£EPA
United States
Environmental Protection
Agency
EPA/600/R-18/025
April 2015
www.epa.gov/ord
Development of a
Methodology for the
Derivation of Aquatic Plant
Water Quality Criteria
Aquatic Plant HC5
10000
FIFRA < Full
'Glyphosate
1000
Chromium
100
Selenium
LO
Metolaqbrfor
Lead,'£
Cadrpfum«j£ pgX
'Tedrfuthylazrfie
lickel
Zinc
(trazine.
MetribuziQ^®-
Moron •linuron
Arsenic V
Prometr,yii 0 J
DUjuat # /
Copper
TricJ<5san
Irgarol
FIFRA > Full
0.01
0.01
0.1
1
10
100
1000
10000
FIFRA Data HC5
Office of Research and Development
National Health and Environmental Effects Research Laboratory
-------
-------
EPA/600/R-18/025
April 2015
www.epa.gov/ord
Development of a Methodology
for the
Derivation of Aquatic
Plant Water Quality Criteria
by
Glen Thursby
Atlantic Ecology Division
National Health and Environmental Effects Research Laboratory
Narragansett, RI
Michael Lewis
Gulf Ecology Division
National Health & Environmental Effects Research Laboratory
Gulf Breeze, FL
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC 20460
-------
NOTICE
The U.S. Environmental Protection Agency through its Office of Research and Development
funded and managed the research described herein. This document has been subjected to the
Agency's peer and administrative review and has been approved for publication as an EPA
document. Mention of trade names or commercial products does not constitute endorsement or
recommendation for use.
This report is contribution number ORD-025196 of the Atlantic Ecology Division, National
Health and Environmental Effects Research Laboratory, Task SSWR 7.1A Highly Targeted
Programmatic Support, Product 7.21B FY2013/2014 OW White Paper.
11
-------
TABLE OF CONTENTS
NOTICE ii
LIST OF ACRONYMS iv
ACKNOWLEDGEMENTS v
EXECUTIVE SUMMARY vi
1. INTRODUCTION 1
2. METHODS 5
2.1 Data Selection 5
2.2 Data Standardization 6
2.3 Data Analyses 7
2.4 SSD Equations 7
2.5 Simulations 9
3. RESULTS AND DISCUSSION 10
3.1 SSD Plots and Representativeness of FIFRA-5 10
3.2 Uncertainty Related to "Correct" Proportional Rank 15
3.3 Examples 23
3.4 Addition ofMyriophyllum Data 27
3.5 Use of Growth Rate as a Standard Endpoint 27
3.6 Representativeness of All Available Data 28
3.7 Summary of the Report's Approach 29
4. CONCLUSIONS 30
5. REFERENCES 31
APPENDICES
A. List of Data Used for Each Toxicant, Organized Alphabetically 34
B. Lognormal Probability Plots for Species Sensitivity Distributions 51
in
-------
List of Acronyms
ALWQC
Aquatic Life Water Quality Criteria
DER
Data Evaluation Record
ECx
Effective Concentration for x% of response
FAV
Final Acute Value
FCV
Final Chronic Value
FIFRA
Federal Insecticide, Fungicide and Rodenticide Act
FPV
Final Plant Value
HC5
Hazard Concentration for 5th percentile
IC50
Inhibition concentration for 50% of organisms
LC50
Lethal Concentration for 50% of organisms
LOEC
Lowest Observed Effect Concentration
MDR
Minimum Data Requirements
NOEC
No Observed Effect Concentration
OPP
Office of Pesticide Programs
OW
Office of Water
SSD
Species Sensitivity Distribution
USEPA
United States Environmental Protection Agency
iv
-------
Acknowledgements
Database searches and QA checks for arsenic, cadmium, chromium, copper, glyphosate, lead,
nickel, selenium, terbuthylazine, triclosan and zinc, were performed by personnel from the Great
Lake Environmental Center, under contract to EPA's Office of Water. Diana Eignor (OW),
Autumn Oczkowski (AED), Peg Pelletier (AED) and Walter Berry (AED) provided useful
comments on an earlier version of the report.
v
-------
EXECUTIVE SUMMARY
This report supplements a recent data harmonization effort between the Office of Pesticide
Programs and the Office of Water (OW) (USEPA 2011). The original effort provided preliminary
evidence for the ability of minimal aquatic plant data sets, primarily for herbicides, to derive
species sensitivity distributions (SSDs). This report expands the scope of the evaluation to include
categories of other compounds and provides a method to evaluate the adequacy of limited toxicity
data sets for the derivation of Aquatic Life Water Quality Criteria (ALWQC) for plants.
Aquatic plants form the base of most aquatic food chains, comprise biodiversity-building habitats
and are functionally important in carbon assimilation and oxygen evolution. The USEPA, as stated
in the Clean Water Act, establishes criterion values for various pollutants found in the waters of
the United States. These criteria serve as guidance for States and Tribes to use in developing their
water quality standards. The current OW methodology for deriving criteria is in the Guidelines for
Deriving Numerical National Water Quality Criteria for the Protection of Aquatic Organisms and
Their Uses (USEPA 1985). These OW guidelines focus primarily on deriving criteria based on
animal toxicity data. Data for aquatic plants, however, has recently become important due to the
need to address the particular modes of action of herbicides. In the USEPA guidelines, an acute
criterion results from using acute toxicity data for animals. A chronic criterion uses the most
sensitive of the final chronic value (FCV) for animals or the final plant value (FPV). There are
some limitations associated with using the FPV because of an insufficient description of minimum
data requirements (MDRs) for plants within the 1985 guidelines. The availability of a more
chemically- and species-diverse phytotoxicity database is thought to be desirable and was
evaluated for this report to provide a more definitive recommendation for the optimal minimum
dataset needed to assess the risk of chemicals to aquatic plants.
The evaluation conducted for this report uses data sets for more chemicals, the majority of which
comes from USEPA's publically accessible ECOTOX database. It includes only compounds for
which data for a minimum of ten species of aquatic plants (both vascular and non-vascular) are
available. Because of the typical scarcity of data, data sets for both freshwater and saltwater were
combined. Unlike requirements for aquatic animals, the report recommends that minimum data
requirements do not necessarily require a fixed number of test species or a particular list of
desirable families. Rather, the recommendations rely on guidance for examining the overall quality
and representativeness of the sensitivity distribution of the aquatic plant community, relying in
part on the ratio of the highest to lowest EC50 values. Similar to the previous evaluation, and based
on our best judgement, test results using only the five recommended FIFRA (Federal Insecticide,
Fungicide and Rodenticide Act) aquatic plant species will provide enough information for a
reasonable estimate of an aquatic plant sensitivity distribution for a chemical of concern1. This
conclusion is preliminary and will need periodic evaluation as additional phytotoxicity information
becomes available, primarily for "non-traditional" taxa such as saltwater and vascular plant
species. This report also provides a means to evaluate small toxicity data sets to estimate the value
of additional data. The method relies on the observed and expected ratio of toxicity values for the
least and most sensitive species in a given data set.
1 FIFRA species selected out of convenience. The report does not recommend these as an MDR; rather it shows the
utility of using these species as a starting point.
vi
-------
1. INTRODUCTION
Historically, there have been a variety of obstacles to the development and use of minimum data
requirements for deriving water quality criteria for aquatic plants. The most well-known is a two
page paper by Kenaga and Moolenar (1979) in which the authors provide their evidence (using
data for thousands of chemicals) for "confirming the generally held view that aquatic animals are
more sensitive to chemical toxicants than aquatic plants". A few years later, Wang (1984) wrote
an editorial questioning the validity of this conclusion based on the plant test procedures used.
Nonetheless, detailed work on the development and interpretation of toxicity test procedures for
aquatic plants still lags behind that for aquatic animals. In addition, the current USEPA
methodology for deriving water quality criteria (USEPA 1985) states that "results of tests with
plants usually indicate that criteria which adequately protect aquatic animals and their uses will
probably also protect aquatic plants and their uses". This assumption clearly has not held true for
herbicides and cannot be assumed correct for other pesticides or even chemical pollutants in
general (Lewis 1990, 1995; Wang and Freemark 1995). Still, there seems to persist among some
the misconception that phytotoxicity information is needed only when phytotoxins are the
chemicals of concern.
A difficulty with deriving hazard concentration values for aquatic plants using data sets of various
sizes is the fact that there are essentially no minimum data requirements (MDRs). The availability
of guidelines for the derivation of Aquatic Life Criteria (ALC) using phytotoxicity results, unlike
for aquatic animals, has been characterized by historical uncertainty. The lack of this foundation
has restricted the use of phytotoxicity data for criteria derivation for most chemicals. The
environmental significance of this exclusion is unknown, but does show a need for
recommendations on how to address protection of aquatic plants. There is a long history within
ALC development of standardized use of aquatic animal toxicity data, which includes the existence
of MDRs for both freshwater and saltwater species. Except for the occasional call for additions to
an MDR (e.g., amphibians), these calculated criteria are generally accepted as the norm for use in
establishing thresholds of acceptable exposure. In contrast, there is no MDR history for aquatic
plants. Consequently, there is no commonly accepted norm for minimum plant data requirements
that sufficiently represents the taxonomic diversity for any given aquatic plant community. In
addition, unlike for animals where the standard endpoints of survival, growth and reproduction are
commonly accepted, there is no consensus on standard endpoints for most vascular and non-
vascular aquatic plants. Even though abundance or growth rate are the more commonly used
endpoints for microalgal tests, there are many other potentially acceptable endpoints for plants,
including phytostimulation. This is especially true for both freshwater and saltwater non-flowering
macrophytes whose life histories generally are much more complex than those of aquatic animals.
The current minimum guidance for derivation of ALC for aquatic plants is to use the lowest toxic
effect value of the available phytotoxicity data. Plant test values acceptable for ALC are "the result
of a 96-hr test conducted with an alga or a chronic test conducted with an aquatic vascular plant"
(USEPA 1985). Although a plant effects value is not required for the derivation of ALC, to
calculate such a value, the minimum data required is one acceptable test with a freshwater alga or
vascular plant for the freshwater criterion. Similarly, the result of only one saltwater species is
required to establish a saltwater plant criterion. If plants are likely to be more sensitive than animals
to a given compound (e.g., herbicides), then the "results with a plant in another phylum (division)
1
-------
should also be available". The 1985 guidelines further describe the calculation of the plant value
as a number that "should be obtained by selecting the lowest result from an acceptable test with an
important aquatic plant species in which the concentrations of test material were measured and the
endpoint was biologically important". The definitions of "important aquatic plant" and
"biologically important endpoint", however, are not given. The lack of specificity for plant toxicity
data was due largely to the fact that at the time of the writing of the 1985 guidelines the "procedures
for conducting tests with plants and interpreting the results of such tests [were] not as well
developed". Current scientific understanding on how to conduct plant toxicity tests and interpret
their results is much better. Nevertheless, standardization of procedures and selection of relevant
endpoints remain issues. There still is no published guidance on how many species and which
taxonomic groups to include to reasonably represent the sensitivity of the aquatic plant community.
The selection of plant test species is essentially based on availability and ease of culture.
Even if there was acceptable guidance for test species and biological endpoint selection,
uncertainty remains for data evaluation. Do we use EC50s, EC20s or some other percent effect
concentration? Do we use the lowest available effect concentration (as current guidance suggests),
or do we use a species sensitivity distribution (as done with aquatic animal data), and if the latter,
at what percentile? The fifth percentile is common, but is that the most appropriate for aquatic
plants? This report will not successfully address these issues due to an insufficient database.
However, some improvement to the current situation is provided.
Before we begin, we define what is included under the category of "aquatic plants". In other words,
what exactly are we trying to protect using aquatic plant criteria values? The definition of "plant"
varies among taxonomists. All vascular plants and some non-vascular aquatic plants are in the
Kingdom Plantae, and some taxonomists reserve the term "plant" to represent only those
organisms in this kingdom. Among the non-vascular aquatic plants, free-living algae include
groups from four separate kingdoms2 including Bacteria (e.g., cyanobacteria), Protozoa (e.g.,
euglenoids and dinoflagellates), Chromista (e.g., diatoms and brown algae) and Plantae (e.g., green
algae and, depending on whose list you use, the red algae are either lumped in with the Plantae or
split off into their own Kingdom—Rhodophyta). To make things simpler, for this report, the term
"plant" includes all autotrophic organisms that contain chlorophyll a—which has members in each
of the above Kingdoms. This diverse group of organisms has members in practically every
conceivable freshwater and saltwater habitat. Aquatic plants have various growth habits, such as
attached (rooted or other holdfast), free-floating, submerged or emergent (only partially submerged
for part or all of their life history).
