EPA 841-R-16-114 October 2024
National Lakes Assessment 2012:
Technical Report
U.S. Environmental Protection Agency
Office of Wetlands, Oceans and Watersheds
Office of Research and Development
Washington, DC 20460
Version 1.1
October 2024
-------
Version History
Version
Date
Revisions or Comments
1.0
April 2017
Final document published with survey report
1.1
October 2024
Updated to address errors identified in Chapter 8, Section 8.4: Stressor
Extent, Relative Risk and Attributable Risk. Corrections to this section
aligned the calculations with the analytical approach used in the NLA
2012 report.
Suggested citation for this document is: USEPA. 2024. National Lakes Assessment 2012: Techical Report,
Version 1.1. EPA 841-R-16-114. U.S. Environmental Protection Agency, Washington, D.C.
Website: https://www.epa.gov/national-aquatic-resource-surveys/nla
2 NLA 2012 Technical Report. October Version 1.1
-------
Table of Contents
Chapter 1: Project Overview 11
1.1 Overview 11
1.2 Objectives of the National Lakes Assessment 11
Chapter 2: Survey Design and Population Estimates 13
2.1 Description of sample design 13
2.1.1 Stratification 13
2.1.2 Unequal Probability Categories 13
2.1.3 Panels 14
2.1.4 Expected Sample Size 14
2.2 Sample frame summary 20
2.3 Survey analysis 20
2.4 Estimated extent of the NLA lake population and implications for reporting 21
2.5 Literature cited 22
Chapter 3: Reference Condition and Condition Benchmarks 24
3.1 Background information 24
3.2 Pre-sampling screening (hand-picked sites only) 24
3.3 Post-sampling screening for biological reference condition 25
3.4 Post-sample screening for nutrient reference condition 28
3.5 Literature cited 29
Chapter 4: Benthic Invertebrates 31
4.1 Background information 31
4.2 Data preparation 31
4.2.1 Standardizing counts 31
4.2.2 Autecological characteristics 31
4.2.3 Tolerance values 32
4.2.4 Functional feeding group and habitat preferences 32
4.2.5 Taxonomic resolution 32
4.3 Multimetric index development 33
4.3.1 Data Set 33
4.3.2 Low Macroinvertebrate Numbers 33
4.3.3 Ecoregion Classification 33
4.3.4 Metric Screening 34
4.3.5 All Subsets MMI selection 34
4.3.6 Setting MMI Thresholds 38
3 NLA 2012 Technical Report. October Version 1.1
-------
4.4 Literature cited 41
Chapter 5: Physical Habitat 42
5.1 Background information 42
5.2 Data preparation 43
5.3 Methods 44
5.3.1 Study area and site selection 44
5.3.2 Field sampling design and methods 44
5.3.3 Classifications 45
5.3.4 Calculation of lake physical habitat metrics 46
5.3.5 Calculation of summary physical habitat condition indices 52
5.3.6 Deriving expected index values under least-disturbed conditions 57
5.3.7 Condition Criteria for Nearshore Lake Physical habitat 58
5.4 Least-disturbed reference distributions and regressions (from sections 5.3.6 and 5.3.7) 60
5.4.1 Disturbance within least-disturbed reference sites 60
5.4.2 Null Model Results for RVegQ, LitCvrQ, and LitRipCvQ: 61
5.4.3 O/E Model Results for RVegQ, LitCvrQ, and LitRipCvQ: 61
5.4.4 Null Model Results for Lake Drawdown and Level Fluctuations: 62
5.5 Precision of physical habitat indicators 63
5.6 Physical habitat index responses to anthropogenic disturbance 64
5.7 Discussion 65
5.8 Literature cited 67
Chapter 6: Water Chemistry 97
6.1 Background information 97
6.2 Threshold development 97
6.2.1 Acidity and Dissolved Oxygen 97
6.2.2 Trophic State 98
6.2.3 Total nitrogen, total phosphorus, chlorophyll-a, and turbidity 98
6.3 Literature cited 101
Chapter 7: Zooplankton 102
7.1 Background information 102
7.2 Methods 103
7.2.1 Field Methods 103
7.2.2 Laboratory Methods 106
7.3 Data Preparation 106
7.3.1 Data Quality Assurance 106
4 NLA 2012 Technical Report. October Version 1.1
-------
7.3.2 Master Taxa List 107
7.3.3 Aggregations and Rarefaction of Count Data 107
7.4 Zooplankton MMI Development 108
7.4.1 Regionalization 108
7.4.2 Least and Most Disturbed Sites 109
7.4.3 Least Disturbed Sites: Calibration versus Validation 110
7.4.4 Candidate Metrics 110
7.4.5 Final Metric Selection Ill
7.4.6 Metric Scoring 112
7.5 Zooplankton MMI Metric Composition and Performance 112
7.5.1 Coastal Plain MMI 112
7.5.2 Eastern Highlands MMI 115
7.5.3 Plains MMI 115
7.5.4 Upper Midwest MMI 118
7.5.5 Western Mountains MMI 118
7.6 Zooplankton MMI Performance 121
7.6.1 Calibration versus Validation Sites 121
7.6.2 Precision of MMIs based on Least Disturbed Sites 121
7.6.3 Responsiveness, Redundancy, and Repeatability of Zooplankton MMIs 121
7.6.4 Responsiveness to a Generalized Stressor Gradient 121
7.6.5 Effect of Natural Drivers and Tow Length on MMI Scores 125
7.7 Thresholds for Assigning Ecological Condition 129
7.8 Discussion 132
7.9 Literature cited 133
7.10 List of Candidate Metrics for Zooplankton 139
Chapter 8: From Analysis to Results 159
8.1 Background information 159
8.2 Population Estimates 159
8.3 Lake Extent Estimates 159
8.4 Stressor Extent, Relative Risk and Attributable Risk 160
8.4.1 Stressor extent 160
8.4.2 Relative risk and attributable risk 161
8.4.3 Considerations When Calculating and Interpreting Relative Risk and Attributable Risk 163
8.5 NLA 2007 versus NLA 2012 Change Analysis 164
8.5.1 Background information 164
5 NLA 2012 Technical Report. October Version 1.1
-------
8.5.2 Data preparation 164
8.5.3 Methods 164
8.6 Literature cited 165
Chapter 9: Quality Assurance Summary 166
6 NLA 2012 Technical Report. October Version 1.1
-------
List of Figures
Figure 2-1. Proportion of Target Population Assessed Versus Not Assessed 22
Figure 3-1. Nine aggregate ecoregions used for reference site classification 25
Figure 4-1. Box and whisker plots showing discrimination between reference (R) and trash (T) sites by
biological ecoregion 38
Figure 4-2. MMI score versus PCA factor 1 disturbance score for NLA macroinvertebrate reference sites.
Higher PCA factor 1 scores indicate more disturbance 40
Figure 5-1. Field sampling design with 10 near-shore stations at which data were collected to
characterize near shore lake riparian and littoral physical habitat in the 2007 and 2012 National Lakes
Assessment (NLA) surveys 82
Figure 5-2. Near-shore anthropogenic disturbance (RDis_IX) in NLA0712 regions, ordered by their
median Reference site RDis 83
Figure 5-3. Near-shore anthropogenic disturbance in NLA0712 least-disturbed reference sites (median
RDisJX), ordered by aggregated region according to the same median level of near-shore disturbance.
84
Figure 5-4. LogSD's for Null-Model and regression-based O/E model for Near-shore RVegQ, LitCvrQ, and
LitRipCvrQ in the set of least-disturbed lakes and reservoirs (Table 5-1) sampled in the combined 2007
and 2012 NLA surveys 85
Figure 5-5. Contrasts in key NLA physical habitat index values among least-disturbed reference (R),
intermediate (S), and highly disturbed (T) lakes in the contiguous 48 states of the U.S. based on
combined NLA 2007 and 2012 data 86
Figure 5-6. Contrasts in key NLA physical habitat index values among least-disturbed reference (R),
intermediat (S), and highly disturbed (T) lakes in the contiguous 48 states of the U.S. shown separately
for the NLA 2007 and 2012 surveys 87
Figure 6-1. Box and whisker plot of Total Phosphorus in GIS screened, outlier removed, reference sites
by ecoregion 99
Figure 6-2. Box and whisker plot of Total Nitrogen in GIS screened, outlier removed, reference sites by
ecoregion 100
Figure 7-1. Five aggregated bio-regions used to develop zooplankton MMIs for the 2012 National Lake
Assessment 109
Figure 7-2. Distribution of six component metrics of the zooplankton MMI for the Coastal Plain bio-
region in least disturbed versus most disturbed sites 114
Figure 7-3. Distribution of six component metrics of the zooplankton MMI for the Eastern Highlands bio-
region in least disturbed versus most disturbed sites 116
Figure 7-4. Distribution of six component metrics of the zooplankton MMI for the Plains bio-region in
least disturbed versus most disturbed sites 117
Figure 7-5. Distribution of six component metrics of the zooplankton MMI for the Upper Midwest bio-
region in least disturbed versus most disturbed sites 119
Figure 7-6. Distribution of six component metrics of the zooplankton MMI for the Western Mountains
bio-region in least disturbed versus most disturbed sites 120
Figure 7-7. Distribution of zooplankton MMI scores in-calibration vs. validation sites for five bio-regions.
123
Figure 7-8. Distribution of zooplankton MMI scores in-least- vs. most disturbed sites for five bio-regions.
124
7
NLA 2012 Technical Report. October Version 1.1
-------
Figure 7-9. Linear regression of NLA 2012 Zooplankton MMI scores vs. first axis score from principal
components analysis (PCA) based on chemical, habitat, and visual assessment stressor variables used to
screen least- and most disturbed sites 126
Figure 7-10. NLA 2012 Zooplankton MMI scores of man-made (shaded boxes) versus natural lakes
(unshaded boxes) for least disturbed sites in five bio-regions 127
Figure 7-11. Zooplankton MMI scores versus lake size class within least disturbed lakes of the 2012 NLA.
128
Figure 7-12. Zooplankton MMI scores versus site depth for least disturbed sites 130
8 NLA 2012 Technical Report. October Version 1.1
-------
List of Tables
Table 2-1. National Lakes 2012 Initial Design 16
Table 2-2. Number of Sites Sampled for NLA 2012 by Design Categories 18
Table 3-1. Least-disturbed reference screening filter thresholds for NLA2012 26
Table 3-2. Most disturbed site screening thresholds for NLA2012 27
Table 3-3. Dichotomous key for defining NLA lakes likely impacted by anthropogenic drawdown 28
Table 3-4. Number of unique reference sites used in analysis - revised ecoregion data 29
Table 4-1. Final NLA 2007-2012 biological ecoregion benthic MMI metrics and their floor/ceiling values
for MMI scoring 36
Table 4-2. Final NLA 2007-2012 biological ecoregion benthic MMI statistics 37
Table 4-3. NLA2012 macroinvertebrate MMI thresholds 40
Table 5-1. NLA reference sites from combined 2007 & 2012 surveys 73
Table 5-2. Assignment of riparian vegetation cover complexity, littoral cover complexity, and littoral-
riparian habitat complexity index variants by aggregated ecoregion 73
Table 5-3. Summary of regression models used in estimating lake-specific expected values of Lake
Physical Habitat variables RVegQx, LitCvrQx and LitRipCvrQx under least-disturbed conditions 74
Table 5-4. Null Model Geometric Means (gMean), geometric Standard Deviations (gSD), 5th percentiles,
and 25th percentiles of habitat index values in least-disturbed reference lakes in the aggregated
ecoregions of the NLA 75
Table 5-5. O/E Physical Habitat Model means (LogMean, gMean), standard deviations (LogSD, gSD), and
percentiles of the distribution of habitat index O/E values for least-disturbed reference lakes in the
aggregated ecoregions of the NLA 77
Table 5-6. Empirical 75th and 95th percentiles of the distribution of vertical and horizontal drawdown... 78
Table 5-7. Precision of the key NLA Physical Habitat indices used as the primary physical habitat
condition measures in the NLA 79
Table 5-8. Association of NLA-2012 Physical Habitat Indices with high and low anthropogenic
disturbance stress classes (RT_NLA12 = R and T), defined as least-disturbed and most disturbed within
NLA regions 80
Table 5-9. Association of NLA 2007 and 2012 Physical Habitat Indices with high and low anthropogenic
disturbance stress classes (RT_NLA12 = R and T), defined as least-disturbed and most disturbed within
NLA regions 81
Table 6-1. Trophic State Classification used in NLA 2012 98
Table 6-2. NLA2012 least, moderately, and most disturbed thresholds (75th/95th percentiles) for TP, TN,
CHLA, and turbidity condition classes 101
Table 7-1. Hypothesized-responses of zooplankton assemblages to disturbance 104
Table 7-2. Component metrics of the zooplankton MMI for the Coastal Plain bio-region 114
Table 7-3. Component metrics of the zooplankton MMI for the Eastern Highland bio-region 116
Table 7-4. Component metrics of the zooplankton MMI for the Plains bio-region 117
Table 7-5. Component metrics of the zooplankton MMI for the Upper Midwest bio-region 119
Table 7-6. Component metrics of the zooplankton MMI for the Western Mountains bio-region 120
Table 7-7. Results of independent assessment and precision tests of NLA 2012 zooplankton MMIs based
on least disturbed sites 123
Table 7-8. Results of responsiveness, redundancy, and repeatability tests for NLA 2012 zooplankton
MMIs 124
Table 7-9. Linear regression statistics of zooplankton MMI scores versus pea-based disturbance score
for each bio-region 131
9 NLA 2012 Technical Report. October Version 1.1
-------
Table 7-10. Thresholds for assigning ecological condition for zooplankton MMI scores based on the
distribution of least disturbed sites in five bio-regions 131
Table 7-11. List of candidate metrics used to develop the zooplankton MMI for the Coastal Plain bio-
region 140
Table 7-12. List of candidate metrics used to develop the zooplankton MMI for the Eastern Highlands
bio-region 143
Table 7-13. List of candidate metrics used to develop the zooplankton MMI for the Plains bio-region 147
Table 7-14. List of candidate metrics used to develop the zooplankton MMI for the Upper Midwest bio-
region 151
Table 7-15. List of candidate metrics used to develop the zooplankton MMI for the Western Mountains
bio-region 155
Table 8-1. Extent estimates for response and stressor categories 161
10 NLA 2012 Technical Report. October Version 1.1
-------
Chapter 1: Project Overview
1.1 Overview
This document, the National Lakes Assessment 2012: Technical Report, accompanies the
National Lakes Assessment 2012: A Collaborative Survey of Lakes in the United States and
related on-line materials. The National Lakes Assessment (NLA) is a collaboration among the
U.S. Environmental Protection Agency (EPA), states, tribes, and other partners. It is part of the
National Aquatic Resource Surveys (NARS) program design to conduct national scale
assessments of aquatic resources. The NLA 2012 provides the second assessment at national
and regional scales of the ecological and recreational condition of lakes. This assessment was
accomplished by collecting and analyzing data from across the conterminous United States.
The National Lakes Assessment 2012: A Collaborative Survey of Lakes in the United States (the
Public Report) is not a technical document, but rather a report geared toward a broad, public
audience. The NLA 2012 presents information from the second National Lakes Assessment. It
provides national-scale assessments and also compares the condition of lakes to those from the
earlier NLA 2007 conducted by EPA and its partners. You can find results for regional scales and
comparisons between natural lakes and reservoirs using our interactive dashboard at
https://nationallakesassessment.epa.gov/. The technical report is a supplemental document
that serves as a technical reference to support findings presented in the public report and on-
line.
1.2 Objectives of the National Lakes Assessment
The objective of the NLA is to characterize aspects of the biological, chemical, physical, and
recreational condition of the nation's lakes throughout the conterminous United States. It
employs a statistically-valid probability design stratified to allow estimates of the condition of
lakes on a national and regional scale.
The NLA is designed to answer the following questions about lakes across the United States.
1. What is the current biological, chemical, physical, and recreational condition of lakes?
a. What is the extent of degradation among lakes?
b. Is degradation widespread (e.g., national) or localized (e.g., regional)?
2. Is the proportion of lakes in the most disturbed condition getting better, worse, or staying
the same over time?
3. Which environmental stressors are most strongly associated with degraded biological
condition in lakes?
A variety of chemical, physical, and biological data were collected and developed into indicators
to address the NLA questions. For each of these indicators, this Technical Report focuses on the
conceptual basis, methods, and procedures used for the NLA. The information described in this
11 NLA 2012 Technical Report. October Version 1.1
-------
Technical Report was developed through the efforts and cooperation of NLA scientists from
EPA, technical experts, and participating cooperators from states, tribes, and academia. While
this Technical Report serves as a comprehensive summary of the NLA procedures, it is not
intended to present an in-depth report of the design, site evaluation process, field sampling,
NLA results, or additional data analysis results. Please see the following documents for
additional details on these aspects of the project.
2012 National Lakes Assessment: Quality Assurance Project Plan (EPA 841-B-11-006)
2012 National Lakes Assessment: Site Evaluation Guidelines (EPA 841-B-11-005)
2012 National Lakes Assessment: Field Operations Manual (EPA 841-B-11-003)
2012 National Lakes Assessment: Laboratory Operations Manual (EPA 841-B-11-004)
12
NLA 2012 Technical Report. October Version 1.1
-------
Chapter 2: Survey Design and Population Estimates
The NLA was designed to assess the condition of the population of lakes, reservoirs, and ponds
in the conterminous United States. The NLA design allows characterization of lakes at national
and regional scales using chemical, physical and biological indicators. It is not intended to
represent the condition of individual lakes. The statistical design also accounts for the
distribution of lakes across the country - some areas have fewer lakes than others - so that
even in areas of the country where there are few sample sites regional and national results still
apply to the broader target population.
2.1 Descriptii esign
The target population for the NLA includes all lakes, reservoirs, and ponds within the 48
contiguous United States greater than 1 hectare (ha) in surface area that are permanent
waterbodies. The word "lake" in the remainder of this document includes lakes, reservoirs and
ponds. Lakes that are saline are excluded as are those used for aquaculture, disposal-tailings,
sewage treatment, evaporation, or other unspecified disposal use.
To select sites for the NLA, EPA statisticians used a Generalized Random Tessellation Stratified
(GRTS) (Stevens and Olsen, 1999; Stevens and Olsen 2004) survey design for a finite resource
with stratification and unequal probability of selection. The design includes reverse hierarchical
ordering of the selected lakes.
2.1.1 Stratification
The overall NLA survey design was stratified by state and by class (NLA12_CLS). NLA12_CLS has
three classes:
NLA07RVT - defined as all NLA 2007 lakes that were target and sampled,
NLA12NEW - remaining lakes in NHD-Plus that are included in the sample frame, and
Exclude - lakes in NHD-Plus that are excluded from the sample frame (see Sample
Frame section below).
The design also included additional sites that states could use to conduct state-scale surveys.
This was accomplished by adding additional sites to the primary draw such that each state had
50 sites. Each state design has two strata, ST_ NLA07RVT and ST_ NLA12NEW (where ST is
replaced by two letter state abbreviation. The total number of strata is 96 (two for each state).
2.1.2 Unequal Probability Categories
The 48 state strata for lakes from the NLA 2007 visited again in 2012 was an equal probability
design within each stratum. The 48 state strata NLA12NEW was an unequal probability design
13
NLA 2012 Technical Report. October Version 1.1
-------
within each state stratum. The unequal probability categories were defined based on lake area:
1 to 4 ha, 4 to 10 ha, 10 to 20 ha, 20 to 50 ha and greater than 50 ha.
2.1.3 Panels
The survey design has four panels: NLA07RVT - identifies lakes from NLA 2007 that will be
visited in 2012, NLA12NAT - identifies new lakes that will be sampled along the lakes in panel
NLA07RVT as part of the NLA2012 national survey design, NLA12ST - identifies additional lakes
that a state may sample to achieve a total sample size of 50 lakes for the state, and OverSamp -
identifies lakes to be used to replace lakes that cannot be sampled for some reason (not a lake,
denied access, physically inaccessible, etc).
The national survey design includes all lakes within a state that are in either panels NLA07RVT
or NLA12NEW.
A state survey design includes all lakes within a state that are either in panels NLA07RVT,
NLA12NEW or NLA12ST.
2.1.4 Expected Sample Size
The expected sample size depends on the strata, panels and lake area category. For the
NLA07RVT strata, the objective was to resample 400 of the NLA 2007 lakes out of the 1028
lakes that were sampled in 2007, i.e., approximately 38% of the lakes. The sample size for each
state in the strata was proportional to the number of lakes sampled in the state in 2007.
Exceptions were made when a state implemented a state-level design in 2007. A total sample
size of 1000 lakes (including revisit sites) was desired for the national design. The sample size
for each state was proportional (approximately 60%) to the state's sample size in NLA 2007. The
minimum number of lakes for a state was set at 8 and the maximum at 43. Although
aggregated ecoregions were not explicitly used in the survey design or setting sample sizes,
they are implicitly used since the NLA 2007 allocated sample sizes using aggregated ecoregions.
Once these two sample sizes were set for a state, an additional sample size was allocated to a
state so that the total number of sites in a state would be 50 lakes. See Table 2-1 for the
expected sample size by state.
Lakes in the NLA 2007 Revisit stratum were selected with equal probability and did not depend
on lake area (NLA 2007 did depend on lake area). New lakes in the design were selected with
unequal probability based on five lake area categories. The total number of lakes for a state in
this strata was divided by five and that sample size (approximately) was assigned to the
"(10,20]" lake area category. Sample sizes for lake area categories "(20,50]" and ">50" were
decreased successively by one and for lake area categories "(4,10]" and "(1,4]" were increased
successively by one. This process was adjusted to meet the total sample size requirement for
the stratum. The rationale for this assignment of sample sizes is based on experience that
smaller lakes are more likely not to be lakes or be inaccessible than larger lakes. When lakes are
14
NLA 2012 Technical Report. October Version 1.1
-------
replaced, the process is expected to more likely result in an equal number of lakes sampled by
lake area category.
15
NLA 2012 Technical Report. October Version 1.1
-------
Table 2-1. National Lakes 2012 Initial Design.
National Lakes 2012 Design
Number of NLA2007
Lakes Revisited in
NLA2012
Number of New Lakes
for NLA2012
Total
Number
of lakes
Lakes
sampled
Total
Number
of Lake
Additional
Lakes
Number
of lakes
Over
Total
Sampled
Sampled
Sampled
Sampled
to be
twice in
Visits
State
State
Sample
Lakes
State
Once
Twice
Once
Twice
Sampled
2012
2012
Design
Design
Lakes
Selected
AL
3
1
3
1
8
2
10
42
50
92
142
AR
3
1
3
1
8
2
10
42
50
92
142
AZ
6
1
5
1
13
2
15
37
50
86
136
CA
7
1
15
1
24
2
26
26
50
84
134
CO
10
1
11
1
23
2
25
27
50
78
128
CT
4
1
4
1
10
2
12
40
50
90
140
DE
3
1
2
1
7
2
9
43
50
46
96
FL
8
1
6
1
16
2
18
34
50
82
132
GA
4
1
5
1
11
2
13
39
50
90
140
IA
6
1
7
1
15
2
17
35
50
86
136
ID
10
1
12
1
24
2
26
26
50
78
128
IL
5
1
6
1
13
2
15
37
50
88
138
IN
16
1
9
1
27
2
29
23
50
66
116
KS
7
1
6
1
15
2
17
35
50
84
134
KY
2
1
5
1
9
2
11
41
50
94
144
LA
5
1
7
1
14
2
16
36
50
88
138
MA
3
1
5
1
10
2
12
40
50
92
142
MD
3
1
3
1
8
2
10
42
50
46
96
ME
9
1
13
1
24
2
26
26
50
80
130
Ml
17
1
19
1
38
2
40
12
50
64
114
MN
21
1
19
1
42
2
44
108
150
256
406
MO
6
1
9
1
17
2
19
33
50
86
136
MS
6
1
6
1
14
2
16
36
50
86
136
16 NLA 2012 Technical Report. October Version 1.1
-------
MT
13
1
16
1
31
2
33
19
50
72
122
NC
4
1
7
1
13
2
15
37
50
90
140
ND
13
1
27
1
42
2
44
8
50
72
122
NE
13
1
13
1
28
2
30
22
50
72
122
NH
4
1
5
1
11
2
13
39
50
90
140
NJ
3
1
6
1
11
2
13
39
50
92
142
NM
4
1
7
1
13
2
15
37
50
90
140
NV
5
1
8
1
15
2
17
35
50
88
138
NY
3
1
5
1
10
2
12
40
50
92
142
OH
6
1
8
1
16
2
18
34
50
86
136
OK
17
1
11
1
30
2
32
20
50
64
114
OR
12
1
15
1
29
2
31
21
50
74
124
PA
6
1
8
1
16
2
18
34
50
86
136
Rl
3
1
3
1
8
2
10
42
50
92
142
SC
2
1
5
1
9
2
11
41
50
94
144
SD
13
1
28
1
43
2
45
7
50
72
122
TN
3
1
4
1
9
2
11
41
50
92
142
TX
15
1
24
1
41
2
43
9
50
68
118
UT
8
1
12
1
22
2
24
28
50
82
132
VA
7
1
12
1
21
2
23
29
50
84
134
VT
3
1
5
1
10
2
12
40
50
92
142
WA
11
1
18
1
31
2
33
19
50
76
126
Wl
10
1
16
1
28
2
30
22
50
78
128
WV
2
1
4
1
8
2
10
42
50
93
143
WY
6
1
11
1
19
2
21
31
50
86
136
Sum
350
48
458
48
904
96
1000
1596
2500
4111
6611
17
NLA 2012 Technical Report. October Version 1.1
-------
Table 2^2, Number of Sites Sampled for NLA 2012 by Design Categories,
Number of Sites Sampled for NLA 2012
State NLA07RVT NLA12NEW NLA12NEW 07RVT
Sampled
Once
Sampled
Twice
Sampled
Once
Sampled
Twice
Sampled
Once
Sampled
Twice
Total
Sites
Total Site
Visits
AL
3
1
3
1
8
10
AR
3
1
3
1
8
10
AZ
4
1
7
1
13
15
CA
7
1
28
1
1
38
40
CO
10
1
11
1
23
25
CT
5
1
4
1
11
13
DE
3
1
2
1
7
9
FL
7
5
16
20
GA
4
1
5
1
11
13
IA
6
1
7
1
15
17
ID
9
1
29
1
40
42
IL
3
1
8
1
13
15
IN
13
1
35
1
50
52
KS
6
1
8
1
16
18
KY
2
1
6
1
10
12
LA
5
1
7
1
14
16
MA
3
1
5
1
10
12
MD
3
1
3
1
8
10
ME
9
1
13
1
24
26
Ml
17
1
34
1
53
55
MN
20
1
28
1
50
52
MO
6
1
9
1
17
19
MS
6
1
6
1
14
16
MT
11
1
19
1
1
33
35
18
NLA 2012 Technical Report. October Version 1.1
-------
NC
3
1
8
1
13
15
ND
12
1
30
1
44
46
NE
13
1
13
1
28
30
NH
4
1
5
1
11
13
NJ
3
1
6
1
11
13
NM
1
1
10
1
13
15
NV
5
1
7
1
1
15
17
NY
1
1
6
1
9
11
OH
6
8
1
1
16
18
OK
16
1
12
1
30
32
OR
11
1
15
1
1
29
31
PA
5
1
9
1
16
18
Rl
2
1
3
1
1
8
10
SC
1
1
5
9
12
SD
11
1
31
1
44
46
TN
2
1
4
9
12
TX
11
1
34
1
47
49
UT
6
1
38
1
46
48
VA
6
1
12
1
1
21
23
VT
3
1
5
1
10
12
WA
10
1
19
1
31
33
Wl
9
1
39
1
50
52
WV
1
1
5
1
8
10
WY
4
1
12
1
18
20
Total
311
48
621
50
6
2
1038
1138
19
NLA 2012 Technical Report. October Version 1.1
-------
2.2 Sample frame summary
The sample frame was derived from the National Hydrography Dataset (NHD). Once the initial
shapefile that included all lake objects in NHD was prepared additional attributes were created
to identify lakes included in the sample frame and other properties used to construct the survey
design.
Lakes included in the sample frame were those lakes with DES_FYTPE values equal to:
Lake/Pond
Lake/Pond: Hydrographic Category = Perennial
Lake/Pond: Hydrographic Category = Perennial; Stage = Average WaterElevation
Lake/Pond: Hydrographic Category = Perennial; Stage = Normal Pool
Reservoir
Reservoir: Reservoir Type = Water Storage
Reservoir: Reservoir Type = Water Storage; Hydrographic Category = Perennial
Lakes excluded in the sample frame were those lakes with DES_FYTPE values equal to:
Lake/Pond: Hydrographic Category = Intermittent
Lake/Pond: Hydrographic Category = Intermittent; Stage = Date of Photography
Lake/Pond: Hydrographic Category = Intermittent; Stage = High Water Elevation
Playa
Reservoir: Reservoir Type = Aquaculture
Reservoir: Reservoir Type = Cooling Pond
Reservoir: Reservoir Type = Disposal
Reservoir: Reservoir Type = Evaporator
Reservoir: Reservoir Type = Tailings Pond
Reservoir; Reservoir Type = Treatment
Swamp/Marsh
Next, lakes were excluded that were evaluated during the NLA 2007 and were identified as
lakes that did not meet definition of a lake for NLA 2012. These were lakes with evaluation
codes of Lake_Saline, Lake_Shallow, Lake_Special_Purpose, Lake_Vegetated, Non_Target, or
Not_Lake".
Finally, lakes that were less than or equal to 1 hectare were excluded.
2.3 Survey analysis
Any statistical analysis of data must incorporate information about the monitoring survey
design. In particular, when estimates of characteristics from a statistical survey such as the NLA
are made for the entire target population are computed, called population estimates, the
statistical analysis must account for any stratification or unequal probability selection in the
design. The statistical estimates for the NLA population estimates were completed using site
weights (see the NLA 2012 Site Information - Data file at https://www.epa.gov/national-
20
NLA 2012 Technical Report. October Version 1.1
-------
aquatic-resource-surveys/data-national-aquatic-resource-surveys.) and the R package
'spsurvey' (Kincaid and Olsen 2013) which implements the methods described by Diaz-Ramos et
al. (1996).
2.4 Estimated extent of the NLA lake population and implications for reporting
Crews evaluated sites from the NLA survey design using a variety of techniques including aerial
photo interpretations, GIS analyses, local knowledge, etc. to identify locations that did not meet
the definition of a lake for NLA. Crews also dropped sites from sampling during field
reconnaissance if they were a non-target type or could not be assessed due to accessibility
issues (land owner denial, too dangerous to access, etc.). Dropped sites were systematically
replaced from a pool of replacement sites from the random design. This process is
implemented to maintain the integrity of the random design and to sample sites consistent
with the original number planned in different categories.
The treatment of sites eliminated from sampling affects how the final population results are
estimated and reported including the total proportion of the target population that we can
assess. Taking into account the sites identified as not being part of the target population (e.g.,
saline lakes, lakes less than 1 hectare in size, etc.), the NLA analysis estimated there were
159,652 lakes in the NLA target population across the conterminous U.S. The area represented
by sites that were part of the target population, but not sampled because of accessibility issues,
is excluded from the assessments because sites which had access issues cannot be assumed to
be randomly distributed. For example, there may be a bias in land-ownership for sites where
access was denied, or sites which were inaccessible may often occur in areas with limited
disturbance. As a result, the final number of lakes represented by the probability sites sampled
and reported by the NLA, i.e., the inference (or sampled) population, was 111,818 lakes or
approximately 70% of the target population. Throughout this report, lake estimates as
percentages are relative to the 111,818 lakes. Figure 2-1 shows the percent of the target
population of lakes that was sampled and the proportions that fell into non-sampleable
categories. The inference population is represented by 1038 probability sites. The not assessed
component of the population is represented by sites 1) where access was denied, 2) that were
inaccessible due to safety considerations or remote location and 3) with other reasons for
dropping.
21
NLA 2012 Technical Report. October Version 1.1
-------
Step 1: EPA found 389,005 lakes in the
National Hydrography Database (NHD)
and identified those lakes that met
eligibility criteria for inclusion in the
sample.
Step 2: EPA excluded lakes that were
not accessible to sampling teams.
Step 3: Field crews collected data from
a random sample of the remaining
111,818 lakes (inference population).
1.038 lakes
were randomly
sampled
I
EPA used the following criteria to
determine eligibility:
Surface area > 1 hectare
Depth > 1 meter
Open water >0.1 hectare
A lake was considered inaccessible for
safety reasons or if the crews were
denied permission by the landowner
Percentages and confidence intervals
reported for a given indicator are
relative to the lakes in the inference
population.
Example: If EPA estimates that 10% of
lakes nationally are most disturbed for
an indicator, this means that 11,181 are
estimated to be in this condition.
Figure 2-1. Proportion of Target Population Assessed Versus Not Assessed.
2.5 Literature cited
Diaz-Ramos, S., D. L. Stevens Jr, and A. R. Olsen. 1996. EMAP Statistical Methods Manual. US
Environmental Protection Agency, Office of Research and Development, NHEERL-
Western Ecology Division, Corvallis, Oregon.
Kincaid, T. M. and A. R. Olsen. 2013. spsurvey: Spatial Survey Design and Analysis. R package
version 2.6.
Omernik, J. M. 1987. Ecoregions of the Conterminous United States. Annals of the Association
of American Geographers 77:118-125.
R Core Team. 2013. R: A language and environment for statistical computing. R Foundation for
Statistical Computing, Vienna, Austria. http://www.R-project.org.
Stevens, D. L., Jr., and A. R. Olsen. 1999. Spatially restricted surveys over time for aquatic
resources. Journal of Agricultural, Biological, and Environmental Statistics 4:415-428.
Stevens, D. L., Jr., and A. R. Olsen. 2003. Variance estimation for spatially balanced samples of
environmental resources. Environmetrics 14:593-610.
22
NLA 2012 Technical Report. October Version 1.1
-------
Stevens, D. L., Jr., and A. R. Olsen. 2004. Spatially-balanced sampling of natural resources.
Journal of American Statistical Association 99:262-278.
USEPA. 2011. Level III Ecoregions of the Continental United States (revision of Omernik, 1987).
US Environmental Protection Agency, National Health and Environmental Effects
Laboratory Western Ecology Division. Corvallis, Oregon.
23 NLA 2012 Technical Report. October Version 1.1
-------
Chapter 3: Reference Condition and Condition Benchmarks
3.1 Background information
NLA analysts used two processes for establishing the least disturbed, moderately disturbed, and
most disturbed findings in the NLA report. For trophic status and recreational indicators,
analysts used fixed, nationally consistent benchmarks. This approach is not covered in detail in
this Technical Addendum although the specific benchmarks are identified in the appropriate
sections. The second approach was to establish regionally consistent reference-based
benchmarks. Detailed information on the regionally consistent approach is presented below. In
refining benchmarks for the NLA 2012, some 2007 benchmark values were revised; therefore,
direct comparisons should not be made between 2012 results and those reported in 2007. For
purposes of identifying change in this report, 2007 results were recalculated based on new
2012 benchmarks.
To assess current ecological condition, it is necessary to compare measurements today to an
estimate of "good" quality. Because of the difficulty of finding minimally disturbed sites in many
parts of the country, NLA 2012 used "least disturbed condition" as the definition of reference
condition. The use of least disturbed condition in the context of defining reference condition is
different than the assessment category of least disturbed used in the NLA report. Least
disturbed condition can be defined as the best available chemical, physical, and biological
habitat conditions given the current state of the landscape - or "the best of what's left"
(Stoddard et al. 2006). Data from reference sites were used to develop ecoregion specific
reference conditions against which test results could be compared. A total of four sets of
reference sites were developed for use in establishing reference condition for the NLA report:
one for the benthic macroinvertebrates indicator, one for the zooplankton indicator, one for
the nutrient indicators, and one for the physical habitat indicators. This section describes the
selection of the biological reference sites which also form the basis for all the nutrient and
habitat reference sites.
3.2 Pre-sampling screening (hand-picked sites only)
In addition to the probability set of lakes, a smaller set of sites were hand selected a priori for
sampling. We were trying to ensure that we captured samples from additional least disturbed
lakes. Potential hand-picked sites were identified as high quality sites by EPA, states, tribes, and
federal partners. When data were available, these potential sites were compared to water
quality screens. When data were not available, sites underwent a high-level visual screen. The
screen was used to minimize human disturbance around potential lakes (Herlihy et al., 2013).
We identified 91 hand-picked lakes for sampling following this coarse screening process. The
hand-picked sites were sampled during the 2012 index period using NLA sampling protocols,
samples were processed and analyzed with the same analytical methods as the probability site
samples, and then both the hand-picked sites and the probability sites were subjected to the
post-sample screening process (Section 3.3). Regardless of whether sites were probability-
24 NLA 2012 Technical Report. October Version 1.1
-------
based or hand-selected, only those that met the final screening criteria for the appropriate
indicator (i.e. benthic macro invertebrates, zooplankton, nutrients, and physical habitat) were
used in developing reference conditions. In an update to 2007, ecoregion designations for each
site were assigned based on the revised ecoregion GIS layer (2015) that accounted for updated
Omernik ecoregion boundaries (Figure 3-1).
Eco regions used in National Aquatic Resource Surveys
5c«t»iv fryhrMwi |-*F* Arc
W- Swtwr IVn
Figure 3-1. Nine aggregate ecoregions used for reference site classification.
3.3 Post-sampling screening for biological reference condition
To maximize the number of reference sites available for data analysis, hand-selected and
probability-based sampled in either NLA 2007 or NLA 2012 were considered potential reference
lakes. For benthic macroinvertebrates, only sites with at least 250 individuals in the sample
were used to establish reference; this criterion did not apply to other sets of reference sites.
Analysts used the chemical and physical data collected at each site to determine whether any
given site was in least-disturbed condition for its aggregate ecoregion following the approach
described by Herlihy et all, (2008). The nine aggregate ecoregions defined in NLA 2007 were
used for the ecoregion classification although in some cases these ecoregions were further
combined or lake types (natural vs. manmade) within an ecoregion treated differently (Figure
25 NLA 2012 Technical Report. October Version 1.1
-------
3-1). In the NLA, screening values were established for twelve chemical and physical
parameters to screen for biological reference sites (Table 3-1). If measurements at a site
exceeded the screening value for any one stressor, it was dropped from reference
consideration. Given that expectations of least disturbed condition vary across regions, the
criteria values for exclusion varied by ecoregion as well. Additional screening for physical
habitat reference are described in Chapter 5.
Details on the calculation and naming of the shoreline habitat disturbance metrics is given in
the physical habitat chapter (Section 5.3). Scoring of the disturbances on the visual assessment
form for agricultural, residential, and industrial disturbance were simply done by summing the
number of checked off disturbances on the form weighting for the noted level of disturbance.
Low disturbance was weighted as 1 point, medium disturbances were weighted as 3 points, and
high disturbances were weighted as 5 points. Fire was not summed in with the industrial
disturbances as it could be an entirely natural disturbance.
All selected lake reference sites were also screened for excessive lake drawdown that was likely
anthropogenic. Evidence of both horizontal and vertical lake level fluctuations were recorded
by field crews. The square root of lake surface area was used as a surrogate for lake diameter
and was used to scale horizontal exposure of littoral lake bottom. Similarly, lake maximum
depth was used to scale vertical lake fluctuations. In addition, the drawdown criteria was
relaxed for lakes with elevated levels of lakeshore disturbance, as indexed by HiiALL_syn > 0.75.
A step by step key to defining NLA lakes impacted by drawdown is provided in Table 3-1. In NLA
2012, 13 otherwise reference lakes were removed due to excessive drawdown of likely
anthropogenic origin.
Table 3-1. Least-disturbed reference screening filter thresholds for NLA2012,
If a lake exceeded any one of the thresholds it was not considered as a least-disturbed reference site for that
ecoregion. Three filters were applied universally across all ecoregions, 1) ANC < 25 ueq/L and DOC < 5 mg/L, 2)
HifPany_Circa_syn& > 0,9, and 3) no excessive lake drawdown (see Table 3-3),
Aggregate
Ecoregion
TP
(ug/L)
TN
(ug/L)
CI
(ueq/L)
S04
(ueq/L)
Turbidity
(NTU)
Hii-
NonAg&
Hii-
Ag&
Assessment5
(Ag/Res/Ind)
WMT
>30@
>400
>100#
>200
>3
>0.6
>0
> 5/5/5
XER
>100
>1000
>500
>1000
>5
>1.5
>0.2
> 5/5/5
NPL
>150
>2000
>1000
>5
>1.5
>0.5
> 10/6/6
SPL
>150*
>2000*
>1000
>5
>1.5
>0.5
> 10/6/6
TPL
>120
>2000
>1000
>5000
>5.5
>1.7
>0.15
> 9/9/9
UMW
>40
>1200
>200
>200
>5
>0.6
>0
> 5/5/5
CPL
>50
>1200
>1000
>400
>5
>1.0
>0
> 6/10/6
SAP
>35
>800
>125
>300
>5
>0.9
>0
> 6/6/6
NAP
>30
>600
>100#
>300
>5
>0.6
>0
> 6/6/6
metric not used for screening
26
NLA 2012 Technical Report. October Version 1.1
-------
& HiiNonAg_syn, HiiAg_syn, and HifPany_Circa_syn are lakeshore physical habitat disturbance indices
(see Section 5.3.4.6).
$ Assessment filters are based on indices of agricultural, residential, and industrial disturbance
calculated from observations on the visual assessment form.
* No nutrient (TP, TN) or Turbidity filters applied in Sand Hills in SPL (Omernik Level III Ecoregion 44)
# No Chloride filter applied in Coastal Ecoregions in NAP (ecoregions 59,82), XER (ecoregion 6), and
WMT (ecoregions 1,2,8)
@ No TP filter used in volcanic ecoregions in WMT (ecoregions 4,5,9,77)
In addition to selecting least disturbed reference sites, analysts also determined most disturbed
sites for each ecoregion. These sites were used primarily in developing biotic MMIs that would
be used in the biological assessment of the nation's lakes and in testing the strength of
association of other indicators to anthropogenic stress. Similar to the reference lake selection
process, thresholds were used to determine which lakes were to be considered most disturbed
in each ecoregion (Table 3-2). If any site exceeded the most-disturbed threshold for any one of
these screening criteria, then the site was classified as most-disturbed.
Note that the NLA did not use data on land-use in the watersheds for the final reference site
screeningsites in agricultural areas (for example) may well be considered least disturbed,
provided that their chemical and physical conditions are among the least-disturbed for the
region. Additionally, the NLA did not use data from the biological assemblages themselves to
define biological reference sites because the reference sites are being used to assess biological
condition and to use biological data to then define reference would constitute circular
reasoning.
Table 3-2, Most disturbed site screening thresholds for NLA2012.
If a lake exceeded any one of the thresholds it was considered a most-disturbed site for that ecoregion. One
screen was applied universally across all ecoregions, ANC < 0 ueq/L and DOC <; 5 mg/L.
Aggregate
Ecoregion
TP
(ug/L)
TN
(ug/L)
CI
(ueq/L)
S04
(ueq/L)
Turbidity
(NTU)
Hii-
NonAg&
Hii-
Ag&
Assessment5
(Ag/Res/Ind)
WMT
>150@
>1500
>1500#
>1500
>10
>2.5
>0.9
> 15/15/15
XER
>400
>4000
>25
>3.5
>1.0
> 15/15/15
NPL
>400
>4000
>50
>3.5
>1.2
> 15/15/15
SPL
>400*
>4000*
>50
>3.5
>1.2
> 15/15/15
TPL
>500
>5000
>5000
>20,000
>50
>4.0
>1.2
> 15/18/15
UMW
>200
>2500
>2500
>2500
>20
>3.5
>0.9
> 15/15/15
CPL
>200
>3000
>5000
>2500
>30
>3.5
>1.0
> 15/15/15
SAP
>150
>2500
>1500
>1500
>20
>3.5
>0.9
> 15/15/15
NAP
>150
>2500
>1500#
>1500
>20
>3.5
>0.9
> 15/15/15
metric not used for screening
& HiiNonAg_syn and HiiAg_syn are lakeshore physical habitat disturbance indices (see Section 5.3.4.6)
$ Assessment filters are based on indices of agricultural, residential, and industrial disturbance
calculated from observations on the visual assessment form.
