7TrrrrrrrrrmT777
'.:'.*•
n ii r iT r J««— i.. k . .
- »
°
d
CN
- in
d
- CO
o
. ^
°
Cl)
o
CD
"g
3
<
Cl)
>
•i=
CD
0
01
2 4 6 8 10 ^2
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Cylindrospermopsis
q
d
2 4 6 8 10 ^2
Secchi Depth (m)
o
0
O
co CD
d T3
C
OM
0
0
*- J5
d 0
Deasonia
Dictyosphaerium
Dinobryon
CN —
0
:= d
.a
5».
0 d
CL
3 d
Q.
ro 8 _
O d
o
.
•
'
.
.•• V '' " :
^- P _
0 "~
d C := °
CD .a
T3 CD
•^ C jS CD
- o _g 00-
^ °
_ CD Q.
.80 ^ g -
0 o
° 0
.
.
•"•:•. . '
S^v-J ••..." ^ „ t
;- P _
0 "~
" ° c 1= ° ~
CD !Q
. T3 CD
o c .a CD
d ^ O o
^ °
CD Q.
.80 CD CN _
do: O °
„ o
0 o
J
.
• •
'••
.'/,'-,
\ . • . ' t
0
1 1
0.34 0.45
bundance
<-|-'
- CM 0
0 >
CD
. ^ Q)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Page 23 of 33
-------
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
Erkenia
Euglena
Fragilaria
o
_^> 00
:= d
!a
CD
.Q CO _
O o
CL
0 •* _
3 °
-1—*
Q.
CD CN
O o -
0
d
""
d ,
___--
'
-in P
d "~
. 5? 8 £3-
d C ^= o
CO _Q
" 2 i "o d ~
3 °
'-I—' -1—'
CD Q.
- ^= 0 tO
•+—I
CD
0
o:
^ _
0
IT co
o o
CD
O
Q- d ~
0
3
^— '
Q. T-
CD ci
O
0
d
.
.
' i :
\m J
d
^
d
^
- o
d
00
- o
d
o
0
CD
"^
f~
£
<
0
>
•+—I
CO
CD
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Gomphosphaeria
^
—
!Q
o
i—
Q-
^
-i—*
Q.
CD
O
s
d
in
'•4— '
CD
~fl5
KL
2 4 6 8 10 12
Secchi Depth (m)
Page 24 of 33
-------
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
00
o
CD
.Q
O
£ 2
0
-I—*
Q_ CN
CD o
o
q
o
Gymnodinium
.• > „....
2 4 6 8 10 12
Secchi Depth (m)
Kephyrion
2 4 6 8 10 12
Secchi Depth (m)
Lobomonas
o
CO
>. CD
o o
22
Q.
0
o
o
q
CD
2 4 6 8 10 12
Secchi Depth (m)
OM
o
0
O
CD
T3
0
0
o:
Lyngbya
Mallomonas
Merismopedia
CD
0
^^
"l—< LO
'-^ • —
o
CO
o ^~ _
0 °
Q. co
0 °
^
^— « . _
Q. o
CD
0 - -
o
0
CD
•
,
•
"
m
-
•
*. " "
*.'
CD
CD
0 >, CD
s ° ~
- <*> C = in
0 CD ^2 CD ~
T3 CD
C ^2 •«-
=3 0 CD ~
O !— '-'
- CN < °- CO
CD 0 0 CD
> ^
'*3 J3 C\l
CD Q. CD
"7u CD
- o (£ 0 - _
CD °
._. o
° 0
•
"
•
*
^
,
* .
* . "
/ ^° . .- -
° 00 _
o
CD "^k
0 -^
05 C :=
~ • CD o ^P _
0 T3 CD °
(~ Q
13 0
.Q >-
CD < Q- •* _
-10 2 °
^-^ ^-^
CD Q. CM
"fT5 CD i ~
o Q^ O
CD
._. o
° 0
•
•
•
1
1 V. .
O)
- OM
CD
0
O
OM C
- OM CD
CD T3
C
^
" ^ >
i^ "CD
- CD — r-
° tt!!
~ O
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 ^2
Secchi Depth (m)
Page 25 of 33
2 4 6 8 10 12
Secchi Depth (m)
-------
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
Microcystis Monomastix Monoraphidium
& 00
:= ci
CD
.a
o
CL
0
-I—*
a.
CD
O
q
ci
2 4 6 8 10 12
Secchi Depth (m)
0> 0
r*~ o
d C
CD
™ "O
co 0
CD
O CO _
O d
Q.
0 •*
CD Q.
CN "m CD CN
d fv- O °
q
d
2 4 6 8 10 12
Secchi Depth (m)
2 8 ^00
H e: =o
CD
o, <
O 0
d >
si
CD
O CD
O d
CL
0 •*
Q.
CD OM
q
d
2 4 6 8 10 12
Secchi Depth (m)
CD
O
CD
q
0
-00
•• - o
Navicula
Nitzschia
Ochromonas
CD _
d
.!_!
:^ ir>
o o
CD
0 d ~
0 d ~
3
"o. d
CD
o - _
0
0
d
.
•
.
.".•.*- • • .. ^.! %
q _
^~
~
0 ^s n-i
C7) CJ -1— ' • —
- If) C. -^ O
d CD !Q
T3 CD
C .Q CD
" E 2 ° "
. 8 < °-
00 0 •* _
> 3 °
_ ^-^ ^—*
CD Q.
^0 CO CN _
d
o
d
.
'
•
•
o
*"• ' • "
Og^j'V' • * » •. .
i
. 0 00 -
d d
i — 0 ^^
- o O •f-'
d C =
CD ^2 o
T3 CD ^ -
. § C ^20
d _g 0
-90 0
0 > ^ °.
'"CD Q.°
- S 0 w
o a: o
o
- o o —
(— i
.
"
•
....••'.-.
_
CD „,
-00
d O
C
CD
T3
C
o -^
d <
0
•+—I
CD
- 5 £
O
~ O
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Page 26 of 33
2 4 6 8 10 12
Secchi Depth (m)
-------
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
Oocystis
Oscillatoria
Pediastrum
•<-
,>> 00
™ d
!a
CD
o to
O d
CL
0 •* _
3 °
-1—*
Q.
CD CN
O o -
0
d
p _
& °P
:= d
o
CD
o to
0 d
CL
, d
- OM O -f^
d C 1=
CD ^2
CM "° ro _
. CN C ^2 °
° E 2 o -
< '-'-
- - 0 2
° > 3 u>
-*-1 -I-1 O
^ CD Q. § ~
r^l fll CD
" d & 0
o
r-i O
— ^_} > — 1 —
o
(O
.00 P _
d ^~
°> f >^ oo
-(DO *J °° _
d C r^ °
CD .Q
"O CD
[n ^ o CD
d ^ O o
< °-
- 00 0 0 ^ _
° ~ 3 °
"CD "o.
- ^ 0 to g _
._. o
0 o
'."
*. i' ' '.
1 1 1 1 1 1
2 4 6 8 10 12
Secchi Depth (m)
Phacotus
•
* o
£: . ' . •
CM °
o 0) >, o
O <-•
C 1= _
CD ^2
"O CD ,-.
oo ^ o y
° E ° °
< CL _
0 2
T- -^ -1-' T- _
~?u CO
D^ O
o
r-i O
— ^_} » — 1 —
o
d ^
d
[0 ?) >,
- o O -t-t
d C = ^r
CD ^2 d
^. T3 CD
' § § 0 <^> _
"CD
10 "0
" d ct
~ O
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Page 27 of 33
-------
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
& oo
:= ci
CD
o
O
CL
0
Q.
CD c\i
O o
q
ci
Pseudanabaena
2 4 6 8 10 12
Secchi Depth (m)
J 8
d C
CD
T3
•<- C
ci Z!
.Q
^ <
o
si
o a:
Pyramichlamys
S
o
o
CD CM
.a d
Q_
0
3 2
0.5
CD
O
o
q
ci
2 4 6 8 10 12
Secchi Depth (m)
o ®
• O
° c
CD
T3
CM .Q
P <
o ^
0
J5
- 0
2 a:
Quadrigula
_g> 00
:= d
CD
O (D _
O o
CL
0 •*
3 °
-I—*
Q.
CD CM
O o
q
ci
2 4 6 8 10 12
Secchi Depth (m)
in
CN
ci
8
CD
s
6 a:
Rhabdoderma
Rhodomonas
Scenedesmus
CO —
d
>,
-1—*
•
O o
CO CM -
.£2 o
0
CL
0
3 ° -
0.°
CD
O
o
r~ !
.
.
<
• 1
. 0 P _
d "~
0
O -*— ' • —
C -^ d
T- CD !Q
- P T3 CD
o C JS oo
13 0 ci
< '-'-
0 0) r- _
. 5 > ^ °
° "CD Q.
0
_ in
— o • ~
.
•
•
a „
* •' '
•.••••' •. . *
%J|[kL.'£jJ_':£:^ '•. . « . . "
*" *
- O) *P _
d °
CO 0 >, ^ -
. r^ O fj o
d C =
CD .Q _,.
^ T3 CD d -
LO ^— t)
" d E 2
^ < °- d ~
- CO 0 0
d > ^ CM
oo ^ °-d "
5 [^ o d -
_ 0
— o • ~~
j
^
•
• •
1
_
. a 0
d <->
0 C
CD
T3
C
CO ^
" 0 <
0
"-^
CD
- s ^
o
- o
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Page 28 of 33
-------
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
Schizochlamys
Schroederia
Selenastrum
in
o
~ d
CD
O d
CL
0 ™ _
13 °
-i—*
Q.
CD T-
O d -
0
0
•
• x
'--— ~~~\
-In P
d "~
. * 8 ^» _
o c: :=o
T3 CD
o D O o
< £
- 2 ^ § S -
"-i—' -i—'
CD Q.
"- "0 CD CN
- d £ 0 d -
0 o
0 d
__--•""
,-^Z-—"""
•
.uciL 'kk-.-A:.n ' •; v
- C\l
d
in
-aS t5"
in ^ ro
T_ C f) o
d ^ O T -
< ol °
- g 0 0
'*3 5 *"
CD Q- S ~
o "0 ro
d Q: o
o
•
.
V .
o
CD
T3
C
-03
.Q
<
0
- ° "CD
a:
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Sphaerocystis
Staurastrum
Stephanodiscus
p _
T~
^> 00 _
:= d
o
CD
O (D
0 d
CL
0 •* _
3 °
"o.
CD OM
O d -
0
d
.
,
•
.
• •' . . •
i&aujb *n' in' at • '
- r- P _
d "~
>^ oo
-(DO *J °° _
d C := °
CD ^2
. 5? 1 & <°. _
r- j 3 O o
5 CL
- CO 0 0 "* _
d .> 3 °
"CD "o.
- £ 0 ro g -
._. o
° 0
.
. .
.
•
. ' :
•
• • °" •
-in P _
d "~
.38 £«? -
d C :^ °
CD ^2
.81 5«_
r- j 3 O o
5 CL
. csi 0 0 ^ _
d .> 3 °
"CD "o.
^ "0 CD CN
" d Q: O d
._. o
° 0
.
• J ' •
•... •
B * •
b ° •
* •" * "
• !• •
• •
• .* • • t
f ' c •*
* * " • " *
r • . ' r
.* . "i , "•• ' .. .. ••. .
. a
d
00 0
- r^ o
d C
CD
_sl
^
- CO 0
d .>
"CD
. | 0
~ O
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Page 29 of 33
-------
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
Stichogloea Synechococcus Synechocystis
CD
O d
tt „
0 d
^
"o. d
CD
o -
o
p
d
2 4 6 8 10 ^2
Secchi Depth (m)
5° c
d CD
T3
^ <£
:= oo
^2 d
CD
.Q
O
Q_ CD
.> 5
-I—* -I—'
JD Q.
0 oo
:= d
CD
O
CL
0
-I—*
Q.
O
p
d
<*•:.*
CN
CO
in CD
CN O
d C
CD
O) f—
_.._.._
T- 0
o >
'•4—'
CD ™
O 0
2 4 6 8 10 ^2
Secchi Depth (m)
Synedra
Synura
Tabellaria
p _
"<~
& °p
:= d
o
CD
O CD
0 d
CL
0 •* _
3 °
^— '
Q.
CD CN
O o -
0
d
t
•
.
i
-CD "^ —
d °
-So -&
d C = « _
CD ^2 °
^ T3 CD
d ^ O
3
^-^ ^-^
CD Q. T-
_ ^ (]) CD d
0 o
0 o
B
"
0
*
• . * •» • B •
. 8
d
^ 0 -^
.
.
"
^ CD
- CN O
d C
CD
T3
C
f 3
d <£
0
- p CD
0 "0
o:
~ O
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
Page 30 of 33
2 4 6 8 10 12
Secchi Depth (m)
-------
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
Tetraedron
Tetrastrum
Uroglena
0
CO _
">•* ^
-1—*
I — _
o
to o
-Q ° -
0 d
Q_
0
13 0
-1— ' T— _
CD °
O
O
O —
t— i
0
t
•
' •
•
C\l
T -
- 0 °
CD *^
o ~
- o ^ ^^
CD ^2
T3 CD §
(~ f^
^ O
- o **•
0 2
• ° "CD "o. o
0 tD
. o ££ O
o
- o o —
l—i
a
•
•. '
p _
^~
*- tD >, „,
.00 -^ °° -
d C 1= °
CD ^2
_ "O CD
C O tD
E 2 ° ; "
- o < Q-
0 0 ^ -
> ^ °
"CD "o.
"0 CD CN
- o Q: o o
- 0 °- -
0
*
a •
\Ji
. CO
d
in 0
- in o
d C
CD
T3
^ c
" ° E
CO <
- C\l 0
d >
'•4— '
- tD
. 2 0
~ O
2 4 6 8 10 ^2
Secchi Depth (m)
2 4 6 8 10 12
Secchi Depth (m)
2 4 6 8 10 ^2
Secchi Depth (m)
Woronichinia
.a
o
Ql
0
Q.
CD
U 8
d
2 4 6 8 10 12
Secchi Depth (m)
in 0
p o
o C
CD
p
d
.a
P 0
'•^
jo
o 0
o a:
O CD
O 6
0
Q.
CD
p
d
Bacillariophyta
Chlorophyta
a • • ,
• ^ *' ' •
'• ..^. ' • ^ " -
*m * ' m *
* *
• • °
•0° * * •
'.!„.'.
. t .„• *g
• f
'•'.'•*'•
'• •' *' ..' * *
•b . • • :* ' '•'•' • •
..Miiv^i .;;:•••
. a
d
CO
d
P5
d
- «
O)
d
- 0
0
O
CD
"^
f~
^2
-------
Capture Probability of Phytoplankton Taxon Along Secchi Depth Gradient
Chrysophyta
_Q
CD
.a
o
Ql
0
-—
Q_ 10
CD ci
O
4 6 8 10 12
Secchi Depth (m)
Cryptophyta
0
o
c
CD
"^
f~
.Q
<
0
>
*-|— «
JD
QX
o
•<~
,>> 0)
:__ o
o
CD
O 00
o d -
Q.
0 £ _
13 °
^—i
Q.
CD to
O d -
in
d
0
-— " -
' "* N"
'..
X
\
x
• • • \
• • • , • *
• \
* "^ •• *. • •
•;*'"•."•••
"' ::- '• '•'••.. • '
• .. . • •. .•.;•• '. r
*H>v ' '• :- • :
S'£tl-*''i •? •
.:iJi»i»rf>-'(..^t../ .: •.
O)
d
(D 0
- r^ o
d C
CD
^ T3
o 5
00 <
00
- CO 0
° .>
0) -^
_ T- 0
d Q£
- 0
2 4 6 8 10 12
Secchi Depth (m)
Cyanophyta
p _
_^> 0)
:= d
o
CD
O 00
O d
ol
0 ^ _
3 °
"o.
CD to
O o -
in
d
;
•' .""
, /
1 " * "o •
"* * •
« ft * * * •
m • " *
. •
*'" ° /• •
- •% "•*"••"
* . ^
* • *
, 0
•' "••" .•
.".".vV! .*V"".:.
^: taX. -.Cs'jx.d
.
•
, •
•
•
"
•
">:•:: c '
- -
- s §
CD
T3
(D C
d 3
<^
. •* 0
° -^
"CD
- d ^
- 0
2 4 6 8 10 12
Secchi Depth (m)
Euglenophyta
p _
^~
^ °°. _
:= °
o
CD
O (D
0 d
CL
0 •* _
3 °
"o.
CD OM
O d -
0
d
B
• ;
•»:.'••.'••
VJ^
d
00 0
- c\i O
d C
CD
^ T3
- CN ^
^
-50
0 >
"CD
. fe 0
o a:
- 0
2 4 6 8 10 12
Secchi Depth (m)
.a
CD
0
p
d
Pyrrhophyta
2 4 6 8 10 12
Secchi Depth (m)
O) 0 >,
to O .-^
d C :=
CD ^2
d 5 P
O 1—
co 0 0
d > =;
o
Rhodophyta
CD Q.
^ "0 tD
'•4—'
JD
o 0
o a:
2 4 6 8 10 12
Secchi Depth (m)
Page 32 of 33
-------
Appendix 4 - Wl TP GAM Models Zooplankton
-------
Capture Probability of Zooplankton Taxon Along TP Gradient
Asplanchnidae
^> 00
:= d
CD
^2
O
Ql
0
-I—*
Q.
CD
O
0
0
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
CD
d
CD
d
0
O
CD
C
.Q
•*
- CN 0
CN —
5 o:
Bosminidae
O o
0 r-
-I—*
Q.
CD co
s
Brachionidae
.
~ "~ ~ """* — — -.- — — "'"
f — " ™" s.
* ^
X \
X \
' \
' \
\
\
•
. • •
. .
• *•» " X • •• • •
V'&*fc«kfciia' . .
IAJ
d
CN
d
^
d
J
0
r^
d
- 0
0
O
c
CD
T3
&
0
ID
a:
P _
"<~
>? 00
™ d
!a
CD
O CD
O d
0) •* _
"Q.
O o
0
d
„
o
• .%• ""• •"
•* ° • •
"•' ' •', •' . .
•••""*•* *
' .' ' .' ''•"'.• "
••*•""*.
!'•'"*•
L :*•'-.' '.'
*• * .•" a
* •'.
. ^
d
CO
- CD
d
- £
d
CO
- CO
d
2
o
- P
i— i
0
O
c
CD
"g
3
0
>
'•4— '
CD
"0
a:
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
0.001
0.003
0.01
0.03
Total Phosphorus (mg/L)
Calanoida
Chydoridae
Collothecidae
P _
"<~
& °p
:= d
o
CD
O CO
0 d
CL
0) •* _
"o.
CD CN
O o i -
0
d
t
. .
•
• " .
. •
•
. . • •• •
. > .'••
• • * * * •
'.'.V"-'"' .'
...*.-__ii.: — LL.
- 0 P _
d ^~
.8 8 ^oo _
d C = °
CD .Q
_,. T3 CD
. S C ^2 3 °
"CD "o.
. 8 0 co CN _
da: o o
p_
1 B
'
*
1
4
•
• •
•' . ".' .'* ' ' .
•-..••tfCifiii-!-^. j. _. .
•»-
d
oo 0)
- o O
d C
CD
. 8?
^
"CD
. 8 0
o a:
- 0
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
Page 1 of 17
0.001
0.003
0.01
0.03
Total Phosphorus (mg/L)
-------
Capture Probability of Zooplankton Taxon Along TP Gradient
Conochilidae
,>> 00 _
:= d
!Q
CD
.Q <°
O o
CL
0 •*
is °
-i—*
Q.
CD CN
O o
q
d
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
oq
d
in CD
to O
d C
CD
O) ^
5 E
CO 0
o >
'•4—'
_CD
^ 0
Copepod.nauplii
:= o
!a
.Q CO _
O o
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
O) 0)
OM O
d C
CD
OM
CN
O
.Q
CD
oo)
da:
Copepodites
:= o
!Q
CD
O d
0
^
"o.
q
d
0.001
0.003
0.01
0.03
Total Phosphorus (mg/L)
ro 0)
- O
d C
CD
T3
•<- C
d 3
o 0
d >
si
o a:
Cyclopidae
Cyclopoida
Daphniidae
q _
^
^* 00
:= °
o
CD
O (D
0 d
CL
0 •* _
3 °
^— '
Q.
CD OM
O o -
0
d
.
, •
• * 0
.
•f
^ *
*
• ••;.'.; '•
• * ''"
• ••"*»''. • . •'
.''''^V:-: '•
" ° * "* »*•* • ••
"" • " ** • "
;- P _
d ^~
® >, m
- T O •<-? m- -
0 C := °
CD !Q
. T3 CD
o C ^2 ^ °
^-^ ^-^
_ CD Q.
- 8 0 tO OS, _
OQ: o o
o-
.
'
• " .
i
. »« •
"" "• "•*""• "*" •*•*
_i.iliJ.-:V!iiiJ.
. 0 P _
d ^~
0
- o O -^
d C = °> _
CD .Q °
"O CD
_ o ^ -Q
d 5 o
< Ql o -
-o0 0)
° .> 3
^-^ ^-^
^ JD Q- i> _
o-
^
a
^
t
. *a
/;.v ;./.:''.
• • ' *nf ,'\f' i^"* .
" «'~:.'$r':'''-'&': .
u^
d
0
„_ O
d CD
T3
C
- § 1
d <
0
. s 1
d 0
o:
- 0
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
0.001
0.003
0.01
0.03
Total Phosphorus (mg/L)
Page 2 of 17
-------
& 00
:= ci
CD
o
O
CL
0
Q.
CD CN
O o
q
ci
Capture Probability of Zooplankton Taxon Along TP Gradient
Dlaptomldae Gastropodldae Holopedlldae
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
2 P
d "~
8 8 ^» _|
d C = °
CD .Q
, T3 CD
g C ^2 CD _
° E 2 °
< °-
k P 9) g> S -I
d >
.^ 3 ~
-I—* -I—'
JD Q.
0 ,tO CN _
q
d
u s
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
^ 0
in o
d C
CD
oo "a
"• i
o 3
CN 0
d >
Si
o o:
i=
CD
.Q
O
CL
0
"o.
CD
O
q _
00
d
CD
d
2 -
CN _
0
d
•
. •
' • •
• a
*
.. . *"' •*-•• * .*
U °
CN 0
O O
d C
CD
T3
P §
.Q
-00
'•^
JD
o 0
o:
0.001
0.003
0.01
0.03
Total Phosphorus (mg/L)
£• 00
:= °
CD
O CD
O d
CL
0 •*
3 °
-I—*
Q.
CD CN
O o
q
d
Lecanidae
- §
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
CD
C
.a
O 0
d >
si
o a:
2
0
a.
CD
q
d
Lepadellldae
Nauplll
.
0
'.
•
•
.
.
. 0
d
^ 0
- o O
o C
CD
T3
3
<
- o 0
>
'"CD
-00
a:
p _
T~
-i— • . _
r^ ^
o
CD
0 d ~
0-
0 ^ _
3 °
"o.
O <=>
0
d
.
,
•'
.
.
•
• .
«-i
d
^r 0
- - O
d C
CD
T3
^ C
d ^
<
- o 0
d >
"Jg
. S 0
o o:
~ O
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
0.001
0.003
0.01
0.03
Total Phosphorus (mg/L)
Page 3 of 1 7
-------
Capture Probability of Zooplankton Taxon Along TP Gradient
Polyphemldae
& 00
:= d
CD
.a
o
CL
0
-I—*
a.
CD
O
q
d
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
Rotifera
8 !>3
CD .Q
T3 CD
d O (D
3 O d
> 5
-i—' -i—'
JO Q.
o tD /O
o: o
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
Sididae
0.001
0.003
0.01
0.03
Total Phosphorus (mg/L)
0
o
CD
to
o
-00
OM
O
0
or
Synchaetidae
Temoridae
Testudinellidae
p _
^~
^ 00
:= °
o
CD
o to
0 d
CL
^ °
"CD "o.
-52 o2-
O Lrl ***'
0 °
H 0
.
•
•
.
" . •* "• •
"* • . • '
.
^J
1
- o
o x -
o
T- 0 >,
d C = „
CD .Q Q -
T3 CD
" °l s
ff CL ™
"^ d
-00 £
"CD "o. ^
- o
-------
Capture Probability of Zooplankton Taxon Along TP Gradient
Trichocercidae
& 00
:= ci
CD
O CD
O d
CL
0 •*
13 °
-i—*
Q.
CD CN
O o
q
o
Trochosphaerldae
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
Unknown
& 00
:= o
CD
O CD
O d
CL
0 •*
Q.
CD CN
q
ci
0.001
0.003
0.01
0.03
Total Phosphorus (mg/L)
CO
CN
8
CD
o, <
O 0
d >
si
o a:
Acanthocyclops. vernalis
Alona
t
o
CD
.a
0
CL
0
3
"o.
CD
O
O _
T~
00
d
CD
d
i; _
0
CN _
0
d
.
•
• . . .
* \ *
% •%
- 0
d
- O
d
- P
- O
o
- o
- 0
0
O
CD
-o
§
^
0
"CD
0
a:
q _
T
i= d
o
CD
O CO _
O o
t
0 ^ _
13 °
"o.
CD CN _
0
d
.
.
% , ' •
. a
d
- O
d
- P
- O
d
- o
- 0
0
O
CD
-o
§
<
0
>
'"CD
0
a:
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
Ascomorpha. ecaudls
^2
CD
q
d
.
• •
• + .r» "i
• d
• % •
•
d
^r 0
. o o
d C
CD
- °- i
c, <
- O 0
d >
'•4— '
CD
.50
o Q:
0.001
0.003
0.01
0.03
Total Phosphorus (mg/L)
Page 5 of 17
-------
Capture Probability of Zooplankton Taxon Along TP Gradient
Ascomorpha. oval is
& °°. _
:= d
CD
O (D _
O d
CL
0 •*
-—
Q.
CD CN
O o
q
ci
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
Asplanchna
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
Bosmina.longirostris
0.001
0.003
0.01
0.03
Total Phosphorus (mg/L)
oo CD
OM O
d C
CD
T3
S §
0 E
. CD
o 0
Calanoid
Ceriodaphnia
Chydorus
p _
T~
_^> 00 _
:= d
o
CD
o to
2 °
Q-
0 •*
3 °
"o.
CD CN
O d -
0
d
.
. •
•
• " .
. .
• < •
• " •
. .'••"
. > .'••
• • * * * •
* " *^j^ **• * •
• o" * * •
. • •" • •
. 0 P _
d "~
. § 8 ^oo _
d C := °
CD .Q
_,. T3 CD
. S C ^2 3 °'
"CD "o.
. 5 0 ro ,
.So -2? °° _
d C = °
CD ^2
T3 CD
. 0 C ^2 3 °
"CD "o.
- o 0) /O ^ -
o: o °
o
d
.
.
.
•
••»'•*•»
• • 0iA njjL^JftSSi^Vi *fc**i ' • «1 . •
. s
d
(D 0)
- o O
d C
CD
- § C
D
0 ^2
- o 0
d .>
"CD
. g 0
- O
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
0.001
0.003
0.01
0.03
Total Phosphorus (mg/L)
Page 6 of 17
-------
Capture Probability of Zooplankton Taxon Along TP Gradient
Collotheca
Collotheca. mutabilis
Colonial, conochilus
_g> 00
:= d
CD
.a
o
CL
0
Q.
CD CN
O o
q
d
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
88 3?»
o C 1= °
CD .a
"
co
°.
o 0
d >
'•4—'
CN ^
O 0
d (V
O CD
O d
0)
a.
CD
q
d
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
:= o
!a
O CD _
O d
a.
CD CN
O o
q
d
0.001
0.003
0.01
0.03
Total Phosphorus (mg/L)
q
d
^ 0
o O
d C
CD
T3
° i
.a
o 0
J5
o CD
a:
Conochiloides
Conochiloides. dossuarius
Conochilus
q _
•<~
& °P
:= d
o
CD
O CD
0 d
CL
0 •* _
3 °
"o.
CD CN
O o ~
0
d
I
tf O _
d T^ ~
. s 8 £«?-
d C :^ °
CD ^2
_,. T3 CD
CN C ^2 cp
o ^ O o
f^ ^
^ °
'-I—' -1—'
CD a.
. 8 0 ^ g _
._. o
° 0
B
.
B
^- p _
d "~
.28 £"P _
d C •= °
CD .a
T3 CD
•^ C .Q CD
' d 13 0 d
^2 <—
CD < CL
-80 CD ^ _
d > ^ °
'•4— ' -I— '
CD a.
. 8 "0 ro CN _
._. o
° 0
^
^
m
*
.
•
k
• • * m
"•
a • * °. ^ * * • V
• *o • *i ••VV * , * | *
' '.''."I*". %' *"lJl.. . _. .
- 00
d
in CD
- CD O
d C
CD
"^
§• C
~ • ^
f}
<;
- CO 0
d >
'•4— '
CD
.20
o a:
~ O
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
0.001
0.003
0.01
0.03
Total Phosphorus (mg/L)
Page 7 of 17
-------
Capture Probability of Zooplankton Taxon Along TP Gradient
Cyclops, varicans. rubellus
Daphnia
Daphnia. dubla
o
•<-
,>> 00
™ d
o
CD
O CO
O d
CL
0 •* _
13 °
-i—*
Q.
CD CN
O d -
0
d
•
• •" .
•
- 0 P _
d "~
.88 ^» _
d C 1= °
CD .Q
_ 8 C % »_
° 2 e °
< Q-
-80 0 * _
° > 3 °
-1—' -1—'
CD Q.
- 8 0 .CO 3 °
-1—' -1—'
CD Q.
- 5 0 ro ^ _
ci Ct O °
._. o
o d -
..
*
•
•
^ •
• • %
0 • ••
•
. 0
d
CN 0
- o o
d C
CD
^ T3
" § E
-50
d >
'•4— '
CD
-00
o:
~ o
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
0.001
0.003
0.01
0.03
Total Phosphorus (mg/L)
Daphnia. longiremis
Daphnia. mendotae
Daphnia. parvula
q _
•<~
^> oq _
:= d
o
CD
O CO
0 d
CL
0 •* _
3 °
Q.
CD CN
O d -
0
d
t
•
• • • *
" •" 0 0 .
'••"*• I
' '^'•'•"•julS: •
. 0 P _
d "~
.88 ^oo _
d C = °
CD .Q
_,. T3 CD
. S C ^2 3
'-I—' -1—'
CD Q.
. 5 0 ro CN _
do: o o
._. o
° 0
t
.
.
'
•'
• b . ' •*
. •
" • • •
• •
•• •
" 1 •• •
.!•••• .
o *"
d °
co 0 >, ^.
- o O -*-• ^. _
d C = °
CD .Q
-. T3 CD
. 8 C ^2 " _
d| !°
-00 0 ^ _
d > ^ °
'•4-* 4-*
CD Q.
- 5 ® 55-
._. o
0 o
^
_.
.
•
'
. •
- 0
d
^ 0
.00
d C
CD
T3
c
.a
-00
"-^
CD
-00
a:
~ O
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
0.001
0.003
0.01
0.03
Total Phosphorus (mg/L)
Page 8 of 17
-------
Capture Probability of Zooplankton Taxon Along TP Gradient
Daphnla.pulex
,>> 00
:= d
CD
.Q
O
CL
0
Q.
CD CM
O o
P
d
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
£
in CD
o O
d C
CD
o?
o 0
d >
'•4—'
JD
o 0
Daphnla.pullcarla
:= o
!a
CD
o to
O d
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
u s
oo CD
o O
d C
CD
« C
o £
CM
O
0
JD
o 0
o o:
Daphnia. retrocurva
0
"o.
0.001
0.003
0.01
0.03
U 8
Total Phosphorus (mg/L)
in CD
o O
d C
CD
* c
o t
O 0
d >
'•4—'
JD
o 0
o o:
Daphnia. rosea
Diacyclops. thomasi
Diaphanosoma
P _
"<~
& °p
:= d
o
CD
o to
0 d
Ql
0 •* _
^ °
"o.
CD CM
O o ~
0
d
t
•
•
. • •
;- P _
d ^~
. - 8 ^» _
o ^ — o
CD !Q
. T3 CD
. fe C ^2 D
"CD "o.
- 8 0 CO cs, _
do: O o
._. o
° 0
u
•
1
"
°
• a
* 1 •
•
•°X • '•
• "^'Vv- . i
• * " * •
*" * . . "" *'?<- " • . J. . .
V N
d
0) CD >, oo
- o o •*"* d ~
d C =
CD ^2
&1 5
" d =3 0 CM
^ t°-
-00 0)
d > ^
- 8 1 & 5 "
d Q: o
„ o
° 0
B
•
. • •
> "
00 0
- P 0
CD
T3
CM ~3
- P E
0
- 5 "CD
0 0
a:
~ o
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
0.001
0.003
0.01
0.03
Total Phosphorus (mg/L)
Page 9 of 17
-------
Capture Probability of Zooplankton Taxon Along TP Gradient
Diaphanosoma. birgei
_g> 00
:= d
CD
.a
o
CL
0
^
"o.
CD
O
q
d
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
o 0
d >
si
Eplschura.lacustrls
8 8 £>»
d C 1= °
CD ^2
o £= -Q ^
° E e °
0
"o.
O
. :j.
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
q
d
^ 0
o O
d C
CD
T3
.Q
<
o 0
•^
JD
o
. "CD
o 0
o a:
Filinia. longispina
Filinia.terminalis
Gastropus
p _
^~
^ CO
:= d
!Q
CD
O CO
0 d
Q-
0 •* _
3 °
Q.
CD CN
O o ~
0
0
.
• .
• .
. 0 P _
0 "~
. & 8 ^» _
d C =°
CD ^2
. §| !«_
Q J O O
-80 ^ °
CD a.
.80 Q g _
0 o
° 0
,
•
• • ' .' '
1
;- P _
0 "~
- o § tS-
CD !Q
. T3 CD
. o c .a co
d ^ O o
-80 ^ °
CD a.
.80 ro g _
._. o
0 o
.
•
.
. s
0
1 1
0.03 0.04
)undance
- O 0
o .>
CD
-00
d Q£
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
0.001
0.003
0.01
0.03
Total Phosphorus (mg/L)
Page 10 of 17
-------
& 00
:= ci
CD
O CD
O ci
CL
0 •*
13 °
•4-»
Q.
CD CN
O o
q
ci
Capture Probability of Zooplankton Taxon Along TP Gradient
Gastropus. hyptopus Gastropus.stylifer Holopedium.gibberum
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
^ 0
co O
o C
CD
CD
*- 0
O >
L §1
o o:
^> 00 _
= 0
CD
O CO _
O ci
CL
0 •*
-—
Q.
CD CN
O o
q
ci
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
in
CN
8
CD
« -i
^> oo _
1=0
CD
O CO _
O ci
CL
0 •*
13 °
•4-'
Q.
CD CN
q
ci
0.001
0.003
0.01
0.03
8
ci
CN 0
q O
ci C
CD
T3
<
O 0
° .>
'•4—'
JD
o 0
o:
Total Phosphorus (mg/L)
Kellicottia. bostoniensis
Kellicottia. longispina
Keratella. cochlearis
q _
*-
^* 00
:= ci
o
CD
O CO
2 °
CL
0 •* _
13 °
•4-'
Q.
CD CN
O o -
0
ci
a
•
•
•
• ^» • •
'*-• "-«--•-'' ** a-M*^i i *
LO ^
0 "~
*- ® >, m
. ^r O 4? °° _
ci C r^ °
CD ^2
T3 CD
. co C ^2 cq _
o ^ O o
f} ^
, m
. ^r O 4? °° _
ci C := °
CD .Q
T3 CD
. co C ^2 CD _
o ^ O o
f^ ^
< °-
. 0! 0 0 •* _
0 > ^0
'•4-* -4-1
CD Q.
-52 o2-
P_
^
.
•
. •
• •
• • •
• **•
* . - . **.• *•
*•* . . .
•»•'••!•
>" •' I ' * •
•t \i-. * •;
o ° '• • f "•)•% * * *Vto "
-• .'-vit-' : « -'
r^
ci
CD 0
- in o
d C
CD
"^
^ ^~
~ • ^
° ^
00 <
00
- CN 0
° .>
"CD
. J 0
o a:
- 0
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
0.001 0.003 0.01 0.03 0.001 0.003 0.01 0.03 0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
Total Phosphorus (mg/L)
Page 11 of 17
Total Phosphorus (mg/L)
-------
& 00
:= ci
CD
o to
O d
CL
0 •*
13 °
-i—»
Q.
CD CN
O o
q
o
Capture Probability of Zooplankton Taxon Along TP Gradient
Keratella. crassa Keratella.earlinae Keratella.hiemalis
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
2 8 ^00
d C 1= °
CD
5 .1
s!
CD
o to
O d
CL
0 •*
3 °
-I—*
Q.
q
ci
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
CM
si
o a:
^> oo _
:= d
CD
o to
O d
CL
0 •*
13 °
-i—*
Q.
CD CN
O o
q
ci
0.001
0.003
0.01
0.03
•* 0
q o
d C
CD
T3
CO ^
q ^
o 3
o 0
d >
'•4—'
JD
o 0
Total Phosphorus (mg/L)
Keratella. quadrata
Keratella. taurocephala
Keratella. testudo
q _
T
^* 00
:= d
o
CD
o to
0 d
CL
0 •* _
3 °
-1—*
Q.
CD CN
O o
0
d
.
•
." '
• .
• < . •
• ''. »*
• • •
. ." •'*" •« '.
• • "•••
.. ." ..'•.*.::; » • .
. 0 P _
d "~
•* 0) >, m
- o O 4? °° _
d C := °
CD ^2
"O CO
. 8 £= ^2 5 °
-1—' -1—'
CD Q.
.50 CO , m
- ^ °
-1—' -1—'
CD Q.
.80 CD
'•4— '
CD
. 8 0
o o:
~ O
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
Page 12 of 17
0.001
0.003
0.01
0.03
Total Phosphorus (mg/L)
-------
& 00
:= ci
CD
O CD
O d
CL
0 •*
13 °
-i—*
Q.
CD CN
O d
q
o
Capture Probability of Zooplankton Taxon Along TP Gradient
Lecane
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
8
8 8 ^»
d C 1= °
CD .Q
CD
O CD
O d
CN
q i
0 E
50 0 ^
"-^ -I—'
CO Q-
5 "0 ,
J5
O 0
o Q:
Leptodlaptomus. ml nut us
q _
_^> 00
:= d
CD
O CD
O d
CL
0 •* _
13 °
"o.
CD CN
O d -
0
d
•
.
• •
••
•
• • •
•"• '\
.i&
jL_. . _. .
0.001
0.003
0.01
0.03
U §
r- 0
o O
d C
CD
•« c
q §
0
-00
Total Phosphorus (mg/L)
Leptodlaptomus. slcllls
Mesocyclops. edax
Monostyla
in _
d ~
-i— • . _
:= o
o
CD
O CO
0 d
CL
0 CM
3 °
"o.
CD T-
O d -
0
d
t
•
•
0
.
'
* •
0 * •
. o q _
d "~
^0 >, _
- o O *? °° _
d C := °
CD ^2
T3 CD
_ 0 C ^2 >. ^t
- o o -f^ • -
d C := °
CD .Q
^ T3 CD
o ^ O o
o ^
< °-
-00 0 ^ _
d .> 3 °
"CD "o.
- 0 0 /? ^ -
o: o °
o
d
B
.
•"•'%•* .•* .
. 6
d
00 0
- o O
d C
CD
"^
- q c
o
CN <
CM
- O 0
d .>
"CD
-00
o a:
~ O
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
i i I i rnrrrrr
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
0.001
0.003
0.01
0.03
Total Phosphorus (mg/L)
Page 13 of 17
-------
CD
O
CL
CD
o
Capture Probability of Zooplankton Taxon Along TP Gradient
Monostyla.lunaris
0
in
o
cq
d
CN
o
0
0
CD ;;
p CD
o -O
CD
o
o
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
Notholca. foliacea
in
d
:= o
!Q
CD
o cq
O d
0 ™
<= d
-i—*
Q.
CD T-
O o
o
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
8
CO 0
o O
d C
CD
CN "2
o b:
_ <
q 0
•i=
CD
Orthocyclops. modestus
:= o
!Q
CD
0
-I—*
Q.
o
0.001
0.003
0.01
0.03
Total Phosphorus (mg/L)
0
O
CD
T3
o 0
o >
'•4—'
JD
o 0
o:
Ploesoma
Ploesoma.lenticulare
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
Polyarthra. dolichoptera
j^*
:_—
o
CD
.Q
0
CL
0
^
"o.
CD
O
p _
T
CO
d
CD
d
•* _
o
CN _
0
d
.
0
• .
'.
"" ' "•
• : . r •.'
•~ '?t5.J'i''i ' «rj-
1 —
d
_ ,_
d
T—
~ o
- O
d
- P
~ O
0
^j
f~
CD
T3
C
^
^2
0
>
"CD
0
p _
^
^* op _
'— O
!Q
CD
O CD
2 °
Q_
0 •* _
3 °
"o.
CD CN
0
d
.
B •
.
• ' '
*
. 0
d
- O
d
- O
—
- 0
- o
0
O
CD
T3
^
f^
<^
0
>
"CD
0
a:
j^*
^_
o
CD
^2
0
Q_
0
^
"o.
p
^
CO
o
CD
o
^r
0
d
0
d
•
•
' i
•' i
•
<••
" • .
' ' "".
U" • «•
I y . ,
Nl
d
^ 0
- - O
d C
CD
- °- §
- o 0
d >
-i!
~ o
0.001
0.003
0.01
0.03
Total Phosphorus (mg/L)
Page 14 of 17
-------
_g> CD
:= d
CD
O CD
O d
CL
0 •*
Q.
CD CN
O o
0
0
Capture Probability of Zooplankton Taxon Along TP Gradient
Polyarthra.major Polyarthra.remata Polyarthra.vulgaris
U 8
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
§ 8 ^oo
d C 1= °
CD .a
^ T3 CD
o C ^2 CD
d 5 P o
° a> a) d
"-^ -I—'
CD a.
o
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
0
o
CD
§1
o Q:
O CD
O d
0
"o.
o
0.001
0.003
0.01
0.03
8
d
r^ 0
CN O
d C
CD
T3
CN C
*-
'•4—'
. CD
o 0
Total Phosphorus (mg/L)
Pompholyx
Skistodiaptomus. oregonensis
Synchaeta
d
•4- »
•^— CO
CD
0
CL ^
0 ° "
3
"Q. ^
CD o
O
0
d
t
•
. .•
p _
d
8 £»» -
CD C := °
- ° CD ^2
° T3 CD
C .Q CD
13 0 d
- § £ CL
d 0 0) . _
> 3 °
" "CD "o.
•§2 52-
„ o
° 0
B
"• ••**" "o
" •" -'• *!/rf"Ytf-» 8 "^ •
T- O _
d ^ ~
.88 &*. -
d C = °
CD .Q
,0 ~& TO
o £= ^2 CD
d 3 O o
< ^
-00 0 "* _
d .> ^ °
"CD "o.
. 8 0 co pi _
OQ: o °
„ o
0 o
B
•
•
..
0"
. •'". •:•»!••
.-/,:••:. i.^
•."»«li»S^'. i'jv*.'. * I
- W
d
r^ 0
- CN O
d C
CD
T3
d 3
^
-20
0 >
. "CD
- 0. 0
~ O
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
Page 15 of 17
0.001
0.003
0.01
0.03
Total Phosphorus (mg/L)
-------
Capture Probability of Zooplankton Taxon Along TP Gradient
Trichocerca
o
CO —
d
CD R _|
o
O
CL
0
Q. o
CD
O
o
q -
d
Trichocerca. birostris
C
CD
- o <
0
O \fS
JD
0
:= o
!a
.a
o
CL
2
^
"o.
O
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
Trichocerca. cylindrica
-V-.
0.001
0.003
0.01
0.03
Total Phosphorus (mg/L)
O) 0
o O
d C
CD
u, <
O 0
d >
si
o a:
Trichocerca. multicrinis
Trichocerca.porcellus
Trichocerca. rousseleti
p _
^~
_^> CO
:= °
o
CD
o to
0 d
CL
0 •* _
3 °
"o.
CD c\i
O d -
0
d
>
a •
•
.
B
4
, •
• • •** *
•..-.' -
' i :•••••
•
•
.
. •
.
• .
•t •
1 , • a
•
- O
d
IT) _
- o o -&"
d C = ^.
CD o • —
-, T3 CD °
- o C ^2
d ^ O co
-Q *~ o ~
< CL
50 0
— • t f\l
• — 3 o" ~
"CD "o.
- 5 0 to ,_
0 o
° 0
.
•
.
.
, „•
'• •
*
p _
^~
- 5
° 0 ^ °° -
T— C "~ ^
- q CD !Q
o T3 CD
C ^2 ^ °
"CD "o.
"0 CD OM
- O QX O °
0 o
° 0
.
*
• •• *
•
"
» •
•
N "
. 0
d
^ 0
_ o O
d C
CO
~u
_ o ^
f^
<;
-50
d >
^
- o 0
a:
~ o
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
0.001
0.003
0.01
0.03
Total Phosphorus (mg/L)
Page 16 of 17
-------
-^ °9
= d
CD
O (D
O d
Ql
Q) ^
Q.
CD OM
O d
q
d
Capture Probability of Zooplankton Taxon Along TP Gradient
Trichocerca. similis Tropocyclops.prasinus. mexicanus
I- 8
OM IU
- q O
d d
CD
~, "O
- o
,>> 00
:= d
CD
O (D _
d _g P d
-00
d >
D
Q.
0
-50 ro g -
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
q
d
0.001 0.003 0.01 0.03
Total Phosphorus (mg/L)
00 0
- q O
d C
CD
(D "2
k °- i
o ^?
-00
d >
LSI
Page 17 of 17
-------
Appendix 5 - Wl TN GAM Models Zooplankton
-------
Capture Probability of Zooplankton Taxon Along TN Gradient
Asplanchnidae
Bosminidae
Brachionidae
o
•<-
,>> 00
™ d
CD
O CD
O d
CL
0) •* _
13 °
^— '
Q.
CD CN
O d -
0
d
C
p _
& °P
:= d
o
CD
O CD
0 d
CL
0) •* _
3 °
"o.
CD CN
O d -
o
d
o
^ «. - - "
'-''^^^
* * * ,/x *•*
--'' /''
S / 0
^^ /
// '/'
• / /' '
/ ' ' . . "
/ *S s •
' .'••".
" " V. * "••*•";""••
.' ii.™'!"'"' *°" "2 * L'. .'" : '
i i i i
) 0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
Calanoida
.
• •
•
• *
•
" ° Q
• • •
• 0»* • » •
;•-"•• • .' %•
• . *;•.,, /•,
•• — — *.. . ——..... .
i i i i
;- P _
d "~
® >> 0)
d c: := d -
CD .Q
, T3 CD
. S C ^2 oo _
d ^ O o
< £
-00 0) ^ _
d > ^ °
^-^ ^-^
CD Q.
-p0 ro d _
— o • ~~
0
C
_ o P _
d ^~
.§8 ^oo _
d C = °
CD .Q
_,. T3 CD
o £= & <°
' d = 00-
< CL
.80 a) . _
0 £ 3
"CD "o.
- 0 0 TO CNj _
do: O °
o
. o d -
^-Zr "" ~ *
^^^\
s
\
\
\
* • • \
\
• • "* • \
• • • \
" * *
* -V -V-* '•.**•
• "• *J»^° •^•^•*g* *
* j^ " "• •* 2*o * • *s* •
. -8'V: "ii.'.«i
l l l l
) 0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
Chydoridae
'
o
,
»'
" « f A * •
. .. ..j&ica..... jkM^._..^ .
l l l l
. 0 P _
d "~
.08 ^» _
d C 1= °
CD .Q
§ C ^ CD
' o =3 Od
O 1—
-80 0) ^ -
d > ^ °
"CD "o.
.00 CO CN _
- o P -
0
C
P _
d ^~
.88 ^» _
d C = °
CD ^2
.- T3 CD
. § C ^2 3 °
"CD "o.
. o 0 CO cs, _
do: o o
o
. o d -
o
•
• "• *
•
" •* •
.'"".'<
.' ^ ; _ ._.
a •* »** •
":'.. t ^"
** •"
8 • • * • •
B . • • * ^«
0 • , •• °
1 1 1 1
) 0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
Collothecidae
•
•
•
*
.
"
. •
•' „ :.;' -': .
*• » %* "*.".'*
"*"""•**•"•,•*"•* *•
i i i i
. ft;
d
m 0)
- CD O
d C
CD
- 9 C
^
O
^. <
- CO 0
d .>
•+—I
CD
. | 0
- P
o
d
oo 0)
- o O
d C
CD
,0 "^
. 8 £=
D
"CD
- 0 0
o o:
- o
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0 0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
Page 1 of 14
-------
^00
:= d
^2
CD
"§ §
Ql
0
52
Q.
CD
O cs,
d
Capture Probability of Zooplankton Taxon Along TN Gradient
Conochilidae Copepod.nauplii Cyclopidae
i •• ...
/ • •••..#.*%. .-
•
•*••
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
CD
d
CD
T3
CD ^
(N >
d w
0
- o:
-1—*
!5
CD
o
£
3
-I—*
Q.
CD
O
o
•<-
CD
d
CD
d
•*
d
CN
°
0
d
•
"
•
*
* " "•
• .. V- ' • •"•"•
. . r •.;• . .
• ••*..••
• :•-• *• " . i
. * .•• ••.
1 1 1 1
M^
d
O)
- OM
d
CN
o
in
d
CD
— O
o
o
0
o
CD
T3
c
^
"0
^>
"CD
0
o
T~
~ d
!a
J2 CD
O d
0 "*
13 °
"o.
O d ~
0
Q
.
. '
• f o
' ' .
' • ." • -
' ' '!*
•* * •• * " •
• " • • v° • •. • •
* • * * • •
• • • • • • •
• 0 , « •
*• .1
*
1 1 1 1
V M
d
d
r^
- o
in
- 0
d
CO
— O
~ o
0
o
CD
T3
^
0
>
'•4— '
CD
0
o:
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
Cyclopoida
Daphniidae
Dlaptomldae
p _
T~
^ CD
:= °
o
CD
O CD
2 °
Ql
0 •* _
3 °
"o.
CD CN
O o
0
d
.
•
•
.
'
• • '.••' • '
_i!^!i^/.l^Y':
i i i i
. 0 P _
d "~
. s 8 ^» _
o C — °
CO ^
"O CO
o ^ o oo
o ^ O o
5 it
-80 0 !> _
d > ^ °
"CD "o.
. 5 0 to cp _
OQ: o °
- 0 g -
.
.
'.
•
•" • • *
•
• «,.•.•''
• '•„•'•• "
• • "• . "* " * .
*" * * • 0
: Y'^:-;V"J'"'
i i i i
. 0 P _
d "~
.88 &m
CD o o
i^ "^3 CD
. o £= ^2
o =5 0
.Q >- CD
.80 0 °
d > ^
CD Q. i> _
. 8 "0 CD o
d Q: O
. o § -
.
•
. •
'_
• • .
. * .""
. • ••'.
• ^ '. • .
, • *
• •• •• «o •.•
1 1 1 1
V M
d
0
. 8 g
d CD
T3
C
^
- p 5
d **•
0
. 8 "CD
d 0
o:
- o
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0 0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
Page 2 of 14
-------
Capture Probability of Zooplankton Taxon Along TN Gradient
Gastropodidae
>. O)
~ o
tO oq
•tr d
0 d
-i—*
Q_ CD
CO d
O
in
d
0 0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
§
o
CO
T3
&
<
0
JO
0
o:
P
d
Holopedlldae
CD
.a
o
-I—*
a.
o
0
0 0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
o
CN 0
O O
d C
CD
T3
P c
d
&
<
0
•^
CO
Lecanidae
:= o
!a
CO
O CD
O d
0
-I—*
Q.
o
U 8
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
r^ 0
o O
d C
CD
p ;=
0 E
O 0
d >
JO
0
Lepadellldae
Rotifera
Sididae
in
d
:= d
CO
J3 r>
0 d ~
CL
0 CN _
3 °
"o.
CD ^
O o -
0
d
f
•
^
•
•
•
1 1 1 1
•<- o _
- d ,-J ~
d
tl> >, m
- § ^ is-
0 CD &
T3 CO
C .Q CD
~ o ^ O o
f^ t_
, „,
.00 ^ °P -
d C = °
CO &
^ ~O CD
. o £= & ^ °
to "o.
.00 tO CN _
do: O o
0 o
0 o
t
.
• ,
*
'
•
m '
•':.ji. •'. •
'"' ''^'8J(L '.••'•> «!<.- —
i ™ ™
i i i i
^~
Q
00 0
.00
d C
CO
"^
. o £=
D
° ^2
^. <
- o 0
o .>
"CD
. 0 0
o a:
~ O
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
Page 3 of 14
-------
Capture Probability of Zooplankton Taxon Along TN Gradient
Synchaetidae Temoridae Testudinellidae
o
"-
_^> CO
:= d
.Q
CD
O CO
O o
Q_
0 •* _
Q_
CO CM
OO ~
0
0
•
•
• •
• •* •
• *
. • • • "5 •
. • '. .
•• '••' I '
'f' ..: • •
•••*•! I
••.',-.'.' ' .
1 1 1 1
-CO P
d "~
. « 8 3?» -
0 c =0
CD ^2
g C ^2 co
d ^ O o
< B-
- CN 0 CD •* _
d > ^ °
u> TO S-
_ !£ 0 CD CN _
d Ct O °
0 °
• Q
, *
X
Xx
\
X%
"^~~-^. '"^
^~~~J ~ ~ "'
""i^""---
• • % ~^-T~~~~— -— — _
~ - - -
i i i i
- O
o
- o O -^
d C 1= ^r
CD ^2 d
T3 CD
^— C^l
1-1 3 O co
< CL °
-00 2 CN
.> U d ~
-t-* -t-*
CD Q.
- o
O '4=
CD
- ° rv
d LH
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
Trichocercidae
Trochosphaerldae
Unknown
O _
^~
^ CD
:= °
CD
O CO
0 ci
CL
0 •* _
3 °
Q.
CD CN
O o ~
0
0
.
•
.
1 0 *
'•":'£•'£-' ':':' '••;.
1 1 1 1
- -T- P _
0 "~
T- 0 >. oo
d C •= °
CD .Q
- 81 1"-
d ^ O o
< ^
.80 a) •* _
_ CD Q.
- 8 0 /O CN _
OQ: o °
0 o
° 0
.
*
1 :
• jj*u— • LJ_ ' '
! !
1 1 1 1
;- P _
0 ^~
. - 8 £>» -
0 C := °
CD !Q
. T3 CD
. o C .£ cp _
d ^ O o
< ^
-80 ^ °
CD Q.
.80 tD CN _
OQ: o °
._. o
0 o
.
I I I I
- CN
0
1 1
0.14 0.19
>undance
- O 0
d >
.81
o Q:
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0 0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
Page 4 of 14
-------
Capture Probability of Zooplankton Taxon Along TN Gradient
Acanthocyclops. vernal is
Ascomorpha. ecaudis
Ascomorpha. ovalis
o
)ture Probability
0.4 0.6 0.8 1
CD CN
O o
o
d
•
« •
1 1 1 1
n P
1 1 1
0.01 0.01 0.01 0.
ative Abundance
)ture Probability
0.4 0.6 0.8 1
- o "0 ^ ^
o: o °
o
. o d -
•
. _ •>» • 1- • •
1 1 1 1
_ 0 P
0.02 0.03 0.03 0.
ative Abundance
)ture Probability
0.4 0.6 0.8 1
.50 (?. ^
OQ: O o
o
. o d -
•
•" •
" •
. V:-' »>%]• •; :/! .*>..'...' •
.... rs— .^.. ?» .
i i i i
£
i i i
0.03 0.04 0.05 0.
ative Abundance
-50
o a:
- 0
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
Asplanchna
Calanoid
>,
:=
o
CD
.a
0
CL
0
3
"o.
CD
O
O _
T~
00
d
CO
d
Q ~~
CN _
0
d
.
•
• d
*
**
"
""•*•*
" • •
* • "• °
, • • * •""•? *•** *^ *••*** • •
1 1 1 1
V N
d
,_
d
£
Q
- O
°
- P
0
O
CD
-o
13
<
0
.^
"CD
0
p _
^
>^ 00
1^ O
o
CO
_Q CD _
O o
ei
0 •* _
^
"o.
CD CN _
q
Q
.
• • *
^
.
.•
'• '
" " f
." "1* . '
" ".^''^ ." I,'"*'*
i
1 1 1 1
. s
d
in
- O
d
- P
CM
- O
°
- 5
d
o
0
Q
CD
-o
13
^
0
.^
"CD
0
a:
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
Ceriodaphnia
p _
^~
-i—' . —
r^ ^
O
CD
O CO _
O o
^
0 •* _
3 °
•4- »
Q.
CD CN _
O °
0
°'
t
.
.
•
.
•
,5 .• . ' "
• • • , •
1 1 1 1
O
d
^ 0
- o O
d C
CD
-o
- °- §
0 .Q
-50
d >
'•*-*
CD
-00
a:
- O
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
Page 5 of 14
-------
Capture Probability of Zooplankton Taxon Along TN Gradient
Chydorus
^> oo _
1=0
CD
O CD _
O d
CL
0 •*
Q.
CD CN
O o
q
d
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
CD 0)
o O
d d
CD
•« c
q iz
.Q
« <
o 0
d >
si
Collotheca
q
d
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
CD
o
^r
o
8
C
CD
"
0)
CD
S 0)
Collotheca.mutabilis
0
"o.
q
d
U °
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
CN 0)
o O
d C
CD
o 0
d >
'•4—'
JD
o 0
Conochiloides
Conochilus
Daphnia
q _
•<~
_^> 00
:= °
o
CD
O CD
0 d
CL
0) •* _
3 °
"o.
CD CN
O o i -
0
d
f
'. '
• ^A "^ ^ • ,
I I I I
"^ O _
d T^ ~
-s§ ts-
CD .Q
_,. T3 CD
CN CI ^2 CD
d ^ O o
f^ t_
^ °'
"CD "o.
- g 0 « ! CN _
OQ: o o
„ o
0 o
^
. .
B
*
.
•
»
* • "•
•
• •
'„•'. "'ft'' '.*•*•' * • .
'• '• •..•/'. ^ °
"CD "o.
2 "0 ro CN
" d Q: O o
._. o
° 0
t
f *
•
.
."
t
V
• . ' '
• I •
•'-•!'••' •'••..
., • • • •
1
1 1 1 1
- 0.
d
CN CD
.00
d C
CD
,%, "^
- 9 £=
f^t
^
.50
o .>
"CD
.50
~ O
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0 0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
Page 6 of 14
-------
Capture Probability of Zooplankton Taxon Along TN Gradient
Daphnia. dubia Daphnia.longiremis Daphnia.mendotae
*
*
\ ,
\
\
\
\ s
\
X. ^ •
^^MJN"-
^^Av^---
- v>\^r
• ~ "" ~ ^ «
• TT 71-771*111701 r^mwrnlnrl OTn«nnm««w«t<* • •
1 1 1 1
D 0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
Daphnia. parvula
-
°
• •
.
'
*
"
0 "~
.88 ^» _
d C 1= °
CD .Q
T3 CD
5 C o co
° _Q Q. °
< Q-
.50 0 ^ _
O > T
"-^ -i—'
CD Q.
~*j\ tn CN
- o 0 >» • —
o: o o
o
§ -
c
q _
- 0
0 tD >, m
O -&• °° _
(~ : — O
- o CD !Q
T3 CD
C .Q co
Q 13 0 d
° < G-
0 0 •* -
- ° .> ^ °
"CD "o.
•0 tD CN _
- O ry O °
o -
~--,^
^> ^
X
^^^- ^^^ N
^* — ^v xv
' " " \ »
' » \ '
' N \ v
' v \ »
' x \ »
' v \ N
• Av
:• A^
• •••.. \\^
* * °» *"*"• • • X \v v •*
_-..%9L._ _. . -Jrj=-—
1 1 1 1
) 0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
Daphnia. pulicaria
•
•
*
* .
,
•
•..• * '
• - — • — ' *••" ••
d ^~
. § 8 ^0° _
d C 1= °
CD .Q
_.. T3 CD
. § £= J2 <0 _
d =3 00
< ^
-80 0 * -
° > 3 °
-1—' -1—'
CD Q.
T~ ^r\ cn c\i
-o0 l" • -
do: o °
o
d -
c
p _
d "~
CO 0 >, m
- o o 4? °° _
d C := °
CD &
"O CO
. 8 £= &
"CD
-00
o Q:
- 0
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0 0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
Page 7 of 14
-------
Capture Probability of Zooplankton Taxon Along TN Gradient
Dlacyclops. thomasl
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
CN
d
O) CD
P O
d C
CD
^ C
o £
in
d >
JD
0
Diaphanosoma. birgei
_^> CD _
!5
CD
O CD
O d
Q.
Q.
o
CD CD
P O
d C
CD
CD ^
o £
- q 0)
sl
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
Eplschura.lacustrls
_^> CD _
0
"o.
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
q
d
^ 0
o O
d C
CD
o 0
"-^
JD
o 0)
o:
Filinia.terminalis
Gastropus
Gastropus. hyptopus
_ _•_•.£.
l
.
•
M
L £%.•*« *.m "
| |
|
T- P
0 "~
. - 8 ^oo _
0 C := °
CD !Q
. T3 CD
- £ C ^ cp _
Q J O O
< °-
-00 ^ °
CD Q.
.80 22-
d
.
•
.
•
I I I I
. 0 P _
0 "~
.38 £3-
d C = °
CD .Q
.8 1 |«.
Q J O O
-80 ^ °
CD Q.
.50 CO CN _
OQ: O °
o
d
V
i i
-,-,,
i i
. 6
0
1 1
0.02 0.03
xindance
- O 0
0 >
CD
-00
o o:
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
Page 8 of 14
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
-------
Capture Probability of Zooplankton Taxon Along TN Gradient
Gastropus.stylifer
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
O) 0
q O
o d
CD
°i
o 3
0
0
sl
o a:
Holopedium. gibberum
,>> 00
:= o
lO
CD
O CO
O o
0 "*
3 °'
-i—*
Q.
CD CN
O o
q
o
0 0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
CN 0
q O
o d
CD
T3
§ I
o 0
o >
•^
JD
o 0
o:
Kellicottia. bostoniensis
,>> 00 _
CD
O CO _
O o
CL
0 •*
-—
Q.