Differences in sensitivities of test species to chemicals may be from the mode of action of the
chemical and a plant's physiology. For instance, since herbicides often target various aspects of
the photosynthetic pathway, there could be differences in response between vascular plants that
have C3 vs. those with C4 pathways3. For example, Lemna species are C3 plants (Longstreth
1989), many saltmarsh grasses (e.g., species of Spartina and Distichlis) are C4 plants, while other
saltmarsh plants (e.g., species of Salicornia and Scirpus) are C3 (Drake 1989). In some species of
amphibious freshwater plants (e.g., the sedges Elocharis vivipara and E. baldwinii), the emergent
2 Note the arrangement of "life" into taxonomic categories has been in a state of flux ever since Linnaeus first
introduced Animale and Vegetabile. The main point here is not to definitively support one particular scheme or
another, but to point out that what we traditionally refer to as plants is an extremely diverse group. Kingdom
titles used above based on Cavalier-Smith (2004).
3 C3 and C4 refer to the first stable carbon compound in CO2 fixation. In C3 plants this compound is 3-
phosphoglyceric acid (a 3 carbon compound) and in C4 plants it is oxaloacetic acid (a 4 carbon compound).
2
-------
phase exhibits C4 photosynthesis while the submerged phase is C3 (Cronk and Fennessy 2001).
In addition, there is a wide variety of photosynthetic pigments among the various algal groups, so
type of accessory pigments (and thus different pigment synthesis pathways) might make a
difference in sensitivity.
Plants have more diverse life history patterns than animals. Plants often have more than one free-
living, "adult" phase, often with separate male and female plants and yet another separate
sporophyte. In some species, the gametophytic phase dominates (relative to size), in others the
sporophyte. Few, if any, studies are available comparing the relative sensitivities of different life
history patterns or even much information on the relative sensitivity of different life stages. Unlike
animal chronic toxicity tests, which may include full and partial life cycle tests and early life stage
tests, many aquatic plant tests, with few exceptions, are conducted using the adult life stage. This
is important since sexual reproduction in seaweeds is among their most sensitive endpoints
(Thursby et al. 1985, Eklund 1998) and sexual reproduction tests generally do not exist for
freshwater algae or aquatic vascular plants.
Given the large diversity in taxonomic position, morphology, physiology, life history, habit and
habitat, the task of selecting technically-defensible MDRs is challenging considering the limited
availability of toxicity data for aquatic plants. This negates using subsets of data representing
different combinations of taxonomic groups. Therefore, the creation of minimum data
requirements analogous to criteria data requirement for aquatic animals (e.g., eight taxa from
different specified Families) is not feasible at this time. As a result, rather than evaluating how
representative a collection of taxa might be of an aquatic plant community, we evaluate how well
a sensitivity distribution from a small data set represents the distribution of a larger data set. To do
this, we arbitrarily define large data set as any compound for which data for at least ten species
with "acceptable" (the methods section defines acceptable) EC50 values were available. The
assumption is this is a reasonable estimate of the community sensitivity range. From these data
sets, a log-normal species sensitivity distribution was constructed, and from this distribution, the
concentration representing the 5th percentile was calculated—referred to as the HC5. The HC
stands for "hazard concentration" and the 5 refers to the 5th percentile. These values were an
estimate of the "true" community effect concentration (or threshold).
We could define small data set numerous ways. To simplify the process, however, we decided to
use the current procedure for pesticide registration where USEPA has more guidance for selection
of aquatic plant data. Under the Federal Insecticide, Fungicide and Rodenticide Act (FIFRA),
EPA's Office of Pesticides Programs (OPP) has the authority to require data in support of the
registration of a pesticide product. With respect to aquatic plants, these are data for five species.
They include four microalgae (representing aquatic non-vascular plants) which usually are
Pseudokirchneriella subcapitata (a freshwater green alga formerly known as Selenastrum
capricornutum), Anabaena flos-aquae (a freshwater cyanobacterium), Navicula pelliculosa (a
freshwater pennate diatom) and Skeletonema costatum (a saltwater centric diatom). The fifth plant
is a species of Lemna (freshwater; duckweed)—usually L. minor or L. gibba (although in practice
most often L. gibba). The "large" data sets had the additional requirement that they must include
compounds for which toxicity results also were included for the standard five species requested by
FIFRA (or a closely related species) for pesticide registration—referred to throughout this
document as the "FIFRA-5". These latter data were used to estimate the parameters for a log-
normal distribution, and an additional HC5 was calculated for comparison with the HC5 calculated
3
-------
using the full data set—often referred to in the remainder of the paper as the "full" HC54. An
additional advantage of using FIFRA-5 species as a starting point is that there are standard test
protocols for their use. We are not proposing these species as a recommended MDR for aquatic
plants. Any set of species are acceptable using our approach. The FIFRA-5 species are a useful
starting point for the reasons given.
We include a variety of endpoints to maximize the number of species for the aquatic plant
sensitivity distributions. In addition, the use of many different endpoints likely results in a better
estimation of the true relative sensitivity of plants. Just as there is no consistently most sensitive
species or other taxonomic grouping, no single endpoint (e.g., growth rate) will always be the most
sensitive endpoint (Hanson and Solomon 2002). The selection of endpoints to include in sensitivity
distributions for aquatic animals is usually not an issue. Acute animal SSDs most often use
survival, while animal chronic SSDs frequently use multiple endpoints (survival, growth and
reproduction). Although the interpretation of chronic SSDs is perhaps less definitive than results
from animal acute SSDs5, they can predict that some portion of species will experience population
reductions ranging from slight to severe following long-term exposures (Suter et al. 2002).
Typically, data from plant tests mostly represent sublethal measurements. These sublethal effects
extend to a variety of parameters including biomass-related (e.g., germination, early seedling
growth, root elongation, total biomass, leaf injury, pigment content, protein amino acid
concentrations), activity-related (e.g., carbon dioxide uptake, oxygen evolution, variable
fluorescence) and biochemical parameters (e.g., ATP levels, enzyme activities). In practice, the
measured endpoints most often used by OPP and OW are some aspect of growth or growth rate.
Activity-related or biochemical parameters are generally not used, but there is no reason not to
include multiple endpoints in plant SSDs. In addition, although not used in SSDs, other authors
propose evaluating multiple plant endpoints simultaneously to estimate a species' relative
sensitivity. Hanson and Solomon (2002, 2004) refer to these data evaluations as Effect Measures
Distributions (EMDs); Brain et al. (2006) refer to them as Intraspecies Endpoint Sensitivity
Distributions (IESDs) sensitivity of aquatic plants.
This report provides a means to evaluate small toxicity data sets to estimate the value of additional
data. The method relies on the observed and expected ratio of toxicity values for the least and most
sensitive species in each data set. We do not recommend minimum data requirements based on a
fixed number or test species or a particular list of desirable species to include. The
recommendations rely on guidance for examining the overall quality and representativeness of the
sensitivity distribution with any available data. We show that using only the recommended FIFRA-
5 aquatic plant species as a starting point, a reasonable estimate of the aquatic plant sensitivity
distribution is possible.
4 Full HC5 is just the best estimate derived using all of the available acceptable data. This value will have more
uncertainty the fewer data points available. The optimal number of test values needed for estimating HC5 for
animal acute data ranges from 15 to 55 (Newman et al. 2000).
5 Acute SSDs not only predict that something will happen for a given exposure, they also approximate what will
happen—some degree of mortality (Suter, et al. 2002).
4
-------
2. METHODS
2.1 Data Selection
Table 1 lists the toxicants used for the analyses. There were two criteria for selection. First, there
needed to be at least ten species in the data set, and second, five of the species had to be the standard
FIFRA-5 species or a close substitute (see paragraph 7 in section 2.2, Data Standardization).
Almost all the data came from EPA's ECOTOX database6 (data inquiries updated January 2015).
Supplemental data came from OPP's DERs and from the open literature. For example, the
ECOTOX inquiry for diuron resulted in 29 species for inclusion in the SSD, but the FIFRA-5 were
not all represented. Therefore, additional data for Anabaena flos-aquae, Naviculapelliculosa and
Lemna gibba were available from an Australian government assessment report (APVMA 2011).
These data rounded out the total needed for the FIFRA-5 for diuron. Data downloaded from the
ECOTOX website and used for SSD calculations were checked for accuracy against the original
citations.
Table 1. Twenty compounds meeting the minimum requirements for inclusion in the analyses. N = number of
toxicity results. PSI = Photosystem I, PSII = Photosystem II.
Compound
CAS
Number
N
Class
Mode of Action7
Arsenic V
7440-38-2.
7778-39-4.
7778-43-0.
13464-38-5
11
Metalloid8
Cell division inhibitor—after
conversion to arsenite
Atrazine
1912-24-9
69
Herbicide
PSII inhibitor
7440-43-9.
Cadmium
10108-64-2.
10124-36-4.
10325-94-70
42
Heavy metal
Enzyme inhibitor
1333-82-0.
Chromium VI
7440-47-3.
7778-50-9.
13907-47-6
34
Heavy metal
Oxidizer
1344-67-8.
Copper
7440-50-8.
7447-39-4.
7758-89-6.
7758-98-7
76
Heavy metal
Reduce photosynthesis—substitution of
Cu for Mg in chlorophyll
Diquat
85-00-7
13
Herbicide
Membrane disruption
Diuron
330-54-1
46
Herbicide
PSII inhibitor
Glyphosate
1071-83-6
11
Herbicide
Enzyme inhibition—shikimate
biosynthetic pathway
Irgarol
28159-98-0
48
Herbicide
PSII inhibitor
7439-92-1.
Lead
7758-95-4.
10099-74-8
19
Heavy metal
Enzyme inhibitor
6 http://cfpub.epa.gov/ecotox/
7 There may be more than one direct or indirect mode of action.
8 Properties between those of a metal and a non-metal.
5
-------
Table 1. Continued
Compound
CAS
Number
N
Class
Mode of Action9
Linuron
330-55-2
10
Herbicide
PSII inhibitor
Metolachlor
51218-45-2
28
Herbicide
Amino acid synthesis inhibitor
Metribuzin
21087-64-9
18
Herbicide
PSII inhibitor
7440-02-0,
Nickel
7718-54-9,
7786-81-4,
13138-45-9
13
Heavy metal
PS II inhibitor, chloroplast membrane
disruption
Pentachlorophenol
87-86-5
27
Wood preservative
Mitochondrial ATPase inhibitor
Prometryn
7587-19-6
13
Herbicide
PSII inhibitor
7446-08-4,
7782-49-2,
Selenium
10102-18-8,
13410-01-0,
26970-82-1
13
Non-metal
Oxidizer
Terbuthylazine
5915-41-3
15
Herbicide
PSII inhibitor, oxidation of membrane
lipids
Triclosan
3380-34-5
10
Antibacterial
Fatty acid synthesis inhibitor
7440-66-6,
Zinc
7646-85-7,
7733-02-0
28
Heavy metal
PS I and PS II inhibitor
2.2 Data Standardization
Toxicity data for each compound was reported for a variety of endpoints and exposure durations.
The following actions culled the data to a standard selection for each chemical.
1. Toxicity data not expressed as EC50, IC50 or LC50 values were not used. This excluded effects for
various percentiles (e.g., EC10, EC25), as well as values forno observable or lowest observable effect
concentrations (NOEC or LOEC). Their elimination does not mean that they would not be useful in
a risk assessment. Rather for consistency we chose the 50th percentile to represent the relative
sensitivity of a given endpoint. All results are given in pg/L.