* No nutrient (TP, TN) or Turbidity filters applied in Sand Hills in SPL (Omernik Level III Ecoregion 44)
27
NLA 2012 Technical Report. October Version 1.1
-------
# No Chloride filter applied in Coastal Ecoregions in NAP (ecoregions 59,82), XER (ecoregion 6), and
WMT (ecoregions 1,2,8)
@ No TP filter used in volcanic ecoregions in WMT (ecoregions 4,5,9,77)
Table 3-3, Dichotomous key for defining NLA lakes likely impacted by anthropogenic drawdown.
Based on field observations of horizontal lake level fluctuations (AH), vertical lake level
fluctuations (AV), and human lakeshore disturbance (physical habitat summary metric
HiiAII_syn).
1. AH < 10 m AND AV < 2 m
Yes - LAKE OK
No - go to 2
2. AH > 10 m and AV > 2 m
Yes - Lake Drawdown, Not Reference
No - go to 3
3. AV > 2 m and AV/Maximum Lake Depth > 10%
Yes - Lake Drawdown, Not Reference
No - go to 4
4. AH < 10 m
Yes - LAKE OK
No - go to 5
5. AH/sqrt(Lakearea) > 5m2
Yes - Lake Drawdown, Not Reference
No - go to 6
6. Lake Disturbed, HiiAII_syn > 0.75
Yes - Lake Drawdown, Not Reference
No - LAKE OK
3.4 Post-sample screening for nutrient reference condition
Setting reference condition for nutrients requires a different process then the one used for
biological reference condition evaluation. Because nutrients (TN, TP) were used to select
biological reference sites, the biological reference sites could not be used as nutrient reference
lakes due to circularity. During the development of nutrient reference sites, we compiled all
sampled sites in NLA 2007 and 2012 as was done for the biological reference condition process
described above. As was the case above, ecoregion designations for each site were assigned
based on the 2015 revised ecoregion GIS layer that accounted for updated Omernik ecoregion
boundaries. All sites were then passed through the NLA 2012 biological reference screening
process for their ecoregion as described with one exception. To avoid complete circularity, TP
and TN thresholds were removed as screening variables in the reference screening process. All
told there were 418 initial reference sites in the combined data, 149 sampled in 2007 and 269
sampled in 2012. For cross-year repeat sites sampled in both years, only the 2012 data was
used. Another modification was made for lakes in the Southern Plains. The nutrient conditions
28
NLA 2012 Technical Report. October Version 1.1
-------
in the natural SPL lakes are so different than the man-made SPL lakes that they need to have
different thresholds. We created SPLman and SPLnat surrogate ecoregions for this analysis.
Screening Reference Sites for Nutrient Thresholds
GIS Screening: There was a fairly strong disturbance signal in the reference sites as evidenced
by looking at relationships with four GIS stressor variables (% Agriculture, %Urban, Road and
Population density). Unfortunately, there was no road and population density available for the
NLA 2007 data so GIS screening was only done using the %Ag and %Urban metrics. In order to
remove this disturbance signal, a GIS stressor filtering approach was used to remove from the
reference site pool those sites that failed the filtering. For %Ag, ecoregional criteria were used:
NAP, WMT, XER (>10%); NPL, SAP, SPL, UMW (>25%); CPL (>40%); TPL (>50%). For%Urban, a
>10% criteria was used for all ecoregions but the CPL where a >15% filtering criteria was used.
Out of the 418 initial nutrient reference sites, 375 passed the GIS stressor screening filter (Table
3-4). Dropped sites due to the GIS screen were most prevalent in the Plains. The TPL lost 11 of
its 26 sites even with a 50% Ag screen. The man-made SPL lost 6 of 22 lakes.
Outlier Screening: As in the original Wadeable Streams Assessment and NLA 2007 threshold
setting, we used a 1.5*IQR outlier screening test to drop outliers from the analysis (sites with
values outside the range of Q1-1.5*IQR or Q3+1.5*IQR were dropped). Outlier screening
removed 18 of the 375 GIS screened reference lakes for TP analysis and 13 of 375 lakes for TN
analysis. For the GIS screened, outlier removed dataset, all ecoregions but the TPL had >10
sites, but only the CPL, NAP, SAP, UMW, and WMT had > 25 sites.
Table 3-4, Number of unique reference sites used in analysis - revised ecoregion data.
Eco
All Nutrient Ref
GIS Screened
GIS Screen with
(Initial screen)
Reference Sites
outliers removed
(TP/TN)
CPL
39
28
27/26
NAP
75
71
68/69
NPL
14
12
12/12
SAP
33
31
30/30
SPL-man
22
16
15/16
SPL-nat
19
19
17/19
TPL
26
15
14/15
UMW
59
56
55/54
WMT
103
103
95/98
XER
28
24
24/23
TOTAL
418
375
357/362
3.5 Lite ted
Herlihy, A. T., S. G. Paulsen, J. Van Sickle, J. L. Stoddard, C. P. Hawkins, and L. L. Yuan. 2008.
Striving for consistency in a national assessment: the challenges of applying a reference
29 NLA 2012 Technical Report. October Version 1.1
-------
condition approach at a continental scale. Journal of the North American Benthological
Society 27:860-877.
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.
Stoddard, J. L., D. P. Larsen, C. P. Hawkins, R. K. Johnson, and R. H. Norris. 2006. Setting
expectations for the ecological condition of running waters: the concept of reference
condition. Ecological Applications 16:1267-1276.
30
NLA 2012 Technical Report. October Version 1.1
-------
Chapter 4: Benthic Invertebrates
4.1 Background information
The taxonomic composition and relative abundance of different taxa that make up the littoral
macroinvertebrate assemblage present in a lake can be used to assess how human activities
affect ecological condition. Two principal types of ecological assessment tools to assess
condition based on macroinvertebrate assemblages are currently prevalent: multimetric indices
and predictive models of taxa richness. The purpose of these indicators is to present the
complex community taxonomic data represented within an assemblage in a way that is
understandable and informative to resource managers and the public. For NLA 2012, we
developed a multimetric index of macroinvertebrate condition.
Multimetric indicators have been used in the U.S. to assess condition based on fish and
macroinvertebrate assemblage data (e.g., Karr and Chu, 2000; Barbour et al., 1999; Barbour et
al., 1995). The multimetric approach involves summarizing various assemblage attributes (e.g.,
composition, tolerance to disturbance, trophic and habitat preferences) as individual "metrics"
or measures of the biological community. Candidate metrics are then evaluated for various
aspects of performance and a subset of the best performing metrics are then combined into an
index, referred to as a multimetric index or MMI. In order to amass the largest dataset possible,
macroinvertebrate data from both the NLA 2007 and NLA 2012 were combined and analyzed
together to develop the MMI and calculate condition class thresholds. Thus, metrics and
subsequent MMI scores were calculated in an identical manner for both NLA datasets.
4.2 Data preparation
4.2.1 Standardizing counts
The number of individuals counted in a sample was standardized to a constant number to
provide an adequate number of individuals that was the same for the most samples and that
could be used for multimetric index development. A subsampling technique involving random
sampling without replacement was used to extract a true "fixed count" of 300 individuals from
the total number of individuals enumerated for a sample (target lab count was 500 individuals).
Samples that did not contain at least 300 individuals were used in the assessment because low
counts can indicate a response to one or more stressors. Only those sites with at least 250
individuals, however, were used as reference sites.
4.2.2 Autecological characteristics
Autecological characteristics refer to specific ecological requirements or preferences of a taxon
for habitat preference, feeding behavior, and tolerance to human disturbance. These
characteristics are prerequisites for identifying and calculating many metrics. A number of
state/regional organizations and research centers have developed autecological characteristics
31
NLA 2012 Technical Report. October Version 1.1
-------
for benthic macroinvertebrates in their region. For the NLA 2012, a consistent "national" list of
characteristics that consolidated and reconciled any discrepancies among the regional lists was
needed before certain biological metrics could be developed and calibrated and an MMI could
be constructed. The same autecological information used in WSA and NRSA was used in NLA.
Members of the data analysis group pulled together autecological information from five
existing sources: the EPA Rapid Bioassessment Protocols document, the National Ambient
Water Quality Assessment (NAWQA) national and northwest lists, the Utah State University list,
and the EMAP Mid-Atlantic Highlands (MAHA) and Mid-Atlantic Integrated Assessment (MAIA)
list. These five were chosen because they were thought to be the most independent of each
other and the most inclusive. A single national-level list was developed based on the following
decision rules:
4.2.3 Tolerance values
Tolerance value assignments followed the convention for macroinvertebrates, ranging between
0 (least tolerant or most sensitive) and 10 (most tolerant). For each taxon, tolerance values
from all five sources were reviewed and a final assignment made according to the following
rules:
1. If values from different lists were all <3 (sensitive), final value = mean.
2. If values from different lists were all >3 and <7 (facultative), final value = mean.
3. If values from different lists were all >7 (tolerant), final value = mean.
4. If values from different lists spanned sensitive, facultative, and tolerant categories,
best professional judgment was used, along with alternative sources of information
(if available) to assign a final tolerance value.
5. Tolerance values of 0 to <3 were considered "sensitive" or "intolerant." Tolerance
values >7 to 10 were considered "tolerant," and values in between were considered
"facultative."
4.2.4 Functional feeding group and habitat preferences
In many cases, there was agreement among the five data sources. When discrepancies in
functional feeding group (FFG) or habitat preference ("habit") assignments among the five
primary data sources were identified, a final assignment was made based on the most
prevalent assignment. In cases where there was no prevalent assignment, the workgroup
examined why disagreements existed, flagged the taxon, and used best professional judgment
to make the final assignment.
4.2.5 Taxonomic resolution
Taxonomic resolution is an import factor in the development of multimetric indices.
Maintaining consistent taxonomic resolution for specific taxa across sites helps ensure that
32
NLA 2012 Technical Report. October Version 1.1
-------
differences between sites are due to environmental factors and not an artifact of taxa
identifications. For most taxa identified the taxonomic resolution was to the generic level,
however the following groups had higher hierarchical taxonomic resolution: oligochaetes,
mites, polychaetes were rolled up to family, ceratopogonids were rolled up to subfamily.
4.3 Multimet development
4.3.1 Data Set
The NLA macroinvertebrate 300 fixed count data was used to calculate the community metrics
used in the MMI. A best ecoregional MMI was developed by scoring and summing the six
metrics that performed best in each ecoregion. We combined the NLA 2007 and 2012 benthic
metric files which were both calculated with common autecology and taxonomic resolution. All
reference sites from both 2007 and 2012 data were defined using the NLA 2012 definitions
described in Section 3 based on nine aggregate ecoregion criteria. The goal was to make the
2007 and 2012 data as comparable as possible so they could be combined for analysis.
Reference sites that had less than 250 individuals were not used as reference for MMI
development. All told, there were 2330 site visits (samples) in the data; 1132 from 2007 and
1198 from 2012. There were 1789 unique sites. Some sites were sampled twice in their
respective years and some sites were sampled in both 2007 and 2012.
4.3.2 Low Macroinvertebrate Numbers
A large number of samples had a very low number of individuals. Examination of these low
number sites did not suggest that this was primarily due to impairment. We think that it is
related to field collection and lake bottom substrate composition. Samples with low bug
numbers will have poor MMI scores because of the strong relationship between sample count
and taxa richness. We decided that samples with less than 100 individuals were not sufficiently
sampled and we would not assess them. They were removed from the process of MMI
development and MMI scores will be set to missing values. These are identified as "not
assessed" in the NLA. In the combined NLA 2007 and 2012 data, 182 samples had < 100
individuals. In the 2012 population, these represent 11,862 lakes (11% of the population).
4.3.3 Ecoregion Classification
For the NLA 2012 assessment, the nine national aggregate ecoregions (Figure 3-1) were
aggregated into five aggregate biological ecoregions by combining some ecoregions together.
Specifically, that consisted of making an Eastern Highlands (EHIGH) region by combining the
SAP and NAP, a PLAINS ecoregion by combining the TPL, SPL, and NPL, and a Western ecoregion
(WMTNS) by combing the WMT and XER regions. The CPL and UMW remain their own
ecoregions. MMIs were developed independently for each of these 5 biological ecoregions.
Ecoregion boundaries were defined by most current (2015) Omernik Ecoregion GIS layers.
33
NLA 2012 Technical Report. October Version 1.1
-------
4.3.4 Metric Screening
All 126 calculated benthic metrics were screened for both signal:noise (S:N) and discrimination
of least-disturbed reference sites from most-disturbed sites (F-test). S:N ratios were calculated
for each metric nationally and within each biological ecoregion using the visit 1 versus visit 2
variance within year as the noise and among site variance as the signal. For calculating F-tests,
and all subsequent MMI development, we only used one visit per site (index visit). The first
sample visit of the year with valid data was used. For sites with valid samples in both years, the
2012 first visit data were used (samples with less than 100 bugs were not considered valid
data). F-tests were run on just the least disturbed reference (R) versus the most disturbed (T)
sites.
Metrics had to pass both F and S:N screens in order to remain in consideration for inclusion in
the final MMI. Metrics had to have S:N > 1.5 either nationally or within their ecoregion in order
to pass. For the F-test, only metrics that had F-values > 4.0 passed. From this screening, 35
metrics from CPL, 42 from EHIGH, 44 from UMW, 29 from PLAINS, and 50 from WMTNS passed
and were considered for the all subsets MMI selection.
4.3.5 All Subsets MMI selection
Passing metrics were assigned to one of the six basic metric classes used to assemble the MMI
as done in the NARS stream MMI (Stoddard et al., 2008). An all subsets procedure was used to
assemble all possible combinations of MMIs using the six metric class framework. There were
8,960 combinations of metrics in the CPL, 12,096 in the EHIGH, 36,855 in the UMW, 3360 in the
PLAINS, and 65,280 in the WMTNs. For each possible MMI combination, the MMI S:N, F-test,
metric correlations, and IQR box delta (separation between least and most disturbed) were
calculated. For correlations, both the mean and maximum correlation among the six metrics
were calculated. IQR box delta or separation is the difference between the 25th percentile of
reference sites and the 75th percentile of most disturbed sites. Thus positive box deltas indicate
separation between the least and most disturbed boxes, negative values indicate overlap in the
IQRs (boxes of box and whisker plot) of the least and most disturbed sites.
To pick the best MMI from the all subsets results, all MMI candidates were first screened for
S:N and maximum metric correlation. Only MMIs that had max correlation < 0.7 and S:N > 3
were considered. MMIs that passed this screen were evaluated for both box delta and F-value
with the goal of picking the MMI that had the best combination of those two values. These two
measures are highly correlated. To do this objectively, we ran a PCA on box delta and F-value
and selected the MMI that had the highest PCA factor 1 score. The intent was to optimize and
pick the model with the best combination of F-value and separation. The six metrics that make
up the final (best) MMI are shown in Table 4-1.
Each of the six selected metrics were scored on a 0-10 scale by interpolating metrics between a
floor and ceiling value. The six metric 0-10 point scaled scores were then summed and
34 NLA 2012 Technical Report. October Version 1.1
-------
normalized to a 0-100 scale by multiplying by 100/60 to calculate the final MMI. Details of this
process are described in Stoddard et al. (2008) for the NARS stream MMI but the NLA process is
the same. The final metrics used in each ecoregion, metric direction, and floor and ceiling
values are summarized in Table 4-1. Scoring equations are different depending on if the metric
responds positively (high values good) or negatively (high values bad) with disturbance. For
positive metrics, values above the ceiling get 10 points, and values below the floor get 0 points.
For negative metrics, values above the ceiling get 0 points, and values below the floor get 10
points. The interpolation equations for scoring the 0-10 points for metrics between the floor
and ceiling values are,
Positive Metrics: Metric Points = 10*((metric value-floor)/(ceiling-floor))
Negative Metrics: Metric Points = 10 * (1 - ((metric value-floor)/(ceiling-floor))).
For positive metrics, floor values are set at the 5th percentile of all samples in the ecoregion,
ceiling values are the 95th percentile of reference sites in the ecoregion. Negative metric
floor/ceilings are calculated the opposite way. Statistics for the final MMI in each ecoregion are
shown in Table 4-2. The overall S:N of the MMI based on visit 1 vs. 2 revisits nationally across
both years was 3.56. Box plots showing the R versus T discrimination of the final MMIs are
shown in Figure 4-1.
35
NLA 2012 Technical Report. October Version 1.1
-------
Table 4-1. Final NLA 2007-2012 biological ecoregion benthic MMI metrics and their floor/ceiling values for MMI
scoring.
Ecoregion
Metric Class
Metric name*
Direction
Floor Value
Ceiling
Value
Coastal Plains
Composition
NOINPTAX
Negative
21.88
55.17
Coastal Plains
Diversity
CHIRDOM3PIND
Negative
38.57
96.08
Coastal Plains
Feeding Group
PREDRICH
Positive
6.00
23.0
Coastal Plains
Habit
SPWLRICH
Positive
5.00
15.0
Coastal Plains
Richness
EPT RICH
Positive
1.00
8.00
Coastal Plains
Tolerance
NTOLPIND
Positive
6.33
64.33
E. Highlands
Composition
NOINPTAX
Negative
13.79
48.72
E. Highlands
Diversity
CHIRDOM3PIND
Negative
39.87
85.94
E. Highlands
Feeding Group
COGARICH
Positive
8.00
27.0
E. Highlands
Habit
CLNGRICH
Positive
3.00
12.0
E. Highlands
Richness
EPOTRICH
Positive
2.00
14.0
E. Highlands
Tolerance
TL23RICH
Positive
1.00
9.00
Plains
Composition
DIPTPTAX
Negative
16.67
60.00
Plains
Diversity
HPRIME
Positive
0.65
3.17
Plains
Feeding Group
PREDRICH
Positive
2.00
19.0
Plains
Habit
CLMBPTAX
Positive
10.0
33.33
Plains
Richness
EPOTRICH
Positive
0
10.0
Plains
Tolerance
TL23PIND
Positive
0
19.67
Upper Midwest
Composition
NOINPIND
Negative
5.33
89.0
Upper Midwest
Diversity
CHIRDOM3PIND
Negative
36.51
87.91
Upper Midwest
Feeding Group
SHRDPIND
Negative
2.67
50.67
Upper Midwest
Habit
CLNGRICH
Positive
3.00
14.0
Upper Midwest
Richness
CRUSRICH
Negative
0
3.00
Upper Midwest
Tolerance
TL23PTAX
Positive
2.17
23.81
Western Mts.
Composition
ODONPIND
Negative
0
17.33
Western Mts.
Diversity
CHIRDOM5PIND
Positive
7.33
98.25
Western Mts.
Feeding Group
SCRPRICH
Negative
0
5.00
Western Mts.
Habit
CLNGRICH
Positive
1.00
8.00
Western Mts.
Richness
TRICRICH
Positive
0
4.00
Western Mts.
Tolerance
TL23PTAX
Positive
0
21.43
*Metric Names
NOINPTAX=% Non-Insect Taxa (Non-Insect Taxa Richness /Total Taxa Richness*100)
DIPTPTAX = % Diptera Taxa (Diptera Taxa Richness / Total Taxa Richness*100)
NOINPIND = % Non-Insect Individuals
ODONPIND = % Odonata Individuals
CHIRDOM3PIND = % Chironomid Individuals in Top 3 most abundant Chironomid Taxa
36 NLA 2012 Technical Report. October Version 1.1
-------
CHIRD0M5PIND = % Chironomid Individuals in Top 5 most abundant Chironomid Taxa
HPRIME = Shannon Diversity Index
PREDRICH = Predator Taxa Richness
COGARICH = Collector-Gatherer Taxa Richness
SHRDPIND = % Shredder Individuals
SCRPRICH = Scraper Taxa Richness
SPWLRICH = SprawlerTaxa Richness
CLNGRICH = Clinger Taxa Richness
CLMBPTAX = % Climber Taxa (Climber Taxa Richness /Total Taxa Richness *100)
EPT_RICH = Ephemeroptera + Plecoptera +Trichoptera Taxa Richness
EPOTRICH = Ephemeroptera + Plecoptera + Trichoptera + Odonata Taxa Richness
CRUSRICH = Crustacean Taxa Richness
TRICRICH = Trichoptera Taxa Richness
NTOLPIND = % Individuals with pollutant tolerance values < 6
TL23RICH = Taxa Richness of taxa with pollutant tolerance values > 2.0 and < 4.0
TL23PIND = % Individuals with pollutant tolerance values > 2.0 and < 4.0
TL23PTAX = % Taxa with pollutant tolerance values > 2.0 and < 4.0
Table 4-2, Final NLA 2007-2012 biological ecoregion benthic MMI statistics.
Ecoregion
F-test
Box Delta
Max Corr.
Mean Corr.
S:N
Coastal Plain
54.7
12.7
0.45
0.17
3.45
E. Highlands
69.0
1.85
0.50
0.26
3.12
Plains
36.2
-2.26
0.68
0.41
3.35
Upper Midwest
64.5
10.4
0.57
0.24
3.00
Western Mts.
88.9
4.46
0.48
0.16
3.66
F-test=F-score for difference between reference and trash site means; Box Delta=Separation difference between
Reference Q1 and most-disturbed Q3 in MMI units; Corr=Pearson correlation among six MMI metrics; S:N =
Ecoregional within year S:N ratio.
37
NLA 2012 Technical Report. October Version 1.1
-------
90
1
10
* * *
0J 1 1 1 1 1 1 1 1 1 1
CPL-R CPL-T EHIGH-R EHIGH-T PLAINS-R PLAINS-T UMW-R UMW-T WMTNS-R WMTNS-T
Ecoregion - Reference/Trash
Figure 4-1. Box and whisker plots showing discrimination between reference (R) and trash (T) sites by biological
ecoregion. Whiskers show the 5th and 95th percentiles
4.3.6 Setting MMI Thresholds
Previous large-scale assessments have converted MMI scores into classes of assemblage
condition by comparing those scores to the distribution of scores observed at least-disturbed
reference sites. See Section 3.3 for information on selecting reference sites. If a site's MMI
score was less than the 5th percentile of the reference distribution, it was classified as in most
disturbed condition; scores between the 5th and 25th percentile were classified as moderately
disturbed and scores in the 25th percentile or higher were classified as least disturbed. This
approach assumes that the distribution of MMI scores at reference sites reflects an
approximately equal, minimum level of human disturbance across those sites. But this
assumption did not appear to be valid for some of the ecoregions.
38
NLA 2012 Technical Report. October Version 1.1
-------
Percentile-based thresholds were adjusted for reference site quality by regressing MMI versus a
PCA Factor 1 disturbance score. For the PCA disturbance factor, all variables used in the NLA
reference site screening (TP, TN, CI, S04, Turbidity, physical habitat disturbance indices, and
assessment indices - Table 3-1) were put into the PCA. Values were log transformed before
analysis. The first principal component (Factor 1) of this PCA well represented a generalized
gradient of human disturbance. There were 247 NLA reference sites with full disturbance data
that was required to calculate the PCA disturbance factor score. Before threshold calculation, a
1.5*IQR outlier analysis was done on the reference site MMIs to remove outliers. Three sites
were dropped as outliers (2 in the UMW and 1 in the WMTNS) leaving 244 reference sites for
analysis.
MMI scores at the reference sites were weakly, but significantly, related to this disturbance
gradient (Figure 4-2). Thus, MMI reference distributions from these regions may be biased
downward, because they include somewhat disturbed sites which may have lower MMI scores.
Herlihy et al. (2008) developed a process that used this PCA disturbance gradient to reduce the
effects of disturbance on threshold values within the reference site population. The process
uses multiple regression modeling to develop adjusted thresholds analogous to the 5th and 25th
percentiles of reference sites in each ecoregion based on the slope of the MMI-disturbance
relationship in each ecoregion. Briefly, the process involves setting the goal for disturbance to
the 25th percentile of the Factor 1 disturbance score for reference sites in each ecoregion. The
ecoregion MMI value at that goal is predicted from the MMI-disturbance regression as,
MMIpred = (GOAL * SLOPE) + INTERCEPT.
Then the percentiles to be used as the adjusted thresholds are calculated assuming there is a
normal distribution around this predicted mean using the RMSE of the regression model as the
standard error,
Least-Moderately Disturbed 25th threshold = MMIpred - 0.675 * RMSE
Moderately-Most Disturbed 5th threshold = MMIpred - 1.650 * RMSE.
The best regression model from the NLA reference site data had a common slope and separate
intercepts by ecoregion. The pooled model RMSE was 11.01, the common slope was -7.953 and
the intercepts were 65.45 in the CPL, 54.30 in the EHIGH, 60.14 in the UMW, 61.47 in the
Plains, and 61.73 in the WMTNS. The resulting adjusted MMI threshold values for the condition
classes in each ecoregion used in the NLA 2012 report are given in Table 4-3.
39
NLA 2012 Technical Report. October Version 1.1
-------
Table 4-3, NLA2012 macroinvertebrate MMI thresholds.
Adjusted 25th
Adjusted 5th
Least-Disturbed
Most Disturbed
Ecoregion
# of Ref Sites
Threshold
Threshold
Coastal Plains
23
>54.8
<44.1
East. Highlands
70
>51.5
<40.8
Plains
48
>46.8
<36.1
Upper Midwest
35
>58.1
<47.3
Western Mountains
68
>64.8
<54.1
NLA Benthic Reference Sites
-2-101234
PCA Factor 1 Score
Figure 4-2, MMI score versus PCA factor 1 disturbance score for NLA macroinvertebrate reference sites. Higher
PCA factor 1 scores indicate more disturbance.
40
NLA 2012 Technical Report. October Version 1.1
-------
4.4 Literature cited
Barbour, M. T., J. Gerritsen, B. D. Snyder, and J. B. Stribling. 1999. Rapid bioassessment
protocols for use in streams and wadeable rivers. EPA 841/B-99/002. Office of Water,
US Environmental Protection Agency, Washington, DC.
Barbour, M. T., J. B. Stribling, and J. R. Karr. 1995. Multimetric approach for establishing
biocriteria and measuring biological condition. Pages 63-77 in W. S. Davis and T. P.
Simon (editors). Biological assessment and criteria: tools for water resource planning
and decision making. Lewis Publishers, Boca Raton, Florida.
Herlihy, A. T., S. G. Paulsen, J. Van Sickle, J. L. Stoddard, C. P. Hawkins, and L. L. Yuan. 2008.
Striving for consistency in a national assessment: the challenges of applying a reference
condition approach at a continental scale. Journal of the North American Benthological
Society 27:860-877.
Karr, J. R., and E. W. Chu. 2000. Sustaining living rivers. Hydrobiologia 422/423:1-14.
Stoddard, J. L., A. T. Herlihy, D. V. Peck, R. M. Hughes, T. R. Whittier, and E. Tarquinio. 2008. A
process for creating multi-metric indices for large scale aquatic surveys. Journal of the
North American Benthological Society 27:878-891.
41
NLA 2012 Technical Report. October Version 1.1
-------
Chapter 5: Physical Habitat
5.1 Background information
Near-shore physical habitat structure in lakes has only recently been addressed by the U.S.
Environmental Protection Agency (EPA) in its National Aquatic Resource Surveys (NARS)
monitoring efforts (e.g., USEPA 2009, Kaufmann et al. 2014a,b,c). Like human activities, aquatic
and riparian biota are concentrated near lakeshores, making near-shore physical habitat
ecologically important, but exposed and vulnerable to anthropogenic perturbation (Schindler
and Scheuerell 2002, Strayer and Findlay 2010, Hampton et al. 2011). Littoral and riparian zones
are positioned at the land-water interface, and tend to be more structurally complex and
biologically diverse than either pelagic areas or upland terrestrial environments (Polis et al.
1997, Strayer and Findlay 2010). This complexity promotes interchange of water, nutrients, and
biota between the aquatic and terrestrial compartments of lake ecosystems (Benson and
Magnuson 1992, Polis et al. 1997, Palmer et al. 2000, Zohary and Ostrovsky 2011). Structural
complexity and variety of cover elements in littoral areas provide diverse opportunities for
supporting assemblages of aquatic organisms (Strayer and Finlay 2010; Kovalenko et al 2012),
while intact riparian vegetation and wetlands surrounding lakes increase near-shore physical
habitat complexity (e.g., Christensen et al. 1996, Francis and Schindler 2006) and buffer lakes
from the influence of upland land use activities (Carpenter and Cottingham 1997, Strayer and
Findlay 2010). Human activities on or near lakeshores can directly or indirectly degrade littoral
and riparian habitat (Francis and Schindler 2006). Increased sedimentation, loss of native plant
growth, alteration of native plant communities, loss of physical habitat structure, and changes
in littoral cover and substrate are all commonly associated with lakeshore human activities
(Christensen et al. 1996, Engel and Pederson 1998, Whittier et al. 2002, Francis and Schindler
2006, Merrell et al. 2009). Such reductions in physical habitat structural complexity can
deleteriously affect fish (Wagner et al. 2006, Taillon and Fox 2004, Whittier et al. 1997, 2002,
Halliwell 2007, Jennings et al. 1999, Wagner et al. 2006), aquatic macroinvertebrates (Brauns et
al. 2007), and birds (Kaufmann et al. 2014b).
The EPA developed standardized, rapid field methods to quantify physical habitat structure and
near-shore anthropogenic disturbances (Kaufmann and Whittier 1997), and piloted them in the
Northeastern U.S. (Larsen and Christie 1993, Whittier et al. 2002b, Kaufmann et al. 2014b).
These methods were modified (USEPA 2007a, Kaufmann et al. 2014a) and applied in 2007 for
the first U.S. national survey of lake physical habitat condition (US EPA 2009, Kaufmann et al.
2014c). The EPA's lake physical habitat methods were once again modified to explicitly assess
habitat structure in exposed drawdown zones (USEPA 2012), and applied in the NLA 2012
survey as part of the EPA's second national survey of the ecological condition of lakes in the
United States (USEPA 2016). The NLA 2012 field method modifications were structured so that
we were able to duplicate of all the lake habitat condition indices that were used in the
previous (2007) national assessment. We calculated habitat metrics and indices described by
Kaufmann et al. 2014a,c) to quantify the variety, structural complexity, and magnitude of areal
42
NLA 2012 Technical Report. October Version 1.1
-------
cover from physical habitat elements within the near shore zones of lakes in the NLA 2012
survey.
Our objectives in this chapter are to describe how we calculated physical habitat indices based
on near-shore physical habitat data collected in the NLA survey, and how we derived physical
habitat condition thresholds relative to least-disturbed conditions. We only briefly describe the
NLA field methods and data reduction procedures, which are published elsewhere (USEPA
2012; Kaufmann 2014a). Finally we evaluate the precision of NLA's key indices of physical
habitat condition and examine their association with anthropogenic disturbances.
5.2 Data preparation
We took the following eight steps to assess physical habitat condition in U.S. lakes based on the
NLA 2012 national probability sample of lakes and reservoirs.
1) Field crews made measurements and observations of near-shore physical habitat structure
and human activities on a national probability sample of lakes and reservoirs (described by
USEPA 2016, and Kaufmann et al. 2014a);
2) Classified survey lakes by aggregated ecoregion (ECOWSA9_2015), and by their relative
levels of anthropogenic disturbance within those ecoregions (RT_NLA12_2015).
3) Calculated a set of physical habitat metrics as described by Kaufmann et al. (2014a) for NLA
2007, but adapted calculations to adjust for the NLA 2012's field method change that
assessed riparian vegetation cover, littoral cover, and human disturbance in the drawdown
zone separate from those above the typical high water mark or inundated by water in the
littoral zone;
4) Calculated multimetric indices of lakeshore anthropogenic disturbance and nearshore
physical habitat cover and structure as described by Kaufmann et al. (2014c) for NLA 2007,
and assigned variants of these indices according to aggregated Ecoregions
(ECOWSA9_2015); also defined a new indicator of lake drawdown;
5) Estimated lake-specific expected ("E") values for physical habitat indices from region-
specific regression models of factors predicting physical habitat in the combined set of
least-disturbed lakes from the NLA 2007 and 2012 surveys. Our modeling approach is very
similar to that employed by Kaufmann et al. (2014c) in the Western Mountain and Xeric
ecoregions for the NLA 2007 report;
6) Set criteria for low, medium and high lakeshore anthropogenic disturbance (good, fair,
poor) based on professional judgement; good, fair, and poor littoral and riparian physical
habitat condition based on deviation from the central tendency of observed/expected (O/E)
values within the group of least-disturbed lakes; and small, medium, and large lake
drawdown based on percentiles of the indicator values themselves in least-disturbed lakes.
7) Examined the precision of NLA 2012 key physical habitat indicators.
8) Examined the association between NLA 2012 physical habitat indicators and anthropogenic
disturbances, comparing the regional distributions of habitat condition in least-disturbed
reference lakes with those in highly disturbed lakes.
43
NLA 2012 Technical Report. October Version 1.1
-------
5.3 Methods
5.3.1 Study area and site selection
The NLA field sampling effort targeted all lakes and reservoirs in the 48 conterminous U.S. with
surface areas >1 ha and depths greater than 1 m. Field crews visited 1131 lakes and reservoirs
between May and October 2012. Of these, 1038 had been selected as a probability sample
from the USGS/EPA National Hydrography Dataset (NHD) with a spatially-balanced, randomized
systematic design that excluded the Great Lakes and Great Salt Lake (Peck et al. 2013). The
remaining 91 lakes were hand-selected to increase the number of lakes in least-disturbed
condition, which were used to estimate potential condition and evaluate response of the
indices to disturbance (following Stoddard et al. 2006). For the NLA 2012 report, we used
physical habitat data collected from 1109 of the 1131 survey lakes, which were those having
surface areas <10,000 ha (1026 probability-selected and 83 hand-picked lakes). Probability and
hand-selected lakes from both 2012 and 2007 were used to develop expected physical habitat
condition models and distributions of O/E values in least-disturbed lakes. Random subsets of 90
probability lakes from NLA 2007 and 88 from NLA 2012 were visited twice during their
respective summer sampling periods to estimate the precision of NLA indicators, including the
habitat measurements and indices (Kaufmann et al. 2014a).
5.3.2 Field sampling design and methods
Our lake physical habitat field methods (USEPA 2007a, USEPA 2012, Kaufmann et al. 2014a)
produced information concerning 7 dimensions of near-shore physical habitat: 1) water depth
and surface characteristics, 2) substrate size and type, 3) aquatic macrophyte cover and
structure, 4) littoral cover for biota, 5) riparian vegetation cover and structure, 6) near-shore
anthropogenic disturbances, and 7) bank characteristics that indicate lake level fluctuations and
terrestrial-aquatic interactions. At each lake, field crews characterized these 7 components of
near-shore physical habitat at 10 equidistant stations along the shoreline. Each station included
a littoral plot (10m x 15m) abutting the shoreline, a riparian plot (15m x 15m) extending
landward from the typical high-water mark, and in a 15m wide drawdown zone plot that
extended a variable distance landward, depending on the amount of lake level drop compared
with typical high water levels (Figure 5-1). Littoral depth was measured 10 m off-shore at each
station. Metrics and indices were calculated for the variable-width drawdown zone plots, the
15m x 15m riparian plots and the 10m x 15m littoral plots. To match the riparian and near-
shore human disturbance indices to those used in the previous (NLA 2007) assessment, we
used information from riparian and drawdown plots along with drawdown horizontal extent
information. These index values are equivalent to the 2007 index values that were directly
calculated from observation the near-shore zone extending from the lake water's edge 15m
outward. See Kaufmann et al. (2014a) for further description of field methods, our approach for
calculating whole-lake physical habitat metrics, and a detailed assessment of habitat metric
precision.
44
NLA 2012 Technical Report. October Version 1.1
-------
5.3.3 Classifications
5.3.3.1 Eco regions
We report findings nationally, and by 9 aggregated Omernik (1987) level III ecoregions (Paulsen
et al. 2008): the Northern Appalachians (NAP), Southern Appalachians (SAP), Coastal Plains
(CPL), Upper Midwest (UMW), Temperate Plains (TPL), Northern Plains (NPL), Southern Plains
(SPL), Western Mountains (WMT), and Xeric West (XER) (Figure 3-1). We used ecoregions as a
first-level classification for defining and evaluating near-shore riparian and littoral condition
indicators (RVegQ, LitCvrQ, and LitRipCvrQ) and their variants (e.g., RVegQ_2, LitCvrQ_b,
LitRipQ^2d). Ecoregions are useful predictors of many characteristics of landform, geology,
climate, hydrology, and potential natural vegetation (Omernik 1987, Paulsen et al. 2008) that
influence physical habitat in lakes (Kaufmann et al. 2014c). Kaufmann et al. (2014c) used a
multivariate classification of lake characteristics including lake chemistry and depth to assign
variants of LitCvrQ, suggesting that such classifications would capture aspects of in-lake habitat
cover complexity better than would ecoregions. We reexamined the 2007 data and found no
substantial difference in assignment of LitCvrQ variants according to Ecoregion (WSAEC09)
versus multivariate cluster analysis (CLUSB). For some aspects of habitat index development,
we grouped ecoregions into broader ecoregions: the Eastern Highlands (EHIGH = NAP + SAP),
the Plains and Lowlands (PLNLOW = CPL + UMW + TPL + NPL + SPL), Central Plains (CENPL =
TPL+ NPL+SPL), and the West (WMT + XER).
5.3.3.2 Anthropogenic disturbance and least-disturbed reference site screening
We used region-specific screening based on water chemistry, near-shore human influences, and
evidence of anthropogenic lake drawdown in NLA survey lakes, 1109 from NLA 2012 and 1101
from NLA 2007, to classify all NLA lakes according to their level of anthropogenic disturbance
(low, medium, high), as described in Chapter 3. Lakes meeting low-disturbance screening
criteria served as least-disturbed reference sites for best-available condition. Low-disturbance
stress (least-disturbed) lakes within each Ecoregion were identified on the basis of chemical
variables (total phosphorus, total nitrogen, chloride, sulfate, acid neutralizing capacity,
dissolved organic carbon, and dissolved oxygen in the epilimnion) and direct observations of
anthropogenic disturbances along the lake margin (proportion of lakeshore with non-
agricultural influences, proportion of lakeshore with agricultural influences, and the relative
extent and intensity of human influences of all types together). For each aggregated ecoregion,
a threshold value representing least-disturbed conditions was established as a "pass/fail"
criterion for each parameter (Table 3-1). Thresholds were values that would be very unlikely in
least-disturbed lakes within each region, and varied by lake type to account for regional
variations in water chemistry and littoral-riparian human activities (Herlihy et al. 2013). A lake
was considered least-disturbed if it passed the screening test for all parameters, and we
identified 214 least-disturbed lakes from NLA 2012 and 168 from NLA 2007. We used the 2012
survey data for the 44 lakes from NLA 2007 that were again sampled in NLA 2012, and still
passed the reference screening, so 124 NLA 2007 lakes remained in the reference set (Table
45
NLA 2012 Technical Report. October Version 1.1
-------
5-1). Lakes that were not classified as least-disturbed were provisionally considered
intermediate in disturbance. The intermediate disturbance lakes were then screened with a set
of high-disturbance thresholds applied to the same variables (Table 3-2) Lakes that exceeded
one or more of the high disturbance thresholds were considered highly disturbed. To avoid
circularity in defining physical habitat alteration, we did not use any of the physical habitat
cover complexity indices or their subcomponent metrics in defining lake disturbance classes.
Our screening process identified 382 least-disturbed, 1309 intermediate, and 519 highly
disturbed lake visits. Of the 338 least-disturbed lakes that did not overlap survey years, 190
were in the WMT, NAP, and UMW aggregated ecoregions (Table 5-1). Even with relaxed
disturbance screening criteria, it was more difficult to find least-disturbed lakes in some other
ecoregions. Respectively, only 11, 20, and 23 least-disturbed lakes were identified in the NPL,
XER, and TPL ecoregions. To increase the useable sample size for estimating expected lake
condition, we grouped least-disturbed lakes from the NPL, SPL, TPL into the Central Plains
(CENPL), and the WMT and XER into the West (for some models). Because of insufficient
numbers of least-disturbed lakes relative to the large amount of lake variability within
ecoregions, we needed all available reference lakes for modeling expected conditions, so were
unable to use totally independent subsets of lakes for developing and validating those models.
5.3.4 Calculation of lake physical habitat metrics
5.3.4.1 Names of habitat metrics
Our variable names are those from the publicly-available NLA 2007 and NLA 2012 datasets
released by the U.S. EPA (http://water.epa.gov/type/lakes/NLA_data.cfm). The first several
letters in the NLA variable names denote the category and type of metric. The initial letters
"hi..." identify human influence metrics. The initial letters "hifp..." specify human influence
frequency of presence metrics and "hii..." specify indices of aggregated or summed human
influences. Riparian vegetation mean presence metrics begin with "rvfp..." and mean riparian
vegetation cover metrics begin with "rvfc...", whereas "rvi..." denotes riparian vegetation cover
sums (e.g., two types of woody cover). The initial letters "fc..." and "am..." indicate,
respectively, fish cover and aquatic macrophyte metrics. These letters followed by "...fp...",
"..fc...", or "..i.." indicate, respectively mean frequency of presence among stations, mean
areal cover, and indices created by summing various metrics. Littoral bottom and exposed
shoreline substrate metrics, respectively, are identified by "bs..." and "ss...". The summary
habitat indices described by Kaufmann et al. (2014c), and used to define habitat condition in
the NLA (RVegQ, LitCvrQ, and LitRipCvQ) all end in the upper case Q, and the NLA summary
human disturbance index is RDis_IX (Riparian Disturbance Intensity and eXtent). Kaufmann et
al. (2014a) describe in detail the definitions and calculation of NLA physical habitat metrics and
quantify their precision.
Many of the physical habitat metrics for NLA 2012 are additionally identified by the suffixes
_rip, _lit, and _DD (e.g., rviWoody_rip, rviWoody_DD, fciNatural_lit, fciNatural_DD),
46
NLA 2012 Technical Report. October Version 1.1
-------
designating that the habitat observations or measurements were from, respectively, the set of
riparian, littoral, or drawdown plots (Figure 5-1).
5.3.4.2 Drawdown Zone Apportioning to match NLA 2007 Riparian and Human Disturbance
metrics:
NLA 2012 retained the measures of "bathtub ring" height and horizontal extent exactly as done
in NLA 2007 to quantify lake drawdown and seasonal lake level fluctuations. However, the near-
shore plot designs of the two surveys differ. In NLA 2007, the 15m x 15m riparian plots abutted
the shoreline. Consequently, exposed littoral bottom may comprise 0 to 100% of NLA 2007
plots, depending upon the extent of drawdown. Near-shore habitat was accurately depicted in
the NLA 2007 data, but because cover and disturbances were not separately assessed in the
drawdown zone, there was no accurate way to separately assess changes in habitat condition
attributable to drawdown (vs. riparian vegetation removal, for example). The NLA 2012 field
methods have separate measures of vegetation and human disturbances for the riparian and
drawdown zone plots, and separate fish cover estimates in littoral and drawdown zone plots.
These field plot changes improve the separation of lake level changes and drawdown from
other stressors in a diagnosis of likely causes of poor nearshore habitat condition in NLA 2012.
We used cover and human disturbance tally data from the riparian and drawdown plots to
calculate cover estimates or disturbance tallies simulating the set often 15m x 15m near-shore
plots abutting the shoreline, as had been used in the NLA 2007 field methods. We calculated
RCsyn, as a synthetic estimate of cover in the 15m band around the shoreline by summing the
areal covers in the drawdown and riparian plots, after weighting each by the proportion of the
15m band that was, respectively, within the drawdown zone or not within the drawdown zone:
RCsyn = (Rpdraw X RCdraw) + (Rprip X RCrip) (Eq 1)
where:
RCsyn = Calculated cover in 15 x 15 m shoreline PHab plot, synthesizing metric values equivalent
to those used in NLA 2007, which represent the riparian condition in the 15m near-
shore band adjacent to the wetted edge of the lake.