CD CN
O o
q
o
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
8
CD
T3
.Q
^ <
o 0
o >
si
o a:
Kellicottia. longispina
Keratella.cochlearis
Keratella.crassa
•
••
« s
. '•" *
0 . " " • •
• ..• '•'••f. "•
.*•"'. •*• 'i*».'./ '
•• :.**•.«'•.
i i i i
0 "~
. ^ 8 £»_
o id := °
CD ^2
T3 CD
^ C o co _
o ^ O o
. ™. y> 0 •* _
0 > ^ °
CD Q.
. T 0 tO CN _
. o § -
•
•* t
o , •
• • ' . •*" •/ v
" V • ' 'i1*'. '"
•• • ' • ••*• *•**
1 1 1 1
o ^ ~
to 0 >, m
. LO O 4? °° _
O id := °
CD ^2
o ^ O o
- c3 0 CD ^ _
o _> D °
^ CD Q.
J "0 tD CN
" d Q^ O o
. o § -
•
•
_
0 ' •
* • ,0»
*" '" .'at'.'' j "x"*"1"^"
_.\. .*.*4*.*j% •.
i i i i
ci
1 1
0.14 0.19
bundance
O >
-§1
- 0
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
Page 9 of 14
-------
Capture Probability of Zooplankton Taxon Along TN Gradient
Keratella. earlinae
Keratella. hiemalis
Keratella. quadrata
& 00
:= d
!O
CD
o to
O d
0 "*
3 °'
-I—*
Q.
CD CN
O d
q
d
0 0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
CM 0
OM O
d C
CD
E
0
o >
si
o a:
£* oo
:= d
CD
o to
O d
Q.
0 J
13 °
-i—*
Q.
CD CN
O d
q
d
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
o
d
•* 0
o o
d C
CD
« c
§ §
c, <
O 0
d >
J5
o 0
o a:
^> 00 _
1=0
CD
o to _
O d
CL
0 •*
13 °
-i—*
Q.
CD CN
O d
q
d
U 8
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
^r 0
o o
d C
CD
« C
q S
0 E
N <
O 0
d >
'•4—'
JD
o 0
o a:
Keratella. taurocephala
Lecane
Lecane.inermis
p _
T~
^ 00
:= d
!Q
CD
o to
0 d
CL
0 •*
3 °
"o.
CD CN
O d -
0
d
.
'
.
' -' .
•• ."• "' " • •
o BV 1 liiAiP*! *• Vnn rf^rc** • •
1 1 1 1
- OM P _
d "~
_ 00 0 >, oq _
d C •= °
CD ^2
_,. ~C CD
C O CD
• o =3 0 d
< ^
-80 0 "* _
° £ 3 °
"CD "o.
. § 0 ro cs, _
OQ: o °
„ o
0 o
.
0
"
0
"
•
.
• • • „ •
• • " " . •
"
1 1 1 1
. 0 P _
d "~
.88 £•» _
d C •= °
CD .Q
. 8 1 & <° _
o 3 O o
< ^
-50 0 "* _
d > ^ °
"CD "o.
.00 CO
"CD
-00
d a:
~ O
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
Page 10 of 14
-------
Capture Probability of Zooplankton Taxon Along TN Gradient
Leptodlaptomus. ml nut us
& oq
:= d
cq
d
q
d
U §
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0
o
CD
in .-
q c
d
l
sl
o o:
Leptodlaptomus. slcllls
in
>, d
2 CO
CL °
£ N
-3 d
Q.
O d
CD
° -Q
d <
>
"CD
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
Mesocyclops. edax
CD
o to
O d
Q.
CD CN
O d
U d
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
'•4—'
JD
o 00 _
:= d
o
CD
o to
0 d
CL
0 •* _
3 °
Q.
CD OM
O d -
0
d
.
" . . . • , . •
1 1 1 1
- q P _
d ^~
.88 ^oo _
d C = °
CD .Q
m "° ro
. 0 £= ^2 D
'-I—' -1—'
CD Q.
'IS Oo-
._. o
° 0
.
•
*
• .
••!^_' • *ii- •'.
« «M^HB «•" ^••••B
1 1 1 1
. 0 P _
d "~
.08 ^oo _
d C = °
CD .Q
in "° ro
. 0 £= ^2 •* _
d > ^ °
'•4-* 4-*
CD Q.
.80 m g _
._. o
° 0
.
•
.
*
• •
* m
\ •'•
•
1 1 1 1
_ o
d
ro CD
- o O
d C
CD
"^
O ^
~ • ^
0 .Q
-50
d .>
'•*-*
CD
.50
~ o
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
Page 11 of 14
-------
Capture Probability of Zooplankton Taxon Along TN Gradient
Ploesoma
& 00
:= d
CO
d
q
d
. U '• '• /'•
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
t~~
d
o C
ro
T3
•<- c
o 0
d >
si
Ploesoma.lenticulare
.a
ro
O CO
O d
0
"Q.
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
Polyarthra. dollchoptera
q
d
^ 0
o O
d C
ro
T3
5- i
0 E
o 0
J5
o
si
Polyarthra. major
^> 00
:= d
tp
d
q
d
Polyarthra. remata
Polyarthra. vulgaris
^
.
•
•
•
•. •.
• .
a ° •
•'iL£i ••**"•'/•'
1 1 1 1
. 0 P _
d "~
in CD >, m
- o o -i? °P _
d C r^ °
ro .a
_,. T3 ro
S C ^2 5 °
^-^ ^-^
ro Q.
.50 ro ,
d C := d
ro .Q
^. T3 ro
" d _g 2 § -
-20 ^ ^r
"Jg "Q °
.80 ro
d Q^ O
d
- 0
t
.
• . : /
0 » •*
a •
* V *" * *• "• •
*° % """•*« •*.•" * * "
• ". " d •"'•L» *,* •
. ' •'/»>' • • *
• a* • •
1 1 1 1
\m j
d
,„
0 ro
0
•* IV
- 0
d
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0 0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
Page 12 of 14
-------
Capture Probability of Zooplankton Taxon Along TN Gradient
Pompholyx
in
d
CD
.Q
O co
Ql d
Q.
f ^
O ci
q
ci
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0 0
oo g
§-§
CD
O
.a
S ®
<=>. >
O '4=
CD
s I
O 0-
Skistodiaptomus. oregonensis
_g> 00
:= d
!Q
CD
O CD
O d
0 "*
3 °'
-I—*
Q.
CD CN
O o
q
ci
0 0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
Synchaeta
oo CD
o O
d C
CD
CD "a
§1
t *^
o 0
d >
"CD Q.
80 co 00
:= d
!Q
CD
O CD
O d
0 "*
3 °'
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
r-
'•4— '
. CD
o 0
o a:
Trichocerca. birostris
Trichocerca. cylindrica
Trichocerca.multicrinis
p _
T~
_^> 00 _
:= d
o
CD
O CO _
O o
CL
0 •* _
3 °
-1—*
Q.
CD CN
O o ~
0
d
.
•
B
._
0 *B
* *. • •
' " " " "•*
« ' « .*
•" S-:.V ' . •• '.".• •
1 1 1 1
. 0 P _
d "~
.88 ^oo _
d C := °
CD ^2
"O CD
g C ^2 CD
d ^ O o
< ^
-50 ^ °
'-I—' -1—'
CD Q.
.50 CO CN _
0 o
0 o
.
" e
•
• ' ': * •.
• .*" *" .
.. . A*. • \ ^' '. t .'.
1 1 1 1
JI q _
d ^~
.88 ^oo _
d C :^ °
CD .Q
. T3 CD
. 0 C .Q CD
d ^ O o
< ^
-80 ^ °
'•4— ' -I— '
CD Q.
.80 CD CN _
OQ: o °
0 o
0 o
.
.
,
•
'
.
"
•. " • • •
• •• * • * •
'.*•
. •• • •• • •
' '* "*" V ' ' %
i
1 1 1 1
_ o
d
CN CD
.00
d C
CD
,%, "^
- P C
5
.50
d >
'•4— '
CD
.50
o o:
~ o
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0 0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
Page 13 of 14
-------
Capture Probability of Zooplankton Taxon Along TN Gradient
Thchocerca.porcellus
Trichocerca. rousseleti
Trichocerca. si mi Us
o
•<-
,>> 00
™ d
o
CD
O CD
O d
CL
CD •* _
.3 °
"o.
CD CN
O o ~
0
d
•
.
B
• ..
•
'
•
1 1 1 1
- 0 P _
d "~
T- CD •>,
.00 -^ °° -
d C — o
CD .Q
T3 CD
_ o c ^2 CD _
0 13 00
o ^_
< '-L-
-00 0) ^ -
> 3 °
"CD "o.
_ "m CD CN
— O VI* s *i ~~
o: o o
- 0 °- -
.
•
X
• • • • x
x'/
" x x *
, -'^^^^^
• ^ """"""^ •
— — '^<"'" ' " '.
--"~*"* * •.
1 1 1 1
- 0 P _
d "~
T- CD •>,
-00 -^ °° -
d C 1= °
CD .a
T3 CD
. o C ^2 CD _
d ^ O o
< ^
.50 22"
+= 5
CD Q.
_ "m CD CN
— O VI* s *i ~~
a: o o
- 0 °- -
.
* B
^ *
• • "
• * . . .
• *
1 1 1 1
- 0
d
CN CD
- o O
d C
CD
CN "°
CN ^
- P §
_Q
-00
0 >
CO
o CD
o a:
~ O
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
Tropocyclops.prasinus.mexicanus
>,
1^
o
CD
^2
0
Ql
0
3
"o.
CD
O
o _
•>-
00
d
CD
d
•* _
°
CN _
0
d
.
.
m
' ' "'.
, ..• ]%•.;
•'• — -*'wv;
ff *
, • f
• •!**• " •
'
T—
d
00
- O
d
- P
- O
d
- P
- 0
0
O
CD
C
13
^
0
>
'•4— '
CD
0
0.1 0.2 0.3 0.4
Total Nitrogen (mg/L)
Page 14 of 14
-------
Appendix 6 - Wl Chi a GAM Models Zooplankton
-------
Capture Probability of Zooplankton Taxon Along Chi a Gradient
Asplanchnidae
Bosminidae
Brachionidae
"-
•O tX)
:= d
CD
in CD
O d
CL
0 •* _
^_«
Q.
CO CN
O, — ; ~~
°
0
d
Q
q _
^* 00
:= d
o
CO
O CD
2 °
CL
0 •* _
2j °
"o.
CD CN
O o
0
d
• •
, - * "
f f ^^^^^^^
~ ~^*" —
.-- •••'• • •
• i • |Mi3J|VB*1^|ar*
1 l l l l ll| l l l l l l ll| l l l l l
.3 1 3 10 30
Chlorophyll a (|ig I )
Calanoida
f
.
•
* • •"** 4* • *
O -^
00 CD >s m
_ 3r o ^ °°. _
d C = o
CO .Q
CD "° tO
. S C J2 & _
dl £°
- CN 0 0 ^ -
0 > ^ °
•4-^ ^-^
CD Q.
_ 2 0) .tO CN _
o Q-; Q o
-_, 0
. o d -
0
P _
d "~
in tD >, _
_ "} O 4? "P _
O ^ I — O
CD !Q
^ "C CO
08 C ^2 CD
o ^ O o
< ^
- CN 0 0 ^ _
d > 21 °
'•4— ' -1— '
CD Q.
. $2 "0 CO CN _
d Ct O °
o
° d ~
r777tr:<<^ -...--•-""
" x^x""11--»^
x
v ^ v
x
• x
B \
\
\
\
\
\
\
.'
*
' . ' ' '
" • « * B
/.:•• :•-.., ':
• /-v^^fci*l^ki^v. „
i 1 1 1 1 ni i i 1 1 1 1 MI i i 1 1 1
3 1 3 10 30
Chlorophyll a (|ig I )
Chydoridae
'
„
f
•
• . • .
• . .;<•• J . '.. ' '
• "•• .••«,VL"«Tlti".'.-yu.i.'.'- •
d "~
„ 0 >, m
CO *\ ^V CO
d c: :=d
co !Q
rsi ~° tD
. ^ C ^2 ZJ °
4-^ ^-^
CO Q.
- 0 0 /O CN _
OQ: o °
„ 0
. o d -
0
P _
d "~
to 0 >, m
.00 4? °° _
d C := °
CO .Q
.- T3 CD
. 0 £= ^2 21 °
'•4— ' -I— '
CO Q.
.80 tO CN _
d Qi O °
o
. o d —
t
• • *
•••.*.."'•'
• • t *+ ' •
• .... v *
"• i '-. •'• '..*.•'
. -Vj*. .-*•.'•• •
•'. .".*«•'••,-. '. '
" ' "••:- s- ":• '
• "*• — • " *•,
* •*"•
«•• « •
i 1 1 1 1 MI i i 1 1 1 1 MI i i 1 1 1
.3 1 3 10 30
Chlorophyll a (|ig I )
Collothecidae
'
,
. e
, ,
.
• . ^ 0
• •' J « '. " •
•- • *i" *« !v£e LA"*"
i
d
CO 0
- CD O
d C
CD
^r ^
" ° E
- CO 0
'•4— I
- s "0
o a:
- o
d
rs, tD
. ^ O
O (~
CD
"^
^r- ^
~ ^ ^
° ^2
- H 0
.>
"-i—'
CO
. 8 0
o o:
- O
0.3 1 3 10 _30
Chlorophyll a (|ig I )
0.3 1 3 10 _30
Chlorophyll a (|ig I )
0.3 1 3 10 _30
Chlorophyll a (|ig I )
Page 1 of 19
-------
& 00
:= ci
ro
o to
O o
CL
0 "*
Q.
CD c\i
O o
q
o
Capture Probability of Zooplankton Taxon Along Chi a Gradient
Conochilidae Copepod.nauplii Copepodites
," ••••
•:rrvA..:^ •
""ft -. .
0.3 1 3 10 _30
Chlorophyll a (|ig I n)
00
ci
0
O
ro
T3
c
3
CD
- 0
.Q
ro
o to
O ci
CO 0 0
Q.
ro
o
q
ci
0.3 1 3 10 _30
Chlorophyll a (|ig I n)
S
o
O) 0
c\i o
ci C
ro
~, "o
T- 0
o >
. "ro
o 0
& 00
:= ci
ro
o to
O ci
CL
0 •*
13 °
-i—*
Q.
ro c\i
q
ci
0.3 1 3 10 30
Chlorophyll a (|ig I n)
ro 0
OM O
d C
CD
*- 0
o >
§1
o a:
Cyclopidae
Cyclopoida
Daphniidae
p _
T~
^ °q _
:= °
o
ro
o to
0 ci
CL
0 •* _
<= 0
"o.
ro c\i
O o -
0
ci
.
. •
...
.
• •
• .8
.... '••
• • "•••*. "•
••:";Vv'*V' '
".'• '»'•$&£'"•• '•
"
;- P _
ci "~
. - g ^oo _
0 C := °
ro !Q
. T3 ro
o £= ^2 <°
"dl l°"
-00 0 ^ _
0 I 3 °
"ro "o.
.80 Cl ! rs _
OQ: o °
„ 0
. o d -
.
•
. .
• •
.
•,, -i&Lfr -^ . •.
-in P -
d ^~
^8^"°
d C = °
ro .Q
_,. T3 ro
ro CI ^2 (D
"d| !°~
. CN 0 0 ^ _
° .> ^ °
"ro "o.
. ^ -0 ro
"ro
. 8 0
o a:
- O
0.3 1 3 10 _30
Chlorophyll a (|ig I )
0.3 1 3 10 _30
Chlorophyll a (|ig I )
0.3 1 3 10 _30
Chlorophyll a (|ig I )
Page 2 of 19
-------
& 00
= d
CD
o to
O d
CL
0 •*
3 °
-i—*
a.
CD OM
O o
q
ci
Capture Probability of Zooplankton Taxon Along Chi a Gradient
Dlaptomldae Euchlanldae Gastropodldae
0.3 1 3 10 _30
Chlorophyll a (|ig I n)
2 8
d C
CD
o, <
O 0
d >
o 0
o a:
>,
'—
o
CD
.Q
2
Q.
2
"o.
CD
O
q _
•<~
00
d
CO
d
•* _
0
OM _
0
d
o
•o •
o
d
OM
- O
d
Q
d
- 0
d
- o
- o
O
CD
T3
f~
.Q
0
>
J5
o:
>,
-i—*
• —
o
CD
.a
o
CL
0
3
"o.
CD
O
q _
•<~
00
d
CO
d
•* _
0
OM _
0
d
f
.
.
.
• • • "•*. a
"* * *'
d
^ 0
_ in o
d C
CD
"^
CO ^
d 1
- CN 0
d >
ID
.20
o a:
- O
0.3 1 3 10 _30
Chlorophyll a (|ig I n)
0.3 1 3 10 30
Chlorophyll a (|ig I n)
Holopedlldae
Lecanidae
Lepadellldae
q _
^
& °P
:= d
o
CD
o to
0 d
CL
0 •* _
3 °
"o.
CD OM
O o
0
d
.
•
.
"
•
«• a
- *"'''•''&"...
*• «.VP fx&f&SEfc •
_ ;! q _
d ^~
.88 ^oo _
d C :^ °
CD .Q
. T3 CD
0 C ^2 (D
d E ° °
^ d
"CD "o.
OM — m in
- O 0 lu O —
d Q: o o
o
- o o —
(— i
.
•
•
•
a • •
•
- 0
d
- o O
d £=
CD
T3
3
^
-00
J5
0 o:
~ O
0.3 1 3 10 _30
Chlorophyll a (|ig I )
0.3 1 3 10 _30
Chlorophyll a (|ig I )
0.3 1 3 10 _30
Chlorophyll a (|ig I )
Page 3 of 19
-------
Capture Probability of Zooplankton Taxon Along Chi a Gradient
Nauplll
^> 00 _
1=0
CD
O CO _
O d
CL
0 •*
13 °
-i—*
Q.
CD CN
O o
q
ci
0.3 1 3 10 _30
Chlorophyll a (|ig I n)
CN 0
in o
d C
CD
O) f—
" §
0 E
CO <
CN 0
d >
a:
o
CO
d
in
^2
P
O m
Q- o
2 o
I.5
CD m
o
P -
d
Notommatidae
Polyphemldae
0.3 1 3 10 _30
Chlorophyll a (|ig I n)
q
o
C 1=
P ^2
O ^1
0
"CD
& 00
:= ci
CD
O CO
O d
CL
0 •*
13 °
-i—*
Q.
CD CN
O o
q
ci
0.3 1 3 10 30
Chlorophyll a (|ig I n)
0
o O
CD
T3
o 0
"-^
JD
o 0
o:
Rotifera
Sididae
Synchaetidae
p _
T~
_^> 00 _
:= d
o
CD
O CO
0 d
CL
0 •* _
3 °
"o.
CD CN
O o
0
d
.
0
*
•
„ . • •
. 00
•....• ••.. •• • . .
• • •» ,* °«o • ' •
. 0 P _
d "~
. s 8 ^»_
d C := °
CD .Q
"O CO
. 8 £= ^2
"CD
. ^ 0
~ O
0.3 1 3 10 _30
Chlorophyll a (|ig I )
0.3 1 3 10 _30
Chlorophyll a (|ig I )
0.3 1 3 10 _30
Chlorophyll a (|ig I )
Page 4 of 19
-------
Capture Probability of Zooplankton Taxon Along Chi a Gradient
Temoridae
Testudinellidae
Trichocercidae
o
•<-
,>> 00
™ d
o
CD
o to
O d
Q.
.
-00 -^ °° -
d C 1= o
CD .£2
T3 CD
. 5 C .£2 CD _
° E 2 °
< °-
-o0 | 2 -
"-^ -i—'
CD Q.
- o 00
:= °
CD
O (D
0 d
Ql
0 •* _
3 °
"o.
CD OM
O o i -
0
d
0
•
§
0.3 1 3 10 _30
Chlorophyll a (|ig I )
Unknown
Acanthocyclops. vernalis
00
d
^
CM
d
CM
d
2
d
fe
o
O
0
o
CD
T3
C
^
^2
0
>
'•4— '
CO
0
tt-
p _
^~
-1—' . —
1— O
!Q
CD
o to
O d
0) ^ _
3 °
-i—*
Q.
tD CM
O o
0
d
t
. ' *
d
O)
d
^"
•
- O
d
- 0
o
~ o
0
O
c
CD
T3
C
^
^2
0
>
'"CD
0
DC.
p _
•<~
-i— • . _
r^ ^
o
CD
o to
O d
0) •* _
3 °
-i—*
Q.
tD CM
O o
0
d
a
•
•
.
°
•
. ' ,
*'•••• •' , '
o
d
CM
- O
d
_ o
- O
d
- o
— o
0
O
C
CD
T3
3
.a
0
>
'•4— '
JD
a:
0.3 1 3 10 _30
Chlorophyll a (|ig I )
0.3 1 3 10 _30
Chlorophyll a (|ig I )
Page 5 of 19
-------
Capture Probability of Zooplankton Taxon Along Chi a Gradient
Alona
i=
CD
.Q
O
CL
0
-1—*
Q.
CD
O
o
00
d
CD
d
2 -
C\l _
d
0
d
.
•
8
^ 0
o O
d C
CD
T3
°- i
0 J2
q
d
0
JD
o 0
o:
Ascomorpha. ecaudls
_g> 00
:= d
CD
O CD _
O d
U "
-I—*
Q.
CD CN
O o
q
d
U 8
1^ 0
o O
d C
CD
•« c
q §
o ^>
0
CO
q
d
sl
o o:
Ascomorpha. ovalis
_g> 00
:= d
CD
O CD
O d
CL
0 •*
3 °
-I—*
Q.
CD CN
O o
q
d
o
in 0
o O
d C
CD
s§
o 3
CO
q
d
0
"-^
JD
o 0
o o:
0.3 1 3 10 _30
Chlorophyll a (|ig I n)
0.3 1 3 10 _30
Chlorophyll a (|ig I n)
0.3 1 3 10 30
Chlorophyll a (|ig I n)
Asplanchna
Bosmina
Bosmina.longirostris
q _
•<~
^ 00
:= °
o
CD
O CD
0 d
CL
0 •* _
3 °
"o.
CD CN
O o -
0
d
^
•
i>
* *
* •
°
• " *\ • *
• * "
•»' •"/•£ •.'.
"• > • Jt^Jlk'f'UH''
CD O _
d ^ ~
. i 8 ^» _
d C := °
CD ^2
. S 1 & «P _
O ^ O O
< c^-
- CN 0 0 "* _
d > ^ °
'•4— ' -I— '
CD Q.
- ^ 0 to rsj _
„ 0
. o d -
t
*
0
•
. •
€
•*
. CO
d
<^> o
d ^
CD
rsj "g
d ^
^
- ^ 0
0 >
'"CD
. 0. 0
O
^00
0 i;
<= d
-i—*
Q.
CD
K_ tD
. 0 0
o o:
0.3 ^ 3 ^0 _30
Chlorophyll a (|ig I )
0.3 1 3 10 _30
Chlorophyll a (|ig I )
0.3 1 3 10 _30
Chlorophyll a (|ig I )
Page 6 of 19
-------
Capture Probability of Zooplankton Taxon Along Chi a Gradient
& 00
:= ci
CD
.a
o
CL
0
-—
Q.
CD CN
O o
q
ci
Calanoid
8
o
- 8
0 c
CD
oo "a
" §
0 E
& 00
:= d
CD
O CO _
O d
CL
CN 0 £ 5j
O > 3 °
'-I—' -I—'
_ CD Q.
<*> — CD CN
O
q
d
Ceriodaphnia
^ 0
o O
d C
CD
T3
O 0
d >
•^
JD
o 0
o:
& 00
:= d
CD
O CO
O d
CL
0 •*
13 °
-i—*
Q.
CD CN
O d
q
d
Chydorus
U §
CO 0
o O
d C
CD
•« c
q §
0
O 0
d >
0.3 1 3 10 _30
Chlorophyll a (|ig I n)
0.3 1 3 10
Chlorophyll a (|ig I
_30
n
0.3 1 3 10 30
Chlorophyll a (|ig I n)
Collotheca
Collotheca. mutabilis
Colonial, conochilus
p _
T~
^ 00
:= °
o
CD
O CO
0 d
CL
0 •* _
2j °
"o.
CD CN
O d
0
d
.
•
• t
.
.
' ^ . •
• ** • * •
- CN P _
d "~
. - g ^00 _
0 C := °
CD !Q
.- T3 CD
10 C ^2 co
" o 13 0 d
.Q >-
^ °
'-I—' -i—*
CD Q.
. § 0 ro CN _
OQ: o o
„ o
° 0
.
•
.
• •
* B ,
•
' .' • V.».Y:.-.
. 0 P _
d ^~
§ 8 ^ra
d C •= °
CD ^2
_,. T3 CD
. S C ^2 cq
o ^ O o
f^ ^
<£ ^~
-80 CD ^ _
d _> D °
'-I—' -i—'
CD a.
.50 CO CN _
OQ: o o
o
O
.
.
£ .'V. • "• i . i i.
u;
d
CN 0
d C
CD
- 8§
f^
<-|-'
CD
- O 0
0 _>
"-i—'
CD
. 8 0
o a:
~ O
0.3 1 3 10 _30
Chlorophyll a (|ig I )
0.3 1 3 10 _30
Chlorophyll a (|ig I )
0.3 1 3 10 _30
Chlorophyll a (|ig I )
Page 7 of 19
-------
Capture Probability of Zooplankton Taxon Along Chi a Gradient
Conochiloides
Conochiloides. dossuarius
Conochilus
^> 00 _
1=0
CD
O CD
O d
CL
0 •*
13 °
-i—*
Q.
CD CN
O d
q
o
CD
0
,>> 00
:= d
CD
O CO _
O d
Q.
0 •*
3 °
-I—*
Q.
CD
& oq
:= d
CD
O CO _
O d
CL
0 •*
13 °
-i—*
Q.
CD CN
q
d
00
d
in 0
CD O
d C
CD
CO 0
d >
a:
0.3 1 3 10 _30
Chlorophyll a (|ig I n)
0.3 1 3 10
Chlorophyll a (|ig I
_30
n
0.3 1 3 10 30
Chlorophyll a (|ig I n)
Conochilus. unicornis
Cyclops, varicans. rubellus
Daphnia
CN —
0
>, 0 _
:= o
CO
o ^
0 d ~
Q.
0 ° _
^ d
"o.
to o
O d
o
O —
(— i
.
•
•
•'
• . •
-?- p _
d "~
.28 ^» _
d C = °
CD .Q
T3 CD
^1 C .Q co
"dl F"
-00 0) ^ _
0 £ 3
_ "CD "o.
.30 ro CN _
OQ: o °
„ o
0 o
.
a
,
• '••*.*
. 0 P _
d "~
.88 ^"o _
d C = °
CD .Q
,n ~& TO
. § C ^2 to _
d| !°
.80 0 ^ _
° £ 3 °
"CD "o.
. 8 0 co CN _
OQ: o °
„ o
0 o
.
. "
•
0 • *
^
- CO
d
in 0
- CN O
d C
CD
"^
c
" 5<
- ^ 0
0 >
"CD
. 8 0
o o:
~ O
0.3 1 3 10 _30
Chlorophyll a (|ig I )
0.3 1 3 10 _30
Chlorophyll a (|ig I )
0.3 1 3 10 _30
Chlorophyll a (|ig I )
Page 8 of 19
-------
Capture Probability of Zooplankton Taxon Along Chi a Gradient
Daphnla.dubla
Daphnia.longiremis
Daphnia. mendotae
& 00
:= o
!Q
CD
O (D
O d
CL
0 "*
3 °'
Q.
CD CN
O o
q
d
u s
i
.02 0.02 0.03
e Abundance
-I—'
JO
-50
o o:
obability
0.6 0.8
0 ^
3 °'
-I—*
Q.
CD CN
q
d
£
in 0
o O
d C
CD
s§
° E
00 <
O 0
d >
•i=
JO
-50
o o:
p _
T~
_^> 00 _
™ d
!a
CD
o to
O d
CL
0 •*
3 °' "
-1—*
Q.
tO (N
O o
0
d
.
.
.
•
• • ' '.' .
•
. . ' * . f .. ' '
''•' .' '
•*""•*"* * •
• "•" "° * *0
•?•••..
- 0.
d
oo 0
- o O
d C
CD
.8?
° E
<
-00
d >
'•4— '
CD
-00
o fV
- 0
0.3 1 3 10 _30
Chlorophyll a (|ig I )
0.3 1 3 10 _30
Chlorophyll a (|ig I )
0.3 1 3 10
Chlorophyll a (|ig
30
n
Daphnia. parvula
Daphnia.pulex
Daphnia.pulicaria
in _
d ~
-*- • . _
:= o
o
CD
O 00
0 d
Q-
0 Os| _
^ °
"o.
CD T-
O o i -
0
d
B
.
d
« B
• '
, .
^
'..'•'
' . •
• ' • •' •
»" .^
.5 ° -
d ^~
*- tD >, _
- o O 4? °° _
d C r^ °
CD .Q
T3 CD
_ C o to
" ° 3 0 d ~
^ -
> 3 °
"CD "o.
- 0 0 /? ^ -
o: o o
._. o
° 0
t
•
•
B
.
.
.
• a • '
• s.
• . ».* •
. 0 , "? _
- 0 O -i^ 0
d C =
CD ^2 —-
^. T3 CD 3 ~
- § § o
-Q i- oo _
-80 0 °
d > ^ CN
"CD "o. °
- o 0 to
do: 0 d -
._. o
° 0
B
*
.'
.
% .
•.••» •:
• "" " "* 0 * "
• ^" V; *.rf. lr*"0"|
_
. S 0
d <->
0 c
CD
T3
C
?! -Q
_
0
"CD
- q &
~ O
0.3 1 3 10 _30
Chlorophyll a (|ig I )
0.3 1 3 10 30
Chlorophyll a '••-•-*
0.3 1 3 10 _30
Chlorophyll a (|ig I )
Page 9 of 19
-------
_g> 00
:= d
CD
o to _
O d
CL
0 •*
Q.
CD CN
O o
0
0
Capture Probability of Zooplankton Taxon Along Chi a Gradient
Daphnia.retrocurva Daphnia.rosea Diacyclops.thomasi
0.3 1 3 10 30
Chlorophyll a (|ig I n)
8 P
d "~
8 8 £>»
d C 1= °
CD .a
^. T3 CD
§ i | §
d > 5
'-I—' -I—'
CD Q.
o
0.3 1 3 10 30
Chlorophyll a (|ig I n)
CN
d
0
° H
CD
^ C
P §
o ^
in
o
0
sl
o a:
:= o
la
CD
O CD
O d
0)
o
. "
0.3 1 3 10 30
Chlorophyll a (|ig I n)
0> 0
o O
d C
CD
P i=
0 E
in
o 0
d >
8 0
o a:
Diaphanosoma
Diaphanosoma. birgei
-1—*
o
CD
.Q
0
CL
0
3
"o.
CD
O
o _
•<~
00
d
CD
d
•* _
o
CN _
0
d
t
•
i'' • •• .
"J-l
d
CO
d
CN
d
. ^
d
. 8
d
~ O
0
o
CD
"^
c
^2
<
0
>
"CD
0
a:
O _
^~
>- 00
j= °
o
CD
O CD
0 d
CL
0) ^ _
3 °
"o.
0
d
B
•
*
.
* »o •
-'~**i?i2£i£jL«L '
•^
°
00
- O
d
- 0
d
- O
d
' d
o
0
O
CD
T3
C
^2
0
>
'"CD
0
>,
:^
!Q
CD
0
0
3
"o.
CN
d
o
CN
d
^2
°
2
d
§
d
o
o
r~i
0.3 1 3 10 _30
Chlorophyll a (|ig I )
0.3 1 3 10 _30
Chlorophyll a (|ig I )
Dlaptomld
>,
!5
CD
0
0
3
-i—*
Q.
U
d
o
CN _
d
o
o _
d
§ _
d
o
o _
r~ i
•
.. . '
. •
•
0.3 1 3 10 _30
Chlorophyll a (|ig I )
r^ 0
o O
d C
CD
§
d
o 0
o >
si
Page 10 of 19
-------
Capture Probability of Zooplankton Taxon Along Chi a Gradient
Diaptomus. oregonensis
CD
O
CL
0
^
0
0
00
o
8
CD
-
- °. "0
0
Eplschura.lacustrls
Eucyclops.agilis
±±
!5
CD
.Q
o
2
^
CD
o
p _
T
d
CO
d
•* _
OM
°
0
d
.
0
%
K
*• ^ ^
,-^T^\"---
- - * ' ••' 'V' ^N^
• ••'*. i r ^ x » •-»_
° • "tL n ..A.,. " "* — ^**^*^
- 0
d
. o
d
o
d
- 0
~ O
0
o
CD
-o
f~
|
0
"0
o:
~ d
o
CD
O OM
Q. °
0
5
CD °
O
0
d
a
0
•
•
•
.
•
•
0
- o 0
CD
-o
C
- 0 .Q
0
'•^
"0
o:
~ o
0.3 1 3 10 _30
Chlorophyll a (|ig I n)
0.3 1 3 10 _30
Chlorophyll a (|ig I n)
0.3 1 3 10 30
Chlorophyll a (|ig I n)
Filinia
Filinia.longispina
Filinia.terminalis
^> 00
:= °
CD
O CD
O d
Ql
0 •*
^ °
-I—*
Q.
CD OM
O o
p
d
r^ 0
CN O
o C
CD
-o
CN C
O 3
^2
" 0
o >
'•4—'
K, ro
fe 0
o Q:
.a
CD
O CD
O d
0
-I—*
Q.
O) CD
p O
o C
CD
^ C
o 0
d >
si
o a:
CD
o
0
Q.
CD
p
d
p
d
•
• •'•J
•
:'.o
I
d
0
' d °
CD
'I 1
- o 0
d >
.81
o a:
0.3 1 3 10 30
Chlorophyll a (|ig I )
0.3 1 3 10 _30
Chlorophyll a (|ig I )
0.3 1 3 10 _30
Chlorophyll a (|ig I )
Page 11 of 19
-------
Capture Probability of Zooplankton Taxon Along Chi a Gradient
Gastropus
^> 00 _
1=0
CD
o
O
CL
0
Q.
CD CN
O o
q
ci
CD 0
q O
d C
CD
•« c
O ^
o 5
«<
q 0
si
o a:
Gastropus. hyptopus
,>> oo
:= d
CD
O CD
O d
CL
0 •*
Q.
CD CN
O o
9
o
^ 0
cq O
d C
CD
^ "O
CO <
T- 0
o >
§1
o a:
Gastropus. stylifer
p _
T~
,>> 00
™ d
o
CD
O CD
O d
CL
0 •* _
13 °
-i—*
Q.
CD CN
O o ~
0
d
.
B
/
•
0 "
•• - • •
•£ .As
• • ••Mk£Z
. . .•+tf«RB&*
•
• .t
• °
••• * •
'.:•»*•
£&&»&• *.. .
•
V N
- CO
d
CD 0
- CN O
d C
CD
"U
T— ^
d -^
^
-20
0 >
'•4— '
-§l
o Q:
- 0
0.3 1 3 10 _30
Chlorophyll a (|ig I n)
0.3 1 3 10 _30
Chlorophyll a (|ig I n)
0.3 1 3 10 30
Chlorophyll a (|ig I n)
Holopedlum
Holopedium. gibberum
Kellicottia. bostoniensis
q _
^
^* 00
:= d
o
CD
O CD
0 d
CL
0 •* _
3 °
"o.
CD CN
O o ~
q
.
•
.
• •
.
_ ;! q _
d "~
0) 0 >, m
- o O 4? °° _
d C r^ °
CD ^2
. T3 CD
^ C .Q CD
" ° E 2 ° "
3 °
., "CD "o.
. 8 0 to CN _
d CC O o
P_
.
a
•
.
• 0
.• . •
'• . "•' *
•' .• v ^
'• . /;??-'*! •
. . .rfc^to':- „ ..
. 0 P _
d ^~
"0 >, m
- o o 4? °° _
d C r^ °
CD .Q
"O CO
. 8 £= ^2 ^ °
"CD "o.
. 5 0 to CN _
d c£ O o
P_
.
•
•
0
" .
•
" • •*
f ' ' «o
V N
- in
d
^ 0
. ^r O
d C
CD
T3
- co £=
°<
- CN 0
d >
"CD
- o £
- 0
0.3 1 3 10 _30
Chlorophyll a (|ig I )
0.3 1 3 10 _30
Chlorophyll a (|ig I )
0.3 1 3 10 _30
Chlorophyll a (|ig I )
Page 12 of 19
-------
& 00
:= ci
CD
o to
O o
CL
0 "*
Q.
CD CN
O o
q
ci
Capture Probability of Zooplankton Taxon Along Chi a Gradient
Kellicottia.longispina Keratella.cochlearis Keratella. crassa
0.3 1 3 10 _30
Chlorophyll a (|ig I n)
t~. 0
•* <->
d C
CD
CD "a
SE
00
:= o
CD
o to _
O d
CL
0 •*
13 °
O
q
ci
0.3 1 3 10 _30
Chlorophyll a (|ig I n)
8
CD
s
o a:
Keratella. earlinae
Keratella. hiemalis
>,
j^
o
CD
.Q
0
CL
0
^
"o.
CD
O
p _
T~
00
d
to
d
•*
d
CN _
0
d
.
.
•
^
>
•
. "
• B •
...-.*/£';• v'
'• 7::%VV?'-'f!L'.'
- OM
d
CN
d
^
~ •
_ ,_
d
- 9
- 0
0
O
CD
T3
C
^
^2
0
>
"CD
0
p _
^
>, CO _
1— O
!Q
CO
O CD
0 ci ~
1_
0 2 -
3 °
"o.
CD "«*. •'.
. 6
d
^r
. o
d
- 9
OM
- O
d
- 0
d
- 0
0
O
CD
T3
C
^
5
0
>
'ro
0
a:
0.3 1 3 10 30
Chlorophyll a (|ig I )
0.3 1 3 10 _30
Chlorophyll a (|ig I )
O o
0 •*
-i—*
Q.
q
d
Keratella. quadrata
«
*. ..*.••-
. 8
0.3 1 3 10 _30
Chlorophyll a (|ig I )
^r 0
o o
d C
CD
T3
o 0
d >
^
o 0
Page 13 of 19
-------
Capture Probability of Zooplankton Taxon Along Chi a Gradient
Keratella.taurocephala Keratella.testudo Lecane
"-
_^> 00 _
™ d
o to
O d
CL
0 •* _
3 °
-i—*
Q.
CD c\i
O o ~
0
d
0
q _
_^> 00
:= °
o
CD
o to
0 d
Q-
0 •* _
3 °
"o.
CD c\i
O o
0
d
- ^
~ ~ - - _--""""
~
^^^~
" •" " ** *.
• ^ •
• • f "
•' •
• • *• .""".* "*
. '* •"•*»"»•«.*• •* *• •• "
i i i i i ill i i i i i i ii| i i i i i
.3 1 3 10 30
Chlorophyll a (|ig I )
Lecane.inermis
'
•
f
.
*•*•..•.'•*•. .' • • •
d -
. - 8 ^»_
o c: — o
CD .Q
"O CD
co c o to
d 3 O d
< CL
. ^ 0 0 ^ _
0 > ^ °
'-I—' -1—'
CD Q.
"~ "0 tO CM
d Q-; O d
„ 0
. o d -
0
P _
d "~
. § 8 £* » _
d C •= °
CD .Q
"O CO
. 8 £= ^2 -
^ °
'-I—' -1—'
CD Q.
- o 0 tO —
< °-
- ^ 0 0 ^ _
d > ^ °
'•4— ' -1— '
CD Q.
- § 0 .CO cs, _
OQ: o °
„ 0
. o d -
0
i —
^ P _
d "~
.28 &°=. -
d C •= °
CD ^2
T3 CD
T— (~ Q CD
' d 3 0 d
.Q >-
^ °
'•4— ' -I— '
CD Q.
. 8 "0 to
'•4— '
CD
-50
- o
d
^ 0
- o O
d C
CD
^ T3
_ 5 ^
d _3
^f
- o Q)
0 _>
"-i—'
CD
-00
o:
~ O
0.3 1 3 10 _30
Chlorophyll a (|ig I )
0.3 1 3 10 _30
Chlorophyll a (|ig I )
0.3 1 3 10 _30
Chlorophyll a (|ig I )
Page 14 of 19
-------
& 00
:= ci
CD
O CO
O d
CL
0 •*
13 °
-i—*
Q.
CD CN
O o
q
o
Capture Probability of Zooplankton Taxon Along Chi a Gradient
Mesocyclops Mesocyclops.edax Monostyla
0.3 1 3 10 _30
Chlorophyll a (|ig I n)
t~. 0
q o
d C
CD
co "2
§1
-00
CD
& 00
:= d
CD
O CO
O d
0
0 00 _
CD
o cq
O d
00 0
d > ^
'•4—' -I—'
JD Q.
5 0 ro
q
d
0.3 1 3 10 30
Chlorophyll a (|ig I n)
0
O
CD
T3
C
3
CN
O 0
d >
'•4—'
JD
o 0
Monostyla. lunaris
Notholca. foliacea
Notommata
0
CO -
!>s d
-i— »
• — _
o
ro ,-,
-Q CN _
0 d
Q-
0
13 o
^- d
CD
O
o
O —
0
.
.
*
^
"
*•'. I ' *• *•
i i i i i MI i i i i i i MI i i i i i
.3 1 3 10 30
Chlorophyll a (|ig I )
p _
T~
_
0) >, m
(0 0 -^ °P _
- 8 C = °
d CD .Q
T3 CD
C .Q co
3 O d
f^ ^
o < °-
d 0 0 ^ -
> ^ °
"CD "o.
_ 0 ro OH _
. o a: o °
d
0 o
0 o
0
.
•
.
*
• •
••_
i ;.
• "
I i i i 1 1 1 1 i i i i i 1 1 1 1 i i i i i
3 1 3 10 30
Chlorophyll a (|ig I )
\- j w;
- O CN —
d d
CO 0 >, 0
- 0 O -t^ CN _
d C := d
CD £)
-. T3 CD ,n
. 8 £= ^2 ^ _
d _g 0 o
< °-
-00 0 ° -
d > ^ o
"CD "o.
. 5 0 ro § _
OQ: O d
o
i— i O
— u — ; —
0
.
• ' •
• •
'
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
.3 1 3 10 30
Chlorophyll a (|ig I )
- 0
d
^ 0
- o O
d C
CD
T3
- o £=
° E
-50
d >
"CD
-00
a:
~ O
Page 15 of 19
-------
Capture Probability of Zooplankton Taxon Along Chi a Gradient
Orthocyclops.modestus Ploesoma Ploesoma.hudsoni
"-
!>\ nn
j"V CO
— d
'o
CO
_Q CD
Or-j
1 — '
CL
0 •* _
-1—*
Q.
CD CN
O o -
0
d
0
0
CO -
!>s d
-i—*
• — _
o
CD
-Q ° _
0 d
CL
0
13 o
-1— ' T— _
Q_ (— i
CD °
O
o
O —
r-i
• ."
*
^ j- — -l —
0 /
/'/^ ^S.
l'/C'~^' \v
'/ ' x\ ^S.
'/ ' • v ^v,
' /ft'*'- •••> ^
---^S'''* •.•'."?• • XN--
.3 1 3 10 30
Chlorophyll a (|ig I )
Ploesoma. lenticulare
•.
B •
•
0
• •*
, •
• • "
0 "~
CN CD ^k
o O i^ ^
d C 1= °
CD ^2
T3 CD
. 5 C ^2 ^ °
'-I—' -1—'
JD Q.
o: o o
o
~ d ""
0
p _
_
^ >, m
^ O 4? °° _
- 0 C := °
d CD .Q
T3 CD
C J3 CD
13 0 d
.Q >-
- o <"t
00 0) "* _
> 3 °
- ° "CD Q.
0 CO CN _
- o a: o °
._. o
° 0
^
_ '
X- *"
-. •• _--:' r-
~_^———~>
~~^-~^~ "
" " f "
^ *" t
•' .••".•
•*.".. •
3 1 3 10 30
Chlorophyll a (|ig I )
Polyarthra
•
*
*
•
•
•
' .
d
d ~
..0 >,
. ™ o £?
0 C 1= ^r
CD ^2 d
in "° ro
C) ^- Q
< o-
- ^~. 0 CD
0 -^ a ° ~
CD Q.
-00 CD _
o
~ d ""
0
. n P -
d ^~
in , m
- CN O 4? °° _
d C := °
CD .Q
"O CO
C o CD
" H ^ O d ~
O s—
^ °
'•4— ' -I— '
CD Q.
.80 Cl ! CN _
OQ: o °
„ o
0 o
•
a
g
•
,
.
•
1 1 1 1 1 ll| 1 1 1 1 1 1 ll| 1 1 1 1 1
.3 1 3 10 30
Chlorophyll a (|ig I )
Polyarthra. dolichoptera
•
t
•
,
• .•
.
V."
' •• r • •
• *° •
J. s«<; ;
. "l-jlir?J. ..."• . •
- 0
0
O
- ° ro
T3
C
<
0
- 0 >
'•4— '
JD
- ° o:
~ O
^-
d
^ 0
- - O
d C
CD
"^
. 0 ^
d _g
„ <
U3
- O 0
d >
"CD
. 8 0
d fV
~ o
0.3 1 3 10 _30
Chlorophyll a (|ig I )
0.3 1 3 10 _30
Chlorophyll a (|ig I )
0.3 1 3 10 _30
Chlorophyll a (|ig I )
Page 16 of 19
-------
Capture Probability of Zooplankton Taxon Along Chi a Gradient
& 00
:= ci
CD
o
O
CL
0
Q.
CD CN
O o
q
ci
Polyarthra. major
00 0
q O
d C
CD
CD "2
§ I
0
0
sl
^> 00
:= d
^2
CD
O CO
O d
0 "*
<= d
-i—*
Q.
CD CN
O d
q
d
Polyarthra. remata
CN 0
co O
d C
CD
* c
CN b:
.>> 00
= ci
!a
CD
O (D
O ci
Ql
0 ^
3 °
CD
T- 0
O >
_ "CD Q.
§ 0 CO CN
dec o °
§
Polyarthra. vulgaris
•„•>
\'
o
s.v*i.i.-
r- 0
OM O
d C
CD
T3
'•4— '
. CD
o 0
0.3 1 3 10 _30
Chlorophyll a (|ig I n)
0.3 1 3 10 _30
Chlorophyll a (|ig I n)
0.3 1 3 10 30
Chlorophyll a (|ig I n)
Polyphemus
Pompholyx
Skistodiaptomus
>,
CD
.£2
0
Ql
0
3
"o.
CD
O
q _
00
CD
d
•* _
0
CN
d
0
d
.
• •
•
. '•
o
o
o
0
o
o
0
o
C
CD
T3
3
<£
0
"CD
0
K.
>,
j^
o
CD
2
Q.
0
3
"o.
CD
O
q _
"~
00
d
CD
d
•* _
o
CN
d
0
d
.
•
" • •• * •
. ^
d
O)
- O
d
. 0
d
- O
ci
. 8
d
~ O
0
o
c
CD
T3
3
<£
0
'"CD
0
o:
>,
j^
o
CD
^2
0
Ql
0
3
"o.
CD
O
q _
"~
00
d
CD
d
•* _
o
CN
d
0
d
.
•
B
•
•^ • B
. ^
d
O)
- O
d
. 8
d
- O
ci
. 8
d
~ O
0
o
c
CD
T3
3
<£
0
"CD
0
a:
0.3 1 3 10 _30
Chlorophyll a (|ig I )
0.3 1 3 10 30
Chlorophyll a '••-•-*
0.3 1 3 10 _30
Chlorophyll a (|ig I )
Page 17 of 19
-------
Capture Probability of Zooplankton Taxon Along Chi a Gradient
Skistodiaptomus.oregonensis Synchaeta Trichocerca
"-
,>> 00
™ d
o
CD
O CD
O d
CL
0 •* _
3 °
-i—*
Q.
CD CN
O °
0
d
0
q _
^* 00
:= d
o
CD
O CD
0 d
CL
0 •* _
3 °
"o.
CD CN
O °
0
d
0
x
x
N
X
s.
V
' — X^x
'""""'"S\N-
''' XN^\ Xv-
* • x ^\. *" *" —
•* • x ^\.
• •• * * v ^^
i*. / ' .'•? • x ^x.
1 1 1 1 1 III 1 1 1 1 1 1 III 1 1 1 1 1
.3 1 3 10 30
Chlorophyll a (|ig I )
Trichocerca. birostris
'
*
•
.
t
m
* • *
« • " •
. "• • o«
•.:JL£ko£L
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 III!
.3 1 3 10 30
Chlorophyll a (|ig I )
O T-
.88 £«» -
d C 1= °
CD .Q
,0 T3 CD
. § C .£2 co _
d =3 00
_Q >—
< °-
-00 0 ^ _
d > ^ °
'-I—' -1—'
CD Q.
- S 0 .CO CN _
d Q^ CJ °
0 o
. o d -
0
P _
d "~
<*> tD >, _
- o O 4? °° _
d C r^ °
CD .Q
"O CO
8 C ^2 CD
° R 2 °
< °-
-00 0 "* _
° £ 3 °
"CD "o.
. 5 0 to CN _
do: O °
._. o
° 0
0
-~l^-~~~
'"^-^^---
f *^s^ ,-*"""
** " f.^^ ~ "
- ""**"* .^^ " '
*" ^r * '
^f *
s *' "
t •
/ . •
/ •
/ 0*
• • % • ^
. • *. " •"•
• •• :;.:- •• •
% ""'•" -^ J>* ':• ' *
-• i*yrlBJHJ|^.*-dSjF* ^ % • " '
i i i 1 1 ii| i i i i 1 1 ii| i i i i i
3 1 3 10 30
Chlorophyll a (|ig I )
Trichocerca. cylindrica
•
B
•
*
• * •
°
*• » o *
" -• •* * •
ik£'/: •
i i 1 1 1 1 1 1 i i i 1 1 1 1 1 1 i i i 1 1
3 1 3 10 30
Chlorophyll a (|ig I )
d "~
- CN 8 -^ °° -
d C 1= °
CD .Q
T3 CD
CN C O CO
d 3 O d
O !—
< °-
.20 0 ^ _
° -^ 3 °
"CD "o.
. o "0 tD CN _
d Q^ O °
0 o
. o d -
0
on
~ q _
d ^~
^•0 >,
^— O -*— ' • —
d C •= °
CD ^2
T3 CD
^1 C ^2 CD
' o 13 0 d
< CL
-00 0 "* _
d .> 3 °
- "to ^>
. S "0 tD CN _
OQ: O °
._. o
0 o
0
*
•
>
"•
• • .
"
•
• .»« "
i i i 1 1 ill i i i i i i ii| i i i i i
.3 1 3 10 30
Chlorophyll a (|ig I )
Trichocerca. multicrinis
•
.
••
* •
• • ". -€ °
'.•-':• J^iiS:", .* -.
, '
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
.3 1 3 10 30
Chlorophyll a (|ig I )
d
^ 0
- o O
d C
CD
T3
^~ p-
o ^
'-I—'
CD
-00
o:
~ O
d
r^ 0
- o O
d C
CD
"^
- o £=
d 5
<£
- O 0
d .>
"CD
. 8 0
o W
~ o
Page 18 of 19
-------
Capture Probability of Zooplankton Taxon Along Chi a Gradient
Thchocerca.porcellus
Trichocerca. rousseleti
Trichocerca. si mi Us
& 00
= d
^2
CD
o to
O d
Ql
0 ^
3
-I—*
Q.
CD CN
O o
q
d
0
o
c
CD
T3
C
.Q
<
0
"-^
CO
(V
LL_
>%
-i—*
I —
!a
CD
.Q
2
Q_
2
3
-1—*
Q_
O _
T~
00
d
CO
d
^r
d
d ~
0
d
m
"
"
. x
• • • • ^
• * - - ' "
^ ^~~~~~-~-^
-^*>;y. - ; ^-7^---^
*" o"* ""--._,
. a
d
- O
d
o
d
- 0
d
- o
~ o
0
O
c
CD
T3
C
^2
<
0
'•^
CD
iV
LL_
>,
-I—*
• —
^2
CD
.a
o
ol
0
^
-1—*
Q_
O _
T~
00
d
(D
d
^r
d
d ~
0
d
.
•
. ••
'. '''. .
. s
d
OM
- O
d
CM
- O
d
- 0
d
- P
~ o
0
O
C
CD
T3
C
^2
<
0
'•^
CD
"0
0.3 1 3 10 _30
Chlorophyll a (|ig I )
0.3 1 3 10 _30
Chlorophyll a (|ig I )
0.3 1 3 10 _30
Chlorophyll a (|ig I )
Tropocyclops.prasinus.mexicanus
q _
•^
^ 00
:= °
o
CD
O CD
0 d
Ql
0 •* _
^ °
-i—*
Q.
CD CN
O d -
0
d
* J ».
• '•
•
.
•• . .*.-
'.' t£wfa
'
* • •
r-w^.
.» a .
T—
d
00 0
- o O
d C
CD
CD ~°
' § 1
- O 0
d >
"CD
o "fl5
" o a;
~ O
0.3 1 3 10 _30
Chlorophyll a (|ig I )
Page 19 of 19
-------
Appendix 7 - Wl Secchi GAM Models Zooplankton
-------
& 00
:= ci
CD
O
O
CL
0
Q.
CD CN
O o
q
ci
Capture Probability of Macroinvertebrate Taxon Along Secchi Depth Gradient
Asplanchnidae Bosminidae Brachionidae
2 4 6 8 10
Secchi Depth (m)
(D
O
0
O
- CN 0
& 00
:= ci
CD .Q
T3 CD
C o co
13 O ci
-° £-
< °-
0 J
13 °
-i—»
Q.
CD CN
O ci
q
ci
2 4 6 8 10
Secchi Depth (m)
CO g >^CO
0 C 1= °
CD !Q
C^ C ^ CD
ci ^ O ci
< £
- 0 £ ^
° .> 3 °
-I—' -I—'
CD Q.
£ "0 ro CN
do: o o
q
ci
• •• v •.
t . •" f \ • •
2 4 6 8 10
Secchi Depth (m)
n 0
to O
d C
CD
d >
a:
Calanoida
Chydoridae
Collothecidae
>,
:=
o
CD
.a
0
CL
0
3
Q.
CD
O
p _
T~
00
ci
CD
ci
•*
°
CN _
0
0
,
.
1
aJbatt&M ••
. CD
0
ID
0
. 8
d
- CN
ci
CO
ci
- o
0
o
CD
T3
/-\
_LJ
0
>
'•4— '
CD
0
a:
>,
:^
o
CD
^2
0
CL
2
3
-1—*
a.
CD
O
O _
T~
00
ci
CD
ci
^r
ci
CN _
0
0
.
B
.
• •
• -i in* •'•!*«&'&*••*''« ' '•
. &
d
CD
- O
ci
. §
ci
CO
- O
ci
. 8
0
o
0
o
CD
T3
^
...j-
0
'•4^
CD
0
a:
•4- »
O
CD
^2
0
CL
0
^
-1—*
a.
CD
O
O
T~
00
ci
CD
ci
^r
ci
CN
0
0
2 4 6 8 10
Secchi Depth (m)
2 4 6 8 10
Secchi Depth (m)
Page 1 of 20
.
•
._ t'.
''f^r*'' "\tk\
•
•
'
• , ' .
"*•*•*
M, '<>•••- .
ci
rs, ®
. ™. O
O (~~
CD
sl
" d<
- o $
.§1
o a:
~ O
2 4 6 8 10
Secchi Depth (m)
-------
Capture Probability of Macroinvertebrate Taxon Along Secchi Depth Gradient
Conochilidae
Copepod.juveniles
Copepod.nauplii
o
"-
^ 00
:= d
o
CD
O CD
O d
CL
CD •* _
Zi °
-i—*
Q.
CD CN
O o -
0
d
-
—
.
:
- **•"" "
m
•••'•.
" • * *• *
^•v^\
«**
_._.--•*"""""
1
K """*""*--
•
' \
0
.
;'*;••" •• '••
^.,B»/ .•""
' .* -O-"" •
- 00
0 o
CO —
- co 0 -£* °
d C 1=
CD .a
o> "° ro o
_ -^- C _Q (M _
d E 2 °
CN „, ~
- CO 0 ^
0 -^ °° -
00 C 1=0
° CD !Q
T3 CD
C o co
E 2 ° "
CO <£ Q.
o 0 CD •* _
> 3 °
-f-* -!-•
CD Q.
. 8 £ o o -
d
- 0 ^ -
• ••
"
•
0
•'
• * • o"
*&••* •*?!** X
,
;
• • ""
' .•'•*t.
."• * . . .
f^-l. . '
•o
. CO
d
co CD
- in O
d C
CD
T3
•* c
d Zi
- CN 0
d >
'•4— '
CD
.20
~ O
2 4 6 8 10
Secchi Depth (m)
2 4 6 8 10
Secchi Depth (m)
2 4 6 8 10
Secchi Depth (m)
Copepodites
Cyclopidae
Cyclopoida
p _
^~
^ 00
:= °
CD
O CO
2 °
CL
0 •* _
3 °
Q.
CD CN
O d -
0
0
.
•
a •
, t t
'' .
•?- P _
0 ^~
d C •= °
CD ^2
T3 CD
•^ C £) co
- o _g oo-
zi °
_ CD Q.
- 8 0 tO ex, _
OQ: o °
„ o
0 o
.
•
•
•
•* • •
' .." '•. \ -°V
' ••»SSwi^^S*^'t^*r' *'•'•
- K P _
0 "~
.38 ^» _
d C :^ °
CD ^2
. 8 c S zi °
CD a.
^ "0 CD CN
" d £ O d
._. o
° 0
.
. .
.
i*L&&a£: '•
1
0
1 1
0.34 0.45
bundance
^
CO
- CN 0
d >
CD
. ^ Q)
2 4 6 8 10
Secchi Depth (m)
2468
Secchi Depth (m)
10
2 4 6 8 10
Secchi Depth (m)
Page 2 of 20
-------
8 J
-i—« O)
•= ci
&
CD 0
O 2
Q.
0 8
-I—*
Si
in
ci
Capture Probability of Macroinvertebrate Taxon Along Secchi Depth Gradient
Daphnlldae Dlaptomldae Euchlanldae
2 4 6 8 10
Secchi Depth (m)
O)
ci
C = o
CD &
C O CD _
T- 0
o >
§1
o a:
o
2 4 6 8 10
Secchi Depth (m)
|n 0 >, n
•* O ±i o
o C •=
CD .Q
2 u p CN
O i— Q
CM /i»
CN =r
JD
^ 0
Q.
O
1§
CD
T3
C
0
- o CD
0 0
o:
2 4 6 8 10
Secchi Depth (m)
Gastropodidae
Holopedlldae
>,
-1— 1
o
CD
.a
0
CL
0
"Q.
CD
O
O _
T~
00
ci
CD
ci
o -
CN _
0
0
.
•
• ,
* \. ' .. ...v
. '. ^t^v£&l&$£*i™tf.. ' : '
- CD
0
- in
ci
. s
ci
in
- OM
ci
CO
ci
- o
0
o
CD
T3
3
<<
0
"CD
0
a:
-i— *
o
CD
s
Q.
§
"Q.
CD
O
O _
T~
00
ci
CD
ci
0
OM
ci
0
0
.
t
t •
• *
I
• . •
'•'."> .:••-.
•L'k* ji&*8a:£» V.
. ^
ci
O)
- O
ci
. fe
ci
- O
ci
. 8
0
o
0
o
CD
T3
3
<<
0
'"CD
0
a:
>,
:=
o
CD
0
Ql
0
"o.
CD
o
p
T~
00
ci
CD
ci
ci
OM
ci
0
ci
2 4 6 8 10
Secchi Depth (m)
2468
Secchi Depth (m)
Page 3 of 20
10
Lecanidae
m •
.
* B
A ' .' J"*S '"
• *.Z ."'Jjfjt"
.
•
.
• •
* "
.. • .
"{"fiii* ' "'•"
- 8
ci
r^ 0
. o o
o C
CD
-§§
o 3
CO
- o 0
ci >
.si
o a:
~ o
2 4 6 8 10
Secchi Depth (m)
-------
Capture Probability of Macroinvertebrate Taxon Along Secchi Depth Gradient
Lepadellldae
in
d
CD
O CO
O d
CL
0 ™
Q.
CD T-
O o
p
d
- o
d
2 4 6 8 10
Secchi Depth (m)
Nauplll
T- CD •>,
00 & °°
d C — o
CD ^2
T3 CD
0 C ^2 <°
0 13 00
.^ Z!
-I—* -I—'
JO Q.
o: o
2 4 6 8 10
Secchi Depth (m)
S
d
CM CD
in o
d C
CD
°> i-
N 0
„ ^0
Notommatidae
:= o
CD
O CO
O d
Q.
O
p
d
2 4 6 8 10
Secchi Depth (m)
p
d
0
O
CD
T3
-00
d >
'•4—'
JO
o CD
o:
Polyphemidae
Rotifera
-1— »
o
CD
^2
0
Ql
0
^
Q.
CD
O
p _
00
d
(D
d
d
CM
d
0
d
.
• •
a
•
.
• 0
•
• •_
.
- o
0
- o 0
CD
T3
" ° 13
.a
-00
~
JD
- o CD
a:
- 0
-1— »
o
CD
0
Ql
2
3
"o.
CD
O
(D
d
in
d
^~
d ~
CO _
d
CM
d
d
0
d
•
•
*
B
•
. '.
•
" * •'. .•.».'..'•"'
. s
d
- 0
d
"
- 0
d
- 0
0
O
CD
T3
C
^
^2
0
~
CD
0
a:
t
o
CD
0
Ql
0
^
Q.
CD
O
p
00
d
(D
d
d
CM
d
0
d
2 4 6 8 10
Secchi Depth (m)
2 4 6 8 10
Secchi Depth (m)
Sididae
.
^
* t
• .% *
• .
j+" • .
&&*':&•<;. ' "• .
. X
d
•3g
CD
CO "g
o 3
u, <
- - 0
o >
.si
o a:
- 0
2 4 6 8 10
Secchi Depth (m)
Page 4 of 20
-------
Capture Probability of Macroinvertebrate Taxon Along Secchi Depth Gradient
Synchaetidae
_g> 00
:= ci
CD
.a
o
CL
0
-I—*
a.
CD
O
p
ci
2 4 6 8 10
Secchi Depth (m)
C\l
CD
CD
T3
C\l 0
° .>
'•4—'
_. CD
Temoridae
_g> 00
:= ci
CD
O d
0 •*
-I—*
Q.