2. Any organisms identified as "algae", "phytoplankton" or some other nonspecific identifier were not
included. Some data sets include species identified only to the genus level, but not if another
representative of that genus was available and identified to species. For example, a data set would
include Chlorella vulgaris, but not Chlorella sp. However, the diquat data set contains Navicula sp.
because data were not available for a specific Navicula species.
3. Data sets exclude tests using unusual isolates, for example, data for a species for which a special
variety is available with specific resistance to an herbicide. These varieties are not representative of
that species "normal" sensitivity.
4. Data sets from tests using pulsed exposures—tests with exposure followed by a period of growth in
control medium before the endpoint is measured—were excluded. Data from tests consisting of single
exposures, but without renewal of media, were included.
9 There may be more than one direct or indirect mode of action.
6
-------
5. The exposure durations were limited to 3, 4 and 5 days for microalgal tests; 7 to 14 days for Lemna
species; and 14 to 35 days for other vascular plants. These durations were selected based on typical
lengths of standard tests using these organisms. Occasionally, durations of 2 days were included for
macroalgae. This occurred when an endpoint such as seaweed spore germination was available. These
lower limits on duration eliminated many of the non-standard endpoints (e.g., most of the variable
fluorescence data and most of the biochemical biomarkers such as enzyme or ATP concentrations).
6. The final culling or streamlining of data was to eliminate any endpoints still on the list that were non-
standard endpoints, such as any remaining enzyme concentrations or non-chlorophyll a pigment
concentrations. The final list of endpoints included in our analysis focused on biomass, cell counts,
length, weight, area, chlorophyll a concentration, photosynthetic rate and growth rate. The final plant
SSDs included only the most sensitive endpoint when multiple endpoints were available for a species.
7. Data subsets from the full data sets contain only values for the standard FIFRA species (FIFRA-5)—
the four standard microalgal species and Lemna. These data sets include Lemna minor and Lemna
gibba when values for both were available10. For most of the compounds, data for the standard FIFRA
species were available. When this was not the case, inclusion of substitutes followed the sequence of
selection from same genus, if not available, then same order, then same class, etc. Appendix A lists
the data selected for each compound.
2.3 Data Analyses
When more than one toxic effects concentration for the same endpoint (for the same species) was
available within the standardized data sets described above, the geometric mean of those data was
used for that endpoint. Results reported as greater thans were included (as absolute values), but
the greater than was retained with the resulting geometric mean. When toxic effect values for more
than one endpoint were available for the same species, the endpoint with the lowest value
represented the sensitivity of that species. One of the limitations of plant data sets is the common
lack of measured test concentrations. In order to maximize the taxonomic representation within
the data sets, values based on unmeasured chemistry were included in our analyses. In addition,
no distinction was made between static and renewal techniques. The Integrated Taxonomic
Information System (www.itis.gov) was used as the authority for any questions on species names.
For example, the currently accepted name for Scenedesmus obliquis (from Chalifour and Juneau
2011) is Scenedesmus acutus var. acutus.
2.4 SSD Equations
Six commonly used distribution equations were initially selected for evaluation: normal, log-
normal, Gumbel (also called Gompertz), log-logistic, Weibull and inverse Weibull (also called
Log Gompertz). Each distribution was evaluated in its linearized form. A function of proportional
rank was plotted on the x-axis and the concentration on the y-axis (concentration for normal and
Gumbel; In of the concentration for the other four). Proportional rank was calculated as R/(N+1),
where R was the cumulative rank of the datum and N was the total number of values in the SSD.
10 EPA's Office of Pesticides Programs maintains an Ecotoxicity Database of available toxicity data submitted as part
of the pesticide registration process. This summary database (March 2009 copy of the spreadsheet) includes the
results of 360 tests using L. gibba, and only 16 for L. minor. The probability is substantially greater that a submitted
FIFRA-5 data set would include L. gibba—which is why the analyses defaulted to that species when data for both
were available.
7
-------
The performance of each of the SSD equations was judged graphically (visually). Graphical
inspections of a linearized form is commonly used to fit a normal distribution (Chapman et al.
2007), and has been used for a variety of other distributions for which a linearized form can be
derived (Lee and Wang 2003). An additional advantage of a linearized plot is that it is easy to
observe if the data deviate from the distribution. In general, the four distributions using the natural
log of the concentration performed well, especially if confining the observations to the most
sensitive half of the SSDs. Because log-normal probability plots consistently performed well, only
those plots were used for the calculations reported herein. While for any given compound used in
this paper, one of the other SSD equations may have had a better statistical fit to the available data;
log-normal distribution was used for consistency.
The linearized form of the distribution was taken from Lee and Wang (2003). The cumulative
distribution function for a log-normal plot is:
Where:
F(c) = proportion of values less than or equal to c, i.e., the proportional rank;
c = in our case is the In of a species EC50 value from the SSD;
u = arithmetic mean of all EC50 values in the statistical population modeled by the
distribution;
o = the standard deviation of all EC50 values in the statistical population; and
-------
(full and FIFRA-5), data were fit to Equation 3 using the method of least squares, and HC5 values
calculated. Regression confidence intervals were calculated using standard procedures.
2.5 Simulations
Because of database limitations, the only effective way to evaluate a large range of potential
relationships between small and large data sets is through computer simulations. Simulations used
Microsoft® Office Excel® 2007 with Oracle's® Crystal Ball add-in (Fusion Edition, Release
11.1.2.0.00). These simulations evaluated various attributes of small data sets as they relate to
accuracy of prediction of HC5 values. For example, the relationship between the ratio of the
highest and lowest EC50 values to the ratio of the calculated and true HC5 values. Simulations
were conducted using a large dataset from a known distribution. This known distribution contained
150 EC50 values with a mean value for the natural logs of the EC50s of zero, and a standard
deviation of the log values (slope) of 2.293. This slope was selected so that the ratio of the 95th to
the 5th proportional rank EC50 values was 10,000. Each simulated data subset began by random
selection of the value for rank 1. This value then became the lower limit for rank 2, and so on for
the selection of all five values. The ratio of the high to low values was recorded, along with the
calculated HC5 using these five values. The ratio of this value to the known HC5 from the parent
distribution was also recorded. The selection of five values was repeated 10,000 times, and the
relationship between the above two ratios was plotted.
9
-------
3. RESULTS AND DISCUSSION
3.1 SSD Plots and Representativeness of FIFRA-5
Appendix A lists the final datasets used for each toxicant. Appendix B contains the lognormal
probability plots for species sensitivity distributions. Each plot shows two graphs, one for all the
available data and the other for just the FIFRA-5 data. Table 2 lists the parameters for the fit of the
linear equation (Equation 3) to each of the full data sets along with the HC5 values and their
corresponding lower 50% and 95% confidence limits. Table 3 shows similar information for the
FIFRA-5 data sets. The simplest comparison is a plot of the HC5 values from the two data sets
(Figure 1). In the absence of zinc, the uncertainty associated with using a data set as small as that
from the FIFRA-5 species is within a factor of four on either side of the one-to-one line. If this
uncertainty were acceptable, then one approach to estimating a water quality criterion for aquatic
plants would be to use a value directly calculated from a FIFRA-5 data set. Presumably, this
estimate would improve as the number of species increased. One way to reward additional data
would be by using lower confidence limits. Figures 2 and 3 are similar plots as Figure 1, except
the x-axis is the lower 50% confidence limit and the lower 95% limit, respectively, for the FIFRA-
5 value. The data points shift to the left—towards the FIFRA-5 value being more conservative
(lower than the value from the full data set). Since confidence limits are a function, in part, of the
number of data points, as the number of data points increases the lower confidence limit should
increase (approach the HC5 value)11.
Since some risk analyses rely on the lowest plant value, Figure 4 displays the relationship between
the lowest FIFRA-5 value and the HC5 calculated from the full data set. There is a general trend
towards the data shifting to the right (FIFRA values being greater than the HC5 estimate) relative
to the data distribution in Figure 1. With the data set we have, using the lowest plant value
increases the likelihood of over estimating the HC5.
Table 2. Parameters from the linear regressions of the full data sets. For compounds listed as "lower half ',
only the data from the most sensitive half of the species were used in the linear regressions. CL = confidence limit,
HC5 = hazard concentration fifth percentile (threshold concentration)
Compound
Portion
of data
N
High/Low
EC50
Ratio
Slope
Intercept
(In
EC50)
r
HC5
(lig/L)
Lower
50% CL
(lig/L)
Lower
95% CL
(lig/L)
Arsenic V
all
11
1200
2.364
4.273
0.965
1.47
1.21
0.79
Atrazine
lower
half
69
6080
1.022
4.466
0.937
16.21
15.82
15.08
Cadmium
all
42
4528
2.336
6.450
0.979
13.57
12.65
11.02
Chromium VI
lower
half
34
6380
1.503
6.677
0.949
66.96
62.76
54.83
Copper
all
76
21,233
2.436
4.345
0.984
1.40
1.33
1.20
11 However, the lower confidence limit for a data set of larger N may not necessarily be better (closer to
the calculated HC5 value) than the lower limit from an N = 5 data set since the new calculated HC5 value
is also dependent on the goodness of fit of all of the data to a straight line.
10
-------
Table 2. Continued
Compound
Portion
of data
N
High/Low
EC50
Ratio
Slope
Intercept
(In
EC50)
r2
HC5
Og/L)
Lower
50% CL
(Hg/L)
Lower
95% CL
(Hg/L)
Diquat
all
13
3920
2
824
4.134
0.908
0.60
0.42
0.20
Diuron
lower
half
46
5246
1
256
2.720
0.944
1.93
1.88
1.78
Glyphosate
all
11
19
0
932
9.224
0.895
2189.35
1909.22
1408.98
Irgarol
lower
half
48
19,051
1
911
0.470
0.939
0.069
0.064
0.055
Lead
all
19
18,615
2
837
7.665
0.985
20.04
17.99
14.41
Linuron
all
10
32
1
198
3.196
0.901
3.40
2.84
1.88
Metolachlor
all
28
6203
2
429
7.424
0.897
30.83
24.01
14.53
Metribuzin
lower
half
18
365
0
667
3.130
0.966
7.64
7.38
6.80
Nickel
all
13
2974
2
632
8.092
0.939
43.07
33.20
18.94
Pentachlorophenol
lower
half
27
8575
1
956
5.721
0.901
12.22
10.63
7.90
Prometryn
all
13
85
1
482
2.662
0.911
1.25
1.05
0.71
Selenium
all
13
1255
2
341
8.212
0.957
78.39
64.75
42.89
Terbuthylazine
all
15
15
1
298
4.107
0.941
7.18
6.40
5.01
Triclosan
all
10
10
2
706
2.908
0.961
0.21
0.17
0.10
Zinc
all
28
28
2
762
7.467
0.965
18.60
16.29
12.48
11
-------
Table 3. Parameters from the linear regressions of the FIFRA-5 data sets (algae and duckweed). CL =
confidence limit, HC5 = hazard concentration fifth percentile (threshold concentration)
Compound
High/Low
EC50
Ratio
Slop
e
Intercept
(In EC50)
r2
HC5
(Hg/L)
Lower
50% CL
(lig/L)
Lower
95% CL
(lig/L)
Arsenic V
204
2.957
5.023
0.9
58
1.17
0.72
0.16
Atrazine
7
0.987
4.165
0.8
92
12.71
9.71
4.14
Cadmium
26
1.674
4.770
0.9
48
7.51
5.52
2.09
Chromium VI
10
1.224
6.217
0.9
06
66.98
44.13
11.80
Copper
21
1.598
4.276
0.9
52
5.20
3.93
1.62
Diquat
>735
3.086
4.049
0.8
93
0.358
0.155
0.011
Diuron
4
0.749
2.464
0.8
81
3.43
2.97
1.89
Glyphosate
2
0.293
9.467
0.9
72
7979.7
8
7674.37
6783.86
Irgarol
32
1.756
-0.316
0.9
71
0.041
0.031
0.013
Lead
562
3.122
7.013
0.8
64
6.53
2.48
0.12
Linuron
5
0.752
3.474
0.9
46
9.36
8.13
5.21
Metolachlor
20
1.568
5.280
0.9
18
14.88
10.42
3.38
Metribuzin
20
1.414
2.930
0.7
36
1.83
0.94
0.12
Nickel
30
1.964
6.667
0.8
42
31.08
15.93
1.93
Pentachlorophenol
7
1.086
4.597
0.9
55
16.62
13.83
7.73
Prometryn
40
1.680
2.135
0.8
76
0.53
0.32
0.07
Selenium
32
1.686
8.462
0.9
78
295.57
242.87
130.57
Terbuthylazine
13
1.291
3.053
0.9
32
2.53
1.93
0.81
Triclosan
53
2.335
2.441
0.9
05
0.247
0.136
0.021
Zinc
2346
4.090
5.291
0.9
42
0.238
0.108
0.009
12
-------
Aquatic Plant HC5
10000
FIFRA < Full
'Glyphosate
1000
100
Selenium
LO
Metolacjtffor 9fc\cke\
Leads'
Cadnrf'um#^ pc^.