Rpdraw and Rprip are the proportions of the 15x15m shoreline PHab plot that are, respectively,
occupied by the drawdown zone and the riparian zone above the high water mark.
Rpdraw = (Horizontal Distance to high water)/(15m) = (bfxHorizDist/15m), and Rpdraw=l-0 if
bfxHorizDist> 15m.
Rprip = (1 - Rpdraw) by definition because RpriP+ Rpdraw= 1.0
Redraw and RcriP are, respectively, the areal cover of vegetation in the drawdown and riparian
zones; RcriP could be single cover type (e.g., canopy layer, or barren ground), or could
be a sum of cover types (e.g., sum of woody cover in 3 layers).
47 NLA 2012 Technical Report. October Version 1.1
-------
Calculated Rcsynfor a hypothetical lake with a mean horizontal drawdown of 10m (est. by
bfxHorizDist), and 100% canopy cover above the high water mark, but 0% cover in the
drawdown zone is as follows:
Rpdraw= 10/15 = 0.67
Rprip =(1.0-0.67) = 0.33
Drawdown Canopy cover: Rcdraw= 0%
Riparian Canopy cover: RcriP = 100%
RCsyn = (0.67 x 0%) + (0.33 x 100%) = 33%
The loss or gain in near-shore riparian habitat cover resulting from lake drawdown or natural
lake level declines can be estimated by the difference in cover between the riparian cover
above the high water mark (RcriP) and that within 15 m of the lakeshore (Rcsyn)-
We conducted a volunteer Drawdown Pilot Survey in 2011 to determine whether modification
of the NLA 2007 field protocols could be made without jeopardizing our ability to track changes
or trends in riparian habitat over time (Anne Rogers 2012 NALMS; Kaufmann et al. Jan 9, 2012
webinar presentation to NLA steering committee and states). NLA 2007 and NLA 2012 field
protocols were applied simultaneously at 210 stations on 21 lakes spread over a range of
drawdown conditions in the states of Texas, Wisconsin, Washington, Oregon, Wyoming, North
Dakota, and Colorado. Kaufmann et al. (2012 webinar) demonstrated that 2007 metric values
for lakeshore vegetation and human disturbances were calculated accurately from the new
(2012) protocol, preserving ability to track changes/trends. The regressions predicting the
measured values of key physical habitat metric values from the NLA 2007 protocol from values
calculated by Eq 1 were virtually 1:1 lines with intercepts very close to 0.0, slopes very close to
1.0, and R2 between 0.87 and 0.94. The drawdown pilot analysis also showed that there was
virtually no difference in whole-lake metric values obtained by applying Eq 1 at each station,
versus applying it once per lake based on values of drawdown extent and cover averaged over
the 10 riparian and drawdown plots on each lake. The drawdown pilot results also
demonstrated that adding separate determinations of habitat cover elements in the drawdown
zone was logistically feasible and resulted in very minor increases in field time.
5.3.4.3 Drawdown Zone Apportioning to Estimate littoral habitat changes due to drawdown:
We used a calculation similar to Eq 1 to simulate the amount of littoral cover that would be
present if, hypothetically, the amount of lake drawdown were zero:
where:
48
LCsim (Lpdraw X LCdraw) + (LpiitX LClit)
NLA 2012 Technical Report. October Version 1.1
(Eq 2)
-------
LcSim = Calculated littoral cover simulating the amount of real or potential cover in a 10 x 15 m
littoral plot abutting the high-water mark, ie., simulating littoral cover that might be
present if there were no drawdown.
Lpdrawand Lput are the estimated proportions of a hypothetical 10m x 15m littoral PHab plot
abutting the highwater mark that are, respectively, occupied by the drawdown zone
(dry) and the littoral zone (wet).
Lpdraw= (Horizontal Distance to high water)/(10m) = (bfxHorizDist/10m), and LPdraw=l-0 if
bfxHorizDist> 10m.
Lpnt= (1 - Lpdraw) by definition because LpriP+ Lpdraw= 1.0
Lent and Lcdraw are, respectively, the areal cover offish habitat elements in the littoral plot, and
exposed (dry) in the drawdown zone, Lc could be single cover type (e.gfcfcSnags) or
could be a sum of cover types (e.g., sum of non-anthropogenic cover types: fcfcNatural).
Calculated LcSim for a hypothetical lake with a mean horizontal drawdown of 10m and 100%
Snag cover in the drawdown zone (dry and exposed), but 0% Snag cover in the littoral
(wet) zone is as follows:
Lpdraw = 10/10 = 1.00
Lput = (1.00-1.00) = 0
Drawdown Snag cover: L.Cdraw= 100%
Littoral Snag cover: Lcat= 0%
LCsim = (1.00 x 100%) + (0 x 0%) = 100%
The loss or gain in littoral habitat cover resulting from lake drawdown or natural lake level
declines can be estimated as the difference between the littoral cover simulated for zero
drawdown conditions (LcSim) the observed cover actually existing in the littoral at the time of
sampling {Lent).
5.3.4.4 Use of Variable suffixes in this report:
Riparian cover or human disturbance metrics calculated by Eq 1 are synthetic values that match
the 2007 metrics, and are designated by the suffixes_sy/7 (e.g., rviWoody_syn and hiiAII_syn) in
the EPA database. For simplicity, we will drop the suffixes on riparian vegetation and human
disturbance metrics in the remainder of this article, and it is understood that we are using the
synthesized variables when no suffix is present (*_syn), and NOT the drawdown zone (*_DD),
or riparian plot (*_r/p) versions of those variables.
49
NLA 2012 Technical Report. October Version 1.1
-------
Littoral cover metrics designated with the suffix_//t are based on field observations that are
conceptually and procedurally identical to those used in NLA 2007. For simplicity, we will drop
the suffixes on littoral cover metrics in the remainder of this article, and it is understood that
we are using the innudated littoral plot version of those variables when no suffix is present
(*_//t), and NOT the drawdown zone (*_DD) or zero-drawdown simulated values (*_s/m)
versions of those variables. Littoral cover metrics calculated using Eq 2 simulate littoral cover
that would be present in the near-shore littoral area if the amount of drawdown were zero, and
are designated by the suffix_s/m (eg.,fciNatural_sim).
5.3.4.5 Near-shore disturbance metrics
We calculated extent of shoreline disturbance around the lakeshore (hifpAnyCirca) as the
proportion of stations at which crews recorded the presence of at least one of the 12
anthropogenic disturbance types as described by Kaufmann et al. (2014a). We calculated the
disturbance intensity metric hiiAII as the sum of the 12 separate proximity-weighted means for
all shoreline disturbance types observed at the 10 shoreline stations (Kaufmann et al. 2014a).
We also calculated subsets of total disturbance intensity by summing metrics for defined
groups of disturbance types. For example, hiiAg sums the proximity-weighted presence metrics
for row crop, orchard, and pasture; hiiNonAg sums the proximity-weighted presence metrics for
the remaining 9 non-agricultural disturbance metrics: 1) buildings, 2) commercial
developments, 3) parks or man-made beaches, 4) docks or boats, 5) seawalls, dikes, or
revetments, 6) trash or landfill, 7) roads or railroads, 8) power lines, and 9) lawns.
5.3.4.6 Riparian vegetation metrics
Field data consisted of visual areal cover % class assignments of the vegetation type and areal
cover for each of 3 layers: canopy (>5 m high), mid-layer (0.5-5 m high), and ground cover (<0.5
m high). Crews estimated large (diameter at breast height [DBH] > 0.3 m) and small (DBH < 0.3
m) diameter tree cover separately in the canopy and mid-layer, distinguished woody from
herbaceous vegetation in the mid-layer and ground cover, and distinguished barren ground
from vegetation inundated by water in the ground layer. To characterize riparian vegetation in
the near-shore zone of the lake, we converted field cover class observations to mean cover
estimates for all the types and combinations of vegetation data (Kaufmann et al. 2014a). We
assigned cover class arithmetic midpoint values to each plot's cover-class observations (i.e.,
absent = 0%, sparse (>0-10%) = 5%, moderate (>10-40%) = 25%, heavy (>40-75%) = 57.5%, and
very heavy (>75-100%) = 87.5%), and then calculated lakeshore vegetation cover as the average
of those cover values across all 10 plots. Metrics for combined cover types (e.g., sum of woody
vegetation in 3 layers) were calculated by summing means for the single-types (see Kaufmann
et al. 1999, 2014a). Metrics describing the proportion of each lakeshore with presence (rather
than cover) of particular features were calculated as the mean of presence (0 or 1) over the 10
riparian plots.
50
NLA 2012 Technical Report. October Version 1.1
-------
5.3.4.7 Littoral cover and aquatic macrophyte metrics
The NLA survey crews made observations of the areal cover attributable to 8 littoral cover types
within each of the 10 littoral plots: rock ledges, boulders, brush, inundated live trees, snags,
overhanging vegetation, aquatic macrophytes, and human structures. Additionally field crews
made separate visual estimates of areal cover for emergent, floating, and submerged aquatic
macrophytes within each of the 10 littoral plots. They used the same % cover classes for these
observations as used for riparian vegetation. Metrics describing the mean cover (and mean
presence) of littoral physical habitat features and aquatic macrophytes were calculated from
these cover class observations as described above for riparian vegetation. Metrics for combined
cover types (e.g. sum of natural types fish cover, floating and emergent aquatic macrophyte
cover) were calculated by summing means for single types.
5.3.4.8 Littoral and shoreline substrate metrics
NLA field crews visually estimated the percent areal cover of 8 substrate types (bedrock,
boulder, cobble, gravel, sand, silt/clay/muck, woody debris, and organic detritus) at each of the
10 near-shore stations (Figure 5-1). These estimates were made separately for the 1 m
shoreline band above the lake margin and for the lake bottom within the littoral plot. In cases
where the bottom substrate could not be observed directly, crews viewed the bottom through
a viewing tube, felt the substrate with a 3 m PVC sounding tube, or observed sediments
adhering to the boat anchor as it was retrieved from the bottom. Cover classes were the same
as for riparian vegetation. We calculated metrics describing the lake-wide mean cover of near-
shore littoral and shoreline substrate in each size category by averaging the cover estimates at
each station, based on the cover class midpoint approach described above.
We adapted the approach of Faustini and Kaufmann (2007) and Kaufmann et al. (2009) for
estimating geometric mean and variance of substrate diameters from systematic pebble-
counts. In this approach (Kaufmann et al. 2014a), we assigned the geometric mean between
the upper and lower diameter bound of each size class for each cover observation before
calculating the cover-weighted mean size index. We calculated the geometric mean diameters
(Dgm) of littoral and shoreline substrate (bsxLdia and ssxLdia) as follows:
Dgm=Antilog{Sum/{P/{[logio(D/u)+logio(D/7 )]/2}}}, (Eq. 3)
where:
P/=areal cover proportion for diameter class /';
D/u=diameter (mm) at upper limit of diameter class /';
Dn =diameter (mm) at lower limit of diameter class /';
Sum, =summation across diameter classes; and
Nominal size class midpoint diameters of 5660 and 0.0077 mm were set, respectively, for the
largest (bedrock and hardpan) and smallest (silt, clay, and muck) diameter classes.
51
NLA 2012 Technical Report. October Version 1.1
-------
Our calculations are identical to those of Faustini and Kaufmann (2007), except that here the
percent cover estimates used to weight diameters were the mean values of 10 visual cover
estimates rather than areal streambed cover determinations derived from the pebble-count
percentages for individual particles in each diameter class.
5.3.4.9 Littoral depth, Lake level fluctuations, bank and water surface characteristics
Field crews measured littoral depth, estimated water level fluctuations and bank heights, and,
and observed water surface and bottom sediment color and odor at each of the 10 nearshore
stations (Figure 5-1). SONAR, sounding lines, or sounding tubes were used to measure lake
depth 10 m offshore. NLA field crews used hand-held levels, survey rods, and laser rangefinders
(rather than unaided visual estimates) to measure vertical and lateral (horizontal) lake level
fluctuation. Field indications of short to medium term fluctuation, drawdown and/or declines in
lake levels were based on measurement of the vertical height and horizontal extent of exposed
lake bottom ("Bathtub Ring") field evidence.
Crews recorded the presence of surface films or scums, algal mats, oil slicks, and sediment color
and odor. They visually estimated the bank angle in the 1 m-wide shoreline band and the
vertical and lateral range in lake level fluctuations, based on high and low water marks. We
calculated whole lake metrics for mean littoral depth and water level fluctuations as arithmetic
averages (sixDepth, bfxVertHeight and bfxHorizDist) and standard deviations of the measured
values at the 10 stations. For bank angle classes and qualitative observations of water surface
condition and sediment color and odor, we calculated the proportion of stations having
observations in each class.
5.3.5 Calculation of summary physical habitat condition indices
We calculated 4 multimetric indices of physical habitat condition and an index of lake
drawdown:
RDis_IX: Lakeshore Anthropogenic Disturbance Index (Intensity and Extent),
RVegQ: Riparian Vegetation Cover Complexity Index,
LitCvrQ: Littoral Cover Complexity Index,
LitRipCvQ: Littoral-Riparian Habitat Complexity Index, and
Drawdown Index: based on bfxVertHeight and bfxHorizDist
5.3.5.1 Lakeshore Anthropogenic Disturbance Index (RDis_IX)
This index was calculated as:
RDis_IX = (Disturbance Intensity + Disturbance Extent)/2; (Eq 4)
where :
52
NLA 2012 Technical Report. October Version 1.1
-------
disturbance intensity was represented by separate sums of the mean proximity-weighted tallies
of near-shore agricultural and non agricultural disturbance types and extent was expressed as
the proportion of the shore with presence of any type of disturbance.
hiiNonAg = Proximity-weighted mean disturbance tally (mean among stations) of up to 9
types of non-agricultural activities.
hiiAg = Proximity-weighted mean tally of up to 3 types of agriculture-related activities
(mean among stations).
hifpAnyCirca = Proportion of the 10 shoreline stations with at least 1 of the 12 types of
human activities present within their 10 x 15 m littoral plots, drawdown plots, or within
15m of the lake shore in their 15 x 15 m riparian plots.
Field procedures classified only 3 types of agricultural disturbances, versus 9 types of non-
agricultural disturbances, limiting the potential ranges to 0-3 for hiiAg and 0-9 for hiiNonAg. In
the combined NLA 2007 and 2012 surveys, the observed ranges of these variables also differed:
hiiAg ranged from 0 to 1.55, whereas hiiNonAg had an observed range almost 5 times as great
(0 to 7.125). To avoid under-representing agricultural disturbances and over-representing non-
agricultural disturbances in the index, we weighted the disturbance intensity tallies for
agricultural land use by a factor of 5 in Equation 2. This weighting factor (ratio of observed
ranges in non-agricultural to agricultural disturbance types) effectively scales agricultural land-
uses equal in disturbance potential to those for non-agricultural land uses. We scaled the final
index from 0 to 1, where 0 indicates absence of any anthropogenic disturbances and 1 is the
theoretical maximum approached as a limit at extremely high disturbance. We applied a single
formulation of the disturbance index RDis_IX throughout the NLA survey in the U.S.
5.3.5.2 Riparian Vegetation Cover Complexity Index (RVegQ)
This index is based on visual estimates of vegetation cover and structure in three vegetation
layers at the 10 near-shore riparian plots along the lake shore. The cover metrics were
calculated for the variable-width drawdown zone plots (metrics with suffix "_DD") and the 15m
x 15m riparian plots (with suffix "_rip"). For the NLA 2012 report, we used areal cover
information from both types of plots along with drawdown horizontal extent information to
calculate RVegQ estimates matching those for the previous report, which are for the near-
shore zone extending from the lake water's edge 15m outward (see Eq. 1). Because the
potential vegetation cover differs among regions, we calculated three variants of the Riparian
Vegetation Cover-Complexity Index (RVegQ^2, RVegQ^.7, or RVegQ_8) for application to
different aggregated ecoregions (Table 5-2). The region-specific formulations reduce the
among-region variation in index values in least-disturbed lakes and reduce ambiguity in their
response to anthropogenic disturbances. If component metrics had potential maximum values
>1, their ranges were scaled to range from 0 to 1 by dividing by their respective maximum
53 NLA 2012 Technical Report. October Version 1.1
RDis IX =
(Eq 5)
2
where:
-------
values based on the NLA 2007 data (see Table 3 in Kaufmann et al. 2014a). Each variant of the
final index was calculated as the mean of its component metric values. Index values range from
0 (indicating no vegetative cover at any station) to 1 (40 to 100 % cover in multiple layers at all
stations).
r rviWoody
2.5
RVegQ_ 2 =
RVegQ _1 =
RVegQ _8 =
+ rvfcGndlnundated
rviLowWood
1.75
+ rvfcGndlnundated
rviWoody
2.5
+ rvfpCanBig + rvfcGndlnundated + ssiNATBedBld
(Eq 6)
(Eq 7)
(Eq 8)
where:
rviWoody = Sum of the mean areal cover of woody vegetation in 3 layers: canopy (large and
small diameter trees), understory, and ground layers (rvfcCanBig + rvfcCanSmall +
rvfcUndWoody + rvfcGndWoody).
rviLowWood = Sum of mean areal cover of woody vegetation in the understory and ground
cover layers (rvfcUndWoody + rvfcGndWoody).
rvfcGndlnundated = Mean areal cover of inundated terrestrial or wetland vegetation in the
ground cover layer.
rvfpCanBig = Proportion of stations with large diameter (>0.3 m dbh) trees present.
ssiNATBedBld = Sum of mean areal cover of naturally-occurring bedrock and boulders
(ssfcBedrock + sfcBoulders), and where the value of ssiNATBedBld was set to 0 in lakes
that have a substantial amount of human-built seawalls and revetments (i.e., hipwWalls
>0.10).
We used RVegQ^2 for mesic ecoregions with maximum elevations <2,000 m (NAP, SAP, UMW,
CPL) where tree vegetation can be expected in relatively undisturbed locations (Table 5-2).
RVegQ_2 sums the woody cover in three lakeside vegetation layers (rviWoody) and includes
inundated groundcover vegetation (rvfcGndlnundated) as a positive characteristic.
We used RVegQ_7 for Central Plains ecoregions (NPL, SPL and TPL). Whereas perennial woody
groundcover and shrubs can be expected on undisturbed lake shorelines throughout the
Central Plains (West and Ruark 2004), the presence or absence of large trees (>5m high) along
lake margins in this region has ambiguous meaning without floristic information (Johnson 2002,
Barker and Whitman 1988, Huddle et al. 2011). RVegQ_7 accommodates lack of tree canopy in
least-disturbed lakes by summing only the lower 2 layers of woody vegetation (rviLowWood)
and includes inundated ground cover vegetation as a positive characteristic.
We used RVegQ_8 for the West (WMT, XER), where climate ranges from wet to arid, and where
lakeshores may have the potential to grow large diameter riparian trees but may lack vegetated
54 NLA 2012 Technical Report. October Version 1.1
-------
lake shorelines at high elevations, or where rock precludes vegetation (Table 5-2). RVegQ^8
sums the woody cover in 3 lakeside vegetation layers and includes inundated groundcover
vegetation as a positive characteristic; it also includes the proportional presence of large
diameter trees around the lakeshore as a positive characteristic. RVegQ_8 includes natural rock
as an undisturbed riparian cover type to avoid penalizing relatively undisturbed lakes in arid
areas or at high elevations above timberline. For lakes where there is a substantial extent or
abundance of constructed seawalls, dikes, or revetments along the shoreline, the substrate
metric was set at 0.
5.3.5.3 Littoral Cover Complexity Index (LitCvrQ)
This index was based on the station-averages for visual estimates of the areal cover of 10 types
of littoral features, including aquatic macrophytes but excluding human structures, within each
of the 10 littoral plots (see Kaufmann et al. 2014a). Note that littoral metrics used to calculate
LitCvrQ are those with the suffix "_lif, which match exactly the NLA 2007 littoral cover metrics
having no suffix. We calculated 3 variants, for application in different ecoregions (Table 5-2).
Each variant of the index was calculated as the mean of its component metric scores, so index
values range from 0 (no cover present at any station) to 1 (very heavy cover at all 10 stations).
Component metrics with potential maximum values >1 were scaled from 0-1 by dividing by
their respective maximum values in the NLA 2007 dataset.
r fcfcSnag
0.2875
LitCvrQ _b =
fciNatural+
(Eq 9)
LitCvrQ _c =
LitCvrQ d =
fciNatural +
fcfcSnag f amfcFltEmg
v
0.2875
1 515
r SomeNatCvr
v is .
fcfcSnag f amfcFltEmg
0.2875
1 515
(Eq 10)
(Eq 11)
where:
fciNatural = summed areal cover of non-anthropogenic fish cover elements (fcfcBoulders +
fcfcBrush + fcfcLedges + fcfcLivetrees + fcfcOverhang + fcfcSnag + fcfcAquatic).
SomeNatCvr = summed cover of natural fish cover elements excluding snags and aquatic
macrophytes {fcfcBoulders + fcfcBrush + fcfcLedges +fcfcLivetrees + fcfcOverhang).
amfcFltEmg = summed cover of emergent plus floating aquatic macrophytes (amfcEmergent +
amfcFloating).
fcfcAquatic = total cover of aquatic macrophytes of any type.
All three variants of LitCvrQ include an expression of the summed cover of naturally occurring
fish or macroinvertebrate cover elements. Snag cover is recognized as a particularly important
element of littoral habitat complexity (Francis and Schindler 2006, Christensen et al. 1996,
Miranda et al. 2010). Therefore, we included snags as a separate contributing cover component
55
NLA 2012 Technical Report. October Version 1.1
-------
in all three variants of the index, and divided cover metrics by their maximum values in the NLA
2007 data to make the weightings of snag cover equal to those of the other two littoral cover
sums. For LitCvrQ^c and LitCvrQ_d, we increased the emphasis on emergent and floating-leaf
aquatic macrophytes relative to other littoral components in response to their reported
importance as cover and their sensitivity to human disturbances in many lake types and regions
(Radomski and Geoman 2001, Jennings et al. 2003, Merrell et al. 2009, Beck et al. 2013).
We used LitCvrQ^b for lakes in the CPL, which includes many generally shallow, warm, low
conductivity lakes. We used LitCvrQ^c for lakes in the SAP, which are all reservoirs, where
disturbed sites commonly have substantial erosion of clay-rich upland soils, large water level
fluctuations, and bare-soil shorelines. These conditions generate abiotic turbidity that
suppresses submerged macrophytes, thereby diminishing the association of abundant
submerged aquatic macrophytes with anthropogenic nutrient inputs that is typically seen in
other regions. LitCvrQ^c emphasizes floating and emergent aquatic macrophytes in addition to
snags, but still includes submerged aquatic macrophytes along with other aquatic macrophytes
and cover types mfciNatural. LitCvrQ_d excludes submerged aquatic macrophytes, and we
used it in the remaining ecoregions (NAP, TPL, NPL, SPL, WMT, and XER), where submerged
aquatic macrophytes provide valuable cover, but high submerged cover is frequently associated
with anthropogenic eutrophication (Hatzenbeler et al. 2004, Merrell et al. 2009).
5.3.5.4 Littoral-Riparian Habitat Complexity Index (LitRipCvrQ)
We averaged the lake values of the littoral cover complexity and riparian vegetation cover
complexity indices to calculate the littoral-riparian habitat complexity index LitRipCvrQ:
r ^ ^ (R VegQ n + LitCvrQ x)
LitRipCvrQ = ; (Eq 12)
where:
RVegQ_n = variant of the riparian vegetation cover complexity index (n=2, 7 or 8, depending on
ecoregion, Table 5-2.
LitCvrQ_x = variant of littoral cover-complexity index (x = b, c, or d, depending on ecoregion,
Table 5-2.
5.3.5.5 Lake Level Drawdown Index (combined use of bfxVertHeigbt and bfxHorizDist)
We used the mean lake values estimating Lake Level Vertical Fluctuation (bfxVertHeight) in
combination with Lake Level Horizontal Fluctuation (bfxHorizDist) to characterize lake
drawdown and natural lake level declines. These metrics are, respectively, the height (meters)
measured from the present lake level to high water, and the horizontal (lateral) distance in
meters from the lake shore to the high water mark in meters. NLA field crews made these
determinations based on the extent and location of vegetation intolerant to frequent or
prolonged inundation, location of flotsom deposits ("trash racks"), evidence of wave action,
and exposed lake bottom. The lake bottom exposure measured by these methods characterizes
seasonal lake level declines and fluctuations on timescales shorter than that required for
56
NLA 2012 Technical Report. October Version 1.1
-------
disintegration of flotsom at the high water mark, or encroachment of perennial terrestrial
vegetation onto the exposed lake bottom area. In most regions, these measurements should be
adequate to document trends in lake level declines attributable to climate change, water
withdrawals, and reservoir management over a decadal timescale. However, more rigorous
tracking of such trends over longer timescales would require that field crews measure lake
levels in relation to established permanent (monumented) reference elevations and/or staff
gauges at sample lakes.
5.3.6 Deriving expected index values under least-disturbed conditions
We based expectations for bfxVertHeight and bfxHorizDist on "Null Models": the expected
value and its dispersion are represented by the central tendency and distribution of these
variables in regional sets of least-disturbed reference sites. In the CENPL and WEST,
expectations were set separately for natural lakes versus man-made reservoirs.
We used lake-specific predictive regression models to estimate physical habitat expectations
for RVegQ, LitCvrQ, and LitRipCvrQ under least-disturbed condition (Table 5-3). We compared
the performance of these regression models with null models (Table 5-4), for which
expectations were simply the mean of logio-transformed physical habitat index scores among
least-disturbed lakes from each ecoregion. Our motivation for using lake-specific models of
expected ("E") condition was to reduce the variance in physical habitat condition indices (in this
case O/E values of RVegQ, LitCvrQ, and LitRipCvrQ) among least-disturbed reference lakes. Air
temperature, precipitation, soils and lithology can vary greatly across ecoregions, resulting in
corresponding variations in potential natural vegetation among least-disturbed lakes. In turn,
that variation results in differences in the amount and complexity of littoral cover, especially for
those elements derived from riparian vegetation. We derived lake-specific expected values by
modeling the influence of important non-anthropogenic environmental factors in relatively
undisturbed lakes, an approach analogous to that used to predict least-disturbed conditions for
multimetric fish assemblage indices (Esselman et al. 2013, Pont et al. 2006, 2009).
For calculating lake-specific expected (E) values of RVegQ, LitCvrQ, and LitRipCvrQ under least-
disturbed condition, we conducted the multiple linear regression (MLR) modeling in 7
aggregated ecoregions (Table 5-3 and Appendix A). These models were based on least-
disturbed lakes from the combined 2007 and 2012 NLA surveys within each region (Table 5-1).
The lake habitat index MLRs employed one to four predictors from among the following:
Latitude, Longitude, Elevation, ElevXLatitude, ElevXLongitude, Lake surface area, Lake origin
(man-made reservoir or natural lake), near-shore anthropogenic disturbance of all types
(RDis_IX), and near-shore anthropogenic agricultural disturbance (hiiAg). Latitude, longitude,
elevation, and ecoregion are surrogates for temperature, precipitation, soil, and other
characteristics that influence potential natural vegetation and littoral cover. Field
measurements of bfxVertHeight and bfxHorizDist were good predictors of riparian and littoral
cover in most of the regions. However, we chose not to use these indicators of level fluctuation
and drawdown to predict expected condition because their use would confound interpretations
and obscure the effects of drawdown on habitat condition. We also did not use lake depth
57 NLA 2012 Technical Report. October Version 1.1
-------
measurements (like maximum depth or littoral mean depth, because of their association with
lake level change. Similarly, survey year was a good predictor of lake physical habitat metrics in
regions where there were marked differences in the amount of lake drawdown between
surveys. We chose not to use survey year as a predictor of expected condition because it would
confound analysis of temporal trends and change between surveys.
Ideally, calculations of expected cover and complexity would be based only on minimally-
disturbed lakes. However, the least-disturbed lakes in most regions include sometimes
substantial disturbances, necessitating inclusion of near-shore disturbance predictors in our
models if they were associated with variance in the habitat indices. The use of RDis_IX or hiiAg
as predictors was supported by the data for all three habitat indicators in the NPL, CPL and
CENPL, and the littoral cover indicator in the SAP (Table 5-3). For predicting expected LitCvrQ
and LitRipCvrQ in the NAP, we had to combine least-disturbed with moderately disturbed lakes
and reservoirs (RT_NLA12_2015 = R or S) to span lake size and elevation gradients affecting
riparian vegetation and littoral cover in that region. The weak association of human disturbance
with habitat indices would not have warranted including RDis_IXas a predictor within NAP least
disturbed sites alone (RT_NLA12_2015=R). However, the human disturbance gradient
introduced by including moderately disturbed NAP lakes (RT_NLA12_2015=S), and the effect of
that disturbance on littoral habitat in the NAP made it necessary to include RDis_IX as a
predictor. Inclusion of RDis_IX or hiiAg as predictors of expected lake habitat index values was
not supported by the data for lakes and reservoirs in the UMW, WMT, and XER. As in most of
the other regions, lake level fluctuation indicators were good predictors of riparian and littoral
cover in the UMW and WEST, but were not used as predictors for reasons we stated in the
previous paragraph.
For regions where RDis_IX or hiiAg were used in modeling expected habitat condition, we set
the value of these variables in the predictive MLR equation to the minimum value observed in
the region before calculating expected values of RVegQ, LitCvrQ, and LitRipCvrQ. In all regions
and subregions there were sites with RDisJX and hiiAg values of 0 (See Appendix A). Setting the
reference expected lake habitat index values slightly higher in this way results in the central
tendency for reference site O/E to be less than 1.0.
5.3.7 Condition Criteria for Nearshore Lake Physical habitat
For the lakeshore anthropogenic disturbance index RDisJX, we used uniform criteria for all
lakes. For RVegQ, LitCvrQ, and LitRipCvQ we set condition criteria based on the distribution of
O/E values of these indices observed in least-disturbed lakes. For bfxVertHeight and
bfxHorizDist, we set condition criteria based on the distribution of the metric values themselves
in least-disturbed lakes (Null model).
5.3.7.1 Condition Criteria for Lakeshore Anthropogenic Disturbance Intensity and Extent
Because RDisJX is a direct measure of human activities, we based criteria for high, medium,
and low levels of disturbance on judgment:
58
NLA 2012 Technical Report. October Version 1.1
-------
Good (Low Disturbance): RDis_IX <0.20
Fair (Medium Disturbance): RDis_IX>0.20 but < 0.75
Poor (High Disturbance): RDis_IX >0.75
Lakes with RDis_IX <0.20 have very low levels of lake and near-lake disturbance, typically having
anthropogenic disturbance on <8% of their shorelines. Those with RDis_IX >0.75 have very high
levels of disturbance, typically having human activities evident on 100% of their shorelines. For
perspective, <21% of the 2364 sample site visits in the combined 2007 and 2012 NLA surveys
had RDis_IX <0.20, and <21% had RDis_IX >0.75. Most of the reference sites in the WMT, UMW,
and NAP regions have RDis_IX <0.20, most of those in SAP, SAP, XER, TPL, and CPL have RDis_IX
<0.40, most NAP reference sites have RDis_IX between 0.40 and 0.6, and no reference sites
have RDis_IX >0.70 (Figure 5-3).
5.3.7.2 Condition Criteria for RVegQ, LitCvrQ, and LitRipCvQ
We calculated physical habitat index observed/expected (O/E) values of RVegQ_OE,
LitCvrQ^OE, and LitRipCvQ_OE for each sample lake by dividing the observed index value at
each lake by the lake-specific expected value derived from regressions in Table 5-3 and
Appendix A. The calculated O/E values of the habitat metrics for each lake express the degree
of deviation of that lake from an estimate of its expected value under least-disturbed
conditions. No model perfectly predicts expected indicator values (E-values) in lakes under
least-disturbed conditions, and field measurements of indicator values ("O" values) include
error and temporal variation. Consequently, O/E values of these indices among reference lakes
have a dispersion (variance) that decreases with the performance of predictive models (i.e.,
how precisely does the model predict reference condition?), and with the precision of the
habitat indicator measurements (i.e., how well do the field methods measure observed
condition?). We set condition criteria for RVegQ, LitCvrQ, and LitRipCvQ with reference to the
distributions of these indices among least-disturbed lakes within each of the 7 merged
ecoregions Table 5-5.
The small number of lakes meeting our low-disturbance criteria in most regions precluded
obtaining reliable percentiles of RVegQ, LitCvrQ, and LitRipCvQ directly from the least-disturbed
lake distributions. Consequently, for all regions, we used the central tendency and variance of
index O/E values in least-disturbed lakes values to model their distributions and to estimate
percentiles (Snedecor and Cochran 1980). The logio-transformed O/E values in the least-
disturbed lakes had symmetrical, approximately normal distributions. We calculated means and
standard deviations of logio-transformed O/E values (Table 5-5, columns 3 and 4), and
estimated the 5th and 25th percentiles (Table 5-5, columns 7 and 8) based on the log-normal
approximation of the index distributions in least-disturbed lakes within each ecoregion.
Because the means and SD's are all log values, a range of + 1SD would be calculated, for
example, by multiplying and dividing the geometric mean by the geometric SD (see Table 5-5
legend for details, including handling of the log-transformation constant).
59
NLA 2012 Technical Report. October Version 1.1
-------
Lakes with 0/E values (MLR model) that are >25th percentile for least-disturbed lakes within
their regions were considered to have habitat in good condition (i.e., similar to that in the
population of least-disturbed lakes of the region). Similarly, lakes with index or O/E values <5th
percentile of least-disturbed lakes were considered to have poor habitat quality (i.e., they have
significantly lower cover and complexity than observed within the sub-population of least-
disturbed lakes of the region). Those with index or O/E values between the 5th and 25th
percentiles of least-disturbed lakes were scored as fair condition.
We emphasize that our designations of good, fair and poor are relative to the least disturbed
sites available in each ecoregion. We define good condition as habitat quality not
distinguishable from the distribution of habitat in least-disturbed sites; and poor condition as
habitat quality that is not likely to be found within the distribution of least-disturbed sites of
the ecoregion. Our designations of poor condition do not indicate impaired water body status.
Conversely, our designations of good condition mean that habitat is similar to the least-
disturbed sites available in a region, which does not mean pristine, only the best available,
which can be relatively disturbed in extensively and highly disturbed regions.
5.3.7.3 Condition Criteria for Lake Drawdown
We based our assessment of Lake Drawdown condition on null models of the expected amount
of drawdown in least disturbed lakes. Specifically, we examined the empirical distributions of
the metrics quantifying vertical and horizontal lake level fluctuations {bfxVertHeight and
bfxHorizDist) in least disturbed lakes within aggregated ecoregions, sometimes stratified by lake
origin (natural lakes versus man-made reservoirs). We used separate null models for the NAP,
SAP, UMW, and CPL regions. For the CENPL (TPL+SPL+NPL) and the West (WMT+XER), we used
separate null models for natural lakes versus man-made reservoirs. Vertical and horizontal
drawdown were considered small if they were <75th percentile of their respective reference
distributions; large if >95th percentile, and medium if in-between (Table 5-6). Overall lake
drawdown condition was considered small if both vertical and horizontal drawdown were
small; medium if one or both were medium (but not large); and large if vertical, horizontal or
both were large.
5.4 Least-disturbed reference distributions and regressions (from sections
5.3.6 and 5.3.7)
5.4.1 Disturbance within least-disturbed reference sites
Near shore human disturbance indexed by RDis_IX varied considerably among least-disturbed
reference sites, and among regions. Reference site RDis_IX was lowest in the WMT and UMW,
intermediate in the NAP, then steadily increasing through SAP, SPL, XER, TPL and CPL to their
highest values in the NPL (Figure 5-2). The level of RDis_IX among all sites within regions did
not cleanly follow their ordering by increasing reference site RDis_IX. For example, the UMW
60
NLA 2012 Technical Report. October Version 1.1
-------
reference sites had very low RDis_IX in relation to the general level of RDis_IX in that region
(Figure 5-2). Conversely, RDis_IX in reference sites of the NPL did not greatly differ from the
distribution of rather high RDis_IX for sites in general within that region.
5.4.2 Null Model Results for RVegQ, LitCvrQ, and LitRipCvQ:
Geometric means for RVegQ, LitCvrQ, and LitRipCvQ in least-disturbed lakes differed among
regions (Table 5-4), but these unsealed null model values are not directly comparable because
the habitat index formulations differed among regions. The RVegQ, LitCvrQ, and LitRipCvQ null-
model logSD's and geometric SD's (Columns 4 and 6 of Table 5-4) were calculated from log-
transformed variables, and therefore are expressions of the proportional variance among least-
disturbed lakes of each region. Whether scaled (divided by the mean) or not, they are directly
comparable as measures of model precision among regions with different geometric means, or
between null and MLR modeling approaches.
Comparing indicators, the precision in modeling least-disturbed condition using null models was
generally better (smaller SDs) for LitRipCvQ than for RVegQ or LitCvrQ, and null models for
RVegQ were generally more precise than for LitCvrQ (Table 5-4, columns 4 and 6). The most
obvious differences, however, were among regions, and the differences were associated with
the level of disturbance in the reference sites. We ordered the seven NLA lake habitat modeling
ecoregions according to increasing reference site median RDisJX for examining variance in the
other lake habitat indicators (Figure 5-3). The regions with the greatest amount of disturbance
in their reference sites (the CENPL, including NPL, SPL, TPL, the CPL, and the XER) generally had
higher within-reference site variance all three lake habitat indices, with the exception of low
variance in all three indicators within reference sites of the relatively high-disturbance CPL
reference sites (Figure 5-4). The precision in modeling least-disturbed condition using null
models was generally best in the UM W and NAP (i.e., lowest gSDs). The smaller the SD of index
values (or O/E values) among least-disturbed lakes, the easier it is to confidently distinguish
disturbed lakes from least-disturbed lakes. The null model SD's serve as an upper bound for the
variance of the indicators among regional reference sites, and are analogous to the RMSE's of
the regressions in Table 5-3. Removing the variance attributed to the predictors reduces the
unexplained variance among reference sites.
5.4.3 O/E Model Results for RVegQ, LitCvrQ, and LitRipCvQ:
The LogSD's of RVegQ^OE, LitCvrQ^OE, and LitRipCvQ^OE among reference sites (Table 5-5,
column 4) were consistently, and in some cases substantially, lower than those for null models
in their respective regions, as evidenced by comparing open circles and black dots plotted in
Figure 5-4. The CPL, CENPL, XER and WMT showed the largest reduction of reference site
variance compared with corresponding null models, denoting improvement in O/E model
performance over null models. As for the null models, however, O/E models in regions with
relatively disturbed reference sites had higher reference site variance (the expected condition
models were less precise). Again, with the exception of the CPL, regions with more disturbance
in their reference sites still had higher SD's than those in regions with less disturbance.
61 NLA 2012 Technical Report. October Version 1.1
-------
Conversely, the four regions with the lowest level of human disturbance in their reference sites
(WMT, UMW, NAP, and SAP) also had the lowest O/E model variance among their reference
sites. These results reinforce the idea that human disturbances are likely responsible for a large
amount of the variance in lake physical habitat structure in reference sites within the disturbed
regions. Therefore, further effort to capture this variance by modeling only non-anthropogenic
("natural") controls would not likely be successful in reducing the variance in O/E values among
reference sites.
Except for regions where O/E models incorporated human disturbance variables (NAP, CPL,
CENPL and LitCvr_OE in SAP), the central tendency of reference site O/E values (Table 5-5,
column 6) was very close to 1 (0.98 to 1.01). This is to be expected. Where E-Models contained
human disturbance predictors, reference O/E values regained the variance modeled out when
observed values were divided by expected values determined with human disturbance
predictors (RDis_IX or hiiAg) set to regional minimum values. If human disturbances decrease
the observed value, the mean O/E will be <1. Accordingly, reference site mean O/E values for
MLR Models in the NAP, CPL, and CPL (and LitCvr_OE in SAP) ranged from 0.79 to 0.91. We
regressed the reference O/E values against the RDis_IX or hiiAg values to obtain y-intercepts for
expected O/E for the minimum disturbance observed in these regions. These are shown in the
Table 5-5 rows with "oe Yint" subscripted after their Ecoregion designation. For example the
NAPoEYint row is the result of this final adjustment on reference O/E results from the NAPMLRModei
row.
Anthropogenic disturbance among reference sites tends to increase the variance in O/E values
within regions, even after the minimum disturbance adjustment. There is a strong relationship
between the LogSDs of null and adjusted O/E models for lake habitat among reference lakes
and the regional level of near-shore anthropogenic disturbance in reference sites (Figure 5-4).
Our modeling improves these models, but it is likely that disturbances other than those
captured by RDis_IX contribute to the uncertainty in predicting habitat characteristics in
minimally-disturbed lakes. These results reinforce the idea that human disturbances are likely
responsible for a large amount of the variance in lake physical habitat structure among least-
disturbed reference sites in the disturbed regions. Therefore, further effort to capture this
variance by modeling only non-anthropogenic ("natural") controls would not likely be
successful in reducing the variance in O/E values among reference sites.
5.4.4 Null Model Results for Lake Drawdov ' Level Fluctuations:
Least-disturbed reference lakes and reservoirs in the NAP, SAP and UMW experienced less
drawdown and level fluctuation than those in the CPL, CENPL, and WEST; particularly in
comparison with marked drawdown observed in man-made reservoirs of the CENPL and WEST
(Table 5-6). Not surprisingly, least-disturbed natural lakes in the CENPL and WEST also
experienced less drawdown and level fluctuation than their human-constructed counterparts.
As a result, the criteria for assessing substantial drawdown in lakes of the Appalachians and
UMW were much smaller than those for lakes (and particularly reservoirs) in the CENPL and
WEST.
62 NLA 2012 Technical Report. October Version 1.1
-------
5.5 Precision of physical habitat indicators
In our synoptic survey context, o2iake is the signal of interest, and o2rep is noise variance; we
define their ratio as S/N. The methods we used to quantify precision, the precision of NLA lake
physical habitat metrics and key habitat condition indices, and the implications of varying
precision levels for monitoring and assessment, are comprehensively evaluated by Kaufmann et
al. (1999, 2014a). Here we summarize findings for key physical habitat indicators based on the
NLA 2012 survey data.
The key NLA physical habitat indices had moderate to high S/N (2.2 - 11.0) over the entire NLA
2012 survey (Table 5-7). Compared with the other composite indices, the human disturbance
index RDis_IX and horizontal drawdown index had the highest S/N (9.1-11), whereas the littoral
cover O/E index had the lowest S/N (2.2). The advantage of S/N as a precision measure is its
relevance to many types of statistical analysis and detecting differences in subpopulation
means (Zar 1999). High noise in habitat descriptions relative to the signal (i.e., low signal: noise
ratio, S/N) diminishes statistical power to detect differences among lakes or groups of lakes.
Imprecise data limit the ability to detect temporal trends (Larsen et al. 2001, 2004). Noise
variance also limits the maximum amount of variance that can be explained by models such as
multiple linear regression (Van Sickle et al. 2005, Kaufmann and Hughes 2006). By reducing the
ability to quantify associations between variables (Allen et al. 1999, Kaufmann et al. 1999),
imprecision compromises the usefulness of habitat data for discerning likely controls on biota
and diagnosing probable causes of impairment. The adverse effects of noise variance on these
types of analysis are negligible when S/N >10; becoming minor as S/N decreases to 6, increasing
to moderate as S/N decreases to 2, and finally becoming severely limiting as S/N approaches 0
(Paulsen et al. 1991, Kaufmann et al. 1999). At S/N=0, all the metric variance observed among
lakes in the survey can be attributed to measurement "noise". Based on these guidelines, the
effects of imprecision are minor for all the indicators except for the Littoral Cover index, for
which the effects are minor-to-moderate.