2 4 6 8 10
Secchi Depth (m)
P
ci
0
O
CD
T3
-0
oq
ci
>,
±± to
1 °
CD
.a
o ^
CL ci
0
••£= ^
JD Q.
0
a:
o
Testudinellidae
O) Q
° ro
T3
C
o _Q
ci <
0
L 8 1
ci 0
a:
2 4 6 8 10
Secchi Depth (m)
Trichocercidae
Trochosphaerldae
Unknown
p _
T~
& °p
:= ci
o
CD
O CD
0 ci
CL
0 •* _
^ °
"o.
CD OM
O o -
0
d
.
•
•
*
• •
• a "
"" • 5°
.*>; ;• t..^..'. 'v
^'\'^^}/^J^^-i::-
-?- P _
ci "~
.58 ^» _
d C = °
CD .Q
T3 CD
^1 C .Q ^ °
.— _^
CD Q.
.30 ro CN _
da: o o
._. o
° 0
.
.
.
•
^
1 '•• • ' '.
'f-m'A !Ji''''
i
. W P -
d ^~
. ^ 8 -^ °° -
d C = °
CD .Q
T3 CD
CM CI O CO
"°l ld"
-20 0 2 _
0 .> D
"CD "o.
- fe 0 /O CH _
OQ: o °
._. o
0 o
.
•.
.": -•
- OM
d
. ™ 8
d C
CD
_,. T3
J £=
" ° E
O)
- O 0
d >
"CD
. § 0
o a:
~ O
2 4 6 8 10
Secchi Depth (m)
2 4 6 8
Secchi Depth (m)
10
2 4 6 8 10
Secchi Depth (m)
Page 5 of 20
-------
Capture Probability of Macroinvertebrate Taxon Along Secchi Depth Gradient
Acanthocyclops
Acanthocyclops. vernal is
Alona
0
ro -
"?>\ o
-i—*
• — _
'o
ro o
-Q ° _
O d
Q_
0
13 0
"o. d ~
CD
o
o
0 -
d
o
• a
. • *
1 1 1 1 1
10
CD ">*
ro <-> •&
- o C 1= ^
d ro & d ~
"O CD
f~ O
13 O ro
-Q *- o ~
o *t ^"
d 0 Q)
•J; a 2 ~
CD Q.
"?i5 ro
- o Q£ O d -
d
o_
f
\
\
\
\ *
\
\
\
w \
^^ \
\^ \
X. XN"
\. Xx^
"*""'"""** ^Sssx? ** * -
"* "* ^^^^^ **"""*"•-
'.'. ' / vt>^^~
.. <• ' •*'•'."'"'•--
i i i i i
p _
OM T~
- O
° Q) >* m
O •& m. -
•*— f~ • ^
• P CO ^
0 "D CO
f~ O CD
T— ^ O O
- q -Q ,!r
o < Q-
0 0 ^
- g .> ^ °
"CD "o.
"0 ro CN
(r ^— '
o_
B
•
•
1 1 1 1 1
d
^ 0
- o O
d C
CD
^ T3
- o ^
. ^
^2
-00
d >
"CD
"?T5
a:
- 0
2468 10 2468 10 2468 10
Secchi Depth (m) Secchi Depth (m) Secchi Depth (m)
Ascomorpha. ecaudis
Ascomorpha. ovalis
Asplanchna
p _
^~
.& °P _
:= o
o
CD
O (D
0 6
Q-
0 •* _
3 °
"o.
CD OM
O o
0
d
^
•
•"
. •
•
•
•B "
* ••••"*
. V " <.'! • . "
. 0 P _
d ^~
.38 ^oo_
d C :^ o
CD £)
. 8 "c 5 ^ °
"CD "o.
. 5 0 ro » -
d C :^ o
CD ^2
sl ^<°
" o = 00-
^ °
"CD "o.
.50 ro CN _
do: O o
._. o
0 o
t
•
o
• 1
•
j
* \ ' , ,'
. " . * • " *•
•j^rt'lsif&'S.Ri.i'rV
i
(D
d
oo CD
- ^r O
d C
CD
- % C
d _g
<-j-<
- CN 0
o .>
"CD
. ^ 0
o Q:
~ O
2 4 6 8 10
Secchi Depth (m)
2 4 6 8 10
Secchi Depth (m)
2 4 6 8 10
Secchi Depth (m)
Page 6 of 20
-------
Capture Probability of Macroinvertebrate Taxon Along Secchi Depth Gradient
Bosmids Bosmina Bosmina.longirosths
0
CO -
-1— »
0 d ~
Q_
0
13 0
S5~
o
o
p -
l—i
•
*
t
.
^
...
p _
d CD .Q
T3 CD
Z! O o i ~
i^ <; Q-
d 0 0) 2 -
.^ 3
"CD "o.
^0 CD c\i _
d
o
.
•
. .
2 4 6 8 10
Secchi Depth (m)
2 4 6 8 10
Secchi Depth (m)
£
oo
d
CD
T3
CD
5 I
§ o:
2 4 6 8 10
Secchi Depth (m)
oo CD
CM O
d C
CD
*- 0)
o >
'•4—'
K_ CD
o 0
Calanoid
Ceriodaphnia
Chydorus
O _
^~
^ 00
:= °
CD
o to
0 d
Q-
0 •* _
3 °
Q.
CD OM
O d -
0
0
,
.
•
-CD P _
0 ^~
- s § ts-
CD !Q
-H |s-
< °-
- CM 0 0) i; _
d .> 3 °
CD Q.
. $2 "0 CD ,
-00 ^ °° -
d C :^ °
CD ^2
T3 CD
- o C ^ ^ °
CD Q.
- o 0) /O ^ -
o: o °
0 o
.
•
'•-. w- *'•'. •.
1 ,
. s
0
1 1
0.05 0.06
bundance
CO
- O 0
d >
-l|
2 4 6 8 10
Secchi Depth (m)
2 4 6 8
Secchi Depth (m)
10
2 4 6 8 10
Secchi Depth (m)
Page 7 of 20
-------
Capture Probability of Macroinvertebrate Taxon Along Secchi Depth Gradient
Collotheca
& 00
:= ci
CD
.a
o
CL
0
-i—»
Q.
CD
O
q
ci
2 4 6 8 10
Secchi Depth (m)
in
CN
0
o
CD
T3
c
.a
0
"-^
JD
0
a:
Collotheca. mutabilis
^ o
CD
.a •*
O o
Q. „
0 o
3 M
Q. o
CD
o -
o
q
ci
2 4 6
Secchi Depth
i
8
(m)
10
0
in O
q C
o CD
T3
C
.a
O
0
JD
. 5 a:
o
o
Colonial, conochilus
CD
O CO _
O d
CL
0 •*
3 °
-i—*
a.
CD CN
O o
q
ci
2 4 6 8 10
Secchi Depth (m)
q
ci
0
O
CD
T3
- o
0
J5
h o 0
a:
Conochiloides
Conochiloides. dossuarius
Conochilus
-&• , m
- co O 4? °° _
d C r^ °
CD ^2
T3 CD
C JS CD
- s E 2 ° "
d^Q.
0 0) ^ -
> 3 °
- ^ "CD "o.
°' "0 CD CN
a: o o
„ o
° 0
.
0
„
° t
. T- P _
d "~
00 0 >,
^— O -*— ' • —
d C •= °
CD ^2
T3 CD
•<- C JS CD
-03 00-
co < CL
-80 0 ^ _
° .> 3 °
"CD "o.
. 8 "0 ro CN _
0 o
° 0
• aa
.
•
^
m
* *
a
*
• '
•• ••
• s s::c-°.:^
*•..." •V*f°-1
\
\: \ '••
'fs • '*"
I '&•:"*. • .
- 00
d
in 0
- CD O
d C
CD
"^
§• C
" ""I
- CO 0
d >
"CD
.20
o a:
~ O
2 4 6 8 10
Secchi Depth (m)
2 4 6 8 10
Secchi Depth (m)
2 4 6 8 10
Secchi Depth (m)
Page 8 of 20
-------
Capture Probability of Macroinvertebrate Taxon Along Secchi Depth Gradient
Conochilus.unicornis Cyclops.varicans.rubellus Daphnia
o
•<-
_^> 00
:= d
o
CD
o to
O d
Q.
0) •* _
13 °
"o.
CD CN
O o i -
0
d
'
,a
• • •
~ in
d °
.Jo £* •* _
d C = °
CD .Q
T3 CD
^1 c .Q oo
o ^ Od
o s—
^ °
- "to ~o-
- o 0 /O - _
OQ: O o
. o § -
^
.'"
- 0 P _
d "~
.88 ^» _
d C = °
CD .Q
in "° ro
- § i | § -
^ °
" ! 1-c,
- O 0 IVJ CM _
OQ: o o
. o § -
'
• t ,
.
•
•
•
• .Y_. ^.'s/.Liii*: •.'•<*. .
_ 00
d
in 0)
- OM O
d C
CD
"^
__ C.
' ° E
<£
- ^ 0
0 >
-l|
- 0
2 4 6 8 10
Secchi Depth (m)
Daphnla.dubla
2 4 6 8 10
Secchi Depth (m)
Daphnia.longiremis
2 4 6 8 10
Secchi Depth (m)
Daphnia. mendotae
p _
"~
.>> 00
:= d
o
CD
o to
0 d
CL
0 •* _
3 °
"o.
CD OM
O o i -
0
d
.
.
*
• •• •
•
'
.
" * • * •
• s. . . '• •• > •
* . : .- .
*",*«• • •
. 0 P _
d ^~
.88 ^» _
d C :^ o
CD .Q
8 C ^2 to
" o = 00-
< °-
-50 3 °
"CD "o.
- 0 0 tO » -
d C :^ o
CD ^2
S? ^ ^ °
"CD "o.
.00 tO
"CD
.50
o Q:
— o
2 4 6 8 10
Secchi Depth (m)
2 4 6 8 10
Secchi Depth (m)
2 4 6 8 10
Secchi Depth (m)
Page 9 of 20
-------
Capture Probability of Macroinvertebrate Taxon Along Secchi Depth Gradient
Daphnia.parvula
Daphnla.pulex
Daphnla.pullcarla
-^ ^~ - -
p _
^~
T- 0 "a..
-00 -^ °° -
d C 1= °
CD .a
T3 CD
C .£2 co _
Z! O o
° < CL
0 0 ^ -
.> D °
- o CD Q.
"0 CD CN _
a: o °
„ o
0 d
0
•
"
•
'" ' •
o •
. 0 P _
0 "~
. o 8 ^» _
d C 1= °
CD .a
- q 1 -g g -
< ^
.80 22~
CD a.
-50 ro ^ -
0 o
0 d
.
•
'
0
1 1
0.12 0.15
>undance
-00
d >
_ CD
- 0 0
o o:
2 4 6 8 10
Secchi Depth (m)
2 4 6 8 10
Secchi Depth (m)
2 4 6 8 10
Secchi Depth (m)
Daphnia. retrocurva
& 00
:= o
CD
.a
o
Ql
0
-i—*
a.
CD
O
q
ci
.
. *
'
'
- 8
d
i i
0.04 0.05
bundance
- O 0
d >
'•4— '
CD
-00
o o:
~ o
2 4 6 8 10
Secchi Depth (m)
Daphnia. rosea
Diacyclops. thomasi
-1—*
o
CD
.a
2
Q-
§
"o.
CD
O
O _
T~
00
d
CD
d
d
CM
d
0
d
.
•
•
•
•
• ' * : ..
V N
d
,_
d
- 0
d
- O
d
- 0
d
~ O
0
o
CD
T3
3
<£•
0
'"CD
0
a:
>,
:^
o
CD
.a
0
Ql
0
"Q.
CD
O
O _
T~
00
d
CD
d
d
CM
d
0
d
.
B
.
.
• .
•\
*.'. •>/.;•:. •
fA* " * ' • fif ftft^Tl «^» *•
- K
d
^
d
. 8
d
CM
- CM
d
d
~ O
0
o
CD
T3
o
0
"CD
0
a:
2 4 6 8
Secchi Depth (m)
10
2 4 6 8 10
Secchi Depth (m)
Page 10 of 20
-------
Capture Probability of Macroinvertebrate Taxon Along Secchi Depth Gradient
Diaphanosoma Diaphanosoma.birgei Diaptomid
_g> 00
:= d
CD
.a
o
CL
0
-I—*
Q.
CO
o
q
d
2 4 6 8 10
Secchi Depth (m)
8
d
co 8 ^oo
Of- 1=0
CD
*-
§1
K c jo CD
0 E £ °
CL
CO
2 4 6 8 10
Secchi Depth (m)
§
CD ?Z
q §
o 3
-o0
sl
ci a:
C = o
CD &
O CD
O ci
a)
2 4 6 8 10
Secchi Depth (m)
r^ 0
o O
d C
CO
o 0
d >
si
o a:
Dlaptomus
Eplschura.lacustrls
Eucyclops.agilis
0
CO _
~>\ °
-1—*
• — _
"CD
o S
0 d ~
0.
0
3 0
3- o
CO
O
O
O —
r~i
.
•
•
•
B •
• •
"
•
..
P _
^~
_
tl> >, m
^ O 4? °° _
- ^r C := 0
d CD .Q
T3 CO
C .Q CD
" E £°~
IT) ^f" M
- CN ^^
00 0 •* _
> 3 °
^ "o.
d
._. o
0 o
.
•
•
*
••
.
.'•'*-"ji* '
' '•{• • •'•?
^ t — '
- O CO -
ci d
T- 0 ">>
- o O -&1
d C =
CD £) a
T3 CO 0| -
. o C J3 o
° E 2
^ 2 _
"CD Q. °
- o a> to
a: O -
O
- o o —
r~i
.
•
•
.
.
.
"•
- o
- o tD
O
C
CO
- 0 T3
C
^
" °
-------
Capture Probability of Macroinvertebrate Taxon Along Secchi Depth Gradient
Eucyclops. elegans
Filinia
Filinia.longispina
o
CO -
~>\ °
-1—*
I — _
o
J2 °
O d
CL
0
13 0
-*- ' T— _
CD
o
o
O —
t— i
m
*
i . • '
p _
CO *~
- o
°' tD >, „
O •& m- -
- (~ • — O
CO -Q
. s c 5 «_
d ^ O o
< G-
0 0) •* _
^ > 3 °
- o ••— ' -*— '
° "0 CD CM
,-, fV O ° ~
- o LL_ ^-^
- 0 ^ -
o
•
.
"-^^
"**••» - - —
• — —
.. * . .
-CO P —
d "~
r- , m
- CM 0 ^ °° -
d C 1= °
CD .a
T3 CD
CM C O tO
d 3 O d
.Q <-
< °-
- ^ 0 0) 2 -
.—- . ^
^ ^ "o.
• -.„, f\ Q ""
o
o d ~
m
-
0>
• • *
. S
d
r^ 0
- o O
d C
CD
0 "?
d ^
o
<
-00
d >
CM ^
O ' (V
LL_
~ O
2 4 6 8 10
Secchi Depth (m)
2 4 6 8 10
Secchi Depth (m)
2 4 6 8 10
Secchi Depth (m)
Filinia.terminalis
Gastropus
Gastropus. hyptopus
^* CD
:= °'
o
CD
^2
2 •*
Ql d
0
3
"Q. ^
CD °
O
0
d
.
•
•'
»*:••'••
••» • •*V*">*J'J,<-' ....
P _
^
, m
- T O 4? °° _
0 C := °
CD !Q
T3 CD
C ^2 tp _
CD D O O
- O o i_
d <£ Q.
0 0) •* _
.> D °
CO ••— ' -i— '
- o CD a.
°' "0 CD CM
a: o o
„ o
° 0
.
•
^
.
' • V •
• .
.
• ..
" . •:
. 0 P _
d "~
to 0) >, m
- o O 4? °° _
d C := °
CD .a
"O CO
. o £= ^2 ^ °
'•4-* 4->
CD a.
.80 tD
'•4— '
CD
. 8 0
o a:
~ O
2 4 6 8 10
Secchi Depth (m)
2 4 6 8 10
Secchi Depth (m)
2 4 6 8 10
Secchi Depth (m)
Page 12 of 20
-------
Capture Probability of Macroinvertebrate Taxon Along Secchi Depth Gradient
Gastropus.stylifer
Holopedlum
Holopedium. gibberum
o
•<-
,>> 00
: — o
.Q
CD
J3 <0
O o
Q.
d °
to , ^
- CM O ••-• • -
d C 1— °
CD .Q
05 C .Q co
° E 2 °
3 °
CD Q.
o fV C3 ^
o-
.
.
•
•
"
" * * "IB-
T- °
0 "~
.88 ^» _
d C 1— °
CD .Q
fe C S to
° E 2 °
CD
-00
d fV
- 0
2 4 6 8 10
Secchi Depth (m)
2 4 6 8 10
Secchi Depth (m)
2 4 6 8 10
Secchi Depth (m)
Kellicottia. bostoniensis
Kellicottia. longispina
Keratella. cochlearis
p _
_^> 00
:= °
CD
O CD
0 d
Q-
0 •* _
Q.
CD CM
O o ~
0
0
•
•
•..•:'.
d T~
d C •= °
CD £)
-O CD
. co C .£ 3 °
CD Q.
. ^ "0 CD » -
d C :^ o
CD ^2
.sl |«.
Q J O O
<£ ^-
- " 0 CD ^ _
d > ^ °
CD Q.
^ "0 CD CM
" d £ O o
0 o
° 0
•
•
• " " • * 0
;.' :\'-:'' ' '-.
• *".^ o *•••
.^ ,v
• jf'. "i'^v "
1 1
o
1 1
0.42 0.56
bundance
^
00
- CM 0
o .>
. J "i
2 4 6 8 10
Secchi Depth (m)
2 4 6 8 10
Secchi Depth (m)
2 4 6 8 10
Secchi Depth (m)
Page 13 of 20
-------
Capture Probability of Macroinvertebrate Taxon Along Secchi Depth Gradient
Keratella.crassa
Keratella. earlinae
Keratella.hiemalis
& 00
:= d
CD
o to
O d
CL
0 •*
13 °
-i—*
Q.
CD CN
O o
0
0
^r
d
d C
CD
_ T3
5 .1
L 8 I
2 4 6 8 10
Secchi Depth (m)
:= o
CD
O d
0 •*
-I—*
Q.
o
2 4 6 8 10
Secchi Depth (m)
£
0
o
CD
kSl
o o:
:= o
!O
.Q (D
O d
0
"o.
o
u s
•<*• 0
o O
d C
CD
« C
o t
-00
d >
'•4—'
JD
1-50
o o:
2 4 6 8 10
Secchi Depth (m)
Keratella. quadrata
Keratella. taurocephala
Keratella. testudo
_g> 00
:= °
CD
o cp
O d
CL
0 •*
Q.
CD CN
O o
q
d
.
*
••
'»• .' -,
' '. . '>%-..:. •' .
•/•. . '•'*': '^:\ •
. &
d
^r
- O
d
- P
- O
- 0
d
- 0
0
O
CD
|
0
"CD
0
a:
2 4 6 8 10
Secchi Depth (m)
p _
•<~
•= o ~
!5
CD
o CD
O o
tl
Q-
0 •* _
3 °
"o.
CD CN _
0
d
f
a
• •
. ' .
\ "
" •! . '
' tl- ' '.
'•« - •_•: • • . .
• % • " •» iff »• "*"?
^
d
d
00
d
. ^
o
d
— o
0
O
CD
T3
C
^
"CD
0
a:
^
o
CD
.a
0
Q-
0
^
"o.
CD
O
p _
•<~
d
(D
d
i; _
°
CN _
0
d
f
„
f
t * .
• • .«.^. •.-«.. •
i
d
00
- OM
d
^
d
_ ^
d
- P
~ O
0
o
CD
T3
^
"CD
0
a:
2 4 6 8 10
Secchi Depth (m)
2 4 6 8 10
Secchi Depth (m)
Page 14 of 20
-------
Capture Probability of Macroinvertebrate Taxon Along Secchi Depth Gradient
Lecane
Lecane.inermis
Lecane.tenuiseta
& 00
:= ci
CD
o
O
CL
0
Q.
CD CN
O o
q
o
u s
2 4 6 8 10
Secchi Depth (m)
ro 0
o O
d C
CD
™ c
q §
o ^>
OM
O 0
d >
J5
o 0
o o:
_g> 00
:= d
CD
o to _
O d
Q_
0 •*
CD CN
O o
q
d
U 8
o
2 4 6 8 10
Secchi Depth (m)
in 0
o O
d C
CD
« C
o t
OM
O 0
d >
J5
o 0
o o:
IT)
O
>> ^r
-t-t ^.
:= o
^2
CD
O CO
O o
Ql
0 <^!
3 °
-I—*
Q.
CD T-
O o
q
ci
2 4 6 8 10
Secchi Depth (m)
ro
o
o
q
0
J2
O 0
o >
'•4— '
JD
o 0
Leptodlaptomus. ml nut us
Leptodlaptomus. slcllls
Mesocyclops
p _
T
^* 00
:= d
o
CD
o to
0 d
CL
0 •* _
3 °
"o.
CD OM
O o -
0
d
.
•
•
0
• *
• ••
* •
.:' ::•'••• ' *' •'•''
1 « 0 "" * • f " •
• 1 . • **" 0J^-o "V, •
tt *i • •ffilhi*iiiti"i"'^i ff "*
i i
T- ^ _
d °
^•0 >,
- ^— O -*— ' • —
d C •= °
CD .Q
T3 CD
•^ C jS oo
" ° E 2 ° "
-00 0 ^ _
d .> ^ °
"CD "o.
.80 ^ g _
0 o
° 0
•
•
•
0
.
• .
. 0 P _
d ^~
*- <1> >^ m
- o O 4? °° _
d C r^ °
CD ^2
T3 CD
. 0 C ^2 ^ °
"CD "o.
- 0 0 /? ^ -
o: o °
._. o
° 0
.
*•
•
. 8
d
1^ 0
- o O
d C
CD
"^
. 8 £=
d _g
<^
- o 0
d >
"CD
. g 0
~ O
2 4 6 8 10
Secchi Depth (m)
2 4 6 8
Secchi Depth (m)
10
2 4 6 8 10
Secchi Depth (m)
Page 15 of 20
-------
Capture Probability of Macroinvertebrate Taxon Along Secchi Depth Gradient
Mesocyclops. edax
& 00
:= ci
CD
.a
o
CL
0
-i—*
a.
CD
O
q
o
8
q o
d C
CD
CN "2
q =r
d 3
-00
'•^
JD
o 0
o a:
2 4 6 8 10
Secchi Depth (m)
Monostyla
& 00
:= d
CD
O CD
O d
Q.
0 •*
Q.
CD CN
O d
q
d
u s
2 4 6 8 10
Secchi Depth (m)
m 0
o O
d C
CD
« C
o t
CN
O 0
d >
J5
o 0
o a:
Monostyla.lunaris
-^ °9
= d
CD
O CO _
O d
CL
0 •*
13 °
-i—*
Q.
CD CN
O d
q
d
U §
2 4 6 8 10
Secchi Depth (m)
r- 0
o O
d C
CD
•« c
q §
0
O 0
d >
Notholca.foliacea
Notommata
-^
"CD
0
a:
>,
'•^
o
CD
o
2
CL
0
^
"o.
CD
u;
d
o
CN _
d
ir-
d
o
d
q -
o
O —
(— i
.
•
• *
. 0
d
- O
d
- O
- O
d
- o
— o
0
O
CD
T3
(~
£
<
0
>
"CD
0
a:
>,
'-^
O
CO
o
0
CL
0
3
"o.
q
^
oo
o
CD
d
^r
0
2
0
d
2 4 6 8 10
Secchi Depth (m)
2 4 6 8
Secchi Depth (m)
10
Orthocyclops. modestus
.- -
• '
•
'
\; .
d
^ 0
_ o O
d C
CO
•D
" °- §
- o Q)
0 >
CO
- o 0)
o:
2 4 6 8 10
Secchi Depth (m)
Page 16 of 20
-------
Capture Probability of Macroinvertebrate Taxon Along Secchi Depth Gradient
Ploesoma
Ploesoma. hudsoni
Ploesoma.lenticulare
o
•<-
_^> 00
:= d
.Q
CD
J3 CO
O o
Q_
0 •* _
13 °
Q.
CD CN
O d -
0
0
X
X
y X
^ - ~ ' ' ' ^^^
— !
--""".
• . '. "»
• •• . "
.-:<"»•. .A*.- *..•#.».:• ;'
i i
- CN ^2 -
0 £= 1=
CD .Q ,1
•n "O CD rS -
_ !£ C ^2 °
d ^ O
< CL d -
- ^ 0 CD
0 > ^ CN
"CD "o. °
. 8 0 co
do: 05-
0
d
•
°
0
•
- 0 °-
- o 0) >, ro
^ = s -
CD ^2
- o T3 CD
C ^2 CD
3 O o
• ° < Q-
0 0) £ -
0 > ^ °
CD Q.
"0 CD CN
- o £ O d
o
o
*
• t
"
• .
" " • *
_ o
d
^ 0
- o O
d C
CO
•D
- o ^
<
-00
>
CD
- o CD
o:
2 4 6 8 10
Secchi Depth (m)
2 4 6 8 10
Secchi Depth (m)
2 4 6 8 10
Secchi Depth (m)
Polyarthra
^> 00
:= d
CD
^2
O
CL
0
^
"o.
CD
O
q
d
8
CD
T3
- - 0
o >
Lsl
2 4 6 8 10
Secchi Depth (m)
Polyarthra. dolichoptera
O CD
O d
Cl !
^ 0
- - O
d C
CD
. 8
2 4 6 8 10
Secchi Depth (m)
in
o 0
d >
si
o Q:
Polyarthra. major
o
0
Q.
CD
q
d
.
•
, "
•
• • '.
L. .'S'l^sL
,
• •
•
• •
lj£iC_-_.'..
. s
d
CD CD
- o O
d C
CD
- o 0
d >
"CD
-00
o Q:
- 0
2 4 6 8 10
Secchi Depth (m)
Page 17 of 20
-------
& 00
:= d
ro
o to
O d
CL
0 •*
Gi-
ro CM
O d
q
d
Capture Probability of Macroinvertebrate Taxon Along Secchi Depth Gradient
Polyarthra. remata Polyarthra.vulgaris Polyphemus
2 4 6 8 10
Secchi Depth (m)
CM 0)
co O
d C
ro
•* c
SE
*- 0)
o >
§1
_g> oo
:= d
!a
ro
o d
Gi-
ro
CO
- CO
2 4 6 8 10
Secchi Depth (m)
c\i
d
0
O
ro
T3
.Q
0
J5
o:
00
d
.Q
ro
.Q
o
-I—*
Q.
0
O
ro
T3
-00
"-^
JD
h o 0)
o:
2 4 6 8 10
Secchi Depth (m)
Pompholyx
Skistodiaptomus
Skistodiaptomus. oregonensis
o
-I— ' CO
:= d ~
ro
2 ^r
Ql d ~
0
3
"Q. c\|
CD d
O
0
d
t
•
;
i
• •• --^« = '•
- . p _
•<~
0) >,
2 O -*— ' • —
~ C -— o
° ro !Q
T3 ro
CD ^ 2 ° ~
d < Ql
0 0) ^ _
> ^ °
. 8 ro Q.
d "0 ro CM
o: o o
P_
B
f
-
. • , "
. .. »-____:
ji p _
d "~
0) 0) >, m
- o O 4? °° _
d C :^ o
ro .Q
. 8 C 5 co _
O ^ O O
< ^
-00 0) •* _
d > ^ °
'•4— ' -1— '
ro Q.
.80 ro
'•4— '
ro
. 8 0
o o:
- 0
2 4 6 8 10
Secchi Depth (m)
2 4 6 8 10
Secchi Depth (m)
Page 18 of 20
2 4 6 8 10
Secchi Depth (m)
-------
Capture Probability of Macroinvertebrate Taxon Along Secchi Depth Gradient
Synchaeta
& 00
:= d
CD
.a
o
CL
0
-I—*
a.
CD
O
q
d
2 4 6 8 10
Secchi Depth (m)
CO
d
0
O
CD
T3
CN
d
T- 0
o >
- -i
Trichocerca
:= o
!a
CD
o to
O d
,-»V^-=- - o
2 4 6 8 10
Secchi Depth (m)
q
d
^ 0
o O
d C
CD
T3
5- i
0 E
_ <
q 0
JD
o 0)
a:
Trichocerca. birostris
:= o
!a
CD
o to
O d
0 "*
"o.
u s
d
2 4 6 8 10
Secchi Depth (m)
ro 0)
o O
d C
CD
OM "2
§ I
O 0
d >
'•4—'
JD
o 0
o a:
Trichocerca. cylindrica
Trichocerca. multicrinis
Trichocerca.porcellus
P _
T~
_^> 00
:= °'
o
CD
o to
0 d
Q-
0 •* _
3 °
-1—*
Q.
CD OM
O o ~
0
d
.
a
.
^
. »" ,
•
, , "
•*•'••' ' •
..!•"?. *>u>^*..*J 5'J»*_ •. •
-?- P _
d ^~
0
- ^— O t^* °P _
d C •= °
CD ^2
T3 CD
. ^ C ^2 -
^ °
'-I—' -1—'
CD Q.
- o 0 tO ^ °
'•4— ' -I— '
CD Q.
.50 tO ^ _
OQ: O °
._. o
0 o
.
•
•
• ^
. ,
•
- 0
d
T- 0
- o O
d C
CD
T3
- ° §
^2
-00
"-^
CD
- o 0)
o:
— o
2 4 6 8 10
Secchi Depth (m)
2 4 6 8 10
Secchi Depth (m)
2 4 6 8 10
Secchi Depth (m)
Page 19 of 20
-------
in
o
CD
O CO
O d
Ql
0 o|
3 °
-i—*
Q.
CD T-
O o
q
o
Capture Probability of Macroinvertebrate Taxon Along Secchi Depth Gradient
Trichocerca. rousseleti Trichocerca. similis Tropocyclops.prasinus
2 4 6 8 ^0
Secchi Depth (m)
o oq
o
^ 0 >,
q O ^
o C •=
CD .Q ,
O ~
CN ==
o CD
d "O
C
^2
q <
CD
.Q
"5 o
•9 °
O d
Q.
0
a: o
o
q
ci
q
ci 0
.a
<
0
1
0
a:
2 4 6 8 ^0
Secchi Depth (m)
2 4 6 8 ^0
Secchi Depth (m)
Tropocyclops.prasinus.mexicanus
q _
"-
_^> 00
:= °
o
CD
O CD
0 ci
Ql
0 •* _
^_«
Q.
CD CN
O o ~
0
d
.
*.
f
\ .
JL •
. _!ii*f .iS
s
•
*
•
.• ''T
• •••. • .
* '• ' ' '
£*••*•.'»••"• •
El.UA_.-i.
T—
d
00 0
- o O
d C
CD
CD ~°
' § 1
- O 0
° "^
CO
o ~fl5
o Q:
- 0
2 4 6 8 10
Secchi Depth (m)
Page 20 of 20
-------
Appendix 8 - Wl TN GAM Models Macroinvertebrates
-------
Capture Probability of Macroinvertebrate Taxon Along TN Gradient
Baslaeschna Janata
Dromogomphus spinosus
Dubiraphia minima
o
•<-
& °°
™ d
o
CD
O CD
O CD
Q.
0 •* _
13 °
"Q.
CD CN
O o i -
0
d
•
.
•
•
•
->^k
Ix^^^jV^.^
- 0 P _
d "~
T- CD •>,
. 5 o -^ °° -
d C 1= o
CD .a
T3 CD
. 5 C ^2 CD
° E 2 °
< °-
-50 CD "* _
d > ^ °
to "o.
- o CD
o ^ :__ o
CD !£2
m "° ro
. " C ^2 CD _
° E 2 °
§ = -
^-^ ^-^
CD Q.
-°2 o2-
o-
.
.
•
5
a
• . ;
*i _ o jj,
••• liiil!'.-
.5 P _
d "~
.58 ^oo _
d C = °
CD .Q
T3 CD
0 C ^2 CD
"d| !°"
• ° .1 § = -
^-^ ^-^
CD Q.
-°2 o2-
o-
.
B
'•
.^ 1
•
. •• i
.. . ^s •* ' "'
. 0
d
CN CD
.00
d C
CD
"^
O ^
" d 1
.50
o .>
•+—I
CD
.50
o o:
- 0
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
Page 1 of 22
-------
Capture Probability of Macroinvertebrate Taxon Along TN Gradient
Haemopis grandis
Hagenius brevistylus
Hyalella azteca
o
•<-
•O CO
™ d
o
CD
o to
O d
Q.
2 ^ -
13 °
^— '
Q.
CD CN
O o
0
d
^
•
•
>
"X~V
^h^
- 0 P _
d "~
*- ® >, m
- 0 O -i^ °°. _
d C •= o
CD .Q
T3 CD
_ 5 C .£! co _
° E 2 °
< °-
-o0 2 S -
> ^ °
^-^ ^-^
CD Q.
- o >, m
-00 ^ °° -
d C 1= °
CD .a
T3 CD
. 5 C J3 (D _
o E 2 °
< °-
-00 0) ^ _
d > ^ °
"co "o.
- o cu /5 ^ -
o: o o
- o ^ -
.
. 5
"
•
•
• •
** • *
• •
*•* f
• B
• J
• . I,;
• ••
•b ••. .'I.
t:.
.. . . in S „
- 0)
d
in CD
- r-- O
d C
CD
"^
LO ^
" d 1
00
- CO 0
d >
•+—I
CD
. 2 0
o a:
~ O
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
Rlplstes paraslta
Slavina appendiculata
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
Stenacron interpunctatum
>,
:_—
o
CD
.a
0
CL
0
"o.
CD
O
O _
T~
00
d
(D
d
^
0
CN _
0
d
.
•
.
•
.. . i^l...
- CO
d
CN
d
CN
Q
^"
d
. fe
d
- 0
0
f~
CD
T3
(~
13
^
f!5
>
"CD
0
a:
p _
^
>, CO _
'— O
o
CO
O CD
O o
el
2 •*: -
"o.
0
d
.
•
t
I
m
;' \
.. . iSvil!1
. 0
d
- O
d
- O
- o
- 0
0
O
CD
T3
f~
D
<
(1)
>
"CD
0
a:
>,
.— —
O
CO
o
0
CL
0
"o.
p
^
00
o
CD
d
^r
d
0
0
d
.
.. . ^L
s_
Li.!'.
- 0
d
^ 0
- o O
d C
CD
^ T3
" °- §
-00
"-^
JD
- o Q)
a:
- 0
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
Page 2 of 22
-------
Capture Probability of Macroinvertebrate Taxon Along TN Gradient
Stenonema femoratum
& 00
:= d
CD
.a
o
CL
0
-I—*
a.
CD
O
0
0
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
S
d
a> CD
CM O
d C
CD
CM "2
^ §
0 E
CD
o0
Stylaria lacustris
:= o
!a
£1 CO
O d
§
-I—*
a.
o
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
CM
d
&
*- 0
o >
§1
Valvata tricarinata
s
d
^ 0 >,
co O ~
d C 1=
0
^
"o.
in
8 -
o
q
d
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
•i 0
^ CD
= 0)
Amnicola
Arigomphus
Basiaeschna
o
CO -
d
-i— • ~~
• —
_£j (— i
CD ° _
-Q d
0
Ql
0
D T- _
0.0
CD
O
O
o _
r~i
t
*%
• J
.. . iu •"..•-... .-
0
CM -\
O
§0
O -i— ' LO
O ^~ ^^ ^~ ^
CD !5 o
T3 CD
c .a
•n E 2 o
" ° < °- d ~
° 0 2
> 3
CD Q. o -
CM "0 ,^0
- P Q-; O
o
o
- o P _
(—1
a
•
| .
" f
' • '.
* ' ;
r^
- T- O
d -^ ~
® >, m
CO O 4? °° _
— T — f~ '^ O
o CD o
T3 CD
C ^2 CD
0) ^ O d
' d < ^
0 0) •* _
> 3 °
•* "m "S
_ o to Q.
00 tO CM _
a: o o
o -
s
0
• ^ .
•• V«.«j *
• • a 9MU^^JSlaBaL _ •• o BOD . . . . a aa
d
^r 0
. o o
d C
CD
"^
- P c
CM <
CM
- O 0
o .>
"CD
-00
- 0
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
Page 3 of 22
-------
Capture Probability of Macroinvertebrate Taxon Along TN Gradient
Bezzia/Palpomyia
Boyerla
Caecidotea
o
•<-
,>> 00
™ d
o
CD
o to
O d
CL
0 •* _
3 °
"o.
CD CN
O o -
0
d
a
•
•
,
-'T^v
^%ijp^.
- O Q ~
d
*- tD >,
.00 ~ (Q
d C :=._;-
CD ^2 °
~C CD
- 5 C ^2
0 =5 0
< Q- d ~
-00 2
.> 3
"CD "a. 3 °
_ o CD Q_
00 CD
ID
.20
- 0
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
Caenis
Callibaetis
Celithemis
p _
^
^* 00
:= d
o
CD
O (D
0 d
Q-
0 •* _
3 °
"o.
CD OM
O o
0
d
.
•
.
•
" .
*• • •
.--..; _i
• *! *y^* ' '. f t
... afeJi..:
. ^ p _
d "~
oo , m
-mo •<-? m- -
d C := °
CD ^2
c5 C ^2 to
' ° E 2 ° "
, m
- 0 O 4? °° _
d C r^ °
CD .Q
. 8 "c 5 ^ °
^ ^ "o.
d Q^ O °
. o g -
.
:
• '
€
.. . »-» "-• '*• •
. S
d
^r 0
. o o
d C
CD
- 8 C
0 _g
"CD
-00
d (Y"
- 0
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
Page 4 of 22
-------
Capture Probability of Macroinvertebrate Taxon Along TN Gradient
Centroptilum
Cernotina
Chaetogaster
in •*
O
T~
^ 00
:= d
o
CD
o to
O o
Q.
0 •* _
"o.
CD CN
O d -
0
d
.
"
i
>
--'/^\
' .ssLt • • ^T^.N
•^' J «' I '• CsL> -
- o d
d
^ 0 >, ^r
-oo ~ d ~
d C •=
CD .Q
73 to _
o CI .Q • —
~ d =3 0 °
.Q <-
.-
0 c 1= °
CD !Q
^ T3 CD
- CM C ^2 -
^o
. "CD "o.
- q "0 tD , _
- o O 4? °° _
d C := °
CD ^2
T3 CD
. 0 C ^2 ^ °
'•4— ' -1— '
CD Q.
- o tD /O , „,
. r- o 4? °° _
d C := °
CD .Q
"O CO
LO ^ O tD
' ° E 2 ° "
^ °
'•4— ' -I— '
CD Q.
2 "0 tD
'•4— '
CD
- S 0
o o:
~ O
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
Page 5 of 22
-------
CD
.a
o
CL
0
-I—*
a.
CD
O
_g> 00
:= d
q
d
Capture Probability of Macroinvertebrate Taxon Along TN Gradient
Dubiraphia Ectopria Enallagma
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
8
d
•* 0
o O
d C
CD
« C
o b:
q 0
o >
'•4—'
JO
-50
o o:
:= o
!a
Q.
CD
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
r- 0
o O
d C
CD
•« c
o t
-00
kSl
o a:
:= o
!a
0)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
u a
m 0)
OM O
d C
CD
T- 0
o >
§1
o a:
Ephemera
Epltheca
Erpobdella
q _
^ 00
:= °'
CD
o to
0 d
Q-
0) •* _
Q.
CD OM
O o ~
0
d
•
.
.. . ^-IL.™
OM O _
d ^ ~
d C = °
CD .Q
- 5 i f o -
. 8 0 23-
^ CD Q.
-30 tO OS, _
OQ: o °
- 0 § -
.
•
. • •
• j
.. . Sailkj
o P _
d "~
.88 ^» _
d C = °
CD .Q
. o C ^2 tp
d ^ O o
< ^
-50 ^ °
^ JD Q.
. o § -
•
.
;
'.
.. . *•*«»• * ' „
- 0
d
i i
0.01 0.01
bundance
-00
JD
- o Q)
o:
- 0
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
Page 6 of 22
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
-------
Capture Probability of Macroinvertebrate Taxon Along TN Gradient
Eurylophella
Fossaria
Gomphus
o
•<-
_^> 00 _
™ d
o
CD
o to
O d
Q.
0 •* _
^_«
Q.
CD c\i
O o ~
0
d
„ - - »
/''" A»
/ • " \*
' • • \\
' tt!/4» ^L-
Nl • —
d
0) CD >, op _
- •<- O -i_» o
d C 1=
CD .a
2 c % § -
" o E 2
< G-
• 5 .1 § 5 -
-1-^ ^-J
in ro Q-
o
o o ~
;
8
1
1
•
O
0 _
•58 t§-
CD !Q
r^ T3 CD
-PC ^2 <°
o J3 od-
u. < °~
- P 0 0 ^ _
0 .> D °
CM ™ S-
- o 75 CD
'-1— «
CD
- § 0
~ O
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
Gyraulus
Gyrinus
Haemopls
p _
^
^* 00
:= d
o
CD
O (D
2 °
CL
0 •* _
^ °
"o.
CD OM
O o ~
0
d
.
... i«JL.
-co P _
d "~
*- tD >, _
- co O •<-? m- -
d C := °
CD ^2
"O CO
CM ^~ o to
o ^ O o
< c^-
- ^ 0 0 ^ -
d > ^ °
'•4— ' -1— '
CD Q.
- g 0 CO ^ _
da: o °
- 0 g -
.
.
*
.
|
.
{
...iaiS
. 0 P _
d "~
^ ® >, m
- 0 O 4? °° _
d C r^ °
CD ^2
T3 CD
. 0 C ^2 ^ °
'•4— ' -I— '
JO Q.
. o g -
.
•
t
\
•..
- 0
d
^ 0
- o O
d C
CD
^ T3
_ o ^
^
-00
>
'•4— '
JD
- o 0
a:
- 0
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
Page 7 of 22
-------
Capture Probability of Macroinvertebrate Taxon Along TN Gradient
Hagenius
& 00
:= d
CD
.a
o
CL
0
-I—*
a.
CD
O
q
d
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
88
CD
T3
C
.Q
-00
sl
o a:
Helisoma
^r
d
CD
O
0
a. ^
CD d
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
in
o
0
O
co c
P CD
0 T3
CM .Q
§ <
0
?; ^o
Hexagenia
:= o
!a
CD
O d
CL
| 5
"o.
CD CM
O o
—Jjj..j..
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
•* 0
in O
d C
CD
^ T3
s|
CM 0
d >
'•4—'
•* —
5 a:
Hyalella
Hydra
Hydroptila
P _
^~
*^
*O 00
:= d
o
CD
o to
0 d
Q-
0 •* _
3 °
-i— »
Q.
CD CN
O o
0
d
.
B
* :
! IT
»t m
q<
*. ^ •
*' • •
*• m I m
'•'•' *! '
• * * °l
/.:. ;•
."• • •*.
nfl* **"••
I'-\ ft*
^•J'l
... gmrJ
. 0) P _
d "~
to CD >, _
. r- o 4? °° _
d C := °
CD .a
. T3 CD
. C5 C ^2 ^ °
'•4-< -1— »
CD Q.
- 2 0 tO <
... SILL
- W P -
d ^~
to CD >, m
- CM 0 -^ °P _
d C = °
CD .Q
T3 CD
CM C O to
d 3 O d
o s—
^ °
'•4-< -1— »
CD a.
. fe "0 tD
"-i—'
JD
- o 0
a:
- 0
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
Page 8 of 22
-------
Capture Probability of Macroinvertebrate Taxon Along TN Gradient
Leptophlebla
& 00
:= d
CD
.a
o
CL
0
-I—*
a.
CD
O
q
d
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
CD
T3
o 0
d >
'•4—'
* ^
o 0
o a:
Leucorrhinia
CD
2
0
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
q
d
^ 0
o O
d C
CD
T3
o 0
J5
o
'•4—'
JD
o 0)
a:
q _
"~
^ 00
:= °
o
CD
o to
0 d
ol
0 •* _
3 °
"o.
CD OM
O o ~
0
d
Mystacides
1
;•
i
• .
t
•
0
1 „
!.
•
kiiJj' J
i i™ "" " "™
. 8
d
^r 0
- o o
d C
CD
m ~&
. 8 £=
d =5
^2
- o 0
d >
"CD
o ~fl5
o a:
~ O
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
Nectopsyche
a _
CD d
O in
§ 2
I.5
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
° 8
g I
0 c
o ^2
d <
>
!s
0
Oecetis
:= °
CD
O CD
O d
si
o o:
Page 9 of 22
-------
Capture Probability of Macroinvertebrate Taxon Along TN Gradient
Orconectes
Oxyethira
Paraleptophlebla
o
•<-
^ 00
:= d
o
CD
o to
O d
CL
0 •* _
13 °
"o.
CD CM
O o -
0
d
_
v' * «^\
V' ."\x
-' ^,,/ji1 ^VC^-.T
-CM P _
d "~
CM g >^ 00
d c: :=d
CD .Q
m "° ro
C .Q 2 ri -
0 > ^ °
"CD "o.
- S 0 Cl ! CM _
do: o °
o
d
o
x^_!V
'/^V
' yo' ."•"• • \\
«C* Ww1#«^V.r
- 0 P _
d "~
.88 ^"o _
d £= = °
CD ^2
rsi ~° tO
. o C ^2 (D _
d ^ O 0
< ^
- S 0 0 * -
d > ^ °
"CD "o.
- 5 0 /O CM _
do: o o
._. o
o d -
• I
•
0
.
^ •
.
. 8
.'.
5
.. . «L'jLu........
- CM
d
co 0
- CM O
d C
CD
r- "g
~ *~. ^
.Q
- ^ 0
0 >
"CD
. § 0
o o:
~ O
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
Phaenopsectra/Trlbelos
Physa
Physella
p _
^
.^
•O 00
:= °
o
CD
O (D
0 d
CL
0 •* _
3 °
-1—*
Q.
CD CM
O o -
0
d
t
,
*
*
0
a
0
1
1
| .
-ft P -
d "~
h* CD •">.
- CN O -^* °^ —
d C •= °
CO &
~O CO
CM C o (O
d D O d
_Q *-
D °
-i—' -i—'
CO Q.
_o0 n 2 ~
o
d
a
0
•
.
•
."
:'."
i "
.. . fc— — •— * „
•*-
d
CD _
oo CD •">. °
o O -^*
d C = ^ _
CD .Q °
^ "C CD
. 8 C ^2 ^ _
d ^ O o
O !—
^
OT Q. d -
'IS 0-:.
O
._. o
0 o
t
.
". . :
. % • •
8 •• "
"
1
CM
- O
d 0
O
CD
^ T3
_ o ^
D
0 .Q
0
- 0 '•*=
d jo
0
- o C£L
~ o
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
Page 10 of 22
-------
Capture Probability of Macroinvertebrate Taxon Along TN Gradient
Pisidium
Planaria
Planorbella
^> 00 _
1=0
CD
O CO _
O d
CL
0 •*
13 °
-i—*
Q.
CD CN
O d
I I I I
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
n 0
CD O
d C
CD
c, <
CO 0
Si
& 00
:= d
CD
O CD
O d
Q.
0 •*
Q.
CD CN
O d
q
d
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
r- 0
co O
d C
CD
oo "a
OM b:
CD
O CD
O d
CL
0 "*
3 °
O)
T- 0
o >
"CD Q.
§ 0 CD CN
OQ: O o
§
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
m 0
o O
d C
CD
* c
q §
0
-00
"-^
JO
o 0
Polycentropus
Probezzia
Procloeon
q _
•<~
& °P
:= d
o
CD
O CO
0 d
CL
0 •* _
3 °
"o.
CD CN
O d -
0
d
B
.
f
a
,"
• !t- .. ,
.. . iA£tii..J
CO O _
d ^ ~
-Z $ ^oo_
d C :^ °
CD ^2
^3 CD
2 C ^2 co
" H 3 O d
o s—
^ °
^ ^ "o.
. o § -
t
. l
' .
i !•
.-• 5
.. . tajftlJl'.,
. 6
d
CO 0
- o O
d C
CD
~^^
. 8 c
0 E
- o 0
d .>
"CD
-00
o Q:
- 0
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
Page 11 of 22
-------
Capture Probability of Macroinvertebrate Taxon Along TN Gradient
Prostoma Ripistes Sialis
"-
^* 00
:= d
o
CD
o to
O d
CL
0 •* _
3 °
"o.
CD CN
O o
q
O
.
r
//^~\s
•v * A»
x/ • A \
'/' •* \\\
/' • B \
1 1 1 1
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
d "~
tl> >, m
-TO jS1 °° _
0 C = °
CD .Q
"O CD
_ § C ^2 cq _
d ^ O o
< ^
-80 0) * _
d > 3 °
"CD "o.
- 8 0 Cl ! CN _
d fV CJ ^
§ -
LJ
*
.
XB/^*~"\^
1 1 1 1
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
d "~
i^ tD >, „,
. Ki O 4? °9 _
d C 1= °
CD .Q
T3 CD
^j C o (o
o ^ O d
< Ol
.20 0 "* _
.—- . ^J
"CD "o.
.00 to CN _
d fV C-) o
d -
u
•
:
_j i
"•( .
0 •
.1 •
..i »« ii llm-.— l... »
1 1 1 1
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
d
in
- ^0
- 0 0
d fy"
- 0
Slavina
Stenacron
Stenelmis
p _
^~
^^
-^ 00
7= °
o
CD
o to
0 d
Ql
0 •* _
3 °
^— '
Q.
CD OM
O o -
0
d
B
q
B
•
B
0
; •!
... ^iJi:.
- 0 P _
d ^~
^- tD >, _
-00 ^ °P -
d C = °
CD .Q
T3 CD
. 0 C ^2 3 °
^-^ ^-^
CD Q.
- ° ^ 0 o -
P_
t
'
*
.
\
0
• ;
v •• • H*
.. . ••***'•
- 0
d ,
- o O ••-• "? _
d C := o
CD .Q
T3 CD ^r
- o c -Q d ~
d _g 0
13 CN
"CD 0.° ~
- ° rv 0 - -
^^ O
. o § -
B
.
. t
•
0
• * * .
..;! ' '
itiil j; L
_
m CD
- K o
d £=
CD
T3
C
3
- 8 <
° 0
~ ^-J
CD
^ 0
- T o:
o
- 0
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
Page 12 of 22
-------
Capture Probability of Macroinvertebrate Taxon Along TN Gradient
Stenonema
& 00
:= ci
CD
o
O
CL
0
Q.
CD CN
O o
q
ci
s
d
O) 0
CM O
d C
CD
™ C
^ §
o 3
0
CD
Stylaria
& 00
:= d
CD
O (D _
O d
Q.
0 •*
.^ U
Q.
CD
q
ci
Sympetrum
s
d
^ 0
co O
d C
CO
« C
^ §
0 E
to
T- 0
o >
_ "CD Q.
§ 0 CO N
OQ: o °
q
ci
,>> 00
:= ci
!a
CD
O (D
O ci
CL
0 "*
=i °
q
ci
^ 0
o O
d C
CD
T3
o 0
"-^
JD
o 0
o:
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
Tanytarsini
Thornia
Triaenodes
P _
T~
_^> 00 _
:= d
o
CO
O (D
2 °
Q-
0 •* _
3 °
"o.
CD OM
O o ~
0
d
,
0
•
I
t
0
I
• * * •*
• " *
.. . iiiJJ.
_ TJ- P _
d ^~
r- ® >, m
- ro o 4? °° _
d C := °
CO .Q
_ T3 CD
?5 C ^2 to
" ° E 2 ° "
™ < °-
.20 0 ^ _
d > ^ °
to "o.
.80 Q g _
- 0 g -
.
.
8
•
j-
.. . tesvk*
. 0 P _
d "~
.88 ^oo _
d C = °
CO .Q
"O CD
_ 8 C ^2 tp
0 E 2 °
3 °
^ ^ "o.
P_
.
*
.
-._
^ i
'
i .
. i. i
... Jsjfciij^
. o
d
CM 0
- o o
d C
CO
^ T3
_ o ^
D
0 ^2
-50
d >
"CD
-00
a:
- 0
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
Page 13 of 22
-------
Capture Probability of Macroinvertebrate Taxon Along TN Gradient
Trlcorythodes
Valvata
Aeshnidae
o
•<-
_^> 00 _
™ d
o
CD
o to
O d
Q.
0 •* _
13 °
"o.
CD c\i
O o ~
0
d
.
.
i
-'7\
^%i-h^^.
- O
d
0
^ 0 >, ^ -
- o o -t-^ °
d C 1=
CO o
,_ "O CD S _
- d | o °
"CD
.50
~ O
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
Ancylidae
Asellidae
Baetidae
p _
•<~
^* 00
:= d
o
CD
o to
0 d
Q-
0 •* _
3 °
-1—*
Q.
CD OM
O o ~
0
d
t
.
•
•
i
. j.
^ A' n*
. 0 P _
d ^~
*- tD >, _
- o O 4? °° _
d C r^ °
CD .Q
T3 CD
. 0 C ^2 tp _
° E 2 °
5 °
-1—' -1—'
CD Q.
- 0 P _
d "~
CM 0 >, oo
d C •= °
CO ^
"O CO
^) C O CD
' ° E 2 ° "
^ °
'•4-* -t-l
CD Q.
" "0 tD
'•4— '
CD
. 8 0
~ O
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
Page 14 of 22
-------
Capture Probability of Macroinvertebrate Taxon Along TN Gradient
Caenidae
Cambaridae
Ceratopogonidae
& 00
= d
!a
CD
O (D _
O d
CL
0 ^r _
Q.
CD OM
O o
q
d
• :
p .%
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
5
s 8
d C
CD
- 0
0 >
'-I—'
CD
80
p _
T~
>^ 00
~ d
!Q
£ CO
00-
°-
(D ^! _
-i—*
/-\
LJ_
CD ,
-i—*
!5
CD
.a
o
CL
/-\
LJ_
°
o _
T
00
d
(D
d
d
<>!
0
0
d
„
• °
•
s ,
"
••
• • *
1 •
" ti.ttJ&:Ji J
*""*"
. &
d
^r
_ r--
d
(D
- in
d
r^
- CO
d
0)
°
~ O
0
o
CD
T3
C
^2
<
0
"-^
"0
a:
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
Chaoboridae
Chironomidae
Cladocera
p _
"
^ 00
:= °
CD
o to
0 d
f^*
LL
0 •* _
^ O
^_«
Q.
CD c\i
O o ~
0
d
,
a
.
•
a
t
.. . *&?.
. 0 S _
d T-
-58 ^g_
° § i °
T3 CD
. o c -Q o
0 J2 2 § -
^ oo _
'*3 -fi O
CD Q.
-°S o§_
d
- 0
. . •
• i--^ •• •
B * *! • " •
• *S • ;
• | " "Lf * \ t
*'•!!" • °.
*• **"l5 «i
• "^ •* •• • "
.'I'ijvj-*
\M *"* * " J ' • *
. • mi \c" -
. ^_
00
d
. s 8 ^
d CD 5 « -
T3 CD °
r^ C o
- "5 3 O
°-9 > _,.
^
o '^3 jr!
JD Q. ts.
d
0
d
.
4
••"
«."
|°i
a 1
:i
" "•!•
o«|
J*"1
... St_
- 0)
d
0
O
- i^ CD
d T3
C
i- -Q
" ° >
^" CO
" ° C?
- 0
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
Page 15 of 22
-------
Capture Probability of Macroinvertebrate Taxon Along TN Gradient
Coenagrionidae Corduliidae Dorylalmldae
"-
^* 00
™ d
o
CD
O CD
O o
Q.
0 •* _
-i—*
Q.
CD CN
O o
0
d
•
f
t
^ ' S
\ * /
\ x ' /
\.* / /
•fc ^v V * ' /
^ ^\.x • • ' /
^^x^ * *' /
v oC - ,. - ' >/ ^
* "**. .- - " ** •
• **• "•
2$M*l5 • i •" • • •
»•
1 1 1 1
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
d "~
on ,
- CN O -*— ' • —
d C •= °
CD &
. T3 CD
!^ c: ^2 CD
r-i ^ O O
^2 £-
^ °
'•4— ' -I— '
CD — S"
- o 0 ro ^ _
OQ: o °
- 0 ^ -
.
•
B
• * •
0 0
t "
*
jl
/"*V "
/A ^»
'A • !J 'AV
^Xh^'r : ^-v-
i i i i
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
d "~
" 0 >, m
- 0 O -i^ °°. _
d C -^ °
CD .Q
-. T3 CD
- q § ^2 CD _
O l—
< °-
-50 0) "* _
° -^ 5 °
CD Q.
-50 CD CN _
OQ: o °
- 0 °- -
,
.
I
m
!'
^^;|^
*""*"*"" ••••••
1 1 1 1
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
d
CN CD
-00
d C
CD
,%, "^
CN *—
- O *—
. ^
^2
-00
d >
'•4— '
CD
-00
o o:
~ o
Elmidae
Enchytraeidae
Ephemerellidae
P _
^~
i= d
CD
O CD
0 o
Q-
0 •* _
Q.
CD CN
O o ~
0
0
.
.
„ •
•
" • B
MiMlJiitiJ °. • • i . .
-CD P _
0 "~
. s 8 ^» _
d C = °
CD .Q
T3 CD
•* C ^2 CD
o 3 O o
O >—
^ °
CD Q.
. 2 "0 ro CN _
OQ: o °
o
d
.
.
5
•
**.f • B !•!•
.. . ..._..,.,....
. 0 P _
0 "~
T- Q) ^>,
.00 -^ °° -
d C = °
CD .Q
T3 CD
. 5 c: ^2 cp _
d =3 00
.§1
o o:
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
Page 16 of 22
-------
Capture Probability of Macroinvertebrate Taxon Along TN Gradient
Ephemeridae
Erpobdellldae
Glossosomatidae
o
•<-
^* 00
™ d
o
CD
o to
O d
Q.
0 •* _
13 °
-i—*
Q.
CD CN
O o i -
0
d
m
.
•
t
_
'jX"*"""*'^ x
->^* ^\~
^\ In 1 * ^Ssllis
... ._....«• -I
-co P _
d "~
•* tD >, _
- LO O -&1 °°. _
d C = °
CD .Q
T3 CD
. 5 C .£2 co _
° E 2 °
3 °
'•4— ' -1— '
-tO Q.
. J -0 CD CN _
- 0 °- -
•
•
s ' S
'^%£X--$^
"" " .«. -r iii ITU
- 0 P _
d "~
^ ® >, m
-00 ^ °° -
d C 1= °
CD .a
T3 CD
. 5 C ^2 CD _
° E 2 °
< °-
-00 0) ^ -
.> D °
-f-* -f-*
CD Q.
- 0
'•4— '
JD
- o Q)
a:
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
Gomphidae
Gyrinidae
Halacaridae
p _
""
& °p
:= d
o
CD
O CD
0 d
Q-
0 •* _
3 °
"o.
CD CN
O o i -
Q
0
.
.
00
I »
!-3 !
1 '^ 1
•' • • "ill • j ' '
. «i-»uH...ii... :_
. =¥ P _
d ^~
- % % ^oo _
o C :^ °
CO _Q
-Q CO
OJ C O CD
" ° E 2 ° "
3 °
"CD "o.
. T 0 tO CN _
. o d -
.
-.'
• • .
1
.. . fauuj*
i^
- O 00
d d
LO 0) >,
- o O -*-•
d C := CD
CD .Q d
_,. T3 CD
. o £= ^2
°| !--
-00 fl)
° .> 3
CD Q. CN
- 1 £ o ° "
. o d -
.
•
•
"» *•
<
-------
Capture Probability of Macroinvertebrate Taxon Along TN Gradient
Heptageniidae
Hirudinidae
Hyalellldae
"-
^ 00
:= d
o
CD
O CO
O d
CL
0 •* _
13 °
-i—*
Q.
CD CN
O d ~
V^ / -~^
0
d
^
*v x
x ^vv '
" V '
" \ \ . /
\\.
VV
VV '
\ '
> ' /
\ / /
\ \ f S
\ vxN ',' /
*• 0"» • tfc^""" """^ '
T^':-'5- • ~~~
m
1 1 1 1
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
d "~
. 83 8 ^°° _
d C = °
CD .Q
CN "° TO
. CN C J2 ^ °
'-I—' -1—'
CD Q.
.00 CO CN _
- 0 °- -
0 o
•
•
•
0
s " S
"'/""N^
^llk.
1 1 1 1
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
d "~
.58 £«» -
d C = °
CD ^2
T3 CD
5 C ^2 CD
" d E 2 d
< °-
-00 0 ^ -
> ZJ °
-1—' -1—'
CD Q.
- o 0 /S ^ -
rv C J O
Ql w "
- 0 °- -
° 0
.
# 5
IB
^
.f 0 *
•/ . .
• '. • ' •
••• ': .
i« .:
>8- -4
."- .•*.
tf* ''-
l&'!
""*"""****
i i i i
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
d
CO 0
- r^ o
d C
CD
^ T3
d 5
<
00 ^
- CO 0
d >
'•4— '
CD
. 2 0
OfV'
LJ_
_ Q
Hydridae
Hydrobiidae
Hydroptilidae
O _
^~
^ °q _
:= °
o
CD
O CD
0 d
CL
0 •* _
<= 0
"a.
CD CN
O o ~
0
d
t
.
t
•
•
1.
0
*
•
oo . Mrr *iii „
. W P -
d "~
.88 3?» _
d C = °
CD .£2
T3 CD
CN C O CO
d 3 O d
< ^
-20 0 ^ _
d .> ^ °
"CD "o.
- fe 0 CD OH _
do: o °
p_
a
"
5
•
•
*
.
•q
; j[ i
... iiiiL..j
in o _
d ^ ~
. * 8 &<°. -
0 C := °
CD !£2
T3 CD
CO C O CO
d 3 O d
< *
. 0! 0 0 ^ _
0 > -1 °
"CD "o.
. ^ "0 CD CN _
o £ O o
P_
B
i •
•4
•
" i'j- :
... SW&b-.-Ji..' .. j .. ..
,_
d
0) 0
- o O
d C
CD
^ T3
o ^
^2
<
^r ^
- o 0
d >
"Jg
§ or
- 0
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
Page 18 of 22
-------
Capture Probability of Macroinvertebrate Taxon Along TN Gradient
Leptocerldae
>} 00
:= d
CD
.a
o
ol
0
-I—*
a.
CD
O
q
d
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
U 8
d
Leptohyphldae
d C = °
CD .Q
^ T3 CD
o 0
d >
si
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
- P
d
o 0
J5
o ,
q O &
d C •=
CD .Q
^ T3 CD
o £ -Q
0
^
"o.