'Tedfuthylazrfie
0/ron 9'Linuron
Zinc
(trazine.
LL
Copper
TricJ<5san
Irgarol
FIFRA > Full
0.01
0.01
0.1
1
10
100
1000
10000
FIFRA Data HC5
Figure 1. Plot of relationship between calculated 5th percentile hazard concentration (HC5) using all available data (see
Table 2) and the HC5 using only the five Federal Insecticide, Fungicide and Rodenticide Act (FIFRA) species (see Table
3). Solid black line is the one-to-one relationship. Data points above this line are situations whereby the FIFRA species-based
value was less than the full data set value. Data points below the line are for a greater FIFRA value than full data set value. Hie
two dashed red lines show a factor of 4 greater than and less than the full data values (one-to-one values divided and multiplied
by 4).
13
-------
Aquatic Plant HC5
10000
FIFRA < Full
1000
100
10
1
0.1
FIFRA > Full
0.01
0.01
100
1000
10000
FIFRA: Lower 50% Confidence Limit
Figure 2. A similar plot as Figure 1 except the FIFRA data set value is the lower 50% confidence limit (see Table 3). Hie
added solid red line represents a factor of 10 lower than the HC5 value from the full data set. Note, data points move to the left
relative to their position in Figure 1.
Aquatic Plant HC5
10000
FIFRA < Full
1000
100
FIFRA > Full
0.01
0.01
0.1
1
10
100
1000
10000
FIFRA: Lower 95% Confidence Limit
Figure 3. A similar plot to Figure 1, expect the FIFRA value is the lower 95% confidence limit (see Table 3). The solid red
line represents a factor of 100 lower than the HC5 value from the full data set. Note, data points move further to the left relative
to botli Figures 1 and 2.
14
-------
10000
FIFRA < Full HC5
'Glyphosate
1000 :
100 :
Selenium
Mefrfibuzin
Ticlosan
I rgar<
FIFRA > Full HC5
0.01
0.01
0.1
1
10
100
1000
10000
FIFRA Data Lowest Value
Figure 4. Plot of HC5 calculated using all available EC50 values against the lowest datum within the FIFRA data subset.
Selection of the lowest value to establish plant criteria is what is currently listed within the 1985 guidelines. The solid black line
is the one-to-one relationship, the dotted red lines represent the same factor of four as in Figure 1, and the solid red line is a factor
of 10 below the one-to-one line. Note, data points move to the right relative to their position in Figure 1. This increases the
likelihood of overestimating the HC5.
3.2 Uncertainty Related to "Correct" Proportional Rank
Much attention is usually given in the literature concerning the repeatability of toxicity test data—
"round-robins", "ring tests", etc. This is important for several reasons, and for species sensitivity
distribution, the reliability of a reported effects value (e.g., EC50) determines the rank assigned to
that species within a data set. Little, if any attention, however, is given to the uncertainty in SSD
calculations related to incorrect proportional rank. We do not mean incorrect with respect to its
calculation within the available data set. Rather, we mean incorrect with respect to a species true
rank within the aquatic plant community. Figure 5 illustrates this point.
Figure 5A represents a hypothetical species sensitivity distribution. With respect to the SSD
equation, the mean of the available data is the "location" factor—it describes the relative location
of the data on the x-axis (EC50s). That is, how far to the left or right on the axis are the data
clustered on this axis. The total spread of the data (the standard deviation) describes the "shape"
of the curve. There is, however, another significant source of uncertainty within toxicity data for
a given compound—that related to the accuracy of the proportional rank calculation. While it is
possible to have a reasonable estimate of the true EC50 value for a species, it can be difficult to be
certain what the true proportional rank of that species is. To estimate this accurately, we would
have to test a very large number of species.
15
-------
Let us consider a couple of examples to illustrate the issue. If there were only one EC50 value in
the database for a given compound, then the proportional rank of that value would be 0.50—
1/(1+1). This means that half the data in the true sensitivity distribution would be less than and
half greater than this value (Figure 5B). However, we do not know exactly where within the full
range of sensitivity this single value resides. Since we would have only one data point, for an SSD,
we would not have much choice except to assume it is in the middle. The actual location of this
value is very likely somewhere else within the distribution. Even if we had a reasonable estimate
of the slope of the SSD, the actual curve shifts to either the left or the right (see Figure 5C as an
example). The accuracy of our toxicity assessment will depend on how close the true proportional
rank of the known value is to 0.50, which we cannot know. If we have two data points, then their
proportional ranks are 0.333 and 0.667. Again, however, the accuracy of our toxicity assessment
will depend on how close the true proportional ranks are to these two values (Figures 5D and 5E).
This process continues as the number of data points increases—with a corresponding potential for
increased complexity in the evaluation of the "correct" assignment of proportional ranks. With the
two data sets we have (full and FIFRA-5), we can compare the position of a species toxicity value
in the larger data set with that fixed by its position within the FIFRA-5 data set (0.167, 0.333,
0.500, 0.667 and 0.833). As expected, there can be quite a difference between these two ranks
(Figure 6). For example, the most sensitive FIFRA-5 species' proportional rank ranged from 0.034
to 0.42 within the corresponding full data set. This suggests caution if using the most sensitive
value as a benchmark.
16
-------
0.001 0.01 0.1 1 10 100 1000
EC50
1.0
o.g
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0.
EC50
Figure 5. Species rank. A) Hypothetical species sensitivity
distribution. B) Red data point represents the 0.50
proportional rank species. C) Hypothetical shift in species
sensitivity distribution for data point whose actual
proportional rank is significantly lower than 0.50. D) Similar
to B, except for two data points whose proportional ranks are,
by definition, 0.333 and 0.667. E) Example of shift in
proportional rank from "true" position to defined positions
based on N of two.
17
-------
1.00
0.90 ¦
0.80 ¦
0.70 ¦
0.60 ¦
0.50 ¦
0.40 ¦
0.30 ¦
0.20 ¦
0.10 ¦
0.00
i
l-
I
I
I
i
2 3 4
Rank within FIFRA-5 group
i
S
Figure 6. Plot of the proportional rank represented by each FIFRA-5 species within their corresponding full data sets.
Data are from the compounds listed in Table 1. For example, the most sensitive FIFRA-5 species (FIFRA rank 1) had a
proportional rank within the full data set that ranges from 0.034 to 0.42. The red data points represent the proportional rank
for each of the five FIFRA species (0.167 for rank 1 up to 0.833 for rank 5).
There is an aspect of the change in proportional ranks as the number of species increases that we
may be able to exploit in our quest to evaluate the representativeness of small data sets. The spread
between the calculated proportional ranks for the lowest and highest values predictably increases
as the number of data points increases (Figure 7). This means that there should be a predictable
relationship between the ratio of the EC50 values corresponding to the least and most sensitive
species in the small data set and that ratio from a much more complete data set. To demonstrate,
we have chosen the ratio of the EC50 values for the 95th and the 5th proportional ranks to represent
the "real" world. Figure 8 shows a plot of a hypothetical SSD where the 95fe and 5th ranks are
marked, along with the 0.833 and the 0.167 ranks (the high and low values for N = 5). The ratio
of the 95th to 5th values (dotted blue lines) is 10,000. If the SSD is accurate, then the ratio for the
small data set values (red dashed lines) should be 210.
Hypothetical SSDs for a range of 95th to 5th ratios (from 2.5 to 25,000) establishes the relationship
between the small data set and full data set ratios whereby we can predict one from the other—
provided we can estimate what the slope (shape factor) for the SSD is. Figure 9 illustrates this
relationship and includes the ratio for the FIFRA-5 ratio for zinc12 (2346) on the x-axis, tracing
the corresponding expected full data set ratio on the y-axis (over 625,000). Since we have the
luxury of the larger data set for zinc (N = 28), we know that the high/low ratio from that larger
data set is 18,541. This suggests that the estimated ratio of over 625,000 is probably too high,
further suggesting that the reason zinc is an outlier in Figure 1 is because the proportional ranks
for its FIFRA data subset are substantially different from their true community sensitivity ranks,
which we will see later, is the case. In the absence of the larger data set, we would not know this.
12 Recall that zinc was the most severe outlier in Figure 1.
18
-------
1.00
0.90
0.80
0.70
S. 0.60
TO
° 0.50
o
O 0.40
Q.
0.30
0.20
0.10
0.00
• •
• Highest Datum
• Lowest Datum
5 10 15
Number of Species in Database
20
Figure 7. Proportional rank for the most and least sensitive species in a data set as a function of the number of data
points. Proportional rank is calculated as the rank divided by N + 1.
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0.0001
0.001
0.01
100
1000
10000
EC50
Figure 8. Hypothetical species sensitivity distribution showing the highest and lowest FIFRA values (red
dashed lines) and the 95th and 5th percentile values (blue dashed lines). Ratio of 95th to 5th percentile values is
10,000. Ratio of FIFRA values is 210.
19
-------
1.00 E+08
1.00 E+07 i
1.00 E+06
1.00 E+05 i
.n
in
5 1.00 E+04 :
IO
o>
1.00 E+03 ;
1.00 E+02
1.00 E+01 i
1.00 E+00
1 10 100 1,000 10,000 100,000
FIFRA high/low
Figure 9. Ratio of the 95th to 5th values plotted against the ratio of the FIFRA high and low values. The vertical
red line is the actual FIFRA ratio for Zinc, and the horizontal red line is the hypothetical full data set ratio that would
correspond to that FIFRA ratio.
As mentioned earlier, the shape of an SSD curve is largely defined by the standard deviation (or
spread) of the data. Another way to envision this is to use the ratio of the highest to the lowest
values as a surrogate for the slope. Figure 10 shows these ratios plotted against the number of data
points within a data set. For comparison, the graph includes the ratio for each of the FIFRA-5 data
subsets. The more species tested, the greater the likelihood of high ratios. Even though large data
sets are not numerous, the plot suggests that there is an upper limit to the ratio of the effects values
for the least and most sensitive species13. We have chosen 5,000 and 25,000 as the potential range
for large data sets. Figure 11 is the same plot shown in Figure 9, except this time the grey bar
shows the expected range for large data. The bar crosses the trend line at the locations where the
high to low ratio of the FIFRA-5 data set should be if their proportional ranks were accurate.
Therefore, if a data set with five data points has a high to low ratio between 150 and 386, then
there is some degree of confidence that this small data set may reasonably represent the range of
sensitivity within the aquatic plant community. If the ratio either is less or greater than this range,
then additional data likely has benefit. Table 4 shows the expected small data set high to low ratios
for a range of N values. Since the proportional ranks change with increasing N (see Figure 7), so
do the expected ratios of high to low values. If the ratio from the small data set is lower than the
predictions listed in Table 4, then the probability the calculated HC5 is an over estimate increases
as this ratio decreases, and vice versa if that ratio is greater than predicted—probability of under
estimation increases.
13 Metsulfuron methyl is a likely outlier (green triangle in Figure 10). Many of the most sensitive species values
were from the same paper (Cedergreen, et al. 2004) and the data plots from which the EC50 values were
calculated show a lot of scatter. Although their calculated values are technically correct, there is reason to
believe that the lowest value could be a factor of 3 or 4 greater than the value used. This would bring the ratio
for that chemical in Figure 11 down to somewhere between 25,000 and 33,000.