Kaufmann et al. (2014a) explain that the S/N ratio may not always be a good measure of the
potential of a given metric to discern ecologically important differences among sites. For
example, a metric may easily discriminate between sparse and abundant littoral cover for fish,
but S/N for the metric would be low in a region where littoral cover does not vary greatly
among lakes. In cases where the signal variance (o2iake) observed in a regional survey reflects a
large range of habitat alteration or a large range in natural habitat conditions, S/N would be a
good measure of the precision of a metric relative to what we want it to measure. However, in
random surveys or in relatively homogeneous regions, o2iake and consequently S/N, may be less
than would be calculated for a set of sites specifically chosen to span the full range of habitat
conditions occurring in a region. To evaluate the potential usefulness of metrics, Kaufmann et
al. (2014a) suggested that an alternate measure of relative precision, orep divided by its
potential or observed range (Rgpot or Rg0bs) offers additional insight. The minimum detectable
difference in means between 2 lakes (or between two times in one lake) is given by Dmin =
63
NLA 2012 Technical Report. October Version 1.1
-------
1.96CTrep(2n)1/2 = 2.Horep, using a 2-sided Z-test with a = 0.05 (Zar 1999). Thus, to detect any
specified difference between 2 lakes in a metric relative to its potential or observed range (Rgpot
or Rg0bs, the standardized within-lake standard deviation, orep/Rg, cannot exceed (Dmin/
Rg)/2.71. By the criteria in Kaufmann et al. (2014a - Table 2), the key NLA physical habitat
indices were precise or moderately precise, with orep/Rgobs between 0.052 - 0.107 (Table 5-7).
Depending on the index, they have the potential to discern differences between single lakes (or
one lake at two different times) that are between l/3rd and l/8th the magnitude of the
observed ranges of these indices.
5.6 Physical habitat index responses : > * thropogenic disturbance
In the U.S. as a whole, RVegQ_OE, LitCvrQ^OE, and LitRipCvQ_OE were significantly higher
(p<0.0001) in least-disturbed lakes (RT_NLA12_2015=R) than in highly-disturbed lakes
(RT_NLA12_2015=T) (Table 5-8, Figure 5-5). The differences were substantial for RVegQ^OE,
and LitRipCvQ_OE, and discrimination was good (no or nearly no overlap in interquartile
ranges). For LitCvrQ^OE, there was an overlap of approximately one-third of the interquartile
range. RDisJX was a major screening variable used to disqualify potential reference sites, so it
is not surprising that the entire range of RDisJX among reference sites had very little overlap
with that for highly disturbed sites. Note that a site with very low RDisJX could be classified as
highly-disturbed on the basis of many other variables, but the converse is not true because
reference sites must all have low RDisJX. Like RDisJX, both vertical and horizontal drawdown
were significantly lower (p<0.0001) in least-disturbed lakes than in highly-disturbed lakes (Table
5-8, Figure 5-5). Except for lake drawdown, contrasts were very similar for the 2007 and 2012
NLA surveys (Figure 5-6). Although the t test between reference and highly disturbed lakes was
similar in both years, the positive relationship between disturbance and in lake level drawdown
was much less evident in the drier year (2007) than in 2012. In 2012 fewer than 5% of reference
lakes showed any drawdown at all, whereas 75 to 95 % of reference lakes showed drawdown in
2007 - with a lot of overlap in the inter-quartile ranges of reference and highly disturbed sites.
RVegQ_OE, LitCvrQ^OE, and LitRipCvQ_OE in sub-sets and sub-regions of the U.S. universally
showed the same pattern of response as the nation, with the mean of reference sites
significantly greater than those for highly-disturbed sites (Table 5-9). Discrimination was
generally greater for RVegQ_OE and LitRipCvQ_OE than for LitCvrQ_OE or the drawdown
indices. Discrimination of these 3 indices was somewhat greater for natural lakes than for
reservoirs, but good in both. RVegQ_OE was strongly and clearly associated with disturbance
(RT_NLA12) in all regions and years except for NPL, and SPL in the NLA 2007 survey year.
LitCvrQ_OE was strongly related to disturbance class in the CPL and NPL, moderately related to
disturbance in the NAP, TPL (2012), SPL, and XER; and associations were with disturbance were
weakest in the SAP, WMT, and TPL (2007). LitRipCvQ_OE was strongly and clearly associated
with disturbance (RT_NLA12) in all regions and both years.
64
NLA 2012 Technical Report. October Version 1.1
-------
5.7 Discussion
The NLA and other lake survey and monitoring efforts increasingly rely upon biological
assemblage data to define lake condition. Information concerning the multiple dimensions of
physical and chemical habitat is necessary to interpret this biological information and
meaningfully assess ecological condition. The controlling influence of littoral structure and
complexity on lake biota has been long recognized, and recent research highlights the roles of
habitat structural components like littoral woody debris in providing refuges from predation
and affecting nutrient cycling and littoral production. NLA field crews characterized lake depth,
water surface characteristics, bank morphology and evidence of lake level fluctuations, littoral
and shoreline substrate, fish concealment features, aquatic macrophytes, riparian vegetation
cover and structure, and human land use activities. These littoral and riparian physical habitat
measurements and visual observations were made in a randomized array of 10 near-shore
littoral-riparian plots systematically spaced along the shoreline of each sample lake. Metrics
describing a rich variety of lake characteristics were calculated from this raw data, and many of
these were determined with moderate precision in the national dataset. For the NLA, we
summarize this information with four integrative measures of lake condition, and one measure
of lake drawdown and lake level fluctuation: RDis_IX, incorporating measures of the extent and
intensity of near-shore human land and water use activities; RVegQ, incorporating the structure
and cover in three layers of riparian vegetation, including inundated vegetation; LitCvrQ, a
combined biotic cover complexity measure including large woody snags, brush, overhanging
vegetation, aquatic macrophytes, boulders, and rock ledges; and LitRipCvrQ, which combines
RipVegQ and LitCvrQ. The measure of lake level drawdown incorporates both horizontal and
vertical fluctuation, comparing them to the regional mean values observed in least-disturbed
lakes and reservoirs.
We modeled expected values of RVegQ, LitCvrQ, and LitRipCvrQ and their divergence from
reference conditions in least-disturbed lakes using regression-based O/E models. The precision
of these O/E indices was moderate to high, and showed good discrimination between least-
disturbed and highly-disturbed lakes nationally, and within ecoregions. These results show that,
compared with least-disturbed reference lakes, those with moderate or high human
disturbances in the same region have reduced cover and extent of multi-layered riparian
vegetation or natural wetlands. In addition, those with moderate or high disturbance generally
also have reduced snag, brush and emergent aquatic macrophyte cover. These results
complement the results of the NLA 2012 Assessment report and those of Kaufmann et al.
2014b, 2014c), confirming our general expectation that near-shore wetland and multi-layered
riparian vegetation and abundant, complex fish concealment features foster native fish,
macroinvertebrate, zooplankton, and avian assemblage integrity, whereas extensive and
intensive shoreline human activities that reduce natural riparian vegetation and reduce littoral
cover complexity are detrimental to these biotic assemblages.
We believe that the metrics and indices derived from the NLA physical habitat field approach
and the O/E indices expressing their divergence from least-disturbed reference conditions
65
NLA 2012 Technical Report. October Version 1.1
-------
describe ecologically-relevant characteristics of lake habitat with sufficient precision to evaluate
near-shore lake habitat structure in national, state, and ecoregional assessments. Their
association with gradients of human disturbance demonstrates that they also describe lake
attributes that are vulnerable to anthropogenic degradation and potential for productive
restoration through lake and land management.
66
NLA 2012 Technical Report. October Version 1.1
-------
5.8 Literature cited
Allen, A. P., T. R. Whittier, P. R. Kaufmann, D. P. Larsen, R. J. O'Connor, R. M. Hughes, R. S.
Stemberger, S. S. Dixit, R. 0. Brinkhurst, A. T. Herlihy, and S. G. Paulsen. 1999.
Concordance of taxonomic composition patterns across multiple lake assemblages: effects
of scale, body size, and land use. Canadian Journal of Fisheries and Aquatic Sciences 56:
2029-2040.
Barker, W. T., and W. C. Whitman. 1988. Vegetation of the Northern Great Plains. Rangelands
10:266-272.
Beck, M. W., B. Vondracek, and L. K. Hatch. 2013. Between- and within-lake responses of
macrophyte richness metrics to shoreline development. Lake and Reservoir Management
29:179-193.
Benson, B. J., and J. J. Magnuson. 1992. Spatial heterogeneity of littoral fish assemblages in
lakes: relation to species diversity and habitat structure. Canadian Journal of Fisheries and
Aquatic Sciences 49:1493-1500.
Brauns, M., X-F Garcia, N. Walz, and M. T. Pusch. 2007. Effects of human shoreline development
on littoral macroinvertebrates in lowland lakes. Journal of Applied Ecology 44:1138-1144.
Carpenter, S. R., and K. L. Cottingham. 1997. Resilience and restoration of lakes. Conservation
Ecology [online] l(l):w. Available from the Internet. URL:
http://www.consecol.org/voll/issl/art2.
Christensen, D. L., B. R. Herwig, D. E. Schindler, and S. R. Carpenter. 1996. Impacts of lakeshore
residential development on coarse woody debris in north temperate lakes. Ecological
Applications 6:1143-1149.
Engel, S., and J. Pederson. 1998. The construction, aesthetics, and effects of lakeshore
development: a literature review. Wisconsin. Dept. of Natural Resources, Report 177.
Madison, Wisconsin http://digital.library.wisc.edu/1711.dl/EcoNatRes.DNRRepl77.
Esselman, P. C., D. M. Infante, L. Wang, A. R. Cooper, D. Wieferich, Y-P. Tsang, D. J. Thornbrugh,
and W. W. Taylor. 2013. Regional fish community indicators of landscape disturbance to
catchments of the conterminous United States. Ecological Indicators 26:163-173.
Faustini, J. M., and P. R. Kaufmann. 2007. Adequacy of visually classified particle count statistics
from regional stream habitat surveys. J. Am.Water Resourc. Assoc. 43(5):1293-1315.
67
NLA 2012 Technical Report. October Version 1.1
-------
Francis, T. B., and D. E. Schindler. 2006. Degradation of littoral habitats by residential
development: woody debris in lakes of the Pacific Northwest and Midwest, United States.
Ambio 35:274-280.
Halliwell, D. 2007. Lake habitat measures. New England Lake News - New England Chapter of
the North American Lake Management Society 2(2):l-4.
Hampton, S. E., S. C. Fradkin, P. R. Leavitt, and E. E. Rosenberger. 2011. Disproportionate
importance of nearshore habitat for the food web of a deep oligotrophic lake. Marine and
Freshwater Research 62:350-358.
Hatzenbeler, G. R., J. M. Kampa, M. J. Jennings, and E. E. Emmons. 2004. A comparison offish
and aquatic plant assemblages to assess ecological health of small Wisconsin lakes. Lake
and Reservoir Management 20:211-218.
Herlihy, A. T., N. C. Kamman, J. C. Sifneos, D. Charles, M. D. Enache, and R. J. Stevenson. 2013.
Using multiple approaches to develop nutrient criteria for lakes in the conterminous USA.
Freshwater Science 32(2):361-384.
Huddle, J. A., T. Awada, D. L. Martin, X. Zhou, S. E. Pegg, and S. J Josiah. 2011. Do invasive
riparian woody plants affect hydrology and ecosystem processes? Papers in Natural
Resources. Paper 298. Digital Commons at University of Nebraska - Lincoln.
http://d igitalcommons.unl.edu/natrespapers/298.
Jennings, M. J., M. A. Bozek, G. R. Hatzenbeler, E. E. Emmons, and M. D. Staggs. 1999.
Cumulative effects of incremental shoreline habitat modification on fish assemblages in
north temperate lakes. North American Journal of Fisheries Management 19:18-27.
Jennings, M. J., E. E. Emmons, G. R. Hatzenbeler, C. Edwards, and M. A. Bozek. 2003. Is littoral
habitat affected by residential development and land use in watersheds of Wisconsin
lakes? Lake and Reservoir Management 19:272-279.
Johnson, W. C. 2002. Riparian vegetation diversity along regulated rivers: contribution of novel
and relict habitats. Freshwater Biology 47: 749-759.
Kaufmann, P. R., and R. M Hughes. 2006. Geomorphic and anthropogenic influences on fish and
amphibians in Pacific Northwest coastal streams. Pages 429-455 in Hughes RM, Wang L,
Seelbach PW (editors). Landscape influences on stream habitat and biological
assemblages. American Fisheries Society Symposium 48, Bethesda, Maryland.
Kaufmann, P. R., and T. R. Whittier. 1997. Habitat Assessment. Pages 5-1 to 5-26 In: J.R. Baker,
D.V. Peck, and D.W. Sutton (Eds.). Environmental Monitoring and Assessment Program -
Surface Waters: Field Operations Manual for Lakes.EPA/620/R-97/001. U.S.
Environmental Protection Agency, Washington, D.C.
68 NLA 2012 Technical Report. October Version 1.1
-------
Kaufmann, P. R., R. M. Hughes, J. Van Sickle, T. R. Whittier, C. W. Seeliger, and S. G. Paulsen.
2014a. Lake shore and littoral habitat structure: A field survey method and its precision.
Lake and Reservoir Management. 30:157-176.
Kaufmann, P. R., R. M. Hughes, T. R. Whittier, S. A. Bryce, and S. G. Paulsen. 2014b. Relevance
of lake physical habitat assessment indices to fish and riparian birds. Lake and Reservoir
Management. 30:177-191.
Kaufmann, P. R., D. V. Peck, S. G. Paulsen, C. W. Seeliger, R. M. Hughes, T. R. Whittier, and N. C.
Kamman. 2014c. Lakeshore and littoral physical habitat structure in a national lakes
assessment. Lake and Reservoir Management. 30:192-215.
Kaufmann, P. R., D. P. Larsen, and J. M. Faustini. 2009. Bed stability and sedimentation
associated with human disturbances in Pacific Northwest streams. Journal of the American
Water Resources Association 45:434-459.
Kaufmann, P. R., P. Levine, E. G. Robison, C. Seeliger, and D. V. Peck. 1999. Quantifying Physical
Habitat in Wadeable Streams. EPA 620/R-99/003. Environmental Monitoring and
Assessment Program, U.S. Environmental Protection Agency, Corvallis, OR.
Kovalenko, K. E., S. M. Thomaz, D. M. Warfe. 2012. Habitat complexity: approaches and future
directions. Hydrobiologia 685:1-17.
Larsen, D. P., and S. J. Christie (editors). 1993. EMAP-Surface Waters 1991 Pilot Report.
EPA/620/R-93/003. U.S. Environmental Protection Agency, Office of Research and
Development, Washington, D.C.
Larsen, D. P., T. M. Kinkaid, S. E. Jacobs, and N. S. Urquhart. 2001. Designs for evaluating local
and regional scale trends. Bioscience 51(12):1069-1078.
Larsen, D. P., P. R. Kaufmann, T. M. Kincaid, and N. S. Urquhart. 2004. Detecting persistent
change in the habitat of salmon-bearing streams in the Pacific Northwest. Canadian
Journal of Fisheries and Aquatic Sciences 61:283-291.
Merrell, K., E. Howe, and S. Warren. 2009. Examining shorelines, littorally. Lakeline 29:1.
Miranda, L. R., M. Spickard, T. Dunn, K. M. Webb, J. N. Aycock, and K. Hunt. 2010. Fish habitat
degradation in U.S. reservoirs. Fisheries 35(4):175-184.
Omernik, J.M. 1987. Ecoregions of the conterminous United States. Annals of the Association of
American Geographers 77:118-125.
69
NLA 2012 Technical Report. October Version 1.1
-------
Palmer, M. A., A. P. Covich, S. Lake, P. Biro, J. J. Brooks, J. Cole, C. Dahm, J. Gibert, W.
Goedkoop, K. Martens, J. Verhoeven, and W. J. Van de Bund. 2000. Linkages between
aquatic sediment biota and life above sediments as potential drivers of biodiversity and
ecological processes. Bioscience 50:1062-1075.
Paulsen, S. G., D. P. Larsen, P. R. Kaufmann, T. R. Whittier, J. R. Baker, D. V. Peck, J. Mcgue, D.
Stevens, J. Stoddard, R. M. Hughes, D. McMullen, J. Lazorchak, and W. Kinney. 1991.
Environmental Monitoring and Assessment Program (EMAP) Surface Waters monitoring
and research strategy. U.S. Environmental Protection Agency, Corvallis, Oregon.
Paulsen, S. G., A. Mayio, D. V. Peck, J. L. Stoddard, E. Tarquinio, S. M. Holdsworth, J. Van Sickle,
L. L. Yuan, C. P. Hawkins, A. T. Herlihy, P. R. Kaufmann, M. T. Barbour, D. P. Larsen, and A.
R. Olsen. 2008. Condition of stream ecosystems in the US: an overview of the first national
assessment. Journal of the North American Benthological Society 27:812-821.
Peck, D. V., A. R. Olsen, M. H. Weber, S. G. Paulsen, C. Peterson, and S. M. Holdsworth. 2013.
Survey design and extent estimates for the National Lakes Assessment. Freshwater
Science 32:1231-1245.
Polis, G. A., W. B. Anderson, and R. D. Holt. 1997. Toward an integration of landscape and food
web ecology: the dynamics of spatially subsidized food webs. Annual Review of Ecology
and Systematics 28:289-316.
Pont, D., B. Hugueny, U. Beier, D. Goffaux, A. Melcher, R. Noble, C. Rogers, N. Roset, and S.
Schmutz. 2006. Assessing river biotic condition at a continental scale: a European
approach using functional metrics and fish assemblages. Journal of Applied Ecology.
43:70-80.
Pont, D., R. M. Hughes, T. R. Whittier, and S. Schmutz. 2009. A predictive index of biotic
integrity model for aquatic-vertebrate assemblages of western U.S. streams. Transactions
of the American Fisheries Society 138:292-305.
Radomski, P., and T. J. Geoman. 2001. Consequences of human lakeshore development on
emergent and floating-leaf vegetation abundance. North American Journal of Fisheries
Management 21:41-46.
Schindler, D. E., and M. D. Scheuerell. 2002. Habitat coupling in lake ecosystems. Oikos 98:177-
189.
Snedecor, G. W., and W. G. Cochran. 1980. Statistical Methods, seventh edition. The Iowa State
University Press, Ames, Iowa, U.S.A. 507 pp.
Strayer, D. L., and S. E. G. Findlay. 2010. Ecology of freshwater shore zones. Aquatic Science
72:127-163.
70 NLA 2012 Technical Report. October Version 1.1
-------
Taillon, D., and M. Fox. 2004. The influence of residential and cottage development on littoral
zone fish communities in a mesotrophic north temperate lake. Environmental Biology of
Fishes 71: 275-285.
USEPA. 2007a. Survey of the Nation's Lakes. Field Operations Manual. EPA 841-B-07-004. U.S.
Environmental Protection Agency, Washington, DC.
USEPA. 2009. National Lakes Assessment: A Collaborative Survey of the Nation's Lakes. EPA
841-R-09-001. U.S. Environmental Protection Agency, Office of Water and Office of
Research and Development, Washington, D.C.
USEPA. 2010. National Lakes Assessment: Technical Appendix. EPA 841-R-09-001a. U.S.
Environmental Protection Agency, Office of Water and Office of Research and
Development, Washington, D.C. 63pp.
USEPA. 2012. 2012 National Lakes Assessment: Field Operations Manual. EPA 841-B-11-003.
U.S. Environmental Protection Agency, Washington, DC.
USEPA. 2016. National Lakes Assessment 2012: A Collaborative Survey of Lakes in the United
States. EPA 841-R-16-113. U.S. Environmental Protection Agency, Washington, DC.
Wagner T, Jubar AK, Bremigan MT. 2006. Can habitat alteration and spring angling explain
largemouth bass nest success? Transactions of the American Fisheries Society 135:843-
852.
West, E., and G. Ruark. 2004. Historical evidence of riparian forests in the Great Plains and how
that knowledge can aid with restoration and management. USDA Forest Service / UNL
Faculty Publications. Paper 7. Digital Commons at University of Nebraska - Lincoln.
http://d igitalcommons.unl.edu/usdafsfacpub/7.
Whittier, T. R., D. B. Halliwell, and S. G. Paulsen. 1997. Cyprinid distributions in northeast U.S.A.
lakes: Evidence of regional-scale minnow biodiversity losses. Canadian Journal of Fisheries
and Aquatic Sciences 54:1593-1607.
Whittier, T. R., S. G. Paulsen, D. P. Larsen, S. A. Peterson, A. T. Herlihy, and P. R. Kaufmann.
2002. Indicators of ecological stress and their extent in the population of Northeastern
lakes: a regional-scale assessment. Bioscience 52(3):235-247.
Whittier, T. R., S. G. Paulsen, D. P. Larsen, S. A. Peterson, A. T. Herlihy, and P. R. Kaufmann.
2002b. Indicators of ecological stress and their extent in the population of northeastern
lakes: a regional-scale assessment. Bioscience 52:235-247.
Zohary, T., and I. Ostrovsky. 2011. Ecological impacts of excessive water level fluctuations in
stratified freshwater lakes. Inland Waters 1:47-59.
71 NLA 2012 Technical Report. October Version 1.1
-------
Zar, J.H. 1999. Biostatistical Analysis, 4th ed. Prentice-Hall, Inc. New Jersey, USA.
72 NLA 2012 Technical Report. October Version 1.1
-------
Table 5-1, NLA reference sites from combined 2007 & 2012 surveys.
Selected using consistent criteria (Alan Herlihy's RT_NLA12_J2015, choosing 2012 visit for sites sampled in both
years). Bold font indicates grouping of reference sites used for modeling expected values for RVegQ, LitCvrQ, and
LitRipCvrQ.
ECQ9 ECOp5 Total 2007 2012
NAP APPAL 67 23 44
SAP. _APPAL 31 14 _1_7_
APPAL (98) (37) (61)
CPL _CPL 28 5 _2_3_
U M W U M W 49 18 _3_1_
TPL CENPL 23 7 16
NPL CENPL 11 3 8
SP L _CEN PL 35 21 _1_4_
CENPL (69) (31) (38)
WMT WEST 74 29 45
XER_ _W E_ST_ 20 4 _1_6_
WEST (94) (33) (61)
Totals for lower 48 states 338 124 214
Table 5-2. Assignment of riparian vegetation cover complexity, littoral cover complexity, and littoral-riparian
habitat complexity index variants by aggregated ecoregion.
Riparian Vegetation Cover Littoral Cover Littoral-Riparian Habitat
Aggregated Complexity Index Complexity Index Complexity Index
Omernik Ecoregion (RVegQ) ( LitCvrQ) (LitRipCvrQ)
CPL
RVegQ_2
LitCvrQ_b
LitRipCvrQ_2b
SAP
RVegQ_2
LitCvrQ_c
LitRipCvrQ_2c
NAP, UMW
RVegQ_2
LitCvrQ_d
LitRipCvrQ_2d
TPL, NPL, SPL
RVegQ_7
LitCvrQ_d
LitRipCvrQ_7d
WMT, XER
RVegQ_8
LitCvrQ_d
LitRipCvrQ_8d
73
NLA 2012 Technical Report. October Version 1.1
-------
Table 5-3. Summary of regression models used in estimating lake-specific expected values of Lake Physical Habitat
variables RVegQx, LitCvrQx and LitRipCvrQx under least-disturbed conditions.
See Appendix A for model details,
REGION y = RVepQ v = LitCvrQ v = LitRipCvrQ
NAP Ly* =f(Lat, Lon, LkOrig, RDisIX,) Ly =f(L_LkArea, RDisIX)
(R =23%, RMSE=0.162L*
(R = 12%, RMSE=0.281L)
Ly =f(Lat, Lon, LkOrig, RDisIX)
(R2=24%, RMSE=0.168L)
SAP Ly =f(Lon)
(R2=16%, RMSE=0.119L)
Ly =f(ElevXLon, RDisIX)
(R2=19%, RMSE=0.267L)
Ly =f(Lon, ElevXLon, Elev)
(R2=31%, RMSE= 0.148L)
CPL y =f(ElevXLat, RDisIX)
(R2=39%, RMSE=0 .0896)
y =f(L_Elev, RDisIX)
(R2=25%, RMSE= 0.174)
y =f( L_Elev, RDisIX)
(R2=44%, RMSE=0.093)
UMW Ly = (mean LRVegQ)
(R2=0%, RMSE=0.153L)
Ly = (mean LitCvrQ)
(R2=0%, RMSE=0.199L)
Ly = (mean LitRipCvrQ)
(R2=0% RMSE=0 .115L)
CENPL Ly=f(hiiAg)
(R2=15%, RMSE=0.318L)
Ly =f(LkOrig, hiiAg)
(R2=9%, RMSE=0.276L)
Ly=f(hiiAg)
(R2=15%, RMSE=0.233L)
WMT Ly =f{Lat, Elev, L_LkArea, LkOrigin) Ly =f(Lat, Elev, L_LkArea, LkOrigin) Ly =f(Lat, Elev, L_LkArea, LkOrigin)
(R2=28% RMSE=0.167L) (R2=16% RMSE=0.244L) (R2=29% RMSE=0.145L)
XER Ly =f(Lat, Elev)
(R2=24% RMSE=0.284L)
Ly =f(Lat, Elev)
(R2=16%, RMSE=0.290L)
Ly =f( Lat, Elev)
(R2=21% RMSE=0.265L)
*Ly refers to Logio-transformed lake habitat metric values.
**L refers to RMSE's that are in Logio units (e.g., 0.162L
74
NLA 2012 Technical Report. October Version 1.1
-------
Table 5-4, Null Model Geometric Means (gMean), geometric Standard Deviations (gSD), 5th percentiles, and 25th
percentiles of habitat index values in least-disturbed reference lakes in the aggregated ecoregions of the NLA,
The gMeans and gSDs are antilogs of mean and SD of logic-transformed index values (LogMean and LogSD), Bold,
italicized text identifies minimum LogSD and gSD values, i.e., the most precise models for each index. Bold,
underlined text marks the least precise models, gSDs calculated from log-transformed variables are expressions of
the proportional variance of these distributions, so are directly comparable among regions with different gMeans,
A range of+lLogSD is equivalent to multiplying and dividing the gMean by the gSD, For example, the gMean +1
gSD for the riparian vegetation cover complexity index in least-disturbed NAP lakes translates to a range of RVegQ
from 0,182 to 0,338: the geometric mean habitat index value of 0,2482 multiplied and divided by 1,363, The 5th
and 25th percentiles were estimated, respectively, as the mean of log-transformed index values minus 1,65 and
0,67 times the SD of log-transformed habitat index values (see Table 5-2 for the variant of each index used). All
percentiles are expressed in the units of the habitat indices, i.e., as antilogs of log-transformed values, (Note that
the constant 0,01 is subtracted from all antilogs because it was added when Q/E values were log-transformed).
Aggregated
Ref07i2
Ref07i2
Ref07i2
Ref07i2
Ref07i2
Ref07i2
ecoregion
Index
LogMean
LogSD
gMean
gSD
est 5th%
est 25th %
Riparian Veaetation Cover Complexity:
NAP null
RVegQ
-0.5881
0.1345
0.2482
1.363
0.1449
0.1998
SAP null
RVegQ
-0.6111
0.1277
0.2348
1.342
0.1407
0.1911
UMWnull
RVegQ
-0.6130
0.1533
0.2338
1.423
0.1262
0.1824
CPLnull
RVegQ
-0.6645
0.2810
0.2065
1.910
0.0644
0.1304
CENPLnull
RVegQ
-0.8346
0.3427
0.1364
2.201
0.0298
0.0760
TPLnull
RVegQ
-0.7295
0.3129
0.1764
2.055
0.0468
0.1050
NPLnull
RVegQ
-1.1352
0.2500
0.0632
1.778
0.0183
0.0398
SPLnull
RVegQ
-0.8093
0.3402
0.1451
2.189
0.0326
0.0817
WMTnull
RVegQ
-0.5900
0.1922
0.2470
1.557
0.1138
0.1811
XERnull
RVegQ
-0.8301
0.3070
0.1379
2.028
0.0360
0.0821
Littoral Cover Complexity:
NAPnull
LitCvrQ
-0.8174
0.2418
0.1423
1.745
0.0508
0.9049
SAPnull
LitCvrQ
-0.6469
0.2873
0.2155
1.938
0.0657
0.1347
UMWnull
LitCvrQ
-0.8756
0.1994
0.1232
1.583
0.0524
0.0879
CPLnull
LitCvrQ
-0.4883
0.2331
0.3049
1.710
0.1240
0.2167
CENPLnull
LitCvrQ
-1.0164
0.2880
0.0863
1.941
0.0222
0.0518
TPLnull
LitCvrQ
-0.9927
0.3190
0.0917
2.084
0.0203
0.0522
NPLnull
LitCvrQ
-0.9974
0.2116
0.0906
1.628
0.0350
0.0626
SPLnull
LitCvrQ
-1.0389
0.2929
0.0814
1.963
0.0200
0.0482
WMTnull
LitCvrQ
-1.0162
0.2578
0.0863
1.811
0.0262
0.0547
XERnull
LitCvrQ
-1.1457
0.2990
0.0615
1.991
0.0130
0.0351
Littoral-Riparian Habitat Complexity:
NAP null
LitRipCvrQ
-0.6740
0.1404
0.2018
1.382
0.1143
0.1606
SAP null
LitRipCvrQ
-0.6069
0.1690
0.2372
1.476
0.1201
0.1805
UMWnull
LitRipCvrQ
-0.7083
0.1149
0.1857
1.303
0.1165
0.1541
CPLnull
LitRipCvrQ
-0.5391
0.1687
0.2796
1.475
0.1422
0.2128
CENPLnull
LitRipCvrQ
-0.8820
0.2508
0.1212
1.782
0.0406
0.0791
TPLnull
LitRipCvrQ
-0.8230
0.2813
0.1403
1.911
0.0416
0.0874
NPLnull
LitRipCvrQ
-1.0442
0.1887
0.0803
1.544
0.0341
0.0575
75
NLA 2012 Technical Report. October Version 1.1
-------
Aggregated
Ref07i2
Ref07i2
Ref07i2
Ref07i2
Ref07i2
Ref07i2
ecoregion
Index
LogMean
LogSD
gMean
gSD
est 5th%
est 25th %
SPLnull
LitRipCvrQ
-0.8698
0.2305
0.1902
1.700
0.0462
0.0846
WMTnull
LitRipCvrQ
-0.7369
0.1677
0.1733
1.471
0.0869
0.1315
XERnull
LitRipCvrQ
-0.9455
0.2818
0.1034
1.913
0.0289
0.0634
76
NLA 2012 Technical Report. October Version 1.1
-------
Table 5-5. 0/E Physical Habitat Model means (LogMean, gMean), standard deviations (LogSD, gSD), and percentiles
of the distribution of habitat index O/E values for least-disturbed reference lakes in the aggregated ecoregions of
the NLA.
See Table 5-3 for the variant of each index used. The gMean and gSD are antilogs of mean and SD of logio-
transformed index values (LogMean and LogSD). Percentiles were estimated, respectively, as the log-transformed
index O/E value of 0.0 (see text) minus 1.65 and 0.67 times the SD of log-transformed habitat index values. Bold,
italicized text identifies minimum SD values, i.e., the most precise models for each index. Bold, underlined text
marks the least precise models. gSDs calculated from log-transformed variables are expressions of the proportional
variance of these distributions, so are directly comparable among regions with different geometric means. A range
of +1SD is calculated by multiplying and dividing the gMean by the gSD. For example, the LogMean + lLogSD for
the riparian vegetation cover complexity O/E index in least-disturbed lakes of the NAP (0.04276 + 0.1255)
translates to a range of O/E values from 0.78 to 1.31: the geometric mean habitat index O/E value of 1.00 (antilog
of+0.04276 = 1.10 minus log-transform constant 0.10) multiplied and divided by 1.34, the antilog of 0.1255. All
percentiles expressed as antilogs of log-transformed values minus constant 0.10. We based physical habitat
condition criteria based on the distribution of O/E index values in least-disturbed lakes within each region. The 5th
and 25th percentiles, respectively, were set as the upper bounds for poor and fair condition.
Aggregated
RefO/E
Ref 0/E
Ref O/E
Ref O/E
Ref O/E
Ref O/E
ecoregion
Index
LogMean
LogSD
gMean
gSD
5th %tile
25th %tile
NAP MLR Model
RVegQ_OE
(-0.00811)
(0.1255)
(0.88)
(1.34)
NAPoEYint
a a
+0.04276
0.1255
1.00
1.34
0.5850
0.8092
SAP MLR Model
RVegQ_OE
+0.04226
0.1105
1.00
1.29
0.6244
0.8295
UMWmlR Model
RVegQ_OE
+0.0428
0.1442
1.00
1.39
0.5381
0.7835
CPL MLR Model
RVegQ_OE
(-0.0617)
(0.2113)
(0.87)
(1.63)
CPLoEYint
a a
-0.00067
0.2129
0.90
1.63
0.3449
0.6191
CENPLMLRMode
RVegQ_OE
(-0.02799)
(0.3165)
(0.84)
(2.07)
I CENPLoEYint
a a
+0.04688
0.2928
1.01
1.96
0.2663
0.6091
WMTmlR Model
RVegQ_OE
+0.04290
0.1535
1.00
1.42
0.5162
0.7711
XERmlR Model
RVegQ_OE
+0.04199
0.2656
1.00
1.84
0.3016
0.6312
NAP MLR Model
LitCvrQ_OE
(+0.04502)
(0.2330)
(l.Ol)
(1.71)
NAPoEYint
a a
+0.04665
0.2330
1.01
1.71
0.3594
0.6772
SAP MLR Model
LitCvrQ OE
(-0.05093)
(0.2500)
(0.79)
(1.78)
SAPoEYint
a a
+0.04287
0.2440
1.00
1.75
0.3368
0.6575
UMWmlR Model
LitCvrQ OE
+0.04422
0.1954
1.00
1.57
0.4245
0.7152
CPL MLR Model
LitCvrQ_OE
(-0.03310)
(0.1909)
(0.83)
(1.55)
CPLoEYint
a a
-0.00743
0.1940
0.88
1.56
0.3704
0.6288
CEN PL MLR Model
LitCvrQ_OE
(+0.00495)
(0.2870)
(0.9l)
(1.94)
CENPLoEYint
a a
+0.02752
0.2839
0.97
1.92
0.2624
0.5876
WMTmlR Model
LitCvrQ_OE
+0.03770
0.2528
0.99
1.79
0.3174
0.6385
XERmlR Model
LitCvrQ_OE
+0.03451
0.2983
0.98
1.99
0.2486
0.5834
NAP MLR Model
LitRipCvrQ_
OE
(+0.00344)
(0.1321)
(0.9l)
(1.36)
NAPoEYint
a a
+0.04230
0.1321
1.00
1.36
0.5672
0.7990
SAP MLR Model
LitRipCvrQ_
OE
+0.04326
0.1329
1.00
1.36
0.5667
0.7999
UMWmlR Model
LitRipCvrQ_
OE
+0.04199
0.1110
1.00
1.29
0.6252
0.8296
CPL MLR Model
LitRipCvrQ_
OE
(-0.0248)
(0.1230)
(0.84)
(1.33)
CPLoEYint
a a
+0.01615
0.1234
0.94
1.33
0.5494
0.7580
CEN PL MLR Model
LitRipCvrQ_
OE
(-0.0121)
(0.2413)
(0.87)
(1.74)
I CENPLoEYint
a a
+0.04303
0.2246
1.00
1.68
0.3703
0.6808
77
NLA 2012 Technical Report. October Version 1.1
-------
Aggregated RefO/E RefO/E RefO/E RefO/E RefO/E RefO/E
ecoregion Index LogMean LogSD gMean gSD 5th %tile 25th %tile
WMTmlr Model LitRipCvrQ_OE +0.04200 0.1366 1.00 1.37 0.5556 0.7922
XERmlr Model LitRipCvrQ_OE +0.04012 0.2552 1.00 L80 0.3159 0.6398
Table 5-6, Empirical 75th and 35th percentiles of the distribution of vertical and horizontal drawdown.
As interpreted from indicators of lake level fluctuation (bfxVertHeight and bfxHorizDist) at least-disturbed
reference lakes sampled by NLA in 2007 and 2012, We used the 75th and 95th percentiles to define the boundaries
between small, medium and large magnitude of drawdown.
Number of Reference Lakes
(2007+2008)
Vertical Drawdown (m)
(bfxVertHeight)
Horizontal Drawdown (m)
(bfxHorizDist)
Ecogion
Lake
Origin
Total
Natural
Man-
Made
median
75th%
95th%
median
75th%
95th%
NAP
All
67
54
13
0.000
0.12
0.470
0.00
0.25
1.65
SAP
All
31
0
31
0.000
0.20
0.760
0.00
0.20
2.15
UMW
All
49
49
0
0.000
0.11
0.50
0.00
0.51
2.65
CPL
All
28
5
23
0.000
0.03
1.00
0.00
0.10
4.00
CENPL
Natural
29
29
0
0.000
0.06
0.28
0.00
0.10
2.85
U 11
Man-
Made
39/4o
0
39/40
0.010
0.36
1.20
0.21
1.55
14.63
WEST
Natural
69
69
0
0.021
0.33
1.00
0.00
0.64
9.43
U 11
Man-
Made
25
0
25
0.232
1.05
2.00
0.27
4.39
11.37
78
NLA 2012 Technical Report. October Version 1.1
-------
Table 5-7, Precision of the key NLA Physical Habitat indices used as the primary physical habitat condition
measures in the NLA,
Precision expressed as: 1) the pooled standard deviation of repeat visits (arep), 2) precision relative to potential or
observed range (orep/Rgpot and arep/Rgpot), and 3) the signal: noise ratio, where signal is among-lakes variance and
noise is within-lake variance during the same year and season (S/N = o2iake/o2reP), Analysis was based on NLA field
measurements on a summer probability sample of 1203 lakes in the 48 conterminous U.S. states, with repeat
sampling on a random subset of 88 of those lakes during the summer of 2012, Six of the sample lakes showed very
large changes in water level, which affected the littoral and riparian indicator values. We excluded these 6 lakes in
this analysis, except for values within perentheses, RDisJX is the Near-shore human disturbance index, RVegQc is
the Riparian vegetation cover & structure index, Log(RVegQcsOE) is the log-transformed O/E index for Riparian
vegetation cover & structure, LitCvrQds the Littoral cover complexity index, LogfLitCvrQcsOE is the log-
transformed O/E index for Littoral cover complexity, LitRipCvrQc is the Littoral-riparian habitat complexity index,
LogfLitRipCvrQaOE) is the log-transformed O/E index for Littoral-riparian habitat complexity, L_VertDD =
Logio(Vertical drawdown +0.1m), and L_HorizDD = Logio(Horizontal drawdown + 1m).
NLA PHab Indices
Orep
RQobs
Orep/RQobs
S/N
RDisJX
0.098
0.0-+0.950
0.103
9.1
L_RVegQc
0.144
-2.0 --0.266
0.083
6.6
L_RVegQc3OE
0.130
-1.0 -+0.666
0.078
5.0
L_LitCvrQc
0.190
-2.0 -+0.0266
0.094
3.4
L_LitCvrQc3OE
0.188
-1.0 -+0.759
0.107
2.2
L_LitRipCvrQc
0.134
-2.0 --0.135
0.072
5.6
L_LitRip CvrQc30E
0.122
-1.0 -+0.681
0.073
4.1
L_ VertDD
0.193 (0.266)
-1.0 -+1.654
0.073 (0.100)
5.9 (2.7)
L_HorizDD
0.148 (0.283)
0.0 -+2.873
0.052 (0.099)
11.0 (3.8)
79
NLA 2012 Technical Report. October Version 1.1
-------
Table 5-8, Association of NLA-2012 Physical Habitat Indices with high and low anthropogenic disturbance stress
classes (RT_NLA12 = R and T), defined as least-disturbed and most disturbed within NLA regions.
The t-values test the null hypothesis that the mean value of the habitat index in Reference sites minus the mean in
most-disturbed sites was zero in the NLA 2012 survey. Positive lvalues indicate that habitat index values are
greater in least-disturbed sites; negative values indicate higher index values in disturbed sites. See Figure 5-6 for
box and whisker plots by NLA regions, presented separately for the NLA 2012 and 2007 surveys.
NLA Physical Habitat Indices
tRT
PRT>I tRT 1
RDis_IX-Near-shore human disturbance index
-25*
<0.0001*
L_RVegQc - Riparian vegetation cover & structure index
13
<0.0001
L_RVegQc30E - O/E index for Riparian vegetation cover & structure
14
<0.0001
L_LitCvrQc - Littoral cover complexity index
8.3
<0.0001
L_LitCvrQc30E O/E index for Littoral cover complexity
9.3
<0.0001
L_LitRipCvrQc-Littoral-riparian habitat complexity index
13
<0.0001
L_LitRipCvrQc30E O/E index for Littoral-riparian habitat complexity
14
<0.0001
L_VertDD - Logio(Vertical drawdown +0.1m)
-4.3*
<0.0001*
L_HorizDD- Logio(Horizontal drawdown +1.0m)
-4.7*
<0.0001*
* Note that RDis_IX was one of the screening variables used to define least-disturbed reference sites
(RT_NLA12=R) and highly-disturbed sites (RT_NLA12=T), and was a very influential. The drawdown
variables bfaVertHeight and bfxHorizDist were also used in the screening process, but had only a minor
influence on the definition of sites.
80
NLA 2012 Technical Report. October Version 1.1
-------
Table 5-9, Association of NLA 2007 and 2012 Physical Habitat Indices with high and low anthropogenic disturbance
stress classes (RT_NLA12 = R and T), defined as least-disturbed and most disturbed within NLA regions.
The t-values test the null hypothesis that the mean value of the habitat index in Reference sites minus the mean in
most-disturbed sites was zero in the Domain specified in column 1, Positive farvalues indicate that habitat index
values are greater in least-disturbed sites; negative values indicate higher index values in disturbed sites. See
Figure 5-6 for box and whisker plots by NLA regions, presented separately for the NLA 2012 and 2007 surveys.
DOMAIN
L_RVegOE
L_LitCvrOE
L_LitRipCvrOE
LJHorizDD
National
07&12
ig****
12****
ig****
_7 7****
National 07&12
Natural
14****
g g****
14****
-3.5***
Man-Made
12****
g g****
12****
-6 o****
National 2007
12****
-j 2****
12****
_g 2****
2012
14****
g 2****
1^****
_4 7****
APPAL 2007
g 4****
3.0***
4 4****
+1.9
2012
g 4****
^ 1****
4 1****
-3 2***
NAP 2007
4 Q***
2.4**
4 1***
+1.1
2012
2 g***
2 g***
4 2****
-2.4*
SAP 2007
4 g****
1.1
2.9**
-0.2
2012
g 2****
1.4
3.3**
-2.4*
CENPL 2007
4 4****
2.5**
^ Q****
_4 Q****
2012
g 2****
^ 5****
g 4****
-0.6
TPL 2007
4 Q***
0.3
2.9**
-1.2
2012
3.6***
3.3**
2 7***
0.6
NPL 2007
1.3
4.6***
4 g***
i****
2012
2.4*
2.4*
2.2*
+1.6*
SPL 2007
1.4
2.1*
2 2**
-1.2
2012
g Q****
4 4****
gi****
-2.2*
CPL 2007
4 5***
1.4
4 g****
-1.3
2012
3.6***
4 2****
^ 4****
-0.5
UMW 2007
g 5****
g 2****
-j 2****
+4 4****
2012
g i****
3.3***
g 5****
-0.5
WEST 2007
g y****
2 ^***
-j y****
_g i****
2012
g 2****
2 2***
7 2****
2****
WMT 2007
g 2****
1.6*
^ 4****
7****
2012
g y****
2.3*
g Q****
g****
XER 2007
g 2****
3.5***
^ g****
-4 g****
2012
4 5****
2.0*
3.6**
-1.4
81
NLA 2012 Technical Report. October Version 1.1
-------
Near-Shore Station NLA-2007:
Near-Shore Station NLA-2012:
Variable-
width
15 m
A
{
Riparian
zone
Drawdown
zone
Littoral
zone
>e-
15 m
1m Shore zone
}
10 m
t
Observation station
Figure 5-1. Field sampling design with 10 near-shore stations at which data were collected to characterize near shore
lake riparian and littoral physical habitat in the 2007 and 2012 National Lakes Assessment (NLA) surveys.