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
in CD
co O
d C
CD
^ £.
d 3
T- 0
o >
"CD
§ 0
Libellulidae
Limnephilidae
Lumbriculidae
p _
^~
^ 00
:= °
CD
o to
0 d
Q-
0 •* _
3 °
Q.
CD OM
O d -
0
0
.
•
*
. "
*
- -
* *
• •
1 *
"
. S P .
0 "~
. § 8 £» _
d C := °
CD .Q
_,. T3 CD
. S C ^2 ^ °
CD a.
.50 CO » -
d C :^ °
CD ^2
_,. T3 CD
. S C ^2 ^ °
CD Q.
.50 CO undance
00
- O 0
d >
. s -i
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
Page 19 of 22
-------
Capture Probability of Macroinvertebrate Taxon Along TN Gradient
Lymnaeidae
Naididae
Phryganeidae
o
•<-
,>> 00
™ d
o
CD
o to
O d
CL
0 •* _
13 °
"o.
CD c\i
O o ~
0
d
•
•
I
•
'v
'/r^\
'/':i\
'^t'j "L '%..
i- P _
d "~
T- 0 >. 00
d C •= °
CD ^2
"O CO
. 8 £= ^2 (D _
° E 2 °
-80 g) * _
d > ^ °
"CD "o.
- | 0 ^ g _
o_
q
.
--V^Sj
<^f£liJi*^V;r
- y P _
d ^~
. s 8 £» _
d C :^ o
CD ^2
in "° ro
2 3 O d
3 °
"CD "o.
. 8 0 ro cs, _
OQ: o o
. o § -
a
"•
• ^
!'
1
• .
.
'i\ i :
.. . ^jlii..j:...i
fi.. Q
d "~
.88 ^» _
d C = °
CD .Q
. T3 CD
fc C ^2 to
" o = 00-
^f n
. co 0 ^ °
"CD "o.
^ -0 CD
"CD
. 8 0
o Q:
- 0
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
Page 20 of 22
-------
^> 00 _
1=0
CD
O CO _
O d
CL
0 •*
13 °
-i—*
Q.
CD CN
O d
q
ci
Capture Probability of Macroinvertebrate Taxon Along TN Gradient
Planorbidae
i i i i
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
00 0
in O
d C
CD
° E
o, <
CN 0
O >
'•4—'
JD
3 0
^> oo _
CD
O CD
O d
CL
0 •*
Q.
CD CM
O d
q
d
Polycentropodldae
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
CD
d
00 0
•* O
d C
CD
CD "a
" §
0 E
•* *^
CN 0
d >
"CD
- 0
o Q:
,>> 00
:= d
!a
CD
O (D
O d
CL
0 ^
3 °
O
q
d
Psephenldae
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
U 8
r- 0
o O
d C
CD
•« c
q §
0
O 0
d >
Sialidae
Tetrastemmatidae
Tubificidae
q _
•<~
^* 00
:= °'
o
CD
O CD
0 d
CL
0 •* _
3 °
"o.
CD CN
O o -
0
d
u
•
•
i
'
,'t •
... JLi*...:
_ ^- p _
d "~
in 0 >,
T— f) I" ^
d C = °
CD ^2
^ -O CD
" 5 =3 0 d ~
^2 <—
^ °
"CD "o.
. o 0 to CN _
do: O o
- 0 § -
B
,
•
... »iid
i- P _
d ^~
, m
- T O -&1 °°. _
0 C := °
CD !Q
. o "c 5 ^ °
_ "CD "o.
. 8 "0 to CN _
do: O o
P_
^
•
. j
^>
d
in 0
- co O
d C
CD
- -1
f}
<
00
- - 0
0 >
"CD
. 8 0
- 0
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
Page 21 of 22
-------
CD
.Q in
O T
CL °
0 o
13 o
"a.
o
q
o
Capture Probability of Macroinvertebrate Taxon Along TN Gradient
Valvatldae
- a
-1 '"I 1 1-
0.2 0.4 0.6 0.8
Total Nitrogen (mg/L)
0
oo O
CO ^
o CD
T3
C
3
o o
ri 0
d .>
^
T CD
Page 22 of 22
-------
Appendix 9 - Wl TP GAM Models Fish
-------
Capture Probability of Fish Taxon Along TP Gradient
Big mouth, buffalo
Black, bullhead
Black.crapple
o
"-
_^> 00 _
™ d
'o
CD
o to
O d
CL
CD •* _
^
-1— »
Q.
CD c\i
O d -
0
d
•'
•
.
. •
•
* •*
- O o
0
oo -^ ° ~
d C 1=
CD .£2 ^r
(M "O CD d
_ o ^ -^
° E e « _
<; CL o
-50 CD
d > ^ CM _
W= .43 0
CD Q.
-IS 05-
o
•
•
•
- <— i O
0 ,- ~
CD
• d g ^» _
c ^
o to !Q
" d ~° tO
0 c .Q 3 °
d *3 -^
CD Q.
^ "0 CD CM _
- §0: o°
. o g -
* "•
8*
a
«
• .•''}•: * •"'*.'.,••
• ' f ''"^A"" 'mmo*m\f' . «7 5*" V"' *. •
\Ji
d
in CD
- T- O
d C
CD
T3
^1 C
d 1
-00
d .>
'•4— '
- tD
- 0 0
o a:
- 0
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
Blackchin. shiner
Blacknose.shiner
Bluegill
p _
^~
•O 00
:= °
o
CD
O (D
0 d
CL
0 •* _
3 °
"o.
CD CM
O o -
0
d
.
t
,
*
.
•
.
•
• • •* *
... . -*, •
V N
d
to _
CD ~>, °
— ^ O ^—*
° C — "? _
CD !Q °
. T3 CD
. fe C ^2 ^ _
d =5 00
.Q <—
^
'*3 -f2 CM
CD Q. d
- ! ^ o -_
°
._. o
0 o
.
' Ji
p _
^~
, m
,_ O .-^ " -
d CD !Q
T3 CD
13 0 d ~
^2 <—
. 3 < °-
°| |5'
13 "o.
d
0 o
0 o
.
^
0
" *o •"
• " * •
•• • " * •
• " . *0 *
• a f s
" • • • !•" * * •
4 • • *
OB • ." •
* • • "
. "•"• a •
• • °
• : ..•':..<"-: • ' ' .
*•* •o'tf'. . /
^H*^; •"•*• •. * * * *
V N
- 00
d
in CD
- to O
d C
CD
"U
^ C
~ • ^
o
"CD
.20
~ O
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
Page 1 of 7
-------
Capture Probability of Fish Taxon Along TP Gradient
Bluegill.x.pumpkinseed. hybrid
Bluntnose. minnow
Bowfin
o
•<-
_^> 00
:= d
o
CD
O d
Q_
» -
d C = °
- o ^ -Q . —
d ^ O o
.Q <-
*^ Q_
-00 CD "* _
d > ^ °
CD Q.
o "55 tO CM
~ (Y O d ~
o
d
•:
.--- --.,
• • "••«•. ^
•. " ""
' • •
• I
' ..." *•*
* • • •
" i '• :•"... • '
' • * • •"«•.• • • • '
*'sjiil-««VVl«'-' • * rN%ir»*
f O
O T-
.88 ^§-
d C = °
Q D O d
<£ ^
. s 0 0 ^ _
.^_ ^
CD Q.
§ "55 CD
CO
_ (\\
a:
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
Brook.silverside
Brown.bullhead
Burbot
>,
j^
o
CD
^2
0
Ql
0
"o.
CD
O
p _
"<~
00
d
(D
d
o -
CN _
0
d
t
•
t
•*
•
•1
. *
. . " •
•
* • •
• *V' .
U3
d
g:
d
CO
d
OM
d
d
— o
0
O
CD
T3
C
^
<^
0
"CD
0
a:
>,
j^
o
CD
^2
0
CL
§
"o.
0
O _
^~
00 _
o
(D
d
^r
d
d ~
0
d
^
'
*•
'.'••'•
*
\m J
d
OM
- O
d
- 0
d
- O
d
- 9
o
0
O
c
CD
T3
C
3
<£
0
'"CD
0
a:
>,
j^
o
CD
0
CL
§
"o.
O
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
ro 0
1-90
o c
CD
T3
OM .Q
d <
0
"-^
JO
- 0
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
Page 2 of 7
-------
& 00
:= ci
CD
o to
O d
CL
0 •*
13 °
-i—*
Q.
CD CM
O o
q
ci
Capture Probability of Fish Taxon Along TP Gradient
Common, carp Common, shiner Fresh water, drum
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
2 8 £»»
d C 1= °
CD .Q
_.. "O CD
^1 CI o to
d 5 P d
§ | § d
"-^ -I—'
CD Q.
S 0 CO oj I
q
ci
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
O) 0
- O
d C
CD
& 00
:= d
CD
O (D
O d
< °-
§00
d >
•^
sl
3
-I—»
Q.
CD CN
q
d
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
00
d
J 8
d C
CD
^ T3
5 I
^ <
O 0
d >
'•4—'
_ CD
o 0
o o:
Golden.shiner
Green, sunfish
Green, sunfish.x.pumpkinseed. hybrid
q _
T~
_^> oq _
:= d
o
CD
O (D
0 d
CL
0 •* _
3 °
"o.
CD CM
O o -
0
d
a
a
"
•
•:'•/' • •*. .
•' V. «/Ai'.-' : ^.il» X*_-...
CO ^
d "~
°> <" >^ 00
- c\i O *? °° _
d C r^ °
CD .Q
R| 1 5 co
d =5 0 d ~
^2 >-
< °-
- ^ 0 0 ^ _
d £ 3 °
"CD "o.
. o 0 ro ^ °
"CD "o.
2 "55 ro CM
"do: O o
. o § -
^
s
* "
.
•
' -. ' -
•
.
T—
d
00 0
- o O
d C
CD
. 8?
^
0 ^2
^. <
- o 0
d >
"CD
. 8 0
o a:
- 0
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
Page 3 of 7
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
-------
Capture Probability of Fish Taxon Along TP Gradient
Iowa.darter
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
CO
p
d 0
O
JN C
P CD
o -o
C
CD
§
Johnny.darter
.Q
CD
O CO
O d
.
5
Q.
CD
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
CO
o
0)
JD
o 0
d Q:
Largemouth.bass
o
8 8 £>»
d C 1= °
CD .£2
§ C J2 cq
d 5 P o
0
^
"o.
••«••
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
O) 0
*- o
d C
CD
CD
§ 0
=> o:
Logperch
Longnose.gar
Mimic.shiner
p _
T
^ °° _
:= 0
!Q
CD
.Q CO _
2 °
CL
0 •* _
3 °
"Q.
CD c\i
O o -
0
d
, "
•
f
•
• o
•
*• •
• '*'•'•''• /
* i *" "M!*" * * £
• "
• • \
'**••• - •-• •
.
^
T- P _
d ^~
in 0 >,
T— f) j^^ 00
d C = °
CD .Q
T3 CD
- ^ C ^2 CD
d ^ O o
f^ t-_
3 °
ID S.
S ±±^
0 ro ^ -
do: o °
„ o
0 o
.
a
•
* •
•
. 0 P _
d "~
*- <1> >^ m
- o O 4? °° _
d C = °
CD ^2
T3 CD
. 0 C ^2 CD _
O ^ O O
f} ^
D °
^0 S.
- o "0) /O csi _
o: o o
0 o
° 0
•
.
9 *
f
*
•
•
. J
•J
.
"
• .
* "
•
• ,
• *0 .
,• .
r«
•
*'.
* ^^ ^ * *
•
- O)
d
^r 0
- r^ o
d C
CD
"U
in ^
- . ^
° ^2
<
- CO 0
d >
ID
. | 0
~ O
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
Page 4 of 7
-------
Capture Probability of Fish Taxon Along TP Gradient
Mottled, sculpln
Mudminnow
Muskellunge
o
•<-
,>> 00
: — o
o
CD
O
d °
P! $ >^ •*
- O O -!-•.—
d C 1— °
CD .Q
CM "° ro
o CI ^2 oo
d =3 00
-Q Jr
< Q-
-50 0 £ _
° > 3 °
CD O_
- o 0 ™ . -
d [£ O °
0 o
0 d
•
X •
\
N
%- -
" * - „
•" —• ' *° "" '
• _ _" ""^^
-""""* """ " ~g^^^^-^
• • ' ' * ""~-T^---
- 0 P
d "~
.88 ^"o _
d C 1— °
en o
rsi "° ro
. 0 C ^2 ^ °
CD Q.
do; O °
._. o
0 d
•
•
q
•
'
. •• •" .'"' "
, „ ', : ':' •_*?.', ::..,,'*L',", ,:;,„;, •'„ •., •.:.
- 0
d
r^ 0
- o o
d C
CD
"^
O -^
• ^
^ <
-00
0 .>
CD
^ "m
d Q^
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
Northern, pike
Pumpkinseed
Rock, bass
P _
•<~
^ 00
:= °
o
CD
O (D
0 d
CL
0 •* _
3 °
"o.
CD CM
O o ~
0
d
f
'
.
•
•
« .
'.•* •.••<*.•»'' •: '. ,;' " •
. s
d
r- 0
- o o
d C
CD
. 8 c
°<
- o 0
d >
"CD
. 8 0
o a:
- 0
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
00
0
Q.
CD
p
d
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
Page 5 of 7
00
d
CM o
d
0
O
CD
^
0 .Q
CM
- 0
O >
si
o a:
p _
!5
CD
0 d ~
0 ^r _
-i—*
Q.
ro CM
0
d
•
•
•
• *
•• • :
:'J*&£
,
<
i. •
%£L-'*,:..'~~z:j-
.•
d
r^ 0
- oo O
d C
CD
- ™ §
o 3
-20
0 >
"CD
. 8 0
o a:
- 0
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
-------
Capture Probability of Fish Taxon Along TP Gradient
Rosyface.shiner
P _
^~
_^> 00
:= d
o
CD
o to
O d
CL
0) •* _
-1—*
Q.
CD CN
O o -
0
d
o
.
. • •
• * * •
Nl
d
^ 0
d C
CD
00 ^
- o ^-
°l
-00
"-^
CD
- 8 0
o o:
- 0
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
Shorthead. redhorse
Smallmouth.bass
O CO
O d
0)
Q.
CD T-
0
•
,
0
m
•
0 •
•* • "
""* * • •• •
VJ^
d
. 2 8
d C
CD
T3
. ^ C
d 3
.Q
CD <
CO
d >
CO ™
. 8 0
° a:
o
P _
"<~
~ d
!Q
CD
O CD
O d
Q-
0 •* _
^ °
"o.
CD CN _
O o
0
d
o
•
•
.
• •
" ': "
*'•".. . .J
•• ** S*» * * • • • •
*\ WJ 7* * I" * •• " " f*
• ^«Vl>'
'-I—'
CD
. ™. Q)
0 QX
~ O
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
Troutperch
CD
.a
o
Ql
0
a.
CD
O
p
d
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
d ®
u Q
co CD
d T3
0
"CD
00
P
Walleye
White, bass
-1—*
^ —
o
CO
o
0
i_
Q_
2
^
"o.
CO
p _
^~
°p _
o
CD
d
^r
d
!s
p _
^~
-1—' . —
1^ O
!Q
CD
O CD
0 d
0) •* _
3 °
"o.
CD CN
O o i -
0
d
B
i
t
.
• *
i.
''f' '
.,
CO
d
^
- CN
d
_ CN
. J
d
- P
~ O
0
o
c
CD
T3
f~
3
^2
<
0
>
"CD
0
a:
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
Page 6 of 7
-------
Capture Probability of Fish Taxon Along TP Gradient
Whlte.crapple
White, sucker
yellow.bullhead
•<-
!>\ nn
j"V CO
— d
o
CD
O CD
O d
CL
Q) •* _
3 °
"o.
CD CN
O o
0
d
C
_
-I— ' O) —
:= d
.Q
-08-
0 d
CL
Q) oo _
13 d
-i—*
Q.
CO g _
O d
in
r~ i
. '. '
.
• • •* .
•
i i i i i i 1 1 i i i i i i i 1 1
.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
Yellow, perch
.
• * *
* . " °
• • ••
t*
... .
t'' *,'••" .*
' •.«•'•.£ , V i.1
* •"* • '*y • * .*•
• ' '. ".".•• •? i '• * . '* M '
• •.••*•:.*« »"'.'. ' «•
• •• •• • • r» _^ •• f • - o
• •,'•• _* ••/ "U • ••• *
• .-iA'A. '-vrf1 ° .«••.*•• 'v>
d
Tf
- O
d
m
- o
d
OM
- 0
d
- o
°
- O
""
_ °p
o
(D
d
_ ^
°
OM
d
- 0
"-
® >, m
o •<-? m. -
C 1= °
CD .a
T3 CD
C O CD
3 O d
.a £-
^ °
"CD "o.
0 to g _
0
Q
C
0
o
c
CD
T3
^
^2
0
"-^
CD
0
a:
~ ~ « ,
_ **•*«. ^
' — ' — ~— - --LJI — —
""--77 --_^^
• * * "***-* ***
. ** ^
. • » • . ^ .
•• " *
•""• A1 .'.."".}* *"" • '
":•' J*': "•*£.„' • •••'.•
• * ,' '$ _y tITii'rr .. * * .11. r.T-.iPr ...'.**...
I i i i i i i | i i i i i i i 1 1
.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
d "~
oo 0) >,
T— t) I" ^
d C 1= °
CD ^2
_.. "O CD
C O CD
o ^ O d
"CD
. S 0
o a:
~ O
0.003 0.01 0.03 0.1
Total Phosphorus (mg/L)
Page 7 of 7
-------
Appendix 10 - Wl TN GAM Models Fish
-------
Capture Probability of Fish Taxon Along TN Gradient
Black.crapple
& 00
:= ci
CD
o
O
CL
0
Q.
CD CN
O o
q
ci
0.2 0.5 1
Total Nitrogen (mg/L)
O)
d
in 0
- O
d C
CD
T3
^1 C
^
.Q
<
o 0
d >
'•4—'
-. tD
Blackchin. shiner
& oo
:= d
CD
O CO _
O d
Q_
0 •*
=i °
-I—*
Q.
q
ci
\- 8
ci
0.2 0.5 1
Total Nitrogen (mg/L)
r- 0
o O
d C
CD
•« c
§1
CO
O 0
d >
si
o a:
Blacknose.shiner
.>> 00
= ci
i!
CD
O (D
O O
Ql
0 ^
=i °
-I—*
Q.
CD OM
O o
q
ci
0.2 0.5 1
Total Nitrogen (mg/L)
$
d
CD 0
r*~ o
d C
CD
^ C
S E
oo <
CO 0
d >
'•4—'
CD
2 0
o o:
Bluegill
Bluegill.x.pumpkinseed.hybrid
Bluntnose. minnow
p _
T~
^> 00 _
:= d
o
CD
O CD
2 °
CL
0 •* _
"o.
CD CN
O o ~
0
d
m
0 .
' V
0 f * • *
« •
,•
» * •
• • *
a**
. . • •
• •
J.
' •
•
•
- r- P _
d "~
0) ® >, on
. in o *? °° _
d C := °
CD ^2
_,. T3 CD
. ^ c .a ^ °
"CD "o.
. ^ 0 to g _
._. o
° 0
.
.
.
•
.
*
» '. •
. 0 P _
d "~
_ g 0 >,cq _
d C •= °
CD .a
"O CO
. o C ^2 tp _
o ^ O o
f} ^
< °-
.50 0 ^ _
. — . ^J
"co "o.
_ o 0 CD ^ _
do: O o
0 o
° 0
.
•
^
••
••
"
" •
• ^
^ • • *
*-M:J» t"
•
*
. . .•
o ^ ^
•
u;
d
00 0
- OM O
d C
CD
^ T3
~ . ^
° ^2
^. <
d >
"CD
. fe 0
~ O
0.2 0.5 1
Total Nitrogen (mg/L)
0.2 0.5 1
Total Nitrogen (mg/L)
0.2 0.5 1
Total Nitrogen (mg/L)
Page 1 of 6
-------
& 00
:= d
CD
o to
O d
CL
0 •*
Q.
CD CN
O o
0
0
Capture Probability of Fish Taxon Along TN Gradient
Brook.silverside
0.2 0.5 1
Total Nitrogen (mg/L)
in
d
cq
d
2
0
O
CD
T3
:= o
!O
CD
o to
O d
'
5
CD Q.
*- "0 CD
'•4—'
_ CD
o 0
o o:
Golden.shiner
Green, sunfish
Green, sunfish.x.pumpkinseed. hybrid
p _
_^> 00
:= °'
CD
o to
0 d
Q-
0 •* _
3 °
Q.
CD OM
O o ~
0
0
•
.
• ''. I '
* ' • • *
• *•_ . • .••*•*'
- -T- P _
d ^~
T- 0 >. 00
d C •= °
CD .Q
- 81 1"-
d ^ O o
.80 a) •* _
_ CD Q.
.80 ^ g -
._. o
0 o
•
•i •
• ^
',_ •
• . s
. 0 P _
d "~
.08 ^» -
d C := °
CD ^2
. 8 1 & <° _
o ^ O o
- 8 0 Q) ^ _
d > ^ °
CD Q.
.50 tO undance
^. <
- O 0
d >
.81
o o:
0.2 0.5 1
Total Nitrogen (mg/L)
0.2 0.5 1
Total Nitrogen (mg/L)
0.2 0.5 1
Total Nitrogen (mg/L)
Page 2 of 6
-------
Capture Probability of Fish Taxon Along TN Gradient
_g> CO
:= d
CD
O .3-
Q- °
0
Is
o
q
d
Iowa.darter
0.2 0.5 1
Total Nitrogen (mg/L)
Johnny.darter
0.2 0.5 1
Total Nitrogen (mg/L)
:= o
!a
0)
Largemouth.bass
0.2 0.5 1
Total Nitrogen (mg/L)
O) 0
*- o
d C
CD
8 I
o a:
Logperch
Mimic.shiner
Mottled.sculpin
p _
T~
,>> 00
:= d
o
CD
O CD
2 °
CL
0 •* _
3 °
"o.
CD CN
O o -
0
d
.
0
.
* > •
'
"••"'.".
\!:.\£v...
T- P _
d ^~
5 8 ^ °°. _
d C •= °
CD .Q
^ T3 CD
" 5 =3 0 d ~
.Q >-
^ °
'-I—' -i—'
CD Q.
-00 tO g _
0 o
0 o
.
*. »••
. „•
. B
. B
• , * *
f
••
• « "
"
.
.• . t
- 0) P _
d "~
. s 8 ^» _
d C := °
CD ^2
"O CO
. Eo £= ^2 D °
'-I—' -1—'
CD Q.
2 "0 tD CN
" d Q: o o
o
d
.
•
•
.
B .
f-
. 0
d
^ 0
- o O
d C
CD
^ T3
_ o ^
f^
^
-00
>
'•4— '
CD
-00
o:
~ O
0.2 0.5 1
Total Nitrogen (mg/L)
0.2 0.5 1
Total Nitrogen (mg/L)
0.2 0.5 1
Total Nitrogen (mg/L)
Page 3 of 6
-------
& 00
:= ci
CO
O CD
O o
CL
0 •*
13 °
-i—*
Q.
CD CN
O o
q
ci
Capture Probability of Fish Taxon Along TN Gradient
Muskellunge
8
d
•* 0
o O
d C
CO
« C
q iz
0 E
N <
q 0
& oq
:= o
CD
O CO _
O ci
Q.
0 •*
JO Q.
50 /O CN
OQ: o o
§
Northern.pike
§
d
•* 0
o o
d C
CO
« c
§1
CN
O 0
d >
& 00
:= d
CD
O CD
O o
CL
0 •*
13 °
-^- -i—•
JO Q.
00 CO CN
o Q:
O
q
ci
Pumpkinseed
2 8
d C
CD
s
o o:
0.2 0.5 1
Total Nitrogen (mg/L)
0.2 0.5 1
Total Nitrogen (mg/L)
0.2 0.5 1
Total Nitrogen (mg/L)
Rock, bass
Shorthead. redhorse
Smallmouth.bass
q _
^~
_^> oq _
:= d
o
CD
O CD
0 d
CL
0 •* _
3 °
"o.
CO CN
O o i -
0
d
*
.*'
• a
.•
^'- .....
.
•„
*v o
_ ^~ . ^
d *~
. ^ 8 ^ °° -
o C r^
CO ^
"O CO
24 ^ o CD
" ° E 2 ° "
, m
- o o •<-? m- -
0 C =°
CD .Q
rxi "° tO
. 0 C ^2 ^ o
to "o.
-00 tO CN _
d Q^ ^ °
o
d
.
.
• .
*
•
t •
•' -: 'i
. %
• s
i^'vV •
• • •
•
- ^
d
0) 0
- co O
d C
CO
"^
OJ ^
0 E
<^
.20
0 .>
JD
~ O
0.2 0.5 1
Total Nitrogen (mg/L)
0.2 0.5 1
Total Nitrogen (mg/L)
0.2 0.5 1
Total Nitrogen (mg/L)
Page 4 of 6
-------
00
o
_Q CD
CD o
.Q
O
CL •*
q
ci
Capture Probability of Fish Taxon Along TN Gradient
Trout perch Walleye White.crappie
0.2 0.5 1
Total Nitrogen (mg/L)
0
o
ro
CM
q
,>> 00
:= ci
!Q
CD
O CO
O o
Q.
0 "*
> 3
-i—' -i—'
CD Q.
"0
o:
o
q
o
0.2 0.5 1
Total Nitrogen (mg/L)
00 0
in o
d C
CD
"i
0 E
O>
OM 0
d >
"CD
- 0
o a:
^> 00 _
CD
O CD
O o
ol
0 ^
'
O
q
ci
0.2 0.5 1
Total Nitrogen (mg/L)
•* 0
q o
d C
CD
T3
C*} f—
q c
0 E
<
O 0
d >
'•4—'
JD
o 0
White.sucker
Yellow, bull head
Yellow, perch
q _
•<~
_^> oq _
:= d
o
CD
O CD
0 d
Q-
0 •* _
3 °
"o.
CD CN
O o -
0
d
B
°
•
q •
•.
° •
*
• \ •'/".'"
• * '"• ;'''7'"' •
CN P —
d ^~
_ oo 0 >, oq _
d C •= °
CD .Q
_,. T3 CD
^ C ^2 CD
" o =3 0 d
.Q <—
^ °
"CD "o.
.80 Q g _
._. o
° 0
B
.
.
,
"• «
V
.
• •. . •• • •
• 0 •
• ' '•'.'• ."
. 0 P _
d "~
.88 ^oo _
d C := °
CD ^2
T3 CD
. o C ^2 CD
H 3 O d ~
O l_
^ °
"CD "o.
- 0 0 /? ^ -
fy Qj o
- o °
r d
t
'
•
• •
. j
• • •
i '•
" / • * • • *
• «• •
• . ." •
• . ** * °
"•"1.". :"• V«*V' ' '• '
00
d
^r 0
- CD O
d C
CD
"^
^ c
~ • ^
0 ^2
- S 0
0 .>
"CD
. ^ 0
d C£L
o
d
0.2 0.5 1
Total Nitrogen (mg/L)
0.2 0.5 1
Total Nitrogen (mg/L)
Page 5 of 6
0.2 0.5 1
Total Nitrogen (mg/L)
-------
Appendix 11 - Wl Chi a GAM Models Fish
-------
Capture Probability of Fish Taxon Along Chi a Gradient
Black.crapple
Blackchin.shiner
Blacknose.shiner
_g> CD
:= d
CD
O CD
O d
CL
0 •*
Q.
CD CN
O o
0
0
in
. CD
:= d
O o
0 •*
-I—*
Q.
o
CN
d
0
5 £
CD
^ C
P §
o ^>
si
CD
O CD
O d
0
Q.
CD
o
._<«.
$
d
CD CD
Is* O
d C
CD
^ C
d ^
co <
CO 0
d >
'•4—'
CD
2 0
2 5 -\Q
Chlorophyll a (|ig I n)
2 5 1Q
Chlorophyll a (|ig I n)
2 5 10
Chlorophyll a (|ig I n)
Bluegill
=
CD
.Q
0
CL
0
"o.
CD
o
P _
00
d
CD
d
2 -
CN _
0
d
•
•
•'.
. B • B •
*.* . • '
".'.••' '.
• * * * *
. ' • • . '
_k i "• '^- -vt.... •:;.•.
2 5 1_Q
Chlorophyll a (|ig I )
Bluntnose. minnow
2 5 1_Q
Chlorophyll a (|ig I )
Burbot
CD
d
^
o
r^
d
CN
d
^!
d
0
0
(_)
(~
CD
T3
C
^
<
0
>
'•4— '
JD
o:
P _
^
-i—' . —
r^ ^
o
CO
O CD
o d -
0 •* _
^ °
-i— »
Q.
O o
d
B
.
,
• •
a • • '
•
^
• _ ' . "
B •
* • • *
.• . . . .
• '• .'•. ' '
• • • • " » • ' .• • *^ *
d
CN
d
^2
d
_ -^
o
. §
d
- 0
0
O
c
CD
"^
c
^
<
0
>
'"CD
0
a:
P _
^
-i—* . —
1— O
!Q
CD
O CD
0 d
CL
Q) ^"
^ °
-1—*
Q.
tD CN
O o
d
.
•"
<
•. .
'
_ .
- 0
d
- O
d
_ o
- 0
- o
- 0
0
O
c
CD
T3
^
<
0
>
'•4— '
JD
a:
2 5 i_q
Chlorophyll a (|ig I )
Page 1 of 5
-------
Capture Probability of Fish Taxon Along Chi a Gradient
Common.shiner
Golden.shiner
Iowa.darter
>? 00
:= ci
CD
O CD
O d
CL
0 •*
13 °
-i—*
Q.
CD CN
O o
q
ci
2 8
d C
CD
O)
O 0
d >
'•4—'
_ ro
0
_g> oo
:= d
CD
O CO _
O d
Q.
0 •*
3 °
-I—*
Q.
q
d
O) 0
CN O
d C
CD
CN "a
CN b:
in
*- 0
o >
"CD Q.
o "0 CO CN
^> 00
:= ci
CD
O CO
O ci
Ql
0 ^
3 °
O
q
ci
8
d
o O
d C
CD
CN ?
q S
0
-00
'•4^
JD
o 0
2 5 1_Q
Chlorophyll a (|ig I n)
2 5 1_Q
Chlorophyll a (|ig I n)
2 5 10
Chlorophyll a (|ig I n)
Johnny.darter
Largemouth.bass
Logperch
p _
^~
^ 00
:= °
CD
O CD
0 d
Q-
0 •* _
3 °
Q.
CD CN
O o -
0
0
.
.
•
• . .
• • " : •.*"•'
. 0 P _
0 "~
.88 ^oo _
d C := °
CD ^2
_,. T3 CD
. S C ^2 ^ °
CD Q.
.50 CO CN _
do: o °
o
" ° 0
.
•
•
.'•'•* ,'.'.*•• '
• . .t;% ^ T'. •• : . •
. •* p _
0 "~
d C •= °
CD .Q
.- T3 CD
CN CI ^2 CD
O ^ O O
< ^
. t 0 0 ^ _
d > ^ °
CD Q.
.80 CD CN _
do: o °
o
" ° 0
.
•
t •
•
.' •' .:'' ' *
••' ' •' *V '• ' .£. ' ' . • •
0
1 1
0.09 0.11
>undance
- O 0
d >
.sl
o o:
2 5 1_Q
Chlorophyll a (|ig I )
2 5 1_Q
Chlorophyll a (|ig I )
2 5 i_q
Chlorophyll a (|ig I )
Page 2 of 5
-------
Capture Probability of Fish Taxon Along Chi a Gradient
Mimic.shiner
Mottled.sculpln
Mudminnow
& 00
:= ci
CD
O CD
O d
CL
0 •*
13 °
-i—*
Q.
CD CN
O o
q
o
u s
d
CD
>n
0
0
O
CD
co 0
d >
a:
IT)
ci
>> ^r
-!-• ^.
= O
'J3
CD
o cq
O ci
U "
-I—*
Q.
CD T-
O o
q
ci
- 9
d
^ 0
o O
d C
CD
T3
§ I
o 0
d >
-^ °9
= ci
CD
O CD
O ci
0
3
CD Q.
00 ro ^
o: o o
§
8
d
o O
o C
CD
CN ?
q S
0
O 0
o >
'•4— '
JD
o 0
2 5 1_Q
Chlorophyll a (|ig I n)
2 5 1_Q
Chlorophyll a (|ig I n)
2 5 10
Chlorophyll a (|ig I n)
Muskellunge
Northern.pike
Pumpkinseed
p _
^~
^ 00
:= °
CD
O CD
0 d
CL
0 •* _
Q.
CD CN
O c i -
0
0
.
•
*, •
'*» ' ' • '
. 0 P _
0 "~
d C •= °
CD ^2
o ^ O o
< ^
.80 0 ^ _
CD Q.
.80 Q g _
._. o
0 o
.
•
•
* •
. • ' • t
. 0 P _
0 "~
.08 ^oo _
d C :^ °
CD .Q
d ^ O o
< ^
.80 0 •* _
d > ^ °
CD Q.
.80 ro CN _
OQ: o °
._. o
° 0
J
.
. •
t •
• .
• • b . •
"'.**• "•
.. . ,,f.,,,:..V.\i°.t, ,,5. „:••"*,,' . '
- CN
0
1 1
0.15 0.19
>undance
- 5 $
.§1
o a:
\ I
2 5 1_Q
Chlorophyll a (|ig I )
2 5 1_Q
Chlorophyll a (|ig I )
2 5 i_q
Chlorophyll a (|ig I )
Page 3 of 5
-------
Capture Probability of Fish Taxon Along Chi a Gradient
Rock, bass
Shorthead. redhorse
Smallmouth.bass
p _
^* 00
— d
o
CD
O CD
O d
CL
0 •* _
3 °
-i—*
Q.
CD CN
O o -
o
d
o
f "• *" ** ~ .. ^
•
'
* q •
•'•''. • . '
* • '/'!'• '". • I ' .
\ •• •**;••''. '.'•-*».• •••••
, •.!>••. .:-4:-o.»'s- ••• ;
i i i i i i 1 1 1
2 5 1Q
Chlorophyll a (|ig I )
d
^
- CO
d
CO
- CN
d
O)
d
O)
- o
d
- o
0
O
CD
T3
C
.Q
0
>
"CD
0
a:
f*O
1=
O
CD
2
Q.
2
^
-i—*
Q.
CD
O
p _
CO
d
CD
d
•*
d
CN
d
o
d
•
•
'
•
• .
• . .
i i i i i i i i i
2 5 1Q
Chlorophyll a (|ig I )
\m .•
d
^~
d
CO
o
d
- 0
d
CO
- o
d
- o
0
O
CD
T3
C
£1
0
>
"CD
0
a:
CD
d
CD
O CD _
O d
0
CD CN
O o
p
d
,
.
.
• , V-
v .:««:^)<"v-
•
U
•
iil .rut.-* \ •*
•
. oo 3
0 c
CD
T3
CD C
" d 3
^2
- o|
"CD
. ^ "0
0 QX
- 0
2 5 1_Q
Chlorophyll a (|ig I n)
Troutperch
Walleye
White.sucker
p _
"<~
& °p
:= d
o
CD
O CD
0 d
CL
0 •* _
3 °
"o.
CD CN
O o -
0
d
a
o
.
» .-•'
" '— r ,
. ^ P _
d "~
CO 0 >, 2 ~
- co O -£? °
d C =
CD .Q °° _
^ T3 CD °
" S 1 2 d -
^
'•^ -i— •
CD Q. in
- ^ 0 /O °
o Q-; O
^
O
~ O
B
.
••
*
B
"
• ""•
• a
° *
. '
" ' "" • *
" * • . •
,V.:. '."'•* •. .
.••...',?''"•/:". •-' ."•'•'
• *" • ** »"• ""i*"»"""^
h- P _
0 "~
8 >,cq _
C -— CD
- in CD !Q
d "O CD
C ^2 CD
=3 0 d
^ °
'•4— ' -1— '
CD Q.
"0 CD CN
^ D^ O o
- P
d
^ o
d
B
•
*
•
.
.
, °
, 0
.'•'•. ' '" • '
»'':&&'£ &'*':
VJ^
d
0) 0
- CN O
d C
CD
"^
CNI ^
" d 1
.50
0 >
'•4— '
CD
- fe 0
o a:
~ o
2 5 1Q
Chlorophyll a (|ig I )
2 5 1Q
Chlorophyll a (|ig I )
2 510
Chlorophyll a (|ig I )
Page 4 of 5
-------
Capture Probability of Fish Taxon Along Chi a Gradient
Yellow, bullhead
& 00
:= ci
CD
o
O
CL
0
Q.
CD CN
O o
q
o
O) 0
o O
d C
CD
,„ T3
.Q
<
Yellow, perch
& O)
:= d
CD
O 00 _
O d
Q.
0 i^
° ^ 13
'•4—' -I—'
CD Q.
S "0 ro <°.
O
in
d
CD 0
to O
d C
CD
T3
O) ^~
^
.Q
co 0
d >
to ^
^ 0
2 5 1_Q
Chlorophyll a (|ig I n)
2 5 1_Q
Chlorophyll a (|ig I n)
Page 5 of 5
-------
Appendix 12 - TVA TN Optima Models Fish
-------
Scatterplot of Black buffalo against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
n nnnn
O
oo
> ooo
I
oo o
oo c
oooo
•DODO O
>
O <
o o
o o
o
0
o
(
>
0
0.009
0.008
0.007
0.006
0.005
0.004
0.003
0.002
0.001
0.000
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.
TN_uf_mg_L
Scatterplot of Black crappie against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
oG
c
o
4>¥
o
o o
"•%,.
O O
0
o
° 0 g o o
o
o
t*°°<
8
>° 0 0
0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.
TN_uf_mg_L
Scatterplot of Black redhorse against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.0014
0.0012
J> 0.0010
b
£ 0.0008
^
03
m 0.0006
0.0004
0.0002
n nnnn
O
O O
C
O
O
» o
o
o o
00
o e c
O OODO
o
o o
9
O O
0
o
o
(
1
OOOOO O OflQDOOODO O O O
0.6 0.8 1.0
TN_uf_mg_L
-------
Scatterplot of Bluegill against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.04
0.03
0.02
0.01
0.00
0.
0.006
0.005
0.004
0.003
0.002
0.001
0.000
O
0 0
0
_
•o°^
•*•
^^«
o
o
o
' 0
90 00 0 °g^
&> o0° °
} 0 _00 C
°0
0
^ O^D^?W °
3
o
o
t o
>
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.
TN_uf_mg_L
Scatterplot of Bluntnose minnow against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
O
e
oo
o
0
o ^
J&P
o
0
o
'9° °
wtWiteQt
o
0
o
o
0 o
o
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.
TN_uf_mg_L
Scatterplot of Brook silverside against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.010
e o.ooe
0.002
-0.002
O
o
o
o
o
o
B °
O
0° 9j „ o
o
o
0 ° ° °
® O aO
O
5 rf>0 o
1
1
0.6 0.8 1.0
TN_uf_mg_L
-------
Scatterplot of Bullhead minnow against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
05
0
m
_n nm
<
IP o° o
o
0 0
o
0.2
0.3
0.4
0.7
0.8
0.5
TN_uf_mg_L
Scatterplot of Channel catfish against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.0035
0.0030
0.0025
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.
TN_uf_mg_L
Scatterplot of Chestnut lamprey against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.00026
0.00024
0.00022
0.00020
0.00018
0.00016
0.00014
0.00012
0.00010
0.00008
0.00006
0.00004
0
0
O
O
9 0
on o
o o a
oo
0
2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.
TN_uf_mg_L
-------
Scatterplot of Common carp against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
Q_
03
O
_n nnnR
0
o
0
*
o
o
%
|°0»o
) VS
OBOGOiO
O
O
OO
0 0
o
00°
o jS°o
$ 0
$$&
o
0
o
b
9 °
o
o o
ko°° o
% 0
0
>
0.0
0.2
0.4
0.6
0.8 1.0
TN_uf_mg_L
1.2
1.6
Scatterplot of Emerald shiner against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.018
0.016
0.014
0.012
0.010
0.008
0.006
0.004
0.002
0.000
_n nm
i
0 0
O
>
o
o
eP
o
p%8
°*8 «
o
o
o o
£ a
« °
008° <
* 8*
0
o
o
°0
• 0
Po o°ot
8
„» .
O
0
0
o
0.2
0.3
0.4
0.5
0.7
0.8
0.9
1.0
TN_uf_mg_L
Scatterplot of Flathead catfish against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.0018
0.0016
0.0014
0.0012
0.0010
0.0008
0.0006
0.0004
0.0002
0.0000
O
8
o
— o
o a
o o c
OOD
o
aomn
0
0
00
O O (
ooo •
>oo oo
o oo
OB 00
o omm
mmom
o
0
1
too
oo
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.
TN_uf_mg_L
-------
Scatterplot of Freshwater drum against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
E
£
LL
_n nm
O
°d
o
pSftoc
0
a
S°8°
»<8BiP
o
o
>08«
Oo
*ri
o
°°c
S*6
RPoc
o
o
l!So
9> 8
o
0 *
9^v»l
0
»o
D°
)
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.
TN_uf_mg_L
Scatterplot of Golden redhorse against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0
a
o
<
o «
0 0
o
o o
> o o
•OOOAHD
2 0
i
O
O
OD O C
•oaooo
4 0
>
O
0
C
o
o
OO O
6 0
O
0
O
8 1
0 1
2 1.
4 1.
TN_uf_mg_L
-------
Scatterplot of Golden shiner against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.0040
| 0.0025
g 0.0020
2
o
0.0005
0.0000
_n nnnc;
C
0
o
o
OB
» 0°
0 *
o oo
o
o
0
0 0
o o
o oo
1 *V
o 8d>
0 o
0
0°
"%
o> <
0
o
x>
o
o
oo
0
0.0
0.1
0.2
0.3
0.4
0.5 0.6 0.7
TN_uf_mg_L
0.8
0.9
1.0
1.1
Scatterplot of Green sunfish against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.05
0.04
| 0.03
=3
C
CD
S 0.02
0.01
0.00
-0.01
0
O
c
A
Sfl^M
o
o
D°0
>
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.
TN_uf_mg_L
Scatterplot of Largemouth bass against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.030
-0.005
0.6 0.8 1.0
TN_uf_mg_L
-------
Scatterplot of Largescale stoneroller against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
n nnnnR
O
o
O O
o
o
oo
(
o
>
0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55 O.I
TN_uf_mg_L
Scatterplot of Logperch against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.65
0.006
0.005
0.004
0.003
0.002
0.001
0.000
-0.001
0
O
O
o 4
O
O O
(
o
b o
o
o
° o _-
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.
TN_uf_mg_L
Scatterplot of Longear sunfish against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.03
-0.01
0.6
TN_uf_mg_L
-------
Scatterplot of Mississippi silverside against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.040
0.035
„ °-030
T3
'Si
S 0.025
'ui
g 0.020
'
Ot flDOOO OOOttD WOOOSDO O O
0.0
0.2
0.4
0.8 1.0
TN_uf_mg_L
1.2
1.4
1.6
Scatterplot of Quillback against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
03
'5
O
n nnnn
o o
o
o o
0
o
o
o o
o
o
0.5 0.6
TN_uf_mg_L
-------
Scatterplot of Redbreast sunfish against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.035
0.030
0.025
0.020
0.015
0.010
0.005
0.000
O
g
„/&>*
**
o
>00 *
lg°^>
o
L?u
-«s*
?0^s
(
1
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.
TN_uf_mg_L
Scatterplot of Redear sunfish against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.012
;§ 0.008
I
b 0.006
0.004
-0.002
o
00
00
0.0
0.2
0.4
0.8 1.0
TN_uf_mg_L
1.2
1.4
1.6
Scatterplot of River redhorse against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.0012
0.0010
0
| 0.0008
0
> 0.0006
CC
0.0004
0.0002
n nnnn
O
0
OO O
1
o
o o
>
0
o
o
o o
OOOOCDOOO OO ODD O O
0.3 0.4
TN_uf_mg_L
-------
Scatterplot of Rock bass against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.0020
0.0018
0.0016
0.0014
» 0.0012
03
5 0.0010
o
^ 0.0008
0.0006
0.0004
0.0002
0.0000
_n nnn9
O
0 0
0 0
o
o
0
o
o
0
at
°000
0 °
0
<
0 OB 00
0 0
0
0
0.0
0.2
0.4
0.8
1.0
1.2
TN_uf_mg_L
Scatterplot of Sauger against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.0012
0.0010
0.0008
0.0006
0.0004
0.0002
0.0000
0.
O
O
o
o
O 00
o
CD
O
O O
o
o o
o oo
O w
o o o
ooo o o o
o
25 0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65 0."
TN_uf_mg_L
Scatterplot of Silver redhorse against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.00030
0.00025
CD
-------
Scatterplot of Skipjack herring against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.003
0.002
0.001
0.
0.0040
0.0035
0.0030
0.0025
0.0020
0.0015
0.0010
0.0005
0.0000
o
°8 f
°» otf
•
o
o
o
>
o
V
^>°«b <
O
o o
p °
>
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.
TN_uf_mg_L
Scatterplot of Smallmouth buffalo against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0
1 0
o
0 %
2 0
oo
0°00
*%°s
3 0
0
°
>_
°a»a
5 0
0°°0
6 0
O
7 0.
O
0
O
8 0
O
9 1
0 1.
TN_uf_mg_L
-------
Scatterplot of Smallmouth redhorse against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.0010
S 0.0008
0
1
5 0.0006
m 0.0004
0.0002
0.0000
0
0
O
O
O
o
0
o
o
0
o <
c
>
o
0 0.1 0.2 0.3 0.4 0.5 0.6 0.
TN_uf_mg_L
Scatterplot of Snail bullhead against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.00030
0.00025
T3
1 0.00020
.a
| 0.00015
0.00010
0.00005
0.00000
0.
O
O
o
o o
o
o
o
(
>
06 0.08 0.10 0.12 0.14 0.16 0.18 0.20 0.22 0.24 0.26 0.;
TN_uf_mg_L
Scatterplot of Snubnose darter against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.00026
0.00024
0.00022
0.00020
0.00018
0.00016
0.00014
0.00012
0.00010
0.00008
0.00006
0.00004
0
0
(
O
0
o oo
c
0 O O
)
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.
TN_uf_mg_L
-------
Scatterplot of Spotfin shiner against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.020
0.018
0.016
0.014
1 0.012
!c
W)
c 0.010
1
eg- 0.008
0.006
0.004
0.002
0.000
O
i
O
9,
00 ^
0 °
o
o
> 00
00° 0 -
0 °
o
o e
0*0
$> oo o*
O
O
°J
^0°
0
o
0
(
>
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.
TN_uf_mg_L
Scatterplot of Spotted bass against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.012
0.008
0.004
-0.002 L-
0.0
0.2 0.4 0.6 0.8 1.0
TN_uf_mg_L
1.2
1.4
1.6
Scatterplot of Spotted gar against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.0014
0.0012
0.0010
0.0008
0.0006
0.0004
0.0002
0.0000
0
O
C
0
I
OOD
O OO
O
o
o o
>
o
o a
0 0
0
>
V V
CO O O O O
O O OOOO O O
o
o
o
0
o
o
2 0.3 0.4 0.5 0.6 0.7 0.8 0.
TN_uf_mg_L
-------
Scatterplot of Spotted sucker against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.008
0.007
0.006
£ 0.005
1
•a 0.004
1
w 0.003
0.002
0.001
0.000
0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.
TN_uf_mg_L
Scatterplot of Striped bass against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.00045
0.00010
n nnnnR
O
G
O
0 O O O
>
O O O OO OMD CO O
0.0
0.2
0.4
0.8 1.0
TN_uf_mg_L
1.2
1.4
1.6
Scatterplot of Threadfin shad against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.14
0.12
0.10
0.08
0.06
0.04
0.02
0.00
0
O
0
o
K»0<8«
O
o
o
o
o
&!*<£ .
0
0
•On n
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.
TN_uf_mg_L
-------
Scatterplot of Walleye against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
n nnnn
0
0
o a
at t
0
O
>a> o
cocao
o
o o
o o
(
>
0.0
0.2
0.4
0.8 1.0
TN_uf_mg_L
1.2
1.4
1.6
Scatterplot of Warmouth against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.006
0.005
0.004
0.003
0.002
0.001
0.000
<£
°8
o
0
3»xo o&ffi
o
0 o
00
bl>c
o
AM
o
o
80
(
>
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.
TN_uf_mg_L
Scatterplot of White bass against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.016
0.014
0.012
0.010
0.008
0.006
0.004
0.002
0.000
_n nm
O
«^
0
00*
3
0
o«oo6
o
o
<
fldtf^n
> o
K)*004
O
ifioo,
> 0
•
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1
TN_uf_mg_L
-------
Scatterplot of White crappie against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.0016
0.0014
0.0012
0.0010
0.0008
0.0006
0.0004
0.0002
0.0000
0
O
o
o
(
O
O O O
o o
0 00
ooo o a
o oaoanooo o
o
>
o o c
oo
ooo o
o
0
(
>
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.
TN_uf_mg_L
Scatterplot of Whitetail shiner against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.0040
0.0035
0.0030
0.0025
0.0020
0.0015
0.0010
0.0005
0.0000
_n nnnc;
O
G
O
CO
6
o
cP o
0
o
o
o o
oo o
oo
bo°8°
o
o
0
o
0.0
0.1
0.2
0.3 0.4
TN_uf_mg_L
0.5
0.6
0.7
0.8
Scatterplot of Yellow bass against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.014
0.012
0.010
0.008
0.006
0.004
0.002
0.000
O
o
V
o
8
9*8°*%
c
Q
0
ortPo ov_
o
o
o
•o
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.
TN_uf_mg_L
-------
Scatterplot of Yellow bullhead against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.00030
0.00025
05
| 0.00020
o
0.00010
0.00005
n nnnnn
O
O
I
>
o
o
OO
I
>
0.38 0.40 0.42 0.44 0.46 0.48 0.50 0.52 0.54 0.56 0.58 0.60 0.62 0.64 0.66 0.68
TN_uf_mg_L
Scatterplot of Yellow perch against TN_uf_mg_L
FishEFCPUENoYOY_AvgSeasonalWaterChem_optima_20140102 180v*464c
0.0010
0.0008
-C
a
S 0.0006
o
CD
0.0004
0.0002
0.0000
0
O
0
coc
o
0 OD
OD
OO O
O O O OD
O O OO OO
OO OD OBBOO O
O
O
9 0
(
>
00 000 OXOOOOaBOO 00 0 0
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.
TN_uf_mg_L
-------
Appendix 13 - 23 are Microsoft Excel Spreadsheets and
are available upon request from:
James Hagy
NHEERL/Gulf Ecology Division
hagy.jim@epa.gov
-------
Appendix 24 - OHLR Maps.pdf
-------
Appendix 1. Maps showing OHLR and soil chemistry data GIS data layers
Climate Class
Wet
Moist
Dry
Semiarid
Arid
Map of OHLR Climate Class
-------
Seasonality Class
Winter or Fall (w)
Spring (s)
Summer (u)
OHLR Seasonality Class
-------
Aquifer Permeability
High
Moderate
Low
OHLR Aquifer Permeability Class
-------
Terrain
Mountain
Flat
Transitional
OHLR Terrain Class.
-------
Soil Permeability
, .^1 High
Moderate
Low
OHLR Soil Permeability Class.
-------
Legend
OR_C_N_TOT_Lambert_HARN
NtotSO
| 0,100000 - 0.456000
H 0.456001 - 0.698000
^B 0.698001 - 0.964000
^H| 0 964001 - 1.344000
^H 1.344001 -2.387000
Total Nitrogen in 0-50 cm depth category.
-------
Legend
OR_C_N_TOT_Lambert_HARN
CtotSO
| 2.690000-6.120000
| 6.120001 -8.580000
^B 8.580001 - 11.460000
| 11.460001 -18.630000
• 18.630001 -46.770000
Total Carbon in 0-50 cm depth category.
-------
Appendix 25 - Wl Variable Explanations
-------
N
min_ob
X50_th_ob
X95_th_ob
max_ob
Opt_WA
Tol_WA
CDF_50_th_Abund
CDF_95_th_Abund
CDF_50_th_PA
CDF_95_th_PA
CDF_wt_50_th
CDF_wt_95_th
LRM_50_th
LRM_95_th
QLRM_50_th
QLRM_95_th
Opt_qlrm
Tol_qlrm
GAM_50_th
GAM_95_th
ROC
Number of Samples
Minimum Observed Value
Median Observed Value
95th Percentile Observed Value
Maximum Observed Value
Weighted Average Optimum value
Weighted Average Tolerance
Abundance Weighted Cumulative distribution function 50th percentile
Abundance Weighted Cumulative distribution function 5th or 95th percentile
Cumulative distribution function 50th percentile
Cumulative distribution function 5th or 95th percentile
Weighted CDF 50th percentile
Weighted CDF 5th or 95th percentile
Linear logistic regression 50% cumulative probability
Linear logistic regression 5% or 95% cumulative probability
Quadratic logistic regression 50% cumulative probability
Quadratic logistic regression 5% or 95% cumulative probability
Quadratic logistic regression optima
Quadratic logistic regression tolerance
Generalized additive logistic regression 50% cumulative probability
Generalized additive logistic regression 5% or 95% cumulative probability
The area under an ROC (Receiver Operating Characteristic) curve
-------
Methods for Computing Downstream Use
Protection Criteria for Lakes and Reservoirs
Final Report
Prepared for
US EPA National Health and Environmental Effects Research Laboratory
Gulf Ecology Division
1 Sabine Island Drive
Gulf Breeze, FL 32561
Prepared by (alphabetical order)
Diane Allen1, Jon Butcher1'2, Michael J. Paul1, Lei Zheng3, Tan Zi1'2
^etra Tech, Inc., Center for Ecological Sciences
2Tetra Tech, Inc.
1 Park Drive, Suite 200
PO Box 14409
Research Triangle Park, NC 27709
and
3Tetra Tech, Inc., Center for Ecological Sciences
400 Red Brook Boulevard, Suite 200
Owings Mills, MD 21117-5172
September 29, 2015
-------
FOREWORD
Nutrient pollution remains a vexing national pollution problem. Numeric nutrient criteria are one
dimension of a national nutrient reduction strategy. The United States Environmental Protection Agency
(USEPA) nutrient criteria guidance recommends the use of multiple lines of evidence (reference, stressor-
response, mechanistic modeling, and scientific literature) in developing nutrient criteria. This same
guidance also encouraged development of other scientifically defensible approaches, consistent with
water quality standards (WQS) regulations [40 CFR §131.10(b)(2)]. Therefore, refining existing
approaches and developing scientifically defensible alternative numeric nutrient criteria analytical
approaches remain technical challenges and, therefore, ripe research opportunities. This project builds
from a previous ORD sponsored project to identify sensitive aquatic life use targets for lakes. This
project takes those sensitive aquatic life use targets and develops empirical models of nutrient
concentrations needed to protect those targets. It then reviewed downstream protective modeling
approaches and describes a novel method for setting watershed based numeric nutrient values to
protective the receiving lake. Two case studies are used to apply these methods: one for the upper
Midwest and one for a southeastern reservoir.
The concept of using a receiving lake/reservoir approach for setting watershed criteria for influent streams
is a viable, creative, and relatively unexplored approach for numeric criteria setting for streams. The
originally proposed Florida stream criteria1 used a conceptually similar approach to derive downstream
protective values (DPV) for lakes and estuaries, which introduced the concept of DPVs. Developing
receiving water based approaches relies, first, on defining sensitive aquatic life uses (ALU) in receiving
lakes/reservoirs. In-lake targets are then developed to protect nutrient sensitive aquatic life. Identifying
nutrient sensitive aquatic life is non-trivial as the definition of lentic ALUs is not always clear and
sometimes conflicts with other uses. Therefore, defining sensitive ALUs is an important part of
developing DPVs. The previous ORD sponsored research project did that. This project develops in-lake
targets using these sensitive endpoints. It then uses a novel approach using spatial network models to
develop downstream protective values for the contributing watershed to protect the lakes.
federal Register, Vol. 75, No. 16, 4174, January 26, 2010. Water Quality Standards for the State of
Florida's Lakes and Flowing Waters; Proposed Rule.
-------
This project was funded by the US Environmental Protection Agency through Work Assignment 2-12 to
Tetra Tech, Inc. under Contract EP-C-12-060. The EPA Work Assignment Contracting Officer's
Representative was Dr. James Hagy III of the US EPA National Health and Environmental Effects
Research Laboratory, Gulf Ecology Division, in Gulf Breeze, FL. The Tetra Tech Work Assignment
Leader was Dr. Michael Paul of the Tetra Tech, Inc. Center for Ecological Research, Research Triangle
Park, NC.
We wish to thank the National Science Foundation funded North Temperate Lakes Long Term Ecological
Research program and the Tennessee Valley Authority for support in using their data for analysis. This
included both facilitating data retrieval as well as answering many questions about the data. Their
patience and help are greatly appreciated.
The primary authors of this document were Dr. Michael Paul, Dr. Jon Butcher, Diane Allen, Tan Zi (Tetra
Tech, Inc., Research Triangle Park, NC), and Dr. Lei Zheng (Tetra Tech, Inc., Owings Mills, MD).
-------
Downstream Use Protection April 2015
Foreword i
Acknowledgment ii
List of Tables v
List of Figures vii
Executive Summary xii
1 Introduction 1
1.1 Overview 1
1.1.1 Developing sensitive aquatic life use indicators 2
1.1.2 Developing numeric nutrient targets for the receiving water 4
1.1.3 Calculating watershed nutrient levels necessary to protect the lake 4
1.2 Study Areas 4
1.3 Study Data 5
1.3.1 North Temperate Lakes Long Term Ecological Research Program Data 5
1.3.2 Tennessee Valley Authority Long-Term Monitoring Data 10
1.4 Methods Overview 11
2 Nutrient Targets to Protect Sensitive Aquatic Life Uses within Lakes 13
2.1 Detailed Methods 13
2.1.1 Data Preparation 13
2.1.2 Analysis Methods 16
2.2 Results -Wisconsin 16
2.2.1 Nutrients and Chlorophyll a 17
2.2.2 Chlorophyll a and Dissolved Oxygen 23
2.2.3 Biota and Dissolved Oxygen, Chlorophyll a, and Nutrients 33
2.2.4 Nutrient Endpoint Summary 36
2.3 Results - Tennessee 38
2.3.1 Nutrient and Chlorophyll a 38
2.3.2 Chlorophyll a and Dissolved Oxygen 38
2.3.3 Biota and Dissolved Oxygen, Chlorophyll a, and Nutrients 40
2.3.4 Nutrient Endpoint Summary 52
3 Downstream Protection Values: Statistical Analysis 53
3.1 Theoretical Issues 54
3.2 Existing Approaches to DPVs 55
3.2.1 Florida Inland Rule 56
TETRATECH
-------
Downstream Use Protection April 2015
3.2.2 Other States 57
3.3 From Lake Targets to Influent Loading Limits 57
3.3.1 Vollenweider Approach 57
3.3.2 BATHTUB Approach 59
3.3.3 Weighted Distribution to Individual Streams 62
3.4 Translating Influent Loads to Concentrations in the Watershed Network 63
3.4.1 Concentration Data Analyses 63
3.4.2 Simplified Assignment of Load 63
3.4.3 Accounting for Spatial Correlation 64
3.4.4 Generation of Complete Networks 66
3.4.5 Monte Carlo Analysis of Sampling Results 67
3.5 Application to Holcombe Flowage, WI 75
3.5.1 Concentration Data Analysis for Holcombe Flowage 79
3.5.2 Simplified Assignment of Load, Holcombe Flowage 85
3.5.3 Network Spatial Correlation Analysis 88
3.5.4 Sampling Results on Spatially Correlated Network 93
3.6 Application to Douglas Reservoir, TN 101
3.6.1 Concentration Data Analysis for Douglas Reservoir 105
3.6.2 Simplified Assignment of Load, Douglas Reservoir 107
3.6.3 Network Spatial Correlation Analysis 109
3.6.4 Sampling Results on Spatially Correlated Network 114
3.7 Conclusions 120
4 References 121
Appendix A. Methods for Correlated Random Variable Generation 124
TETRATECH
IV
-------
Downstream Use Protection April 2015
Table 1-1. North Temperate Lakes Long Term Ecological Research program data reviewed for
use in this analysis. Shaded cells indicate data selected for use 6
Table 1 -2. Asynchronous benthic macroinvertebrate and water chemistry data 9
Table 1-3 Tennessee Valley Authority program data used in this analysis 10
Table 2-1. Spearman correlations between temperature, nutrients, chlorophyll, and DO. Method
1 defined hypolimnetic depth based on Secchi depth. Method 2 was based on temperature
change. Uf = unfiltered, f = filtered, 26
Table 2-2. Fish species identified as "game fish" for the purposes of this analysis 44
Table 3-1. Concentration Data Analysis for Total Phosphorus, Holcombe Flowage Watershed. 82
Table 3-2. Comparison of Total Phosphorus Data to UCL on Downstream Criterion, Holcombe
Flowage Watershed 84
Table 3-3. Revision to Upstream Targets based on SPARROW Attenuation 85
Table 3-4. Stratified Regression Analysis of Loads at Holcombe Flowage Gages 86
Table 3-5. Comparison of SPARROW Flow-weighted Concentrations to Results of Stratified
Regression Analysis, Holcombe Flowage 88
Table 3-6. Spatial Model Components Tested for the Holcombe Flowage Watershed Model... 88
Table 3-7. Evaluation of Regression Model Evaluation with All the Regression Factors,
Holcombe Flowage 89
Table 3-8. Additional Regression Models Tested, Holcombe Flowage 89
Table 3-9. Spatial Covariance Models Evaluated, Holcombe Flowage 90
Table 3-10 Performance of Spatial Covariance Models, Holcombe Flowage 91
Table 3-11. Exceedance Thresholds (Count) for Different Sample Sizes, Holcombe Flowage.. 96
Table 3-12. TP Concentration Results as a Function of Sample Size and Downstream
Concentration Target, Holcombe Flowage Watershed 98
Table 3-13. Range of Probabilities that a Downstream Target Concentration is Esceeded as a
Function of Sample TP Concentration Mean, Holcombe Flowage 100
Table 3-14. Percent of All TP Observations by Station Greater than 0.090 mg-P/L 105
Table 3-15. Comparison of Total Phosphorus Data to UCL on Downstream Criterion, Douglas
Reservoir Watershed 106
Table 3-16. Comparison of SPARROW Flow-weighted Concentrations to Results of Stratified
Regression Analysis, Douglas Reservoir 109
Table 3-17. Spatial Model Components Tested for the Douglas Reservoir Watershed Model. 109
Table 3-18. Evaluation of Regression Models for Douglas Reservoir with all the Regression
Factors 110
TETRATECH
-------
Downstream Use Protection April 2015
Table 3-19. Selected Predictive Model for Douglas Reservoir Ill
Table 3-20. Spatial Covariance Models Evaluated, Douglas Reservoir Ill
Table 3-21. Performance of Spatial Covariance Models, Douglas Reservoir 112
Table 3-22. Exceedance Thresholds (Count) for Different Sample Sizes, Douglas Reservoir .116
Table 3-23. TP Concentrations for Different Sample Sizes, Douglas Reservoir Watershed .... 117
Table 3-24. Exceedance Probability for Different Sample TP Concentrations, Douglas Reservoir
119
TETRATECH
VI
-------
Downstream Use Protection April 2015
Figure 1-1. Schematic of central research question 1
Figure 1 -2. Example nutrient optima derivation plot to identify indicator values 3
Figure 1-3. Zooplankton nutrient optima derivation plots 3
Figure 1-4. TP inference model based on WI fish data 3
Figure 1-5. Simple conceptual model of nutrient effects on aquatic life in the study systems. ... 11
Figure 2-1. Lake temperature profiles from the same sampling station in one year in April (left),
August (center), and October (right). The thermocline is represented as a dashed line. The
proportion of the water column for which dissolved oxygen (DO) was less than 4 mg/L,
effectively reducing habitat size for intolerant fish species, was lowest in the middle panel
(August). The relative thermal resistance, a unitless index of stratification strength, is
printed in the upper right corner 14
Figure 2-2. Box and whisker plots offish species richness and Shannon-Wiener diversity scores
for fish samples from mainstem (Main) and tributary (Trib) TVA reservoirs. Reservoirs on
the main stem of the river had higher fish biodiversity index scores than tributary storage
reservoirs 16
Figure 2-3. Intra- and inter-annual chlorophyll a concentration (ng/L) variation in WI lakes. Both
mean and 1 standard deviation by year are shown 17
Figure 2-4. Intra- and inter-lake variations of chlorophyll a concentrations (ng/L) in WI lakes.