20
-------
o 10,000
1,000 ¦ ¦
100 ¦ ¦
10
. *
••
•: •
• •
» %
M*
• • •
10 20 30 40 50 60
Number of Species in Data set
70
Figure 10. Plot of the ratio of the highest EC50 to the lowest from a given data set against the number of species within
that data set. The red data points are the ratios from the FIFRA-5 data subsets. The green triangle is a likely outlier for
metsulfuron methyl (outlier for reasons explained within footnote 12). Hie grey horizontal box represents a likely range for ratios
within full data sets with large numbers of species—5,000 to 25,000.
.00 E+08
.00 E+07
.00 E+06
.00 E+05
.00 E+04
.00 E+03
.00 E+02
.00 E+01
.00 E+00
1
10
100
1,000
10,000 100,000
F IF RA high/low
Figure 11. Same plot as Figure 9, except the expected range of full data set high to low ratios (horizontal box) is indicated
along with the corresponding FIFRA high to low ratio values (red vertical lines) from Table 4.
21
-------
Table 4. Expected high low ratios for small data sets ranging in size from 2 to 19 data points—if those data
points accurately represented their true proportional rank. The columns of ratios represent two different situations:
first, if the full data set had a high low ratio of 5000, and second, if the high low ratio was 25,000.
Expected High/Low Ratio Within Small Data Set
N
95th/5th = 5000
95th/5th = 25,000
2
9
14
3
33
64
4
78
178
5
150
386
6
252
715
7
386
1191
8
556
1836
9
762
2670
10
1006
3715
11
1289
4986
12
1611
6501
13
1973
8273
14
2375
10317
15
2818
12643
16
3302
15265
17
3827
18192
18
4394
21434
19
5000
25000
The relationship between high/low EC50 ratios and the probability of over or under estimating the
HC5 is presented in Figure 12. This figure displays simulation results from a known distribution
within which the ratio of the 95th to 5th ranked value is 10,000. Plots of simulations for known
distributions with ratios of 5000 or 25,000 would be similar, but shifted slightly to the right or left,
respectively. The distribution used to generate Figure 12 is the same distribution shown in Figure
5. Data selected for each of the five ranks was restricted based on the expected ranges presented
in Figure 6. The verticle blue dashed line in Figure 12 shows the high low ratio if the most and
least sensitive species selected lined up perfectly on the distribution curve (in other words, if the
EC50 values associated with 0.167 and 0.833 proportaional ranks were selected). The vertical
scatter in Figure 12 illustrates how the high to low ratio is not a perfect predictor of slope. This is
becase the middle three EC50 values not only contribute to the slope calculation, but also to the
intercept (the mean of the EC50 values). The smaller a high/low ratio is the greater the probability
calculating an HC5 value that is significantly higher than the true value. The greater the ratio, the
greater the probability of an HC5 being significantly less than the true value.
22
-------
Slope for 95th/5th = 10,000
100
10
g
T3
E
LLJ
0.1
0.01
•
*
A?.
.
1
= 5 Over Esti
nates HC5
•
••
ovl S
*Ar25ii
'• |
jjtyi
• •••«
* •«
•
• ir. •
•
•
•
-•jHL
•t «*•••*"
T
i * •
¦ ¦ •
*• » a *
•l .
•" • •
i .# • ••
N = 5 Unc
er Estimates H
i • •
i •••
C5
«
•
• ¦
•
10 100 1,000 10,000
Ratio of High to Low EC50
100,000
Figure 12. Plot of data from a simulation selecting five data points at a time using a fixed species sensitivity
distribution. Data selection was restricted to be between the 5th and 95th percentiles. Hie distribution contained 150
species with "known" EC50 values (calculated using Equation 3), and is the same distribution shown in Figure 5. The
two dashed red lines show factors of 4 above and below 1. The vertical dashed blue line is the high to low ratio if all
five data points lined up exactly on the distribution. See the text for explanation of the simulation process.
3.3 Examples
Now that we have this expected ratio information, how can we use it to assist in the evaluation of
small datasets? The information allows us to evaluate the likelihood additional data will contribute
significantly to the estimation of the HC5. We examine several examples below. The first is zinc.
Zinc is of interest because the relationship between the values for the HC5 from the FIFRA data
subset and the full data set is an obvious outlier relative to that relationship for the other 19
toxicants (Figure 1). Of course, if we were beginning with zinc as a new toxicant, we would not
have the luxury of the full data set. Figures 13 A and 13B show the relationship between the
position of the five data points selected to represent the FIFRA-5 within the full data set. Figure
13A is a standard SSD plot, and Figure 13B is the linearized transformation version. The red
arrows trace the FIFRA-5 data points from their position within the full data set. The HC5 EC50
calculated from the full data set is 18.6 ug/L. The value based on the FIFRA-5 is 0.24 ug/L. We
can either accept the zinc FIFRA-5 value (after all, the r2 value is 0.942—Table 3), or we can use
Table 4 to help decide whether or not additional data would be a benefit.
The high to low ratio for the zinc FIFRA-5 data set is 2346. Figure 12 shows how this ratio
indicates a high probability of under estimating the true HC5 value. This ratio cannot decrease no
matter how much additional data is collected. It can only stay the same or increase—either by
values less than the current lowest value, greater than the highest value, or both. Anything between
the current high and low leaves the ratio intact. Lacking any other knowledge about the SSD for
23
-------
zinc, we might assume zinc follows the high to low relationship presented in Figure 10. If this
assumption is true, then Table 4 suggests the minimum number of data points required by zinc
would be nine. This is the first row in Table 4 where the ratio of 2346 is bounded by the minimum
and maxium expected. Therefore, it is likely the additional data would improve the estimate.
10000
100000
Figure 13. Plot showing the relationship between the FIKKA-5 data (solid points) and their ranks within the original full
dataset (open points) for zinc. A) Hie original data set showing the displacement of the FIFRA-5 data. B) The same data,
except this time the plot is log-normal (the solid line is the regression using all of the data, dashed line is the regression for the
FIFRA data). C) The FIFRA data including the value for Myriophylliim spicatum (red datum). D) The same data as C except
there was an assumption of an N of 12—see text for explanation. Plots A, B and C are based on Equation 3. The vertical red
dashed line is the location of the 5th percentile.
Now the question becomes, what additional data? In general, any data will be useful. There is,
however, a recommendation to include the rooted freshwater macrophyte Myriophyllum spp. in
standard testing procedures (Maltby et al. 2010), and there are now guidelines for conducting
tested with Myriophyllum species (OECD 2014). It would seem logical, therefore to add this
species as the best choice for a sixth value. Figure 13C displays the results when we include the
24
-------
available EC50 for Myriophyllum spicatum (the only Myriophyllum species in the full data set).
The HC5 for zinc only improves slightly (0.39 ug/L) relative to the HC5 of the full data set. An
additional issue arises; however, since the Myriophyllum EC50 is now the highest value, the new
high to low ratio is 5109. Going back to Table 4 we see that the number of data points now needed
is 12. We could continue to seek additional data, or we can assume that because the ratio is greater
than 5000, the probability is reasonably high that any new data will be between the current most
and least sensitive species. If we assume an N of 12 and re-plot14 the data in Figure 13C (see Figure
13D), the HC5 estimate improves to 2.1 ug/L—now only a factor of 9 less than the HC5 for the
full data set, rather than the original factor of 78. Clearly, in the case of zinc there would be added
benefit from collecting more data. However, we would likely need to more than double the number
of data points within the small data set, or to assume a higher N based on Table 4.
Two other compounds, diquat and lead, have FIFRA-5 ratios that exceed the expected range for
an N = 5. Their ratios (>735 and 563, respectively), are not as high as that for zinc. In fact,
information contained in Table 4 would suggest that we need only one more datum to bring their
ratios within the expected range—as long as those data points were between their current high and
low values. Also, Figure 12 indicates that the probability of under estimating the HC5 within the
range of these ratios is not too great. Both of these lines of evidence suggest that the calculated
HC5 values for diquat and lead may be close enough. If we had additional data, then we would
need to re-evaluate. For example, if in addition to the FIFRA-5 EC50 values for lead, the EC50
tor Myriophyllum spicatum (the least sensitive species in the full data set) were available the HC5
value decreases from 6.53 to 3.48 ug/L (HC5 value from full data set is 20.0 ug/L). More
importantly, the high to low ratio is now over 10,000. The information in Table 4 shows that this
is exceptionally high for a data set with only six values. In fact, it would correspond to a full data
set 95th to 5th value ratio of over 1.6 million (Figure 10 suggests this is unlikely). In this case, more
data would likely improve the estimated HC5.
There was only one toxicant, arsenic V, for which the FIFRA high to low ratio (204) was within
the expected range listed in Table 4. If we have the EC50 tor Myriophyllum as a sixth value, the
small data set calculation of the HC5 goes from 1.17 to 2.19 ug/L—still not very different from
the full data HC5 of 1.39 ug/L. The high to low ratio stays the same since the sixth EC50 value is
between the previous high and low values. This ratio, however, now is slightly lower that the
expect limits listed in Table 4. However, based on the information in Figure 12, this ratio is
probably still close enough to expected so that the probability of the HC5 being very different from
what might be expected with a larger data set is low.
The FIFRA-5 high to low EC50 ratio for sixteen of the twenty toxicants was between 2 and 53
(Table 3, significantly lower than the expected lower limit for data sets with only five EC50 values
(Table 4). Figure 12 suggests that the HC5 values calculated for these toxicants have a high
probability of over estimating the true HC5. Yet, Figure 1 shows that there are no extreme over
estimation outliers analogous to zinc's under estimation. One plausible explanation is that Figure
12 assumes that the data for each FIFRA-5 subset comes from representatives that cover most of
the entire range of possible toxicity values. The actual data for the FIFRA-5 species from the
toxicants with lower high to low ratios are largely restricted to the more sensitive portions of the
full data set—the upper forth of the distribution is not commonly present in this sub set. Figure 14
includes two additional simulations plotted along with the original data from Figure 12. The data
14 The proportional rank for the lowest value was set at 1/13, the rank for the highest value set a 12/13, and the
ranks for the other four values were evenly distributed between these.
25
-------
plotted with green markers restricted the data selection to no greater than the 0.75 proportional
rank. The data plotted in red restricted the simulation to EC50s less than the 0.50 rank. As the
upper bound of the sensitivity is restricted, the range of values on the y-axis shrinks. The
probability of over estimating declines substantially relative to the full simulation plot (blue),
especially for ratios greater than 10.
Slope for 95th/5th = 10,000
100
N = 5 Over EstimatesHCo
N = 5 UnderEstimatesHCo
0.01
10 100 1,000 10,000
Ratio of High to Low EC50
100,000
Figure 14. Plots of data from simulations selecting five data points at a time using a fixed species sensitivity
distribution. The blue markers represent data shown in Figure 12. Hie green markers are the data from the simulation
restricting the upper limit of the five data points to the 75th percentile. The red markers are from the simulation restricting
the upper limit to the 50th percentile.
For those six situations where the ratio is less than ten, the probability of over estimation increases;
however, based on Figure 1, this is not a big probability. Four of these data sets (atrazine,
chromium, diuron and pentachlorophenol) exhibit a biphasic distribution in the sensitivity
distribution (see Appendix B). It is not clear why this occurred. It could be a solubility issue;
however, examining only measured values does not necessarily remove the biphasic nature.
Perhaps different modes of action are at work at lower versus higher concentrations. On the other
hand, in some cases, it could be a matter of too few EC50 values at the high end relative to the rest
of the data set. This could distort the shape of the curve due entirely to inaccuracy in the ranks of
those values. Whatever the cause, we restricted the HC5 calculations for these full data sets to the
lower half of the sensitivity curve. Recognition that small data sets associated with very small high
to low EC50 ratios generally have their data concentrated in the lower half to three quarters of the
distribution suggests that the likelihood of substantially over estimating the true HC5 is small.
Additional data, however, can help to confirm whether a biphasic condition exists—minimizing
the likelihood of under estimating the HC5.