The 10 stations were systematically spaced around the shore of the lake from random starting point. Insert shows
riparian plot, shoreline band, littoral plot, and (for NLA-1012 only) drawdown zone plot located at each station.
Littoral-Riparian Plot
82
NLA 2012 Technical Report. October Version 1.1
-------
X
I
Ln
Q
cc
CD
'lo
M
CD
cc
1 WMT 2-UMW
3-NAP
4-SAP
5-SPL
DRankEcoRef
T r
6-XER 7-TPL
8-CPL
9-NPL
1 -
x
l
Ln
Q
cc
IS)
CD
"lS)
<
0-
1 1 1 1 1 1 1 H
1_WMT 2-UMW 3-NAP 4-SAP 5-SPL 6-XER 7-TPL 8-CPL 9-NPL
DRankEcoRef
Figure 5-2. Near-shore anthropogenic disturbance (RDis_IX) in NLA0712 regions, ordered by their median
Reference site RDis.
Upper plot: Least-disturbed reference sites. Lower plot: all sites. Unweighted sample statistics are shown; box
midline and lower and upper ends show median and 25th and 75th percentile values, respectively; whiskers show
maximum and minimum observations within 1.5 times the interquartile range above/below box ends; circles
show outliers.
83
NLA 2012 Technical Report. October Version 1.1
-------
c
ro
T3
CD
CD
to
CD
0£
RDislXjmed
0.3-
0.25-
0.2-
0.15-
0.1-
0.05-
o-
i i i i i i i
1-WMT 2-UMW 3-NAP 4-SAP 5-XER 6-CENPL 7-CPL
RegionDRk
Figure 5-3. Near-shore anthropogenic disturbance in NLA0712 least-disturbed reference sites (median RDisJX),
ordered by aggregated region according to the same median level of near-shore disturbance.
The NLA EC09 regions NPL, SPL, and TPL are combed into the Central Plains (CENPL) region.
84
NLA 2012 Technical Report. October Version 1.1
-------
Log(RVegQ):
RVegLSDrNul
RVegLSDrOEaj
0.35-
O
-0.35
0.3-
0.25-
o
o
-0.3
-0.25
0.2-
Q
-0.2
0.15-
9
-0.15
0.1-
9
o
-0.1
0.05-
-0.05
0-
-0
1-WMT 2-UMW 3-NAP 4-SAP 5-XER 6-CENPL 7-CPL
RegionDRk
IQ RVegLSDrNul | | % RVegLSDrOEaj |
Log(LitCvrQ):
LtCvLSDrNul
LtCvLSDrOE
0.35-
-0.35
0.3-
o
#
-0.3
0.25-
w
-0.25
0.2-
-0.2
0.15-
-0.15
0.1-
-0.1
0.05-
-0.05
0-
-0
1-WMT 2-UMW 3-NAP 4-SAP 5-XER 6-CENPL 7-CPL
RegionDRk
IQ LtCvLSDrNuT| 9 LtCvLSDrpj]
Log(LitRipCvrQ):
LtRpCvLSDrNul
LtRpCvLSDrOEaj
0.3-
-0.3
O
0.25-
o
-0.25
0.2-
-0.2
O
o
O
0.15-
-0.15
8
9
0.1-
-0.1
0.05-
-0.05
0-
-0
I
1-WMT
i
2-UMW
i i i i
3-NAP 4-SAP 5-XER 6-CENPL
I
7-CPL
RegionDRk
| O LtRpCvLSDrNul # LtRpCvLSDrOEaj
Figure 5-4. LogSD's for Null-Model arid regression-based O/E model for Near-shore RVegQ, LitCvrQ, and LitRipCvrQ
in the set of least-disturbed lakes and reservoirs (Table 5-1) sampled in the combined 2007 and 2012 NLA surveys.
X-axis shows the 7 modeling regions ordered by increasing median RDisJX in the reference sites. The NLA EC09
regions NPL, SPL, and TPL are combed into the Central Plains (CENPL) region. Low variance among reference sites
denotes greater precision in estimating expected reference condition. The smaller variance in regression-based
O/E models (black dots) illustrate their greater precision compared with null models (open circles) for a given
indicator and region.
85
NLA 2012 Technical Report. October Version 1.1
-------
10-"
0.5-
0.5-
uo
LU
o
co
o
O
o>
o
0.0-
0
-0.5-
-1.0-
O
o
0
g
1
R
i
T
1-
0.8-
0.6-
£2
b
1.0-
00
3
1
G
O
0.5-
¦n
,
1
I
0.0-
-0.5-
-1.0-
8
i
o
@
I
R
r~
S
"X"
T
3.0-
2.5-
3
I
Q
Q 15-1
c
O
x
_i 1.0H
0.5-
0.0-
8
8
~
8
0
1
I-
S
n-
T
Figure 5-5. Contrasts in key NLA physical habitat index values among least-disturbed reference (R),
intermediate (S), and highly disturbed (T) lakes in the contiguous 48 states of the U.S. based on combined NLA
2007 and 2012 data.
Unweighted sample statistics are shown; box midline and lower and upper ends show median and 25th and
75th percentile values, respectively; whiskers show maximum and minimum observations within 1.5 times the
interquartile range above/ below box ends; circles show outliers. See Table 5-9 for t and p values for the
differences between means for reference (R) and disturbed (T) sites.
86
NLA 2012 Technical Report. October Version 1.1
-------
RDis IX L1_RVegQc30E15
Figure 5-6. Contrasts in key NLA physical habitat index vaiues among least-disturbed reference (R), intermediat (S),
and highly disturbed (T) lakes in the contiguous 48 states of the U.S. shown separately for the NLA 2007 and 2012
surveys.
Unweighted sample statistics are shown; box midline and lower and upper ends show median and 25th and 75th
percentile values, respectively; whiskers show maximum and minimum observations within 1.5 times the
interquartile range above/ beiow box ends; circles show outliers. See Table 5-9 for t and p values for the
differences between means for reference (R) and disturbed (T) sites.
87
NLA 2012 Technical Report. October 2024 Version 1.1
-------
Lake Physical Habitat Expected Condition Models Appendix A
Table 3 from TSD Chapter. Summary of regression models used in estimating lake-specific
expected values of Lake Physical Habitat variables RVegQx, LitCvrQx and LitRipCvrQx under
least-disturbed conditions. Variable definitions and model details on following pages.
REGION y = RVepQ
y = LitCvrQ
y = LitRipCvrQ
NAP Ly* =f(Lat, Lon, LkOrig, RDisIX,) Ly =f(L_LkArea, RDisIX)
(R2=23%, RMSE=0.162L*
(R2= 12%, RMSE=0.281L)
Ly =f(Lat, Lon, LkOrig, RDisIX)
(R2=24%, RMSE=0.168L)
SAP Ly =f(Lon)
(R2=16%, RMSE=0.119L)
Ly =f(ElevXLon, RDisIX)
(R2=19%, RMSE=0.267L)
Ly =f(Lon, ElevXLon, Elev)
(R2=31%, RMSE= 0.148L)
CPL y =f(ElevXLat, RDisIX)
(R2=39%, RMSE=0 .0896)
y =f(L_Elev, RDisIX)
(R2=25%, RMSE= 0.174)
y =f( L_Elev, RDisIX)
(R2=44%, RMSE=0.093)
UMW Ly = (mean LRVegQ)
(R2=0%, RMSE=0.153L)
Ly = (mean LitCvrQ)
(R2=0%, RMSE=0.199L)
Ly = (mean LitRipCvrQ)
(R2=0%, RMSE=0 .115L)
CENPL Ly=f(hiiAg)
(R2=15%, RMSE=0.318L)
Ly =f(LkOrig, hiiAg)
(R2=9%, RMSE=0.276L)
Ly=f(hiiAg)
(R2=15% RMSE=0.233L)
WMT Ly =f(Lat, Elev, L_LkArea, LkOrigin) Ly =f(Lat, Elev, L_LkArea,
(R2=28% RMSE=0.167L) (R2=16% RMSE=0.244L)
LkOrigin) Ly=f(Lat, Elev, L_LkArea, LkOrigin)
(R2=29% RMSE=0.145L)
XER Ly =f(Lat, Elev)
(R2=24% RMSE=0.284L)
Ly =f(Lat, Elev)
(R2=16%, RMSE=0.290L)
Ly =f( Lat, Elev)
(R2=21% RMSE=0.265L)
*Ly refers to Logio-transformed lake habitat metric values.
"*L refers to RMSE's that are in Logio units (e.g., 0.162L)
88
NLA 2012 Technical Report. October 2024 Version 1.1
-------
VARIABLE DEFINITIONS
On following pages variables are defined as follows:
Observed Habitat Indicator values are: (in the TSD text, these are abbreviated as RVeaQ, LitCvrQ, and
LitRipCvrQ)
RVegQclS, LitCvrQcl5, LitRipCvrQcl5
L_RVegQcl5 = Logi 0(RVegQcl5 +0.01)
L_LitCvrQcl5 = Log10(LitCvrQcl5 +0.01)
L_LitRipCvrQcl5 = Log10(LitRipCvrQcl5 +0.01)
Expected Condition Regression Models have the form (in the TSD text. Expected condition variables are
abbreviated as RVeaQX. LitCvrQX. and LitRipCvrQX):
L_RVegQc3xl5 = f(predictors) or RVegQc3xl5 = f(predictors)
L_LitCvrQc3xl5 = f(predictors) or LitCvrQc3xl5 = f(predictors)
L_LitRipCvrQc3xl5 = f(predictors) or LitRipCvrQc3xl5 = f(predictors)
Observed/Expected Condition Variables are defined as follows (in the TSD text. O/E variables are
abbreviated as RVeciQ OE, LitCvrQ OE, and LitRipCvrQ OE):
RVegQc30E15= (RVegQcl5/RVegQc3xl5) and Ll_RVegQc30E15 = Logw(RVegQc30E15 +0.1)
LitCvrQc30E15= (LitCvrQcl5/LitCvrQc3xl5) and Ll_LitCvrQc30E15 = Log10(LitCvrQc3OE15 +0.1)
LitRipCvrQc30E15= (LitRipCvrQcl5/LitRipCvrQc3xl5) and Ll_LitRipCvrQc30E15 =
Log10(LitRipCvrQc3OE15 +0.1)
Predictors defined from variables in prk datafile NLA12 pc.nla lakeinfo all 20150415 are as follows:
LATdd_use = LAT_DD_N83 = latitude in decimal degrees
LONdd_use = LON_DD_N83 = longitude in decimal degrees
ELEV_use = ELEVATION = lake surface elevation (meters above mean sea level)
L_ELEV_use = Logw(ELEV_use)
LkArea_km2 = LAKEAREA = lake surface area (km2)
L_LkAreakm2 = Logio (LkArea_km2)
Lake_Origin_use = LAKE_ORIGIN (with values: 'NATURAL' or 'MAN-MADE')
Reservoir = an indicator variable of Lake Origin, where
If Lake_Origin_use = 'MAN-MADE' then Reservoir= 1;
If Lake_Origin_use = 'NATURAL' then Reservoir=0;
Field human disturbance variables:
RDis_IX index of near-shore human disturbance intensity and extent (see TSD text equation 5)
hiiAg proximity-weighted mean tally of up to 3 near-shore agricultural disturbances (mean among
stations).
89
NLA 2012 Technical Report. October 2024 Version 1.1
-------
NAP Expected PHab Reference Condition Models:
L_RVegQc3xl5 = 2.34593-(0.03705*LATdd_use)+(0.01723*LONdd_use)-(0.07954*Reservoir)
-(0.31865 * RDisJX);
Note: Reservoir = 0 for natural lakes, 1 for man-made reservoirs.
Rsq=0.2331 RMSE=0.16177 p<.0001 n=166/170;
Sites: All non-overlapping 2007-2012 NAP RT_NLA12=R or S;
Set RDis_IXto zero (14% of 2007-&12 NAP sample sites have RDis_IX=0);
RVegQc3xl5= 10 * * (L_RVegQc3xl5)-0.01;
Applied simple dirty models for LitCvr and LitRipCvr (see powerpoint file of regressions 6/13/14) that
better define the influence of lake area but then MUST include RDisJX, because it is the strongest
predictor of any of the 3 PHab indices if RT_NLA12_2015 S or T sites are included with reference (R)
sites;
Adjustment for reference distribution of O/E values:
L_RVegQc30E15= +0.04276 - (0.29150 RDisJX);
Rsq= 0.2026 RMSE=0.14469 p<0.0001 n=166/170;
Sites: All non-overlapping 2007-2012 NAP RT_NLA12=R or S;
Ref O/E distribution based on Y-intercept of adjustment regression, but SD of ref sites only (not S sites)
L_LitCvrQc3xl5= -0.8598 -(0.08109*L_LkAreakm2) - (0.28562*RDisJX);
Rsq=0.1228 RMSE=0.2808 p<0.0001 n=166/170;
Set RDisJX to zero (14% of 2007-2012 NAP sample sites have RDis_IX=0);
Sites: All non-overlapping 2007-2012 NAP RT_NLA12_2015=R or S;
LitCvrQc3xl5=10**(LJitCvrQc3xl5)-0.01;
Adjustment for reference distribution of O/E values:
LJitCvrQc30E15= +0.04665 - (0.28240 RDisJX);
Rsq= 0.0592 RMSE=0.26819 p=0.0009 n=166/170;
Sites: All non-overlapping 2007-2012 NAP RT_NLA12=R or S;
Ref O/E distribution based on Y-intercept of adjustment regression, but SD of ref sites only (not S sites)
LJitRipCvrQc3xl5= 2A1606-(0.03964*LATdd_use)+(0.01798*LONdd_use) -(0.08301* Reservoir)
-(0.34039* RDisJX);
Note: Reservoir = 0 for natural lakes, 1 for man-made reservoirs.
Rsq=0.2407 RMSE=0.16783 p<0.0001 n=166/170;
Set RDisJX to zero (14% of 2007-2012 NAP sample sites have RDis_IX=0);
Sites: All non-overlapping 2007-2012 NAP RT_NLA12_2015=R or S;
LitRipCvrQc3xl5=10**(LJitRipCvrQc3xl5)-0.01;
Adjustment for reference distribution of O/E values:
LJitRipCvrQc30E15= +0.04230 - (0.31323 RDisJX);
Rsq= 0.2075 RMSE=0.15095 p<0.0001 n=166/170;
Sites: All non-overlapping 2007-2012 NAP RT_NLA12=R or S;
Ref O/E distribution based on Y-intercept of adjustment regression, but SD of ref sites only (not S sites).
90
NLA 2012 Technical Report. October 2024 Version 1.1
-------
SAP - Expected PHab Condition Models:
L_RVegQc3xl5= 0.24710 +(0.01012* LONdd_use);
Rsq=0.1637 RMSE=0.11878 p=0.0240 n=31/31 ;
Sites: All non-ovelapping 2007-2012 SAP RT_NLA12_2015=R;
RVegQc3xl5= 10 * * (L_RVegQc3xl5)-0.01;
Ref O/E distribution based on mean and SD of ref sites.
L_LitCvrQc3xl5= -0.66613 -(0.00000410* ElevXLon_use) -(0.51350*RDis_IX);
Rsq=0.1942 RMSE=0.26697 p=0.0487 n=31/31;
Set RDis_IX to zero (2% of 2007-2012 SAP sample sites have RDis_IX=0);
Sites: All non-overlapping 2007-2012 SAP RT_NLA12_2015=R;
LitCvrQc3xl5=10**(L_LitCvrQc3xl5)-0.01;
Adjustment for reference distribution of O/E values:
L_LitCvrQc30E15= +0.04287 - (0.46211 RDisJX);
Rsq= 0.0790 RMSE=0.24397 p=0.1255 n=31/31;
Sites: All non-overlapping 2007-2012 SAP RT_NLA12=R;
Ref O/E distribution based on Y-intercept and RMSE of adjustment regression.
L_LitRipCvrQc3xl5=1.92708 -(0.000115130* ElevXLon_use) + (0.03141*LONdd_use) -
(0.00923* ELEV_use);
Rsq=0.3083 RMSE=0.14817 p=0.0175 n=31/31;
Sites: All non-overlapping 2007-2012 SAP RT_NLA12_2015=R;
LitRipCvrQc3xl5=10**(L_LitRipCvrQc3xl5)-0.01;
Ref O/E distribution based on mean and SD of ref sites.
91
NLA 2012 Technical Report. October 2024 Version 1.1
-------
CPL Expected PHab Condition Models:
RVegQc3xl5=0.35438 -0.00003019{ElevXLat_use) - 0.15193[RDisJX);
Rsq= 0.3868 RMSE=0.08963 p<0.0001 n=28/28;
Sites: All non-overlapping 2007-2012 CPL RT_NLA12_2015=R;
Set RDis_IX to lowest value in the region (4.4% have RDis_IX=0 in CPL);
Adjustment for reference distribution of O/E values:
L_RVegQc30E15= -0.0006653 - (0.22746 RDisJX);
Rsq= 0.0235 RMSE=0.21279 p=0.4362 n=28/28;
Sites: All non-overlapping 2007-2012 CPL RT_NLA12=R;
Note: Regression keeping one low outlier with very little leverage;
Ref O/E distribution based on Y-intercept and RMSE of adjustment regression.
LitCvrQc3xl5= 0.71804 - (0.19300*£_£/ei/_i/se) - (0.12565*RDisJX);
Rsq= 0.2526 RMSE=0.17393 p<0.0001 n=28/28;
Sites: All non-overlapping 2007-2012 CPL RT_NLA12_2015=R;
Set RDis_IX to lowest value in the region (0 in CPL);
Adjustment for reference distribution of O/E values:
L_LitCvrQc30E15= -0.00743 - (0.09579 RDisJX);
Rsq= 0.0051 RMSE=0.1940 p=0.7178 n=28/28;
Sites: All non-overlapping 2007-2012 CPL RT_NLA12=R;
Ref O/E distribution based on Y-intercept and RMSE of adjustment regression.
LitRipCvrQc3xl5= 0.59561 - (0.15322*L_Elev_use) - (0.14358* RDisJX);
Rsq= 0.4423 RMSE=0.09293 p<0.0001 n=28/28;
Sites: All norepeat 2007-2012 CPL RT_NLA12_2015=R;
Set RDisJX to lowest value in the region (0 in CPL);
Adjustment for reference distribution of O/E values:
LJitRipCvrQc30E15= 0.01615 - (0.15265 RDisJX);
Rsq= 0.0312 RMSE=0.1234 p=0.3685 n=28/28;
Sites: All non-overlapping 2007-2012 CPL RT_NLA12=R;
Ref O/E distribution based on Y-intercept and RMSE of adjustment regression.
92
NLA 2012 Technical Report. October 2024 Version 1.1
-------
UMW Expected PHab Condition Models:
L_RVegQc3xl5= -0.61298;
****Dropped LON and LkArea - USED geometric (Log mean) NULL MODEL;
Rsq=0 RMSE=0.15333 n=49/50 ;
Sites: All non-overlapping 2007-2012 UMW RT_NLA12_2015=R;
RVegQc3xl5= 10 * * (L_RVegQc3xl5)-0.01;
Ref O/E distribution based on mean and SD of ref sites.
L_LitCvrQc3xl5= -0.87559;
****Dropped survey year - USED geometric (Log mean) NULL MODEL;
Rsq=0 RMSE=0.19944 p=N/A n=49/50;
Sites: All non-overlapping 2007-2012 UMW RT_NLA12_2015=R;
LitCvrQc3xl5=10**(L_LitCvrQc3xl5)-0.01;
Ref O/E distribution based on mean and SD of ref sites.
L_LitRipCvrQc3xl5=-0.7083 0;
***** Dropped Lake Area - USED geometric (Log mean) NULL MODEL;
Rsq=0 RMSE=0.11487 p=N/A n=49/50;
Sites: All non-overlapping 2007-2012 UMW RT_NLA12_2015=R;
LitRipCvrQc3xl5=10**(L_LitRipCvrQc3xl5)-0.01;
LitCvrQc3xl5=10**(L_LitCvrQc3xl5)-0.01;
Ref O/E distribution based on mean and SD of ref sites.
93
NLA 2012 Technical Report. October 2024 Version 1.1
-------
CENPL fNPL + SPL + TPL) Expected PHab Condition Models:
L_RVegQc3xl5=-0.75460- {0.0.86385*hiiAg);
Rsq=0.1532 RMSE=0.3178 p<0.0009 n=69/71;
Sites: All non-overlapping 2007-2012 CENPL_2015 RT_NLA12_2015=R, Excluding KS-R02 SD-101 (Oahi Res)
which has inadequate no of transects, but Includes Mound City res KS-R02 with corrected Elevation;
Set hiiAg to lowest value in the region (0)
Note: 2007-2012 NLA sites in CENPL with hiiAg=0 in NPL(>25%) SPL(>50%) TPL(75%)
RVegQc3xl5= 10 * * (L_RVegQc3xl5)-0.01;
Adjustment for reference distribution of O/E values:
L_RVegQc30E15= 0.04688 - (0.80799 hiiAg);
Rsq= 0.1571 RMSE=0.29278 p=0.0007 n=69/71;
Ref O/E distribution based on Y-intercept and RMSE of adjustment regression.
L_LitCvrQc3xl5= -1.03378 + 0.10822*Reservoir-(0.38197*hiiAg);
Note: Reservoir = 0 for natural lakes, 1 for man-made reservoirs.
Rsq=0.0855 RMSE= 0.27579 p<0.0572 n=69/71;
Sites: All non-overlapping 2007-2012 CENPL_2015 RT_NLA12_2015=R
Set hiiAg to lowest value in the region (0)
Note: 2007-2012 NLA sites in CENPL with hiiAg=0 in NPL(>25%) SPL(>50%) TPL(75%)
LitCvrQc3xl5=10**(L_LitCvrQc3xl5)-0.01;
Adjustment for reference distribution of O/E values:
L_LitCvrQc30E15= 0.02752 - (0.35038 hiiAg);
Rsq= 0.0359 RMSE=0.28386 p=0.1255 n=69/71;
Ref O/E distribution based on Y-intercept and RMSE of adjustment regression.
L_LitRipCvrQc3xl5=-0.82455-(0.61960* hiiAg);
Rsq=0.1471 RMSE=0.23336 p=0.0011 n=69/71;
Sites: All non-overlapping 2007-2012 CENPL_2015 RT_NLA12_2015=R
Set hiiAg to lowest value in the region (0)
Note: 2007-2012 NLA sites in CENPL with hiiAg=0 in NPL(>25%) SPL(>50%) TPL(75%)
LitRipCvrQc3xl5=10**(L_LitRipCvrQc3xl5)-0.01;
Adjustment for reference distribution of O/E values:
L_LitRipCvrQc30E15= 0.04303 - (0.59485 hiiAg);
Rsq= 0.1465 RMSE=0.22462 p=0.0012 n=69/71;
Ref O/E distribution based on Y-intercept and RMSE of adjustment regression.
**** Note: If remove sites East of approximately -95 degrees LON that removes all hiiAg so association
with LON is largely assoc with hiiAg - adopted conservative model without LON. See dirty models for all
three indices with hiiAg alone (prk 3/13/15 SAS EnterpriseGuide projects) for all three of the above, they
all have higher Rsq, similar RMSE, similar intercepts, similar slopes p<0.0001 n= 669/694 to 673/694.
94
NLA 2012 Technical Report. October 2024 Version 1.1
-------
WMT Expected PHab Condition Models:
L_RVegQc3xl5=0.53572-{0.00008953*ELEV_use)-{0.25957*Reservoir)+{0.07296*L_LkAreakm2)
-(0.01939* LATdd_use);
Note: Reservoir = 0 for natural lakes, 1 for man-made reservoirs.
Rsq=0.2825 RMSE=0.16743 p=0.0001 n=74/75;
Sites: All non-overlapping 2007-2012 WMT RT_NLA12_2015=R;
RVegQc3xl5= 10 * * (L_RVegQc3xl5)-0.01;
Ref O/E distribution based on mean and SD of ref sites.
L_LitCvrQc3xl5= -1.10550-(0.00004299*ELEV_use)-(0.05083*L_LkAreakm2)+(0.00407*LATdd_use)
-(0.18384* Reservoir);
Note: Reservoir = 0 for natural lakes, 1 for man-made reservoirs.
Rsq=0.1555 RMSE=0.24373 p=.0187 n=74/75;
Sites: All non-overlapping 2007-2012 WMT RT_NLA12_2015=R;
LitCvrQc3xl5=10**(L_LitCvrQc3xl5)-0.01;
Ref O/E distribution based on mean and SD of ref sites.
L_LitRipCvrQc3xl5=-0.08802-(0.00006666*ELEV_use)+(0.04200*L_LkAreakm2)-(0.01015*LATdd_use)-
(0.22650* Reservoir);
Note: Reservoir = 0 for natural lakes, 1 for man-made reservoirs.
Rsq=0.2922 RMSE=0.14513 p<.0001 n=74/75;
Sites: All no-repeat 2007-2012 WMT RT_NLA12_2015=R;
LitRipCvrQc3xl5=10 * * (L_LitRipCvrQc3xl5)-0.01;
Ref O/E distribution based on mean and SD of ref sites.
95
NLA 2012 Technical Report. October 2024 Version 1.1
-------
XER Expected PHab Condition Models:
L_RVegQc3xl5= 0.44708 -(0.02612 *LATdd_use) -(0.00013249*ELEV_use) ;
Rsq=0.2365 RMSE=0.28355 p=0.1009 n=20/21 ;
Sites: All no-repeat 2007-2012 XER RT_NLA12_2015=R;
RVegQc3xl5= 10 * * (L_RVegQc3xl5)-0.01;
Ref O/E distribution based on mean and SD of ref sites.
L_LitCvrQc3xl5=0.08706-(0.02849*LATdd_use)-(0.00003932*ELEV_use);
Rsq=0.1578 RMSE=0.29004 p=0.2322 n=20/21;
Sites: All no-repeat 2007-2012 XER RT_NLA12_2015=R;
*** Note this was 8th best in All Subsets Regression models with <=2 predictors ranked by Cp;
*** Note this was 6th best in All Subsets ranked by Rsq;
*** Consistent model across all the indicators and across full set of sites;
LitCvrQc3xl5=10**(L_LitCvrQc3xl5)-0.01;
Ref O/E distribution based on mean and SD of ref sites.
L_LitRipCvrQc3xl5=0.24931 - (0.02529*LATdd_use)-(0.00010090*ELEV_use) ;
Rsq=0.2115 RMSE= 0.26455 p=0.1327 n=20/21;
Sites: All no-repeat 2007-2012 XER RT_NLA12_2015=R;
LitRipCvrQc3xl5=10 * *(L_LitRipCvrQc3xl5)-0.01;
Ref O/E distribution based on mean and SD of ref sites.
NOTE 3/13/15 prk: Reexamined models. The p-values (and of course also r2 and RMSE) not improved by
using
single predictors (ELEV_use LATdd_use and ELEVxLatdd_use). The mechanisms and univariate plots of
these single predictors all convincing and support the 3 models above;
96
NLA 2012 Technical Report. October 2024 Version 1.1
-------
Chapter 6: Water Chemistry
6.1 Background information
The NLA report summarizes water quality stressor data collected at the deepest part of each
study lake (up to 50 m). Field sampling included a depth profile and a 0-2 m depth integrated
water sample. Variables analyzed for the NLA 2012 report include: total nitrogen (TN), total
phosphorus (TP), chlorophyll-a (CHLA), turbidity, acidity, and dissolved oxygen. Acidity,
dissolved oxygen and trophic state class thresholds were based on established criteria and
applied consistently across the nation. Least, moderate, and most disturbed condition classes
were established for TP, TN, CHLA, and turbidity using the same percentile of reference sites
approach that was used in NLA 2007 (Herlihy and Sifneos, 2013). Thresholds, however, were
recalculated to include additional nutrient reference sites sampled in 2012. This more than
doubled the number of nutrient reference sites available in each ecoregion allowing for better
estimation of the percentiles used to calculate the thresholds. Separate thresholds were
established for each of the nine ecoregions reported on in NLA 2012. As a result of threshold
refinement 2007 benchmark values were revised; therefore, direct comparisons should not be
made between 2012 results and those reported in 2007.
6.2 Threshold development
6.2.1 Acidity and Dissolved Oxygen
For setting acidity classes, concentrations of acid neutralizing capacity (ANC) and dissolved
organic carbon (DOC) were analyzed following the scheme developed by Herlihy et al. (1991).
Sites with acid neutralizing capacity (ANC) > 50 ueq/L were considered to be non-acidic and
least disturbed for acidification. Sites with ANC < 50 |aeq/L and DOC values > 6 mg/L were
classified as naturally acidic due to organic acids. Sites with ANC < 0 |aeq/L and DOC values < 6
mg/L were classified as acidic due to either acidic deposition or acid mine drainage and
considered most disturbed. Sites with ANC between 0 and 50 |aeq/L and DOC < 6 mg/L were
considered acid-influenced but not currently acidic. These low ANC sites typically become acidic
during high flow events (episodic acidity) and were considered moderately disturbed.
Depth profiles of dissolved oxygen were collected at the deepest of the lake. Surface water
dissolved oxygen was calculated by removing all duplicate depth observations and taking the
mean of all dissolved oxygen values between 0 and 2 meters depth, inclusive. If the lake was
shallower than 2 m depth, the entire depth profile was used. Surface water dissolved oxygen
was classified into three classes, least disturbed (>5 mg/L), moderately disturbed (3-5 mg/L),
and most disturbed (<3 mg/L).
97
NLA 2012 Technical Report. October 2024 Version 1.1
-------
6.2.2 Trophic State
Lakes have long been classified according to their trophic state. By the dictionary, "trophic" is
defined as of or relating to nutrition. A eutrophic lake has high nutrients and high algal and/or
macrophyte plant growth. An oligotrophic lake has low nutrient concentrations and low plant
growth. Mesotrophic lakes fall somewhere in between eutrophic and oligotrophic lakes and
hypereutrophic lakes have very high nutrients and plant growth. Lake trophic state is typically
determined by a wide variety of natural factors that control nutrient supply, climate, and basin
morphometry. Trophic state can be defined based on a number of different nutrient or plant
biomass variables. For NLA 2012, trophic state was defined using specific numeric criteria for
concentrations CHLA (Table 6-1). The same trophic state classification was used for all
ecoregions.
\n;il> le
(Jliuoii'iiphic
Mcsoliiiphic
1 Miliophic
11\ pcrciiirophic
Chlorophyll-a (|ig/L)
<2
>2 and <7
>7 and <30
>30
6.2.3 Total nitrogen, total phosphorus, chlorophyll-a, and turbidity
TN, TP, CHLA, and turbidity were classified into least, moderate, or most, disturbed condition
classes based on percentiles of the nutrient reference site distribution (Herlihy and Sifneos,
2008, 2013). See Section 3.4 for more information on selecting reference sites for nutrients.
Once the nutrient reference lakes were selected, nutrient levels for separating least disturbed,
moderately disturbed, and most disturbed were determine from the distribution of reference
lake nutrient concentrations from each ecoregion (and for the Southern Plains for natural and
manmade lakes separately). Nutrient levels were determined for both total phosphorus (TP)
and total nitrogen (TN). The cutoff between least disturbed and moderately disturbed lakes was
set at the 75th percentile (Q3) of reference lakes, and the cutoff between moderately disturbed
and most disturbed lakes was set at the 95th percentile (P95) of reference lakes. If a nutrient
ecoregion had < 20 lakes, then the cutoff between the moderately disturbed and most
disturbed lakes was the maximum nutrient concentration (P95 = maximum) for reference lakes
in that nutrient ecoregion.
In addition to developing thresholds for nutrients, we determined thresholds from population
percentiles in the reference lakes in each of the nutrient ecoregion for chlorophyll-a and
turbidity. Like the nutrient thresholds, these percentile-based thresholds were used to
determine least disturbed, moderately disturbed, and most disturbed lake conditions for the
NLA. With the cutoff between least disturbed and moderately disturbed lakes set at the 75th
percentile (Q3), and the cutoff between the moderately disturbed and most disturbed lakes set
at 95th percentile (P95).
There was a very large difference in the absolute concentrations of TP and TN among
ecoregions in the nutrient reference sites (Figure 6-1 and Figure 6-2). Looking at the data, it is
also evident why the natural lakes in the SPL need their own threshold versus man-made SPL
98
NLA 2012 Technical Report. October 2024 Version 1.1
-------
lakes. Table 6-2 reports the 75th and 95th percentile-based thresholds used to define the least,
moderately, and most, disturbed condition classes for TP, TN, CHLA, and turbidity for each of
the ecoregions.
1000
O)
-------
100000
10000
O)
c
-------
Table 6-2, NLA2012 least, moderately, and most disturbed thresholds (75th/95th percentiles) for TP, IN, CHLA, and
turbidity condition classes.
TP (ng/L)
TP (ng/L)
TN (ng/L)
TN (ng/L)
75th
95th
75th
95th
Ecoregion
Least-
moderately
Moderately-Most
Least-moderately
Moderately-Most
CPL
37.0
51.0
510
801
NAP
14.5
22.0
400
600
NPL
69.5
82.0
866
1,620
SAP
19.0
33.0
309
407
SPL-manmade
34.0
56.0
657
830
SPL-natural
486
839
7,925
12,875
TPL
49.0
82.0
1,105
1,699
UMW
28.0
41.0
722
920
WMT
29.0
53.0
245
380
XER
48.0
84.0
465
746
CHLA (|ag/L)
CHLA (|ag/L)
Turbidity (NTU)
Turbidity (NTU)
75th
95th
75th
95th
Ecoregion
Least-
moderately
Moderately-Most
Least-moderately
Moderately-Most
CPL
11.5
28.0
3.38
4.05
NAP
3.81
7.76
1.10
1.46
NPL
8.53
13.0
3.19
4.46
SAP
5.23
11.5
2.83
3.94
SPL-manmade
6.85
13.8
3.32
4.67
SPL-natural
118.4
218.7
73.5
172.0
TPL
13.9
22.7
3.70
5.38
UMW
6.70
9.60
2.13
2.89
WMT
1.83
3.04
0.760
1.43
XER
6.65
12.2
2.97
4.84
6.3 Literature cited
Herlihy, A. T., P. R. Kaufmann, and M. E. Mitch. 1991. Chemical characteristics of streams in the
Eastern United States: II. Sources of acidity in acidic and low ANC streams. Water
Resources Research 27:629-642.
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., N. C. Kamman, J. C. Sifneos, D. Charles, M. D. Enache, and R. J. Stevenson. 2013.
Using multiple approaches to develop nutrient criteria for lakes in the conterminous
USA. Freshwater Science 32:367-384. doi: 10.1899/11-097.
101
NLA 2012 Technical Report. October 2024 Version 1.1
-------
Chapter 7: Zooplankton
7.1 Background information
Zooplankton assemblages have several attributes that make them potentially useful for
assessing the ecological condition of lakes (Stemberger and Lazorchak 1994, Jeppesen et al.
2011). Zooplankton are typically the dominant pelagic consumer in lakes (in terms of both
biomass and numbers (Larsen and Christie 1993). Taxa richness tends to be high in nearly all
lakes. Zooplankton species or guild structure can respond to abiotic stressors such as
eutrophication and acidification, and possibly climate change. Zooplankton occupy an
intermediate level in the overall food web of lakes, and thus can respond to stress responses
from within lower (e.g., phytoplankton) or higher trophic levels (e.g., fish). Zooplankton taxa
demonstrate a range of life history strategies and patterns (e.g., parthenogenesis, resting eggs)
that can be related to environmental stress, both natural and anthropogenic.
The use of zooplankton assemblages in the context of bioassessment appears to be limited,
with many studies focused mainly on taxa richness and taxonomic composition changes in
response to disturbance. Gannon and Stemberger (1978) discussed the potential of using
zooplankton communities to help determine trophic state in lakes, primarily through the use of
"indicator species" that were associated with either oligotrophic or eutrophic conditions.
Sprules and Holtby (1979) and Sprules (1980) examined the utility of using metrics related to
body size and feeding ecology of zooplankton to evaluate lake condition. Duggan et al. (2001,
2002) investigated the potential for developing bioindicators of trophic state using rotifer
assemblages. Dodson et al. (2005) concluded that zooplankton assemblages are indirectly
associated with land use through effects on riparian vegetation and lake characteristics such as
typology and water chemistry. Dodson et al. (2009) examined changes in zooplankton
community structure within a set of lakes in northern Wisconsin in relation to a variety of
within-lake and watershed level characteristics (including human disturbance in the riparian
zone). Stemberger and Lazorchak (1994) calculated 14 metrics based on taxonomy, body size,
life history stage, and trophic guild in 19 lakes in the northeastern USA representing a gradient
of human disturbance, lake type, and land use. Stemberger and Miller (1998) discussed
expected changes in zooplankton assemblage trophic structure and species composition in
response to changes in the N:P ratio that might result from increased anthropogenic
disturbance.
More recently, there have been attempts to develop indices of biotic condition in lakes using
plankton assemblages, following two approaches. The multimetric approach pioneered by Karr
(e.g., Karr 1981, Karr 1991) has been implemented successfully for other assemblages (e.g., fish,
benthic invertebrates) in streams. Kane et al. (2009) combined zooplankton and phytoplankton
metrics from Lake Erie into a single multimetric index (MMI), the Planktonic Index of Biotic
Integrity, to reflect the response of the plankton to eutrophication. The second approach
(predictive model approach) compares the observed taxa collected at each site to the list of
taxa expected at that site under least disturbed conditions by means of an Observed/Expected
102
NLA 2012 Technical Report. October 2024 Version 1.1
-------
index (0/E, e.g., Wright 1995, Hawkins et al. 2000, Hawkins 2006, Hawkins et al. 2010). The
predictive modelling approach has been used successfully for other assemblages, principally
benthic invertebrates, but also fish, in streams. The National Lake Assessment 2007 (NLA 2007)
used an O/E model that combined zooplankton and phytoplankton assemblages to assess
ecological condition of lakes in the conterminous US (Yuan et al. 2008, USEPA 2009). Table 7-1
summarizes current knowledge regarding the hypothesized responses of zooplankton
assemblages to different types of disturbance.
For the NLA 2012, we decided to develop a MM I for pelagic zooplankton assemblages to assess
biological condition in lakes. We followed the approach described by Stoddard et al. (2008) to
screen candidate metrics for possible inclusion in an MMI. We then computed a large number
of MMIs based on all possible combinations of the metrics that passed the screening process,
following Van Sickle (2010), and selected the MMI that showed the best combination of
responsiveness to disturbance, repeatability, and low redundancy among component metrics.
7.2 Methods
7.2.1 Field Methods
Sample collection procedures for zooplankton are described in the NLA 2012 field operations
manual (USEPA 2012a). Field crews collected two samples at the index site (deepest area of a
lake or the midpoint of a reservoir) of each lake. The crew collected a "Coarse" sample (ZOCN)
using a 1-m long, 30-cm diameter plankton net having a mesh size of 150 |am. The crew
collected a "Fine" sample (ZOFN) using a 1-m long net with a reducing collar (20-cm diameter)
with a mesh size of 50 |am. The total tow length for each net was 5 m, with the number of tows
being dependent on the site depth. At lakes deeper than 6 m, a single 5 m tow was done. At
lakes between 3 and 6 m deep, two 2.5-m tows were done. At lakes shallower than 3 m, five 1-
m tows were done. Results from pilot studies suggested that a total tow length of 5 m would
provide sufficient numbers of taxa and organisms to develop the MMI from nearly all lakes.
103
NLA 2012 Technical Report. October 2024 Version 1.1
-------
Table 1-1. Hypothesized-responses of zooplankton assemblages to disturbance
Assemblage component or
metric
Type of disturbance
Hypothesized
response
References
Species richness
Nutrients; Agricultural
land use; riparian buffer
presence
Decrease
Gannon and
Stemberger (1978),
Dodson et al. (2005)
Native species richness,
abundance, or biomass
Invasive species
Decrease
Kane et al. (2009)
Large-sized species richness
(e.g., Daphnia spp., calanoid
copepods)
Nutrients, land use
Decrease
Stemberger and
Lazorchak(1994)
Small-sized species richness
(e.g., Ceriodaphnia, rotifers)
Nutrients, land use
Increase
Stemberger and
Lazorchak(1994)
Proportion of calanoid copepod
taxa
Nutrients
Decrease
Jeppesen et al. (2000),
Du et al. (2015)
Proportion of cyclopoid
copepod taxa
Nutrients
Increase
Jeppesen et al. (2000),
Du et al. (2015)
Rotifer assemblage composition
Nutrients, chlorophyll a,
Secchi transparency,
temperature, dissolved
oxygen
Change
Duggan et al. (2001),
(2002)
Mean size
Nutrients
Decrease
Gannon and
Stemberger (1978)
Total biomass
Nutrients
Increase
Gannon and
Stemberger (1978)
Ratio of calanoid copepods to
(cyclopoid copepods +
cladocerans)
Nutrients
Decrease
Gannon and
Stemberger (1978),
Kane et al.
(2009) ENREF 11
Biomass of rotifers and
cyclopoid copepods
Nutrients (total P)
Increase
Du et al. (2015)
Biomass of cladocerans and
cyclopoid copepods
Nutrients (total P)
Decrease
Du et al. (2015)
Biomass of small cladocerans
Catchment
development
increase
Gelinas and Pinel-Alloul
(2008), Beaver et al.
(2014)
104
NLA 2012 Technical Report. October 2024 Version 1.1
-------
Assemblage component or
metric
Type of disturbance
Hypothesized
response
References
Proportion of cladoceran
biomass
Nutrients
Decrease
Jeppesen et al. (2000),
Du et al. (2015)
Abundance of large-bodied
zooplankton
Decrease in acid
neutralization
capacity/calcium
concentrations
Decrease
Tessierand Horwitz
(1990)
Abundance of small daphnids
and cladocerans
Catchment
development
Increase
Gelinas and Pinel-Alloul
(2008), Dodson et al.
(2009), Van Egeren et
al. (2011), Beaver et al.
(2014)
Relative abundance of calanoid
copepods
Nutrients
Decrease
Brooks (1969), Gannon
and Stemberger (1978)
Relative abundance of cyclopoid
copepods and small-bodied
cladocerans
Nutrients
Increase
Brooks (1969), Attayde
and Bozelli (1998)
Omnivorous taxa richness,
abundance, or biomass
Nutrients
Increase
Stemberger and
Lazorchak (1994),
Stemberger et al.
(2001)
105
NLA 2012 Technical Report. October 2024 Version 1.1
-------
7.2.2 Laboratory Methods
Laboratory methods for zooplankton samples are described in the NLA 2012 laboratory
operations manual (USEPA 2012b). For both the ZOCN and ZOFN samples, the objective was to
subsample a sufficient volume to enumerate and identify at least 400 individuals. In the ZOCN
samples, all taxa were enumerated. In the ZOFN samples, only "small" taxa were enumerated
(Cladocera < 0.2 mm long, copepods < 0.6 mm long, rotifers, and nauplii). Veligers were not
enumerated in the ZOFN sample. Individuals were identified to species where possible. A
"Large/Rare" search of the entire subsample was done to identify larger taxa (e.g., Chaoborus,
Leptodora, Mysidae, Ostracoda, and Hydracarina). Only the presence of these taxa in the
subsample was noted (i.e., they were not enumerated).