Both mean and 1 standard deviation by lake are shown 18
Figure 2-5. Inter- and intra-annual total nitrogen (TN, mg/L) and total phosphorus (TP, mg/L)
variation. Both mean and 1 standard deviation by year are shown 19
Figure 2-6. TP (mg/L) variations within and among WI lakes. Both mean and 1 standard
deviation by lake are shown 20
Figure 2-7. TN (mg/L) variations within and among WI lakes. Both mean and 1 standard
deviation by lake are shown 21
Figure 2-8. Relationships between paired nutrient and chlorophyll a concentrations in the large
WI lake dataset. Uf = unfiltered, spec = spectrophotometric, fluor = fluorometric. All
regressions are significant (p<0.05) 22
Figure 2-9. Epilimnetic DO and temperature fluctuation during 2012 23
Figure 2-10. DO and temperature variations at different depths of Big Muskellunge in 1989 (top)
and 2001 (bottom) 24
Figure 2-11. Vertical chlorophyll a fluctuation at Big Muskellunge Lake during different seasons
in 1989 (left) and 2001 (right) 25
Figure 2-12. Relationships between chlorophyll a and dissolved oxygen (DO) concentrations in
all instantaneous samples expressed with a loess fit to raw paired data (a) and a logistic
regression model fit to binned data (b). The blue fitted line is the 95th quantile loess fit, the
TETRATECH
VII
-------
Downstream Use Protection April 2015
vertical line is the change point, and the gray shaded area is the 90% confidence interval
(CI) 26
Figure 2-13. Long-term mean chlorophyll a concentrations and surface epilimnetic DO at each
lake (each dot represents a lake) 27
Figure 2-14. Water column maximum (upper panel) and mean (lower panel) chlorophyll a
concentrations vs. lowest dissolved oxygen concentrations at all locations during all seasons.
Logistic regression models (right side panel) were also fit to the DO data in the left side
converting DO data to greater or less than 2 mg/L. The probability of hypoxia (hypolimnion
DO< 2mg/L) increased dramatically with elevated chlorophyll a concentrations. Blue lines
in left side panel are loess fits. Values above the left side panels are the change point (center
value), and the 95th percentile confidence interval around the change point (left and right
values) 28
Figure 2-15. Proportion of Hypoxic samples (DO <2 mg/L) in the entire water column and
chlorophyll a concentration in different seasons. Colored lines are loess fits 29
Figure 2-16. Hypolimnetic DO vs. water column chlorophyll a. Change point (red line) occurred
at less than 5 |ig/L. Values above the figure is the change point (center value), and the 95th
percentile confidence interval around the change point (left and right values) 30
Figure 2-17. Proportion of hypolimnetic hypoxia (DO <2 mg/L) vs. temperature, chlorophyll a,
and nutrients in the summer. Red lines are loess fits 31
Figure 2-18. Multiple regression models showing the minimum DO concentration in the
hypolimnion and its relationships with mean (a) and maximum (b) chlorophyll a and
temperature concentrations. The solid lines are predicted minimum DO concentrations 32
Figure 2-19. Multiple regression models showing the proportion of hypoxia samples in the
hypolimnion and its relationships with mean (a) and maximum (b) chlorophyll a and
temperature concentrations. The solid lines are predicted proportion hypoxia. Both
temperature and chlorophyll a are significant predictors 32
Figure 2-20. Summer lake mean chlorophyll a and hypolimnetic DO concentrations as raw data
(left) and as binomials with DO > 2 mg/L as the response variable. Each data point at left
graph represents one lake-year. The gray areas are potential change points. The figure on the
right is a logistic regression of the probability of DO > 2mg/L in the hypolimnion as a
function of chlorophyll a 33
Figure 2-21. Relationship between invertebrate taxa richness in Hester_Dendy samples and a)
minimum DO proportion in a water column; 2).annual minimum DO in water column 34
Figure 2-22. Zooplankton responses to nutrients and DO concentrations 35
Figure 2-23. Annual mean and summer mean chlorophyll a vs. TP relationships. Each data point
represent mean value 37
Figure 2-24. Annual mean and summer mean chlorophyll a vs. TN relationships 37
Figure 2-25. Average summer chlorophyll a values in response to nutrient concentrations
averaged across spring and summer. Both linear relationships (green lines) are significant
(p-value < 0.01). R-squared values for the linear models are in the top left corner 38
TETRATECH
VIM
-------
Downstream Use Protection April 2015
Figure 2-26. A. Proportion of DO samples > 5mg/L in the hypolimnion in response to
epilimnetic chlorophyll a averaged across spring and summer. B. Proportion of DO samples
> 2mg/L in the hypolimnion. Both logistic regression models (red) were statistically
significant (p < 0.01). Mean values for mesotrophic (4.7 |ig/L), eutrophic (14.3 |ig/L), and
upper eutrophic (25 |ig/L) lake chlorophyll a values are indicated with dashed grey lines
(EPA 2000, OECD 1982) 39
Figure 2-27. A. Proportion of DO samples > 5mg/L in the water column in response to
epilimnetic chlorophyll a averaged across spring and summer. B. Proportion of DO samples
> 5mg/L in the epilimnion. Both logistic regression models were statistically significant (p-
value < 0.01). Very few lake profiles had DO concentration in the epilimnion < 5mg/L
below a chlorophyll a concentration of 4.7 |ig/L 40
Figure 2-28. Lake temperature profiles from the same sampling station in one year in April (left),
August (middle), and October (right). The thermocline is represented as a dashed line. The
proportion of the water column for which dissolved oxygen (DO) is less than 5 mg/L,
effectively reducing habitat size for intolerant fish species, is highest in the middle panel
(August). The relative thermal resistance, a unitless index of stratification strength, is
printed in the upper right corner 41
Figure 2-29. Reservoirs on the main stem of the river had higher fish biodiversity indices than
tributary storage reservoirs 41
Figure 2-30. Fish diversity indices - species richness (left) and evenness (right) - are expressed as
a function of A. minimum August DO values, B. proportion of water column samples with
DO > 5 mg/L and C, proportion of hypolimnion samples with DO > 5 mg/L 43
Figure 2-31. Game fish diversity indices - species richness (left) and evenness (right) - as a
function of A. minimum August DO values, B. proportion of water column samples with
DO > 5 mg/L and C., proportion of hypolimnion samples with DO > 5 mg/L. Evenness
increased with higher minimum DO values, an increasing proportion DO > 5 mg/L in the
hypolimnion and the water column (p < 0.05). No other relationships were statistically
significant 45
Figure 2-32. Bluegill (blue) and largemouth bass (brown) relative abundances in the fall as a
function of minimum August DO (top) and DO concentration (proportion of samples with
DO > 5 mg/L) in the hypolimnion (middle) and in the entire water column (bottom).
Bluegill abundance declined with increasing DO, whereas largemouth bass abundance
increases with increasing DO. All relationships were statistically significant (p < 0.05) 47
Figure 2-33. Fish species richness increases with increasing spring-summer chlorophyll a
concentration. There was no apparent relationship between chlorophyll a concentration and
species evenness 48
Figure 2-34. When the analysis was narrowed to game fish alone, species richness increased with
increasing chlorophyll a, but species evenness declined. Statistically this means that
although there might have been more species present, additional fish caught were more
likely to be of the same species, as there were larger numbers of fewer species 48
Figure 2-35. Bluegill and largemouth bass abundances both increased with increasing
chlorophyll a concentrations 49
TETRATECH
IX
-------
Downstream Use Protection April 2015
Figure 2-36. Fish species richness (left) increased with increasing total phosphorus (TP) and total
nitrogen (TN) concentrations (p < 0.05). There was no statistically significant relationship
between species evenness (right) and nutrient concentrations (p<0.05) 50
Figure 2-37. Among game fish alone, species richness (left) had no relationship with TP or TN
concentrations, whereas species evenness (right) declined with increasing TN (p < 0.05), but
notTP 51
Figure 2-38. Bluegill and largemouth bass abundances increased with increasing nutrient
concentrations, with the exception of bluegill and TP. Solid lines indicate statistically
significant relationships (p < 0.05) 52
Figure 3-1. TP concentration distribution of upstream stream network of Douglas reservoir with
different sample sizes and downstream concentration (DC, mg/L) values 68
Figure 3-2. Kolmogorov-Smirnov test statistics for upstream TP concentration as a function of
sample size (Douglas Reservoir, TN example) 69
Figure 3-3. Kolmogorov-Smirnov test statistics for upstream TP concentration as a function of
sample size (Holcombe Flowage, WI example) 70
Figure 3-4. Cumulative Distribution Function (CDF) of downstream endpoint TP concentration
using fitted log-normal distribution 71
Figure 3-5 the number of sites which TP concentration is higher than the criterion TP
concentration (Douglas reservoir) 72
Figure 3-6. Generic 95% confidence interval range on the binomial mean as a function of
sample size when the number of observed exceedances equals 1 73
Figure 3-7. Fitted log-normal distributions of mean sample TP concentration as a function of
downstream concentration 74
Figure 3-8. Schematic plot of the connections between sampled TP concentration and
downstream endpoint concentration 75
Figure 3-9. Watershed of Holcombe Flowage, WI 76
Figure 3-10. Land Use in Holcombe Flowage (WI) Watershed 77
Figure 3-11. WWTP in the Downstream Portion of Holcombe Flowage (WI) Watershed 78
Figure 3-12. Sampling Stations in Holcombe Flowage (WI) Watershed 79
Figure 3-13. Distribution of TP Concentrations after Achieving Flow-Weighted Concentration
of 0.030 mg/L, Chippewa River 87
Figure 3-14. Distribution of TP Concentrations after Achieving Flow-Weighted Concentration
of 0.030 mg/L, Flambeau River 87
Figure 3-15. TP Concentrations Estimated by Spatial Regression Model, Holcombe Flowage. 92
Figure 3-16 Histogram of Predicted TP Concentrations, Holcombe Flowage 93
Figure 3-17. PDFs of Reference and Critical Stream TP Concentration Scenarios at the
Downstream Pour Point, Holcombe Flowage 94
TETRATECH
-------
Downstream Use Protection April 2015
Figure 3-18. Exceedance Ratio and One-sided 95% Confidence Interval for Reference Scenario,
Holcombe Flowage 95
Figure 3-19. Exceedance Ratio and One-sided 95% Confidence Interval for Critical Scenario,
Holcombe Flowage 96
Figure 3-20. Exceedance Probability for Different Sample TP Concentrations, Holcombe
Flowage 100
Figure 3-21. Watershed of Douglas Reservoir, TN 102
Figure 3-22. Land Use in Douglas Reservoir (TN) Watershed 103
Figure 3-23. Sampling Stations in Douglas Reservoir (TN) Watershed 104
Figure 3-24. Distribution of TP Concentrations after Achieving Flow-Weighted Concentration
of 0.090 mg/L, French Broad River (TN) 108
Figure 3-25. Distribution of TP Concentrations after Achieving Flow-Weighted Concentration
of 0.090 mg/L, Nolichucky River (TN) 109
Figure 3-26. TP Concentrations Estimated by Spatial Regression Model, Douglas Reservoir. 113
Figure 3-27 Histogram of Predicted TP Concentrations, Douglas Reservoir 114
Figure 3-28. PDFs of Reference and Critical Stream TP Concentration Scenarios at the
Downstream Pour Point, Douglas Reservoir 115
Figure 3-29. Exceedance Ratio and One-sided 95% Confidence Interval for Reference Scenario,
Douglas Reservoir 115
Figure 3-30. Exceedance Ratio and One-sided 95% Confidence Interval for Critical Scenario,
Douglas Reservoir 116
Figure 3-31. Exceedance Probability for Different Sample TP Concentrations, Douglas
Reservoir 119
TETRATECH
XI
-------
Downstream Use Protection
April 2015
EXECUTIVE SUMMARY
Numeric nutrient criteria are being developed to protect the designated uses, including aquatic life uses,
of waterbodies nationwide. Developing numeric nutrient criteria that protect aquatic life is a scientific
challenge because nutrients are natural components of healthy ecosystems in natural levels, the effects of
nutrient pollution vary depending on many covariables and confounding stressors, and the measures that
states apply to protect aquatic life vary state by state.
Traditional analytical approaches for developing numeric criteria include reference distributions, stressor-
response relationship modeling, and mechanistic modeling, with scientific literature frequently used to
support these approaches. Derivation of stream criteria using reference approaches has been challenged
on the grounds of over-protection while derivation using stressor-response modeling has been challenging
due to the wide range of co-variates that affect stream responses (e.g., flow, substrate, light, grazing) and
the panoply of co-occurring stressors (e.g., sediment). For this reason, developing stream numeric criteria
has been difficult.
Water quality standards regulations require that criteria "ensure... .attainment and maintenance of the
water quality standards of downstream waters" [40 CFR 131.10(b)]. This requires that any stream criteria
in addition to protecting in-stream uses also be required to protect receiving water uses downstream.
Stressor-response models in such receiving waters, such as lakes, are frequently more precise because
there are fewer confounding variables affecting nutrient response in open water systems with longer
residence times. As a result, the central research question of this research is identifying concentrations
that need to be met in tributary streams in order to assure protection of sensitive aquatic life uses in
receiving waters. Such research is not without precedent, and the development of downstream protective
values was pursued in the United States Environmental Protection Agency (USEPA) proposed Florida
numeric criteria effort (Federal Register, Vol.
75, No. 16, 4174, January 26, 2010). This
research effort is a multi-step process: 1)
develop sensitive aquatic life use indicators for
receiving waters; 2) develop numeric nutrient
criteria for the receiving water to protect
sensitive aquatic life; 3) calculate watershed
loads and stream concentrations necessary to
protect the lake. The first step of this process
was addressed in previously sponsored ORD
research (Paul et al. 2014). This report
addresses steps 2 and 3 in this process and
consists of nutrient target development to
protect sensitive aquatic life uses in two lakes
- Holcombe Flowage in Wisconsin and
Douglas Reservoir in Tennessee and
downstream protective value development.
Streams
Central Research Question:
What concentrations need to be met
here...
...to protect uses here?
Study Area
The study areas selected for this work were Wisconsin and Tennessee. Wisconsin was selected because it
represented a region rich in natural temperate cool, typically dimictic lakes. It was also selected because
of the presence of the North Temperate Lakes Long Term Ecological Research Program (NTL LTER,
lter.limnology.wisc.edu) site in northern Wisconsin, with a long history of natural lake data collection,
including a wide range of biological responses. Tennessee was chosen because it represented a region
with subtropical/temperate, typically warm monomictic reservoirs that provided a contrast in lentic
waterbody type as well as aquatic life. While sport fishing in both regions is valued, there is a natural and
I TETRATECH
XII
-------
Downstream Use Protection
April 2015
expected fauna for cool, dimictic lakes that defines a natural aquatic life community, even for an
impoundment in Wisconsin. In contrast, warm monomictic southeastern reservoirs are unnatural systems
placed on riverine landscapes with no expected lentic fauna. Aquatic life uses for such systems do not
have a natural expectation and, therefore, recreational or sports fisheries pose a realistic and defensible
endpoint for management.
For the DPV section, two specific lentic waterbodies were chosen: Holcombe Flowage in Wisconsin and
Douglas Reservoir in Tennessee. These waterbodies were chosen because of their location relative to the
regions for which we had sufficient data to complete the nutrient target development portion of this work
(Chapter 2) and because each had a substantial amount of watershed sample data. The Holcombe Flowage
is in northeastern Wisconsin on the Chippewa River. It supports a diverse fishery and recreation and
drains a sizeable watershed. Douglas Reservoir is in eastern Tennessee on the French Broad River. It
supports a constrained warmwater fishery and according to TV A has scored in the poor to low end of the
fair range for ecological health since 1994).
Nutrient Targets to Protect Sensitive Aquatic Life Uses within Lakes
Nutrient
Phytoplankton
For deriving nutrient targets for sensitive uses,
we used a standard stressor-response based
approach using a simple conceptual model. In
this model, nutrients stimulate phytoplankton
productivity and increase phytoplankton
biomass (estimated as chlorophyll a). This
biomass affects oxygen directly via
photosynthesis and respiration by primary
producers, and indirectly via microbial
respiration of detrital algal biomass. The
biomass also influences biota directly by
providing food/energy through bottom-up
control, but also via mediated changes in food
quality. Lastly, oxygen availability has a direct
effect on biota. We modeled nutrient-
chlorophyll, chlorophyll-dissolved oxygen,
chlorophyll-biota, and oxygen-biota relationships. We also explored models of the nutrient-biota
relationship, which treats the intermediate steps implicitly. Using targets for biota or oxygen, we
identified chlorophyll endpoints and then nutrient targets needed to meet those chlorophyll endpoints.
Wisconsin
In Wisconsin, nutrients related to chlorophyll and chlorophyll to dissolved oxygen, especially hypoxia.
Dissolved oxygen and chlorophyll, however, were only loosely related to most direct biotic measures in
Biota
Simple conceptual model of nutrient effects on
aquatic life in the study systems.
TETRATECH
XIII
-------
Downstream Use Protection
April 2015
these lakes, except zooplankton. We believe
these were due to sample size (phytoplankton)
and methodological (benthic invertebrates)
reasons. Zooplankton exhibited the strongest
responses, but none that provided a directly
extractable association with DO that could be
linked to chlorophyll.
As a result, for deriving nutrient targets, we
relied mostly on the relationships of chlorophyll
to DO and nutrients to chlorophyll. DO values
less than 2 mg/L represent hypoxia and values
less than 5 mg/L are stressful to many fish
species and are the default dissolved oxygen
criteria in WI [WI NR 102.04 (4) (a)]. The
analyses above indicate that instantaneous
maximum chlorophyll above 10-20 ug/L, and
average chlorophyll concentrations of
approximately 5 ug/L are associated with
increased risk of observing increase
All observed samples
A
o
Cl
Vi
^5-.—.—.- -_-.
nll
10
n n n
50 100
n i n
5001000
Chlorophyll a (ug I )
Mean Chi a '/s.hypolimnion DO
|
10
Chlorophyll a big I )
hypolimnetic hypoxia (< 2 mg/L) and epilimnetic and water
column means less than 5 mg/L DO in these cool dimictic
lakes. Mean chlorophyll a concentrations of 5 ug/L are
generally associated with TP concentrations in the range of
20-30 ug/L and these are consistent with TP predictions using
global chlorophyll-TP equations for lakes (OECD 1982). In
addition, change points in annual mean and summer mean TP
and TN relationships against chlorophyll occurred at 20 ug/L
TP and 500 ug/L TN respectively. Therefore, a target TP value
of 20-30 ug/L for the Holcombe Flowage reservoir would be
defensible. Lastly, WI actually has a TP criterion for
reservoirs, such as Holcombe Flowage, and that value is 30
ug/L. Again, such a concentration would appear to be
protective of sensitive aquatic life based on this independent
analysis.
Tennessee
In Tennessee, fish species did change in response to nutrient
enrichment, chlorophyll increases, and changes in DO. Of
greatest focus, was the effect of DO. We did not observe a
distinct threshold in fish to 5 mg/L, however, richness
declined as the proportion of the water column and
hypolimnetic oxygen declined below 5 mg/L. DO
concentrations above 5 mg/L are important for fish species,
even in warm monomictic lakes. A summer, littoral DO
value of > 5 mg/L represents an optimal level for bluegill
(Stuber et al. 1982b) and the level below which largemouth
bass show distress (Stuber et al. 1982b). A substantially
•*
d
• • R*= 0.03
0.0 0.2 0.4 0.6 0.8 1.0
Proportion water column DO > 5 mg/L
TETRATECH
XIV
-------
Downstream Use Protection
April 2015
higher average value, 8 mg/L, is preferred for normal largemouth bass growth. Coincidentally, the
Tennessee state water quality criterion states that "DO shall not be less than 5 mg/L" [Tennessee Rules
Chapter 1200-04-03-.03 (3)(a)], therefore we focused on this DO endpoint in deriving TP and TN targets.
Chlorophyll a concentrations consistent with meeting
a DO value of 5 mg/L (> 40% of the water column,
greater than 50% of the hypolimnion, and greater than
95% of the epilimnion), were approximately 14 ug/L,
which is also the mean annual chlorophyll a value for
eutrophic lakes (OECD 1982). Epilimnetic samples
showed a threshold effect whereby there was a
decreasing proportion of samples with DO
concentrations > 5mg/L above a chlorophyll a
concentration of approximately 4.7 Lig/L; and this
proportion declined below 95% above 14 ug/L. For
the hypolimnion, the preferred summer habitat for
some game species that are stressed by higher
epilimnetic summer temperatures, the model
predicted, at a target chlorophyll a value of 14 ug/L,
just over 50% of the hypolimnion will maintain a DO
value > 5 mg/L. Furthermore, at the same target
chlorophyll a value of 14 Lig/L, about 70% of the
hypolimnion will maintain DO values > 2mg/L
(Figure 2-26). This would suggest sufficient habitat
protection of hypolimnetic habitat if average
chlorophyll values are maintained at 14 ug/L for
Douglas Reservoir.
We next converted this average summer growing
season chlorophyll a endpoint of 14 ug/L into
nutrient targets. The nutrient-chlorophyll models for
this dataset were similar in precision to other
chlorophyll yield models (OECD 1982). The
phosphorus model developed predicts a TP
concentration of approximately 0.04 mg/L TP to
meet the 14 ug/L chlorophyll a endpoint. The same
chlorophyll endpoint was associated with a target
TN concentration of 0.71 mg/L. We believe
meeting these average nutrient concentrations will
result in chlorophyll levels that will prevent DO
concentrations that threaten fish, especially
important game species.
A. Proportion of hypolimnion with DO > 5 mg/L
0.5 1.0 1.5
Log average chlorophyll a concentration (Mg/L)
A. Proportion of water column with DO > 5 mg/L
Meso Eu Upper Eu
s
-------
Downstream Use Protection April 2015
procedures using methods such as LOADEST. Next, methods for accounting for spatial correlation in
watershed nutrient concentrations were applied and were followed by methods to generate a complete
network using the covariance structure developed with the spatial correlation analysis. Finally, a Monte
Carlo simulation approach was used to simulate a complete nutrient dataset for a watershed to test various
sampling strategies and exceedance profiles to test against downstream targets.
Results presented in this section address only two example watersheds and the analysis of Monte Carlo
simulation analyses of those two watersheds is open to further data exploration. Nonetheless, it is clear
that the problem of assessing instream concentrations relative to a downstream receiving water target is
amenable to a statistical analysis that evaluates whether or not sample means observed upstream in the
watershed are consistent with achieving the downstream target.
When land uses and other sources of nutrient loading are relatively homogenous and distributed evenly
throughout the watershed an analysis of evidence from watershed sampling sites could be made based
solely on the measured or estimated lognormal distribution CDF at the downstream pour point. This
approach breaks down when there are different types and sources of loads in the watershed, in which case
measurements at different observation sites are likely to exhibit strong spatial correlation. This situation
can be addressed through the development of spatial covariance representations (SSN/STARS) coupled
with either a regression analysis of loads based on landscape features or a simple mean representation for
homogenous source distributions. The regression analyses reported herein met with only moderate
success, but could likely be improved through better accounting of point source discharges in particular.
Taking into account stream network-based autocorrelation enables estimation of the expected value of
concentrations throughout the stream network. This distribution is dependent on the concentration and
load present at the downstream pour point. A relatively straightforward solution to the assessment
problem can be achieved by estimating a lower confidence limit on the distribution across all watershed
segments, based on the assumption that the downstream is at relatively undisturbed reference
concentration levels, and an upper confidence limit calculated under the assumption that the target
concentration is just met. This leads
to a three-tiered approach: Site
concentrations below the lower c
confidence limit can be deemed
"safe" in that they do not present ^ "
statistically significant results that the |
downstream pour point concentration | "
will exceed targets. In contrast, sites J „
where the flow-weighted mean
concentration is above the upper ' £
confidence limit are not consistent
with the distribution of site loads that
would be expected to achieve the
downstream criterion. SampledTP •a™"**™ *™«
We propose methods for determining the appropriate confidence limits when network-based spatial
correlation is present. This requires estimation of the spatial correlation structure and underlying
regression on watershed characteristics (if appropriate), from which an empirical distribution of the
relationship between the desired downstream pour point concentration and concentrations at sites
distributed throughout the watershed can be obtained. Additional research is needed to determine simpler
methods for evaluating appropriate adjustment factors to relate the distribution of the downstream pour
point concentration to concentrations in individual segments throughout the watershed.
| TETRATECH
xvi
-------
Downstream Use Protection April 2015
(This page intentionally left blank)
TETRATECH
XVII
-------
Downstream Use Protection
April 2015
1 INTRODUCTION
1.1 Overview
Numeric nutrient criteria are being developed to protect the designated uses, including aquatic life uses,
of waterbodies nationwide. Developing numeric nutrient criteria that protect aquatic life is a scientific
challenge because nutrients are natural components of healthy ecosystems in natural levels, the effects of
nutrient pollution vary depending on many covariables and confounding stressors, and the measures that
states apply to protect aquatic life vary state by state.
The United States Environmental Protection Agency (USEPA) nutrient criteria guidance (USEPA 2000)
recommends the use of various approaches for developing numeric criteria including reference
distributions, stressor-response relationship modeling, and mechanistic modeling, with scientific literature
frequently used to support these approaches. Derivation of stream criteria using reference approaches has
been challenged on the grounds of over-protection while derivation using stressor-response modeling has
been challenging due to the wide range of co-variates that affect stream responses (e.g., flow, substrate,
light, grazing) and the panoply of co-occurring stressors (e.g., sediment). For this reason, developing
stream numeric criteria has been difficult. Beyond these standard approaches, USEPA guidance,
including the lakes guidance (USEPA 2000), offered additional approaches and encouraged development
of other scientifically defensible approaches, consistent with water quality standards (WQS) regulations
[40 CFR §131.10(b)(2)]. Therefore, developing scientifically defensible numeric nutrient criteria
analytical approaches remains a technical challenge and, therefore, a research opportunity.
Water quality standards regulations also require that criteria "ensure... .attainment and maintenance of the
water quality standards of downstream waters" [40 CFR 131.10(b)]. This requires that any stream criteria
in addition to protecting in-stream uses also be required to protect receiving water uses downstream.
Stressor-response models in such receiving
waters, such as lakes, are frequently more
precise than those for streams because there
are fewer confounding variables affecting
nutrient response in open water systems with
longer residence times. As a result, the focus
of this research is identifying concentrations
that need to be met in tributary streams in
order to assure protection of sensitive aquatic
life uses in receiving waters (Figure 1-1-1).
The concept of using a receiving
lake/reservoir approach for setting watershed
criteria for influent streams is viable,
creative, and relatively unexplored. It has the
added benefit of meeting two criteria
requirements - protecting uses and
downstream receiving waters. The stream
criteria proposed by USEPA for Florida2
were developed using conceptually similar analyses to derive downstream protective values, which
introduced the concept; this research furthers that approach.
Streams
Central Research Question:
What concentrations need to be met
'here...
...to protect uses here?
Figure 1-1. Schematic of central research question.
2 Federal Register, Vol. 75, No. 16, 4174, January 26, 2010. Water Quality Standards for the State of
Florida's Lakes and Flowing Waters; Proposed Rule.
-------
Downstream Use Protection April 2015
Developing receiving water based approaches is a multi-step process: 1) develop sensitive aquatic life use
indicators for receiving waters; 2) develop numeric nutrient targets for the receiving water to protect
sensitive aquatic life; and 3) calculate watershed loads and stream concentrations necessary to protect the
lake. The first step in this process was completed under a previous Task Order (Task Order 0005 under
Contract EP-C-11-037, Paul et al. 2014), the second two steps are the focus of this Report. In the
following section, we briefly review the work to develop sensitive aquatic life use indicators.
1.1.1 Developing sensitive aquatic life use indicators
The first step relies on defining sensitive ALUs in receiving lakes/reservoirs. This is non-trivial as the
definition of lentic ALUs is not always clear and sometimes conflicts with other uses. Another problem
is that development of lake bioassessment tools lags behind streams (USEPA 1998, USEPA 2009);
therefore, defining discrete, direct targets has been difficult and has relied, principally, on the trophic state
concept (Carlson 1977). Also, management of many lakes and reservoirs focuses on productive sport
fisheries, resulting in active fertilization programs (e.g., VWRRC 2005). Such goals may produce algal
targets that pose a trade-off between competing uses like fishing (low clarity, high production) and
swimming (low production, high clarity). But, criteria must protect the most sensitive use [40 CFR
131.1 l(a)(l)]. Therefore, defining sensitive ALUs was an important part of this research.
Under the previous effort (Task Order 0005 under Contract EP-C-11-037, Paul et al. 2014), a review of
state water quality standards and consolidated assessment and listing methodologies was conducted to
characterize the measures currently being used to assess aquatic life use in lakes in Midwest (WI, MN,
SD, IA, IL, MO) and Southeast (KY, TN, NC, SC, MS, AL, GA, FL) states, the spatial focus of this
research. The results of this review were surprising. Most of these states evaluated aquatic life use in
lakes and reservoirs using indirect measures (chlorophyll a, Secchi depth clarity, and dissolved oxygen)
rather than direct biological measures, as frequently occurs in assessing stream aquatic life uses. Of these
states, only Florida had a direct use measure (macrophytes). However, several states either supplemented
water chemical information with biological information (e.g., IA used macrophyte information in their
assessment; IL used percent macrophyte cover as part of an aquatic life use index along with chlorophyll
a, clarity, and sediment) or they were developing direct biological indices (e.g., WI was developing a
macrophyte index; MN was collaborating with the National Lakes Assessment to pursue a lake fish
index). We chose to focus on common indirect assessment endpoints (chlorophyll a and dissolved
oxygen), given the frequency of their application, but also to explore biological assessment endpoints
(phytoplankton, benthic invertebrates, zooplankton, and fish in WI; fish in TN) because we were able to
identify datasets with available paired water quality and biological data from lakes to use for that purpose.
The US National Science Foundation Long-Term Ecological Research (LTER) program, Northern
Temperate Lakes (NTL) site in Wisconsin proved invaluable in housing a wealth of paired water quality
and biological assemblage data for a range of lakes in Wisconsin that varied in nutrient concentrations.
The core long-term monitoring data and the more than sixteen associated research studies provided data
on phytoplankton, zooplankton, macroinvertebrates, and fish for a variety of lakes in Wisconsin. The
data were available online and LTER staff facilitated its assembly into a useful database of water
chemical and biological response data.
Similarly, the Tennessee Valley Authority (TVA) manages more than 30 reservoirs in the southeastern
US and has been monitoring water chemistry and fish assemblages in their lakes for decades. These data
were also made available for use in comparing responses in Midwestern lakes to those in southeastern
reservoirs.
-------
Downstream Use Protection
April 2015
An indicator value approach (Yuan 2006) was used to
identify sensitive aquatic life along gradients in nutrient
conditions across lakes and over time. Nutrient optima
were calculated for different taxa within each
assemblage group and used to evaluate the range in
sensitivities to nutrients (Figure 1-2). Some taxa
combination models were also explored for their utility
in deriving nutrient sensitive indicators. Optima were
converted into ranks from 2 (nutrient sensitive) to 5
(nutrient insensitive).
In the Wisconsin data, phytoplankton taxa were
relatively poor
predictors of nutrient
concentration and
exhibited fairly
: = uncertain optima in
; § response to nutrient
Optimum
Caenis
.a
o
0)
Q.
(0 CM
O °
o
o
Diacyclops thomasi
o
0005 0.01 002
Total Phosphorus (mg/L)
0.001 0003 OOi 0.03
Total Phosphorus (mg/L)
Mesocyclops. edax
a
3.
0.001 0003 001 003
Total Phosphorus (mg/L)
Figure 1-3. Zooplankton
nutrient optima derivation
plots
Figure 1-2. Example nutrient optima
derivation plot to identify indicator values.
gradients. This was a
surprise as these taxa, and diatoms in particular, almost universally
exhibit strong sensitivity to water chemistry (including nutrients) and
are frequently used to infer water quality conditions. The fairly narrow
gradient over which there were sufficient data, small number of actual
lakes with data, and the coarse level of the taxonomic information that
was available to use likely explain the lack of more convincing
phytoplankton responses.
In contrast, Wisconsin zooplankton data produced better nutrient
inference models and a range of nutrient sensitivities (1-3). There were
more than 53 zooplankton taxa available and more than 12 showed
consistent responses on both ends of the sensitivity spectrum.
Combination metrics were constructed based on these sensitivities to
produce low nutrient and high nutrient sensitive multi-taxa indicators
that showed significant responses to nutrient gradients.
Macroinvertebrate data from Wisconsin was intermediate in terms of its
sensitivity to nutrient gradients
in lakes, but there was also
limited range in nutrient
conditions for the lakes with macroinvertebrate data. Models
were better than phytoplankton, but not as precise as zooplankton.
•i.o -
, -1.5 -
Fish produced surprisingly precise nutrient inference models from
the Wisconsin data and exhibited a wide range in nutrient
sensitivities (Figure 1-4). This was somewhat surprising given the
"trophic" distance offish taxa from nutrient inputs but may reflect
a pronounced influence of nutrient loading on oxygen
concentrations in lakes. The fish indicator values were also used
to produce multi-taxa indicators that were responsive to nutrient
gradients.
TVA fish data also produced robust models and also generated a
variety of nutrient sensitivities among fish taxa.
-2.5 -
''€; = .-e: T=
Figure 1-4. TP inference model
based on Wl fish data.
-------
Downstream Use Protection April 2015
In summary, nutrient sensitivity appeared consistently across taxonomic assemblages. Zooplankton and
fish produced more precise inference models than phytoplankton or macroinvertebrates, but the basis for
this was hypothesized to be methodological (Paul et al. 2014). Nutrient sensitive aquatic life use
measures could be derived easily from single or multiple assemblages from the indicator value/optima
analysis. The USEPA National Lake Assessment has shown this is possible with phyto- and zooplankton.
1.1.2 Developing numeric nutrient targets for the receiving water
Once sensitive uses and indicators were defined (Step 1), the next step (Step 2) was developing numeric
nutrient targets for the receiving water. The first part of this study focused on this step. The regional
nutrient and biological response data from Wisconsin and Tennessee compiled in the first project were
used to develop regional stressor-response models to derive numeric nutrient targets protecting sensitive
lake aquatic life for use in the study lakes used for Step 3. Chapter 2 of this report describes the methods,
results, and nutrient targets derived from those analyses.
1.1.3 Calculating watershed nutrient levels necessary to protect the lake
Once in-lake nutrient targets were developed to protect sensitive aquatic life uses, the final step in the
process (Step 3) was quantifying nutrient targets in the contributing watershed that need to be met in
order to support ALU attainment in the downstream lakes and reservoirs. The second part of this study
describes methods for projecting receiving water nutrient targets into downstream protection value (DPV)
nutrient targets for streams to meet in the contributing watershed to protect the lake. Chapter 3 of this
report covers theoretical issues, existing approaches to DPVs, converting lake targets to influent loading
limits, and converting influent loading limits to concentrations in the watershed network. It then applies
these concepts in two lakes examples: one from a temperate lake in Wisconsin and the other from a
reservoir in Tennessee.
1.2 Study Areas
The study areas selected for this work were Wisconsin and Tennessee. Wisconsin was selected because it
represented a region rich in natural temperate cool, typically dimictic lakes. It was also selected because
of the presence of the North Temperate Lakes Long Term Ecological Research Program (NTL LTER,
lter.limnology.wisc.edu) site in northern Wisconsin, with a long history of natural lake data collection,
including a wide range of biological responses. Tennessee was chosen because it represented a region
with subtropical/temperate, typically warm monomictic reservoirs that provided a contrast in lentic
waterbody type as well as aquatic life. While sport fishing in both regions is valued, there is a natural and
expected fauna for cool, dimictic lakes that defines a natural aquatic life community, even for an
impoundment in Wisconsin. In contrast, warm monomictic southeastern reservoirs are unnatural systems
placed on riverine landscapes with no expected lentic fauna. Aquatic life uses for such systems do not
have a natural expectation and, therefore, recreational or sports fisheries pose a realistic and defensible
endpoint for management.
For the DPV section, two specific lentic waterbodies were chosen: Holcombe Flowage in Wisconsin and
Douglas Reservoir in Tennessee. These waterbodies were chosen because of their location relative to the
regions for which we had sufficient data to complete the nutrient target development portion of this work
(Chapter 2) and because each had a substantial amount of watershed sample data. The Holcombe Flowage
is in northeastern Wisconsin on the Chippewa River. It supports a diverse fishery and recreation and
drains a sizeable watershed. Douglas Reservoir is in eastern Tennessee on the French Broad River. It
supports a constrained warmwater fishery and according to TV A has scored in the poor to low end of the
fair range for ecological health since 1994. More details on these specific waterbodies are provided in
Chapter3.
-------
Downstream Use Protection April 2015
1.3 Study Data
1.3.1 North Temperate Lakes Long Term Ecological Research Program Data
A comprehensive lakes data set was acquired from the North Temperate Lakes Long Term Ecological
Research Program (NTL LTER, lter.limnology.wisc.edu. downloaded July 2013). The data set consisted
of 150 data files representing both data collected as part of the baseline monitoring for the NTL LTER
and data collected for 16 additional projects that were also available from the NTL LTER program
website (Table 1-1).
-------
Downstream Use Protection
April 2015
Table 1-1. North Temperate Lakes
data selected for use.
Project
Biocomplexity at North Temperate
Lakes Long-Term Ecological Research
Cascade Project
Cross Lake Comparison
Crustacean Zooplankton Species
Richness in 66 North American Lakes
EPA Eastern Lake Survey, Upper
Midwest Region
EPA Environmental Research Lab
Duluth Lake Survey
Fluxes project at North Temperate
Lakes LTER: Random Lake Survey
Lake Mendota Phosphorus
Entrainment 2005
Lake Metabolism
Lake Wingra
Landscape Position Project
Little Rock Lake
Madison Lakes Zooplankton
North Temperate Lakes LTER (Base
Tl \
Long
Term Ecological
u T i Water
# Lakes _ ... _. ,
Quality Data
61
8
19
11
254
435
7,588
1
31
1
32
1
2
13
2001-2004
1984-2007
2006
19923
1984
1979-1982
2004
2005
2000
1996-2013
1998-2000
1983-2000
1981-2013
Research program data reviewed for use in this analysis. Shaded cells indicate
Nutrient Data Fish Data Invertebrate Data Phytoplankton Data
2001-2004 2001-2004
1991-1999 1984-2003 1984-1995
2006 20032
1984
1979-1982
2004
2005
2000
1996-2013
1998-2000 1998-1999 1998-1999
1996-2000 2002-20042
1981-2013 1981-2012 1981-2004 1984-2012
Zooplankton Data
2001-20041
1984-2007
2006
1992
1983-2007
1976-1994
1981-2012
Program)
North Temperate Lakes LTER (High
Frequency Data)
1989-2012
-------
Downstream Use Protection
April 2015
Project # Lakes _ ... _. , Nutrient Data Fish Data
J Quality Data
Invertebrate Data Phytoplankton Data Zooplankton Data
Primary Production and Species
Richness in Lake Communities4
Wisconsin Historical Lakes Data
Yahara Lakes Fisheries
13,093
4
1925-2009
1925-2009
1997-20005
1987-19986
1997-2000
1997-2000
Presence/Absence data only
Paired with Biocomplexity water quality and nutrient data
Conductivity data only
Richness data only; taxonomic abundance data are not available.
Paired with NTL LTER (base program) water quality and nutrient data
Paired with Crustacean Zooplankton Species Richness water quality and nutrient data
-------
Downstream Use Protection April 2015
To prepare the most complete data sets possible for the analysis, lake identifying features (name or other
identifier, geographic coordinates, lake size, and county) were extracted from each file containing that
information. Source filenames were maintained along with the lake information for tracking purposes.
Lake information was compared in order to identify lakes that were the same across projects. In many
cases, it was not possible to determine if similarly named lakes were, in fact, the same lake. Geographic
coordinates were not available for lakes in the Cascade, Cross-Lake Comparison, Lake Metabolism, or
Wisconsin Lake Plants Historical Data projects, nor were they available for approximately 10% of the
lakes in the Wisconsin Historical Lakes Data.
Next, water quality and nutrient data across all projects for which they were available were combined.
Different projects contained data for one or more of the following parameters (parenthetical is number of
studies with relevant data):
• Conductivity (7/18) • Total nitrogen (TN uf)(7/18)
• Dissolved oxygen (8/18) • Total dissolved nitrogen (TN f)(4/18)
• pH (10/18) • Dissolved inorganic phosphate (PO4
• Water temperature (9/18) f)(6/18)
_. , , . , /T^m/m/io\ * Total phosphorus (TPuf)( 12/18)
• Dissolved organic carbon (DOC)( 10/18) r r
~ . , • , /-Tr»r>w/:/io\ * Total dissolved phosphorus (TP f)(5/l8)
• Total organic carbon (TOC)(6/18)
_. , , . ^TTTS^/TOS • Secchi depth (9/18)
• Dissolved ammonia (NH3)(6/18)
TV i A •+ + i •+•+ ™r» ^ • Chlorophyll a (4/18)
• Dissolved nitrate plus nitrite (N Lh +
NO3)(4/18) • Phaeophytin(4/18)
• Total Kj eldahl nitrogen (TKN)(4/18)
Prior to combining the data, all collection, laboratory analysis, and reporting methods and units were
reviewed to ensure that only comparable data were combined. All nitrogen analytes were reported on an
"as Nitrogen" basis, and all phosphorus analytes were reported on an "as Phosphorus" basis. Due to lack
of sufficient data, total nitrogen or total phosphorus was not calculated for samples lacking those analytes.
Where necessary, data were converted to ensure that reported units were identical.
Water quality and nutrient data were reviewed to identify flagged data, outliers, and likely erroneous
values (negative or zero values considered to be erroneous). All data that were flagged for quality reasons
were removed from the data set as were negative or zero values for analytes having a method detection
limit. Dissolved oxygen values greater than 25 mg/L and water temperature values greater than 35 °C
were also removed.
All projects contained chlorophyll data that were corrected for phaeophytin, but projects variously applied
fluorometric methods, spectrophotometric methods, or both. Spectrophotometric data were selected over
fluorometric data when both were available, since those data were more prevalent across the different
projects. Likewise, spectrophotometric data for phaeophytin were chosen over fluorometric data when
both were available.
Nutrient data (DOC, TOC, NH3, NO2 + NO3, TKN, TN, PO4, and TP) were logio-transformed prior to
calculating site averages. Water quality data, Secchi depth, chlorophyll, and phaeophytin were not
transformed. Site-specific water quality and nutrient data were then averaged over replicates (samples
obtained on the same day from the same location using the same methods that represent either field
duplicates or laboratory duplicates). After averaging replicate data, site-specific averages over depths and
time were calculated as follows:
-------
Downstream Use Protection
April 2015
• Water data to be combined with benthic macroinvertebrate data were first averaged over all
sampled depths (since samples were in the littoral), then annually.
• Water data to be combined with fish data were first averaged over sampled depths ranging
between 0 and 2 m (to estimate photic zone conditions), then annually.
• Water data to be combined with zooplankton data were first averaged over sampled depths
ranging between 0 and 2 m (to estimate photic zone conditions), then (separately) annually and
seasonally.
• Water data to be combined with phytoplankton data were not averaged over depths or time.
Unfiltered TP and TN data were used to model spectrophotometric chlorophyll, since there were more
paired data for these variables than for filtered values.
Biological data were similarly combined. Fish data were segregated by method: separate data files of
combined project data were prepared for electrofishing, gill net, and seine data. Similarly, benthic
macroinvertebrate data were segregated by sampling method: separate data files of combined project data
were prepared for Hester-Dendy (passive sampler), coarse woody habitat, and SCUBA vacuum data. All
zooplankton data were collected using the same methods, as were phytoplankton data. Therefore, all
project data were combined for these two data categories into one file for each category.
Lastly, water quality and nutrient data were merged with biological data. Specifically, for all but the
phytoplankton data, water data for each lake/site were joined with biological data for that lake/site where
sampling occurred during the same time frame as the water data averaging (e.g., same year, same season)
and for the same project. When a biological sample did not match a water sample from the same project,
an attempt was made to identify appropriate samples from a different project. A water sample was
considered an appropriate match if it was sampled in the same time frame as the biological sample, or in
the case of benthic macroinvertebrate s and fish, if the water sample was obtained during the preceding 2
years (using the closest year's data). Temporally shifted data for 1 lake that was sampled for fish (West
Long Lake, Cascade project, fish data from 2001 and 2002, water data from 2000) and for 8 lakes
sampled for benthic macroinvertebrate s were used (Table 1-2).
Table 1-2. Asynchronous benthic macroinvertebrate and water chemistry data.
Lake Name
Pallete Lake
Vandercook Lake
Jute Lake
Found Lake
Towanda Lake
Camp Lake
Moon Lake
Little Rock Lake
Invertebrate Data Source
Cross-Lake Comparison Project
Cross-Lake Comparison Project
Cross-Lake Comparison Project
Cross-Lake Comparison Project
Cross-Lake Comparison Project
Cross-Lake Comparison Project
Cross-Lake Comparison Project
Little Rock Lake Experiment
Invertebrate
Data Year
2003
2003
2003
2003
2003
2003
2003
2004
Water Data Source
Biocomplexity Project
Biocomplexity Project
Biocomplexity Project
Biocomplexity Project
Biocomplexity Project
Biocomplexity Project
Biocomplexity Project
Biocomplexity Project
Water Data
Year
2001
2002
2002
2002
2002
2002
2002
2003
Phytoplankton data were joined to water data collected within 30 days (before or after) of the
phytoplankton data collection, as part of the same project.
-------
Downstream Use Protection
April 2015
1.3.2 Tennessee Valley Authority Long-Term Monitoring Data
A second data set was acquired from the Tennessee Valley Authority (TVA, Table 1-3). Water quality
and nutrient data collected prior to 2000 from EPA's Legacy STORET database were downloaded
September 30 - October 2, 2013). TVA personnel provided water quality and nutrient data collected since
2000 (TVA, personal communications with Tyler Baker, September - November 2013) and with fish
community data. TVA has been monitoring water quality since 1960, and has collected fish community
data since 1993.
Table 1-3 Tennessee Valley Authority program data used in this analysis.
Data Source
EPA's Legacy STORET Database
TVA Water Chemistry Data
TVA Fish Community Data
# Lakes
32
31
31
Water Quality Data
1960-1997
2000-2006
Nutrient Data
1961-1997
2000-2006
Fish Data
1993-2012
The TVA data sets included the following water quality and nutrient parameters:
Conductivity
Dissolved oxygen
pH
Water temperature
Turbidity
Total dissolved solids (TDS)
Total suspended solids (TSS)
Total organic carbon (TOC)
Ammonium, unfiltered (NH4>
Nitrite, unfiltered (NO2)
Nitrate plus nitrite, unfiltered (NO2 +
N03)
Nitrate, unfiltered (NO3)
Organic nitrogen, filtered (DON)
Organic nitrogen, unfiltered (TON)
Total kjeldahl nitrogen (TKN)
Total nitrogen, unfiltered (TN uf)
Phosphate, filtered (PO4 f)
Total organic phosphorus, unfiltered
(TOP)
Total phosphorus, unfiltered (TP uf)
Total phosphorus, filtered (TP uf)
Secchi depth
Chlorophyll a
Phaeophytin
10
-------
Downstream Use Protection
April 2015
TVA data were prepared following the quality control requirements as for the NTL LTER data. Unlike
the NTL LTER data, the TVA data reported Method Detection Limits (MDLs). For each reported value of
less than the MDL for any analyte, final result value was set equal to one-half the value of the MDL. Site
replicates were next averaged, and site-specific average calculated for samples obtained between 0 and
2m deep. Because TVA collected fish community data multiple times between August and November at
each site, both monthly and seasonal averages were prepared for the water quality and nutrient data.
Young-of-year fish were excluded from the fish community data, because they typically are under
sampled and are not representative of the fish community. Next, catch per unit effort (CPUE) was
calculated from the abundance and effort data. Lastly, the fish community data were separated by
sampling method (electrofishing or gill net). Fish data were then joined with the monthly and seasonal
water quality and nutrient data to create the final data sets used in the analysis.
TVA classifies their reservoirs into main and tributary reservoirs. Main stem reservoirs appear on the
main stem Tennessee River although two labeled main are at the bottom of the Clinch, and Little
Tennessee. The main stem reservoirs also have, presumably, lock and dam structures or canals that
enable barge traffic. Tributary reservoirs are on major tributaries of these main stem rivers and appear to
lack the lock and dam structures. We analyzed these systems together, except we noted that fish metrics
differed obviously between the two types: main and tributary, therefore we standardized metrics to the
central tendency of these two classes of TVA reservoirs (explained in detailed methods in Chapter 2). WI
lakes were not classified, but data were primarily limited to ecoregion 50, the LTER region and location
of the study lake for Chapter 3.
Nutrient
Phytoplankton
1.4 Methods Overview
For deriving nutrient targets for sensitive uses,
we used a standard stressor-response based
approach using a simple conceptual model
(Figure 1-4). In this model, nutrients stimulate
phytoplankton productivity and increase
phytoplankton biomass (estimated as
chlorophyll a). This biomass affects oxygen
directly via photosynthesis and respiration by
primary producers, and indirectly via
microbial respiration of detrital algal biomass.
The biomass also influences biota directly by
providing food/energy through bottom-up
control, but also via mediated changes in food
quality. Lastly, oxygen availability has a direct
effect on biota. We modeled nutrient-
chlorophyll, chlorophyll-dissolved oxygen,
chlorophyll-biota, and oxygen-biota
relationships (dark lines in Figure 1-5). We also explored models of the nutrient-biota relationship, which
treats the intermediate steps implicitly. Using targets for biota or oxygen, we identified chlorophyll
endpoints and then nutrient targets needed to meet those chlorophyll endpoints.
Oxygen
Biota
Figure 1-5. Simple conceptual model of nutrient
effects on aquatic life in the study
systems.
TETRATECH
11
-------
Downstream Use Protection April 2015
For calculating downstream protection levels, a few approaches currently applied to the problem are
described in detail (Vollenweider, BATHTUB, and weighted distribution). Next, a general approach for
translating loads into concentrations is described. It starts with a description of concentration data
analyses using a few approaches such as setting downstream concentrations to upstream or using an upper
confidence interval of the downstream target upstream. It then describes simplified assignment of load
procedures using methods such as LOADEST. Next, methods for accounting for spatial correlation in
watershed nutrient concentrations are applied and are followed by methods to generate a complete
network using the covariance structure developed with the spatial correlation analysis. Finally, a Monte
Carlo simulation approach is used to simulate a complete nutrient dataset for a watershed to test various
sampling strategies and exceedance profiles to test against downstream targets.
Each of the subsequent chapters provides greater detail about the methods applied for each analysis.
TETRATECH
12"
-------
Downstream Use Protection April 2015
2.1
This section describes the detailed methods used in developing numeric nutrient targets to protect
sensitive aquatic life uses in the lakes. A description of some of the detailed data preparation is followed
by a description of the statistical methods used.
2.1.1
Regression models to characterize the relationships described in Figure 1-4 were developed. We accept
that compounding error may have been an issue, but we believed we lacked the data structure to pursue
multi-level models to reduce this effect. We explored the conceptual model relationships using a variety
of data. First, paired observational data were used. This consisted of the synoptic pairs of existing data to
the extent provided by the data. At times, if chemical (including dissolved oxygen) or chlorophyll data
were not contemporaneous, we identified the closest date. Next, we explored models using paired
average lake-seasonal data (e.g., summer average dissolved oxygen data with summer average
chlorophyll a). Some of these models also explored seasonal lags (e.g., late summer or august dissolved
oxygen profiles as a function of spring or summer lake average chlorophyll a). We focused on critical
late summer periods for dissolved oxygen when organic matter loading is expected to express the greatest
effect on respiration and oxygen stress during the end of summer stratification.
For dissolved oxygen (DO) analyses, we needed to identify the hypolimnion. We explored a few methods
to do this. For the WI dataset, the hypolimnion was first estimated as that point when sample depth was
more than maximum Secchi depth of the lakes, a fairly crude method. We also used a temperature driven
approach. We first specifically selected summer lake samples for which depth, water temperature, and
DO concentration were reported. For each of these samples, the change in temperature relative to the
change in depth (Atemperature/Adepth) between adjacent measurements were calculated and rates of
temperature change >=0.75 were identified as potential thermocline locations. Samples containing a
clearly identified thermocline were selected for the final data set. These included only those samples
where minimal change occurred prior to the thermocline, and minimal change occurred after the
thermocline (in other words, the thermocline boundaries were reasonably clearly demarcated). In some
samples, depths with rapid rates of change (>=0.75) were interrupted by a gap of 1 or 2 measurements of
lesser change (<0.75). In these cases, if the rate of change over the entire suspected thermocline
(including the gap) was >=0.75, the entire region was identified as thermocline. If not, the region of
greater change was selected. In some cases, there was a strong change near the surface. These were not
included as they might result from solar heating of surface waters. After identifying the top and bottom
of the thermocline, the average thermocline depths were calculated and attributed to either the epilimnion
or the hypolimnion. The proportion of DO measurements < 2 mg/L (count DO <2 mg/L divided by total
count DO) in the hypolimnion and the average chlorophyll a in the epilimnion layer were then calculated
for use in those regression models.
For the TVA data, we used lake profile data from summer months (June to September) to summarize
patterns in DO concentrations. Lake profiles with fewer than 4 sampling depths were excluded. The
thermocline for each lake profile was identified by finding the depth at which the greatest rate of decrease
TETRATECH
13"
-------
Downstream Use Protection
April 2015
in dissolved oxygen (DO) per foot increase in sample depth occurred, that was less than or equal to 15 m
in depth (which is technically the oxycline). DO was used rather than temperature to identify the
thermocline because it appeared to be more accurate based on visual inspection of lake profiles, but
thermocline and oxycline profiles are generally coincident in this dataset. Correct identification of the
thermocline was then confirmed by visual inspection of the lake temperature profiles.
We characterized hypolimnetic DO by determining the proportion of DO measurements at or below the
thermocline that were > 5 mg/L. This process includes the metalimnion with the hypolimnion. Including
the metalimnion was necessary, because, for many lake profiles, the lakes were not deep enough for the
temperature to stabilize below the thermocline. We similarly characterized epilimnetic DO by calculating
the proportion of samples above the thermocline that were > 5 mg/L.
As in WI, because we were interested in identifying conditions most stressful for fish in terms of DO, we
narrowed subsequent analysis to the month of August for which 1) stratification as measured by relative
thermal resistance (RTR) was strong and 2) dissolved oxygen levels were lowest on average (Figure 2-1).
Dissolved oxygen (mg/L)
Dissolved oxygen (mg/L)
Dissolved oxygen (mg/L)
* Dissolved oxygon
"5. "
° Temperature
* Dissolved oxygon
s Temperature
• Dj&sotved oxygen
10 20
Temperature (C)
10 20
Temperature (C)
10 20 30
Temperature (C)
Figure 2-1. Lake temperature profiles from the same sampling station in one year in April (left),
August (center), and October (right). The thermocline is represented as a dashed line.
The proportion of the water column for which dissolved oxygen (DO) was less than 4
mg/L, effectively reducing habitat size for intolerant fish species, was lowest in the
middle panel (August). The relative thermal resistance, a unitless index of stratification
strength, is printed in the upper right corner.
RTR is a unitless index of lake stratification strength (Kalff, 2002) that is based on density measures and
not temperature change. This is important as the density of water changes as a function of temperature
and in warm, southeastern monomictic reservoirs, the same absolute temperature difference as northern
temperature lakes does not equal the same density difference. We calculated RTR for each lake profile as
the difference in density between the two samples above and two samples below the thermocline:
RTR = (5bebw- 5above)/(8 x 10"6); where the denominator was the density difference between water at 5° C
and 4° C.
Density (5) was estimated as:
5=1- 6.63 x 106 (T- 4)2; where T was mean temperature in °C. We then used the magnitude of RTR to
identify the strongest stratifications.
I TETRATECH
14
-------
Downstream Use Protection April 2015
To assess biologic response to lake DO, chlorophyll and nutrient concentrations, we used biological data
from the WI dataset. For the invertebrates, six different collection methods were used, including benthic
samples (vacuum, N=140), Coarse woody habitat (vacuum, CWH, N=140), Vacuum (N=151) methods,
and a Hester-Dendy substrate (N=390) method. However, only two lakes (Sparkling Lake and little Rock
Lake) had samples collected from natural habitats whereas 32 of 33 lakes used artificial substrate to
collect invertebrates due to the depth of the lakes. Therefore, only Hester-Dendy samples were analyzed.
According to the LTER website, the Hester-Dendy samples were mainly collected by the Landscape
Position Project, while other projects also collected macroinvertebrate samples from these lakes using
similar methods:
"Two Hester-Dendy samplers were set at a depth of one meter on each of three substrate
types (cobble, sand and silt) within each lake for four weeks in late June through late
July ... Within each lake, areas of different substrate types were identified using WI-DNR
depth contour lake maps, and substrate type was verified by direct observation. Different
substrates were sampled to account for invertebrate associations with specific substrate
characteristics."
224 unique taxa were identified, and OTU (operational taxonomic unit) were developed for different
levels of taxonomy, but in the end, species, genus, and family levels were used for the analysis. Seventy-
eight invertebrate metrics were calculated to relate to environmental variables. Due to the difficulty of
matching invertebrate samples collected from different times of the year and locations of the samples with
environmental variables, we used a yearly summary statistics of DO data collected in the same lake. The
majority of zooplankton samples were collected in six lakes, including Allequash Lake (31), Big
Muskellunge Lake (32), Crystal Lake (33), Little Rock Lake (36), Sparkling Lake (33), and Trout Lake
(33). Due to the size of the lakes and different sampling years, we treated each sample as independent.
We also evaluated fish data from WI. Most fish were assigned tolerance values according to the Rapid
Bioassessment Protocol (RBP, Barbour et al. 1999). Tolerance values developed from previous analysis
were also used to link biological (fish) responses to nutrient concentrations.
For the TVA dataset, only fish data were provided. We used fish relative abundance data collected each
fall by TVA. To generate fish relative abundance data, TVA electrofished fifteen 1000m transects along
each station in the fall of each year and data were abundances expressed as catch per unit effort. We
calculated fish species richness and Shannon-Wiener biodiversity index for all species. Using a subset of
species we considered game fish (Table 1). The Shannon-Wiener index includes a richness component,
but also weights samples higher for relative evenness. We also computed percent tolerant species and
percent intolerant species for each sample using tolerance data from the RBP (Barbour et al. 1999).
Because tributary storage reservoirs and main stem reservoirs (see Study Area in Chapter 1) had
significantly different richness and Shannon-Wiener index values (p < 0.0001, one-way ANOVA), we
relativized those metrics to their class means (Figure 2-2). To assess the effect of August DO
concentration on fish species diversity, we used linear models to compare the diversity indices to the
proportion of DO samples > 5 mg/L in the hypolimnion and in the water column and minimum August
DO. As supporting evidence, we also examined the effect of summer chlorophyll a concentrations and
spring - summer nutrient fish diversity.
TETRATECH
15"
-------
Downstream Use Protection
April 2015
I
* 8
2
8
s-
Main
Trib
Main
Trib
Figure 2-2. Box and whisker plots offish species richness and Shannon-Wiener diversity scores
for fish samples from mainstem (Main) and tributary (Trib) TV A reservoirs. Reservoirs
on the main stem of the river had higher fish biodiversity index scores than tributary
storage reservoirs.
Finally, to examine the effect of lake chemistry on individual taxa, we examined the relative abundance of
bluegill and largemouth bass in relation to the minimum August DO, and proportion of the water column
and of the hypolimnion that had DO measurements > 5 mg/L using linear models. We chose these species
for their presence in the majority offish samples (bluegill: 100%, largemouth bass: 99%). We also looked
at the taxa in relation to summer chlorophyll a and spring - summer concentrations. We transformed
relative abundance data with a fourth root to improve normality.
2.1.2 Analysis Methods
For the Wisconsin analysis, scatterplots with loess smoothed fits of mean and 95th quantiles were
developed for several relationships. We then conducted correlation, regression and change-point analyses.