26
-------
3.4 Addition of Myriophyllum Data
As mentioned earlier, species of Myriophyllum are a logical first choice for an additional sixth
value. The intent for adding Myriophyllum is to cover situations where a compound would be more
toxic to dicots than monocots (Maltby et al. 2010). Myriophyllum is a dicot, and Lemna (the
standard FIFRA aquatic vascular plant) is a monocot. Myriophyllum species also are rooted in
sediment, and Lemna is floating. Figure 15 shows the relationship between and HC5 value
estimated using the FIFRA-5 data with that estimated using these data plus a value for
Myriophyllum. None of the compounds in our data set specifically target dicots. It is not surprising,
therefore, that the addition of & Myriophyllum datum has minimal effect on the estimated HC5
value.
Myriophyllum Effect
100
HC5 Increases
Atrazinfe
Copper'
Terbuthyl^zine/
my m/L
Zinc S _/
/ w ^Diuron
/ Metfibuzin
z
Cadmium
*
' Lead
E
3
Q.
O
Diquat
5
io
o
X
arol
HC5 Decreases
0.01
0.01
0.1
1
10
100
HC5 without Myriophyllum (N =5)
Figure 15. Comparison of HC5 with (N= 6) and without (N = 5) EC50 values included for Myriophyllum. Red dashed lines
represent a factor of two from the one-to-one line (solid black line).
3.5 Use of Growth Rate as a Standard Endpoint
A common dilemma that arises when constructing SSDs for aquatic plants is which endpoints to
include in the data set. There are those who favor only using growth or relative growth rate
(commonly used endpoints) so there is a common link among all the test values. Even with these
endpoints, however, there are inconsistencies. Growth rate of microalgal species essentially always
refers to population growth rate, while growth of aquatic vascular plants (with the possible
exception of Lemna spp.) usually refers to growth of individual plants. There are situations, such
as the atrazine work by Erickson (2010), when there is a reason for standardization based on
growth rate as the endpoint. In many situations, however, the selection of only growth rate may
not be justified and may even be counter to the need of best representing the true sensitivity of
27
-------
species. Ultimately, for an aquatic life criterion, the question being asked is, for a given exposure
scenario, what proportion of the species of aquatic plants are likely to be adversely affected?
Confining the definition of adverse toxicity effect to only effects on growth would ignore the fact
that growth often is not the most sensitive endpoint. In addition, available growth results may not
be the most ecologically relevant growth endpoint. Growth results of aquatic vascular plants most
often are available only for shoots. In many instances, when data are available for both shoots and
roots, growth of roots is substantially more sensitive than that of shoots (Lin et al. 2002, Arts et al.
2008). For aquatic plants that are rooted in the sediment, this can make a substantial difference in
their survivability. Including a broader base of endpoints (beyond just growth rate) will provide a
more complete representation of the range of species sensitivities within the aquatic plant
community.
3.6 Representativeness of All Available Data
Understanding the relationship of conclusions drawn from phytotoxicity data sets of various sizes
is the primary objective of the data analyses presented in this paper. The accuracy of this
comparison, however, has some uncertainty associated with how well the results within "full" data
sets represent the distribution of aquatic plant sensitivities in the "real world". While this source
of uncertainty is also true for aquatic animal data sets, at least with the animal data there exists
currently accepted MDRs assumed to be a reasonable surrogate for representing that community
at large. The representativeness of the usual surrogates for aquatic plants, which are algal
dominated, is difficult to access since an analogous MDR for plants does not exist. While there is
no evidence to suggest that the underrepresented plant taxonomic groups have some special
sensitivity to pesticides, neither is there any evidence that they do not. As stated above, the true
goal of the compilation of aquatic plant toxicity results is to determine a relevant distribution of
sensitivity. The validity of this distribution, however, cannot be fully assessed without some
consideration (and testing) of taxonomic groups and species that are rarely, if ever, tested. This is
especially true for marine macrophytes, both vascular (seagrass, mangroves) and non-vascular
(i.e., seaweeds), and for freshwater macroalgae. In addition, the relative sensitivities of freshwater
and saltwater plants are unknown for most chemicals. Although, for the few compounds within
the analysis in this paper for which large numbers of both freshwater and saltwater species are
present, there does not appear to be any consistent difference between the sensitivities of
freshwater and saltwater species15. In addition, even though there is often significant
underrepresentation of certain aquatic plant groups, when the total number of species within a
given database is large, most groups that might be of interest have at least one species in the mix.
These additional groups do not show up as being particularly sensitive relative to the other,
typically tested species. This conclusion should occasionally be re-examined as phytotoxicity data
availability increases, especially relative to endpoints that represent reproduction.
15 Arsenic V is a possible exception (see Appendix A). Saltwater species are all more sensitive than freshwater ones.
This could be because of the "salt"; however, test nutrient conditions can influence results significantly. For
example, if freshwater tests have higher levels of phosphate in the medium than saltwater tests, the effect of
arsenate could be greatly reduced.
28
-------
3.7 Summary of the Report's Approach
It is important to recognize that the specifics of the approach presented in this report are based on
data sets which include only EC50 values and combine a variety of endpoints. In addition, the data
sets combine values for both freshwater and saltwater species, as well as include all taxa (algal and
vascular) in a single species sensitivity distribution for each toxicant of interest. We also made our
comparisons based on the widely used 5th percentile value (HC5). While the approach presented
herein is sound, the details illustrated in Figure 10 and Table 4 will need to be adjusted for data
sets that deviate from the above. For example, if EC20 values are chosen or algae and vascular
plants are segregated into separate distributions, the information presented in Figure 10 should be
re-evaluated. This will likely result in a different set of expected ranges for high to low ratios of
large datasets. This, in turn, will result in a different set of values for Table 4.
Our approach does not require a fixed number of required data values, nor a fixed set of specific
species with which to conduct toxicity tests. We focused on the FIFRA-5 dataset out of
convenience, and because there already exists a precedence of using these species within the
Agency—therefore, there already exist some degree of acceptance. We provided some insight into
the expected value of adding data from a sixth species—Myriophyllum spp. Myriophyllum was
selected because it represents a critical group needed if a given toxin is selective towards dicots,
as is the case with auxin-like compounds. Also, as with the FIFRA-5 species, an accepted protocol
for the test procedure is already available.
The approach outlined below does not eliminate the need for judgement. Judgement can never be
totally eliminated from the process; however, the approach does provide a basis for guiding that
judgement.
The approach has the following simple steps.
1) Gather all of the available data and decide which values will be used for the analysis. The
section above entitled Data Standardization describes the way we culled the available data—
it is not the only way.
2) Calculate the ratio of the highest value to the lowest value. Look up in Table 4 the
expected range for this value based on the total number of data points (N). If your value
falls between the two values from the table, then additional data may not be needed. If your
value falls outside these two values, then additional data may make a significant difference
in the final calculated HC5.
3) Plot a linearized form of the data (e.g., Equation 3). If the high to low ratio is acceptable
according to Table 4, and the data are fairly evenly distributed between the low and high
values, then the need for more data is less likely.
Step 4: If additional data are collected, repeat Steps 2 and 3.
Step 5: Calculate HC5 using Equation 3.
29
-------
4. CONCLUSIONS
Minimum data requirements for the derivation of aquatic plant water quality criteria focusing on
a search for a definitive group of species, genera or some other fixed set of required taxonomic
representatives is a difficult task due to the lack of a taxa-diverse toxicity database. Even a cursory
examination of the ECTOX database shows that the vast majority of data are for freshwater
species, primarily algae (mostly microalgae). In addition, it is hard to imagine that species of
Lemna, the most commonly tested aquatic vascular plant, is universally representative of all
vascular aquatic plants—a very large and morphologically and physiologically diverse group.
Lemna is a floating plant, with roots suspended in the water column. It is unlikely that this route
of exposure (i.e., via the suspended roots and undersurface portion of leaves) is representative of
that for rooted submerged and emergent aquatic plant species. We have shown when the focus
shifts to the representativeness of the distribution of sensitivities rather than the representativeness
of the taxonomic groups, Lemna contributes significantly to the conclusion that a phytotoxicity
data set consisting as few as five EC50s provides a reasonable evaluation of the expected
toxicological response of the aquatic plant community. Scientific judgement using a taxa-limited
database, however, formed the basis of this conclusion (although with a diverse set of chemicals).
As such, the conclusion may change in the future as the phytotoxicity database increases. We
provide an approach to evaluating the representativeness of small data sets using the ratio of the
highest to lowest toxicity values. This evaluation includes examining the likelihood of over
estimating, as well as under estimating, the HC5. With the toxicants selected, four chemicals had
similar HC5 values from both the FIFRA and full data sets. Over estimation of the HC5 using the
FIFRA data set occurred for six chemicals and was always less than a factor of four. Under
estimation occurred more frequently using the small data set (ten chemicals) and the differences
could be much greater (e.g., zinc). Under estimation would error on the side of environmental
protection.
30
-------
5. REFERENCES
APVMA. 2011. Diuron Environmental Assessment. Australian Pesticides and Veterinary
Medicines Authority. July 2011. URL.
www.apvma.gov.au/products/review/docs/diuron_environment.doc
Arts, GHP, JDM Delgers, CH Hoekzema and JTNM Thissen. 2008. Sensitivity of submersed
freshwater macrophytes and endpoints in laboratory toxicity tests. Environmental
Pollution 153:199-206.
Bao, VWW, KMY Leung, J-W Qiu and MHW Lam. 2011. Acute toxicities of five commonly
used antifouling booster biocides to selected subtropical and cosmopolitan marine
species. Marine Pollution Bulletin 62:1147-1151.
Brain, RA, H Sanderson, PK Sibley and K Solomon. 2006. Probabilistic ecological hazard
assessment: Evaluating pharmaceutical effects on aquatic higher plants as an example.
Ecotoxicology and Environmental Safety 54:128-135.
Cavalier-Smith, T. 2004. Only six kingdoms of life. Proceedings of the Royal Society of London
B. 271:1251-1262.
Cedergreen, N, JC Streibig, and NH Spliid. 2004. Sensitivity of aquatic plants to the herbicide
metsulfuron-m ethyl. Ecotoxicology and Environmental Safety 57:153-161.
Chalifour, A and P Juneau. 2011. Temperature-dependent sensitivity of growth and photosynthesis
of Scenedesmus obliquits, Navicular pelliculosa and two strains of Microsystis aeruginosa to
the herbicide atrazine. Aquatic Toxicology 103:9-17.
Chapman, PF, M Reed, A Hart, W Roelofs, T Aldenbert, K Solomon, J Tarazona, M Liess, P
Byrne, W Powley, J Green, S Ferson and H Galicia. 2007. Methods of Uncertainty
Analyses: Work Package 4 in EUFRAM: Concerted action to develop a European
Framework for probabilistic risk assessment of the environmental impacts of pesticides.
Versin 3, March 2007. URL. www.eufram.com
Cronk, JK and MS Fennessy. 2001. Wetland Plants: Biology and Ecology. CRC Press. 463 pp.
Drake, BG. 1989. Photosynthesis of salt marsh species. Aquatic Botany 34:167-180.
Ebenezer, V and J-S Ki. 2012. Evaluation of the sub-lethal toxicity of Cu, Pb, bisphenol A and
polychlorinated biphenyl to the marine dinoflagellate Cochlordinium polykrikoides.
Algae 27(l):63-70.
Eklund, B. 1998. Reproductive performance and growth response of the red alga Ceramium
strictum under the impact of phenol. Marine Ecology Progress Series 167:119-126.
Erickson, R. 2010. Proposed methodology for specifying atrazine levels of concern for
protection of plant communities in freshwater ecosystems. Draft Report to Environmental
Fate and Effects Division, Office of Pesticide Programs, USEPA. September 15, 2010.
31
-------
Guo, R. V Ebenezer and J-S Ki. 2012. Transcriptional responses of heat shock protein 70
(Hsp70) to thermal, bisphenol A, and copper stresses in the dinoflagellate Prorocentrum
minimum. Chemosphere 89:512-520.
Hanson, ML and KR Solomon. 2002. New technique for estimating thresholds of toxicity in
ecological risk assessment. Environmental Science and Technology 36:3257-3264.
Hanson, ML and KR Solomon. 2004. Haloacetic acids in the aquatic environment. Part I:
macrophyte toxicity. Environmental Pollution 130:371-383.