Besides the number of individuals enumerated in the subsample (abundance), we estimated
the volume of water sampled by the tow using the tow length and the radius of the net mouth
for the sample. We used this tow volume to estimate density (no. individuals/L) of each taxon:
r Sample Vol. (mL) .. , ^
- - x Abundance
K Vol. Counted (mL)
Density = ± -
Tow Vol. (L)
The biomass (mg dry mass/L) of each taxon in a sample was estimated by measuring the length
of 20 individuals (if possible). Length was converted to a biomass factor (mg dry
mass/individual) using proprietary equations developed by the laboratory that processed the
majority of the zooplankton samples. Biomass was then calculated as:
Biomass = Density (lndiv.IL) x Biomass Factor (mg / Indiv.)
One state laboratory did not estimate biomass for their samples. For these samples, we
estimated biomass as the mean biomass of a taxon from samples collected from surrounding
states, or used a national mean (all samples collected that included the taxon) if the regional
sample size was too small.
7.3 Data Preparation
7.3.1 Data Quality Assurance
We reviewed field data to correct recording errors and, when possible, to fill in missing values,
especially for critical variable like tow length. We reviewed the raw count files from each
laboratory to correct spelling errors in taxon names, and to make the taxonomy consistent
across laboratories (using the national lab taxonomy as the standard for all labs). We used
106
NLA 2012 Technical Report. October 2024 Version 1.1
-------
range checks on count, density, and biomass estimates to identify outliers, and corrected them
if they were due to recording errors.
7.3.2 Master Tax a List
We developed a master taxa list that included all taxa identified in the ZOFN and ZOCN samples.
The master taxa list included taxonomic information (e.g., phylum, class, order, suborder,
family, subfamily, genus, species, and subspecies). Autecological information for each taxon
included feeding guild (Predator, Omnivore, or Herbivore), Cladocera size class (LARGE vs.
SMALL), based on data from Stemberger and Lazorchak (1994) and the Northeastern Lakes
Survey (Whittier et al. 2002), and a size class variable (NET_SZECLS_NEW) based on whether a
taxon was collected in the ZOCN samples vs. only in the ZOFN samples. Additional attributes for
a limited number of taxa that are included in the list but were not used include trophic
assignments from Sprules and Holtby (1979), and some trait information from Barnett et al.
(2007, 2013).
The laboratory identified 535 unique taxa in the NLA 2012 ZOCN and ZOFN samples
(variable=TAXANAME). We combined some of these unique taxa using a different variable
(TARGET_TAXON), which resulted in 481 unique taxon names as used in metric calculations.
We also had some information regarding non-native zooplankton taxa based on the USGS
Nonindigenous Aquatic Species (NAS) database (Fuller and Neilson 2015). Bosmina coregoni (or
Eubosmina coregoni), Daphnia lumholtzi, and Sinocalanus doerri were considered to be
introduced to North America. Eutymora affinis was considered to be introduced to inland
waters of the US. Pseudodiaptomus forbesi has been introduced into San Francisco Bay, and so
we considered it to be non-native if collected from nearby lakes. Arctodiaptomus dorsalis has
been introduced into lakes in Arizona, Hawaii, and Indiana.
7.3.3 Aggregations and Rarefaction of Count Data
We aggregated some values of TARGET_TAXON within a given ZOCN or ZOCN sample. We
combined copepodites and nauplii with adults of the same taxon if both were present in a
sample. If a species and a lower level taxon (i.e., subspecies, variety, or form) were both
present in a single sample, we aggregated the count data to the species level.
After aggregating at the sample level, we combined the results for each ZOCN and ZOFN sample
to create a separate site-level count file. We assumed that individuals collected in the ZOCN
samples that were also present in the ZOFN sample represented smaller individuals that passed
through the coarse-mesh net, and so we added the counts from the two samples together.
Because not all zooplankton individuals in a sample can be confidently identified to species,
there is a risk of overestimating taxa richness. For each sample, we reviewed the list of taxa to
determine whether they were represented at more than one level of resolution. For example, if
a "Daphnia sp." was collected, and it was the only representative of the genus in the sample (or
107
NLA 2012 Technical Report. October 2024 Version 1.1
-------
at the site), we assigned it as distinct. If any other members of the genus were collected, then
we considered the unknown as not distinct. We used only the number of distinct taxa in the
sample to calculate any metrics based on species richness. We calculated distinct taxa for both
the sample-level aggregated count file and the site-level count file. Taxa that were identified
(but not enumerated) during the Large/Rare search were included in calculating richness
metrics.
We created an additional count file to use for metric calculation by subjecting the sample-level
aggregated count data to a rarefaction procedure to randomly select 300 individuals per
sample (for those samples that had > 300 individuals enumerated and identified). We repeated
the sample level aggregation of taxa on the 300-count file, thus the resultant site-level count
file typically had a total count of 600 individuals. We did not calculate density on the 300-count
files, but did calculate biomass.
7.4 Zooplankton MMI Development
7.4.1 Regionalization
We divided the conterminous US into five "bio-regions" based on nine aggregated Omernik
Level III ecoregions (Omernik 1987, Stoddard 2004, Herlihy et al. 2008, Omernik and Griffith
2014) that were developed for use on NARS reporting (Figure 7-1). We combined the Northern
and Southern Appalachian regions (NAP, SAP) into a single bio region (Eastern Highlands,
EHIGH). We combined the three "plains" regions (Northern, Southern, and Temperate [NPL,
SPL, and TPL]) into a single bio-region (PLAINS). In the western US, we combined the Xeric and
Western Mountains regions (XER, WMT) into a single "Western Mountains" bio-region
(WMTNS). Despite relatively small sample sizes of least disturbed sites, we kept the Coastal
Plain (CPL) and Upper Midwest (UMW) as separate bio-regions.
108
NLA 2012 Technical Report. October 2024 Version 1.1
-------
Figure 7-1. Five aggregated bio-regions used to develop zooplankton MMIs for the 2012 National Lake Assessment
(CPL=Coastal Plains; EHIGH=Eastern Highlands, PLAINS= Plains, UMW=Upper Midwest, and WMTNS=Western
Mountains). Solid dots indicate least disturbed sites used for developing the zooplankton MMI. White circles
indicate least disturbed sites that we excluded because of atypical samples (too few taxa or number of individuals
collected).
7.4.2 Least and Most Disturbed Sites
For the zooplankton MMI, we used the same list of sites as those selected for benthic
macroinvertebrates (RT_NLA12; see Section 3.3). We identified two least disturbed sites that
appeared to have abnormal zooplankton samples. The ZOCN sample collected from McDonald
Lake, ID (NLA12ID-142) did not have any individuals in the ZOCN sample, and < 100 individuals
enumerated from the ZOFN sample. For Waldo Lake, OR (NLA12_OR-109), only 6 individuals
were collected in the ZOCN sample, and 53 individuals were collected in the ZOFN sample. We
created a new variable (RT_ZOOP) to use for zooplankton, and these two sites were assigned a
value of "B" for RT ZOOP.
109
NLA 2012 Technical Report. October 2024 Version 1.1
-------
7.4.3 Least Disturbed Sites: Calibration versus Validation
As an independent check on the MMI developed for each bio-region, we set aside a small
number of least disturbed sites as "validation" and did not include them in any MMI or metric
evaluations or performance testing. We used revisit sites (typically VISIT_N0=2) as validation
sites because they are not used in any metric or MMI testing. We then supplemented the list of
revisit sites in each region by randomly selecting sites from the list of least disturbed sites.
Where possible, we withheld ~10% of the least disturbed sites in each bio-region as validation
sites, leaving at least 15 least disturbed sites available for developing and evaluating metrics
and MMIs. For the CPL and UMW bio-regions, the small number of least disturbed sites
prevented setting aside 10% of the site for validation. Numbers of validation sites were as
follows: CPL (8), EHIGH (16), PLAINS (14), UMW (10), and WMTNS (18).
7.4.4 Candidate Metrics
We used the count data file and the master taxa list file to calculate candidate metrics. We
assigned candidate metrics to one of six metric categories, with each category reflecting a
different attribute of assemblage structure or ecological function.
The Abundance category included metrics based on abundance, density, or biomass. We
calculated these metrics separately for the ZOFN samples, the ZOCN samples, and for the
combined samples. Within the combined sample, we also calculated abundance metrics
separately for the net-based size classes (COARSE vs. FINE).
The Richness category included metrics based on taxa richness and metrics related to taxa
diversity or dominance. Richness metrics included total distinct taxa richness, number of
genera, and number of families. We calculated these metrics separately for the ZOCN, ZOFN,
and combined sample. We calculated diversity and dominance metrics for the combined
sample based on abundance, density, and biomass. Diversity metrics included Shannon-Weiner
and Simpson indices, and Hurlbert's Probability of Interspecific Encounter (PIE, Hurlbert 1971,
Jeppesen et al. 2000). We developed dominance metrics for the most dominant taxon and for
the three and five most dominant taxa in each sample.
We assigned separate categories for each of the three principal taxonomic components of the
zooplankton assemblage: Cladoceran, Copepod, and Rotifer. Metrics in these three categories
included abundance and richness metrics calculated separately for each taxonomic group. For
copepods, we also calculate the ratio of calanoids to the sum of cladocerans and cyclopoids,
following Gannon and Stemberger (1978) and Kane et al. (2009).
The sixth metric category was trophic guild. We identified three major guilds, herbivores,
omnivores, and predators. Each taxon was assigned to a trophic guild based on information
from the Northeast Lakes Survey (Stemberger and Lazorchak 1994, Stemberger et al. 2001).
110
NLA 2012 Technical Report. October 2024 Version 1.1
-------
We calculated metrics using both the entire sample and for the 300-count rarefied samples.
Metrics derived from the rarefied sample have "300" in the variable name.
For many metrics, we could calculate six different variants: the number of distinct taxa
(metn'c_NTAX), total biomass (metric_BIO), density (metr/'c_DEN), percent of individuals
(metr/'c_PIND), percent of total biomass (metric_PBIO) and percent of total density
(metr/'c_PDEN). We did not calculate density-based metrics for the 300-count rarefied samples.
Each variant was calculated based using all the individuals in the sample, and for just the native
individuals in the sample. We calculated a total of 374 candidate metrics for the whole sample
count data, and an additional 272 metrics from the 300-count rarefied sample data.
7.4.5 Final Metric Selection
We subjected all of the candidate metrics to five screening procedures, following Stoddard et
al. (2008). The first was a range test. We excluded richness metrics (metr/'c_NTAX) with a range
of <4 from further consideration. We excluded metrics based on biomass (metric_BIO), density
(metr/'c_DEN), diversity metrics, and zooplankton ratio if the 90th percentile (P90) was 0. We
excluded percentage metrics (metricJPTAX, metric_PBIO, metr/'c_PDEN) if the 75th percentile
(P75) was <10%.
The second screen was a signal to noise (S:N) test, following Kaufmann et al. (1999). We
compared the total variance observed across all sites (signal) against the variance observed for
sites that were sampled twice in the same index period (noise). We excluded metrics that had
S:N values < 1.25.
The third screen was for responsiveness to disturbance. For each metric, we calculated the t-
statistic for each metric comparing values for the set of least disturbed sites with those for the
set of most disturbed sites. We considered metrics having 1t\ values < 1.73 as non-responsive
to disturbance.
The fourth screen was to determine if metrics required adjustment for lake size. We generated
plots of linear regressions of each metric with lake area (AREA_HA) to determine if the metric
response changed with increasing lake size. For all metrics, the upper 95% prediction interval at
the minimum response value overlapped the lower 95% prediction interval at the maximum
response value, indicating there was no significant effect of lake size on the metric response.
For each bio-region, we used the set of candidate metrics that had passed the four screens
describe above to develop candidate MMIs. We constrained the MMIs to contain at least one
metric from each of the six metric categories (abundance, richness, crustacean, copepod,
rotifer, and trophic). If no metrics within a category passed all of the screens, we selected one
or more metrics that had the highest t values and had S:N values near 1 (if possible). Values of
S:N <1 indicate that that variation within a site is equal to or greater than the variation among
sites, so the metric cannot discriminate among sites.
ill
NLA 2012 Technical Report. October 2024 Version 1.1
-------
Finally, we evaluated the redundancy among candidate metrics using correlation analysis.
Historically, we have evaluated redundancy based on the establishing a maximum allowable
correlation coefficient (r) between two metrics (e.g., r >0.7; Stoddard et al. 2008). Van Sickle
(2010) demonstrated that MMIs containing a suite of metrics that have a low average
correlation among them perform better that simply using a maximum threshold value of r to
reduce redundancy within the suite of metrics. We included correlations in the procedure
below, computing correlations among metrics for each candidate MMI, rather that evaluating
individual input metrics within a category and choosing only non-redundant metrics to include
in a final MMI, as described by Stoddard et al. (2008).
Candidate metrics that we considered for inclusion into an MMI for each of the five bio-regions
are listed in Section 7.10. For each bio-region, we computed MMIs from all possible
combinations of candidate metrics from the six categories. We evaluated each MMI for
responsiveness (t test of least disturbed vs. most disturbed sites) and repeatability (S:N). For
each bio-region, we selected MMI that had a combination of high t value, a reasonable value
for S:N, low mean r among the suite of metrics, and, when possible, a maximum value of r for
the suite of metrics that was <0.7.
7.4.6 Metric Scoring
We followed the approach described by Stoddard et al. (2008) to transform metric responses
into a metric score that ranged between 0 and 10 (Blocksom 2003). For positive metrics (i.e., t
>0), we used the 5th percentile of all sites in the bio-region as the "floor" value, and the 95th
percentile of the set of least disturbed sites as the "ceiling" value. For negative metrics (i.e., t
<0), we used the 5th percentile of least disturbed sites in the bio-region as the "floor" value, and
the 95th percentile of all sites as the "ceiling" value. When metric response values were less
than the floor value, we assigned a score of 0. When metric response values were greater than
the ceiling, we assigned a score of 10. We estimated scores for response values that were
between the floor and ceiling values by linear interpolation.
We calculated the final MMI score for each bio-region by summing the six component metric
scores, and then multiplying by 100/6. This resulted in an MMI score that ranged between 0
and 100.
7.5 Zooplankton MMI Metric Composition and Performance
7.5.1 Coastal Plain MMI
The component metrics for the Coastal Plain MMI are presented in Table 7-2. Information
related to the performance of the Coastal Plain MMI are presented in Section 7.6. Figure 7-2
compares the distributions of the six metrics in least disturbed vs. most disturbed sites. Three
metrics are "negative" metrics (t <0) values, indicating that the response is greater in most
disturbed sites compared to least disturbed sites. No abundance or cladoceran metrics passed
both the responsiveness and repeatability screens. The abundance metric (FINE_BIO [biomass
112
NLA 2012 Technical Report. October 2024 Version 1.1
-------
of smaller-sized taxa]) had a t value and an S:N value that were just below the screening
criterion. The cladoceran metric (SIDID_PIND [percent of individuals of the cladoceran family
Sididae]) had an S:N value that was below the screening criterion.
The abundance metric (FINE_BIO), the cladoceran metric (SIDID_PIND), the richness metric
(FAM300_NAT_NTAX), and the trophic metric (OMNI_PTAX) responded to disturbance as
expected (Figure 7-2; Table 7-1). The copepod metric (DOM1_300_COPE_PBIO) and the rotifer
metric (COLLO_PBIO) decreased in response to disturbance (Figure 7-2). Declines in the
proportion of total biomass contributed by either dominant copepods or a subgroup of rotifers
might be expected if the total richness, abundance, and total biomass of cyclopoid copepods
and rotifers increased with disturbance (Table 7-1).
113
NLA 2012 Technical Report. October 2024 Version 1.1
-------
Table 7-2. Component metrics of the zooplankton MMI for the Coastal Plain bio-region. Evaluations for
responsiveness (t-value) and signal:noise (S:N) based on index visits and do not include least disturbed "validation"
sites. Negative values for t indicate response is greater in most disturbed sites vs. least disturbed sites. Metrics
having values marked with an asterisk were among the best performing metric of that category, but failed one or
more evaluation screens. Floor and ceiling values are used to derive a score for the metric. See Section 7.10 for
metric descriptions.
Metric Type
Metric Variable Name (floor, ceiling)
(value
S:N (bio-region)
Abundance/Size
FINE_BIO (2.913623, 173.279784)
-1.67*
1.2*
Cladoceran
SID1 D_PI ND (0, 24.88)
-1.80
0.5*
Copepod
DOM1_300_COPE_PBIO (45.90, 100)
+1.82
1.9
Richness/Diversity
FAM300_NAT_NTAX (5, 15)
+2.66
1.8
Rotifer
COLLO_PBIO (0, 5.90)
+1.85
7.6
Trophic
OMNI_PTAX (10.53, 47.06)
-3.35
4.3
ABUNDANCE
CLADOCERAN
COPEPOD
o
m
400
300
j 200
100
0
100
80
t 60
q'
5 40
-------
7.5.2 Eastern Highlands MM!
The component metrics for the Eastern Highlands MMI are presented in Table 7-3. Information
related to the performance of the Eastern Highlands MMI are presented in Section 7.6. Figure
7-3 compares the distributions of the six metrics in least disturbed vs. most disturbed sites. The
suite of metrics includes both positive (2) and negative (4) metrics. No richness metrics passed
the screens for responsiveness or repeatability. The richness metric (ZOCN300_FAM_NTAX) had
a t value (1.64) just below the screening criterion, while the S:N value (0.3) was well below the
screening criterion.
The cladoceran metric (SMCLAD_PBIO), the richness metric (COARSE_NAT_PTAX ), the rotifer
metric (ROT_PBIO), and the trophic metric (OMNI300_PTAX) responded as expected to
increased disturbance (Figure 7-3; Table 7-1). The abundance metric (ZOCN_DEN) and the
copepod metric (COPE_NAT_DEN) both increased in response to disturbance (Figure 7-3). An
increase in cyclopoid copepods expected with increased disturbance (Table 7-1) would help to
explain the observed response in both of these metrics.
7.5.3 Plains MMI
The component metrics for the Plains MMI are presented in Table 7-4. Information related to
the performance of the Plains MMI are presented in Section 7.6. Figure 7-4 compares the
distributions of the six metrics in least disturbed vs. most disturbed sites. The MMI was
comprised of two negative and four positive metrics. All metrics passed the screening criteria
for both responsiveness and repeatability.
The copepod (COPE_RATIO_300_BIO), richness (FAM300_NAT_TAX), and the trophic
(COPE_HERB_PDEN) metrics responded as expected to increased disturbance (Figure 7-4; Table
7-1). The abundance (FINE300_NAT_PBIO), cladoceran (SMCLAD_NAT_PIND), and the rotifer
(ROT_NTAX) metrics all decreased with response to increased disturbance. If herbivorous
cyclopoid copepods are becoming more dominant in terms of richness, abundance, and
biomass, that may result in a decline in the relative biomass of individuals collected in the fine-
mesh net (principally rotifers), a decline in the relative abundance of smaller cladocerans, and a
decline in rotifer taxa richness.
115
NLA 2012 Technical Report. October 2024 Version 1.1
-------
Table 7-3. Component metrics of the zooplankton MMI for the Eastern Highland bio-region. Evaluations for
responsiveness (t-value) and signal:noise (S:N) based on index visits and do not include least disturbed "validation"
sites. Negative values for t indicate response is greater in most disturbed sites vs. least disturbed sites. Floor and
ceiling values are used to derive a score for the metric. See Section 7.10 for metric descriptions.
Metric Type
Metric Variable Name (floor, ceiling)
(value
S:N (bio-region)
Abundance/Size
ZOCN_DEN (0.096200402, 115.2464653)
-1.89
7.1
Cladoceran
SMCLAD_PBIO (0, 51.41)
-2.84
1.4
Copepod
COPE_NAT_DEN (7.5388,385.279)
-1.74
1.5
Richness/Diversity
COARSE_NAT_PTAX (22.22,57.14)
+1.64*
0.3*
Rotifer
ROT_PBIO (1.69, 89.89)
-1.89
1.3
Trophic
OMNI300_PTAX (12.50, 43.75)
-2.60
1.5
ABUNDANCE
CLADOCERAN
COPEPOD
250
200
8 100
N
50
0
m
O-
q'
<
_i
o
-------
Table 7-4, Component metrics of the zooplankton MM I for the Plains bio-region. Evaluations for responsiveness it-
value) and signal:noise (S:N) based on index visits and do not include least disturbed "validation" sites. Negative
values for t indicate response is greater in most disturbed sites vs. least disturbed sites. Floor and ceiling values
are used to derive a score for the metric. See Section 7,10 for metric descriptions.
Metric Type
Metric Variable Name (floor, ceiling)
(value
S:N (bio-region)
Abundance/Size
FINE300_NAT_PBIO (0.66, 85.12)
+1.74
6.2
Cladoceran
SMCLAD_NAT_PIND (0, 49.03)
+3.11
1.8
Copepod
COPE_RATIO_300_BIO (0, 62.81)
+2.41
3.0
Richness/Diversity
FAM300_NAT_NTAX (5, 14)
+2.21
2.6
Rotifer
ROT_NTAX (3, 17)
+2.63
1.7
Trophic
COPE_HERB_PDEN (0, 21.07)
-2.13
13.0
ABUNDANCE
CLADOCERAN
COPEPOD
m
o.
100
80
60
40
20
LD MD
60
Q 50
°"l 40
30
Q
<
20
O
oj 10
LD MD
80
I 60
O
i= 40
g
w1
o. 20
O
O
LD MD
RICHNESS
ROTIFER
TROPHIC
16
12
8
4
0
20
15
10
5
0
30
20
w
Q
Q.
m'
QL
LU
X
LU 10
Q.
O
o
LD MD
LD MD
LD MD
Figure 7-4, Distribution of six component metrics of the zooplankton MMI for the Plains bio-region in least
disturbed versus most disturbed sites. Dots indicate the 5th and 95th percentiles.
117
NLA 2012 Technical Report. October 2024 Version 1.1
-------
7.5.4 Upper Midwest MM!
The component metrics for the Upper Midwest MMI are presented in Table 7-5. Information
related to the performance of the Upper Midwest MMI are presented in Section 7.6. Figure 7-5
compares the distributions of the six metrics in least disturbed vs. most disturbed sites. The
MMI is composed of four negative and two positive metrics. No abundance metrics passed the
screen for responsiveness. The abundance metric (ZOCN_NAT_PDEN [the percent of total
density represented by native individuals in the coarse net sample]) had a t-value that is below
the screening criteria for responsiveness. Repeatability (S:N values) of the metrics in this bio-
region are higher than in other bio-regions, but interpretation of the S:N values is constrained
somewhat by a limited number of revisit samples (5).
Only three of the six metrics responded to disturbance as expected (Figure 7-5; Table 7-1). The
abundance metric (TOTL_NAT_PIND) showed a slight decrease with disturbance, indicating the
effect of non-native taxa in this bio-region. The rotifer metric (DOMl_ROT_PBIO) indicates a
reduction in species richness (i.e., increased dominance by one or a few taxa) with increased
disturbance. The trophic metric (COPE_HERB300_PBIO) indicates an increase in herbivorous
taxa (possibly cyclopoid copepods) with increased disturbance. The cladoceran metric
(BOSM300_NAT_PTAX) was expected to increase with increased disturbance, but the response
may reflect a larger increase in the taxa richness of other forms of smaller zooplankton (e.g.,
cyclopoid copepods). The copepod metric (CALAN300_NAT_BIO) indicates an increase in larger
forms of zooplankton. Such a response might occur if the least disturbed population of lakes is
dominated by oligotrophic lakes that do not support large populations of zooplankton. The
richness metric (FINE_PTAX) decreased in response to disturbance. This response may be
similar to that observed for the cladoceran metric, where other forms of smaller zooplankton
(e.g., cyclopoid copepods) increase in tax richness compared to rotifers, which are the
dominant taxa collected in the fine-mesh net.
7.5.5 Western Mountains MMI
The component metrics for the Western Mountains MMI are presented in Table 7-6.
Information related to the performance of the Western mountains MMI are presented in
Section 7.6. Figure 7-6 compares the distributions of the six metrics in least disturbed vs. most
disturbed sites. The MMI is composed of three negative and three positive metrics. No richness
metrics passed the screen for responsiveness. The richness metric (ZOFN300_NTAX [Number of
distinct taxa in the 300-count rarefied sample from the fine net sample]) had a t value that was
below our acceptance criteria for responsiveness.
The abundance (COARSE300_NAT_PBIO), cladoceran (LGCLAD300_NAT_PTAX), richness
(ZOFN300_NTAX), rotifer (PLOIMA_PTAX), and trophic (COPE_OMNI_PTAX) metrics responded
as expected to increased disturbance (Figure 7-6, Table 7-1). The copepod metric
(COPE300_BIO) would respond as expected to disturbance if the increase in biomass was due
primarily to smaller forms (e.g., cyclopoid copepods).
118
NLA 2012 Technical Report. October 2024 Version 1.1
-------
Table 7-5. Component metrics of the zooplankton MMI for the Upper Midwest bio-region. Evaluations for
responsiveness (t-value) and signal:noise (S:N) based on index visits and do not include least disturbed "validation"
sites. Negative values for t indicate response is greater in most disturbed sites vs. least disturbed sites. Metrics
having values marked with an asterisk were the best performing metric of that category, but failed one or more
evaluation screens. Floor and ceiling values are used to derive a score for the metric. See Section 7.10 for metric
descriptions.
Metric Type
Metric Variable Name (floor, ceiling)
(value
S:N (bio-region)
Abundance/Size
TOTL_NAT_PIND (96.75, 100)
+1.47*
Noise=0
Cladoceran
BOSM300_NAT_PTAX (0, 12.5)
+2.73
1.4
Copepod
CALAN300_NAT_BIO (0, 58.429968)
-2.17
9.2
Richness/Diversity
FINE_PTAX (37.50, 77.78
+1.87
1.4
Rotifer
DOMl_ROT_PBIO (25.03, 93.60)
-2.46
3.5
Trophic
COPE_HERB300_PBIO (0.19, 59.65)
-1.96
5.1
100
§ 90
E
§ 80
.I
P 70
60
ABUNDANCE
T
LD MD
20
X
<
!r 15
10
-------
Table 7-6. Component metrics of the zooplankton MMI for the Western Mountains bio-region. Evaluations for
responsiveness (t-value) and signal:noise (S:N) based on index visits and do not include least disturbed "validation"
sites. Negative values for t indicate response is greater in most disturbed sites vs. least disturbed sites. Metrics
having values marked with an asterisk were the best performing metric of that category, but failed one or more
evaluation screens. Floor and ceiling values are used to derive a score for the metric. See Section 7.10 for metric
descriptions.
Metric Type
Metric Variable Name (floor, ceiling)
(value
S:N (bio-region)
Abundance/Size
COARSE300_NAT_PBIO (10.94, 99.26)
+1.88
5.7
Cladoceran
LGCLAD300_NAT_PTAX (0, 30.385)
+2.12
2.3
Copepod
COPE300_BIO (0.074, 150.462701)
-2.76
2.0
Richness/Diversity
ZOFN300_NTAX (3, 15)
-1.69*
1.9
Rotifer
PLOIMA_PTAX (20, 70.835)
+2.28
4.3
Trophic
COPE_OMNI_PTAX (0, 22.22)
-2.52
1.5
m
Q.
100
80
60
CO 40
LLl
-------
7.6 Zooplankton MMI Performance
We evaluated each of the five regional MMIs in several ways.
7.6.1 Calibration versus Validation Sites
To provide an independent assessment of MMI performance, we compared the distribution of
MMI scores between the set of validation sites (which we did not use in MMI development) and
the calibration sites using a t-test. The null hypothesis was that the mean values of the two
groups would be equal. Mean values of the two groups were not significantly different (p <
0.05) for any bio-region (Table 7-7). Figure 7-7 shows the distribution of MMI scores between
the calibration and validation sites in the five bio-regions.
7.6.2 Precision of MMIs based on Least Disturbed Sites
We evaluated the precision of the regional MMIs using the sets of least disturbed calibration
sites, following Van Sickle (2010). We rescaled the MMI scores in each bio-region by dividing
each site score by the mean MMI score, which resulted in a mean rescaled MMI score of 1. We
calculated the standard deviation of the rescaled MMI scores (Table 7-7). The smaller the
standard deviation, the more precise the index is, and the better the ability to detect sites that
are not in least disturbed condition. Standard deviations were generally small except for the
Plains, where site MT-104 had a large influence.
7.6.3 Responsiveness, Redundancy, and Repeatability of Zooplankton MMIs
We compared the MMI scores from the set of least disturbed sites to the set of most disturbed
sites (excluding the validation sites) using a t-test. We calculated the S:N values using the set of
revisit sites within each bio-region (again excluding the validation sites). Table 7-8 presents the
results of these tests, along with the maximum and average correlations observed for the
component metrics. The tvalues for responsiveness are comparable to MMIs developed for
other resource types and assemblages (e.g., benthic invertebrates). Figure 7-8 shows the
distribution of MMI scores between least- and most disturbed sites in the five bio-regions.
SignakNoise values are comparable to other MMIs that have been developed for other
assemblages. The S:N value for the UMW bio-region is constrained by the small number of
revisit sites (5) available. When MMI scores from all bio-regions are considered, the national-
level estimate of S:N is 6.7.
7.6.4 Responsiveness to a Generalized Stressor Gradient
We performed an additional evaluation of the MMIs for responsiveness to disturbance. We
performed principal components analysis (PCA) on the set of chemical, physical habitat, and
visual assessment stressor variables used to screen for least disturbed and most disturbed sites.
Chemical stressor variables included chloride, sulfate, turbidity, and acid neutralizing capacity
121
NLA 2012 Technical Report. October 2024 Version 1.1
-------
(CL, S04, TURB, and ANC, respectively). Habitat stressor variables (Kaufmann et al. 2014; see
Chapter 5 for descriptions and calculations) included shoreline disturbance due to non-
agricultural activities (hiiNonAg), shoreline disturbance due to agricultural activities (hiiAg_Syn),
and the proportion of shoreline stations with at least one type of disturbance present in either
the littoral zone or shoreline plots (hifpAnyCirca_syn). Stressor variables from the visual
assessment included the intensity of observed types of agricultural activities (AGR_SCORE),
intensity of observed types of residential activities (RES_SCORE), and intensity of observed
types of commercial and industrial activities, excluding evidence of fire (IND_NOFIRE). We
transformed the chemical variables (logio[x+l]), and standardized all variables to mean=0 and
variance=l. The first PCA axis explained 38% of the total variance, and the highest variable
loadings were for the chemical and agricultural-related habitat variables. The second PCA axis
explained an additional 18% of the total variance, and the highest variable loadings were for
the non- agricultural habitat variables and the intensity of residential activities. Linear
regression of the MMI score versus the PCA axis 1 scores yielded an r2 of 0.32 (r= 0.56) for PCA
axis 1 (Figure 7-9), and 0.006 for PCA axis 2 scores. These results indicate the zooplankton MMI
is principally responsive to nutrient conditions resulting from agricultural disturbance, and less
responsive to other types of habitat disturbance.
122
NLA 2012 Technical Report. October 2024 Version 1.1
-------
Table 7-7. Results of independent assessment and precision tests of NLA 2012 zooplankton MMIs based on least
disturbed sites. None of the t-values were significant at p = 0.05. Standard deviations were calculated using only
calibration sites.
Calibration vs. Validation
Standard Deviation
Sites
of Standardized
Regional MMI
(t-value)
MMI scores
Coastal Plains (CPL)
0.73
0.164
Eastern Highlands (EHIGH)
-1.08
0.116
Plains (PLAINS)
1.87
0.332
Upper Midwest (UMW)
0.86
0.115
Western Mountains (WMTNS)
0.49
0.122
100
^ n> *
-Ss vr t^y
A^ A^
; _X' a' A' .A' A' a'
-------
Table 7-8. Results of responsiveness, redundancy, and repeatability tests for NLA 2012 zooplankton MMIs.
Metrics having values marked with an asterisk were the best performing metric of that category, but failed one or
more evaluation screens.
Redundancy
Responsiveness
(Mean pairwise
t-test of Least
Redundancy
correlation
Repeatability
disturbed vs.
(Maximum pairwise
among
Signal: Noise
Most disturbed
correlation among
component
ratio based on
Bio-Region
Sites
component metrics)
metrics)
revisit sites
Coastal Plains
(CPL)
4.68
0.58
0.26
2.7
Eastern
Highlands
(EHIGH)
5.42
0.48
0.17
2.5
Plains (PLAINS)
4.47
0.72*
0.25
3.6
Upper Midwest
(UMW)
5.84
0.48
0.26
19.0
Western
Mountains
(WMTNS)
6.30
0.63
0.24
3.1
100
Least Disturbed (LD) vs. Most Disturbed (MD)
(Index visits only. For LD: Calibration sites only)
v
-------
7.6.5 Effect of Natural Drivers and Tow Length on MM! Scores
The set of lakes sampled for the 2012 NLA included both natural and man-made lakes, and
included a wide range of sizes (as estimated by lake area as represented in NHD). In addition,
the sampling protocol did not include a vertical tow through the entire water column. Any one
of these factors might produce a bias in the MMI scores that would require assessing ecological
condition separately for one or more of these groups of lakes (natural vs. man-made, small vs.
large lakes, or shallow versus deeper lakes). We use the set of least disturbed sites (calibration
and validation) to evaluate the potential differences in MMI scores in these groups of lakes.
7.6.5.1 Lake Origin
We compared the distributions of MMI scores in least disturbed natural lakes vs. man-made
reservoirs for each of the five bio-regions (Figure 7-10). The distributions are similar within each
bio-region except the WMTNS, where man-made lakes appear to have much lower MMI scores
than natural lakes. In the Coastal Plain, man-made lakes have higher MMI values than natural
lakes, but interpretation is constrained by the small number of least disturbed natural lakes
(n=3). In the WMTNS, the sample size for least disturbed man-made lakes is relatively small
(n=16) and is influenced to some extent by the presence of outliers with low MMI scores (Figure
7-10). We did not feel the observed differences were large enough to treat MMI scores from
lakes and reservoirs differently in terms of setting thresholds for condition.
125
NLA 2012 Technical Report. October 2024 Version 1.1
-------
MMI=-11.70(PCA Axis 1 Score) +55.94 (Adj. R2=0.41)
O
i_
o
o
(0
100
80 -
60 -
40 --
C
O
+->
c
ro
§ 20
N
-4 -2 0 2 4 6
PCA Axis 1 Score
Figure 7-9. Linear regression of NLA 2012 Zooplankton MMI scores vs. first axis score from principal components
analysis (PCA) based on chemical, habitat, and visual assessment stressor variables used to screen least- and most
disturbed sites.
126
NLA 2012 Technical Report. October 2024 Version 1.1
-------
LEAST DISTURBED SITES
0)
o
o
(0
100
.8 80
60
40
20
0
o
c
JO
Q.
O
o
N
CPL EHIGH PLAINS UMW WMTNS
(20,3) (26,35) (23,16) (0,31) (17,43)
Bio-region
Figure 7-10. NLA 2012 Zooplankton MMI scores of man-made (shaded boxes) versus natural lakes (unshaded
boxes) for least disturbed sites in five bio-regions. See Figure 7-1 for bio-region codes. Sample sizes for each type
are in parentheses. Dots indicate 5th and 95th percentiles.
7.6.5.2 Lake Size
We examined the set of least disturbed sites for evidence of difference in MMI scores due to
lake size (Figure 7-11). We noted earlier than we did not have to calibrate individual metrics for
lake size (Section 7.4.5). Distributions of MMI scores were similar in median values and ranges
for all size classes except for the largest (> 500 ha), which had a similar median but a wider
range.
127
NLA 2012 Technical Report. October 2024 Version 1.1
-------
Least Disturbed Sites
0)
o
o
(/)
c
o
100
80
60 -
3 40
c
re
Q.
O
o
N
20 -
0 -
# ^
kS>v *0
,° ^ ^ JT j?
+' f $'
n -7
Lake Area Class (ha)
Figure 7-11. Zooplankton MMI scores versus lake size class within least disturbed lakes of the 2012 NLA. Sample
sizes are in parentheses. Dashed lines are mean values. Dots indicate the 5th and 95th percentiles.
128
NLA 2012 Technical Report. October 2024 Version 1.1
-------
7.6.5.3 Site Depth
We had some concerns that the 5-m tow length used to collect zooplankton samples might be
less effective in deeper lakes, where larger taxa may migrate to deeper waters during the day
to avoid fish predation, and thus be underrepresented in the samples. We examined MMI
scores in least disturbed sites as they related to the depth of the index site where samples were
collected (Figure 7-12). There was no apparent pattern in relation to site depth, and the
distribution of MMI scores was similar for least-disturbed lakes that were < 6 m deep (the
maximum depth where the tow length encompassed the entire water column), and for lakes >
6 m deep (where part of the water column would not be subject to sampling).
7.7 Thresholds for Assigning Ecological Condition
We followed Stoddard et al. (2008) in using the set of least disturbed sites (including calibration
and validation sites) to set threshold values to assign ecological condition based on the
zooplankton MMI. We used the 25th percentile value to distinguish sites in "good" condition
(similar to least disturbed) from sites in "fair" condition (slightly deviant from least disturbed).
We used the 5th percentile value to distinguish sites in "fair" condition from sites in "poor"
condition (different from least disturbed).
Because of varying quality of least disturbed sites within each bio-region, we adjusted the
percentiles using the same process as for the NLA 2012 benthic macroinvertebrate indicator
(Herlihy et al. 2008; see Chapter 4). We performed principal components analysis (PCA) based
on all variables used in the screening of least disturbed sites (TP, TN, CI, S04, Turbidity, physical
habitat disturbance indices, and assessment indices). We transformed values (logio[x] or
logio[x+l]) before analysis. Initially, there were 214 least disturbed sites for zooplankton. We
performed a linear regression of zooplankton MMI score versus the score for the first principal
component. Before calculating thresholds, we performed a 1.5*IQR outlier analysis on the set
of least disturbed site MMIs to remove outliers. We excluded three sites based on this test (one
each in the CPL EHIGH, and WMTNS), leaving 211 least disturbed sites. Of the 211 least
disturbed sites, 9 sites (8 in WMTNS and 1 in PLAINS) were missing data required for the PCA
analysis, and so do not have principal component scores (mostly missing turbidity in CA). Thus,
there were a total of 202 sites used for the threshold adjustment statistical analysis.
The best regression model had two different slopes and separate intercepts for each bio-region
(Table 7-9). The pooled model RMSE was 10.86. We used a pooled RMSE (based on all sites) to
provide an adequate sample size for estimating the distribution of MMI scores about the
intercept value for each bio-region. The regression models for the CPL, EHIGH and UMW bio-
regions had no relationship with disturbance and their slopes were set to zero. The slopes for
the PLAINS and WMTNS bio-regions were similar enough that a single value (-6.113) was used
for both. The intercepts were 74.16 in the CPL, 78.75 in the EHIGH, 74.10 in the UMW, 58.32 in
129
NLA 2012 Technical Report. October 2024 Version 1.1
-------
Least Disturbed Sites
LU
0£
O
o
o
H
*
Z
<
_l
Q.
O
o
N
10 20 30 40 50
INDEX SITE DEPTH (m)
Reference line=6m (maximum tow length=5m)
60
Least Disturbed Sites
UJ 100
O
o
6 m
(113) (97)
Depth Class
Figure 7-12. Zooplankton MMI scores versus site depth for least disturbed sites. Upper panel shows MMI scores
versus actual site depth. The reference line of 6 m separates shallower lakes where the entire water column was
sampled and deeper lakes where part of the water column was not sampled. The lower panel compares
distribution of MMI scores in shallow lakes (<6 m; n=113) versus deeper lakes (> 6 m, n=97). Dots indicate the 5th
and 95th percentiles.
130
NLA 2012 Technical Report. October 2024 Version 1.1
-------
Table 7-3, Linear regression statistics of zooplankton MM I scores versus pea-based disturbance score for each bio-
region.
Bio-Region
Slope
Intercept
RMSE (Pooled)
Coastal Plains
0
64.94
10.01
(CPL
Eastern Highlands
0
76.50
10.01
(EHIGH
Plains (PLAINS)
-6.143
54.55
10.01
Upper Midwest
0
72.49
10.01
(UMW)
Western
-6.143
63.48
10.01
Mountains
(WMTNS)
Table 7-10, Thresholds for assigning ecological condition for zooplankton MM I scores based on the distribution of
least disturbed sites in five bio-regions. Poor condition indicates a site is different from least disturbed condition.
Fair condition indicates a site is somewhat deviant from least disturbed condition. Good condition indicates a site
is similar to least disturbed condition. Values in bold (adjusted based on the regressions of MM I scores to PCA-
based disturbance scores) are used to assign condition.
Range of MMI
scores in Least
Good/Fair Threshold (P25)
Fair/Poor Threshold (P5)
disturbed
Bio-Region
na
Adjusted
Unadjusted
Adjusted
Unadjusted
Sites
Coastal
22
57.7
59.4
48.4
49.7
38.80 to 94.47
Plains
(CPL)
Eastern
59
57.2
58.0
60.0
57.3
46.37 to 92.62
Highlands
(EHIGH)
Plains
37
42.4
37.8
33.2
17.4
4.42 to 78.57
(PLAINS)
Upper
31
73.3
73.7
56.0
58.0
53.37 to 92.01
Midwest
(UMW
Western
51
69.2
63.6
54.6
53.9
31.24 to 97.94
Mountains
(WMTNS)
' Number of least disturbed sites remaining after exc
-based disturbance scores.
uding statistical outliers and sites with missing PCA
131
NLA 2012 Technical Report. October 2024 Version 1.1
-------
the PLAINS, and 74.39 in the WMTNS. Table 7-10 shows both the raw (unadjusted sample) 5th
and 25th percentiles and the regression model adjusted percentiles that we are using as the
MMI thresholds. In three bio-regions (CPL, EHIGH, and UMW), the adjustment resulted in as
slight lowering (< 2 points) of the Good/Fair threshold value. In the PLAINS and WMTNS bio-
regions, the Good/Fair threshold values were increased (4.6 to 5.6 points). Adjustment
lowered the Fair/Poor threshold values in the CPL, EHIGH, and UMW bio-regions by 2.7 to 6.7
points. The Fair/Poor threshold value was increased by 14.5 points in the PLAINS bio-region,
and 3.9 points in the WMTNS bio-region.
7.8 Discussion
We were able to develop regional MMIs for pelagic zooplankton assemblages that were
sufficiently responsive and repeatable to allow us to assess ecological condition for the 2012
NLA. The zooplankton assemblage appears to be responsive principally to disturbance resulting
from increased nutrients and from increases in agricultural-related activity, which is consistent
with previous studies (e.g., Gannon and Stemberger 1978, Stemberger and Lazorchak 1994).
We did not observe a strong response of the zooplankton assemblage to shoreline habitat
disturbance, as has been noted by others (e.g., Stemberger and Lazorchak 1994).
Based on our evaluations, the zooplankton MMIs we developed do not appear to be affected by
lake origin (except possibly in the WMTNS), lake size, or by the use of a restricted tow length
that does not collect individuals which might be occupying waters deeper than 6 m. Presence of
these effects requires dealing with different types or sizes of lakes differently, either in terms of
developing separate MMIs for them, or in setting different threshold values for them based on
a very small number of least disturbed lakes.
The regional zooplankton MMIs we developed for the 2012 NLA do have some limitations.
Samples must be collected using the same protocols and nets. Individuals were identified to the
lowest practical taxon (with species being the target level). However, total richness metrics did
not perform well in terms of responsiveness or repeatability, so coarser level identification may
be possible in the future. However, coarser-level identification will constrain the development
of predictive models based on taxa richness (O/E models; see Section 7.1), and would reduce
the precision associated with biomass estimates due to lumping of taxa to coarser levels. While
many richness metrics may not have performed well, many density- and biomass-based metrics
did, thus laboratory analyses require determination of biomass, which increases costs and
requires the use of conversion equations that may not be easily available to outside users.
In some bio-regions, our requirement for inclusion of at least one metric from each of the six
categories resulted in using metrics that were either not very responsive to disturbance or were
not very repeatable, and, in some bio-regions, including metrics that were highly correlated.