Simple linear regression models were most commonly used, followed by logistic regression for
proportion data (e.g., proportion of hypoxic hypolimnia). For models of DO as a function of temperature
and chlorophyll a, multiple linear regression was used and the minimum DO plotted as contours on a
scatterplot of temperature vs. chlorophyll a. Finally, we also used change point analysis (deviance
reduction method, Qian et al. 2003), to evaluate non-linear changes in response conditions as a function
of independent variables. For the TVA analysis, simple linear and logistic regression were used. We
transformed regression equations of chlorophyll as a function of nutrients to derive nutrient targets. We
performed all analysis in R (R Core Team 2014).
2.2 Results - Wisconsin
This section follows the conceptual model described in Chapter 1 (Figure 1-4) and describes the results of
the Wisconsin lakes analysis.
TETRATECH
16
-------
Downstream Use Protection
April 2015
2.2.1 Nutrients and Chlorophyll a
There were two types of chlorophyll a (fluorometric and spectrophotometric), TN (filtered and
unfiltered), and TP (filtered and unfiltered) measurements in the dataset. Chlorophyll a concentrations
were measured either as a composite from the water column, or at certain depths, depending on different
projects. It was difficult to organize the thousands of samples from the project descriptions for each of
them.
Chlorophyll a varied within and between years and among lakes, as did nutrient concentrations (Figure
2-3 to Figure 2-7). This provided a sufficient range for generating nutrient-chlorophyll response models.
10
Chi a
15
Figure 2-3. Intra- and inter-annual chlorophyll a concentration (ug/L) variation in Wl lakes. Both
mean and 1 standard deviation by year are shown.
TETRATECH
17
-------
Downstream Use Protection
April 2015
V/est Lutiy Lake
V/ait)-(reported as)
luesdt-ty Lake
1 -otrf Lake
1 rout Bog ^ •"
1 enderfool L%ke
c liJnny Lake
lat Ldke
p%rkling Lake
nipe-trftke ^
eddiiiuluti Buy (leported as)
lum "(reported as)
F eter Lake—•—
Rau-Wiake
a#t~(reported as)
Flalnihii g.dkhi
aate Bog (reported as)
l\|laryLake *
Lorrg^reported as)
L ttle ftiawling Stone Lake
L ttle Arbor Vitae Lake
LakeWingra '
Lak^Ldura
Uac flu Lune
ickapoo Reported as)
idiairLakB
h ummingbird (ibpuifed db)
h iawathat_ake
h etnet Lake
E ast Long Lake*(reported as)
C temondLake
yfystal Lake^
ranberry Bfly (leported as)
ratnpton Lake
diftp (reported as)
rown Lake
E olger EJfcg (reported as)
E og Pot (reported asj —
ig F^ildge Lake
ig*Muskellunge Lake
nvil Lalte -
lleqtfesti Lake
10
20
30
40
Chi a
Figure 2-4. Intra- and inter-lake variations of chlorophyll a concentrations (^g/L) in Wl lakes. Both
mean and 1 standard deviation by lake are shown.
TETRATECH
18
-------
Downstream Use Protection
April 2015
0.01 0.02 0.03 0.04 0.05
TP
Figure 2-5. Inter- and intra-annual total nitrogen (TN, mg/L) and total phosphorus (TP, mg/L)
variation. Both mean and 1 standard deviation by year are shown.
TETRATECH
19
-------
Downstream Use Protection
April 2015
V/est t*mg Lake
V/ard (reputed as)
Tuesday Lake
1 'ouf1*ake
1 "out Bog ~~*
Tenderfoot Lake-*
£ ttftmy Lake
£ taftake
£ paTflfnq Lake
£ nipe Lak%
F eddinglun Buy (leported as)
F lum (reprnted as)
F eter tfeke
F a* Lake
F auh(teported as)
F aimer Lake *
r'ortrafet5_Bog (reported as)
Mary LaVe
Lyn*-Lake
Longjfrftported as)
L ttle~"trawling Stone Lake
L ttle Arbor Vitae Lake
Lake Wingra-*—
Lake gdura
LaC*Su Lune
Y- ickapoo (tepftiled as)
Indian l*dke
h ummingbintfreported as)
h iawattTJhlake
h elmertake
E ast Long~tfeke (reported as)
C iAnondLake
C fystal Lake
C rantferry Bog (reported as)
C rafnpton Lake
C raw-take
C ampfmported as)
E rown Lake —*
E olger Bog (tepoiled as)
E og Pot (repoileitab)
E ig-Mortage Lake
E ig^/luskellunge Lake
f nvil tdke
/ lleauash Lake *
0.05
0.10
0.15
TP
Figure 2-6. TP (mg/L) variations within and among Wl lakes. Both mean and 1 standard deviation
by lake are shown.
TETRATECH
20
-------
Downstream Use Protection
April 2015
V/est Long Lake •"
Tuesda^Lake
F eter
F aul LaVe
j i i i i i i
Lake Wingra-
I- ummingbird (reported
E ast Long Lake
C raiiifJtuii Lake
\ i i i i i r
0.4 0.6 0.8 1.0 1.2 1.4
TN
Figure 2-7. TN (mg/L) variations within and among Wl lakes. Both mean and 1 standard deviation
by lake are shown.
A preliminary examination of the relationships between different nutrient parameters showed that
spectrophotometric chlorophyll a had the most paired samples with unfiltered TP and TN concentrations
(Figure 2-8, more than 5300 paired samples). However, the strongest simple linear regression relationship
was between fluorometric chlorophyll a concentrations and unfiltered TP concentrations. The coefficient
was almost strong as the correlations between TN and TP. In either case, chlorophyll was significantly
correlated with nutrients, as expected.
TETRATECH
21
-------
Downstream Use Protection
April 2015
133 -
* 10 -
100 T
tf 10
D
ul
0.005 0.01
0.05 0.1
odio o o
130 -
t' 10 -
_. . . .
,:, o * Ciooocitxoo 0x0*00 o o o .:
O O O0QO0OO 00000* « «00 O O0 O O O
O O 0000000 0030 OOKEOOO OB O OO
O O O O O O O 00O000 OOCOOX TOD O O OOO O O
O O O O O O
I I I I I I I
0.005 0.01
0.05 0.1
TP uf
O.E
n.lQ.05 -
h-
0.01 -
0.005 -
3.1
13
13
Figure 2-8. Relationships between paired nutrient and chlorophyll a concentrations in the large Wl
lake dataset. Uf = unfiltered, spec = spectrophotometric, fluor = fluorometric. All
regressions are significant (p<0.05).
This analysis included data from all ecoregions (not only ecoregion 50). If only ecoregion 50 was
selected, then only a small number of samples (262) samples was available in the fluorometric dataset.
More than 1,600 samples had both fluorometric chlorophyll a samples and TP measurements. Direct
factors affecting chlorophyll a concentrations included nutrients (TN and TP), light (depth of samples),
and temperature (season, month). A linear model including all four factors as predictors found no
I TETRATECH
22
-------
Downstream Use Protection
April 2015
significant effect of depth, while water temperature also dropped out when seasonality was considered. In
summer, chlorophyll a concentrations were significantly correlated with both TN and TP. However,
overall, the effect of season was not significant for this particular dataset.
2.2.2 Chlorophyll a and Dissolved Oxygen
Dissolved Oxygen (DO) was collected in many of the WI lakes from different projects. Surface (depth < 1
meter) and/or vertical profile DO measurements were monitored at different times of the dayand were
used to characterize the spatial and temporal variations of DO in these lakes. In addition, a few
continuous surface DO monitoring efforts were available to explore seasonal and diel patterns.
Both temperature and primary production were the primary drivers for DO fluctuation in the water
column, so surficial DO and temperature varied seasonally. As expected, DO reached the highest peak in
early spring when nutrients and algal production were high and temperatures were still low (Figure 2-9).
DO was lowest in summer when temperatures were highest, then DO began to rise and temperature
decline during the fall. Compared to seasonal variation, epilimnetic DO daily variation was minor in this
lake (± 1 mg/L).
O)
O
Q
12 -
11 -
10 -
9 -
8 -
- 25
- 20
- 15
10
I- 5
Apr
Jun Aug
Month
Oct
Figure 2-9. Epilimnetic DO and temperature fluctuation during 2012.
Seasonal temperatures and DO varied with depth in the WI lakes, following a typical cool dimictic pattern
as evidenced by Big Muskellunge Lake (Figure 2-10). Temperature was uniform and lakes well mixed
during fall (October to November) and spring (May) mixis. Stratification occurred at around 10 meter
TETRATECH
23
-------
Downstream Use Protection
April 2015
depth during the rest of the year. Thermal stratification was less in winter but DO was as stratified in
winter as summer in this northern WI lake.
Q. ^ -
1
i
1
i
i
j
1
1
i
i
!
j
i
i / ,
! ' /.
\ f
\'i
i '; .-''
i f.) / i
i if r
S :
j I.1! i
5
...
' >•
t
''
1
10
t
.:•,
,'' f
,-•' /
,-"'
^'
I
15
---""
_
•—
-------
Downstream Use Protection
April 2015
1
0
1
; ":
v\
1 /•
I .'
t\
1 1
1
1
1
I
!
i .
'•
1 \
!| '.
,'
1 !,i /
!l
! '''
1 il
i 'l1
!'v /
^ \
--11., }
r ""'"•-•--
j / ""•-
( i "•""-..
N /
I1, \
>\ -
i c- \
1 \
1 1 l-
1 1
5 10
Jan.17
--- Feb.28
•- Mar.31
— - May.02
Jun.12
- Jun.27
Jul.25
Aug.22
Sep. 20
— - Oct.17
- Nov.01
Nov. 14
I
15
o.
a
a
\
.'
f
\
\
\
\
I
0
• is \ i \
y\\ t
A \\ : j \
,' '. V'l I > !
\ \A X /
;• \\ < \
\( \ \ \
/ \ N
\ K
^- \ \ /
:' ">V-X. S
\ ^ \ \/
V S
\ ^
\ /' i '''•••
\ ,'/ \
V '' •'' / \
V f \ i -i \
\i-4- ('
V f --/-/-. /
h .' —f--.....
i\i / 1 /
1 1 1 1
2468
Jan.18
--- Feb.23
-- Apr.06
— - May.04
Jun.12
- Jun.27
Jul.25
Aug.21
Sep. 17
— - Oct.16
- Oct.31
Nov.12
I I
10 12
Chlorophyll a ijig I")
Chlorophyll a (ng I)
Figure 2-11. Vertical chlorophyll a fluctuation at Big Muskellunge Lake during different seasons in
1989 (left) and 2001 (right).
A number of factors affect the DO - chlorophyll relationship, including temperature and light intensity.
Ecoregion 50 had 87,263 samples in the final dataset. However, only 60 lakes out of 1,616 lakes (based
on lake name) had more than 10 DO samples, and only 13 lakes had more than 100 DO samples. Only 10
lakes had lake DO < 2 mg/L in their samples. All of the small lakes (with 10 or fewer DO measurements)
lacked paired DO and chlorophyll a data.
Dissolved oxygen declined (and hypoxia increased) as chlorophyll a and nutrient concentrations increased
(Figure 2-12, Table 2-1). We first examined instantaneous data and the instantaneous DO and
chlorophyll concentration relationship at all depths of various lakes was interesting (Figure 2-12, left). A
95th quantile loess smoothing line was fit to the dataset and the maximum observed DO declined when
instantaneous chlorophyll a concentrations were above 20 ug/L. This was confirmed with change point
analysis. Logistic regression was also used to model the relationship between water column chlorophyll a
concentration and the probability of DO > 5 mg/L (Figure 2-12, right).
TETRATECH
25
-------
Downstream Use Protection
April 2015
All observed samples
16-20-22
All observed samples
20 -\
15 -
P>
£ 10 -
5 -
0 -
Chlorophyll a Qig f )
CO
o
O
Q
o
CL «
O
O
5 10
n i rn
50 100
I | I rrq
5001000
Chlorophyll a (ug f )
Figure 2-12. Relationships between chlorophyll a and dissolved oxygen (DO) concentrations in all
instantaneous samples expressed with a loess fit to raw paired data (a) and a logistic
regression model fit to binned data (b). The blue fitted line is the 95th quantile loess fit,
the vertical line is the change point, and the gray shaded area is the 90% confidence
interval (Cl).
Table 2-1. Spearman correlations between temperature, nutrients, chlorophyll, and DO. Method 1
defined hypolimnetic depth based on Secchi depth. Method 2 was based on
temperature change. Uf = unfiltered, f = filtered,
Proportion Hypolimnetic DO < 2
mg/L (method 1)
Proportion Hypolimnetic DO < 2
mg/L (method 2)
Minimum Hypolimnetic DO
(method 1)
Minimum Hypolimnetic DO
(method 2)
Surface Mean DO
Proportion Water Column DO <
2mg/L
Mean Hypolimnetic DO (method
1)
Mean Hypolimnetic DO (method
2)
Water
Temp
0.37
0.58
-0.33
-0 16
-0.80
0.37
-0.49
-0.41
TP_f
0.37
0.46
-0.32
0.34
-0.37
-0.41
TP_uf
0.36
0.55
0.32
-0.33
-0.50
TN_uf
0.40
0.61
-0.35
-0.34
0.38
-0.40
-0.59
TN_f
0.38
0.59
-0.35
-0.31
0.36
-0.39
-0.59
IgNOx
-0.48
0.46
Mean
Chla
0.42
-0.34
Max
Chla
0.39
-0.35
TETRATECH
26
-------
Downstream Use Protection
April 2015
Most of the lakes did not have hypoxic epilimnia, though DO tended to decline with rising chlorophyll a
concentrations in the water column (r = - 0.25) and long-term epilimnetic dissolved oxygen declined as a
function of long-term mean chlorophyll a (Figure 2-13), which could be due to increased organic matter
loading from algal production to the surface layer. However, these responses may have co-varied with
and been driven by temperature as well.
Longterm Lake Average
10.0 -
9.5 -
3
e»
§ 9.0 H
CD
£ 8.5 H
D
•g 8.0 H
7.5 ~
7.0 -
Ra= 0.66
i i i i i |
5 10
Chlorophyll a (pig I 1)
i
20
Figure 2-13. Long-term mean chlorophyll a concentrations and surface epilimnetic DO at each lake
(each dot represents a lake).
The relationship between epilimnetic chlorophyll a concentration (maximum and mean) in the water
column and minimum DO concentrations were also evaluated across all seasons (Figure 2-14). A value of
2 mg/L was selected as the traditional threshold for hypoxia to convert DO data into binomials. DO
concentrations declined above maximum epilimnetic chlorophyll a values of 10 to 20 (ig/L and above a
mean epilimnetic chlorophyll a concentration between 4 and 8 (ig/L. Change point analysis (Qian et al.
2003) confirmed that decline occurred at thresholds of approximately 19 (ig/L maximum chlorophyll a
and less than 10 (ig/L mean chlorophyll a. The maximum chlorophyll a relationship with DO was more
significant than the mean DO relationship. Logistic regression models were used to assess the probability
of hypoxia in these lakes, and again increased between 10-20 (ig/L maximum chlorophyll a and 5-10
(ig/L mean chlorophyll a.
TETRATECH
27
-------
Downstream Use Protection
April 2015
11-19-20
Maximum Chi a
d
CT'
c
O
14 -
12 -
10 -
00
O)
KJ
o -
14 H
12 -
10 -
6 -
4 -
2 ~
0 -
100
Chlorophyll a (MQ f )
4-4-15
1000
10
Chlorophyll a
f )
M O
E
CM
V tt,
O °
Q
lity o
0.4
obabi
.2
o
o
5 10 50 100
Chlorophyll a (ng f1)
Mean Chi a
500 1000
• 1
10
20
50
Chlorophyll a (ug I")
Figure 2-14. Water column maximum (upper panel) and mean (lower panel) chlorophyll a
concentrations vs. lowest dissolved oxygen concentrations at all locations during all
seasons. Logistic regression models (right side panel) were also fit to the DO data in
the left side converting DO data to greater or less than 2 mg/L. The probability of
hypoxia (hypolimnion DO< 2mg/L) increased dramatically with elevated chlorophyll a
concentrations. Blue lines in left side panel are loess fits. Values above the left side
panels are the change point (center value), and the 95th percentile confidence interval
around the change point (left and right values).
We next focused on summer hypolimnetic dissolved oxygen, because with an absence of continuous
reaeration the hypolimnetic DO deficit is a good indicator of organic load, especially epilimnetic
productivity (Kalff 2002). The DO deficit in the hypolimnion is expected to be most pronounced in
I TETRATECH
28
-------
Downstream Use Protection
April 2015
summer, which was supported by observations (Figure 2-15). In addition, the hypolimnion is important
habitat for many aquatic species, including many fish.
05 -
04 -
Summer
A Spring
+ Other
5 10 50
Mean Water Column Chi a
Figure 2-15. Proportion of Hypoxic samples (DO <2 mg/L) in the entire water column and
chlorophyll a concentration in different seasons. Colored lines are loess fits.
These hypolimnetic responses were similar to the water column responses. Hypolimnetic DO decreased
with chlorophyll a, expressed as raw data (Figure 2-16). Change point analysis indicated that a value of
water column average chlorophyll less than 5 ug/L was associated with a decline in proportion hypoxia,
although the trend in the figure appears linear, so any change point should be considered carefully. The
same trends held when the hypolimnetic DO data were converted into proportion of hypoxia using a 2
mg/L definition. Proportion hypoxia increased with water column mean chlorophyll a and nutrient
concentrations, as well as with temperature (Figure 2-17).
TETRATECH
29
-------
Downstream Use Protection
April 2015
Hypolimnion DO
2-2-5
1.5 -
I 1.0
0.5 -
r = -0.2
1 5 10 50
Chlorophyll a (ng I"1)
Figure 2-16. Hypolimnetic DO vs. water column chlorophyll a. Change point (red line) occurred at
less than 5 ug/L. Values above the figure is the change point (center value), and the
95th percentile confidence interval around the change point (left and right values).
TETRATECH
30
-------
Downstream Use Protection
April 2015
1 1.0 H
E
™ 08 -
O
Q 06 -
I 0.2 H
CL
10 12 14 16 18 20
Mean Water Column Temperature
1 0 H
08
06
0.4
02
T
= 042
1 5 10 50
Mean Water Column Chi a
d 10"
E 08 -
CM
V
O 06 -\
Q
"S 04 H
1 02 H
o
Q.
2
= 06
I |
02 0.5 1
Mean Total Nitrogen (mg/L)
1 0 -
0.8 -
06 -
04 -
0.2 -
0005 001 005 01
Mean Total Phosphorus (mg/L)
Figure 2-17. Proportion of hypolimnetic hypoxia (DO <2 mg/L) vs. temperature, chlorophyll a, and
nutrients in the summer. Red lines are loess fits.
We explored the effects of water temperature using a multiple regression model of minimum DO in the
hypolimnion as well as proportion hypoxia using temperature and chlorophyll a as predictors. The
responses were plotted as contours on scatterplots of temperature vs. chlorophyll. Minimum hypolimnetic
DO concentrations in summer (Figure 2-18) responded most strongly to chlorophyll. After chlorophyll
was included in the model, the temperature effect was not significant (p>0.1). On the other hand,
proportion of hypoxia in the hypolimnion was correlated with both temperature and chlorophyll a (Figure
2-19).
TETRATECH
31
-------
Downstream Use Protection
April 2015
20 -
18 -
16 -
14 -
12 -
10 -
I |- T 1 [ -I j |
5 10 20
Mean Chi a dig I"1}
20 -
18 -
16 -
14 -
12 -
10 -
50
-, -, -, i i-| -i -. i , i-.-rq-
5 10 50 100
Maximum Chi a (pig I"1)
500 1000
Figure 2-18. Multiple regression models showing the minimum DO concentration in the
hypolimnion and its relationships with mean (a) and maximum (b) chlorophyll a and
temperature concentrations. The solid lines are predicted minimum DO
concentrations.
20 -
18 -
16 -
14 -
12 -
10 -
20 -
18 -
16 -
14 -
12 -
10 -
5 10 20
Mean Chi a ((ig f1}
50
! :!
10
50 100
500 1000
Maximum Chi a (|ig I )
Figure 2-19. Multiple regression models showing the proportion of hypoxia samples in the
hypolimnion and its relationships with mean (a) and maximum (b) chlorophyll a and
temperature concentrations. The solid lines are predicted proportion hypoxia. Both
temperature and chlorophyll a are significant predictors.
TETRATECH
32
-------
Downstream Use Protection
April 2015
These results are also found when the data are combined into lake-year averages (hypolimnetic DO and
chlorophyll a). Hypolimnetic DO decreases with chlorophyll a and change point analysis indicated lake-
year average values from 2-4 (ig/L associated with a sharp change in mean hypolimnetic DO. Similarly,
the probability of observing DO < 2 mg/L is essentially zero above 5 (ig/L lake-year average chlorophyll
a (Figure 2-20).
Summer Mean Values
2-2-4
Mean Chi a '/s.hypolimnion DO
o
Q
2 -
0 -
r = -0.36
CM
A
O
Q
5 10 20
Chlorophyll a Qig I"1)
50
I i i n^ i
5 10 20
Chlorophyll a Qig I"1)
Figure 2-20. Summer lake mean chlorophyll a and hypolimnetic DO concentrations as raw data
(left) and as binomials with DO > 2 mg/L as the response variable. Each data point at
left graph represents one lake-year. The gray areas are potential change points. The
figure on the right is a logistic regression of the probability of DO > 2mg/L in the
hypolimnion as a function of chlorophyll a.
2.2.3 Biota and Dissolved Oxygen, Chlorophyll a, and Nutrients
We analyzed phytoplankton, benthic invertebrate, zooplankton, and fish responses to dissolved oxygen,
chlorophyll a, and nutrients.
A total of 547 phytoplankton samples were collected from only 6 WI lakes from 1984 to 2012 in the
entire dataset. However, the majority of samples were collected in Lake Mendota (281) and Lake Monona
(181), south of ecoregion 50. As a result, insufficient data were collected from lakes of ecoregion 50.
Phytoplankton taxonomic composition response could not be used in this analysis.
Atotal of 821 invertebrate samples were collected from 33 lakes in ecoregion 50 from 1981 to 2010.
Most of the correlations between invertebrate metrics and nutrient concentrations were relatively weak.
However, an interesting relationship was found between total number of invertebrate taxa and low DO
condition or proportion of low DO in the lakes (Figure 2-21), which was counter-intuitive. Considering
the different zones in which the chemistry and invertebrate samples were taken and different littoral vs.
TETRATECH
33
-------
Downstream Use Protection
April 2015
open water habitat, we do not consider these associations causal and they may actually be reflecting
differences in productivity not captured in the data.
16 -
14 -
OJ
£ 12
£
In
-g 10 -
ID
.0
ZJ
6 -
4 -
2 -L
R2= 0 27
(to o
o
.0
U
16 -
14 -
12 -
10 -
8 -
6 -
4 -
R2= 0.37
0.05 0.10 0.15
Proportion of Samples with DO < 2 mg/L
Annual Minimun DO
Figure 2-21. Relationship between invertebrate taxa richness in Hester_Dendy samples and a)
minimum DO proportion in a water column; 2).annual minimum DO in water column.
A total of 208 of 268 zooplankton samples from 16 lakes were available from ecoregion 50 in the dataset.
The majority of these samples were collected from six of these lakes, including Allequash Lake (31), Big
Muskellunge Lake (32), Crystal Lake (33), Little Rock Lake (36), Sparkling Lake (33), and Trout Lake
(33). Due to the size of the lakes and different sampling years, we treated each sample as independent.
Tolerance values were previously developed for zooplankton taxa (Paul et al. 2014). We associated
zooplankton abundance and richness with annual geometric mean nutrient concentrations in these lakes
(Figure 2-22). Both TN and TP correlated with abundance of tolerant taxa in this dataset. In addition,
tolerant taxa abundance declined with mean DO concentrations, but increased with the proportion of DO
measurements less than 2 mg/L.
TETRATECH
34
-------
Downstream Use Protection
April 2015
0.35 -
0.30 -
CD
X
CD
t 0.25 -
CD
-------
Downstream Use Protection April 2015
A number of environmental variables, including DO, proportion hypolimnetic hypoxia, chlorophyll a and
nutrients were correlated to fish metrics. However, the associations between calculated fish metrics and
environmental variables were relatively weak (|r| < 0.2) and, therefore, not pursued.
2.2.4 Nutrient Endpoint Summary
Nutrients were related to chlorophyll and chlorophyll to dissolved oxygen, especially hypoxia. Dissolved
oxygen and chlorophyll, however, were only loosely related to most direct biotic measures in these lakes,
except zooplankton. We believe these were due to sample size (phytoplankton) and methodological
(benthic invertebrates) reasons. Zooplankton exhibited the strongest responses, but none that provided a
directly extractable association with DO that could be linked to chlorophyll.
As a result, for deriving nutrient targets, we relied mostly on the relationships of chlorophyll to DO and
nutrients to chlorophyll. DO values less than 2 mg/L represent hypoxia and values less than 5 mg/L are
stressful to many fish species and are the default dissolved oxygen criteria in WI [WINR 102.04 (4) (a)].
The analyses above indicate that instantaneous maximum chlorophyll above 10-20 (ig/L, and average
chlorophyll concentrations of approximately 5 (ig/L are associated with increased risk of hypolimnetic
hypoxia (< 2 mg/L) and epilimnetic and water column means less than 5 mg/L DO in these cool dimictic
lakes. Mean chlorophyll a concentrations of 5 (ig/L are generally associated with TP concentrations in the
range of 20-30 (ig/L (Figure 2-8 and Figure 2-23) and these are consistent with TP predictions using
global chlorophyll-TP equations for lakes (OECD 1982). In addition, change points in annual mean and
summer mean TP and TN relationships against chlorophyll occurred at 20 (ig/L TP and 500 (ig/L TN
respectively (Figure 2-23 and Figure 2-24). Therefore, a target TP value of 20-30 (ig/L for the Holcombe
Flowage reservoir would be defensible. Lastly, WI has adopted a TP criterion of 30 (ig/L for reservoirs
such as Holcombe Flowage. Again, such a concentration would appear to be protective of sensitive
aquatic life based on this independent analysis.
TETRATECH
36"
-------
Downstream Use Protection
April 2015
50
0.018-0.02-0.021
0.02-0.021-0.023
ra 10 -
.c
O
5 -
50 -
10 -
5 -
R==0.47
0.005 0.01 0.02
Total Phosphorus (mg/L)
|
0.005 0.01 0.02 0.05
Total Phosphorus (mg/L)
Figure 2-23. Annual mean and summer mean chlorophyll a vs. TP relationships. Each data point
represent mean value.
0.443-0.548-0.587
0.329-0.674-0.678
50 -
.
o
10 -
5 -
0.2
0.5
Total Nitrogen (mg/L)
0.2
Total Nitrogen (mg/L)
0.5
Figure 2-24. Annual mean and summer mean chlorophyll a vs. TP relationships. Each data point
represent mean value..
TETRATECH
37
-------
Downstream Use Protection
April 2015
2.3 Results - Tennessee
2.3.1 Nutrient and Chlorophyll a
In total, there were 2,686 station samples in the dataset. After averaging water chemistry data across
relevant months, there were 397 station annual averages (a median of 7 values per station) from which we
determined nutrient - chlorophyll a relationships.
As expected, epilimnetic chlorophyll a concentration increased with increasing total phosphorus (TP) and
total nitrogen (TN) concentrations (Figure 2-25). Both linear models were significant (p< 0.01).
A. Chlorophyll a vs. total phosphorus
B. Chlorophyll a vs. total nitrogen
0.507
• * ..I
£ •
logTP=-0136
TP=0.04 mg/L
g>
0.424
log TN=-0 15
TN=0 71 mg/L
-2.5 -2.0 -1.5
Log total phosphorus mg/L
-1.0
-1.0 -0.8 -0.6 -0.4 -0.2
Log total nitrogen mg/L
0.0
Figure 2-25. Average summer chlorophyll a values in response to nutrient concentrations
averaged across spring and summer. Both linear relationships (green lines) are
significant (p-value < 0.01). R-squared values for the linear models are in the top left
corner.
2.3.2 Chlorophyll a and Dissolved Oxygen
Of the 2,686 station samples in the dataset, 1,597 were collected in summer months from which the
chlorophyll - DO relationships were developed. The proportion of late summer vertical profile DO
samples in the hypolimnion <5 mg/L or 2 mg/L, and in the water column and epilimnion less than 5 mg/L
as a function of mean growing season chlorophyll were modeled. Other time periods gave similar models,
but linking summer DO response to growing season mean algal biomass made theoretical sense, since it is
a continuous growing season organic load that drives DO. If we had the data to calculate a cumulative
primary production, we would have tested that as well. We define hypoxia as a concentration of 2 mg/L,
a concentration many species will avoid and one that has been used frequently as a breakpoint for hypoxia
(USEPA 2001). The value of 5 mg/L is a stressful level for fish and is the state of TN water quality
I TETRATECH
38
-------
Downstream Use Protection
April 2015
standard for DO. Logistic regression models revealed decreasing oxygenated proportions of the
hypolimnion, water column and epilimnion with increasing chlorophyll a concentrations (Figure 2-26 and
Figure 2-27).
A, Proportion of hypolimnion with DO > 5 mg/L
B. Proportion of hypolimnion with DO > 2 mg/L
C OD QfXQ OCDOO OOOI
o o
0 o 0 °
ff^^VC^f*1
(ft> O 001 f» ~ °
0.0 0.5 1.0
Log average chlorophyll a concentration •: ng'_;
1.5
0.5 1.0
Log average chlorophyll a concentration (ug'L)
1.5
Figure 2-26. A. Proportion of DO samples > 5mg/L in the hypolimnion in response to epilimnetic
chlorophyll a averaged across spring and summer. B. Proportion of DO samples >
2mg/L in the hypolimnion. Both logistic regression models (red) were statistically
significant (p < 0.01). Mean values for mesotrophic (4.7 ug/L), eutrophic (14.3 ug/L),
and upper eutrophic (25 ug/L) lake chlorophyll a values are indicated with dashed grey
lines (EPA 2000, OECD 1982).
TETRATECH
39
-------
Downstream Use Protection
April 2015
A. Proportion of water column with DO > 5 mgIL
Meso En Upper Eu
B. Proportion of epilimnion with DO > 5 mg/L
to
o
0
00
0.5
10
15
ra
§
I
I
O O OOO OCO-B OO OOC O O
5mg/L in the water column in response to epilimnetic
chlorophyll a averaged across spring and summer. B. Proportion of DO samples >
5mg/L in the epilimnion. Both logistic regression models were statistically significant
(p-value < 0.01). Very few lake profiles had DO concentration in the epilimnion < 5mg/L
below a chlorophyll a concentration of 4.7 ug/L.
2.3.3 Biota and Dissolved Oxygen, Chlorophyll a, and Nutrients
Because we were interested in identifying conditions most stressful for fish in terms of DO, we narrowed
subsequent analysis to the month of August for which 1) stratification as measured by relative thermal
resistance (RTR) was strong and 2) dissolved oxygen levels were lowest on average (Figure 2-28).
TETRATECH
40
-------
Downstream Use Protection
April 2015
Dissolved oxygen (mg/L)
« 19 3* » M
Dissolved oxygen (mg/L)
Dissolved oxygen (mg/L)
RTH=16
. .
* if
* Tenure laluie
* * * DissoK-ed oxygen
15 20 30
Temperature (C)
£
s. "
I
? -
R
, ^RTR=223
*
* o
* •?
•
« »
« 4
*
•J Temper Hlure
* * • DissohnBd oxygon
0 10 20 3D
Temperature (C)
"E
I *-
Q
S
A
RTR=95
* •>
* 0
*
* 6
* o
* 0
* *
* «
° Ternperuture
* * * DtssohAx) oxygon
0 10 20 3D
Temperature (C)
Figure 2-28. Lake temperature profiles from the same sampling station in one year in April (left),
August (middle), and October (right). The thermocline is represented as a dashed line.
The proportion of the water column for which dissolved oxygen (DO) is less than 5
mg/L, effectively reducing habitat size for intolerant fish species, is highest in the
middle panel (August). The relative thermal resistance, a unitless index of stratification
strength, is printed in the upper right corner.
We also noted that tributary storage reservoirs and main stem reservoirs had significantly different
richness and Shannon-Wiener index values (p < 0.0001, one-way ANOVA, Figure 2-29). Therefore, we
relativized those metrics to the class means).
S
Figure 2-29. Reservoirs on the main stem of the river had higher fish biodiversity indices than
tributary storage reservoirs.
TETRATECH
41
-------
Downstream Use Protection April 2015
We first examined the relationship between fish species diversity indices and the proportion of DO
samples > 5 mg/L in the hypolimnion and in the water column and minimum August DO. In general these
relationships were weak, however, that was a slight tendency for decreased species richness at stations
with a lower proportion of DO values > 5 mg/L in the hypolimnion and in the water column (Figure 2-
30).
TETRATECH
42"
-------
Downstream Use Protection
April 2015
A.
CO
in
"ro
B.
in
in
01
c
0
01
>
^
o
QL
C.
^ •* ***• «A» • •« •
JH* •*•*• ••*•« (* *
A
* * • • Main
M> • •» *
*. • Trib
i i i i i
02468
Minimum August DO (mg/L)
* . R2=002 ^ «
• •
t . s
}'..*•
Irt^A^A^L
* *•* * * *
• • • • *» *• • •
*• •••• * •
• * * • Mair
'•Trib*
i i i i i
0.0 0.2 0.4 0.6 0.8 1.C
Proportion hypolimnion DO > 5 mg/L
• • R*=0.03
4t* *» *
* • * •
• •• •* • t
-V ... . !
*\ • • • •
^S^^Tj^/ *J :
• /••* *. • i
• •
. **".* . * •
• • • • * I Mail}
• • « Trib
i i i i i
3.0 0.2 0.4 0.6 0.8 1.C
X
~° -d-
c t _
J_
is
rv
Dt CM
o
X
01
0>
c -
01
> 0
0
c
c -
(D
.c
Q —
«
01
Q: CM
O
1
X
0)
j_
c: —
5 0
c ^~
o
c
c: —
CD
J=
CO CD
5 mg/L
. •
•* • •
•M . . .
* * *i*I? *.r* * **^ •*
*^'*» **isv *^r *^»
*1*5 *«*****'**•• •**
• *)••***:•' •».*/
« •* •
• % Main
• Trib
i i i i
).0 0.2 0.4 0.6 0.8 1.C
Proportion water column DO > 5 mg/L
Proportion water column DO > 5 mg/L
Figure 2-30. Fish diversity indices - species richness (left) and evenness (right) - are expressed as
a function of A. minimum August DO values, B. proportion of water column samples
with DO > 5 mg/L and C, proportion of hypolimnion samples with DO > 5 mg/L
TETRATECH
43
-------
Downstream Use Protection
April 2015
Fish richness decreased slightly with an increasing proportion DO > 5 mg/L in the hypolimnion and the
water column (p < 0.05). No other relationships were statistically significant.
We narrowed the analysis to game fish only (Table 2-2). Again, there were no strong relationships
between measures of hypoxia and fish diversity indices. However, the Shannon-Wiener index (a measure
of evenness) increased with every measure of increasing oxygenation (Figure 2-31).
Table 2-2. Fish species identified as "game fish" for the purposes of this analysis.
American Fisheries Society
(AFS) Common Name
AFS Scientific Name
Black crappie
Bluegill
Channel catfish
Flathead catfish
Largemouth bass
Redbreast sunfish
Redear sunfish
Smallmouth bass
White crappie
Yellow perch
Pomoxis nigromaculatus
Lepomis macrochims
Ictalurus punctatus
Pylodictis olivaris
Micropterus salmoides
Lepomis auritus
Lepomis microlophus
Micropterus dolomieu
Pomoxis annularis
Percaflavescens
TETRATECH
44
-------
Downstream Use Protection
April 2015
A.5-
2 -
en
(n CM
« - -
_c
« -^
Ei
- op
,S o
o ~
4.
ci
«* •» *
• • • *• **
• «• * «*•* • • *« » •. •
**•<*• • ** ** ^"J"
» * * • •• *» • • »
• • Main
» Trib
i i i i i
02468
X
•****.**. » •* : *
^* ••* *. • »•*•, *. *.
• * **v^ **5i * *• .»j— — - •
'V_|i_*— — yj •»5^^«» • %l
^ v» ••
5/.- * . • •* •
*.. ..«.*•
f • •
• Mam
• Trib
i i i i
2468
Minimum August DO (mg/L)
Minimum August DO (mg/L)
B. 2 -
•*
in
<" ^i
a> w
C r-
r:
o
-c o
"Jo co
"^ o
Of
<£>
o
o ~
c 5 -
•* _
tf)
£ « _
£ ^
o
C 0
^
? CO _
a> o
rS*
EC,
CD
0
• •**« • • • • •
t % •«•• **«t^ •••«•• 1* • 9
• ••«• ••••• •
• * •* *••*• MW4* » •»** *
• • • • • •
^ • Main
• Trib
i i i i i i
0.0 0.2 0.4 06 0.8 1.0
Proportion hypolimnion DO > 5 mg/L
...... • .
. ..4t . ....
V *«»*•***•*•*«*• \ff 9
• «•«* ..«• .«•* .
•«• * • * • •
. • Main
• Trib
i i i i i i
3.0 0.2 04 06 08 1.0
Proportion water column DO > 5 mg/L
X
•S
TD
— to
flj T~
C
' CD
O "?~ ~
*•
C
1
° ~
i
s
0- CD
o
«
1
_E ^o
te "
(D
>
0 «- ~
1
(/3
1 6 -
~ra
OJ
o:
o
(
s R2= S.evf .
1 .. > ' •'•"• * * •
> ^ • .
. •• ... *
* . • * .
•t * . f .*
^ — ' *•
_t * J^ *- — ^*~^ • *
^ +
• A* **
' •» *
. v * «• *t •
• . • *
»• * *
• Main
•* Trib
i i i i
JO 02 0.4 0.6 08 1.0
Proportion hypolimnion DO > 5 mg/L
n .
R2= 0.05 • .
****** * '
** ** •• *v **•
• . ** «•*• • . *
*» » **% *' * '*** «*
!_#^^^**~^^"^S^ * ^
. * ^ *.
*• •! *• •* **
. . ^^ . . • . f .
. . .
j»**» • .
*t .* • Main
•*Tnb
i i i i
J.O 0.2 0.4 06 08 1.0
Proportion water column DO > 5 mg/L
Figure 2-31. Game fish diversity indices - species richness (left) and evenness (right) - as a
function of A. minimum August DO values, B. proportion of water column samples
with DO > 5 mg/L and C., proportion of hypolimnion samples with DO > 5 mg/L.
Evenness increased with higher minimum DO values, an increasing proportion DO > 5
mg/L in the hypolimnion and the water column (p < 0.05). No other relationships were
statistically significant.
TETRATECH
45
-------
Downstream Use Protection April 2015
Finally, we looked at individual taxa - largemouth bass and bluegill - in relation to measures of lake
oxygenation (Figure 2-32). The relative abundance of bluegill decreased with increasing oxygenation as
measured by all three parameters: minimum August DO, and proportion of the hypolimnion and of the
water column with DO > 5 mg/L. In contrast, largemouth bass relative abundance decreased as oxygen
decreased. While low DO levels were not likely beneficial to bluegill, they may have reflected higher
primary productivity in the lake or reduced predation pressure from more oxygen sensitive piscivores,
such as largemouth bass. Because the bluegill were relatively less sensitive to low DO than largemouth
bass, at these levels, there did not appear to be an adverse effect on their abundance. In fact, they
increased. For the largemouth bass, the lower DO appeared to be more important in limiting their
abundance than any effect of increased primary productivity (see below discussion on chlorophyll
relationship).
To further explore these ideas, we also looked at the relationship between the fish diversity indices and
chlorophyll a (Figure 2-33 to Figure 2-35) and nutrient concentrations (Figure 2-36 to Figure 2-38). In
general, these analyses reinforce the relationships noted between fish diversity indices and DO. That is,
fish species richness increased with increasing chlorophyll a concentration for both all fish and game fish
alone (Figure 2-33 and Figure 2-34). However, evenness decreased with increasing chlorophyll a
concentration among game fish (Figure 2-34). This suggests that only a subset of species may have
benefitted from increased primary productivity, while other species were reduced. Bluegill and
largemouth bass relative abundances both increased with increasing chlorophyll a concentration (Figure
2-35).
Results for nutrient concentrations not surprisingly mirror those of chlorophyll a. Fish species richness
increased with increasing TP and TN, but evenness was not affected (Figure 2-36). In contrast, for game
fish alone, evenness decreased with increasing TN, but richness was not affected (Figure 2-37). Bluegill
and largemouth bass relative abundances increased with increasing TN concentration (Figure 2-38).
TETRATECH
46"
-------
Downstream Use Protection
April 2015
A...
R2= 0 09
2468
Minimum August DO (mg/L)
Minimum August DO (mg/L)
B.
o
E
0)
cr
R2=004
00 02 04 06 0,8 1.0
Proportion hypolimnion DO > 5 mg/L
i r
0.0 02 04 06 08 1.0
Proportion hypolimnion DO > 5 mg,'L
C.
-Q
0)
•S CM -
™
QJ
s
cr
2= 0 05
0.0
02
0.4
06 08
1 0
0.0
i
0.2
0.4
0.6
08
r
1.0
Proportion water column DO > 5 mg/L
Proportion water column DO > 5 mg/L
Figure 2-32. Bluegill (blue) and largemouth bass (brown) relative abundances in the fall as a
function of minimum August DO (top) and DO concentration (proportion of samples
with DO > 5 mg/L) in the hypolimnion (middle) and in the entire water column (bottom).
Bluegill abundance declined with increasing DO, whereas largemouth bass abundance
increases with increasing DO. All relationships were statistically significant (p < 0.05).
TETRATECH
47
-------
Downstream Use Protection
April 2015
Main
Trib
R2=0.15
• Main
• Trib
0.4 0.6 0.8 1.0 1.2 1.4
Log chlorophyll a concentration (ng/L)
1.6
0.4 0.6 0.8 1.0 1.2 1.4
Log chlorophyll a concentration (ng/L)
1.6
Figure 2-33. Fish species richness increases with increasing spring-summer chlorophyll a
concentration. There was no apparent relationship between chlorophyll a
concentration and species evenness.
W CN
CO ^J
CD
o
-*
CD
R2= 0.02
• ••
*** *•* *»•»•**•*
Main
Trib
• • • *•»•••••
• •*
* *
•• ••
1.0
1.4
0.4 0.6 0.8 1.0 1.2 1.4 1.6
Log chlorophyll a concentration (jig/L)
i i i
0.4 0.6 0.8 1.0 1.2 1.4
Log chlorophyll a concentration (jig/L)
Figure 2-34. When the analysis was narrowed to game fish alone, species richness increased with
increasing chlorophyll a, but species evenness declined. Statistically this means that
although there might have been more species present, additional fish caught were
more likely to be of the same species, as there were larger numbers of fewer species.
TETRATECH
48
-------
Downstream Use Protection
April 2015
0-5 1.0 1.5
Log chlorophyll s concentration (Lig/L>
0.5 10 1.5
Log chlorophyll a concentration (ng/L)
Figure 2-35. Bluegill and largemouth bass abundances both increased with increasing chlorophyll
a concentrations.
TETRATECH
49
-------
Downstream Use Protection
April 2015
-2.5 -2.0 -1.5 -1.0
Log TP concentration (mg/L)
to
o
OL
Main
Trjb
-2.5
-2.0
-1.5
-10
Log TP concentration (mg/L)
QL
• Main R=0.21 , .
• Trib • •
-1.0 -0.8 -0.6 -0.4 -0.2 0.0
Log TN concentration (mg/L)
Main
Trib
i i i i
-1.0 -0.8 -0.6 -0.4 -0.2 0.0
Log TN concentration (mg/L)
Figure 2-36. Fish species richness (left) increased with increasing total phosphorus (TP) and total
nitrogen (TN) concentrations (p < 0.05). There was no statistically significant
relationship between species evenness (right) and nutrient concentrations (p<0.05).
TETRATECH
50
-------
Downstream Use Protection
April 2015
Main
Trib
-2.5 -2.0 -1.5 -1.0
Log TP concentration (mg/L)
• Main
Trib
•
' *"*5V* ••* * *
».£ •* *•».••,
• • • . * • .*•>•• ti*
•.«>• ViH***
•<:,VV:V?C*
-2.5
-2.0 -1.5 -1.0
Log TP concentration (mg/L)
Main
Trib
-1.0 -0.8 -0.6 -0.4 -0.2 0.0
Log TN concentration (mg/L)
(O
-------
Downstream Use Protection
April 2015
a
•-
c
to
T3
•
•-
£ 03 -
O
O
CO
03
~U
,.R2=0-25
-2.5 -2.0 -1.5 -1.0
Log TP concentration (mg/L)
I I I I
-2.5 -2.0 -1.5 -1.0
Log TP concentration (mg/L)
-2 04 -
.
• J* 7/v* <£%•», •
* ••" ;* »-
I I I I I
-1.0 -0.8 -0.6 -0.4 -0.2 0.0
Log TN concentration (mg/L)
en
(n
03
0)
CO
.
c
03
•a
c
ZJ
. R2=0.14.*
• s •••
.*».»••
./»
Ill
-1.0 -0.8 -0.6 -0.4 -0.2 0.0
Log TN concentration (mg/L)
Figure 2-38. Bluegill and largemouth bass abundances increased with increasing nutrient
concentrations, with the exception of bluegill and TP. Solid lines indicate statistically
significant relationships (p < 0.05).
2.3.4 Nutrient Endpoint Summary
We used the data from the above analyses to derive numeric nutrient targets. Fish species did change in
response to nutrient enrichment, chlorophyll increases, and changes in DO. Of greatest focus, was the
effect of DO. We did not observe a distinct threshold in fish to 5 mg/L, however, richness declined as the
proportion of the water column and hypolimnetic oxygen declined below 5 mg/L (Figure 2-30). DO
concentrations above 5 mg/L are important for fish species, even in warm monomictic lakes. A summer,
littoral DO value of > 5 mg/L represents an optimal level for bluegill (Stuber et al. 1982b) and the level
below which largemouth bass show distress (Stuber et al. 1982b). A substantially higher average value, 8
mg/L, is preferred for normal largemouth bass growth. Perhaps not coincidentally, the Tennessee state
water quality criterion states that "DO shall not be less than 5 mg/L" [Tennessee Rules Chapter 1200-04-
03-.03 (3)(a)], therefore we focused on this DO endpoint in deriving TP and TN targets.
TETRATECH
52
-------
Downstream Use Protection April 2015
Chlorophyll a concentrations consistent with meeting a DO value of 5 mg/L (> 40% of the water column,
greater than 50% of the hypolimnion, and greater than 95% of the epilimnion, Figure 2-26 and Figure 2-
27), were approximately 14 (ig/L, which is also the mean annual chlorophyll a value for eutrophic lakes
(OECD 1982). Epilimnetic samples showed a threshold effect whereby there was a decreasing proportion
of samples with DO concentrations > 5mg/L above a chlorophyll a concentration of approximately 4.7
(ig/L; and this proportion declined below 95% above 14 (ig/L (Figure 2-27). For the hypolimnion, the
preferred summer habitat for some game species that are stressed by higher epilimnetic summer
temperatures, the model predicted, at a target chlorophyll a value of 14 (ig/L, just over 50% of the
hypolimnion will maintain a DO value > 5 mg/L (Figure 2-26). Furthermore, at the same target
chlorophyll a value of 14 (ig/L, about 70% of the hypolimnion will maintain DO values > 2mg/L (Figure
2-26). This would suggest sufficient habitat protection of hypolimnetic habitat if average chlorophyll
values are maintained at 14 (ig/L for Douglas Reservoir.
We next converted this average summer growing season chlorophyll a endpoint of 14 (ig/L into nutrient
targets. The nutrient-chlorophyll models for this dataset were similar in precision to other chlorophyll
yield models (OECD 1982). The phosphorus model developed (Figure 2-25) predicts a TP concentration
of approximately 0.04 mg/L TP to meet the 14 (ig/L chlorophyll a endpoint. The same chlorophyll
endpoint was associated with a target TN concentration of 0.71 mg/L. We believe meeting these average
nutrient concentrations will result in chlorophyll levels that will prevent DO concentrations that threaten
fish, especially important game species.
3 DOWNSTREAM PROTECTION VALUES: STATISTICAL
ANALYSIS
The basic question for Downstream Protection Values (DPVs) is the following: Suppose there is a
nutrient criterion for a lake, defined as an average seasonal or annual concentration. Can this be used to
define an acceptable concentration (a DPV) in streams where they enter a lake? Then, given targets at the
stream mouth, what does this imply about allowable concentrations at monitoring points upstream in the
flow network?
These questions are more complicated than first seem at first. First, the average concentration in a lake is
determined by mass load, residence time and internal nutrient processing, not by raw concentration series
of the inflows. Second, when there are multiple streams entering a lake, or when we look at monitoring
stations higher up in the network, it may matter little if one tributary has a high concentration because it is
really the total load that determined downstream water quality..
There are two end members of this problem that are of lesser theoretical interest. On the one hand, if one
simply said that the stream DPVs always had to meet the lake criterion it is highly likely that the lake
criterion would be met (unless there is significant loading from internal regeneration of nutrients from the
sediment or atmospheric deposition), but would be imposing much more stringent requirements than
actually needed. At the other end, the issue could be addressed on a site-specific basis through the
construction of a detailed watershed simulation model, as has been done in many lake TMDLs. That type
of approach can yield accurate results provided that model itself is accurate, but at a high cost. The
interest is in evaluating more cost-effective approaches that lie between these extremes. Such an
approach could be valuable for protecting large numbers of streams and receiving water lakes, potentially
including waters attaining uses, before a costly TMDL is needed or can be funded.
TETRATECH
53
-------
Downstream Use Protection April 2015
3.1 Theoretical Issues
The stated objective of this work is to derive "numeric nutrient criteria for flowing waters that will
support attainment of identified numeric criteria and aquatic life uses in downstream lakes." EPA
expressed specific interest in the issue of allocating the load that protects a downstream lake into the
upstream subwatershed areas. The criteria need to be evaluated relative to concentrations observed at
points in time and space, C(tiX^ so that it can be used to assess use support or to derive permit limits for a
discharge.
Criteria to protect uses in lakes are most naturally expressed as a seasonal average concentration in the
epilimnion of the lake. This seasonal average concentration arises from the mass balance of inputs and
outputs and is related more closely to the influent load than to the influent concentration. For example,
the in-lake concentration could arise as a result of one day on which 1000 kg was loaded by a flow of 20
m3/s and 99 days on which 0 kg was loaded by a constant flow of 1 m3/s. The average influent
concentration (0.058 mg/L) is not the same as the total load divided by the flow (0.097 mg/L).
Working backward from the in-lake target to the stream input involves the following steps:
1. In-lake, seasonal average concentration target is based on protection of the uses.
2. A mass balance analysis converts the in-lake concentration target to a net loading rate.
3. If the rate of net recycling between the epilimnion and the hypolimnion and sediment can be
estimated, the net loading rate can be converted to an estimate of external load.
4. After further correcting for any direct atmospheric and groundwater loading sources, the external
load can be converted to a seasonal watershed load.
5. Dividing the seasonal watershed load by the seasonal watershed flow yields a seasonal flow-
weighted concentration target at the point where the stream enters the lake.
6. The flow-weighted concentration, CFW, is related to the instantaneous concentration at the
downstream pour point, C^^, as CFW = J C^o) Q(t,o)dt/ J Q(t,o) dt. This does not readily yield
the distribution of C(t/0) unless the relationship of C(t)to Q(t) is known and either the distribution
of Q(t,o) is known or C(t 0) and Q(t,o) are independent. These conditions generally do not hold, in
which case there is not a unique solution for the distribution of C(t0) time series. However,
sufficient statistics might be derivable.
Several issues are raised by this analysis:
Issue 1: Relationship of CFW and C(t;0). There will be a general desire to relate observations of
downstream concentration, C(t 0) to the DPV, which is really a flow-weighted concentration, C(FMr). As
noted above, it may be difficult to make inferences about what individual values of C(t;0) should be based
on C(FMr) unless the relationship of C(t,o) and Q(t,o) can be described. In reality, flow and concentration
are typically not independent and positive and/or negative correlations often exist in different parts of the
hydrograph. These relationships may also differ for N and P, by location, and according to the principal
sources of nutrients. Nonetheless, the relationship is obviously not completely uninformative. For
instance, if C(t/0) and Q(t,o) were observed such that the product exceeded CFW * / Q(t,o) dt (i.e., the load
on that day was more than the total implied by the flow-weighted concentration), then the observed
would clearly be incompatible with the target expressed as
TETRATECH
54~
-------
Downstream Use Protection April 2015
The Florida Inland Rule (which is discussed in more detail in Section 3.2.1 below) seems to define the
DPV as an annual geometric mean - but it would appear that an annual flow-weighted concentration is
what is really intended or should be intended. If a simple geometric mean is used this has the advantages
of establishing a mathematical relationship between any C(t,o) and the DPV target because the correlation
with Q is ignored.
There are distributional and regression approaches to addressing this problem. The distributional
approach would assume a distribution form for C(t/0) that is consistent with meeting C(FMQ. We could
then test the hypothesis that an observation, COBS, is consistent with being drawn from the distribution
of C(t 0) | C(FMQ. The problem is that the actual form of C(t;0) is unlikely to follow a clean parametric
distribution that could be described under a null hypothesis as having C(FW^ as a sufficient statistic.
Indeed, the distribution is likely to be a mixture - i.e., dilution of point sources and groundwater plus a
storm component positively correlated with flow.
Another approach would be to create a stratified regression estimator of C(t,o) = f(Q> 0) using a tool like
LOADEST (Runkel et al., 2004) or the Weighted Regressions on Time, Discharge, and Season (WRTDS)
method (Hirsch et al., 2010). Stratified regression allows different relationships in different flow regimes.
LOADEST creates confidence bounds on the regression and one could test whether C(oss) \Q falls within
the confidence bounds. Because deviations at low flow don't contribute much to load it may be better to
run this sort of test on load as a function of flow - although this gets into issues with flow being on both
sides of the equation such that autocorrelation is introduced into the error structure.
Issue 2: Translating back into the watershed. Given a DPV at the pour point into a lake, how is this
distributed back to concentration targets in the headwater reaches? There are losses in transit, although
these may tend to be small if we are interested in true losses rather than temporary retention within the
stream network. These losses could be accounted for by using a discount or decay factor, as in the USGS
regional SPARROW applications. However, what would observing a concentration greater than some
implied target signify in a headwater reach if a load from a single headwater segment is only a small
fraction of the total load to a downstream lake? One approach would be to say that all headwater
segments must meet the (discounted) criterion derived from the downstream DPV - thus ensuring
compliance. However, it would also be possible to perform a statistical analysis of the allowable
distribution of headwater concentrations that is likely to be consistent with achieving the DPV.
A significant problem is that we are likely to have measurements at only a few scattered points so it
would be difficult to infer if an elevated concentration at one upstream monitoring location was
significant relative to achieving the DPV, which integrates across the entire upstream network.
The simplest approach is to require that the expectation of the upstream concentrations be consistent with
the pour point concentration distribution: Eic i = C(t/0)/R, where R is the discount factor accounting
for reduction between location x and the pour point. This works, but as noted above is not really
appropriate. Instead, we want the sum of the upstream loads to meet the loading target, i.e.:
// L(t,X)dt dx or J/[C(t^) * Q(t,x)] = CFW * E(Qi0).
3.2 Existing Approaches to DPVs
TETRATECH
55
-------
Downstream Use Protection April 2015
3.2.1 Florida Inland Rule
The Florida Inland Rule DPV approach is one variant of an attempt to define DPVs. Specifically, the FL
2010 final inland rule and (second) 2012 proposed inland rule establish DPVs at the point of flow into a
lake using the BATHTUB model or other tool. The TSD for the original proposed rule presents some
calculations to convert an in-lake P target using the Vollenweider approach into a DPV concentration at
the pour point, where the conversion was based on the lake residence time and the fraction of inflow due
to surface flow. In all three cases, the DPV is defined as an annual geometric mean concentration.
Although not stated, the procedure shown in the original proposed rule TSD calculates a flow-weighted
mean concentration (total load divided by total flow), and not the geometric mean of the concentration
time series. These calculations are omitted in the TSD for the final rule.
Both the 2010 rule (75 FR 75762) and the 2012 proposed rule (77 FR 74985) note, but do not solve the
problem of translating the DPV back into the watershed. The issue is described as follows at 77 FR
75001:
EPA selected the point of entry into the lake as the location to measure water quality because the
lake responds to the input from the pour point, and all contributions from the stream network
above this point in a watershed affect the water quality at the pour point. When a DPV is
exceeded at the pour point, the waters that collectively comprise the network of streams in the
watershed above that pour point are considered to not attain the DPV for purposes ofCWA
section 303(d). The State may identify these impaired waters as a group rather than individually.
Contributions of TN and/or TP from sources in stream tributaries upstream of the pour point are
accountable to the DPV because the water quality in the stream tributaries must result in
attainment of the DPV at the pour point into the lake. The spatial allocation of load within the
watershed is an important accounting step to ensure that the DPV is achieved at the point of
entry into the lake. How the watershed load is allocated may differ based on watershed
characteristics and existing sources (e.g., areas that are more susceptible to physical loss of
nitrogen; location of towns, farms, and dischargers), so long as the DPV is met at the point of
entry into the downstream lake. Where additional information is available, watershed modeling
could be used to develop allocations that reflect hydrologic variability and other water quality
considerations. For protection of the downstream lake, what is important is an accounting for
nutrient pollution loadings on a watershed scale that results in meeting the DPV at the point of
entry into the downstream lake.
As in the December 6, 2010 final rule, this proposal provides that additional DPVs may be
established in upstream locations to represent sub-allocations of the total allowable loading or
concentration. Such sub-allocations may be useful where there are differences in hydrological
conditions and/or sources of TN and/or TP in different parts of the watershed.
The last paragraph reflects a concern that equal distribution of the DPV back up into the watershed may
be unattainable for watersheds in which portions are affected by intensive anthropogenic uses. One
potential way to address this issue would be to assign upstream DPVs as 100 percent of natural
background plus some proportionate reduction of anthropogenic load, at least for estuarine endpoints.
The concept makes some sense in theory: First, figure out the natural condition delivered (annual) load
from each upstream subwatershed (where the rates may differ according to local conditions in different
subwatersheds). Then take the remaining assimilative capacity (the difference between the natural annual
load and the DPV load) and distribute it as a proportion of the existing anthropogenic loads - thus taking
an equal reduction approach. This is analogous to a typical TMDL implementation planning approach
TETRATECH
56"
-------
Downstream Use Protection April 2015
and can have the same high bar of modeling/analytical work. Equal reduction is only one option for the
anthropogenic loads. There are various alternatives, such as obtaining reductions at the cheapest cost,
with the greatest degree of assurance, or through a market-based cap-and-trade approach.
Note that this kind of approach is intrinsically load-based, not concentration-based. It should be based on
a summary measure (e.g., geometric mean) of the flow-weighted concentrations, not the simple observed
concentrations, just as in the case of the DPV at the pour point.
3.2.2 Other States
A survey of regulations in states other than Florida revealed frequent acknowledgments that discharges
must be regulated to control uses in downstream waterbodies. For instance, Arizona's rules say "the
Director shall take into consideration the applicable water quality standards for downstream waters and
shall ensure that the water quality standards that are established for an upstream surface water also
provide for the attainment and maintenance of the water quality standards of downstream waters."
Georgia and some other states have established by rule nutrient loading criteria for lakes that are binding
on the upstream watershed. We did not, however, find any cases in which states have proposed detailed
procedures for establishing upstream nutrient targets based on protection of downstream uses.
3.3 From Lake Targets to Influent Loading Limits
As noted above, nutrient targets for lakes are expected to be expressed either as loading limits or as
seasonal or annual average concentrations. Specification of an average concentration (or integrative
measure of concentration) depends on the load along with the mixing and reactive characteristics of the
lake, rather than the instantaneous concentrations of the influent. Thus, lake targets are most readily
transformed into loading targets in the influent.
Translating a lake nutrient target to an influent load is a tractable problem, although not without
uncertainties. The major difficulty is likely to lie in properly accounting for loading that does not enter
the lake through streams - including atmospheric deposition, direct groundwater inflow, and internal
loading from lake sediment sources.
Going from a concentration target to a load requires an accounting of lake flushing and nutrient loss rates.
Options range from simple equations to fully detailed lake models. In developing the 2010 Phase I
Florida Inland Rule, EPA proposed use of an empirical formulation (Vollenweider 1975, 1976) to make
the translation.. Florida proposed use of a BATHTUB model approach instead, and it is that approach
that is described in the final rule (FR 75(233):75782, December 6, 2010). Both will be described here.
The final rule also explicitly allows other options, including detailed, calibrated lake models such as
WASP. We do not discuss detailed model applications for this purpose, but do consider additional
medium complexity options.
The 2012 Phase II Florida Inland Rule (FR 77(243):74985-75005, December 18, 2012) retained the
BATHTUB and WASP approaches, but also added several other options, including a linear regression
relationship between stream and lake concentrations specifically developed for Florida, default values for
unimpaired lakes, and a reference condition approach.
3.3.1 Vollenweider Approach
The simplified Vollenweider approach was omitted from the final version of the Technical Support
Document (TSD) for the 2010 rule (USEPA, 2010); however, it appears in earlier drafts. The following
material is taken from the January 13, 2010 unpublished draft of the TSD:
TETRATECH
57
-------
Downstream Use Protection April 2015
The analysis compares TP criteria in streams and rivers to those proposed to protect lakes using the
simple phosphorus loading model developed by Vollenweider (1975, 1976) based on a nutrient
budgeting approach. It has been updated with adjustments made by other authors, including some
specific to Florida (Larsen andMercier 1976; Canfield and Bachmann 1981; Baker et al. 1985;
Steward and Lowe 2010). A similar approach could be used for nitrogen criteria comparison using
comparable nitrogen loading models developed for lakes (Shannon andBrezonik 1972).