Kenaga, EE and RJ Moolenaar. 1979. Fish and Daphnia toxicity as surrogates for aquatic
vascular plants and algae. Environmental Science and Technology 13(12): 1479-1480.
Larras, F, A Bouchex, F Rimet and B Montuelle. 2012. Using bioassays and species sensitivity
distributions to assess herbicide toxicity towards benthic diatoms. PLOS ONE Vol 7
Issue 8. E44458. www.plosone.org.
Lee, ET and JW Wang. 2003. Statistical Methods for Survival Data Analysis. 3rd Ed. Wiley &
Sons. 513 pp.
Lewis, MA. 1990. Are laboratory-derived toxicity data for freshwater algae worth the effort?
Environmental Toxicology and Chemistry 9: 1279-1284
Lewis, MA. 1995. Use of freshwater plants for phytotoxicity testing: a review. Environmental
Pollution 87:319-336.
Lin, Q, IA Mendelssohn, MT Suidan, K Lee and AD Venosa. 2002. The dose-response
relationship between No. 2 fuel oil and the growth of the salt marsh grass, Spartina
alterniflora. Marine Pollution Bulletin 44:897-902.
Longstreth, DJ. 1989. Photosynthesis and photorespiration in freshwater emergent and floating
plants. Aquatic Botany 34:287-299.
Maltby, L, D Arnold, G Arts, J Davies, F Heimbach, C Pickl and V Poulsen. 2010. Aquatic
Macrophyte Risk Assessment for Pesticides. CRC Press. 135 pp.
Newman, MC, DR Ownby, LCA Mezin,DC Powell, TRL Christensen, SB Lerberg and B-A
Anderson. 2000. Applying species-sensitivity distributions in ecological risk assessment:
Assumptions of distribution type and sufficient numbers of species. Environmental
Toxicology and Chemistry 19(2): 508-515.
OECD 2014. Test No. 239. Water-sedimentMyriophyllum spicatum toxicity test, OECD
Guidelines for Testing of Chemicals Section 2: Effects on Biotic Systems. OECD
Publishing, Paris, http://dx.doi.oru/10.178719789264224155-en
Suter II, GW, TP Traas and L Poshuma. 2002. Issues and practices in the derivation and use of
species sensitivity distributions, pp 437-474 (in) L Posthuma, GW Suter II and TP Traas
(eds) Species Sensitivity Distributions in Ecotoxicology. Lewis Publishers 587 pp.
TenBrook, PL and RS Tjeerdema. 2006. Methodology for derivation of pesticide water quality
criteria for the protecton of aquatic life in Sacramento and San Joaquin River basins.
32
-------
Phase I: review of existing methodologies. Report to the Central Valley Regional Water
Quality Board. April 2006.
Thursby, GB, RL Steele and ME Kane. 1985. Effects of organic chemicals on growth and
reproduction in the marine red alga Champiaparvula. Environmental Toxicology and
Chemistry 4:797-805.
USEPA. 1985. Guidelines for Deriving Numerical National Water Quality Criteria for the
Protection of Aquatic Organisms and Their Uses. EPA Report 822/R-85-100 (NTIS
Report PB85-227049.
USEPA. 2011. Appendix F. Estimating Aquatic Plant Community Hazard Concentrations for
Pesticide Effects. http://www.regulations.gOv/#!documentDetail;D=EPA-HQ-OPP-2011-
0898-0012.
Wang, W. 1984. Uses of aquatic plants in ecotoxicology. Environmental International 10:i-iii.
Wang, W and K Freemark. 1995. The use of plants for environmental monitoring and assessment.
Ecotoxicology and Environmental Safety 30:289-301.
33
-------
Appendix A. List of Data Used for Each Toxicant, Organized
Alphabetically.
These data sets do not necessarily represent all possible available data. January 2015 was the last
ECOTOX database search update. A quick search of the internet added a few addition data values,
and notations for these are in the following tables. Shaded cells show the data included in the
FIFRA-5 subsets. Column labeled Lowest Geometric Mean is the EC50 for the most sensitive
endpoint for that species. When more than one value was available for a given endpoint, the
geometric mean was calculated.
Table A.l. Data for arsenic V.
Lowest
Geometric
Medium
Species Scientific Name
Species Category
Mean (ug/L)
SW
Isochrysis galbana
Haptophyte
2
SW
Amphidinium carterae
Dinoflagellate
10
SW
Skeletonema costatum
Diatom
11.7
SW
Chaetoceros ingolfianum
Diatom
20
SW
Cylindrotheca closterium
Diatom
> 100
SW
Tetraselmis contracta
Green Algae
> 100
SW
Thalassiosira pseudonana
Diatom
> 100
FW
Scenedesmus acutus var.
acutus
Green Algae
159.3
FW
Pseudokirchneriella subcapitata
Green Algae
228
FW
Lemna minor
Duckweed
630
FW
Anabaena flosaquae
Blue-Green Algae
>2400
34
-------
Table A.2. Data for atrazine.
Medium
Species Scientific Name
Species Category
Lowest
Geometric
Mean (ug/L)
FW
Pseudanabaena galeata
Blue-Green Algae
12.96
FW
Chlorella vulgaris
Green Algae
14.71
SW
Amphidinium operculatum
Dinoflagellate
17.19
SW
Tetraselmis chuii
Prasinophyte
20
FW
Elodea canadensis
Waterweed
21
FW
Scenedesmus subspicatus
Green Algae
21
FW
Ceratophyllum demersum
Coon-Tail
22
SW
Storeatula major
Algae
22.17
SW
Synechococcus leopoliensis
Blue-Green Algae
24.1
FW
Oscillatoria limnetica
Blue-Green Algae
24.2
SW
Isochrysis galbana
Haptophyte
30
FW
Potamogeton perfoliatus
Pondweed
30
FW
Lemna gibba
Inflated Duckweed
31.46
SW
Ankistrodesmus sp.
Green Algae
32.36
FW
Najas sp.
Water Nymph
32.55
FW
Scenedesmus acutus var acutus**
Green Algae
33.02
SW
Skeletonema costatum
Diatom
35.67
FW
Microcystis aeruginosa**
Blue-Green Algae
39.97
FW
Desmodesmus subspicatus
Green Algae
41
FW
Scenedesmus quadricauda
Green Algae
41.69
SW
Chaetoceros sp.
Diatom
43
FW
Synechococcus elongatus
Blue-Green Algae
49.68
FW
Navicula pelliculosa
Diatom
60
FW
Lemna minor
Duckweed
60
SW
Zostera marina
Eelgrass
60
SW
Phaeodactylum tricornutum
Diatom
60.61
FW
Chlorella fusca ssp. vacuolata
Green Algae
66
SW
Dunaliella tertiolecta
Green Algae
66.81
FW
Lemna aequinoctiales
Duckweed
69.12
FW
Pse udokirchneri el la s ubcapi ta ta
Green Algae
73.19
FW
Myriophyllum spicatum
Eurasian Watermilfoil
91
FW
Chlorella pyrenoidosa
Green Algae
93.27
FW
Fragilaria rum pens*
Diatom
106
FW
Chlamydomonas reinhardtii
Green Algae
109.37
FW
Scenedesmus abundans
Green Algae
110
SW
Pavlova sp.
Chrysophyte
130.60
FW
Myriophyllum heterophyllum
Two-Leaf Water-Milfoil
132
FW
Craticula accomoda
Pennate Diatom
164
FW
Nitzschia accomodata
Diatom
164
35
-------
Table A.2. Continued.
sw
Rhodomonas salina
Marine Microalga
165
sw
Nannochloropsis gaditana
Microalgae
209
FW
Anabaena flosaqucie
Blue-Green Algae
224.94
FW
Ulnaria ulna*
Diatom
232
FW
Chlamydomonas geitleri
Green Algae
243.54
FW
Hydrilla verticillata
Hydrilla
258.64
FW
Asterionella formosa
Diatom
261
SW
Porphyridium cruentum
Red Algae
308
FW
Vallisneria americana
Water-Celery, Tapegrass
329.89
FW
Cryptomonas pyrenoidifera
Cryptomonad
500
FW
Fragilaria capuchina varvaucheriae*
Diatom
635
FW
Achnanthidium minutissimum*
Diatom
748
FW
Cyclotella meneghiniana*
Diatom
812
FW
Gomphonema parvulum*
Diatom
907
FW
Craticula accomoda*
Diatom
919
FW
Mayamaea fossalis*
Diatom
929
FW
Anabaena variabilis
Blue-Green Algae
1689
FW
Myriophyllum sibiricum
Water Milfoil
2092
FW
Sellaphora minima*
Diatom
2510
FW
Stichococcus bacillaris
Green Algae
3368
FW
Nitzschia palea*
Diatom
3988
FW
Oscillatoria laetevirens
Blue-Green Algae
4968
FW
Anacystis alpicola
Blue-Green Algae
5360
FW
Encyonema silesiacum*
Diatom
5995
FW
Chlorella saccharophila
Green Algae
8457
SW
Ruppia maritima
Widgeon-Grass
10571
FW
Chlorella kessleri
Green Algae
19852
FW
Aulacoseira granulata
Diatom
56000
FW
Lepidium sativum
Garden Cress
> 66960
FW
Euglena gracilis
Flagellate Euglenoid
78799
*Larras et al. 2012
**Chalifour and Juneau 2011. Note, Scenedesmus acutus var acutus is the accepted name for 5.
obliquis.
36
-------
Table A.3. Data for cadmium.
Medium
Species Scientific Name
Species Category
Lowest
Geometric
Mean (ug/L)
FW
Scenedesmus acutus
Green Algae
10.8
FW
Staurastrum cristatum
Green Algae
11.4
FW
Navicula pelliculosa
Diatom
31
FW
Scenedesmus subspicatus
Green Algae
32
FW
Pseudokirclineriella subcapitata
Green Algae
43.1
FW
Spirodela polyrhiza
Large Duckweed
54.1
SW
Chaetoceros calcitrans
Diatom
60
SW
Ditylum brightwellii
Diatom
60
FW
Chlorella saccharophila
Green Algae
107.5
FW
Gonium pectorale
Green Algae
109
FW
Anabaena flosaquae
Blue-Green Algae
120
SW
Thalassiosira pseudonana
Diatom
160
SW
Skeletonema costatum
Diatom
178.1
SW
Ulva pertusa
Green Algae
188.7
FW
Lemna minor
Duckweed
191.1
SW
Asterionella japonica
Diatom
224.8
SW
Gonyaulax polyedra
Dinoflagellate
300
FW
Navicula closterium
Diatom
476
FW
Chlamydomonas reinhardtii
Green Algae
497.1
FW
Wolffia globosa
Duckweed
500
SW
Gracilaria tenuistipitata
Red Algae
590.1
FW
Prorocentrum minimum
Dinoflagellate
674.5
FW
Chlorella vulgaris
Green Algae
770.3
FW
Lemna gibba
Inflated Duckweed
800
SW
Isochrysis galbana
Haptophyte
965.4
FW
Chlamydomonas acidophila
Green Algae
1562
SW
Tetraselmis gracilis
Green Flagellate
1800
SW
Laminaria saccharina
Kelp, Brown Algae
2150
FW
Entomoneis cf punctulata
Diatom
2400
FW
Navicula incerta
Diatom
3009.0
FW
Prasinococcus sp.
Green Algae
5900
SW
Dunaliella tertiolecta
Green Algae
6000
FW
Myriophyllum spicatum
Eurasian Watermilfoil
7400
SW
Tetraselmis tetrahele
Prasinophyte
8223.1
FW
Chaetoceros gracilis
Diatom
8500
FW
Cylindrotheca sp.
Diatom
9300
SW
Tetraselmis suecica
Prasinophyte
9380
FW
Chlorococcum littorale
Green Algae
9700
SW
Phaeodactylum tricornutum
Diatom
12362.5
FW
Heterocapsa sp.
Dinoflagellate
13800
37
-------
Table A.3. Continued.
FW Spirulina platensis
SW Dunaliella salina
Blue-green algae
Green Algae
18350
48900
Table A.4. Data for chromium VI.