Eliminating the poor-performing metrics from the suite of metrics did not appear to improve
the MMI performance, so we retained them for consistency across bio-regions. Moreover, in
those cases where we had a pair of highly correlated metrics, the mean correlation among all
132
NLA 2012 Technical Report. October 2024 Version 1.1
-------
pairs of component metrics was low, so we did not feel the correlation unduly influenced the
performance of the MMI (Van Sickle 2010). Future research might eliminate the requirement of
metric categories and just include the best performing metrics regardless of metric category to
determine if the resulting MMIs prove to be more responsive and repeatable than those
developed for the 2012 NLA.
We observed that the responses of some metrics were contradictory to what we expected with
increased disturbance (Table 7-1). However, little information is available, other than
generalization about taxa richness and assemblage composition, and possibly feeding ecology,
to support or refute the responses we observed in metrics related to density or biomass.
We also worked with a limited set of autecological information for the zooplankton taxa that
were collected (essentially taxonomic and coarse-level feeding ecology). Additional information
is available for a limited number of taxa (e.g., Sprules and Holtby 1979, Barnett et al. 2007,
2013, Vogt et al. 2013), but it is uncertain if this information can be assigned to related taxa.
We did not have any information regarding the tolerance of zooplankton taxa either to specific
stressors or to a generalized disturbance variable. These values have been developed for large
numbers of fish taxa as well as benthic invertebrate taxa (Yuan 2004, Carlisle et al. 2007,
Whittier et al. 2007, Meador et al. 2008, Whittier and Van Sickle 2010), and for rotifers in New
Zealand (Duggan et al. 2001). Data are available from the 2007 NLA that would allow tolerance
values to be developed and applied to the 2012 NLA, albeit at a coarser taxonomic level than
species, and tolerance values derived from the 2012 NLA would be available for future
assessments.
Finally, it is well known that predation by fish and larger invertebrate predators can affect
zooplankton assemblages. Predation by planktivorous fish can result in smaller-sized taxa
becoming more abundant. The 2012 NLA did not collect any detailed information about fish
assemblages, so interpretations of response of metrics or the MMI to increased nutrients may
be confounded with an increase in the number offish species (including planktivorous species)
that might accompany an increase in nutrients and a shift in the temperature regime from cold
water to warm water.
7.9 Literature cited
Attayde, J. L., and R. L. Bozelli. 1998. Assessing the indicator properties of zooplankton
assemblages to disturbance gradients by canonical correspondence analysis. Canadian
Journal of Fisheries and Aquatic Sciences 55:1789-1797.
Barnett, A. J., K. Finlay, and B. E. Beisner. 2007. Functional diversity of crustacean zooplankton
communities: towards a trait-based classification. Freshwater Biology 52:796-813.
133
NLA 2012 Technical Report. October 2024 Version 1.1
-------
Barnett, A. J., K. Finlay, and B. E. Beisner. 2013. Functional diversity of crustacean zooplankton
communities: towards a trait-based classification. Freshwater Biology, 52, 796-813. DOI:
10.1111/j. 1365-2427.2007.01733.x. Freshwater Biology 58:1755-1765.
Beaver, J. R., C. E. Tausz, T. R. Renicker, G. C. Holdren, D. M. Hosier, E. E. Manis, K. C. Scotese, C.
E. Teacher, B. T. Vitanye, and R. M. Davidson. 2014. The late summer crustacean
zooplankton in western U.S.A reservoirs reflects ecoregion, temperature and latitude.
Freshwater Biology 59:1173-1186.
Blocksom, K. A. 2003. A performance comparison of metric scoring methods for a multimetric
index for Mid-Atlantic Highlands streams. Environmental Management 31:0670-0682.
Brooks, J. L. 1969. Eutrophication and changes in the composition of the zooplankton. Pages
236-255 Eutrophication: causes, consequences, correctives. National Academy of
Sciences, Washington, DC.
Carlisle, D. M., M. R. Meador, S. R. Moulton II, and P. M. Ruhl. 2007. Estimation and application
of indicator values for common macroinvertebrate genera and families of the United
States. Ecological Indicators 7:22-33.
Dodson, S. I., R. A. Lillie, and S. Will-Wolf. 2005. Land use, water chemistry, aquatic vegetation,
and zooplankton community structure of shallow lakes. Ecological Applications 15:1191-
1198.
Dodson, S. I., A. L. Newman, S. Will-Wolf, M. L. Alexander, M. P. Woodford, and S. Van Egeren.
2009. The relationship between zooplankton community structure and lake
characteristics in temperate lakes (Northern Wisconsin, USA). Journal of Plankton
Research 31:93-100.
Du, X., E. Garcfa-Berthou, Q. Wang, J. Liu, T. Zhang, and Z. Li. 2015. Analyzing the importance of
top-down and bottom-up controls in food webs of Chinese lakes through structural
equation modeling. Aquatic Ecology 49:199-210.
Duggan, I. C., J. D. Green, and R. J. Shiel. 2001. Distribution of rotifers in North Island, New
Zealand, and their potential use as bioindicators of lake trophic state. Hydrobiologia
446:155-164.
Duggan, I. C., J. D. Green, and R. J. Shiel. 2002. Distribution of rotifer assemblages in North
Island, New Zealand, lakes: relationships to environmental and historical factors.
Freshwater Biology 47:195-206.
Fuller, P., and M. E. Neilson. 2015. The U.S. Geological Survey's Nonindigenous Aquatic Species
Database: over thirty years of tracking introduced aquatic species in the United States
(and counting). Management of Biological Invasions 6:159-170.
134
NLA 2012 Technical Report. October 2024 Version 1.1
-------
Gannon, J. E., and R. S. Stemberger. 1978. Zooplankton (Especially Crustaceans and Rotifers) as
Indicators of Water Quality. Transactions of the American Microscopical Society 97:16-
35.
Gelinas, M., and B. Pinel-Alloul. 2008. Relating crustacean zooplankton community structure to
residential development and land-cover disturbance near Canadian Shield lakes.
Canadian Journal of Fisheries and Aquatic Sciences 65:2689-2702.
Hawkins, C. P. 2006. Quantifying biological integrity by taxonomic completeness: its utility in
regional and global assessments. Ecological Applications 16:1277-1294.
Hawkins, C. P., Y. Cao, and B. Roper. 2010. Method of predicting reference condition biota
affects the performance and interpretation of ecological indices. Freshwater Biology
55:1066-1085.
Hawkins, C. P., R. H. Norris, J. M. Hague, and J. M. Feminella. 2000. Development and
evaluation of predictive models for measuring the biological integrity of streams.
Ecological Applications 10:1456-1477.
Herlihy, A. T., S. G. Paulsen, J. V. Sickle, J. L. Stoddard, C. P. Hawkins, and L. L. Yuan. 2008.
Striving for consistency in a national assessment: the challenges of applying a reference-
condition approach at a continental scale. Journal of the North American Benthological
Society 27:860-877.
Hurlbert, S. H. 1971. The non-concept of species diversity: a critique and alternative
parameters. Ecology 52:577-586.
Jeppesen, E., P. Noges, T. Davidson, J. Haberman, T. Noges, K. Blank, T. Lauridsen, M.
S0ndergaard, C. Sayer, R. Laugaste, L. Johansson, R. Bjerring, and S. Amsinck. 2011.
Zooplankton as indicators in lakes: a scientific-based plea for including zooplankton in
the ecological quality assessment of lakes according to the European Water Framework
Directive (WFD). Hydrobiologia 676:279-297.
Jeppesen, E., J. Peder Jensen, M. S0ndergaard, T. Lauridsen, and F. Landkildehus. 2000. Trophic
structure, species richness and biodiversity in Danish lakes: changes along a phosphorus
gradient. Freshwater Biology 45:201-218.
Kane, D. D., S. I. Gordon, M. Munawar, M. N. Charlton, and D. A. Culver. 2009. The Planktonic
Index of Biotic Integrity (P-IBI): An approach for assessing lake ecosystem health.
Ecological Indicators 9:1234-1247.
Karr, J. R. 1981. Assessment of biotic integrity using fish communities. Fisheries 6:21-27.
135
NLA 2012 Technical Report. October 2024 Version 1.1
-------
Karr, J. R. 1991. Biological integrity: a long neglected aspect of water resource management.
Ecological Applications 1:66-84.
Kaufmann, P. R., P. Levine, E. G. Robison, C. Seeliger, and D. V. Peck. 1999. Quantifying physical
habitat in wadeable streams. EPA 620/R-99/003, Office of Research and Development,
US Environmental Protection Agency, Washington, DC.
Kaufmann, P. R., D. V. Peck, S. G. Paulsen, C. W. Seeliger, R. M. Hughes, T. R. Whittier, and N. C.
Kamman. 2014. Lakeshore and littoral physical habitat structure in a national lakes
assessment. Lake and Reservoir Management 30:192-215.
Larsen, D. P., and S. J. Christie, editors. 1993. EMAP-Surface Waters: 1991 pilot report. EPA
620/R-93/003. U.S. Environmental Protection Agency, Washington, DC.
Meador, M. R., D. M. Carlisle, and J. F. Coles. 2008. Use of tolerance values to diagnose water-
quality stressors to aquatic biota in New England streams. Ecological Indicators 8:718-
728.
Omernik, J. M. 1987. Ecoregions of the conterminous United States. Annals of the Association
of American Geographers 77:118-125.
Omernik, J. M., and G. E. Griffith. 2014. Ecoregions of the Conterminous United States:
Evolution of a Hierarchical Spatial Framework. Environmental Management 54:1249-
1266.
Sprules, W. G. 1980. Zoogeographic patterns in the size structure of zooplankton communities,
with possible applications to lake ecosystem modeling and management. Pages 642-656
in W. C. Kerfoot, editor. Evolution and Ecology of Zooplankton Communities. Special
Symposium Volume 3, American Sociey of Limnology and Oceanography. University
Press of New England, Hanover, New Hampshire.
Sprules, W. G., and L. B. Holtby. 1979. Body size and feeding ecology as alternatives to
taxonomy for the study of limnetic zooplankton community structure. Journal of the
Fisheries Research Board of Canada 36:1354-1363.
Stemberger, R. S., D. P. Larsen, and T. M. Kincaid. 2001. Sensitivity of zooplankton for regional
lake monitoring. Canadian Journal of Fisheries and Aquatic Sciences 58:2222-2232.
Stemberger, R. S., and J. M. Lazorchak. 1994. Zooplankton assemblage responses to disturbance
gradients. Canadian Journal of Fisheries and Aquatic Sciences 51:2435-2447.
Stemberger, R. S., and E. K. Miller. 1998. A Zooplankton-N:P-Ratio Indicator for Lakes.
Environmental Monitoring and Assessment 51:29-51.
136
NLA 2012 Technical Report. October 2024 Version 1.1
-------
Stoddard, J. 2004. Use of ecological regions in aquatic assessments of ecological condition.
Environmental Management 34:S61-S70.
Stoddard, J. L., A. T. Herlihy, D. V. Peck, R. M. Hughes, T. R. Whittier, and E. Tarquinio. 2008. A
process for creating multimetric indices for large-scale aquatic surveys. Journal of the
North American Benthological Society 27:878-891.
Tessier, A. J., and R. J. Horwitz. 1990. Influence of Water Chemistry on Size Structure of
Zooplankton Assemblages. Canadian Journal of Fisheries and Aquatic Sciences 47:1937-
1943.
USEPA (United States Environmental Protection Agency). 2009. National Lakes Assessment: a
collaborative survey of the Nation's lakes. EPA 841/R-09/001, U.S. Environmental
Protection Agency, Office of Water and Office of Research and Development,
Washington, DC.
USEPA (United States Environmental Protection Agency). 2012a. 2012 National Lakes
Assessment Field Operations Manual. EPA/841/B-11/004, EPA/841/B-11/004. U.S.
Environmental Protection Agency, Office of Water, Washington, DC.
USEPA (United States Environmental Protection Agency). 2012b. 2012 National Lakes
Assessment Laboratory Operations Manual. EPA/841/B-11/004. U.S. Environmental
Protection Agency, Office of Water, Washington, DC.
Van Egeren, S. J., S. I. Dodson, B. Torke, and J. T. Maxted. 2011. The relative significance of
environmental and anthropogenic factors affecting zooplankton community structure in
Southeast Wisconsin Till Plain lakes. Hydrobiologia 668:137-146.
Van Sickle, J. 2010. Correlated metrics yield multimetric indices with inferior performance.
Transactions of the American Fisheries Society 139:1802-1917.
Vogt, R. J., P. R. Peres-Neto, and B. E. Beisner. 2013. Using functional traits to investigate the
determinants of crustacean zooplankton community structure. Oikos 122:1700-1709.
Whittier, T. R., R. M. Hughes, G. A. Lomnicky, and D. V. Peck. 2007. Fish and amphibian
tolerance values and an assemblage tolerance index for streams and rivers in the
western USA. Transactions of the American Fisheries Society 136:254-271.
Whittier, T. R., S. G. Paulsen, D. P. Larsen, S. A. Peterson, A. T. Herlihy, and P. R. Kaufmann.
2002. Indicators of ecological stress and their extent in the population of northeastern
lakes: a regional-scale assessment. Bioscience 52:235-247.
137
NLA 2012 Technical Report. October 2024 Version 1.1
-------
Whittier, T. R., and J. Van Sickle. 2010. Macroinvertebrate tolerance values and an assemblage
tolerance index (ATI) for western USA streams and rivers. Journal of the North American
Benthological Society 29:852-866.
Wright, J. F. 1995. Development and use of a system for predicting the macroinvertebrate
fauna in flowing waters. Australian Journal of Ecology:181-197.
Yuan, L. L. 2004. Assigning macroinvertebrate tolerance classifications using generalised
additive models. Freshwater Biology 49:662-677.
Yuan, L. L., C. P. Hawkins, and J. V. Sickle. 2008. Effects of regionalization decisions on an O/E
index for the US national assessment. Journal of the North American Benthological
Society 27:892-9
138
NLA 2012 Technical Report. October 2024 Version 1.1
-------
7.10 List of Candidate Metrics for Zoopiankton
This section provides additional details for the candidate metrics we considered when
developing the MMIs for each bio-region. Table 7-11 through Table 7-15 list each metric by its
variable name, which of the six metric categories it was assigned to (see Section 7.4.4), and a
description of the metric for the Coastal Plains, Eastern Highlands, Plains, Upper Midwest, and
Western Mountains bio-regions, respectively. In addition, the responsiveness to disturbance
and repeatability of each metric is provided (t-value for responsiveness, and S:N value for
repeatability).
139
NLA 2012 Technical Report. October 2024 Version 1.1
-------
Table 7-11. List of candidate metrics used to develop the zooplankton MMI for the Coastal Plain bio-region.
f value
Mean Value for
Mean Value for
(Least disturbed vs.
Metric
Least disturbed
Most disturbed
Most disturbed
SignakNoise
Category
Metric Name
Description
Sites
Sites
Sites)
Value
Abundance/
Biomass of individuals of smaller-sized taxa
Biomass/
(NET_SIZECLS_NEW=FINE; coarse and fine net
Density
FINE BIO
samples combined)
14.73941733
50.21840118
4. -1,67
1.2
Abundance/
Biomass represented by individuals collected in
Biomass/
fine mesh net (50-um for 2012 samples, 80-um for
Density
ZOFN BIO
2007 resamples)
20.49135593
67.15372044
5. -1.79
1.2
Percent of total individuals that are within the
cladoceran family Sididae (coarse and fine net
Cladoceran
SIDID PIND
samples combined)
2.10
8.18
-1.80
0.4
Total density of individuals within the copepod
order Calanoida (coarse and fine net samples
Copepod
CALAN DEN
combined)
2.806313333
15.22849706
-1.46
2.2
Number of families represented by distinct native
Richness/Diversity
FAM NAT NTAX
taxa (coarse and fine net samples combined)
11.9
9.3
2.62
1.9
Number of families represented by distinct taxa
Richness/Diversity
FAM NTAX
(coarse and fine net samples combined)
11.9
9.4
2.55
2.0
Number of genera represented by distinct taxa
Richness/Diversity
GEN NTAX
(coarse and fine net samples combined)
15.4
12.1
2.21
1.5
Number of genera represented by distinct native
Richness/Diversity
GEN NAT NTAX
taxa (coarse and fine net samples combined)
15.4
12
2.27
1.3
Number of families represented by distinct native
Richness/Diversity
ZOFN FAM NAT NTAX
taxa in the fine mesh net (50-um)
7.4
5.4
2.32
1.4
Total density of individuals within the rotifer order
Collothecaceae (coarse and fine net samples
Rotifer
COLLO BIO
combined)
0.198623267
0.021970559
1.79
3.3
Percent of total individuals within the rotifer order
Collothecaceae (coarse and fine net samples
Rotifer
COLLO PIND
combined)
2.27
0.32
1.87
2.0
Percent of total biomass within the rotifer order
Collothecaceae (coarse and fine net samples
Rotifer
COLLO PBIO
combined)
1.08
0.15
1.8
7.6
Number of distinct predator taxa (coarse and fine
Trophic
PRED NTAX
net samples combined)
2.5
1.3
2.56
4.6
Percent of distinct taxa that are predators (coarse
Trophic
PRED PTAX
and fine net samples combined)
12.01
6.59
2.71
2.2
Number of distinct herbivore taxa (coarse and fine
Trophic
HERB NTAX
net samples combined)
11.9
8.7
2.27
2.1
Percent of distinct taxa that are omnivorous (coarse
Trophic
OMNI_PTAX
and fine net samples combined)
22.03
34.10
-3.35
4.3
140
NLA 2012 Technical Report. October 2024 Version 1.1
-------
f value
Mean Value for
Mean Value for
(Least disturbed vs.
Metric
Least disturbed
Most disturbed
Most disturbed
SignakNoise
Category
Metric Name
Description
Sites
Sites
Sites)
Value
Percent of total density represented by omnivorous
Trophic
OMNI PDEN
individuals (coarse and fine net samples combined)
18.31
40.85
-2.42
1.6
Number of distinct rotifer taxa that are predators
Trophic
ROT PRED NTAX
(coarse and fine net samples combined)
2.2
1.1
2.50
4.5
Trophic
ROT PRED PTAX
Percent of distinct rotifer taxa that are predators
10.78
5.64
2.70
1.9
Number of distinct rotifer taxa that are herbivores
Trophic
ROT HERB NTAX
(coarse and fine net samples combined)
6.8
4.6
2.00
1.8
Biomass represented by rotifer individuals that are
Trophic
ROT OMNI BIO
omnivores
4.7929874
35.027427794
-1.76
1.4
Percent of rotifer individuals represented by
Trophic
ROT OMNI PIND
omnivores
13.41
26.55
-1.88
2.0
Trophic
ROT OMNI PTAX
Percent of distinct rotifertaxa that are omnivorous
17.26
27.95
-3.34
2.6
Percent of rotifer density represented by
Trophic
ROT OMNI PDEN
omnivores
18.15
40.57
-2.42
1.6
Metrics Derived from 300-count Subsamples of Coarse and Fine Net Samples
Abundance/
Biomass
Total biomass in 300-count subsample of fine-mesh
Density
ZOFN300 BIO
net sample (50-urn)
10.962325
35.92416574
-1.89
0.9
Percent of distinct taxa in the 300-count
subsamples that are in the family Bosminidae
Cladoceran
BOSM300 PTAX
(coarse and fine net samples combined)
7.357333333
3.916470588
2.77
0.3
Percent of individuals within the cladoceran family
Sididae in 300-count subsamples (coarse and fine
Cladoceran
SIDID300 PIND
net samples combined)
2.95
9.10
-1.68
0.7
Percent of biomass in dominant copepod taxon in
the 300 count subsamples (coarse and fine net
Copepod
DOM1 300 COPE PBIO
samples combined)
90.00
76.87
1.82
1.9
Number of genera represented by distinct taxa
Richness/Diversity
GEN300 NTAX
(coarse and fine net samples combined)
14
11.1
2.13
1.6
Number of genera represented by distinct native
Richness/Diversity
GEN300 NAT NTAX
taxa (coarse and fine net samples combined)
14
11.0
2.18
1.4
Number of families represented in 300 count
subsamples (coarse and fine net samples
Richness/Diversity
FAM300 NTAX
combined)
10.9
8.6
2.61
1.9
Number of native families represented in 300 count
subsamples (coarse and fine net samples
Richness/Diversity
FAM300 NAT NTAX
combined)
10.9
8.5
2.66
1.8
Number of distinct native families in 300-count
Richness/Diversity
ZOFN300 FAM NAT NTAX
subsample of fine-mesh net sample (50-^m)
6.7
4.8
2.49
1.3
Biomass represented by individuals of the rotifer
order Collothecaceae in the 300-count subsamples
Rotifer
COLLO300_BIO
(coarse and fine net samples combined)
0.0838373333
0.0125823235
1.76
3.4
141
NLA 2012 Technical Report. October 2024 Version 1.1
-------
f value
Mean Value for
Mean Value for
(Least disturbed vs.
Metric
Least disturbed
Most disturbed
Most disturbed
SignakNoise
Category
Metric Name
Description
Sites
Sites
Sites)
Value
Percent of biomass within the rotifer order
Collothecaceae in the 300-count subsamples
Rotifer
COLLO300 PBIO
(coarse and fine net samples combined)
0.96
0.16
1.75
5.9
Number of distinct taxa that are predators in 300
count subsamples (coarse and fine net samples
Trophic
PRED300 NTAX
combined)
1.7
1.0
1.94
2.7
Biomass of predator individuals in 300 count
subsamples (coarse and fine net samples
Trophic
PRED300 BIO
combined)
0.4595966
0.1407230588
2.45
1.5
Number of distinct taxa that are herbivores in 300
count subsamples (coarse and fine net samples
Trophic
HERB300 NTAX
combined)
11.0
7.9
2.41
1.7
Percent of omnivorous individuals in 300 count
subsamples (coarse and fine net samples
Trophic
OMNI300 PIND
combined)
15.53
28.44
-1.86
1.4
Percent of distinct taxa that are omnivores in 300
count subsamples (coarse and fine net samples
Trophic
OMNI300 PTAX
combined)
23.38
37.04
-3.27
4.9
Percent of biomass represented by omnivorous
individuals in 300 count subsamples (coarse and
Trophic
OMNI300 PBIO
fine net samples combined)
27.224
35.48058824
-2.96
4.7
Number of distinct rotifer taxa that are predators in
300 count subsamples (coarse and fine net samples
Trophic
ROT PRED300 NTAX
combined)
1.7
1.0
1.940
2.7
Biomass represented by rotifer individuals that are
predators in 300 count subsamples (coarse and fine
Trophic
ROT PRED300 BIO
net samples combined)
0.4595966
0.1407230588
2.45
1.5
Number of distinct rotifer taxa that are herbivores
in 300 count subsamples (coarse and fine net
Trophic
ROT HERB300 NTAX
samples combined)
6.1
4.0
2.24
1.4
Percent of rotifer individuals that are omnivorous in
300 count subsamples (coarse and fine net samples
Trophic
ROT OMNI300 PIND
combined)
12.24
25.10
-2.00
1.9
Percent of distinct rotifertaxa that are omnivorous
in 300 count subsamples (coarse and fine net
Trophic
ROT_OMNI300_PTAX
samples combined)
18.47
30.13
-3.00
4.3
142
NLA 2012 Technical Report. October 2024 Version 1.1
-------
Table 7-12. List of candidate metrics used to develop the zooplankton MMI for the Eastern Highlands bio-region
f value
Mean Value for
Mean Value for
(Least disturbed vs.
Metric
Least disturbed
Most disturbed
Most disturbed
SignakNoise
Category
Metric Name
Description
Sites
Sites
Sites)
Value
Abundance/
Density represented by individuals collected in
Biomass/
coarse mesh net (150-um for 2012 samples, 243 um
Density
ZOCN DEN
for 2007 resamples)
12.56848
34.33432549
-1.89
7.1
Abundance/
Density represented by native individuals collected
Biomass/
in coarse mesh net (150-um for 2012 samples, 243
Density
ZOCN NAT DEN
um for 2007 resamples)
12.56848
34.33106863
-1.89
2.1
Abundance/
Density represented by individuals of taxa collected
Biomass/
in coarse mesh net (150-um; coarse and fine net
Density
COARSE DEN
samples combined)
21.26666667
53.84573922
-2.13
2.4
Abundance/
Biomass represented by individuals of taxa
Biomass/
collected in coarse mesh net (150-um; coarse and
Density
COARSE PBIO
fine net samples combined)
68.49155556
56.48058824
1.86
1.7
Abundance/
Density represented by individuals of native larger-
Biomass/
sized taxa (NET_SIZECLS_NEW=COARSE; coarse and
Density
COARSE NAT DEN
fine net samples combined)
21.266666667
53.80877451
-2-12
1.5
Abundance/
Biomass represented by individuals of native larger-
Biomass/
sized taxa (NET_SIZECLS_NEW=COARSE; coarse and
Density
COARSE NAT PBIO
fine net samples combined)
68.491555556
56.44254902
1.86
1.5
Abundance/
Biomass represented by individuals of smaller-sized
Biomass/
taxa (NET_SIZECLS_NEW=FINE; coarse and fine net
Density
FINE PBIO
samples combined)
31.508444444
43.519411765
-1.86
1.7
Density of native individuals within the suborder
Cladoceran
CLAD DEN
Cladocera (coarse and fine net samples combined)
6.813766667
27.71694902
-1.94
1.9
Density of native individuals within the suborder
Cladoceran
CLAD NAT DEN
Cladocera (coarse and fine net samples combined)
6.813766667
27.71382549
-1.94
1.8
Biomass represented by large cladoceran
individuals (SUBORDER=CLADOCERA and
CLADOCEAN SIZE=LARGE; coarse and fine net
Cladoceran
LGCLAD BIO
samples combined)
25.780533111
10.663794725
2.16
1.3
Biomass represented by native large cladoceran
individuals (SUBORDER=CLADOCERA and
CLADOCEAN SIZE=LARGE; coarse and fine net
Cladoceran
LGCLAD NAT BIO
samples combined)
25.780533111
10.656975706
2.16
1.3
Biomass represented by small cladoceran
individuals (SUBORDER=CLADOCERA and
CLADOCEAN SIZE=SMALL; coarse and fine net
Cladoceran
SMCLAD BIO
samples combined)
2.985147667
31.80179637
-2.37
2.6
Density represented by small cladoceran individuals
(SUBORDER=CLADOCERA and
CLADOCERAN SIZE=SMALL; coarse and fine net
Cladoceran
SMCLAD DEN
samples combined)
2.476364444
22.86743922
-1.99
2.4
143
NLA 2012 Technical Report. October 2024 Version 1.1
-------
f value
Mean Value for
Mean Value for
(Least disturbed vs.
Metric
Least disturbed
Most disturbed
Most disturbed
SignakNoise
Category
Metric Name
Description
Sites
Sites
Sites)
Value
Percent of small cladoceran individuals
(SUBORDER=CLADOCERA and CLAD-SIZE=SMALL;
Cladoceran
SMCLAD PIND
coarse and fine net samples combined)
9.58
17.42
-2.73
1.6
Percent of total density represented by small
cladoceran individuals (SUBORDER=CLADOCERA
and CLADOCERAN SIZE=SMALL; coarse and fine net
Cladoceran
SMCLAD PDEN
samples combined)
1.03
3.34
-1.91
19.1
Biomass represented by native small cladoceran
individuals (SUBORDER=CLADOCERA and
CLADOCERAN SIZE=SMALL; coarse and fine net
Cladoceran
SMCLAD NAT BIO
samples combined)
2.985147667
31.79812541
-2.37
2.5
Density represented by native small cladoceran
individuals (SUBORDER=CLADOCERA and
CLADOCERA SIZE=SMALL; coarse and fine net
Cladoceran
SMCLAD NAT DEN
samples combined)
2.476364444
22.86662549
-1.99
2.2
Percent of total density represented by native small
cladoceran individuals (SUBORDER=CLADOCERA
and CLADOCERAN SIZE=SMALL; coarse and fine net
Cladoceran
SMCLAD NAT PDEN
samples combined)
1.03
3.33
-1.91
19.1
Density of individuals within the family Daphniidae
Cladoceran
DAPHNIID DEN
(coarse and fine net samples combined)
3.223097778
16.27482549
-2.09
2.5
Density of native individuals within the family
Cladoceran
DAPHNIID NAT DEN
Daphniidae (coarse and fine net samples combined)
3.223097778
16.27251961
-2.09
2.5
Density represented by individuals within the
subclass Copepoda (coarse and fine net samples
Copepod
COPE DEN
combined)
81.931315556
139.66798235
-1.74
1.5
Density represented by native individuals within the
subclass Copepoda (coarse and fine net samples
Copepod
COPE NAT DEN
combined)
81.931315556
139.66784314
-1.74
1.5
Number of distinct taxa within the copepod order
Copepod
CALAN NTAX
Calanoida (coarse and fine net samples combined)
1.3
1.1
2.10
2.4
Percent of total density represented by taxa of the
copepod order Calanoida (coarse and fine net
Copepod
CALAN PDEN
samples combined)
3.82
1.64
1.80
35.0
Number of distinct native taxa within the copepod
order Calanoida (coarse and fine net samples
Copepod
CALAN NAT NTAX
combined)
1.3
1.0
2.22
1.3
Percent of total density represented by individuals
of native taxa within the copepod order Calanoida
Copepod
CALAN NAT PDEN
(coarse and fine net samples combined)
3.81
1.64
1,80
35.0
Percent of distinct larger-sized native taxa
(NET SIZECLS NEW=COARSE; coarse and fine net
Richness/Diversity
COARSE_NAT_PTAX
samples combined)
40.65
37.17
1.64
0.3
144
NLA 2012 Technical Report. October 2024 Version 1.1
-------
f value
Mean Value for
Mean Value for
(Least disturbed vs.
Metric
Least disturbed
Most disturbed
Most disturbed
SignakNoise
Category
Metric Name
Description
Sites
Sites
Sites)
Value
Percent total biomass from rotifers (coarse and fine
Rotifer
ROT PBIO
net samples combined)
23.72
34.91
-1.88
1.3
Percent of distinct taxa that are omnivorous (coarse
Trophic
OMNI PTAX
and fine net samples combined)
23.38
27.56
-2.36
1.6
Density of herbivorous cladocerans
(suborder=CLADOCERA; coarse and fine net
CLAD HERB DEN
samples combined)
6.8127244444
27.71694902
-1.94
1.9
Percent density represented by herbivorous
copepods (order=COPEPODA; coarse and fine net
COPE HERB PDEN
samples combined)
4.22
1.92
1.86
20.0
Metrics Derived from 300-count Subsamples of Coarse and Fine Net Samples
Percent of biomass represented by individuals of
taxa collected in coarse mesh net (150-um;
Abundance/
NET_SIZECLS_NEW=COARSE) in 300 count
Biomass/
subsamples (coarse and fine net samples
Density
COARSE300 PBIO
combined)
70.74
58.61
1.96
1.7
Percent of biomass represented by individuals of
native taxa collected in coarse mesh net (150-um;
Abundance/
NET SIZECLS NEW=COARSE ) in 300 count
Biomass/
subsamples (coarse and fine net samples
Density
COARSE300 NAT PBIO
combined)
70.738666667
58.570196078
1.96
1.5
Percent biomass represented by individuals of
Abundance/
smaller-sized taxa (NET_SIZECLS_NEW=FINE) in
Biomass/
300-count subsamples (coarse and fine net samples
Density
FINE300 PBIO
combined)
29.26
41.39
-1.96
1.7
Biomass represented by large cladoceran
individuals (SUBORDER=CLADOCERA and
CLADOCEAN_SIZE=LARGE) in 300-count subsamples
Cladoceran
LGCLAD300 BIO
( coarse and fine net samples combined)
15.692285844
7.0078742941
2.02
1.4
Biomass represented by native large cladoceran
individuals (SUBORDER=CLADOCERA and
CLADOCEAN_SIZE=LARGE) in 300-count subsamples
Cladoceran
LGCLAD300 NAT BIO
( coarse and fine net samples combined)
15.692285844
7.0031208824
2.02
1.4
Biomass represented by small cladoceran
individuals (SUBORDER=CLADOCERA and
CLADOCEAN_SIZE=SMALL) in 300-count subsamples
Cladoceran
SMCLAD300 BIO
( coarse and fine net samples combined)
1.8545441111
21.410646353
-2.40
2.6
Percent of small cladoceran individuals
(SUBORDER=CLADOCERA and
CLADOCEAN_SIZE=SMALL) in 300-count subsamples
Cladoceran
SMCLAD300 PIND
( coarse and fine net samples combined)
10.90
19.03
-2.72
1.7
145
NLA 2012 Technical Report. October 2024 Version 1.1
-------
f value
Mean Value for
Mean Value for
(Least disturbed vs.
Metric
Least disturbed
Most disturbed
Most disturbed
SignakNoise
Category
Metric Name
Description
Sites
Sites
Sites)
Value
Percent of biomass represented by small
cladoceran individuals (SUBORDER=CLADOCERA
and CLADOCEAN_SIZE=SMALL) in 300-count
subsamples ( coarse and fine net samples
Cladoceran
SMCLAD300 PBIO
combined)
5.50
16.12
-2.82
1.6
Biomass represented by native small cladoceran
individuals (SUBORDER=CLADOCERA and
CLADOCEAN_SIZE=SMALL) in 300-count subsamples
Cladoceran
SMCLAD300 NAT BIO
( coarse and fine net samples combined)
1.8545441111
21.410646353
-2.40
2.5
Percent of native small cladoceran individuals
(SUBORDER=CLADOCERA and
CLADOCEAN_SIZE=SMALL) in 300-count subsamples
Cladoceran
SMCLAD300 NAT PIND
( coarse and fine net samples combined)
10.90
19.03
-2.72
1.4
Number of distinct taxa within the copepod order
Calanoida in 300-count subsamples (coarse and fine
Copepod
CALAN300 NTAX
net samples combined)
1.3
1.0
1.94
2.8
Number of distinct native taxa within the copepod
order Calanoida in 300-count subsamples (coarse
Copepod
CALAN300 NAT NTAX
and fine net samples combined)
1.3
1.0
2.08
1.4
Percent distinct native taxa in 300-count subsample
Richness/Diversity
ZOCN300 NAT PTAX
of coarse net sample (150-um)
100
98.55
1.88
0.1
Number of distinct native taxa in coarse net
Richness/Diversity
ZOCN300 FAM NTAX
samples (150-um) based on 300-count subsample
5.1
4.7
1.47
0.8
Percent biomass from rotifers in 300-count
subsamples (coarse and fine net samples
Rotifer
ROT300 PBIO
combined)
22.26
34.91
-1.89
1.3
Percent of distinct taxa that are omnivorous in 300-
count subsamples (coarse and fine net samples
Trophic
OMNI300_PTAX
combined)
23.31
28.29
-2.60
1.5
146
NLA 2012 Technical Report. October 2024 Version 1.1
-------
Table 7-13. List of candidate metrics used to develop the zooplankton MMI for the Plains bio-region
f value
Mean Value for
Mean Value for
(Least disturbed vs.
Metric
Least disturbed
Most disturbed
Most disturbed
SignakNoise
Category
Metric Name
Description
Sites
Sites
Sites)
Value
Abundance/
Percent of total biomass represented by individuals
Biomass/
collected in coarse mesh net (150-um for 2012
Density
COARSE PBIO
samples, 243 um for 2007 resamples)
57.38
70.00
-1.75
6.3
Abundance/
Percent of total biomass represented by native
Biomass/
individuals collected in coarse mesh net (150-um
Density
COARSE NAT PBIO
for 2012 samples, 243 um for 2007 resamples)
57.38
69.94
-1.74
6.3
Abundance/
Percent of biomass represented by individuals of
Biomass/
smaller-sized taxa (NET_SIZECLS_NEW=FINE; coarse
Density
FINE PBIO
and fine net samples combined)
42.62
30.00
1.75
6.3
Percent of biomass represented by native
Abundance/
individuals of smaller-sized taxa
Biomass/
(NET_SIZECLS_NEW=FINE; coarse and fine net
Density
FINE NAT PBIO
samples combined)
42.62
29.99
1.75
6.2
Percent of total individuals within the suborder
Cladocera that are "small"
(CLADOCERA_SIZE=SMALL; coarse and fine net
Cladoceran
SMCLAD PIND
samples combined)
19.26
9.03
3.09
1.8
Percent of native individuals within the suborder
Cladocera that are "small"
(CLADOCERA_SIZE=SMALL; coarse and fine net
Cladoceran
SMCLAD NAT PIND
samples combined)
19.26
8.94
3.11
1.8
Percent of total biomass represented by native
small cladoceran individuals
(SUBORDER=CLADOCERA and
CLADOCEAN SIZE=SMALL; coarse and fine net
Cladoceran
SMCLAD NAT PBIO
samples combined)
13.35
7.02
1.74
1.4
Copepod
Percent of total individuals within the subclass
COPE PIND
Copepoda (coarse and fine net samples combined)
29.45
41.97
-2.46
1.4
Copepod
Percent of native individuals within the subclass
COPE NAT PIND
Copepoda (coarse and fine net samples combined)
29.45
41.97
-2.46
1.4
Percent of distinct taxa that are within the copepod
order Calanoida (coarse and fine net samples
Copepod
CALAN PTAX
combined)
6.38
10.16
-2.32
2.0
Percent of total density represented by individuals
within the copepod order Calanoida (coarse and
Copepod
CALAN PDEN
fine net samples combined)
1.20
6.52
-2.06
14.1
Percent of total density represented by native
individuals within the copepod order Calanoida
Copepod
CALAN_NAT_PDEN
(coarse and fine net samples combined)
1.20
6.52
-2.06
14.1
147
NLA 2012 Technical Report. October 2024 Version 1.1
-------
f value
Mean Value for
Mean Value for
(Least disturbed vs.
Metric
Least disturbed
Most disturbed
Most disturbed
SignakNoise
Category
Metric Name
Description
Sites
Sites
Sites)
Value
Ratio of Calanoid to (Cladoccera+Cyclopoids) based
on number of individuals (coarse and fine net
samples combined). Adapted from Kane et al.
(2009) Lake Erie plankton IBI. Calculated as
Copepod
COPE RATIO NIND
CALANOID_NIND/(CLAD_NIND+CYCLOPOID_NIND)
17.435
0.812
1.84
38.9
Ratio of Calanoid to (Cladoccera+Cyclopoids) based
on biomass (coarse and fine net samples
combined). Adapted from Kane et al. (2009) Lake
Erie plankton IBI. Calculated as
Copepod
COPE RATIO BIO
CALANOID_BIO/(CLAD_BIO+CYCLOPOID_BIO)
7.325729723
1.327404241
2.31
4.6
Total distinct taxa richness (coarse and fine net
Richness/Diversity
TOTL NTAX
samples combined)
17.3
14..6
2.27
2.2
Total distinct native taxa richness (coarse and fine
Richness/Diversity
TOTL NAT NTAX
net samples combined)
17.3
14.5
2.34
2.2
Number of genera represented by distinct taxa
Richness/Diversity
GEN NTAX
(coarse and fine net samples combined)
13.8
11.6
2.45
2.2
Number of genera represented by distinct native
Richness/Diversity
GEN NAT NTAX
taxa (coarse and fine net samples combined)
13.8
11.5
2.56
2.2
Number of families represented by distinct taxa
Richness/Diversity
FAM NTAX
(coarse and fine net samples combined)
10.7
9.1
2.32
1.9
Number of families represented by distinct native
Richness/Diversity
FAM NAT NTAX
taxa (coarse and fine net samples combined)
10.7
9.1
2.41
2.2
Number of distinct taxa in fine net sample (ZOFN;
Richness/Diversity
ZOFN NTAX
80-um mesh)
12.4
9.8
2.69
1.7
Number of distinct native taxa in fine net sample
Richness/Diversity
ZOFN NAT NTAX
(ZOFN; 80-um mesh)
12.4
9.8
2.73
1.7
Number of genera represented by distinct taxa in
Richness/Diversity
ZOFN GEN NTAX
fine net sample (ZOFN; 80-um mesh)
8.1
5.8
3.36
3.8
Number of genera represented by distinct native
Richness/Diversity
ZOFN GEN NAT NTAX
taxa in fine net sample (ZOFN; 80-um mesh)
8.1
5.8
3.42
3.8
Number of families represented by distinct taxa in
Richness/Diversity
ZOFN FAM NTAX
fine net sample (ZOFN; 80-um mesh)
6.6
4.7
3.48
3.0
Number of families represented by distinct native
Richness/Diversity
ZOFN FAM NAT NTAX
taxa in fine net sample (ZOFN; 80-um mesh)
6.6
4.7
3.56
3.0
Number of distinct taxa collected only in the fine-
Richness/Diversity
FINE NTAX
mish net (80-um; NET_SIZECLS_NEW=FINE)
10.5
8.0
2.61
1.8
Number of distinct native taxa collected only in the
Richness/Diversity
FINE NAT NTAX
fine-mish net (80-um; NET_SIZECLS_NEW=FINE)
10.5
8.0
2.63
1.7
Percent of total biomass represented in top 5 taxa
Richness/Diversity
DOM5 PBIO
(coarse and fine net samples combined)
91.31
94.16
-1.77
2.5
Number of distinct rotifer taxa (coarse and fine net
Rotifer
ROT_NTAX
samples combined)
10.5
8.0
2.63
1.7
148
NLA 2012 Technical Report. October 2024 Version 1.1
-------
f value
Mean Value for
Mean Value for
(Least disturbed vs.
Metric
Least disturbed
Most disturbed
Most disturbed
SignakNoise
Category
Metric Name
Description
Sites
Sites
Sites)
Value
Percent of total density represented by herbivorous
Trophic
COPE HERB PDEN
copepods (coarse and fine net samples combined)
1.23
6.58
-2.13
13.0
Metrics Derived from 300-count Subsamples of Coarse and Fine Net Samples
Percent of biomass represented by individuals of
Abundance/
taxa collected in coarse mesh net (150-um) in 300
Biomass/
count subsamples (coarse and fine net samples
Density
COARSE300 PBIO
combined)
59.0316
71.48616279
-1.77
5.2
Percent of biomass represented by native
Abundance/
individuals of taxa collected in coarse mesh net
Biomass/
(150-um) in 300 count subsamples (coarse and fine
Density
COARSE300 NAT PBIO
net samples combined)
59.0316
71.42267442
-1.76
5.1
Percent of biomass represented in individuals of
Abundance/
smaller-sized taxa (NET_SIZECLS_NEW=FINE) in the
Biomass/
300-count subsample (coarse and fine mesh
Density
FINE300 PBIO
samples combined)
42.15
28.64
1.89
6.0
Percent of biomass represented in native
individuals of smaller-sized taxa
Abundance/
(NET_SIZECLS_NEW=FINE) in the 300-count
Biomass/
subsample (coarse and fine mesh samples
Density
FINE300 NAT PBIO
combined)
42.15
28.63
1.90
5.8
Percent of small cladoceran individuals
(SUBORDER=CLADOCERA and
CLADOCEAN_SIZE=SMALL) in 300-count subsamples
Cladoceran
SMCLAD300 PIND
( coarse and fine net samples combined)
19.788
9.848139535
2.97
2.0
Percent of biomass represented by small
cladoceran individuals (SUBORDER=CLADOCERA
and CLADOCEAN_SIZE=SMALL) in 300-count
subsamples (coarse and fine net samples
Cladoceran
SMCLAD300 PBIO
combined)
14.17
7.52
1.74
1.4
Percent of native small cladoceran individuals
(SUBORDER=CLADOCERA and
CLADOCEAN_SIZE=SMALL) in 300-count subsamples
Cladoceran
SMCLAD300 NAT PIND
( coarse and fine net samples combined)
19.788
9.760930233
2.99
2.0
Percent of biomass represented by native small
cladoceran individuals (SUBORDER=CLADOCERA
and CLADOCEAN SIZE=SMALL) in 300-count
subsamples (coarse and fine net samples
Cladoceran
SMCLAD300 NAT PBIO
combined)
14.17
7.47
1.76
1.4
Percent of individuals within the subclass Copepoda
in 300-count subsamples (coarse and fine net
Copepod
COPE300 PIND
samples combined)
30.94
43.16
2.42
1.3
149
NLA 2012 Technical Report. October 2024 Version 1.1
-------
f value
Mean Value for
Mean Value for
(Least disturbed vs.