If the benchmark nutrient criterion is higher than the stream concentration necessary to protect a
lake, the benchmark criterion will not be protective of the lake. Accordingly, critical stream
concentrations derived by loading models to lakes based on proposed lake criteria that are lower
than the benchmark stream criteria would indicate potential concern for downstream protection. The
Vollenweider model for loading (Vollenweider 1976; Larsen andMercier 1976) follows:
L(P) = [TP]Lf- (1 + TW) /27
t-w
Where,
L(P) is loading rate, g m~2 y'1
[TP]i is lake mean TP concentration, g m3
z is mean depth, m
TW is hydraulic retention time, y...J
The term (1 + TW) represents the phosphorus settling rate
TP load for a lake is the loading rate times the lake surface area:
TL(P)=L(P)xA [3]
Where,
TL(P) is total load, in g/y
A is lake surface area, m2
It is assumed that total stream/low into a lake (Qt, in m3 y'1) cannot be more than the net outflow, or
lake volume divided by retention time:
Qi<^ [4]
Tw
or,
_ zA
Qi = cf-
Tw
where c/is the fraction of inflow due to stream/low, 0 < c/< 1, and includes all streams flowing into
the lake.
If there is net evaporation from the lake, this assumption would be untrue; but in Florida direct
rainfall exceeds evaporation, so Eq. 4 holds.
It is further assumed that all phosphorus loading is from the stream. Then,
- TL(P) -
- Qi ~
TETRATECH
58"
-------
Downstream Use Protection April 2015
Rearranging,
TL(P)TW
[TP]S =
:ZA
J
Substituting Eq.3for TL(P):
= Hfllw
CfZ
Substituting Eq. 2forL(P):
Tw Cf Z
resulting in:
pro] _ _Lprp] r-\ +T ~\ rsi
\_l r\s — c I1 r\L \L^iw) LJJ
By substituting the lake critical TP concentration into [TP]i, derived from chlorophyll-nutrient
regressions, the maximum stream TP, [TP]S, can be estimated for values ofc/andTw. Because c/is
between 0 and 1, the most stringent critical inflow concentration will occur when c/is close to 1 (i.e.,
when a lake's inflow is almost entirely due to stream/low).
Florida lakes tend to be shallow, and internal loadings to the water column following a decrease in
external loading (e.g., from storm-induced resuspension or diagenetic release of sediment
phosphorus) may be substantial. Resuspension or diagenetic release of sediment phosphorus could
be modeled to simulate this phenomenon mechanistically, and it would likely require substantial site-
specific data for calibration. EPA notes that simpler alternatives, such as excluding the settling/loss
term from Equation 5, or reversing the sign on the settling/loss term so that it becomes a net source
term. For general regional application, however, EPA used the model described in Equation 5.
3.3.2 BATHTUB Approach
In the TSD for the 2010 Final Rule, "EPA is specifying that, where sufficient data and information is
available, the U.S. Army Corps of Engineers BATHTUB reservoir model (Walker 1999) may be used to
compute downstream protective values (DPV) necessary in the stream to protect a given lake."
BATHTUB performs steady-state water and nutrient mass balance calculations, accounting for advective
and diffusive transport, to predict annual or seasonally averaged nutrient concentrations. The BATHTUB
model (both Windows and DOS versions) is available from the U.S. Army Corps of Engineers
Waterways Experiment Station at
http://el.erdc .usace.army.mil/products.cfm?Topic=model&Type=watqual.
Application of the BATHTUB model to calculate influent loads consistent with meeting an in-lake target
consists of two steps: (1) development of a calibrated representation of lake response based on observed
loads of water and nutrients, and (2) adjustment of influent concentrations until the specified in-lake
criteria for TN, TP, and chlorophyll a are predicted to be attained. Note that the concentrations that are
specified are flow-weighted concentrations over the period of interest; thus the user specification of
concentration in combination with the specified inflow volume is actually a statement of the allowable
mass loading rate.
Use of the BATHTUB model requires calibration. Calibration of BATHTUB can take place at varying
levels of complexity depending on the availability of data. The user must first assemble or estimate
available information on lake morphometry, inflow, outflow, and observed water quality. For many
TETRATECH
59
-------
Downstream Use Protection April 2015
smaller lakes, representation of the waterbody as a single mixed segment may be sufficient; however,
multiple segment models can be developed for more complex lakes.
The general sequence for the calibration of BATHTUB is described by Walker (1987, 1999) and consists
of the following steps:
1 . Determine appropriate averaging period over which water and mass balance calculations are
performed. This may be annual or seasonal. Criteria in terms of the calculated nutrient
turnover ratio are provided by Walker (1999).
2. Calibrate mass balance for water.
3 . (Optional) For multi-segment lake representations, calibrate rates of constituent exchange
between lake segments for a conservative tracer.
4. Calibrate mass balance for nutrients over the appropriate annual or seasonal averaging period.
Calibration factors should be adjusted only within the ranges recommended by Walker.
5. After obtaining a satisfactory representation of growing season nutrient concentrations,
calibrate chlorophyll-a concentrations and Secchi depth.
In work in support of the 2010 Rule, Tetra Tech (unpublished) tested a rapid, simplified BATHTUB
application to approximately 40 lakes throughout Florida, selected to represent the lake classes, different
trophic states within the lake classes, and a regional distribution of lakes that is representative of Florida's
lakes. This approach used only already-assembled water quality data and estimates of inflow based on
area adjusting the nearest USGS gage records. Detailed morphometric data were not available, and lake
volumes were estimated using an assumed average lake depth of 1.5 meters. Even with these simplifying
assumptions, a reasonable fit to observed water quality was obtained for the majority of the lakes. For
TP, the most restrictive DPVs calculated by BATHTUB generally fell between the in lake criterion and
the instream criterion for the region. For TN, DPVs were only occasionally lower than the instream
criteria. It is noted, however, that some of the rapid model applications were not operating within realistic
ranges for calibration adjustments because general assumptions were not adequate to represent the lake.
For these waterbodies, more site-specific information would be required to set up accurate BATHTUB
models and establish DPVs.
The TSD does not describe the details of the BATHTUB simulation of in lake nutrient concentrations and
the Rule and TSD do not recommend specific choices of BATHTUB model options. The BATHTUB
simulation represents steady state concentration as a function of loading, mixing, and sedimentation
losses.
The unit sedimentation or loss rate is represented in general form as K Al CA2, where K is a calibration
adjustment factor, A 1 is an empirical parameter representing the effective decay or loss rate, and A2 takes
a value of either 1 or 2, defining either a first-order or second-order model. For a first-order model, the
resulting in-lake steady-state concentration, C, is C = G / (1 + K Al T), where G is the influent
concentration and T is the hydraulic residence time. For the second-order model, C = [-1 + (1+4KA1
Ci T) °5 ] / (2 K Al T). Walker (1987) specifically recommends the second-order type of model as "the
most generally applicable formulation for representing phosphorus and nitrogen sedimentation in
reservoirs." Walker also notes the following:
Effective second-order sedimentation coefficients are on the order ofO. 1 m3 /mg-year for total
phosphorus and 0.0032 m3 /mg-year for total nitrogen... With these coefficients, nutrient
sedimentation models explain 83 and 84 percent of the between-reservoir variance in average
TETRATECH
60"
-------
Downstream Use Protection April 2015
phosphorus and nitrogen concentrations, respectively. Residuals from these models are
systematically related to inflow nutrient partitioning (dissolved versus paniculate or inorganic
versus organic) and to surface overflow rate over the data set range of 4 to 1,000
m/year. Effective rate coefficients tend to be lower in systems with high ortho-P/total P (and
high inorganic N/total N) loading ratios or with low overflow rates (4 to
10 m/year). Refinements to the second-order formulations (Models 1 and 2) are designed to
account for these dependencies...
Nutrient sedimentation Models 1 and 2 use different schemes to account for effects of inflow
nutrient partitioning. In the case of phosphorus, Model 1 performs mass balance calculations
on "available P, " a weighted sum ofortho-P and non-ortho-P which places a heavier emphasis
on the ortho-P (more biologically available) component. Model 2 uses total phosphorus
concentrations but represents the effective sedimentation rate as inversely related to the
tributary ortho-P/total P ratio, so that predicted sedimentation rates are higher in systems
dominated by nonortho (paniculate or organic) P loadings and lower in systems dominated by
ortho-P or dissolved P loadings. The nitrogen models are structured similarly, although
nitrogen balances are much less sensitive to inflow nutrient partitioning than are phosphorus
balances, probably because inflow nitrogen tends to be less strongly associated with suspended
sediments.
Model 1 accounts for inflow nutrient partitioning by adjusting the inflow concentrations, and
Model 2 accounts for inflow nutrient partitioning by adjusting the effective sedimentation rate
coefficient. While Model 2 seems physically reasonable, Model 1 has advantages in reservoirs
with complex loading patterns because a fixed sedimentation coefficient can be used and effects
of inflow partitioning are incorporated prior to the mass balance calculations. Because
existing data sets do not permit general discrimination between these two approaches, each
method should be tested for applicability to a particular case. In most situations, predictions
will be relatively insensitive to the particular sedimentation model employed, especially if the
ortho-P/total P loading ratio is in a moderate range (roughly 0.25 to 0.60). Additional model
application experiences suggest that Method 2 may have an edge over Model 1 in systems with
relatively long hydraulic residence times (roughly, exceeding 1 year), although further testing is
needed. Because the coefficients are concentration- or load-dependent and because the models
do not predict nutrient partitioning in reservoir outflows, Sedimentation Model 2 cannot be
applied to simulations of reservoir networks (Scheme 6 in Figure 4.3).
Based upon error analysis calculations, the models discussed above provide estimates of
second-order sedimentation coefficients which are generally accurate to within a factor of 2 for
phosphorus and a factor of 3 for nitrogen. In many applications, especially reservoirs with low
hydraulic residence times, this level of accuracy is adequate because the nutrient balances are
dominated by other terms (especially, inflow and outflow). In applications to existing
reservoirs, sedimentation coefficients estimated from the above models can be adjusted within
certain ranges (roughly a factor of 2 for P, factor of 3 for N) to improve agreement between
observed and predicted nutrient concentrations. Such tuning of sedimentation coefficients
should be approached cautiously because differences between observed and predicted nutrient
levels may be attributed to factors other than errors in the estimated sedimentation rates,
particularly if external loadings and pool concentrations are not at steady state.
TETRATECH
61"
-------
Downstream Use Protection April 2015
3.3.3 Weighted Distribution to Individual Streams
The methods in the prior two sections can estimate a total input loading limit to a lake or other
downstream waterbody, but do not address the issue of partitioning that load between multiple influent
streams. DPVs can always be assessed through detailed watershed modeling, which also allows
evaluation of how pollutant losses or attenuation during transit from distant points in the watershed
affects the delivered load.
As part of the 2010 Florida Phase I Estuarine Rule (Federal Register, 75(16): 4173-4226, January 26,
2010), USEPA proposed an intermediate complexity modeling approach to address this issue using the
USGS SPARROW (Spatially Referenced Regressions on Watershed attributes) model. Specifically,
USEPA proposed using the SPARROW model for nitrogen in the southeastern United States (Hoos and
McMahon, 2009). Although this approach was later withdrawn, it provides an example of one
intermediate complexity approach to partitioning downstream loads into the watershed. USEPA
described the approach as follows:
EPA's approach to developing nutrient criteria for streams to protect downstream estuaries in
Florida involves two separate steps. The first step is determining the average annual nutrient
load that can be delivered to an estuary without impairing designated uses. This is the
protective load. The second step is determining nutrient concentrations throughout the network
of streams and rivers that discharge into an estuary that, if achieved, are expected to result in
nutrient loading to estuaries that do not exceed the protective load. These concentrations,
called "downstream protection values" or DPVs, depend on the protective load for the
receiving estuary and account for nutrient losses within streams from natural biological
processes. In this way, higher DPVs may be appropriate in stream reaches where a significant
fraction of either TN or TP is permanently removed within the reach before delivery to
downstream receiving waters.
While this refers to "determining nutrient concentrations," what SPARROW provides is estimates of
average annual load, so the text should properly refer to flow-weighted concentrations (total load divided
by total flow), not ambient concentrations.
USGS has developed SPARROW models for many regions of the U.S. (see summary in Preston et al.,
2011). These are empirical statistical models that predict the natural logarithm of annual average load as
estimated from USGS flow and water quality monitoring throughout a region. Each regional model
differs, but all contain three major explanatory components: a regression model of nutrient sources
expressed as mass per unit time or as area of specific land uses, exponential land-to-water delivery
functions, and first-order aquatic decay (attenuation) relationships. The coefficient of determination (R2)
on total yield rates (mass per unit area) in logarithmic space ranges from 0.72 to 0.86 for total nitrogen
and from 0.60 to 0.80 for total phosphorus, although significant bias can be present in estimates for
individual sites. It is important to note that the SPARROW estimates are dependent on the methods used
to estimate total load from intermittent monitoring data (see discussion in Saad et al., 2011).
SPARROW model variables for predicting nitrogen sources typically include measures related to point
source discharges, area in urban land, agricultural fertilizer, livestock, and atmospheric deposition.
Variables for predicting total phosphorus typically include measures related to agricultural fertilizer,
livestock, and background sources. Parameters describing delivery to streams are variable, but often
include measures of precipitation, temperature, and stream density. Attenuation is usually based on time
of travel, with rates that decrease as average annual flow increases.
TETRATECH
62
-------
Downstream Use Protection April 2015
In theory, SPARROW models could be used to partition downstream load to progressively smaller
upstream areas, although there are likely to be discrepancies in any specific local application and the
uncertainty bounds on SPARROW predictions are generally large. One potentially promising approach is
a Bayesian analysis in which the distribution of the SPARROW parameters and their associated model
predictions (as a prior) are combined with information from local monitoring to obtain an updated
posterior distribution. The reader is referred to Qian et al. (2005) and Wellen et al. (2014) for further
exploration of this approach.
3.4 Translating Influent Loads to Concentrations in
the Watershed Network
3.4.1 Concentration Data Analyses
Monitoring data from the Holcombe Flowage (WI) and Douglas Reservoir (TN) watersheds are used to
examine questions related to how concentrations of total phosphorus observed throughout real watersheds
relate to downstream concentrations and loads of phosphorus at the pour point into the terminal reservoir.
For each case we begin with some simple statistical analyses. First, we examine the implications of
simply setting the target for all upstream stations in the watershed equal to the downstream target (as a
seasonal average). This is examined both under actual observed conditions and under an assumption that
all observations in the watershed are scaled to exactly achieve the seasonal average concentration at the
pour point. This simple analysis helps examine a baseline for the expected frequency at which watershed
sample concentrations can be expected to exceed the downstream target.
A companion analysis then evaluates the implications of setting the target for stations upstream in the
watershed at an upper confidence limit on the downstream target, calculating using lognormal distribution
assumptions. This represents an assessment strategy under which observations are flagged if they appear
to be inconsistent with attaining the downstream target as a long term average. The lognormal statistics
are based on the arithmetic coefficient of variation (CV) for samples at each station, using the same
calculation methods described in USEPA (1991). The upper bound confidence limit (UCL) for a
lognormally distributed variable relative to a target concentration, T, is given by
UCL = T • exp(cr z — 0.5 cr2), where o is the standard deviation in log space, estimated from the
arithmetic CV as a = y ln(CV2) + 1 and z is the critical value of the standard normal distribution, equal
to 1.645 for a 95th percentile probability basis
3.4.2 Simplified Assignment of Load
As noted above, the true objective is generally best framed in terms of controlling annual or seasonal load
to the downstream receiving water. Concentrations observed in the stream network may be more or less
significant depending on their implications for total load delivered from the watershed. Nutrient mass
may be lost during transport, in which case concentrations observed at locations far from the downstream
terminus may be less significant. Obviously, concentration measurements that are associated with larger
flows (and thus larger loads) are more significant to determining the total load. Further, different areas
within a watershed will be expected to have different loading rates, and the downstream use can be
protected if the net effect of different load sources is consistent with the overall loading goal.
The analyses of Holcombe Flowage and Douglas Reservoir look at the load delivery piece separately by
use of SPARROW loss rates. We then examine two simple methods of partitioning the delivered load
TETRATECH
63
-------
Downstream Use Protection April 2015
back into the watershed. The first employs an inverse application of load estimation to estimate
concentration as a function of flow (see below). The second uses the SPARROW approach (described
above in Section 3.3.3) to predict both load generation rates (as a function of land use) and rate of
delivery to the downstream terminus. Both analyses can be performed on either the actual observed data
or on observed data scaled to the assumption that the downstream criterion is exactly met. Both analyses
are also simplified analyses that do not take spatial correlation into account and are not calibrated to local
observed data.
Direct load estimation was evaluated in two ways, using LOADEST or using a simpler stratified
regression. The USGS LOADEST program (Runkel et al, 2004) develops a rating curve regression to
estimate loads. The regression models can be of varying degrees of complexity, but the simplest
LOADEST model estimates load, L, at time / as
where Q is flow, H is a bias correction factor, and a0 and cii are regression coefficients fitted to observed
samples of load (as concentration times flow). Given a complete time series of Q, total load over a period
of interest can then be estimated as the sum of the estimates of Li.
In the regression equation, ai can be thought of as representing the portion of concentration that is related
to flow in the system. The constant cto (although it is multiplicative after back transform) would include
the effects of relatively constant loads from point sources, along with other non-flow-related inputs, such
as direct atmospheric deposition. Attaining a target load could be done by adjusting the coefficients until
the desired load is met at a specified level of assurance over a long flow series. Similarly, LOADEST
results can be adjusted to attain a target specified as a flow-weighted mean concentration by dividing the
predicted load by the mean flow.
Once LOADEST is adjusted to the target load it can be used to calculate permissible concentrations
conditional on flow, written as Ci\Qf.
u = -•
i-lns;gjl „
- — : — '-^ •
Assessment relative to protection of downstream uses could compare observations to d\Qt - either as a
strict limit or with some frequency specifications. The alternative stratified regression approach uses a
broken line log -log linear regression of concentration versus flow, with the breakpoint assigned at the
point that maximizes the resulting correlation coefficient.
This approach can be extended up into the watershed network by assuming that the regression coefficients
are constant (unless there is a need to correct for the location of point sources). If soils and geometry are
relatively consistent the flow could be subdivided in the network based on relative area only. Otherwise,
a simple hydrologic model could be used.
3.4.3 Accounting for Spatial Correlation
One task for the DPV analysis is determining how to evaluate observation series at miscellaneous
upstream points within a stream network. What do observations at one location tell us about other
locations in the watershed and about loads from the watershed as a whole?
In analyzing data of this sort it is important to take into account the strong spatial correlation that is likely
to be present among monitoring sites on a stream network. For example, if Site B is downstream of Site
TETRATECH
64~
-------
Downstream Use Protection April 2015
A, observations at Site A provide partial information on conditions at Site B. Conversely, adding Site B
to a monitoring network that already contained Site A would not add as much information as monitoring a
site on a separate (but same-sized) tributary due to the correlation of observations at A and B. Further, if
empirical models are fit to observations at a suite of sites on a watershed network failure to take into
account spatial correlation could produce biased parameter estimates and autocorrelated error structures.
These issues can be addressed through use of spatial statistical models that are designed to work on the
topology of a bifurcating stream network rather than on a Euclidean grid. Techniques for accomplishing
this were recently developed by Peterson and Ver Hoef (Peterson and Ver Hoef, 2010; Ver Hoef and
Peterson, 2010). A good example application of the method is in Isaak et al.'s (2010) predictive analysis
of stream temperatures.
The geospatial approach builds a covariance structure using up to three types of variance components.
One is a standard geospatial (kriging) Euclidean distance model, which uses direct distance between two
sites on a two-dimensional mapping to calculate spatial correlation. The other two potential variance
components reflect geospatial analysis on a network and are referred to as tail-up and tail-down
covariance.
The most general form of the spatial statistical model is:
where X is a vector of site characteristics, (3 is a vector of regression parameters, zu contains spatially-
autocorrelated random variables with a tail-up autocovariance, zd and ze are random variables with a tail-
down and Euclidean autocovariance respectively. W is a design matrix for random effects, while y and £
contain independent random variables.
In what follows, we include the regression parameters and spatial covariance matrix in the spatial
statistical model. The general form of spatial regression model is:
The first three terms on the right hand side of equation are components of ordinary linear regression: Xis
a vector of landscape characteristics used as regressors, (3 is a vector of regression coefficients, £ is a
random disturbance term, and u> is the bias correction vector based on the spatial covariance matrix.
The covariance matrix has the form:
COV(Y} = I = ff2fl(au) + o2dR(ad) + o2R(ae} + a2!
The different spatial covariance models are created using moving-average constructions proposed by Ver
Hoef and Peterson (2010). If a moving average function is non-zero only upstream of the start point, it is
called a "tail-up" model. On the contrary, it is called a "tail-down" model if the moving average function
is also non-zero downstream of the start point.
For a tail-up model, weighting factor between two different locations r and s are calculated using the
equation (Isaak et al., 2010):
Q(^l^u) if r and s are flow — connected,
if r and s are flow — unconnected,
Here h is the distance between locations r and s, the 0 V^fc are weights due to branching characteristics
of the flow-connected reaches, and Ct(h\9u~) is a kernel variogram function that can take various forms
TETRATECH
65"
-------
Downstream Use Protection April 2015
(linear-with-sill, spherical, exponential, Mariah, and so on). This covariance component is analogous to
the variogram used in kriging, but is based on hydrologic distance. The variogram is often expressed in
exponential form, in which case (Isaak et al., 2010):
Here h is the hydrologic distance, o2ru is the tail-up partial sill or variance component, and a is a spatial
range parameter. The tail-up autocovariance also requires calculation of spatial weights (wk) reflecting
relative discharge volume.
The tail-down covariance is also analogous to a variogram used in kriging based on hydrologic distance.
For example, the kernel function of the tail-down linear-with-sill model is:
ad
Cd(a,b,h) =
a^ ' ' J
°™(l }l(h\ad < 1) if flow-connected,
V Ctrl)
r i,\\
max(a, b}
°TD I 1 -- 1 /(max(a, b} \ad < 1) if flow — unconnected,
where / is an indicator function and o2xo is the tail-down partial sill or variance component. When two
sites are flow-unconnected, a and b denote the distances from each site to a common downstream
junction. If two sites are flow-connected, h denotes the separation distance between two sites via stream
network.
The data needed to develop the spatial covariance structure include hydrologic distances and spatial
weights. These can be calculated with the STARS ArcGIS toolset (Peterson and Ver Hoef, 2014) in
conjunction with the FLoWS toolset (Theobald et al., 2006). The output of STARS is a
SpatialStreamNetwork object. This object can then be imported into SSN (Ver Hoef et al., 2014), an R
package that accomplishes the generalized linear mixed model estimation and application. The entire set
of tools is actively supported by the Rocky Mountain Research Station of the U.S. Forest Service.
A spatial covariance structure created in this way can be used in both retrospective and prospective
modes. In retrospective mode it may be of interest to fit predictive models to stream networks that have
monitoring at multiple points. Such a predictive model could form the basis for evaluating
comprehensive sets of stream reach concentrations consistent with attaining a DPV. In prospective mode,
a description of the spatial autocovariance structure of observations on a stream network provides the
foundation for generating realistic comprehensive test data sets, incorporating uncertainty, with which to
examine the effects of different sampling and assessment strategies.
3.4.4 Generation of Complete Networks
Evaluation of the relationship between nutrient concentrations at points in the stream network and those
in a downstream receiving water body is complicated because the true distribution of concentrations at
every point in the stream network is not known. The spatial covariance structure analysis described in
Section 3 .4.3 provides a basis for generating a complete population of stream concentrations that takes
into account both the network-based and distance-based correlation that is likely to be present between
different segments. Once a realization of the complete population of concentrations in the network is
generated, experiments to assess the impacts of small sample sizes can be conducted.
TETRATECH
66
-------
Downstream Use Protection April 2015
Generation of a complete population of concentration data on the stream network is done through use of a
hybrid Monte Carlo data generation technique. The hybrid analysis combines statistical simulation with
continuous observation data.
Techniques for direct simulation of loads following an assumed distribution pattern are well established
(e.g., Press et al., 1992). The approach is relatively simple for a two-variable system, but is fully
generalizable to multiple correlated variables. The key idea is that the full set of correlated variables
needs to be generated simultaneously so that the correlation structure is maintained. The mathematics of
the process are explained in the appendix.
Monte Carlo simulations were used to explore the relationship of upstream exceedances to downstream
conditions under conditions of network spatial correlation. We assumed the outlet (pour point) of the
stream network is at the receiving waterbody of interest. To explore cases in which we wish to hold the
concentration at this point to a specified target value and allow the remainder of the network to vary
randomly in accordance with the spatial correlation structure, we set the row of the spatial covariance
matrix that corresponds to the pour point to a fixed value that is equal to the desired mean or regression
JT~ / n W2
result. The standard deviation of the remaining normal deviates in ~ was then inflated by (—— \ to
reflect the reduction of one degree of freedom caused by fixing the value at the pour point.
3.4.5 Monte Carlo Analysis of Sampling Results
In practice, the status of nutrient water quality in a watershed will almost always be assessed based on a
limited number of samples at a small number of sampling stations. What do such samples tell us about
whether a downstream target based on a seasonal or annual load or flow-weighted concentration is likely
to be achieved?
3.4.5.1
The methods described in Section 3.4.4 were used to generate complete populations of TP concentrations
within the stream network, using different assumptions about the relationship between the expected value
at the downstream pour point and the long-term average target flow-weighted concentration target.
Specifically, we tested values of the ratio ( 2) of the expected concentration to the target concentration
ranging from 0.1 to 1.9. A second layer of Monte Carlo analysis is then used to randomly select and
analyze a realistic sample of results from the complete network.
Water quality samples at sites on the highest order streams integrate the loading from a majority of the
drainage area. The mean TP concentration at those sites are much more informative as to the ultimate
downstream mean than would be results from a first-order tributary as they approximate the full
distribution at the pour point. Results at these points are of less interest for examination of how a DPV
should be propagated further up into the network. Thus a randomized sampling scheme on lower order
(flow accumulation area <= 500 km2) tributaries was combined with Hybrid Monte Carlo Data
Generation to investigate the relationship between monitored TP concentration in tributaries and the total
loading to the terminal waterbody.
For 10 different downstream endpoint TP concentration, 500 realization of the TP concentrations at sites
within the stream network were generated. For each realization, 5,000 different random sampling
combinations (with 1 to 20 sampling sites) were conducted. For each site, the mean TP concentration was
sampled with a lognormal distribution about "true mean" (concentration derived using spatial regression
TETRATECH
§7
-------
Downstream Use Protection
April 2015
function) and a coefficient of variation equals to one. The mean TP concentration across all the sampled
sites was calculated, and compared to the proposed critical TP concentration.
3.4.5.2 Representative Sample Size
It is commonly noted that the distribution of concentrations of water quality constituents approximates a
log-normal probability distribution (Van Buren et al. 1997; Novotny, 2004). For different sample sizes
and different downstream endpoint TP concentration (DC) scenarios, a log-normal distribution was used
to fit the sampled mean TP concentration. An Example of the dependence of the fitted distribution on
sample size and DC for the spatial network generated for Douglas Reservoir, TN is shown in Figure 3-1,
which displays the histograms of sampled TP concentration and the fitted probability density function
lines. As sample sizes increases, the log-normal distribution has a better fit.
Sample size=l,DC=0.045
Sample size=l,DC=0.099
Sample size=l,DC=0.135
DO 02
Sample size=10,DC=0.045
Sample size=10
DC=0.099
Sample size=10
DC=0.135
Sample size=20
DC=0.045
Sample size=20
DC=0.099
Sample size=20
DC=0.135
Figure 3-1. TP concentration distribution of upstream stream network of Douglas reservoir with
different sample sizes and downstream concentration (DC, mg/L) values
The larger the sample size (number of sampling sites), the less uncertainty is present in estimates of the
distribution of upstream TP concentrations. However, larger sample sizes also increases the cost of
monitoring and there are usually practical limitations on samples. Thus it is crucial to ascertain the
appropriate sample size for acquiring a representative upstream TP concentration. From a statistical point
of view, we measure the goodness of fit of log-normal distributions of different sample sizes (from 1
sample to 20 samples), then define the threshold of sample size above which the goodness of fit for the
log-normal distribution is stable. The Kolmogorov-Smirnov test (K-S test) was used to test the goodness
of fit. For a given cumulative distribution function (CDF) CDF(x), the Kolmogorov-Smirnov test
calculates the maximum distance between the predicted CDF curve and the observed one:
I TETRATECH
68
-------
Downstream Use Protection
April 2015
D = max(CDF(p) - CDF(o))
The smaller the value of D is, the better the fit between the observed value and the predicted values.
For different numbers of sampling sites, we calculated the mean concentration of TP from all the
sampling sites, fitted the mean TP concentration values with the log-normal distribution and applied the
Kolmogorov-Smirnov test. In Figure 3-2, diamonds with different colors are from different downstream
concentration scenarios, with decreasing concentration from top to bottom. As shown in Figure 3-2, the
K-S test statistics became stable if the sample size is larger than 10. This implies that little improvement
in the fitted distribution occurs if the sample size exceeded 10. So for the watershed of Douglas reservoir,
a set of 10 sample locations can be considered an appropriate size.
o
CM
in
o
O
¥1 o
T3 t-
(I)
QJ
CO
if)
o
o
o
o
I
* *
*
* *
10
DC=0.009
DC=0.027
DC=0.045
DC=0.063
DC=0.081
DC=0.099
DC=0.12
DC=0.14
DC=0.15
DC=0.17
15
I
20
Size of sampling sites
Figure 3-2. Kolmogorov-Smirnov test statistics for upstream TP concentration as a function of
sample size (Douglas Reservoir, TN example)
A similar analysis for Holcombe Flowage shows that a sample size greater than or equal to 5 results in
stabilization of the K-S test (Figure 3-3). The smaller sample size requirement compared to Douglas
Reservoir is due to less spatial variation in observed TP concentrations in the stream network.
TETRATECH
69
-------
Downstream Use Protection
April 2015
8
i5
to
OJ
V)
ID
O
o
O
u;
O
o
o
o
DC=0.005
DC=0.01
DC=0.03
DC=0.05
DC=0.07
DC=0.09
DC=0.11
DC=0.13
DC=0.15
« ! i
: ; ;
i
15
20
5 10
Size of sampling sites
Figure 3-3. Kolmogorov-Smirnov test statistics for upstream TP concentration as a function of
sample size (Holcombe Flowage, Wl example)
3.4.5.3 Analysis of Criterion Exceedance Ratio of Sampled Sites
Two types of sampling information from Monte Carlo simulations were used to estimate the downstream
concentration and its relationship to a target concentration. One is the number of sampling sites that have
a higher TP concentration than the criterion long term average concentration at the network pour point.
The other information is the mean TP concentration of the samples. Both types of information are
indicators of downstream TP concentration given the spatial correlation relationship between upstream
and downstream TP concentrations.
Specific results for the two studied watersheds are provided in Sections 3.5 and 0. Details of analytical
methods are discussed here. We applied statistical analysis based on conditional probability to derive the
downstream TP concentration information with sampling from upstream area. For both watersheds, we
did Monte Carlo simulations for 12 different downstream TP concentration scenarios. The range of the
12 TP concentration scenarios is set to cover most of the expected variation in TP concentrations at the
downstream site. The assumption was made that the downstream endpoint TP concentration also follows
a log-normal distribution. With some prior information of the endpoint TP concentration, the log-normal
distribution can be fitted to the 12 different concentrations and the probability of each concentration can
be calculated (Figure 3-4).
TETRATECH
70
-------
Downstream Use Protection
April 2015
Q
O
0.05
0.10 0.15
TP concentration (mg/L)
020
Figure 3-4. Cumulative Distribution Function (CDF) of downstream endpoint TP concentration
using fitted log-normal distribution
Two different log-normal distributions were used to fit the downstream endpoint concentration in this
project. One assumes a reference waterbody with high water quality and downstream endpoint
concentrations well less than the target criterion. The other one assumes a "critical" waterbody, where the
downstream nutrient concentration is very close to the target value. The use of these two different
distributional fits allows us to analyze the variance of upstream indicators given a reasonable range of
downstream concentrations.
For each TP concentration scenario, the same spatial regression (or spatial mean) model was applied to
the whole stream network. The spatial covariance of stream network TP concentration was multiplied by
the ratio of downstream TP concentration and the criterion TP concentration to take account the
concentration changes at the downstream endpoint. The mean exceedance ratio and the average number
of exceedance sites can then be calculated for different sample sizes given different downstream endpoint
TP concentrations. The exceedance ratio is defined as the number of sample sites with TP concentration
higher than the criterion TP concentration divided by the total number of sample sites. Figure 3-5 shows
the average number of sites with higher TP concentration than the criterion TP concentration with
different sampling size for each downstream TP concentration scenario at Douglas Reservoir. The
number of exceedance sampling sites were calculated based on the sampled exceedance ratio of the whole
upstream network.
TETRATECH
71
-------
Downstream Use Protection
April 2015
14 15 16
Samplingsize
I DC=0.009 • DC=0.027 • DC=0.045 • DC=0.063 • DC=0.081
I DC=0.099 • DC=0.117 • DC=0.135 • DC=0.153 • DC=0.171
Figure 3-5 the number of sites which TP concentration is higher than the criterion TP
concentration (Douglas reservoir)
Real world samples within the stream network provide limited information about the downstream
endpoint concentration. Thus, conditional probability was used to connect the upstream sampling
information to the downstream endpoint concentration.
Varying weights based on the downstream endpoint concentration distribution were assigned to
exceedance ratios from different downstream endpoint concentration scenarios to calculate the weighted
mean exceedance ratio:
r =
Here r is the exceedance ratio, w is the probability based weight factor, and / is downstream endpoint
concentration scenarios.
Given the relatively small sample size (1-20), site average results do not provide definitive information as
to whether a site actually exceeds a specific concentration target. Therefore, a confidence interval
approach was used. The Agresti-Coull interval is an approximate binomial confidence interval (Agresti
and Coull, 1998). Given thatXsites exceed the criterion concentration from n sampling sites, define
and
where p is the proportion of exceedances from the samples. A confidence interval forp is then given by
p±Z ~
TETRATECH
72
-------
Downstream Use Protection
April 2015
Where Zis the 1 — - a percentile of a standard normal distribution. For a 95% confidence interval,
a=0.05,Z=1.96.
Figure 3-6 shows an example of the convergence rate of 95% confidence intervals on the binomial
proportion for different sample sizes for the case when the number of exceedances is one. The range of
the confidence intervals converges to around 0.2 when the sample size is larger than 27.
Samplingsize
Figure 3-6. Generic 95% confidence interval range on the binomial mean as a function of sample
size when the number of observed exceedances equals 1
The large uncertainties of the weighted mean exceedance ratio cannot provide us much information about
what the downstream endpoint concentration would be given the sampled exceedance ratio. However, the
upper and lower bounds of the weighted mean exceedance ratio are good indicators of whether the
concentration at the endpoint exceeds the criterion concentration or not. If the sampled exceedance ratio
is higher than the upper bound of the weighted mean exceedance ratio, the downstream concentration has
a high probability of exceeding the criterion concentration, and vice versa. Given two different
downstream concentration fit assumptions, the upper bound of exceedance ratio at the critical situation
and the lower bound of the reference situation are used to be indicators of the situations of downstream
concentration. One-sided 95% confidence intervals were calculated for both downstream concentration
scenarios, and were combined to evaluate the range of the exceedance ratios. The lower confidence limit
calculated in this way is referred to as the "safe" threshold and the upper confidence limit as the "danger"
threshold in this analysis.
3.4.5.4 Analysis of Sampled Mean TP Concentration
The mean TP concentration of watershed samples can also be connected to the downstream concentration
using conditional probability. For 10 different downstream concentration scenarios, the log-normal
distribution was used to fit the mean TP concentration of samples. Figure 3-7 shows the distributions of
mean sample TP concentrations associated with different downstream TP concentrations. The flatter bell
curves are associated with higher downstream concentrations. As the downstream TP concentration
increases, the probability of encountering large sample mean TP concentrations in the stream network
increases.
TETRATECH
73
-------
Downstream Use Protection
April 2015
cq
o
J3 ^
O O
CL
O
O
DC=0.009
DC=0.027
DC=0.045
DC=0.063
DC=0.081
DC=0.099
DC=0.12
DC=0.14
DG=0.15
DC=0.17
0.00
0.05
0.10 0.15 0.20
TP concentration (mg/L)
0.25
0.30
Figure 3-7. Fitted log-normal distributions of mean sample TP concentration as a function of
downstream concentration
We then mapped different sample distributions to the distribution of downstream TP concentration. As
shown in Figure 3-8, we derived distributions of sampled TP concentration with different downstream
endpoint concentration (the orange curves) and the distribution of downstream endpoint concentration itself
(the light green curve). Thus for any given sampled TP concentration, the probabilities of this TP
concentration value could be calculated using the derived sample TP distributions (the orange polygons).
We assign different weights (light blue shaded polygons) to derived sample probabilities based on the
derived downstream endpoint TP concentration distribution. The ensemble probability of sampled TP
concentration was be calculated as
P =
where P is the probability of sampled TP concentration.
To connect the sampled TP concentration to the exceedance probability of the downstream TP
concentration, the following equation was used:
where Pe is the probability of the downstream concentration exceeds the criterion value given a mean
sample TP concentration, and / is the count of bins with downstream concentration larger than the criterion
TP value. The ratio between the sum of weighted probabilities at the right hand side of the vertical dashed
TETRATECH
74
-------
Downstream Use Protection
April 2015
line in Figure 3-8 for a given sample TP concentration and the weighted probability of that sample TP
concentration tells us the probability of exceedance of the target at the downstream location.
Criterion downstream concentration
I
0.04
0.08 0.10
TP concentration (mg/L)
0.12
0.14
Figure 3-8. Schematic plot of the connections between sampled TP concentration and
downstream endpoint concentration.
Notes: Each bar is a discretized probability of endpoint TP concentration for a given range (w2 and w7 are areas of
the shaded trapezoids, which are probabilities of endpoint TP concentration within ranges of [0.035mg/L, 0.055mg/L]
and [0.125mg/L to 0.145mg/L], respectively). The two small curves are probability density functions (PDFs) of
sampled TP concentration given certain endpoint TP concentration ranges. Two small shaded areas(P2 and P7)
are the probabilities of sampled TP concentration for a given range. The vertical dashed line identifies the criteria
endpoint TP concentration. This figure is only used for demonstration purpose..
The fitted sample TP concentration log-normal curve is flatter when the downstream TP concentration
increases (as shown in Figure 3-7), which causes the estimated area of low sample TP concentration with
high downstream TP concentration to be larger than it is with a low downstream TP concentration. To
avoid this 'tail effect' from the empirical probability density distribution, a lower boundary of the sample
TP concentration was set for the analysis. If the sampled TP concentration is lower than the threshold,
current probability method is not valid. If the mean sample TP concentration is low, the exceedance
probability is small. Thus more interest is given to higher mean sample TP concentration.
The exceedance probability calculation of different sample TP concentration was also applied to both the
reference and critical situations. The exceedance probability values for these two conditions can be treated
as lower and upper boundaries.
3.5 Application to Holcombe Flowage, Wl
Holcombe Flowage is an impoundment of the Chippewa River in Wisconsin (Figure 3-9). There are three
major tributaries to the Flowage, each of which has a USGS gage: the Chippewa River itself (Chippewa
River near Bruce, gage 05356500), the Flambeau River (Flambeau River near Bruce, gage 05360500),
and the Jump River (Jump River at Sheldon, gage 05362000). The Chippewa and Flambeau Rivers join
I TETRATECH
75
-------
Downstream Use Protection
April 2015
prior to reaching the Flowage, but downstream of the USGS gage, and there are also several minor
tributaries. Land use in the watershed is largely northern hardwood forest, with a relatively even
scattering of small towns, farms, and pasture (Figure 3-10). A number of small WWTP discharges are
spread throughout the watershed. Those nearer to Holcombe Flowage are shown in Figure 3-11.
Wisconsin \ Michigan
Holcombe
Flowage
SIream/River
Lake/Reservoir
Holcombe Flowage Watershed
County Boundary
State Boundary
Holcombe Flowage
Location Map
Figure 3-9. Watershed of Holcombe Flowage, Wl
TETRATECH
76
-------
Downstream Use Protection
April 2015
Wisconsin \ Michigan
Legend
— Stream/River
| Lake/Reservoir
| | Holcombe Flowage Watershed
_j County Boundary
II I State Boundary
Holcombe Flowage
Land Use/Land Cover (NLCD 2006)
[3 Developed - Low | | Hay/Pasture
^| Developed * Medium ^^B Cropland
| Developed - High | | wetlands
I N
A
TETRA TECH
Figure 3-10. Land Use in Holcombe Flowage (Wl) Watershed
TETRATECH
77
-------
Downstream Use Protection
April 2015
Wisconsin \ Michigan
Village of
Glen Flora
(0029963)
Village of
Weyerhaeuser
(0020761)
Village of
Hawkins
(0024201)
Catawba Kennan
Joint Sewage Commission
(0061701)
City of
Ladysmith
(0021326)
Village of Tony
(0026000)
Village of
Conrath
(0032522)
Village of
Sheldon
(00252453)
Holcombe
Flowage
Legend
A Municipal WWTPs
— Stream/River
| Lake/Reservoir
I Holcombe Flowage Watershed
\ County Boundary
State Boundary
Holcombe Flowage
Municipal Wastewater Treatment Plants
Figure 3-11. WWTP in the Downstream Portion of Holcombe Flowage (Wl) Watershed
There are various upstream impoundments in the watershed, each of which must meet its own protective
criteria. For the purposes of this analysis, only those sampling stations located between upstream
impoundments and Holcombe Flowage are evaluated. There are 26 stations with adequate data (at least 9
observations) within this area, and these proximate stations are shown in Figure 3-12.
TETRATECH
78
-------
Downstream Use Protection
April 2015
Wisconsin \ Michigan
553171
553173
10030672
Legend
USGS Flow Gage
l\ DNR Water Quality Gage
— Stream/River
| Lake/Reservoir
I Holcombe Flowage Watershed
j County Boundary
State Boundary
Holcombe
Flowage
Holcombe Flowage
Gage and Sampling Locations
Figure 3-12. Sampling Stations in Holcombe Flowage (Wl) Watershed
3.5.1 Concentration Data Analysis for Holcombe Flowage
Wisconsin Administrative Code (NR102.06(4) establishes a total phosphorus criterion for stratified
reservoirs of 30 ug/L, (0.03 mg/L), which is thus the downstream phosphorus criterion for Holcombe
Flowage. The Administrative Code is silent as to the averaging period across which this criterion should
be applied.
Under current conditions, the average TP concentration at every station evaluated in the watershed is
greater than the downstream criterion, and in many cases 100 percent of samples are greater than 0.03
mg/L (columns 3 and 4 in
TETRATECH
79
-------
Downstream Use Protection April 2015
Table 3-1). To examine what happens when each of the three tributaries on average met the criterion, all
observations were reduced by the ratio of 0.030 to the average concentration at the main tributary with the
highest average concentration (Jump River, station 553042, average 0.053 mg/L) or a factor of 0.566.
After this adjustment (column 5 in
TETRATECH
80"
-------
Downstream Use Protection April 2015
Table 3-1), the concentrations at many stations are still greater than 0.03 mg/L 100 percent of the time.
Of the downstream stations, Chippewa River (station 553003) exceeds 0.03 mg/L 15 percent of the time,
Jump River (553042) 40 percent of the time, and Flambeau River (553149) 8 percent of the time.
TETRATECH
81"
-------
Downstream Use Protection
April 2015
Table 3-1. Concentration Data Analysis for Total Phosphorus, Holcombe Flowage Watershed
Station
513188
513189
553003
553042
553063
553097
553126
553131
553137
553138
553149
553156
553158
553167
553169
553170
553171
553172
553173
613199
10029123
10029539
10030672
10030673
10031838
10031884
Station Name
Marsh Creek at Little Rapids Road
Wolf Creek at Wolf Creek Road
Chippewa River at USH8 Bridge
Jump River at CTH C, at Sheldon, Wl
Thornapple River - CTH A NW Ladysmith
Devils Creek - Low Site at Hwy 40 Brg
Main Creek at Broken Arrow Road
Deer Tail Creek at Broken Arrow Rd
McDermott Creek at CTH F, near
Weyerhauser, Wl
McDermott Creek at Horseshoe Lk Rd
Flambeau River at USGS Station, Hwy E
Flambeau River - Near Bruce, Wl
Little Soft Maple Creek at Kief Rd, near
Weyerhauser
Jump River - CTH G Bridge about 4 Miles
Downstream Of Sheldo
Deer Tail Creek at CTH B
Main Creek, North Fork at Cutoff Road
Bear Creek at STH 73
Alder Creek at STH 73
Unnamed Creek at STH 73
Levitt Creek at CTH D
MEADOW BROOK at STH 27
Big Weirgor Creek-downstream of Short
Cut Road
Jump River at Highway 73
South Fork Jump River along CTH I
Mud Creek at CTH D
Chippewa River at boat landing near
CTH/H and STH 40
Average
TP (mg/L)
0.081
0.061
0.040
0.053
0.053
0.066
0.081
0.114
0.160
0.169
0.042
0.044
0.064
0.053
0.065
0.143
0.055
0.107
0.089
0.077
0.116
0.044
0.072
0.080
0.151
0.046
% >0.03,
Current
100%
100%
71%
80%
83%
100%
100%
100%
100%
100%
85%
62%
96%
85%
100%
100%
100%
100%
100%
100%
100%
58%
100%
100%
100%
92%
% > 0.03,
Adjusted
82%
55%
15%
40%
58%
50%
58%
100%
100%
100%
8%
8%
46%
46%
64%
90%
50%
90%
100%
100%
100%
25%
75%
92%
100%
25%
Note: The "Adjusted" column shows results obtained with all observations adjusted by the ratio needed to meet the
target concentration at the downstream pour point. Station locations are shown in Figure 3-12
It
TETRATECH
82
-------
Downstream Use Protection April 2015
The median CV across all stations is 0.047. Under assumptions of a lognormal distribution, the
corresponding UCL on the downstream criterion is 0.056 mg/L. The average observed concentration at
some, but not all stations in the watershed exceed this level, and 53 percent of individual observations are
greater than 0.056 mg/L. Even after the adjustment to guarantee that all three of the downstream stations
achieve a sample average of 0.030 mg/L, rates of excursion of up to 100 percent of samples occur, and
only two stations have no excursions (Table 3-2). This demonstrates that frequent observations in the
watershed in excess of the downstream criterion do not necessarily indicate failure to meet that
downstream criterion.
TETRATECH
83"
-------
Downstream Use Protection
April 2015
Table 3-2. Comparison of Total Phosphorus Data to UCL on Downstream Criterion, Holcombe
Flowage Watershed
Station
513188
513189
553003
553042
553063
553097
553126
553131
553137
553138
553149
553156
553158
553167
553169
553170
553171
553172
553173
613199
10029123
10029539
10030672
10030673
10031838
10031884
Station Name
Marsh Creek at Little Rapids Road
Wolf Creek at Wolf Creek Road
Chippewa River at USH8 Bridge
Jump River at CTH C, at Sheldon, Wl
Thornapple River - CTH A NW Ladysmith
Devils Creek - Low Site at Hwy 40 Brg
Main Creek at Broken Arrow Road
Deer Tail Creek at Broken Arrow Rd
McDermott Creek at CTH F, near
Weyerhauser, Wl
McDermott Creek at Horseshoe Lk Rd
Flambeau River at USGS Station, Hwy E
Flambeau River - Near Bruce, Wl
Little Soft Maple Creek at Kief Rd, near
Weyerhauser
Jump River - CTH G Bridge about 4 Miles
Downstream Of Sheldo
Deer Tail Creek at CTH B
Main Creek, North Fork at Cutoff Road
Bear Creek at STH 73
Alder Creek at STH 73
Unnamed Creek at STH 73
Levitt Creek at CTH D
MEADOW BROOK at STH 27
Big Weirgor Creek-downstream of Short
Cut Road
Jump River at Highway 73
South Fork Jump River along CTH I
Mud Creek at CTH D
Chippewa River at boat landing near
CTH/H and STH 40
% >0.056,
Current
82%
55%
10%
40%
50%
50%
58%
100%
100%
100%
8%
8%
46%
46%
64%
90%
30%
90%
80%
100%
100%
25%
67%
75%
100%
92%
% > 0.056,
Adjusted
36%
9%
5%
10%
0%
21%
25%
25%
100%
85%
0%
8%
13%
8%
9%
50%
0%
30%
40%
9%
50%
8%
17%
17%
73%
25%
Note: The "Adjusted" column shows results obtained with all observations adjusted by the ratio needed to meet the
target concentration at the downstream pour point. Station locations are shown in Figure 3-12
It
TETRATECH
84
-------
Downstream Use Protection
April 2015
3.5.2 Simplified Assignment of Load, Holcombe Flowage
The significance of some excursions of the downstream criterion within the watershed network may be
mitigated by transit losses. The regional SPARROW model of Robinson and Saad (2011), developed for
the Upper Midwest, including the drainages to the Great Lakes, the Upper Mississippi River Basin, and
the Ohio River Basin, estimates exponential attenuation coefficients for TP (day"1) as a function of travel
time. For average flows less than 1.416 m3/s the calibrated attenuation coefficient is 0.198 day"1, while
for flows between 1.416 and 2.265 m3/s the attenuation coefficient is 0.298 day"1. No attenuation is
assigned when average flows are above 2.265 m3/s.
A GIS analysis was conducted using NHDPlus Version 2 coverages to generate river miles and time of
travel estimates from the flowage to the upstream 26 stations, respectively. The NHDPlus data included
reach lengths, estimated annual average flow, and estimated average velocity. Travel time was calculated
in days. Application of the resulting total estimated attenuation yielded only minor changes in results as
travel times from the stations examined to Holcombe Flowage are short and many of the stream segments
exceed the average annual flow cutoff of 2.265 m3/s (80 cfs) above which no attenuation occurs. The
targets at individual stations adjusted for attenuation (Table 3-3) range up to 0.053 mg/L, but are all less
than the UCL analysis target presented above.
Table 3-3. Revision to Upstream Targets based on SPARROW Attenuation
Station
513188
513189
553003
553042
553063
553097
553126
553131
553137
553138
553149
553156
553158
553167
553169
553170
553171
553172
553173
613199
10029123
10029539
10030672
10030673
Adjusted Target TP (mg/L)
0.030
0.030
0.030
0.030
0.030
0.031
0.030
0.032
0.034
0.036
0.030
0.030
0.036
0.030
0.053
0.040
0.036
0.035
0.042
0.032
0.030
0.031
0.030
0.030
TETRATECH
85
-------
Downstream Use Protection
April 2015
A load partitioning analysis was undertaken using the stratified regression approach. The stratified
regression approach provided a better fit to the observed data than simple LOADEST approach (although
it should be noted that LOADEST contains a whole array of more complex models that would likely
provide a better fit, but which are difficult to rearrange for the purpose of predicting concentration as a
function of flow). The stratified regression was applied to develop a relationship between the natural log
of load and the natural log of flow at the three downstream USGS stations (Chippewa River near Bruce,
05356500; Flambeau River near Bruce, 05360500; and Jump River at Sheldon, 05362000), which
together account for 87 percent of the upstream drainage area. The resulting R2values were 0.49, 0.44,
and 0.16 for the Chippewa, Flambeau, and Jump River stations, respectively.
We used the stratified regression relationships to develop a complete time series of predicted loads and
concentrations for water years 1951 - 2013. Application of the rating curve (Table 3-4) suggests that the
Chippewa River contributes a somewhat larger percentage of the TP load than is accounted for by its
drainage area.
Table 3-4. Stratified Regression Analysis of Loads at Holcombe Flowage Gages
USGS ID (WQID)
05356000 (553003)
05360500(553149)
05362000 (553042)
Station
Chippewa R nr Bruce
Flambeau R nr Bruce
Jump R at Sheldon
Drainage Area (mi2, %)
1,650(40.4%)
1,860(45.5%)
576(14.1%)
TP Load (%)
46.1%
43.8%
10.1%
For the Chippewa River, the long-term flow-weighted concentration predicted by the stratified regression
is 0.055, or nearly twice the downstream target of 0.030 mg/L. A compliance scenario was developed by
iteratively reducing the concentrations in the upper stratum only (above 1,552 cfs) until the flow-weighted
concentration met the desired target of 0.030. The analysis includes a random error term based on the
root mean squared error in the estimates of ln(P). Achieving the 0.030 target required a reduction of 67.7
percent in the upper stratum. After the target flow-weighted concentration was achieved, 46.1 percent of
the individual observations were still greater than 0.03 mg/L.
TETRATECH
86
-------
Downstream Use Protection
April 2015
h 60% £
- 40% £
Total P (mg/l)
Figure 3-13. Distribution of TP Concentrations after Achieving Flow-Weighted Concentration of
0.030 mg/L, Chippewa River
A similar analysis for the Flambeau River yields somewhat different results. For the Flambeau, a
reduction in the upper stratum (greater than 804 cfs) of 40.5 percent was needed to achieve the flow-
weighted concentration target of 0.030 mg/L, resulting in 38 percent of individual observations remaining
greater than the target (Figure 3-14).
100%
90%
80%
70%
Total P (mg/l)
Figure 3-14. Distribution of TP Concentrations after Achieving Flow-Weighted Concentration of
0.030 mg/L, Flambeau River
A load-based analysis can also be done using the USGS SPARROW model. Holcombe Flowage and its
watershed fall within the SPARROW total phosphorus model for the Great Lakes, Ohio, Upper
Mississippi, and Souris-Red-Rainy region (Robertson and Saad, 2011). Reaches in the SPARROW
I TETRATECH
87
-------
Downstream Use Protection
April 2015
model are based on Enhanced River Reach File 2.0. which is defined at a much coarser scale than
NHDPlus. The SPARROW model results are available on a companion decision support website and
include incremental and cumulative SPARROW inputs and outputs for each model reach. Outputs
include total delivered model load and estimated annual average flow, from which an annual average
flow-weighted concentration was calculated.
The SPARROW model predicts a flow-weighted concentration of total phosphorus within Holcombe
Flowage of 0.0484 mg/L, suggesting a need for a 38 percent reduction in loads (much less than the load
reduction suggested by the stratified regression analysis). In addition, the flow-weighted concentrations
from SPARROW differ in relative ordering and magnitude from those obtained with the site-specific
regression analysis (Table 3-5). This likely primarily reflects the uncertainty in SPARROW load
estimates at the local watershed scale.
Table 3-5. Comparison of SPARROW Flow-weighted Concentrations to Results of Stratified
Regression Analysis, Holcombe Flowage
Gage Location
0535600 (Chippewa)
05360500 (Flambeau)
05362000 (Jump)
SPARROW flow-weighted TP
concentration (mg/L)
0.048
0.040
0.068
Stratified regression flow-
weighted TP concentration (mg/L)
0.055
0.048
0.037
3.5.3 Network Spatial Correlation Analysis
3.5.3.1 Model Estimation
Twenty-seven observation sites within Holcombe Flowage watershed stream network were used to derive
the spatial regression model. For each site, the accumulated upstream area of different land uses and the
area of local land uses (both in km2) were calculated and considered as regression factors. A variety of
different spatial covariance models were tested with the SSN package (Table 3-6).
Table 3-6. Spatial Model Components Tested for the Holcombe Flowage Watershed Model
Regression factor
Total drainage area
Accumulation area of water body
Accumulation area of agriculture
Accumulation area of wetland
Regression
factor
acronyms
acc_area
acc_wb
acc_agri
acc_wet
Spatial covariance
models
Tail. up
Tail. down
Euclidean
Kernel function
type
Linear with sill
Spherical
exponential
Cauchy
TETRATECH
88
-------
Downstream Use Protection
April 2015
Regression factor
Accumulation area of agriculture
Upstream distance
Regression
factor
acronyms
acc_graze
upDist
Spatial covariance
models
Kernel function
type
Mariah
To evaluate the potential role of different regressors the potential models were first tested without spatial
covariance. Fitted results for the complete model are shown in Table 3-7.
Table 3-7. Evaluation of Regression Model Evaluation with All the Regression Factors, Holcombe
Flowage
Coefficients
Estimate Parameter
Std. Error
t-value
Pr(>|t|)
Significance
code
Model: TP ~ acc_area + acc_wb + acc_agri +acc_urban+ acc_wet + upDist
(Intercept)
acc_wet
acc_graze
acc_wb
acc_agri
acc_area
upDist
1.09E-01
1.34E-05
-3.27E-05
2.28E-04
4.99E-05
-1.25E-05
-3.10E-02
2.44E-02
3.50E-05
6.90E-05
2.00E-04
3.56E-05
9.46E-06
4.05E-02
4.488
0.383
-0.474
1.14
1.401
-1.32
-0.766
0.00023
0.70607
0.64061
0.26759
0.17666
0.20189
0.4528
***
Note: Significance codes: <0.001 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 " 1
The full regression model has a generalized R2 value of 0.32; however, none of the regressors was found
to have a significant relationship with TP concentration - possibly because the distribution of land uses is
relatively homogenous in this watershed.
We tested seven other combinations of regression factors (Table 3-8). No estimated parameters were
significantly different from zero with the exception of the intercept term, suggesting that a model based
on the mean and spatial covariance structure would be most appropriate.
Table 3-8. Additional Regression Models Tested, Holcombe Flowage
Function
Number
Formula
TP ~ acc_area + acc_wb + acc_agri +acc_graze+ acc_wet + upDist
TP ~ acc_area + acc_wb + acc_agri +acc_graze + upDist
TETRATECH
89
-------
Downstream Use Protection
April 2015
Function
Number
3
4
5
6
7
Formula
TP ~ acc_area + acc_wb + acc_agri + upDist
TP ~ acc_area + acc_wb + upDist
TP ~ acc_area + upDist
TP ~ acc_area
TP~1
We then investigated different spatial covariance model combinations (Table 3-9). The exponential
kernel function was used to compare different spatial covariance components.
Table 3-9. Spatial Covariance Models Evaluated, Holcombe Flowage
Number
1
2
3
4
5
6
7
Formula
TP~1
TP~1
TP~1
TP~1
TP~1
TP~1
TP~1
Variance Components
Exponential. tailup + Exponential. taildown + Exponential. Euclid + Nugget
Exponential. tailup + Exponential. taildown+ Nugget
Exponential. tailup + Exponential. Euclid+ Nugget
Exponential. taildown + Exponential. Euclid+ Nugget
Exponential. tailup+ Nugget
Exponential. taildown+ Nugget
Exponential. Euclid+ Nugget
By comparing the negative log-likelihood (neg2LogL) and the standardized mean-squared prediction
error (std.MSPE), the spatial variance model with tailup and taildown components but no Euclidean
distance component (No.2) was determined to provide the best fit (Table 3-21). The negative log-
likelihood is the primary indicator of our selection. We first selected the models with the lowest negative
log-likelihood value, then rank the candidates by smallest standardized mean-squared prediction error.
For the Holcombe Flowage, model No.2 and model No.4 both have the lowest negative log-likelihood
value and no significant difference for the standardized mean-squared prediction error(~ 0.0001). Model
No.2 was selected to take account for spatial impacts from both flow connected and flow unconnected
sites.
TETRATECH
90
-------
Downstream Use Protection
April 2015
Table 3-10 Performance of Spatial Covariance Models, Holcombe Flowage
Number
1
2
3
4
5
6
7
Variance Components
Exponential. tailup + Exponential.taildown +
Exponential. Euclid + Nugget
Exponential. tailup + Exponential.taildown+ Nugget
Exponential. tailup + Exponential. Euclid+ Nugget
Exponential.taildown + Exponential. Euclid+ Nugget
Exponential. tailup+ Nugget
Exponential. taildown+ Nugget
Exponential. Euclid+ Nugget
neg2Logl_
-105.5865
-107.4239
-107.3983
-101.7381
-107.4239
-101.7385
-100.6893
std.MSPE
0.838332
0.880429
0.873515
0.807122
0.880318
0.807028
0.922747
Different kernel functions for the selected two spatial covariance were tested. Using the same
performance indicators (neg2LogL, std.MSPE), the best-fit spatial covariance components are
Exponential Tailup and Taildown. The finalized spatial regression model has the following form:
TP = 0.0824 + Scov(Exponential. tailup + Exponential, taildown)
where TP is the concentration of total phosphorus (mg/L) SCOv is the spatial covariance corrector vector.
This model has a median residual of-0.002 mg/L and the standard error of the residuals is 0.03.
3.5.3.2 Population Level Results for Holcombe Flowage Watershed
The selected spatial covariance model was used to estimate the average TP concentration of stream
segments through the reach network (Figure 3-26). The histogram of TP concentrations is shown in
Figure 3-16. The mean of the TP concentration for all segments is 0.08 mg/L and the standard deviation
is 0.012 mg/L. The minimum predicted TP concentration within the stream network is 0.034 mg/L and
the maximum predicted TP concentration is 0.16 mg/L. The 75th percentile predicted TP concentration is
less than 0.082 mg/L, and more than 50% of predicted TP concentrations were within a range from 0.075
mg/L to 0.082 mg/L.
TETRATECH
91
-------
Downstream Use Protection
April 2015
l/l
QJ
CUD
QJ
T3
"ro
£
'o
QJ
~oT
T3
to —
uo
in —
co
UD
XT
-91.4 -91.2 -91.0 -90.8 -90.6
Longitude(decimal degrees)
-90.4
Standard Errors
0 0.0357
0 0.0356
° 0.0354
o 0.0349
0 0.032
O 0.00171
Predictions
0.03 to 0.06
0.06 to 0.07
0.07 to 0.07
0.07 to 0.08
0.08 to 0.08
0.08 to 0.08
0.08 to 0.08
0.08 to 0.08
0.08 to 0.08
0.08 to 0.15
Figure 3-15. TP Concentrations Estimated by Spatial Regression Model, Holcombe Flowage
TETRATECH
92
-------
Downstream Use Protection
April 2015
IT
O
O
LO
O
O
O
O
O
O -
0.04 0.06 0.08 0.10
TP (mg/L)
0.12
0.14
0.16
Figure 3-16 Histogram of Predicted TP Concentrations, Holcombe Flowage
3.5.4 Sampling Results on Spatially Correlated Network
3.5.4.1 Prior Information on TP Targets
As noted above, the Wisconsin Administrative Code specifies a total P criterion of 0.030 mg/L for
stratified reservoirs, but does not state the averaging period. For the purposes of this analysis, we depart
from the published criterion and instead examine the statistical distribution of reference conditions. The
Level III ecoregional study of reference conditions conducted by USEPA (2000) for lakes in this area has
a mean of 0.032 mg/L and a 95th percentile of 0.095 mg/L. In the remainder of this section, we used the
95th percentile lake TP value (0.095 mg/L) as a daily TP target.
Based on this prior information, two downstream endpoint TP concentration scenarios were created for
the upper and lower boundaries. The first downstream endpoint TP concentration scenario is a reference
scenario, based on the 25th percentile value of 0.009 mg/L for the ecoregional reference lake population.
The critical scenario has a mean concentration of 0.095 mg/L, and the observed concentrations have a CV
of 1.88. Log-normal probability distribution functions (PDFs) were fit to both distributions (Figure 3-17).
TETRATECH
93
-------
Downstream Use Protection
April 2015
IT)
IN
O
CN
c
-------
Downstream Use Protection
April 2015
Exceedance Ratio with 95% Confidence Interval
0.80
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0.00
8
10 11 12 13 14 15 16 17 18 19 20
Seriesl
Senes3
0.33 0.33 0.33 0.33 0,33 0.33 0.33 0.33 0.33 0.33 0.33 0.33 0.33 0.33 0.33 0.33
0.68 0.65 0.63 0.61 0.60 0.58 0,57 0.56 0.55 0.55 0.54 0,53 0.53 0.52 0.52 0.51
Number of sampling stations
Figure 3-18. Exceedance Ratio and One-sided 95% Confidence Interval for Reference Scenario,
Holcombe Flowage
TETRATECH
95
-------
Downstream Use Protection April 2015
Exceedance Ratio with 95% Confidence Interval
0.30
0.25
0.20
0.15
0.10
0.05
0.00
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Seriesl 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16
— — Series2 0.01 0.02 0.02 0.03 0.03 0.04 0.04 0.04 0.05 0.05 0.05 0.05 0.06 0.06 0.06 0.06
Number of sampling stations
Figure 3-19. Exceedance Ratio and One-sided 95% Confidence Interval for Critical Scenario,
Holcombe Flowage
We define two thresholds for the number of sample means exceeding the target. The lower bound
exceedance ratio was used to calculate the "safe" threshold. This threshold implies that if the number of
exceedance sites is smaller than the given threshold, there is only a small chance (less than 5%) that the
downstream endpoint TP concentration is exceeding the criterion TP concentration. In this case, we
could conclude with a high level of confidence that the TP concentration at the downstream pour point is
below the target. Similarly, the upper bound exceedance ratio was used to calculate the "danger"
threshold. If the number of sites with concentrations above this boundary is larger than the "danger"
threshold, there is a high probability that the TP concentration at the downstream pour point is higher than
the criterion TP concentration. Table 0-3 lists the thresholds for different sample sizes.
Table 3-11. Exceedance Thresholds (Count) for Different Sample Sizes, Holcombe Flowage
TETRATECH
96
-------
Downstream Use Protection
April 2015
Sample Size
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Safe Threshold
0
0
0
0
0
0
0
1
1
1
1
1
1
1
1
1
Danger Threshold
4
4
5
5
6
6
7
7
8
8
9
9
9
10
10
11
3.5.4.3 Mean TP Concentration at Sampling Sites
For different downstream endpoint TP concentrations, the mean TP concentration and the first and third
quantiles for sampling sites were calculated for different sampling sizes and downstream concentration
(DC) targets (Table 3-12). The mean sample TP concentration differs little across DC when sample size
is larger than five. The sample TP concentrations were fitted with log-normal probability distribution
functions for different downstream endpoint TP concentrations to derive the 25th and 75th percentile
values.
TETRATECH
97
-------
Downstream Use Protection
April 2015
Table 3-12. TP Concentration Results as a Function of Sample Size and Downstream Concentration Target, Holcombe Flowage
Watershed
Sample
Size
5
6
7
8
9
10
11
12
13
14
15
16
DC=0.005
0.082,
[0.054,0.081]
0.082,
[0.053,0.082]
0.082,
[0.053,0.082]
0.082,
[0.057,0.082]
0.082,
[0.061 ,0.082]
0.082,
[0.053,0.082]
0.082,
[0.061 ,0.082]
0.082,
[0.060,0.082]
0.082,
[0.064,0.082]
0.082,
[0.065,0.082]
0.082,
[0.065,0.082]
0.082,
[0.065,0.082]
DC=0.03
0.082,
[0.052,0.081]
0.082,
[0.049,0.081]
0.082,
[0.056,0.082]
0.082,
[0.059,0.082]
0.082,
[0.060,0.082]
0.082,
[0.060,0.082]
0.082,
[0.060,0.082]
0.082,
[0.061 ,0.082]
0.082,
[0.064,0.082]
0.082,
[0.064,0.082]
0.082,
[0.064,0.082]
0.082,
[0.064,0.082]
DC=0.05
0.082,
[0.049,0.081]
0.082,
[0.049,0.081]
0.082,
[0.051 ,0.082]
0.082,
[0.056,0.082]
0.082,
[0.057,0.082]
0.082,
[0.058,0.082]
0.082,
[0.055,0.082]
0.082,
[0.059,0.082]
0.082,
[0.058,0.082]
0.082,
[0.058,0.082]
0.082,
[0.061 ,0.082]
0.082,
[0.061 ,0.082]
DC=0.07
0.082,
[0.037,0.080]
0.082,
[0.043,0.081]
0.082,
[0.044,0.082]
0.082,
[0.045,0.082]
0.082,
[0.050,0.082]
0.082,
[0.049,0.082]
0.082,
[0.051 ,0.082]
0.082,
[0.048,0.082]
0.082,
[0.053,0.082]
0.082,
[0.054,0.082]
0.082,
[0.056,0.082]
0.082,
[0.053,0.082]
DC=0.09
0.082,
[0.025,0.080]
0.082,
[0.029,0.081]
0.082,
[0.037,0.081]
0.082,
[0.032,0.082]
0.082,
[0.036,0.082]
0.082,
[0.037,0.082]
0.082,
[0.040,0.082]
0.082,
[0.036,0.082]
0.082,
[0.037,0.082]
0.082,
[0.044,0.082]
0.082,
[0.049,0.082]
0.082,
[0.048,0.082]
DC=0.11
0.083,
[0.018,0.080]
0.082,
[0.019,0.081]
0.083,
[0.022,0.081]
0.083,
[0.026,0.082]
0.082,
[0.031 ,0.082]
0.082,
[0.032,0.082]
0.082,
[0.035,0.082]
0.082,
[0.033,0.082]
0.082,
[0.033,0.082]
0.083,
[0.030,0.082]
0.082,
[0.031 ,0.082]
0.083,
[0.037,0.082]
DC=0.13
0.083,
[0.010,0.080]
0.083,
[0.016,0.081]
0.083,
[0.016,0.082]
0.083,
[0.023,0.082]
0.083,
[0.020,0.082]
0.083,
[0.025,0.083]
0.083,
[0.027,0.083]
0.083,
[0.028,0.083]
0.083,
[0.031 ,0.083]
0.083,
[0.034,0.083]
0.083,
[0.031 ,0.083]
0.083,
[0.031 ,0.083]
DC=0.15
0.085,
[0.003,0.080]
0.084,
[0.006,0.082]
0.084,
[0.011,0.082]
0.084,
[0.014,0.083]
0.085,
[0.017,0.083]
0.084,
[0.016,0.084]
0.084,
[0.022,0.084]
0.084,
[0.022,0.084]
0.084,
[0.028,0.084]
0.084,
[0.026,0.084]
0.084,
[0.026,0.084]
0.084,
[0.029,0.084]
DC=0.17
0.086,
[0.001,0.080]
0.086,
[0.010,0.082]
0.085,
[0.010,0.083]
0.085,
[0.014,0.084]
0.085,
[0.015,0.084]
0.085,
[0.016,0.084]
0.086,
[0.015,0.085]
0.086,
[0.023,0.085]
0.085,
[0.015,0.085]
0.085,
[0.020,0.085]
0.085,
[0.026,0.085]
0.085,
[0.026,0.085]
DC=0.3
0.103,
[0.000,0.079]
0.103,
[0.000,0.094]
0.103,
[0.000,0.096]
0.103,
[0.002,0.098]
0.103,
[0.000,0.099]
0.103,
[0.003,0.100]
0.103,
[0.006,0.100]
0.103,
[0.004,0.100]
0.103,
[0.007,0.100]
0.103,
[0.012,0.100]
0.103,
[0.012,0.100]
0.103,
[0.013,0.100]
TETRATECH
98
-------
Downstream Use Protection
April 2015
Sample
Size
17
18
19
20
DC=0.005
0.082,
[0.067,0.082]
0.082,
[0.066,0.082]
0.082,
[0.069,0.082]
0.082,
[0.069,0.082]
DC=0.03
0.082,
[0.066,0.082]
0.082,
[0.066,0.082]
0.082,
[0.066,0.082]
0.082,
[0.066,0.082]
DC=0.05
0.082,
[0.063,0.082]
0.082,
[0.063,0.082]
0.082,
[0.064,0.082]
0.082,
[0.064,0.082]
DC=0.07
0.082,
[0.053,0.082]
0.082,
[0.058,0.082]
0.082,
[0.057,0.082]
0.082,
[0.057,0.082]
DC=0.09
0.082,
[0.049,0.082]
0.082,
[0.049,0.082]
0.082,
[0.051,0.082]
0.082,
[0.051,0.082]
DC=0.11
0.083,
[0.042,0.082]
0.083,
[0.041 ,0.082]
0.083,
[0.040,0.082]
0.083,
[0.040,0.082]
DC=0.13
0.083,
[0.036,0.083]
0.083,
[0.038,0.083]
0.083,
[0.036,0.083]
0.083,
[0.036,0.083]
DC=0.15
0.084,
[0.032,0.084]
0.084,
[0.030,0.084]
0.084,
[0.031,0.084]
0.084,
[0.031,0.084]
DC=0.17
0.085,
[0.022,0.085]
0.086,
[0.028,0.085]
0.086,
[0.031,0.085]
0.086,
[0.031,0.085]
DC=0.3
0.103,
[0.014,0.100]
0.103,
[0.018,0.100]
0.103,
[0.019,0.100]
0.103,
[0.019,0.100]
Note: Results show the mean with the interquartile range in parentheses. DC = downstream concentration target.
TETRATECH
99
-------
Downstream Use Protection
April 2015
For Holcombe Flowage, we chose the Monte Carlo simulation results with sample size equal to five for
the mean TP concentration analysis. The exceedance probability at the downstream pour point for
different sampled mean TP concentrations was calculated using the methods described in Section 3.4.5.4.
The reference and critical scenarios were also applied to the mean TP concentration analysis. The
exceedance probability range is derived from these two scenarios (Figure 3-20, Table 3-13). For different
sampled mean TP concentrations, the exceedance probability calculated with the reference situation is
considered the lower bound and upper bound for the value calculated with the critical situation.
-------
Downstream Use Protection
April 2015
TP(mg/L)
0.115
0.12
0.125
0.13
0.135
0.14
0.145
0.15
0.155
0.16
0.165
0.17
0.175
0.18
0.185
0.19
Exceedance Probability (%)
61.3-78.1
74.8-84.0
83.5-88.6
89.2-92.1
93.0-94.6
95.5-96.4
97.1-97.6
98.2-98.5
98.9-99.0
99.3-99.4
99.6-99.6
99.7-99.8
99.8-99.9
99.9-99.9
99.9-99.9
100.0-100.0
Note: Results shown are based on a sample size of 5.
3.6 Application to Douglas Reservoir, TN
Douglas Reservoir, operated by the Tennessee Valley Authority, lies primarily in Jefferson County with
small overlap into Cocke (upstream) and Sevier (downstream) counties (Figure 3-21. The watershed
encompasses three 8-digit HUCs (06010105: Upper French Broad, 06010106: Pigeon, and 06010108:
Nolichucky), with an upstream drainage area is 4,289 mi2 in Tennessee and North Carolina. Land use in
the watershed is shown in (Figure 3-102).
TETRATECH
101
-------
Downstream Use Protection
April 2015
^g Laurel Cree*
Douglas
Reservoir
North
Carolina
Legend
Stream/River
Lake/Reservoir
Douglas Reservoir Watershed
County Boundary
South
Carolina
Douglas Reservoir
Location Map
Figure 3-21. Watershed of Douglas Reservoir, TN
TETRATECH
102
-------
Downstream Use Protection
April 2015
Douglas
Reservoir
Stream/River
Lake/Reservoir
Douglas Reservoir
County Bourvdary
State Boundary
North
Carolina
J Developed - Open
Developed • Low
• Developed - Medium
H Developed - High
Barren
H Forest
J HerBaceousj'Shrub, Shrub
Hay/Pasture
• Cropland
_] V* Hands
South
Carolina
Douglas Reservoir
Land Use/Land Cover (NLCD 2006)
Figure 3-22. Land Use in Douglas Reservoir (TN) Watershed
TETRATECH
103
-------
Downstream Use Protection
April 2015
Water quality monitoring in the watershed is relatively sparse compared to Holcombe Flowage. The 19
stations that had at least 10 observations for total phosphorus were selected for analysis (Figure 3-23).
03465500
03465500
03466208
03466208
03455000
03455000
Douglas
Reservoir
E4770000 \ §
co
03460000
E6450000
E6480000
FRB047C
FRB047B
FRB047A
E5600000
E5495000
FRBLK2
FRBLK1
E3520000
E2730000
E2120000
North
Carolina
Legend
• USGS Flow Gage
A Water Quality Gage
— Stream/River
I Lake/Reservoir
Douglas Reservoir Watershed
County Boundary
State Boundary
03448800
E0150000
South
Carolina
Douglas Reservoir
Gage and Sampling Locations
Figure 3-23. Sampling Stations in Douglas Reservoir (TN) Watershed
TETRATECH
104
-------
Downstream Use Protection
April 2015
3.6.1 Concentration Data Analysis for Douglas Reservoir
The state of Tennessee has developed numeric interpretations of their narrative nutrient criterion (Denton
et al., 2001). Ecoregion 67g, Southern Shale Valleys, is located at the upstream portion of the reservoir.
The recommended interpretation from Denton et al. for streams in ecoregion 67g for TP is 0.090 mg/L.
As Tennessee has not proposed a lake criterion we adopt this stream-based target as the downstream
criterion (DC) for illustrative purposes in this analysis.
Average TP in the upstream watershed is quite variable, ranging from 0.023 to 0.208 mg/L (Table 3-14)
and all stream stations have at least some observations greater than the downstream criterion of 0.090
mg/L. To examine what happens when each of the three major tributaries on average met the criterion,
all observations were reduced by the ratio of 0.090 to the average concentration in the main tributary with
the highest average (French Broad River, station 03455000) or a factor 0.434. After this adjustment,
excursions of the criterion still appear, but the frequency is greatly reduced.
Table 3-14. Percent of All TP Observations by Station Greater than 0.090 mg-P/L
HUC8
06010105
Upper French
Broad
06010106
Pigeon
WQ Sta.
ID
03448800
03455000
E01 50000
E1490000
E2120000
E2730000
E3520000
E4280000
E4770000
FRBLK1
FRBLK2
E5495000
E5600000
E6450000
Name
Swannanoa R at I-40 at Black
Mtn, NC
French Broad R nr Newport, TN
French Broad R at NC-178 at
Rosman
Mills R at end of SR-1 337 nr
Mills R
Mud Cr at SR-1 508 nr Balfour
French Broad R at SR-3495
Glenn Bridge Rd nr Skyland
Hominy Cr at SR-3413 nr
Asheville
French Broad R at SR-1 348 at
Asheville X Ref E3420000
French Broad R at SR-1 634 at
Alexander
Lake Kenilworth
Lake Kenilworth
Pigeon R at NC-215 nr Canton
Pigeon R at SR-1642 at Clyde
Cataloochee Cr at SR-1 395 nr
Cataloochee
Average TP
(mg/L)
0.100
0.208
0.042
0.037
0.046
0.073
0.067
0.079
0.151
0.048
0.032
0.074
0.195
0.036
% > 0.09,
Current
36%
50%
3%
4%
10%
14%
17%
25%
74%
13%
0%
15%
72%
6%
% > 0.09,
Adjusted
18%
11%
0%
4%
0%
5%
8%
6%
17%
0%
0%
15%
31%
3%
TETRATECH
105
-------
Downstream Use Protection
April 2015
HUC8
06010108
Nolichucky
WQ Sta.
ID
E6480000
FRB047A
FRB047B
FRB047C
03465500
03466208
03467609
Name
Pigeon R at SR-1338 nr Hepco
Lake Junaluska
Lake Junaluska
Lake Junaluska
Nolichucky R at Embreeville, TN
Big Limestone Cr nr Limestone,
TN
Nolichucky R nr Lowland
Average TP
(mg/L)
0.168
0.029
0.023
0.025
0.046
0.151
0.151
% > 0.09,
Current
49%
0%
0%
0%
7%
58%
58%
% > 0.09,
Adjusted
10%
0%
0%
0%
7%
13%
13%
The median CV across all stations is 0.91, much higher than for Holcombe Flowage. Assuming a
lognormal distribution, the corresponding UCL on the downstream criterion is 0.239 mg/L. Only 6.8
percent of individual observations are greater than 0.239 mg/L. The percentage of observations greater
than the UCL in individual samples range from 0 to 24 percent. Even after the adjustment to guarantee
that all three of the downstream stations achieve a sample average of 0.090 mg/L, some excursions are
still present (Table 3-15).
Table 3-15. Comparison of Total Phosphorus Data to UCL on Downstream Criterion, Douglas
Reservoir Watershed
HUC8
06010105
Upper French
Broad
WQ Sta.
ID
03448800
03455000
E01 50000
E1490000
E21 20000
E2730000
E3520000
E4280000
E4770000
FRBLK1
FRBLK2
Name
Swannanoa R at I-40 at Black Mtn, NC
French Broad R nr Newport, TN
French Broad R at NC-178 at Rosman
Mills R at end of SR-1337 nr Mills R
Mud Cr at SR-1508 nr Balfour
French Broad R at SR-3495 Glenn Bridge
Rd nrSkyland
Hominy Cr at SR-341 3 nr Asheville
French Broad R at SR-1 348 at Asheville X
Ref E3420000
French Broad R at SR-1 634 at Alexander
Lake Kenilworth
Lake Kenilworth
% > 0.239,
Current
18%
11%
0%
4%
0%
5%
5%
6%
12%
0%
0%
% > 0.239,
Adjusted
0%
4%
0%
0%
0%
0%
0%
0%
2%
0%
0%
TETRATECH
106
-------
Downstream Use Protection
April 2015
HUC8
06010106
Pigeon
06010108
Nolichucky
WQ Sta.
ID
E5495000
E5600000
E6450000
E6480000
FRB047A
FRB047B
FRB047C
03465500
03466208
03467609
Name
Pigeon R at NC-215 nr Canton
Pigeon R at SR-1642 at Clyde
Cataloochee Cr at SR-1395 nr
Cataloochee
Pigeon R at SR-1338 nr Hepco
Lake Junaluska
Lake Junaluska
Lake Junaluska
Nolichucky R at Embreeville, TN
Big Limestone Cr nr Limestone, TN
Nolichucky R nr Lowland
% > 0.239,
Current
15%
24%
3%
10%
0%
0%
0%
0%
9%
5%
% > 0.239,
Adjusted
0%
4%
0%
5%
0%
0%
0%
0%
5%
1%
3.6.2 Simplified Assignment of Load, Douglas Reservoir
Douglas Reservoir has a large watershed with relatively long travel times from more distant locations.
We used the southeastern regional SPARROW phosphorus model (Garcia et al., 2011) to calculate
attenuation. In contrast with SPARROW models in which attenuation is solely a function of travel time
(but sorted into different flow range bins), this model calculates attenuation relative to the product of
travel time divided by mean water depth, with a coefficient of 0.048 m/day. Unfortunately, average
depths were not readily available for the reach network. For example, if a depth of 1 m is assumed, the
adjusted targets at the monitoring stations would increase from 0.090 to a maximum of 0.119 mg/L.
A stratified regression loading analysis was performed at two downstream stations, French Broad River
near Newport (03455000) and Nolichucky River near Lowland (03467609), which together account for
83 percent of the total drainage area of Douglas Reservoir. The analysis was restricted to water years
1997-2013 due to changes in point source loads and land use in the watershed relative to earlier time
periods. Gaps in the flow record for Nolichucky River were filled by a linear regression against
Nolichucky River near Embreeville (03465500), with an R2 of 0.59. The correlation is lower than would
be expected given the locations of the stations on the stream network because of the influence of the
Nolichucky Dam, which lies between the stations. .
As with the analysis for Holcombe Flowage, the stratified regression approach provided a better fit to the
loads derived from observed data than the simple LOADEST approach. The stratified regression was
applied to develop a relationship between the natural log of load and the natural log of flow at the two
downstream stations. The resulting R2values were 0.66 for the French Broad River and 0.21 for the
Nolichucky River.
We used the stratified regression relationships to develop a complete time series of predicted loads and
concentrations for water years 1997 - 2013. For the French Broad River, the long-term flow-weighted
total phosphorus concentration estimate is 0.147 mg/L. For the Nolichucky River, the long-term flow-
weighted total phosphorus concentration estimate is 0.089 mg/L, which is just below the downstream
I TETRATECH
107
-------
Downstream Use Protection
April 2015
target of 0.090 mg/L. A compliance scenario was developed for the French Broad by iteratively reducing
the concentrations in the upper stratum only (above 3,562 cfs) until the flow-weighted concentration met
the desired target of 0.0390. The analysis includes a random error term based on the root mean squared
error in the estimates of ln(P). Achieving the 0.090 target required a reduction of 64.7 percent in the
upper stratum. After the target flow-weighted concentration was achieved, 31.1 percent of the individual
observations were still greater than 0.09 mg/L (Figure 3-24). The distribution is strongly skewed, so
some individual observations are much greater than 0.09 mg/L.
1 finn
1 Ann -
1 ?nn -
j. -i nnn
3
5snn
tc
tt £nn
Ann
?nn
J^-*-~——
^
/
/
r
1 1 .
V I i i i i i i i i i i i i i
000000000000000
omLnooomLnoo*-imiooo*-imio
OOOO'H'H'H'Hoioioioirororo
ooooooooooooooo
Total P (mg/l)
- 100%
90%
80%
70%
60% £
+-i
cno/ !r
40% |
30%
20%
10%
no/
Figure 3-24. Distribution of TP Concentrations after Achieving Flow-Weighted Concentration of
0.090 mg/L, French Broad River (TN)
For the Nolichucky River, direct estimates with the stratified regression have a flow-weighted mean
concentration that is below the target; however, the series created with random perturbations in ln(P)
exceeds the target by a small amount at 0.11 mg/L. For the Nolichucky, a reduction in the upper stratum
(greater than 1,429 cfs) of 20.0 percent was needed to achieve the flow-weighted concentration target of
0.090 mg/L, resulting in 18 percent of individual observations remaining greater than the target (Figure
3-14).
TETRATECH
108
-------
Downstream Use Protection
April 2015
2,000
1,800
1,600
1,400
c 1,200
3
8 1,000
5 800
600
400
200
0
^
100%
90%
80%
70%
60% £
50%
40%
30%
20%
10%
0%
ii
QJ
000000000000000
dddddddddddddcid
Total P (mg/l)
Figure 3-25. Distribution of TP Concentrations after Achieving Flow-Weighted Concentration of
0.090 mg/L, Nolichucky River (TN)
A load-based analysis was also be done using the USGS SPARROW model. Douglas Reservoir and its
watershed fall within the SPARROW total phosphorus model for the Southeastern United States region
(Garcia et al., 2011). The SPARROW model predicts a flow-weighted concentration within Douglas
Reservoir of 0.102 mg/L, suggesting a need for an 11.5 percent reduction in loads (much less than the
LOADEST analysis). In addition, the flow-weighted concentrations from SPARROW differ in relative
ordering and magnitude from those obtained with the site-specific regression analysis (Table 3-5). As
with Holcombe Flowage, the SPARROW load estimates differ substantially from the estimates obtained
from analysis of data from an individual station.
Table 3-16. Comparison of SPARROW Flow-weighted Concentrations to Results of Stratified
Regression Analysis, Douglas Reservoir
Gage Location
03455000 (French Broad River near
Newport)
03467609 (Nolichucky River near
Lowland)
SPARROW flow-weighted TP
concentration (mg/L)
0.042
0.154
Stratified regression flow-
weighted TP concentration (mg/L)
0.147
0.089
3.6.3 Network Spatial Correlation Analysis
3.6.3.1 Model Estimation
Twenty-six observation sites within the Douglas Reservoir stream network were used to derive the spatial
regression model. For each site, the accumulated upstream area of different land uses and the area of
local land uses (both in km2) were calculated and considered as regression factors. A variety of different
spatial covariance models were tested with the SSN package (Table 3-17).
Table 3-17. Spatial Model Components Tested for the Douglas Reservoir Watershed Model
TETRATECH
109
-------
Downstream Use Protection
April 2015
Regression factor
Total drainage area
Accumulation area of water body
Accumulation area of agriculture and pasture
Accumulation area of urban area
area of local RCA
Agriculture and pasture area of local RCA
Urban area of local RCA
Urban area ratio of local RCA
Agriculture and pasture area ratio of local RCA
Water body area ratio of local RCA
Upstream distance
Regression
factor
acronyms
acc_area
acc_wb
acc_agrigr
acc_urban
rca_area
rca_agrigr
rca_urban
urbanratio
agrigrrati
wbratio
upDist
Spatial covariance
models
Tail. up
Tail. down
Euclidean
Kernel function
type
Linear with sill
Spherical
exponential
Mariah
Cauchy
To evaluate the potential role of different regressors the potential models were first tested without spatial
covariance. Fitted results for the complete model are shown in Table 3-18.
Table 3-18. Evaluation of Regression Models for Douglas Reservoir with all the Regression
Factors
Coefficients
Estimate Parameter
Std. Error
t-value
Pr(>|t|)
Significance
code
Model: TP ~ acc_area + acc_wb + acc_agrigr +acc_urban+ rca_agrigr +
urban_ratio+agrigrrati + wbratio+upDist+rca_area
(Intercept)
acc_area
acc_wb
acc_agrigr
acc_urban
rca_agrigr
1.10E-01
-1.73E-04
-1.90E-02
1.92E-03
9.07E-08
5.91 E-02
8.50E-02
7.83E-05
1.98E-02
8.05E-04
2.49E-03
1.70E-01
1.294
-2.214
-0.957
2.378
0
0.349
0.2165
0.0439
0.3547
0.0322
1
0.7326
*
*
TETRATECH
110
-------
Downstream Use Protection
April 2015
Coefficients
Urbanratio
Agrigrrati
Wbratio
upDist
rca_area
rca_urban
Estimate Parameter
-2.52E-02
-4.24E-01
-2.70E-01
-3.39E-03
5.41 E-03
2.06E-02
Std. Error
2.20E-01
4.83E-01
1.93E-01
2.68E-02
2.07E-02
2.04E-01
t-value
-0.115
-0.878
-1.4
-0.126
0.261
0.101
Pr(>|t|)
0.9104
0.3948
0.1834
0.9012
0.798
0.921
Significance
code
Note: Significance codes: <0.001 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 " 1
The full regression model has a relatively high generalized R2 value (0.65); however, the regressors are
calibrated with one another and the fitted regression function implies that urban land use is associated
with better water quality, which is counter-intuitive. Further, most of the coefficients are not statistically
significant. Therefore, the full regression model was simplified by stepwise elimination of non-
significant factors, resulting in a single regressor model in which TP is predicted as a function of
acc_agrigr.
The selected model is shown in Table 3-19. Both the intercept and acc_agrigr have a significance level of
0.001 or better and suggests that area in agriculture is a significant predictor of TP concentration in the
streams of this watershed which is an expected result.
Table 3-19. Selected Predictive Model for Douglas Reservoir
Coefficient
Intercept
acc_agrigr
Parameter Estimate
5.40E-02
2.36E-04
Std. Error
1.30E-02
6.48E-05
t-value
4.15
3.638
Pr(>|t|)
0.00036
0.00131
Significance
code
***
**
We then investigated different spatial covariance model combinations together with the regression
function (Table 3-20). The exponential kernel function was used to compare different spatial covariance
components.
Table 3-20. Spatial Covariance Models Evaluated, Douglas Reservoir
Number
1
2
3
4
5
Formula
TP ~ acc_agrigr
TP ~ acc_agrigr
TP ~ acc_agrigr
TP ~ acc_agrigr
TP ~ acc_agrigr
Variance Components
Exponential. tailup + Exponential.taildown + Exponential. Euclid + Nugget
Exponential. tailup + Exponential. taildown+ Nugget
Exponential. tailup + Exponential. Euclid+ Nugget
Exponential.taildown + Exponential. Euclid+ Nugget
Exponential. tailup+ Nugget
TETRATECH
111
-------
Downstream Use Protection
April 2015
Number
6
7
Formula
TP ~ acc_agrigr
TP ~ acc_agrigr
Variance Components
Exponential. taildown+ Nugget
Exponential. Euclid+ Nugget
As was presented above in the discussion for the Holcombe Flowage example, spatial covariance models
are selected first for lowest negative log-likelihood (neg2LogL), with ties resolved by the standardized
mean-squared prediction error (std.MSPE). On the basis of neg2LogL, the spatial variance model with
taildown and Euclidean distsance components (No.4) was determined to provide the best fit (Table 3-21).
Choice of a Euclidean distance model makes some sense in a watershed where agricultural land is a major
source of load, but is distributed in small patches throughout the watershed. The observation points along
the stream network may not represent the spatial impact of agriculture very accurately if the sampling
density is low. The Euclidean distance thus may serve as a surrogate for upstream network influences. It
also may reflect spatial patterns of native soil phosphorus concentrations.
Table 3-21. Performance of Spatial Covariance Models, Douglas Reservoir
Number
1
2
3
4
5
6
7
Variance Components
Exponential. tailup + Exponential. taildown +
Exponential. Euclid + Nugget
Exponential. tailup + Exponential. taildown+ Nugget
Exponential. tailup + Exponential. Euclid+ Nugget
Exponential. taildown + Exponential. Euclid+ Nugget
Exponential. tailup+ Nugget
Exponential. taildown+ Nugget
Exponential. Euclid+ Nugget
neg2Logl_
-68.8421
-67.8262
-68.8428
-69.5757
-67.8258
-69.0542
-67.8686
std.MSPE
0.976457
0.94896
0.976362
1.011669
0.94914
1.015237
0.987441
The taildown plus Euclidean spatial covariance components were combined with the spatial regression
model based on the results of Table 3-21. Different kernel functions for the selected two spatial
covariance were then tested. Using the same performance indicators (neg2LogL, std.MSPE), the best-fit
spatial covariance components are Exponential Taildown and Cauchy Euclidean.
The final spatial regression model has the following form:
TP = 0.0565 + 2.27 * lO~4Aag + Scov(Exponential. taildown + Cauchy. Euclidean),
where TP is the concentration of total phosphorus (mg/L), Aag is the accumulated agriculture and pasture
area (km2), and SCOv is the spatial covariance corrector vector. This model has a generalized R2 of 0.34
and a median residual of-0.017 mg/L.
3.6.3.2 Population Level Results for Douglas Reservoir Watershed
TETRATECH
112
-------
Downstream Use Protection
April 2015
The selected spatial regression model was used to estimate the average TP concentration of stream
segments through the reach network (Figure 3-26). The histogram of TP concentrations is shown in
Figure 3-27. The mean of the TP concentration for all segments is 0.065 mg/L and the standard deviation
is 0.034 mg/L. The minimum predicted TP concentration within the stream network is 0.02 mg/L and the
maximum predicted TP concentration is 0.6mg/L. The 75th percentile predicted TP concentration is less
than 0.06mg/L, and more than 50% of predicted TP concentrations were within a range from 0.05 mg/L to
0.06 mg/L. This implies that the structure of spatial variation of TP concentrations within the stream
network has a limited impact, in which case random sampling of stream sites would be a simple and
direct way to assess the population level TP concentrations.
ID
4-1
ra
c
TJ
0
0
O
(D
CO
(N
10
P3
(0
CO
GO
CO
(D
in
CO
(N
id
CO
Standard Errors
• 0.154
0 0.0484
0 0.0483
0 0.0431
0 0.0475
00.00741
Predictions
0.02 to 0.05
0.05 to 0.05
0.05 to 0.05
0.05 to 0.05
0.05 to 0.05
0.05 to 0.05
0.05 to 0.05
* 0.05 to 0.06
* 0.06 to 0.07
* 0.07 to 0.6
-83.5
-83.0
-82.5
-82.0
x-coordinate
Figure 3-26. TP Concentrations Estimated by Spatial Regression Model, Douglas Reservoir
TETRATECH
113
-------
Downstream Use Protection
April 2015
o
o
o
o
o
o
t 8
01
I s
u- e
o
o
o
0.0
0.1
1
02
0.3
TP (mg/L)
1
0.4
0.5
06
Figure 3-27 Histogram of Predicted TP Concentrations, Douglas Reservoir
3.6.4 Sampling Results on Spatially Correlated Network
3.6.4.1 Prior Information on TP Targets
The Douglas Reservoir watershed lies with Level IV ecoregion 67g. The Tennessee nutrient criterion
study (Denton et al., 2001) proposes a criterion TP concentration for this ecoregion as 0.09 mg/L, the 90th
percentile of the reference TP concentrations in the ecoregion. The reference values have a mean
concentration of 0.052 mg/L and a standard deviation of 0.085 mg/L.
Section 2.3.4 derived a target TP concentration for Douglas Reservoir of 0.04 mg/L. An alternative
calculation was done based on the Level IV stream criterion. Level IV results are not available for lakes,
but quantile mapping can be used to project the stream criterion into the lake based on quantile mapping
(Panofsky and Brier, 1968) from the Level III lake criterion to Level IV results for the sub-ecoregion:
TPsubJak = FTPecolak(FTPecostr(TPsub_strV
where Vindicates the cumulative distribution function and F~l is the inverse cumulative distribution
function. In the nutrient criterion study, the critical value of TP in stream is defined as the 90th percentile
TP value. Given the median and 95th percentile value of TP concentration, a log-normal distribution was
fit to the TP concentration in both streams and lakes. Then the percentile of critical TP value (0.09 mg/L)
in the level III ecoregion streams was retrieved and was used to project the lake critical TP concentration.
In this case, the projected critical lake TP concentration in sub-ecoregion 67g is 0.05 mg/L and projected
mean lake TP concentration is 0.04 mg/L.
Based on this prior information, two downstream endpoint TP concentration scenarios were created for
the upper and lower boundaries. The first downstream endpoint TP concentration scenario is a reference
scenario, based on the 90th percentile value of 0.05 mg/L. The critical scenario has a mean concentration
of 0.05 mg/L (as opposed to a 90th percentile), and the observed concentrations have a CV of 0.695. Log-
normal probability density functions were fit to both distributions (Figure 3-28).
TETRATECH
114
-------
Downstream Use Protection
April 2015
!=
0)
Q
CO
-Q
O
in
(Si
o
r-j
Reference Situation
Critical Situation
-o-o-o-o-o-O'
-8-8-9-Q-0-S-8-O—©—e—e—o
0.00
0.05
0.10 0.15 0.20
TP concentration (mg/L)
0.25
0.30
Figure 3-28. PDFs of Reference and Critical Stream TP Concentration Scenarios at the
Downstream Pour Point, Douglas Reservoir
3.6.4.2 Thresholds for the Number of Exceedances
The weighted mean exceedance ratio and one-sided 95% confidence intervals were calculated for both
endpoint TP concentration scenarios (Figure 3-29, Figure 3-30). Only cases in which the number of
sampling sites is greater than or equal to 10 are presented. The upper bound of critical scenario and the
lower bound of reference scenario were used to calculate the "safe" and "danger" thresholds for number
of sample means exceeding the target.
Exceedance Ratio with 95% Confidence Interval
0.30
0.25
0.20
0.15
0.10
0.05
0.00
10 11 12 13 14 15 16 17 18 19 20
•mean 0.24 0.25 0.24 0.24 0.24 0.24 0.24 0.24 0.24 0.24 0.24
'lowerbound 0.08 0.09 0.09 0.10 0.10 0.10 0.11 0.11 0.12 0.12 0.12
Number of sampling stations
Figure 3-29. Exceedance Ratio and One-sided 95% Confidence Interval for Reference Scenario,
Douglas Reservoir
TETRATECH
115
-------
Downstream Use Protection
April 2015
Exceedance Ratio with 95% Confidence Interval
0.60
0.50
0.40
0.30
0.20
0.10
0.00
10 11 12 13 14 15 16 17 18 19 20
0.30 0.30 0.30 0.30 0.30 0.29 0.29 0.29 0.29 0.29 0.30
upper bound 0.56 0.55 0.54 0.53 0.52 0.51 0.50 0.50 0.49 0.49 0.48
Number of sampling stations
Figure 3-30. Exceedance Ratio and One-sided 95% Confidence Interval for Critical Scenario,
Douglas Reservoir
We define two thresholds for the number of sample means exceeding the target. The lower bound
exceedance ratio was used to calculate the "safe" threshold. This threshold implies that if the number of
exceedance sites is smaller than the given threshold, there is only a small chance (less than 5%) that the
downstream endpoint TP concentration is exceeding the criterion TP concentration. In this case, we
could conclude with a high level of confidence that the TP concentration at the downstream pour point is
below the target. Similarly, the upper bound exceedance ratio was used to calculate the "danger"
threshold. If the number of sites with concentrations above this boundary is larger than the "danger"
threshold, there is a high probability that the TP concentration at the downstream pour point is higher than
the criterion TP concentration. Table 3-22 lists the thresholds for different sample sizes.
Table 3-22. Exceedance Thresholds (Count) for Different Sample Sizes, Douglas Reservoir
Sample Size
10
11
12
13
14
15
16
17
18
19
Safe Threshold
1
1
1
1
1
2
2
2
2
2
Danger Threshold
6
7
7
7
8
8
9
9
9
10
TETRATECH
116
-------
Downstream Use Protection
April 2015
Sample Size
20
Safe Threshold
2
Danger Threshold
10
3.6.4.3 Mean TP Concentration at Sampling Sites
For 12 different downstream endpoint TP concentrations, the mean TP concentration and first and third
quantiles for sampling sites were calculated for different sampling sizes and downstream concentration
(DC) targets (Table 3-23). The mean sample TP concentration differs little across DC when sample size
is larger than 10. The sample TP concentrations were fitted with log-normal probability distribution
functions for different downstream endpoint TP concentrations to derive the 25th and 75th percentile
values.
Table 3-23. TP Concentrations for Different Sample Sizes, Douglas Reservoir Watershed
Sample
Size
10
11
12
13
14
15
16
17
18
19
20
10
11
DC=0.005
0.063,
[0.042,0.060]
0.063,
[0.043,0.061]
0.063,
[0.044,0.061]
0.063,
[0.045,0.061]
0.063,
[0.044,0.061]
0.063,
[0.045,0.061]
0.063,
[0.045,0.061]
0.063,
[0.044,0.062]
0.063,
[0.046,0.062]
0.063,
[0.047,0.062]
0.063,
[0.046,0.062]
0.070,
[0.002,0.069]
0.071,
[0.005,0.069]
DC=0.045
0.057,
[0.040,0.057]
0.057,
[0.041,0.057]
0.057,
[0.043,0.057]
0.057,
[0.041,0.057]
0.057,
[0.041,0.057]
0.057,
[0.044,0.057]
0.057,
[0.044,0.057]
0.057,
[0.044,0.057]
0.057,
[0.045,0.057]
0.057,
[0.045,0.057]
0.057,
[0.045,0.057]
0.074,
[0.001,0.073]
0.074,
[0.002,0.073]
DC=0.063
0.058,
[0.028,0.057]
0.058,
[0.029,0.057]
0.058,
[0.028,0.057]
0.057,
[0.031,0.057]
0.057,
[0.031,0.057]
0.058,
[0.032,0.057]
0.057,
[0.031,0.057]
0.057,
[0.034,0.057]
0.057,
[0.035,0.057]
0.057,
[0.035,0.057]
0.057,
[0.035,0.057]
0.078,
[0.003,0.077]
0.078,
[0.001,0.077]
DC=0.081
0.063,
[0.019,0.062]
0.063,
[0.020,0.062]
0.063,
[0.024,0.062]
0.063,
[0.022,0.062]
0.063,
[0.024,0.062]
0.063,
[0.023,0.062]
0.063,
[0.025,0.062]
0.063,
[0.025,0.063]
0.063,
[0.027,0.063]
0.063,
[0.029,0.063]
0.063,
[0.028,0.063]
0.077,
[0.001,0.075]
0.077,
[0.003,0.076]
DC=0.099
0.065,
[0.008,0.064]
0.065,
[0.014,0.064]
0.065,
[0.016,0.064]
0.065,
[0.014,0.064]
0.065,
[0.014,0.064]
0.065,
[0.017,0.064]
0.065,
[0.023,0.064]
0.065,
[0.019,0.064]
0.065,
[0.018,0.064]
0.065,
[0.023,0.064]
0.065,
[0.025,0.064]
0.081,
[0.000,0.079]
0.081,
[0.001,0.079]
DC=0.12
0.067,
[0.005,0.066]
0.067,
[0.010,0.066]
0.067,
[0.010,0.066]
0.067,
[0.004,0.066]
0.067,
[0.012,0.066]
0.067,
[0.016,0.066]
0.067,
[0.016,0.066]
0.067,
[0.012,0.066]
0.067,
[0.017,0.066]
0.067,
[0.017,0.067]
0.067,
[0.018,0.067]
0.119,
[0.000,0.120]
0.119,
[0.000,0.120]
TETRATECH
117
-------
Downstream Use Protection
April 2015
Sample
Size
12
13
14
15
16
17
18
19
20
DC=0.005
0.070,
[0.008,0.069]
0.070,
[0.010,0.069]
0.071,
[0.011,0.070]
0.071,
[0.009,0.070]
0.071,
[0.007,0.070]
0.070,
[0.016,0.070]
0.070,
[0.015,0.070]
0.071,
[0.011,0.070]
0.071,
[0.017,0.070]
DC=0.045
0.074,
[0.008,0.073]
0.074,
[0.008,0.073]
0.074,
[0.008,0.073]
0.074,
[0.010,0.073]
0.074,
[0.010,0.073]
0.074,
[0.010,0.073]
0.074,
[0.012,0.073]
0.074,
[0.011,0.073]
0.074,
[0.015,0.073]
DC=0.063
0.078,
[0.007,0.077]
0.078,
[0.005,0.077]
0.078,
[0.006,0.077]
0.078,
[0.010,0.077]
0.078,
[0.007,0.077]
0.078,
[0.008,0.077]
0.078,
[0.010,0.077]
0.078,
[0.012,0.077]
0.078,
[0.010,0.077]
DC=0.081
0.077,
[0.000,0.076]
0.077,
[0.008,0.076]
0.077,
[0.003,0.076]
0.077,
[0.007,0.076]
0.077,
[0.003,0.076]
0.077,
[0.009,0.076]
0.077,
[0.010,0.076]
0.077,
[0.010,0.076]
0.077,
[0.013,0.076]
DC=0.099
0.081,
[0.001,0.079]
0.081,
[0.004,0.080]
0.081,
[0.002,0.080]
0.081,
[0.005,0.080]
0.081,
[0.007,0.080]
0.081,
[0.010,0.080]
0.081,
[0.008,0.080]
0.081,
[0.009,0.080]
0.081,
[0.012,0.080]
DC=0.12
0.119,
[0.000,0.120]
0.119,
[0.001,0.120]
0.119,
[0.002,0.120]
0.119,
[0.006,0.120]
0.119,
[0.008,0.120]
0.119,
[0.002,0.120]
0.119,
[0.008,0.120]
0.119,
[0.008,0.120]
0.119,
[0.016,0.120]
Note: Results show mean; first and third quantile values shown in parentheses.
For Douglas Reservoir, all stations have at least 10 samples and Monte Carlo simulation results with
sample size equal to 10 are presented for the mean TP concentration analysis. The exceedance
probability at the downstream pour point for different sampled mean TP concentrations was calculated
using the methods described in Section 3.4.5.4.
The reference and critical scenarios were also applied to the mean TP concentration analysis. The
exceedance probability range is derived from these two scenarios (Figure 3-31, Table 3-24). For different
sampled mean TP concentrations, the exceedance probability calculated with the reference situation is
considered the lower bound and upper bound for the value calculated with the critical situation.
TETRATECH
118
-------
Downstream Use Protection
April 2015
ooooooooooooooooo
CO
o
_
ro
_Q CD
e o
Q.
01
o
c
(D -rt-
E o
o>
o
ffl
OJ
o
p
o
Critical Situation
° Reference Situation
0.10
0.15 0.20
Sampled TP concentration (mg/L)
0.25
0.30
Figure 3-31. Exceedance Probability for Different Sample TP Concentrations, Douglas Reservoir
Table 3-24. Exceedance Probability for Different Sample TP Concentrations, Douglas Reservoir
TP (mg/L)
0.055
0.06
0.065
0.07
0.075
0.08
0.085
0.09
0.095
0.1
0.105
0.11
Exceedance Probability (%)
1.5-16.2
1.1-14.9
1.8-18.6
4.0-25.6
7.8-33.2
12.5-40.5
17.5-47.2
22.6-53.7
27.9-60.0
33.5-66.2
39.5-72.1
46.0-77.5
TP (mg/L)
0.18
0.185
0.19
0.195
0.2
0.205
0.21
0.215
0.22
0.225
0.23
0.235
Exceedance Probability (%)
97.9-99.7
98.4-99.8
98.8-99.8
99.1-99.9
99.4-99.9
99.5-99.9
99.7-100.0
99.8-100.0
99.8-100.0
99.9-100.0
99.9-100.0
99.9-100.0
TETRATECH
119
-------
Downstream Use Protection
April 2015
TP (mg/L)
0.115
0.12
0.125
0.13
0.135
0.14
0.145
0.15
0.155
0.16
0.165
0.17
0.175
Exceedance Probability (%)
52.7-82.2
59.4-86.1
65.8-89.4
71.7-92.0
77.0-94.1
81.6-95.6
85.4-96.8
88.6-97.7
91.2-98.3
93.3-98.8
94.9-99.1
96.1-99.4
97.1-99.6
TP (mg/L)
0.24
0.245
0.25
0.255
0.26
0.265
0.27
0.275
0.28
0.285
0.29
0.295
0.3
Exceedance Probability (%)
100.0-100.0
100.0-100.0
100.0-100.0
100.0-100.0
100.0-100.0
100.0-100.0
100.0-100.0
100.0-100.0
100.0-100.0
100.0-100.0
100.0-100.0
100.0-100.0
100.0-100.0
3.7 Conclusions
Results presented in this section address only two example watersheds and the analysis of Monte Carlo
simulation analyses of those two watersheds is open to further data exploration. Nonetheless, it is clear
that the problem of assessing instream concentrations relative to a downstream receiving water target is
amenable to a statistical analysis that evaluates whether or not sample means observed upstream in the
watershed are consistent with achieving the downstream target.
When land uses and other sources of nutrient loading are relatively homogenous and distributed evenly
throughout the watershed an analysis of evidence from watershed sampling sites could be made based
solely on the measured or estimated lognormal distribution CDF at the downstream pour point. This
approach breaks down when there are different types and sources of loads in the watershed, in which case
measurements at different observation sites are likely to exhibit strong spatial correlation. This situation
can be addressed through the development of spatial covariance representations (SSN/STARS) coupled
with either a regression analysis of loads based on landscape features or a simple mean representation for
homogenous source distributions. The regression analyses reported herein met with only moderate
success, but could likely be improved through better accounting of point source discharges in particular.
Taking into account stream network-based autocorrelation enables estimation of the expected value of
concentrations throughout the stream network. This distribution is dependent on the concentration and
load present at the downstream pour point. A relatively straightforward solution to the assessment
problem can be achieved by estimating a lower confidence limit on the distribution across all watershed
segments, based on the assumption that the downstream is at relatively undisturbed reference
concentration levels, and an upper confidence limit calculated under the assumption that the target
I TETRATECH
120
-------
Downstream Use Protection April 2015
concentration is just met. This leads to a three-tiered approach: Site concentrations below the lower
confidence limit can be deemed "safe" in that they do not present statistically significant results that the
downstream pour point concentration will exceed targets. In contrast, sites where the flow-weighted
mean concentration is above the upper confidence limit are not consistent with the distribution of site
loads that would be expected to achieve the downstream criterion.
In the preceding sections we propose methods for determining the appropriate confidence limits when
network-based spatial correlation is present. This requires estimation of the spatial correlation structure
and underlying regression on watershed characteristics (if appropriate), from which an empirical
distribution of the relationship between the desired downstream pour point concentration and
concentrations at sites distributed throughout the watershed can be obtained. Additional research is
needed to determine simpler methods for evaluating appropriate adjustment factors to relate the
distribution of the downstream pour point concentration to concentrations in individual segments
throughout the watershed.
4 REFERENCES
Agresti, A., and B.A. Coull. 1998. Approximate is better than 'exact' for interval estimation of binomial
proportions. The American Statistician, 52: 119-126.
Baker, L.A., P. L. Brezonik, E.S. Edgerton, and R. W. OGBURN III. 1985. Sediment acid neutralization
in softwater lakes. Water Air SoilPollut., 25: 2 15-230.
Barbour, M. T., J. Gerritsen, B. D. Snyder, and J. B. Stribling. 1999. Rapid Bioassessment Protocols for
Use in Wadeable Streams and Rivers: Periphyton, Benthic Macroinvertebrates, and Fish. 2nd ed. U.S.
Environmental Protection Agency, Office of Water, Washington, DC. EPA 841-B-99-002.
Canfield, D.E., Jr., and R.W. Bachmann. 1981. Prediction of total phosphorus concentrations,
chlorophyll a and Secchi depths in natural and artificial lakes. Canadian Journal of Fisheries and
Aquatic Sciences, 38(4):414-423.
Carlson RE. 1977. A trophic state index for lakes. Limnology and Oceanography 22:361-369.
Denton, G.M., D.H. Arnwine, and S.H. Wang. 2001. Development of Regionally-based Interpretations
of Tennessee's Narrative Nutrient Criterion. Tennessee Dept. of Environment and Conservation,
Nashville, TN. http://www.tennessee.gov/assets/entities/environment/attachments/nutrient_fmal.pdf
Garcia, A.M., A.B. Hoos, and S. Terziotti. 2011. A regional modeling framework of phosphorus sources
and transport in streams of the southeastern United States. Journal of the American Water Resources
Association, 47(5): 991-1010.
Hirsch, R. M., D.L. Moyer, and S.A. Archfield, 2010. Weighted Regressions on Time, Discharge, and
Season (WRTDS), with an Application to Chesapeake Bay River Inputs. Journal of the American Water
Resources Association, 46: 857-880. doi: 10.1111/j.1752-1688.2010.00482.x
Hoos, A.B. and G. McMahon. 2009. Spatial analysis of instream nitrogen loads and factors controlling
nitrogen delivery to stream in the southeastern United States using spatially referenced regression on
watershed attributes (SPARROW) and regional classification frameworks. Hydrological Processes,
doi:10.1002/hyp.7323.
Isaak D., C. Luce, B. Rieman, D. Nagel, E. Peterson, D. Horan, S. Parkes, and G. Chandler. 2010.
Effects of recent climate and fire on thermal habitats within a mountain stream network. Ecological
Applications, 20(5): 1350-1371.
Kalff, J. 2002. Limnology: Inland Water Ecosystems. Prentice-Hall, Upper Saddle River, NJ
TETRATECH
-------
Downstream Use Protection April 2015
Larsen, D.P., and H.T. Mercier. 1976. Phosphorus retention capacity of lakes. Journal of the Fisheries
Research Board of Canada, 33:1742-1750.
Novotny, V. 2004. Simplified databased Total Maximum Daily Loads, or the world is log-normal.
Journal of Environmental Engineering, 130:674-683.
Organization for Economic Cooperation and Development (OECD). 1982. Eutrophication of Waters:
Monitoring, Assessment and Control. Organization for Economic Cooperation and Development, Paris,
France.
Panofsky, H.A., and G.W. Brier. 1968. Some Applications of Statistics to Meteorology. Pennsylvania
State University, University Park, PA
Paul, M.J., A. Herlihy, D. Bressler, L. Zheng, and A.Roseberry-Lincoln. 2014. Methodologies for
Development of Numeric Nutrient Criteria for Freshwaters. Prepared by Tetra Tech, Inc., Research
Triangle Park, NC and Tetra Tech, Inc., Center for Ecological Sciences, Research Triangle Park, NC
Peterson E.E., and J.M. Ver Hoef. 2010. A mixed-model moving-average approach to geostatistical
modeling in stream networks. Ecology, 91(3): 644-651.
Peterson E.E., and J.M. Ver Hoef. 2014. STARS: An ArcGIS toolset used to calculate the spatial
information needed to fit spatial statistical models to stream network data. Journal of Statistical
Software, 56(2).
Press, W.H., B.P. Flannery, S.A. Teukolsky, and W.T. Vetterling. 1992. Numerical Recipes, the Art of
Scientific Computing. Cambridge University Press, Cambridge.
Preston, S.D., R.B. Alexander, G.E. Schwarz, and C.G. Crawford. 2011. Factors affecting stream
nutrient loads: A synthesis of regional SPARROW model results from the continental United States.
Journal of the American Water Resources Association, 47(5): 891-915.
Qian, S.S., R.S. King, and C.J. Richardson. 2003. Two statistical methods for the detection of
environmental thresholds. Ecological Modeling 166: 87-97.
Qian, S.S., K.H. Reckhow, J. Zhai, and G. McMahon. 2005. Nonlinear regression modeling of nutrient
loads in streams: A Bayesian approach. Water Resources Research, vol. 41, W07012,
doi: 10.1029/2005WR003986.
R Core Team (2014). R: A language and environment for statistical computing. R Foundation for
Statistical Computing, Vienna, Austria.URL http://www.R-project.org/.
Robertson, D.M., and D.A. Saad. 2011. Nutrient inputs to the Laurentian Great Lakes by source and
watershed estimated using SPARROW watershed models. Journal of the American Water Resources
Association, 41'(5): 1011-1033.
Runkel, R.A., C.G. Crawford, and T.A. Cohn. 2004. Load Estimator (LOADEST): A FORTRAN
Program for Estimating Constituent Loads in Streams and Rivers. Techniques and Methods Book 4,
Chapter A5. U.S. Geological Survey, Reston, VA.
Robertson, D.M., and D.A. Saad. 2011. Nutrient inputs to the Laurentian Great Lakes by source and
watershed estimated using SPARROW watershed models. Journal of the American Water Resources
Association, 41'(5): 1011-1033.
Runkel, R.L., C.G. Crawford, and T.A. Cohn. 2004. Load Estimator (LOADEST): A FORTRAN
Program for Estimating Constituent Loads in Streams and Rivers. U.S. Geological Survey Techniques
and Methods Book 4, Chapter A5. U.S. Geological Survey, Reston, VA.
TETRATECH
122
-------
Downstream Use Protection April 2015
Saad, D.A., G.E. Schwarz, D.M. Robertson, and N.L. Booth. 2011. A multi-agency nutrient dataset used
to estimate loads, improve monitoring design, and calibration regional nutrient SPARROW models.
Journal of the American Water Resources Association, 47(5): 933-949.
Shannon, E.E., and P.L. Brezonik. 1972. Limnological characteristics of north and central Florida lakes.
Limnology and Oceanography, 17:97-110.
Steward, J.S., and E.F. Lowe. 2010. General empirical models for estimating nutrient load limits for
Florida's estuaries and inland waters. Limnology and Oceanography, 55(1): 433-445
Theobald, D., J. Norman, E.E. Peterson, S. Feraz, A. Wade, and M.R. Sherburne. 2006. Functional
Linkage of Waterbasins and Streams (FLoWS) vl User's Guide: ArcGIS Tools to Analyze Freshwater
Ecosystems. Colorado State University, Fort Collins, CO
TDEC. 2001. Development of Regionally Based Interpretations of Tennessee's Narrative Nutrient
Criterion. Tennessee Department of Environment and Conservation, Tennessee Water Quality Control
Board. Nashville, TN.
Theobald, D., J. Norman, E.E. Peterson, S. Feraz, A. Wade, and M.R. Sherburne. 2006. Functional
Linkage of Waterbasins and Streams (FLoWS) vl User's Guide: ArcGIS Tools to Analyze Freshwater
Ecosystems. Colorado State University, Fort Collins, CO.
United States Environmental Protection Agency (USEPA). 1991. Technical Support Document for
Water Quality-based Toxics Control. EPA/505/2-90-001. Office of Water, U.S. Environmental
Protection Agency, Washington, DC.
USEPA. 1998. Lake and reservoir bioassessment and biocriteria. Technical guidance document. United
States Environmental Protection Agency, Office of Water, Washington, DC. EPA 841-B-98-007.
USEPA. 2000. Nutrient Criteria Technical Guidance Manual, Lakes and Reservoirs. United States
Environmental Protection Agency, Office of Water, Washington, DC. EPA-822-BOO-001.
USEPA. 2001. Nutrient Criteria Technical Guidance Manual, Estuarine and Coastal Marine Waters.
United States Environmental Protection Agency, Office of Water, Washington, DC. EPA-822-B-01-003.
USEPA. 2009. National Lakes Assessment: A Collaborative Survey of the Nation's Lakes. U.S.
Environmental Protection Agency, Office of Water and Office of Research and Development,
Washington, D.C. EPA 841-R-09-001.
USEPA. 2010. Technical Support Document for U.S. EPA's Final Rule for Numeric Criteria for
Nitrogen/Phosphorus Pollution in Florida's Inland Surface Fresh Waters.
Van Buren, M.A., W.E. Watt, and J. Marsalek. 1997. Application of the log-normal and normal
distributions to stormwater quality parameters. Water Research, 31(1): 95-104. .
Ver Hoef, J.M., and E.E. Peterson. 2010. A moving average approach to spatial statistical models of
stream networks. The Journal of the American Statistical Association, 489: 6-18.
Ver Hoef, J.M., E.E. Peterson, D. Clifford, and R. Shah. 2014. SSN: An R package for spatial statistical
modeling on stream networks. Journal of Statistical Software, 56(3).
Virginia Water Resources Research Center (VWRRC). 2005. Issues Related to Freshwater Nutrient
Criteria for Lakes and Reservoir in Virginia. Virginia Water Resources Research Center, Blacksburg, VA.
VWRRC Special Report SR27-2005.
Vollenweider, R.A. 1975. Input-output models with special reference to the phosphorus loading concept
in limnology. Schweizerische Zeitschriftfur Hydrologie (Swiss Journal of
TETRATECH
123
-------
Downstream Use Protection April 2015
Vollenweider, R.A. 1976. Advances in defining critical loading levels for phosphorus in lake
eutrophication. Memorie dell'Istituto Italicmo di Idrobiologia, 33:53-83.
Walker, W.W. 1987. Empirical Methods for Predicting Eutrophication in Impoundments; Report 4,
Phase III: Applications Manual. Technical Report E-81-9. U.S. Army Engineer Waterways Experiment
Station, Vicksburg, MS.
Walker, W.W., 1999. Simplified Procedures for Eutrophication Assessment and Prediction: User
Manual. Instruction Report W-96-2. U.S. Army Corps of Engineers Waterways Experiment Station,
Vicksburg, MS.
Wellen, C., G.B. Arhonditsis, T. Labencki, and D. Boyd. 2014. Application of the SPARROW model in
watersheds with limited information: A Bayesian assessment of the model uncertainty and the value of
additional monitoring. Hydrological Processes, 28: 1260-1283, doi:10.1002/hyp.9614.
Yuan, L. 2006. Estimation and Application of Macroinvertebrate Tolerance Values. United States
Environmental Protection Agency, National Center for Environmental Assessment, Office of Research
and Development, U.S. Environmental Protection Agency, Washington, B.C. EPA/600/P-04/116F.
APPENDIX A METHODS FOR CORRELATED RANDOM VARIABLE
GENERATION
This appendix describes the mathematical methods used to generate multiple random variables that
preserve a correlation structure. Derivation of the generating equations is most readily illustrated in
vector-matrix notation, but, for simplicity, we focus on the bivariate case in algebraic notation.
Suppose Gx is a column vector of length two, containing realizations (at location x) of the variables ax and
bx,Mis a column vector of length two containing the means of the variables, /ua and /Ub, and Ex is a
disturbance vector of length two containing the deviations of each realization from its mean value at
location x, sa,x, and Sb,x. Then
GX=M + EX
or, in algebraic notation,
ax = /ua + sa,x
The two variables are correlated, so the disturbance terms are not independent of one another. The
variance -covariance matrix, S, ofE can be written as
S =
Sn 812
S21 S22
2
-tT
- V
C
b,a
a,b
where a2a is the variance of series a, CVa,b is the covariance of series a and b, and so on. Now suppose S
is decomposed into two identical matrices R, such that RRT=S, where "T" stands for the transpose of the
matrix. Algebraically, this means
TETRATECH
124~
-------
Downstream Use Protection
April 2015
R»RT =
rum
l2 T22
= s
2 , 2 _ _ 2
rn + rn - sn - crfl
r2i rn + r22 rn = s2i = CVa,b = C
If the random variables are generated by
GX=M + RWX, or
where Wx is a vector of standard normal variates with mean 0 and standard deviation of 1, then the proper
variance and covariance of the correlated variables will be reproduced. This is easily shown, as the
variance -covariance matrix of Gt is
Var(Gx)= (Gx - M)(GX - A/f = (M+RWT -
x RT = ai* RRr = c>2w*
It remains to find the matrix R such that RRT = S. For any covariance matrix, R can be expressed as a
lower triangular matrix, such that all entries above the diagonal are zero (the Cholesky decomposition).
For the two variable case,
RRT =S, or
Sll Sn
$21 $22
For the general case, this may be solved by using the Cholesky decomposition procedure, found in most
texts on linear algebra or numerical analysis (e.g., Press et al, 1992).
In this project, Gxis the TP concentration at location x, Mis the linear regression component based on the
sources/sinks terms (land uses) and the Exis the spatial correlation matrix of all the prediction sites.
Monte Carlo simulations were used to explore the relationship of upstream exceedances to downstream
conditions under conditions of network spatial correlation. We assumed the outlet (pour point) of the
stream network is at the receiving waterbody of interest. To explore cases in which we wish to hold the
concentration at this point to a specified target value and allow the remainder of the network to vary
randomly in accordance with the spatial correlation structure, we set the row of the spatial covariance
matrix that corresponds to the pour point to the first row. After Cholesky decomposition, the spatial
covariance value associated with this point is determined by the single top left entry (rn) in the
decomposed lower triangle matrix. If the first random variate in the vector Wx is set to zero, the generated
concentration at the pout point will be equal to the desired mean or regression result. The standard
TETRATECH
125
-------
Downstream Use Protection April 2015
W f n W2
deviation of the remaining normal deviates in x was then inflated by 1 to reflect the reduction of
& J Vn-l/
one degree of freedom caused by fixing the value at the pour point.
TETRATECH
126
------- |