Metric
Least disturbed
Most disturbed
Most disturbed
SignakNoise
Category
Metric Name
Description
Sites
Sites
Sites)
Value
Percent of native individuals within the subclass
Copepoda in 300-count subsamples (coarse and
Copepod
COPE300 NAT PIND
fine net samples combined)
30.94
43.16
30.93
1.3
Percent of distinct taxa within the copepod order
Calanoida in 300-count subsamples (coarse and fine
Copepod
CALAN300 PTAX
net samples combined)
7.51
11.20
-2.07
4.6
Ratio of Calanoid to (Cladoccera+Cyclopoids) based
on number of individuals in 300-count subsamples
(coarse and fine net samples combined). Adapted
from Kane et al. (2009) Lake Erie plankton IBI.
Calculated as
Copepod
COPE RATIO 300 NIND
CALANOID_NIND/(CLAD_NIND+CYCLOPOID_NIND)
12.675
0.800
1.83
19.6
Ratio of Calanoid to (Cladoccera+Cyclopoids) based
on biomass in 300-count subsamples (coarse and
fine net samples combined). Adapted from Kane et
al. (2009) Lake Erie plankton IBI. Calculated as
Copepod
COPE RATIO 300 BIO
CALANOID_BIO/(CLAD_BIO+CYCLOPOID_BIO)
5.712
1.003
2.41
3.0
Total distinct native taxa richness in 300-count
subsamples (coarse and fine net samples
Richness/Diversity
TOTL300 NAT NTAX
combined)
14.8
12.9
1.76
1.4
Total distinct generic richness in 300-count
subsamples (coarse and fine net samples
Richness/Diversity
GEN300 NTAX
combined)
12.3
10.6
2.03
2.7
Total distinct native generic richness in 300-count
subsamples (coarse and fine net samples
Richness/Diversity
GEN300 NAT NTAX
combined)
12.3
10.5
2.13
2.9
Total distinct family richness in 300-count
subsamples (coarse and fine net samples
Richness/Diversity
FAM300 NTAX
combined)
9.8
8.4
2.11
2.3
Total distinct native family richness in 300-count
subsamples (coarse and fine net samples
Richness/Diversity
FAM300 NAT NTAX
combined)
9.8
8.4
2.22
2.6
Number of distinct genera in 300-count subsample
Richness/Diversity
ZOFN300 GEN NTAX
of fine-mesh net sample (50-^m)
6.8
5.3
2.45
2.7
Number of distinct native genera in 300-count
Richness/Diversity
ZOFN300 GEN NAT NTAX
subsample of fine-mesh net sample (50-^m)
6.8
5.2
2.48
2.9
Number of distinct families in 300-count subsample
Richness/Diversity
ZOFN300 FAM NTAX
of fine-mesh net sample (50-^m)
5.6
4.3
2.74
3.1
Number of distinct native families in 300-count
Richness/Diversity
ZOFN300 FAM NAT NTAX
subsample of fine-mesh net sample (50-^m)
5.6
4.3
2.79
3.1
Percent of biomass represented in top 5 taxa in
300-count subsamples (coarse and fine net samples
Richness/Diversity
DOM5_300_PBIO
combined)
91.38
94.27
-1.78
1.9
150
NLA 2012 Technical Report. October 2024 Version 1.1
-------
Table 7-14, List of candidate metrics used to develop the zooplankton MMI for the Upper Midwest bio-region
f value
Mean Value for
Mean Value for
(Least disturbed vs.
Metric
Least disturbed
Most disturbed
Most disturbed
SignakNoise
Category
Metric Name
Description
Sites
Sites
Sites)
Value
Abundance/
Biomass/
Percent of native individuals (coarse and fine net
Density
TOTL NAT PIND
samples combined)
100
98.02
1.47
2348
Abundance/
Biomass/
Percent of density represented by native individuals
Density
ZOCN NAT PDEN
in coarse net sample (150-um)
100
95.90
1.52
Noise=0
Number of distinct taxa within the cladoceran
family Daphniidae (coarse and fine net samples
Cladoceran
DAPHNIID NTAX
combined)
1.4
1.8
-1.91
3.1
Density of individuals within the cladoceran family
Bosminidae (coarse and fine net samples
Cladoceran
BOSM DEN
combined)
28.20401905
6.857369231
1.85
2.8
Percent of individuals within the cladoceran family
Bosminidae (coarse and fine net samples
Cladoceran
BOSM PIND
combined)
15.31
8.35
1.85
19.5
Biomass of native individuals within the cladoceran
family Bosminidae (coarse and fine net samples
Cladoceran
BOSM NAT BIO
combined)
16.33606357
3.165346051
1.89
1.8
Density of native individuals within the cladoceran
family Bosminidae (coarse and fine net samples
Cladoceran
BOSM NAT DEN
combined)
28.204019048
5.0981051282
2.01
4.9
Percent of native individuals within the cladoceran
family Bosminidae (coarse and fine net samples
Cladoceran
BOSM NAT PIND
combined)
15.31
6.71
2.29
9.6
Percent of distinct native taxa within the
cladoceran family Bosminidae (coarse and fine net
Cladoceran
BOSM NAT PTAX
samples combined)
5.59
3.96
2.16
1.6
Percent of biomass represented by native
individuals within the cladoceran family Bosminidae
Cladoceran
BOSM NAT PBIO
(coarse and fine net samples combined)
10.01
2.57
2.07
4.9
Shannon Diversity based on the number of
cladoceran individuals (coarse and fine net samples
combined). Calculated as SUM{p(i)*Log[p(i)]},
where p(i) is proportion of individuals of taxon i,
Cladoceran
HPRIME CLAD
and Log= natural logarithm.
0.579
0.772
-1.91
1.3
Biomass of individuals within the copepod order
Copepod
CALAN BIO
Calanoida (coarse and fine net samples combined)
12.010544048
27.035772872
-1.73
12.7
Biomass of native individuals within the copepod
order Calanoida (coarse and fine net samples
Copepod
CALAN_NAT_BIO
combined)
12.010544048
27.025444897
-1.73
12.8
151
NLA 2012 Technical Report. October 2024 Version 1.1
-------
f value
Mean Value for
Mean Value for
(Least disturbed vs.
Metric
Least disturbed
Most disturbed
Most disturbed
SignakNoise
Category
Metric Name
Description
Sites
Sites
Sites)
Value
Percent of distinct native taxa (coarse and fine net
Richness/Diversity
TOTL NAT PTAX
samples combined)
100
98.05
2.65
21.7
Richness/Diversity
Percent of distinct taxa represented by native
ZOCN NAT PTAX
individuals in coarse net sample (150-um)
100
95.84
2.59
8.9
Richness/Diversity
Percent of distinct larger-sized taxa
(NET SIZECLS NEW=COARSE; coarse and fine net
COARSE PTAX
samples combined)
39.74
45.09
-1.89
1.4
Richness/Diversity
Percent of distinct smaller-sized taxa
(NET_SIZECLS_NEW=FINE; coarse and fine net
FINE PTAX
samples combined)
60.26
54.91
-1.89
1.4
Percent of distinct taxa within the phylum Rotifera
Rotifer
ROT PTAX
(coarse and fine net samples combined)
60.26
54.91
1.87
1.4
Density of individuals within the rotifer order
Flosculariaceae (coarse and fine net samples
Rotifer
FLOS DEN
combined)
290.0439619
115.22284872
1.82
7.6
Shannon Diversity based on the number of rotifer
individuals (coarse and fine net samples combined).
Calculated as SUM{p(i)*Log[p(i)]}, where p(i) is
proportion of individuals of taxon i, and Log=
Rotifer
HPRIME ROT
natural logarithm.
1.524
1.264
2.12
1.4
Simpson Diversity based on the number of rotifer
individuals (coarse and fine net samples combined).
Calculated as SUM{p(i)*p(i)} where p(i) is the
Rotifer
SIMPSON ROT
proportion of taxon 1 in the sample.
0.325
0.414
-1.79
2.4
Hurlbert's Probability of Interspecific Encounter
(PIE) based on the number of rotifer individuals
(coarse and fine net samples combined).
Calculated as SUM{p(i)*[N-n(i)/N-l]} where p(i) is
the proportion of taxon 1 in the sample, N is the
total number of rotifer individuals in the sample,
and n(i) is the number of rotifer individuals of taxon
Rotifer
PIE ROT
i in the sample.
0.678
0.590
1.76
2.5
Percent of rotifer individuals in top 3 Rotifer taxa
Rotifer
DOM3 ROT PIND
(coarse and fine net samples combined)
78.89
86.34
-2.35
1.6
Percent of rotifer individuals in top 5 Rotifer taxa
Rotifer
DOM5 ROT PIND
(coarse and fine net samples combined)
91.39
94.46
-1.81
2.6
Percent of rotifer biomass in dominant rotifer taxon
Rotifer
DOM1 ROT PBIO
(coarse and fine net samples combined)
45.30
59.27
-2.46
3.5
Percent of rotifer density in top 3 Rotifer taxa
Rotifer
DOM3 ROT PDEN
(coarse and fine net samples combined)
78.89
86.34
-2.35
1.6
Percent of density in top 5 rotifertaxa (coarse and
Rotifer
DOM5 ROT PDEN
fine net samples combined)
91.39
94.46
-1.81
2.6
152
NLA 2012 Technical Report. October 2024 Version 1.1
-------
f value
Mean Value for
Mean Value for
(Least disturbed vs.
Metric
Least disturbed
Most disturbed
Most disturbed
SignakNoise
Category
Metric Name
Description
Sites
Sites
Sites)
Value
Metrics Derived from 300-count Subsamples of Coarse and Fine Net Samples
Number of distinct taxa within the cladoceran
family Daphniidae in 300-count subsamples (coarse
Cladoceran
DAPHNIID300 NTAX
and fine net samples combined)
1.2
1.7
-2.3
3.1
Number of distinct native taxa within the
cladoceran family Daphniidae in 300-count
subsamples (coarse and fine net samples
Cladoceran
DAPHNIID300 NAT NTAX
combined)
1.4
1.7
-2.3
3.1
Biomass of native individuals within the cladoceran
family Bosminidae in 300-count subsamples (coarse
Cladoceran
BOSM300 PIND
and fine net samples combined)
16.74
9.15
1.87
15.4
Density of native individuals within the cladoceran
family Bosminidae in 300-count subsamples (coarse
Cladoceran
BOSM300 NAT BIO
and fine net samples combined)
9.9940477143
2.211484641
1.84
2.1
Percent of native individuals within the cladoceran
family Bosminidae in 300-count subsamples (coarse
Cladoceran
BOSM300 NAT PIND
and fine net samples combined)
16.74
7.12
2.42
15.3
Percent of distinct native taxa that are within the
cladoceran family Bosminidae in 300-count
subsamples (coarse and fine net samples
Cladoceran
BOSM300 NAT PTAX
combined)
6.48
4.08
2.73
1.4
Biomass of biomass represented by native
individuals within the cladoceran family Bosminidae
in 300-count subsamples (coarse and fine net
Cladoceran
BOSM300 NAT PBIO
samples combined)
10.56
2.78
211
4.7
Biomass of individuals within the copepod order
Calanoida in 300-count subsamples (coarse and fine
Copepod
CALAN300 BIO
net samples combined)
6.3444415238
17.540568538
-2.17
9.2
Percent of distinct native taxa in 300-count
subsamples (coarse and fine net samples
Richness/Diversity
TOTL300 NAT PTAX
combined)
100
97.87
2.66
8.2
Percent of distinct native taxa in the coarse net
sample (150-um) based on the 300-individual
Richness/Diversity
ZOCN300 NAT PTAX
subsamples
100
95.92
2.76
Noise=0
Percent of distinct taxa represented by the rotifer
order Ploima in 300-count subsamples (coarse and
Rotifer
PLOIMA300 PTAX
fine net samples combined)
48.72
42.16
2.05
9.8
Shannon Diversity based on the number of rotifer
individuals in 300-count subsamples (coarse and
fine net samples combined). Calculated as
SUM{p(i)*Log[p(i)]}, where p(i) is proportion of
Rotifer
HPRIME_ROT300
individuals of taxon i, and Log= natural logarithm.
1.515
1.254
2.12
1.4
153
NLA 2012 Technical Report. October 2024 Version 1.1
-------
f value
Mean Value for
Mean Value for
(Least disturbed vs.
Metric
Least disturbed
Most disturbed
Most disturbed
SignakNoise
Category
Metric Name
Description
Sites
Sites
Sites)
Value
Simpson Diversity based on the number of rotifer
individuals in 300-count subsamples (coarse and
fine net samples combined). Calculated as
SUM{p(i)*p(i)} where p(i) is the proportion of taxon
Rotifer
SIMPSON ROT300
1 in the sample.
0.324
0.416
-1.86
2.1
Hurlbert's Probability of Interspecific Encounter
(PIE) based on the number of rotifer individuals in
300-count subsamples (coarse and fine net samples
combined). Calculated as SUM{p(i)*[N-n(i)/N-l]}
where p(i) is the proportion of rotifer taxon 1 in the
sample, N is the total number of rotifer individuals
in the sample, and n(i) is the number of individuals
Rotifer
PIE ROT300
of taxon i in the sample.
0.680
0.590
1,78
2.2
Percent of rotifer individuals in dominant rotifer
taxon in 300-count subsamples (coarse and fine net
Rotifer
DOM1 300 ROT PIND
samples combined)
45.70
54.61
-1.74
2.1
Percent of rotifer individuals in top 3 Rotifer taxa in
300-count subsamples (coarse and fine net samples
Rotifer
DOM3 300 ROT PIND
combined)
78.91
86.25
-2.26
1.4
Percent of rotifer individuals in top 5 Rotifer taxa in
300-count subsamples (coarse and fine net samples
Rotifer
DOM5 300 ROT PIND
combined)
91.50
94.71
-1.91
3.7
Percent of rotifer biomass in dominant Rotifer
taxon in 300-count subsamples (coarse and fine net
Rotifer
DOM1 300 ROT PBIO
samples combined)
47.97
58.94
-1.95
2.0
Percent of biomass represented by predator
individuals in 300-count subsamples (coarse and
Trophic
PRED300 PBIO
fine net samples combined)
2.06
0.93
1.86
95.5
Percent of biomass represented by predaceous
rotifer individuals in 300-count subsamples (coarse
Trophic
ROT PRED300 PBIO
and fine net samples combined)
2.06
0.93
1.86
95.5
Percent of biomass represented by herbivorous
Trophic
COPE_HERB_PBIO
copepods (coarse and fine net samples combined)
16.04
24.53
-1.96
5.0
154
NLA 2012 Technical Report. October 2024 Version 1.1
-------
Table 7-15. List of candidate metrics used to develop the zooplankton MMI for the Western Mountains bio-region
f value
Mean Value for
Mean Value for
(Least disturbed vs.
Metric
Least disturbed
Most disturbed
Most disturbed
SignakNoise
Category
Metric Name
Description
Sites
Sites
Sites)
Value
Percent of distinct native taxa within the
cladoceran family Bosminidae (coarse and fine net
Cladoceran
BOSM NAT PTAX
samples combined)
5.59
3.96
2.16
1.3
Number of distinct taxa within the subclass
Copepod
COPE NTAX
Copepoda (coarse and fine net samples combined)
2.6
3.3
-2.15
1.7
Percent of distinct taxa within the subclass
Copepod
COPE PTAX
Copepoda (coarse and fine net samples combined)
14.33
18.08
-2.29
1.9
Number of distinct native taxa within the subclass
Copepod
COPE NAT NTAX
Copepoda (coarse and fine net samples combined)
2.6
3.3
-2.07
1.7
Percent of distinct native taxa within the subclass
Copepod
COPE NAT PTAX
Copepoda (coarse and fine net samples combined)
14.33
18.00
-2.21
1.9
Total density of individuals within the subclass
Copepod
COPE DEN
Copepoda (coarse and fine net samples combined)
177.8479619
156.08843077
0.3
1.6
Total biomass of individuals within the copepod
order Calanoida (coarse and fine net samples
Copepod
CALAN BIO
combined)
12.010544048
27.035772872
-1.73
4.4
Total biomass of native individuals within the
copepod order Calanoida (coarse and fine net
Copepod
CALAN NAT BIO
samples combined)
12.010544048
27.025444897
-1.73
4.4
Percent of distinct larger-sized taxa
(NET SIZECLS NEW=COARSE; coarse and fine net
Richness/Diversity
COARSE PTAX
samples combined)
39.75
45.09
-1.87
2.3
Percent of distinct taxa collected only in the fine-
mesh net (50-um; NET SIZECLS NEW=FINE; coarse
Richness/Diversity
FINE PTAX
and fine net samples combined)
60.25
54.91
1.87
2.3
Simpson Diversity based on the total density
individuals (coarse and fine net samples combined).
Calculated as SUM{p(i)*p(i)} where p(i) is the
Richness/Diversity
SIMPSON DEN
proportion of density of taxon i in the sample.
0.288
0.353
-1.46
1.25
Percent distinct rotifer taxa (coarse and fine net
Rotifer
ROT PTAX
samples combined)
60.26
54.91
1.87
2.5
Percent distinct taxa that are within the rotifer
order Ploima (coarse and fine net samples
Rotifer
PLOIMA PTAX
combined)
48.72
42.00
2.28
4.3
Simpson Diversity based on the number of rotifer
individuals (coarse and fine net samples combined).
Calculated as SUM{p(i)*p(i)} where p(i) is the
Rotifer
SIMPSON ROT
proportion of taxon 1 in the sample.
0.325
0.414
-1.79
1.4
Percent of distinct taxa that are omnivorous
Trophic
COPE OMNI PTAX
copepods (coarse and fine net samples combined)
5.44
8.65
-2.526
1.5
155
NLA 2012 Technical Report. October 2024 Version 1.1
-------
f value
Mean Value for
Mean Value for
(Least disturbed vs.
Metric
Least disturbed
Most disturbed
Most disturbed
SignakNoise
Category
Metric Name
Description
Sites
Sites
Sites)
Value
Metrics Derived from 300-count Subsamples of Coarse and Fine Net Samples
Abundance/
Total biomass of individuals in 300-count
Biomass/
subsamples (coarse and fine net samples
Density
TOTL300 BIO
combined)
90.072878905
270.55043706
-3.09
1.4
Abundance/
Total biomass of native individuals in 300-count
Biomass/
subsamples (coarse and fine net samples
Density
TOTL300 NAT BIO
combined)
90.072878905
269.19077886
-3.07
1.4
Abundance/
Biomass/
Biomass of individuals in 300-count subsample of
Density
ZOCN300 BIO
coarse net sample (150 um)
81.538501524
226.56640233
-2.68
2.2
Abundance/
Biomass/
Biomass of native individuals in 300-count
Density
ZOCN300 NAT BIO
subsample of coarse net sample (150 um)
81.538501524
225.20674414
-2.65
2.2
Biomass represented by individuals of large-sized
Abundance/
taxa in 300-count subsamples
Biomass/
(NET_SIZE_CLS=COARSE; coarse and fine net
Density
COARSE300 BIO
samples combined)
83.550340952
235.93896061
-2.77
3.0
Biomass represented by native individuals of large-
Abundance/
sized taxa in 300-count subsamples
Biomass/
(NET_SIZE_CLS=COARSE; coarse and fine net
Density
COARSE300 NAT BIO
samples combined)
62.150708119
234.5793024
-2.74
3.1
Percent biomass of native individuals of large-sized
Abundance/
taxa in 300-count subsamples
Biomass/
(NET_SIZE_CLS=COARSE; coarse and fine net
Density
COARSE300 NAT PBIO
samples combined)
85.15
75.20
1.88
5.7
Biomass of individuals within the suborder
Cladocera in 300-count subsamples (coarse and fine
Cladoceran
CLAD300 BIO
net samples combined)
62.150708119
173.03849657
-2.301
2.2
Biomass of native individuals within the suborder
Cladocera in 300-count subsamples (coarse and fine
Cladoceran
CLAD300 NAT BIO
net samples combined)
61.59444164
171.73934691
-2.28
2.2
Biomass represented by large cladoceran
individuals (SUBORDER=CLADOCERA and
CLADOCEAN_SIZE=LARGE) in 300-count subsamples
Cladoceran
LGCLAD300 BIO
(coarse and fine net samples combined)
54.826014262
142.47459983
-1.92
2.2
Percent of large cladoceran individuals
(SUBORDER=CLADOCERA and
CLADOCEAN_SIZE=LARGE) in 300-count subsamples
Cladoceran
LGCLAD300 PIND
(coarse and fine net samples combined)
20.42
14.14
2.22
1.8
Biomass represented by native large cladoceran
individuals (SUBORDER=CLADOCERA and
CLADOCEAN_SIZE=LARGE) in 300-count subsamples
Cladoceran
LGCLAD300_NAT_BIO
(coarse and fine net samples combined)
54.826014262
142.37664379
-1.91
2.2
156
NLA 2012 Technical Report. October 2024 Version 1.1
-------
f value
Mean Value for
Mean Value for
(Least disturbed vs.
Metric
Least disturbed
Most disturbed
Most disturbed
SignakNoise
Category
Metric Name
Description
Sites
Sites
Sites)
Value
Percent of native large cladoceran individuals
(SUBORDER=CLADOCERA and
CLADOCEAN_SIZE=LARGE) in 300-count subsamples
Cladoceran
LGCLAD300 NAT PIND
(coarse and fine net samples combined)
20.41
13.47
2.49
1.8
Percent of distinct native taxa that are large
cladocerans (SUBORDER=CLADOCERA and
CLADOCEAN_SIZE=LARGE) in 300-count subsamples
Cladoceran
LGCLAD300 NAT PTAX
(coarse and fine net samples combined)
16.37
12.90
2.12
2.3
Biomass of individuals within the family Daphniidae
in 300-count subsamples (coarse and fine net
Cladoceran
DAPHNIID300 BIO
samples combined)
54.749187071
150.72825063
-2.08
3.0
Biomass of native individuals within the family
Daphniidae in 300-count subsamples (coarse and
Cladoceran
DAPHNIID300 NAT BIO
fine net samples combined)
54.749187071
150.63029459
-2.08
3.0
Total biomass of individuals within the subclass
Copepoda in 300-count subsamples (coarse and
Copepod
COPE300 BIO
fine net samples combined)
22.109055071
66.786813029
-2.76
2.0
Total biomass of native individuals within the
subclass Copepoda in 300-count subsamples
Copepod
COPE300 NAT BIO
(coarse and fine net samples combined)
22.109055071
66.726304529
-2.75
2.0
Total biomass of individuals within the copepod
order Calanoida in 300-count subsamples (coarse
Copepod
CALAN300 BIO
and fine net samples combined)
14.414470595
36.214300186
-2.00
3.2
Total biomass of native individuals within the
copepod order Calanoida in 300-count subsamples
Copepod
CALAN300 NAT BIO
(coarse and fine net samples combined)
14.414470595
36.153791686
-1.99
3.2
Number of distinct taxa in the 300-count subsample
Richness/Diversity
ZOFN300 NTAX
from the fine net sample (50-um)
7.3
8.4
-1.69
1.9
Simpson diversity based on number of individuals
Richness/Diversity
SIMPSON300 NIND
(coarse and fine net samples combined)
0.307
0.306
0.08
0
Percent of distinct taxa that are within the rotifer
family Asplanchnidae in 300-count subsamples
Rotifer
ASPLAN300 PTAX
(coarse and fine net samples combined)
0.88
2.25
-2.04
1.3
Biomass of herbivorous individuals in 300-count
subsamples (coarse and fine net samples
Trophic
HERB300 BIO
combined)
75.625607619
201.15711961
-2.56
3.1
Percent biomass of herbivorous individuals in 300-
count subsamples (coarse and fine net samples
Trophic
HERB300 PBIO
combined)
76.31
65.36
2.06
3.6
Number of distinct taxa that are omnivorous in 300-
count subsamples (coarse and fine net samples
Trophic
OMNI300 NTAX
combined)
3.0
3.6
-1.94
1.8
157
NLA 2012 Technical Report. October 2024 Version 1.1
-------
Metric
Category
Metric Name
Description
Mean Value for
Least disturbed
Sites
Mean Value for
Most disturbed
Sites
f value
(Least disturbed vs.
Most disturbed
Sites)
SignakNoise
Value
Trophic
CLAD PRED300 PTAX
Percent of distinct taxa that are predaceous
cladocerans in 300-count subsamples (coarse and
fine net samples combined)
0.87
0
2.67
Noise=0
Trophic
CLAD HERB300 BIO
Percent biomass of herbivorous cladoceran
individuals in 300-count subsamples (coarse and
fine net samples combined)
62.140336143
173.03849657
-2.30
2.2
Trophic
COPE OMNI300 BIO
Biomass of omnivorous copepod individuals in 300-
count subsamples (coarse and fine net samples
combined)
4.7491737381
24.176607243
-2.38
2.0
Trophic
COPE_OMNI300_PTAX
Percent of distinct taxa represented by omnivorous
copepod individuals in 300-count subsamples
(coarse and fine net samples combined)
8.16
11.5
-2.15
2.1
158
NLA 2012 Technical Report. October 2024 Version 1.1
-------
Chapter 8: From Analysis to Results
8.1 Background information
In the NLA 2012 report, lake condition estimates based on chemical, physical and biological
information are expressed as percent of lakes or number of lakes; therefore, site weights from
the probability design must be used to generate population estimates along with the data from
the probability sites sampled (1038). Extent estimates for biological indicators and other
measures are used to calculate relative and attributable risk.
8.2 Population Estimates
The survey design for the NLA, discussed in Chapter 2 of this report, produces a spatially-
balanced sample using the NHD+ as the sample frame. Each lake has a known probability of
being sampled (Stevens and Olsen 1999, Stevens and Olsen 2000, Stevens and Olsen 2004), and
a sample weight is assigned to each individual site as the inverse of the probability of that lake
being sampled. Sample weights are expressed in units of lakes.
The probability of a site being sampled was stratified by state and other factors. Site weights for
the survey were adjusted to account for additional sites (i.e., oversample lakes) that were
evaluated when the primary sites were not sampled (e.g., due to denial of access, being non-
target). These site weights are explicitly used in the calculation of lake condition and extent
estimates, so results can be expressed as estimates of lakes (i.e., numbers of lakes or percent of
the entire resource) in a particular condition class for the entire conterminous U.S. For
examples of how this has been done for other National Aquatic Resource Survey (NARS)
assessments, see USEPA (2006), Olsen and Peck (2008), and USEPA (2009). It is important to
note that the NLA was not designed to report on individual lakes or states, but to report at
national and regional scales.
8.3 Lake Extent Estimates
Each NLA probability site is designated as least disturbed, moderately disturbed or most
disturbed based on the appropriate indicator values and the thresholds established for that
indicator and ecoregion. Next, the site weights from the probability design are summed across
all sites in each condition class to estimate the percent of lakes nationally or in other sub
populations (ecoregions, natural vs. manmade lakes, etc) in each condition category for the
inference population. The survey design allows calculation of confidence intervals around these
condition estimates and allows for estimates of the whole resource not just those lakes
sampled. Note that only Visit 1 (i.e., the index visit) data and only probability sites are used in
the calculation of extent. Hand-selected sites have a weight of zero. Using this method, the
lakes in a particular condition class is estimated and reported in percent of lakes or number of
lakes.
159
NLA 2012 Technical Report. October 2024 Version 1.1
-------
8.4 Stressor Extent, Relative Risk and Attributable Risk
A major goal of the National Aquatic Resource Surveys is to assess the relative importance of
stressors that impact aquatic biota on a national basis. The EPA assesses the influence of
stressors in three ways: stressor extent, relative risk, and population attributable risk. In NLA,
each targeted and sampled lake was classified as being in either Good, Fair, or Poor condition,
separately for each stressor variable and for each biological response variable. From this data,
we estimated the stressor extent (prevalence) of lakes in Poor condition for a specified stressor
variable. We also estimated the relative risk of each stressor for a biological response. Relative
risk is the ratio of the probability of a poor biological condition when the stressor is poor to the
probability of a poor biological condition when the stressor is not poor (Van Sickle et al. (2006)).
Finally, we estimated the population attributable risk (AR) of each stressor for a biological
response. AR combines RR and stressor extent into a single measure of the overall impact of a
stressor on a biological response, over the entire population of lakes (Van Sickle and Paulsen
(2008)).
8.4.1 Stressor extent
For each particular stressor, the stressor extent (SE) may be reported as the number of lakes,
the proportion of lakes, or the percent of lakes in Good, Fair, Poor, or Not Assessed condition. If
the SE is reported as the proportion of lakes, then it can be interpreted as the probability that a
lake chosen at random from the population will be in Poor condition for the stressor.
Stressor extent in Poor condition is estimated as
(1) SEV, the sum of the sampling weights for sites that are assessed in Poor condition
SEp Wpj
i=l
t
(2) SEPp, as the ratio of the sums of the sampling weights for the probability selected sites
that are assessed in Poor condition divided by the sum of the sampling weights of all the
selected sites regardless of condition, i.e.,
v-i Tip
sep =5i=i^i
p Z?=1wt
, or
(3) SERp, the percent of stressor extent in Poor condition (i.e., stressor relative extent)
v 'V
Zj-_i Wr
SER == 100 * SEP = 100 # ¦ 1-1 P
P P Z?=1w,
where wpi is the weight for the /'th selected site in the Poor condition category, Wj is the weight
for the /'th selected site regardless of condition category, np is the number of selected sites that
are in Poor condition, and n is the total number of sites regardless of their condition category. A
stressor condition category may use other terminology to identify if a site is in poor condition
but generically, we use the term Poor. Note that the extent for a response variable is defined
similarly.
160
NLA 2012 Technical Report. October 2024 Version 1.1
-------
8.4.2 Relative risk and attributable risk
To estimate relative risk and attributable risk, we restrict the sites to those that both the
stressor and response variable assessed as Good, Fair, or Poor (or their equivalents). That is, if a
site is Not Assessed for either the stressor or response variable, it is dropped. Next, for these
sites the condition classes are combined to be either Poor or Not Poor for the stressor and
response variables. For example, Not Poor combines the Good and Fair condition classes. Thus,
each sampled lake was designated as being in either Poor (P) or Not Poor (NP) condition for
each stressor and response variable separately.
To estimate the relative risk and attributable risk for one stressor (S) and one response (B)
variable, we compiled a 2x2 table (Table 8-1), based on data from all lakes that were included in
the probability sample and that had both the stressor and response variable measured. A
separate table must be compiled for each pair of stressor and response variables.
Table 8-1. Extent estimates for response and stressor categories
Response(B)
Stressor (S)
Not Poor (NP)
Poor(P)
Not Poor (NP)
nnn
a ~ wnni
(=1
nnp
b=^j wnpi
(=1
npn
nPP
Poor(P)
II
\y.
$
2.
d y Wppi
(=i
i=1
Table entries (a, b, c, d) are the sums of the sampling weights of all sampled lakes that were
found to have each combination of Poor or Not Poor condition for stressor and response. For
example, d = wppi where npp is the number of sites with both the stressor and response
in poor condition and wppi is the weight for the /'th site. Note that the estimates in Table 10-1
may differ from the stressor extent estimates since both the stressor and response variables
must be measured at each site.
¦ Relative risk
Relative risk (RR) is the ratio of the probability of a Poor biological condition when the stressor
is Poor to the probability of a Poor biological condition when the stressor is Not Poor. That is,
Pr(5 = P\S = P)
RR ~ Pr(B = P\S = NP)
Using the simplified notation in Table 10-1, relative risk (RR) is estimated as:
d/{b + d)
est c/(a + c)
A RR = 1.0 indicates there is no association between the stressor and response. That is, a Poor
response condition in a lake is equally likely to occur whether or not the stressor condition is
Poor. A RR > 1.0 indicates that a Poor response condition is more likely to occur when the
161
NLA 2012 Technical Report. October 2024 Version 1.1
-------
stressor is Poor. For example, when the RR is 2.0, the chance that a lake is in Poor biological
(response) condition is twice as likely when the stressor is Poor than when the stressor is Not
Poor. Further details of RR and its interpretation, including estimation of a confidence interval
for RRest, can be found in Van Sickle et al. (2006).
¦ Attributable risk
Population attributable risk (AR) measures what percent of the extent in Poor condition for a
biological response variable can be attributed causally to the Poor condition of a specific
stressor. AR is based on a scenario in which the stressor in Poor would be entirely eliminated
from the population of lakes, e.g., by means of restoration activities. That is, all lakes in Poor
condition for the stressor are restored to the Not Poor condition. AR is defined as the
proportional decrease in the extent of Poor biological response condition that would occur if
the stressor were eliminated from the population of lakes. Mathematically, AR is defined as
(Van Sickle and Paulsen (2008))
Pr{B = P)-Pr{B = P\S = NP)
AR ~ Pr(B = P)
We estimated AR as
BEPV c/(a + c)
ARest = ' -
est BEPp
where
(c + d)
BEPp = - ^-
(a + b + c + d)
and is the estimated proportion of the biological response that is in Poor condition. We
calculated a confidence interval for ARest following Van Sickle and Paulsen (2008).
An AR can take a value between 0 and 1. A value of 0 indicates either "No association" between
stressor and response, or else a stressor has a zero extent, i.e., is not present in the population.
A strict interpretation of AR in terms of stressor elimination, as described above, requires one
to assume that the stressor-response relation is strongly causal and that stressor effects are
reversible. Van Sickle and Paulsen (2008) discuss the reality of these assumptions, along with
other issues such as interpreting them when multiple, correlated stressors are present, and
using them to express the joint effects of multiple stressors.
However, AR can also be interpreted more informally, as a measure that combines RR and SE
into a single index of the overall, population-level impact of a stressor on a response. Van Sickle
and Paulsen (2008) show that the population attributable risk can be written as
SEPJRR - 1)
AD = 1_ _
1 +SEPp(RR - 1)
This shows that the numerator of AR is the product of the SE of Poor stressor condition and the
"excess" RR, i.e., RR-1, of that stressor. The denominator standardizes this product to yield AR
values between 0 and 1. Thus, a high AR for a stressor indicates that the stressor is widely
162
NLA 2012 Technical Report. October 2024 Version 1.1
-------
prevalent (has a high SE of Poor condition), and the stressor also has a large effect (high RR) in
those lakes where it does have Poor condition.
8.4.3 Considerations When Calculating and Interpreting Relative Risk and Attributable
Risk
It is important to understand that contingency tables are created using a categorical, two-by-
two matrix; therefore, only two condition classes / stress levels can be used. There are three
ways in which condition classes / stress levels can be used for contingency tables:
Good vs. Poor
Good vs. Not-Good
Not-Poor vs. Poor
where, "Not Good" combines fair and poor condition classes, and "Not Poor" combines good
and fair condition classes. In the first bulleted method, "Good vs. Poor" data associated with
the fair condition class is excluded from the analysis. Therefore, the results of the associated
calculation of relative risk are affected by which one of the above combinations is used to make
the contingency tables, and it is crucial that the objectives of the analysis are carefully
considered to help guide this decision. For the NLA, for non-biological condition indicators (e.g.,
nutrients, physical habitat, etc.), a condition / stressor-level contingency table was created,
comparing the Not Poor condition class (i.e., a combination of good condition and fair
condition) to Poor condition class. This decision was made to indicate which stressors policy
makers and managers may want to prioritize for management efforts to improve poor
condition. After creating contingency tables, relative risk for each indicator was calculated.
A second consideration is that relative risk does not model joint effects of correlated stressors.
In other words, each stressor is modeled individually, when in reality, stressors may interact
with one another potentially increasing or decreasing impact on condition. This is an important
consideration when interpreting the results associated with relative risk.
To appropriately interpret attributable risk, it is important to understand that attributable risk
is associated with the following three major assumptions:
Causality, or that the stressor causes an increased probability of poor condition;
Reversibility, or that if the stressor is eliminated, causal effects will also be eliminated; and,
Independence, or that stressors are independent of each other, so that individual stressor
effects can be estimated in isolation from other stressors.
These assumptions should be kept in mind when applying these results to management
decisions.
163
NLA 2012 Technical Report. October 2024 Version 1.1
-------
Attributable risk provides much needed insight into how to prioritize management for the
improvement of our aquatic ecosystems - lakes, in the case of the NLA. While the results of
attributable risk estimates are presented as percent area in poor condition that could be
reduced if the effects of a particular stressor were eliminated, these estimates are meant to
serve as general guidance as to what stressors are affecting condition and to what degree
(relative to the other stressors evaluated).
8.5 NLA 2( ^ versus NLA 2012 Change Analysis
8.5.1 Background information
One of the objectives of the National Lakes Assessment (NLA) is to track changes over time. The
NLA conducted in 2012 was the second statistically valid survey of the nation's lakes and
reservoirs. Previously, EPA and partners reported on the condition of the nation's natural and
man-made lakes in the 2007 National Lakes Assessment. In NLA 2007, lakes 4 hectares and
larger were sampled. As discussed earlier in the technical report, the NLA 2012 expanded the
target population to include lakes within a smaller size class category (1-4 hectares). Because of
this change in design between the two surveys, the change analysis can only assess lakes equal
to or greater than 4 hectares. As with other NLA 2012 analyses, differences in the population
condition estimates between surveys included both natural and man-made lakes.
8.5.2 Data preparation
All sites from NLA 2007 and all but 87 lakes (those from 1-4 hectares in size) from NLA 2012
were used in the change analysis. Due to changes in methodologies between NLA 2007 and NLA
2012, change estimates could not be made for some indicators, including zooplankton, total
mercury, and methyl mercury. Additionally, change analysis was not conducted for acidification
due to the relatively small percentage of lakes in condition classes other than least disturbed.
Additionally, no changes analysis was conducted for atrazine since this indicator was not
included in NLA 2007. All other indicators reported on in the NLA 2012 report were included in
the change analysis.
8.5.3 Methods
Change analysis was conducted through the use of the spsurvey 3.3 package in R (Kincaid and
Olsen, 2016). Within the GRTS (Generalized Random Tessellation Stratified) survey design,
change analysis can be conducted on continuous or categorical response variables (e.g. least
disturbed, moderately disturbed, and most disturbed). The analysis measures the difference
between response variables of two separate surveys. For NLA 2012, the categorical response
variables were used to compare changes between NLA 2007 and NLA 2012. When using
categorical response variables, change is estimated by the difference in category estimates
from the two surveys. Category estimates are defined as the estimated proportion of values in
each category, for example least disturbed, moderately disturbed, and most disturbed
164
NLA 2012 Technical Report. October 2024 Version 1.1
-------
categories. Change between the two years is statistically significant when the resulting error
bars around the change estimate do not cross zero.
8.6 Literature cited
Kincaid, T. M., and A. R. Olsen. 2016. spsurvey: Spatial Survey Design and Analysis. R package
version 3.3.
Olsen, A. R., and D. V. Peck. 2008. Survey design and extent estimates for the Wadeable
Streams Assessment. Journal of the North American Benthological Society 27:822-836.
Stevens, D. L., Jr, and S. F. Jensen. 2007. Sampling design, implementation, and analysis for
wetland assessment. Wetlands 27:515-523.
Stevens, D. L., Jr, and A. R. Olsen. 1999. Spatially restricted surveys over time for aquatic
resources. Journal of Agricultural, Biological, and Environmental Statistics 4:415-428.
Stevens, D. L., Jr, and A. R. Olsen. 2000. Spatially restricted random sampling designs for design-
based and model based estimation. Pages 609-616 in Accuracy 2000: Proceedings of the
4th International Symposium on Spatial Accuracy Assessment in Natural Resources and
Environmental Sciences. Delft University Press,The Netherlands.
Stevens, D. L., Jr, and A. R. Olsen. 2004. Spatially-balanced sampling of natural resources.
Journal of American Statistical Association 99:262-278.
Van Sickle, J., J. L. Stoddard, S. G. Paulsen, and A. R. Olsen. 2006. Using relative risk to compare
the effects of aquatic stressors at a regional scale. Environmental Management 38:1020-
1030.
Van Sickle, J., and S. G. Paulsen. 2008. Assessing the attributable risks, relative risks, and
regional extents of aquatic stressors. Journal of the North American Benthological
Society 27:920-931.
USEPA. 2006. Wadeable Streams Assessment: A Collaborative Survey of the Nation's Streams.
US Environmental Protection Agency, Office of Water and Office of Research and
Development, Washington, DC.
USEPA. 2009. National Lakes Assessment: A Collaborative Survey of the Nation's Lakes. US
Environmental Protection Agency, Office of Water and Office of Research and
Development, Washington, DC.
165
NLA 2012 Technical Report. October 2024 Version 1.1
-------
Chapter 9: Quality Assurance Summary
The NLA has been designed as a statistically valid report on the condition of the Nation's lakes
at multiple scales, i.e., ecoregion (Level II), and national, employing a randomized site selection
process. The NLA is an extension of the EMAP methods for assessing lakes, similar to the 1997
Northeastern Lakes Assessment; therefore, it uses similar EMAP-documented and tested field
methods for site assessment and sample collection as the Northeast Lakes Assessment.
Key elements of the NLA Quality Assurance (QA) program include:
Quality Assurance Project Plan - A Quality Assurance Project Plan (QAPP) was developed and
approved by a QA team consisting of staff from EPA's Office and Wetlands, Oceans and
Watersheds (OWOW) and Office of Environmental Information (OEI) and a Project QA Officer.
All participants in the program signed an agreement to follow the QAPP standards. Compliance
with the QAPP was assessed through standardized field training, site visits, and audits. The
QAPP addresses all levels of the program, from collection of field data and samples and the
laboratory processing of samples to standardized/centralized data management.
Field training and sample collection - EPA provided training sessions throughout the study area
(with at least one instructor in each session) for all field crew members of each field crew team.
All field teams were audited on site within the first few weeks of fieldwork. Adjustments and
corrections were made on the spot for any field team problems. To assure consistency, EPA
supplied standard sample/data collection equipment and site container packages for all random
site, reference site, and repeat site sample collections.
Water chemistry laboratory QA procedures - NLA used the same single lab for all water
chemistry samples. The Western Ecology Division (WED) was responsible for QA oversight in
implementing the NLA QAPP and lab standard operating procedures (SOPs) for sample
processing.
Zooplankton laboratory QA procedures - NLA used four labs, all four were audited for
adherence to the NLA QAPP/SOP for benthic sample processing. This included internal quality
control (QC) checks on sorting and identification of zooplankton and the use of the Integrated
Taxonomic Information System for correctly naming species collected, as well as the use of a
standardized data management system. Independent taxonomists were contracted to perform
QC analysis of 10% of each labs samples (audit samples).
Benthic macroinvertebrate laboratory QA procedures - NLA used one lab, this lab was audited
for adherence to the NLA QAPP/SOP for benthic macroinvertebrate sample processing. This
included internal quality control (QC) checks on sorting and identification of benthic
macroinvertebrates and the use of the Integrated Taxonomic Information System for correctly
naming species collected, as well as the use of a standardized data management system.
Independent taxonomists were contracted to perform QC analysis of 10% of each labs samples
(audit samples).
Entry of field data - NLA used a standardized data management structure, i.e., the same
standard field forms for data collected in the field, with centralized data entry through scanning
166
NLA 2012 Technical Report. October 2024 Version 1.1
-------
in to electronic data files. Internal error checks were used to confirm data sheets were filled out
properly.
Records management - These records include (1) planning documents, such as the QAPP,
SOPs, and assistance agreements and (2) field and laboratory documents, such as data sheets,
lab notebooks, and audit records. These documents are ultimately to be maintained at EPA. All
data will eventually be archived in the STORET data warehouse at www.epa.gov/STORET.
167
NLA 2012 Technical Report. October 2024 Version 1.1
------- |