Medium
Species Scientific Name
Species Category
Lowest
Geometric
Mean (ug/L)
SW
Champia parvula
Red Algae
42.43
FW
Chlorella protothecoides
Green Algae
104
FW
Pseudokirchneriella subcapitata
Green Algae
122.2
FW
Euglena gracilis
Flagellate Euglenoid
139.4
FW
Gomphonema parvulum
Diatom
150
FW
Scenedesmus subspicatus
Green Algae
161.2
FW
Chlorella kessleri
Green Algae
203.7
FW
Nitzschia linearis
Diatom
208
FW
Chlorella vulgaris
Green Algae
270.12
FW
Navicula seminulum
Diatom
286.4
FW
Chlamydomonas reinhardtii
Green Algae
287.7
FW
Scenedesmus pannonicus
Green Algae
396.7
FW
Gonium pectorale
Green Algae
431.1
FW
Stichococcus bacillaris
Green Algae
581.2
FW
Anacystis aeruginosa
Blue-Green Algae
760.0
FW
Scenedesmus quadricauda
Green Algae
894.5
FW
Myriophyllum spicatum
Eurasian Watermilfoil
915.8
FW
Lemna minor
Duckweed
990.3
FW
Oscillatoria agardhii
Blue-Green Algae
1040.7
FW
Chlorella pyrenoidosa
Green Algae
1042.4
FW
Ankistrodesm us falcatus
Green Algae
1050
FW
Chilomonas Paramecium
Cryptomonad
1100
FW
Synechococcus leopoliensis
Blue-Green Algae
1171
SW
Skeletonema costatum
Diatom
1200
FW
Dictyosphaerium chlorelloides
Algae
1540
SW
Gracilaria tenuistipitata
Red Algae
1687.3
FW
Wolffia globosa
Duckweed
1700
SW
Nitzschia closterium
Diatom
2444
SW
Macrocystis pyrifera
Giant Kelp
5000
SW
Galdieria sulphuraria
Red Algae
6864
SW
Dunaliella tertiolecta
Green Algae
17004
SW
Dunaliella bioculata
Green Algae
27258
SW
Ecklonia radiata
Brown Algae
41378
FW
Nitellopsis obtusa
Green Algae
270680
38
-------
Table A.5. Data for copper.
Lowest
Geometric
Medium
Species Scientific Name
Species Category
Mean (ug/L)
SW
Bellerochea polymorpha
Diatom
0.6
FW
Microcystis aeruginosa
Blue-green alga
<0.75
SW
Champia parvula
Red Algae
1.4
SW
Isochrysis galbana
Haptophyte
1.85
SW
Micromonas pus ilia
Algae
1.93
FW
Scenedesmus acutus
Green Algae
3.83
SW
Proteomonas sulcata
Cryptophyte
4.2
SW
Scrippsiella faeroense
Dinoflagellate
5
SW
Macrocystis pyrifera
Giant Kelp
5.5
FW
Synechococcus leopoliensis
Blue-Green Algae
5.98
SW
Thalassiosira pseudonana
Diatom
6.56
SW
Heterocapsa niei
Dinoflagellate
8.76
FW
Trachelomonas sp.
Green Flagellate
9.8
SW
Enteromorpha intestinalis
Green Algae
9.9
SW
Phaeodactylum tricornutum
Diatom
10.21
SW
Ceramium strictum
Red Algae
12.2
SW
Asterionella japonica
Diatom
12.7
FW
Pseudokirclineriella subcapitata
Green Algae
14.83
SW
Gephyrocapsa oceanica
Coccolithophore
17
SW
Nitzschia closterium
Diatom
17.14
SW
Lessonia nigrescens
Brown Kelp
17.20
FW
Chlorella protothecoides
Green Algae
17.24
SW
Entomoneis punctulata
Green Algae
17.27
SW
Coccolithus huxleyi
Coccolithophorid
17.32
SW
Chlorella autotrophica
Green Algae
19.22
SW
Gymnodinium splendens
Dinoflagellate
20
SW
Nitzschia cf. paleacea
Green Algae
24
FW
Staurastrum chaetoceras
Green Algae
25.42
SW
Nannochloris atomus
Microalgae
27.62
FW
Anabaena flosaquae
Blue-Green Algae
29
SW
Rhodomonas salina
Marine Microalga
30
FW
Staurastrum cristatum
Green Algae
30.4
FW
Staurastrum manfeldtii
Green Algae
32
FW
Chlamydomonas reinhardtii
Green Algae
52.19
SW
Fucus serratus
Toothed Wrack
53.15
SW
Ulva pertusa
Green Algae
53.57
SW
Fucus vesiculosus
Bladder Wrack
60
FW
Scenedesmus dimorphus
Green Algae
62.3
FW
Nephrocytium lunatum
Algae
63.55
FW
Stichococcus bacillaris
Green Algae
66.39
39
-------
Table A.5. Continued.
sw
Chaetoceros calcitrans
Diatom
70
FW
Lemna minor
Duckweed
82.61
SW
Gracilaria tenuistipitata
Red Algae
84.34
FW
Scenedesmus quadricauda
Green Algae
102
SW
Hormosira banksii
Neptune'S Necklace
108.17
SW
Skeletonema costatum
Diatom
114.2
FW
Chlorella vulgaris
Green Algae
114.8
SW
Gonyaulax polyedra
Dinoflagellate
120
SW
Tetraselmis tetrathele
Green Flagellate
124.9
FW
Navicula pelliculosa
Diatom
125
SW
Nannochloropsis gaditana
Microalgae
137
FW
Scenedesmus subspicatus
Green Algae
211.1
FW
Myriophyllum spicatum
Eurasian Watermilfoil
250
FW
Parachlorella kessleri
Green Algae
284.4
SW
Tetraselmis suecica
Prasinophyte
311.0
FW
Lemna gibba
Inflated Duckweed
314.4
SW
Tetraselmis chuii
Prasinophyte
330
FW
Hydrilla verticillata
Hydrilla
440
SW
Chlorella pyrenoidosa
Green Algae
509.1
SW
Dunaliella tertiolecta
Green Algae
530
FW
Chlorella saccharophila
Green Algae
550
SW
Ulva reticulata
Algae
605.4
FW
Navicula seminulum
Diatom
804.9
FW
Nitzschia linearis
Diatom
805
FW
Lemna aequinoctiales
Duckweed
943.4
FW
Aulacoseira granulata
Diatom
1027
SW
Prorocentrum minimum*
Dinoflagellate
1070
SW
Chaetoceros gracilis
Diatom
1200
FW
Aphanothece clathrata
Cyanobacteria
1815
FW
Spirodela polyrhiza
Large Duckweed
3300
SW
Prasinococcus sp.
Algae
5400
SW
Cylindrotheca sp.
Diatom
7700
FW
Chlamydomonas acidophila
Green Algae
8961
FW
Navicula incerta
Diatom
10429
FW
Euglena gracilis
Flagellate Euglenoid
11925
SW
Cochlodinium polykrikoides
Dinoflagellate
12740
*Guo et al. 2012
40
-------
Table A.6. Data for diquat.
Lowest
Geometric
Medium
Species Scientific Name
Species Category
Mean (ug/L)
FW
Spirodela punctata
Large duckweed
0.75
FW
Lemna minor
Duckweed
4
FW
Oscillatoria clialybea
Blue-green algae
12.2
FW
Navicula sp.
Diatom
19
FW
Ochromonas danica
Diatom
23
FW
Cryptomonas ozolini
Cryptomonad
35
FW
Anabaena flosaquae
Blue-green algae
42
FW
Anacystis aeruginosa
Blue-green algae
65
FW
Pseudokirclineriella subcapitata
Green algae
66
FW
Myriophyllum sibiricum
Water milfoil
84.72
FW
Chlorella vulgaris
Green algae
>2940
FW
Euglena gracilis
Flagellate euglenoid
>2940
SW
Skeietonema costatum
Diatom
>2940
Table A.7. Data for diuron.
Medium
Species Scientific Name
Species Category
Lowest
Geometric
Mean (ug/L)
FW
Chlorella pyrenoidosa
Green algae
1.3
SW
Coccolithus huxleyi
Coccolithophorid
2.26
FW
Scenedesmus quadricauda
Green algae
2.7
FW
Apium nodiflorum
European Marshwort
2.81
SW
Zostera marina
Eelgrass
3.2
SW
Ceramium tenuicorne
Red Algae
3.4
FW
Fragilaria capuchina var vaucheriae*
Diatom
4.03
FW
Scenedesmus acutus acutus
Green Algae
4.09
FW
Chlorella vulgaris
Green algae
4.3
SW
Thalassiorira pseudonana
Diatom
4.3
FW
Myriophyllum spicatum
Eurasian watermilfoil
5
SW
Dunaliella tertiolecta
Green algae
5.81
SW
Skeietonema costatum"
Diatom
5.9
FW
Pseudokirclineriella subcapitata
Green algae
7.09
FW
Scenedesmus subspicatus
Green algae
7.2
FW
Encyonema silesiacum*
Diatom
8.79
FW
Fragilaria rum pens*
Diatom
8.89
FW
Lemna aequinoctiales
Duckweed
9.32
FW
Lemna perpusilla
Duckweed
15
FW
Lemna gibba
Duckweed
15.7
FW
Navicula peiiicuiosa
diatom
16.2
41
-------
Table A.7. Continued.
FW
Nitzschia closterium
Diatom
17
SW
Gracilaria tenuistipitata
Red algae
17.32
SW
Phaeodactylum tricornutum
Diatom
20.98
FW
Cyclotella meneghiniana*
Diatom
23
FW
Anabaena flow-aquae
Blue-green algae
23.2
FW
Oscillatoria chalybea
Blue-green algae
23.31
SW
Entomoneis punctulata
Algae
24
SW
Synechococcus leopoliensis
Blue-green algae
24.9
FW
Chlorococcum hypnosporum
Green algae
25
FW
Lemna minor
Duckweed
25
FW
Navicula forcipata
Diatom
26.46
SW
Chaetoceros gracilis
Diatom
36
FW
Spirodela polyrhiza
Large duckweed
41
FW
Ulnaria ulna*
Diatom
42
FW
Anabaena variabilis
Blue-green algae
48.99
FW
Achnanthidium minutissimum*
Diatom
56
FW
Mayamaea fossalis*
Diatom
139
FW
Oscillatoria laetevirens
Blue-green Algae
489.51
FW
Ulothrix fimbriata
Green algae
540
FW
Chlamydomonas moewusii
Green algae
559.44
FW
Gomphonema parvulum*
Diatom
1423
FW
Craticula accomoda*
Diatom
1426
FW
Nitzschia palea*
Diatom
1539
FW
Sellaphora minima*
Diatom
2606
SW
Hormosira banksii
Neptune's Necklace
6820
*Larras et al. 2012
**Bao et al. 2011
Table A.8. Data for Glyphosate.
Medium
Species Scientific Name
Species Category
Lowest
Geometric
Mean (ug/L)
FW
Chi or el la pyrenoidosa
Green Algae
3530
FW
Lemna aequinoctialis
Lesser Duckweed
3889
FW
Chlorella vulgaris
Green Algae
4696
FW
Scenedesmus quadricauda
Green Algae
7200
FW
Pseudokirclineriella subcapitata
Green Algae
9513
FW
Scenedesmus acutus
Green Algae
10200
SW
Skeletonema costatum
Diatom
12000
FW
Lemna gibba
Inflated Duckweed
12400
FW
Anabaena flosaquae
Blue-Green Algae
15000
FW
Navicula pelliculosa
Diatom
17000
FW
Chlorococcum hypnosporum
Green Algae
68000
42
-------
Table A.9. Data for Irgarol.
Lowest
Geometric
Medium
Species Scientific Name
Species Category
Mean (ug/L)
FW
Apium nodiflorum
European Marshwort
0.013
FW
Ulnaria ulna*
Diatom
0.056
FW
Navicula pelliculosa
Diatom
0.1
SW
Tetraselmis sp.
Green flagellate
0.1
SW
Synechococcus sp.
Blue-green algae
0.16
SW
Coccolithus huxleyi
Coccolithophorid
0.25
SW
Thalassiosira weissflogii
Diatom
0.28
SW
Porphyra yezoensis
Red algae
0.37
SW
Thalassiosira pseudonana
Diatom
0.40
SW
Skeletonema costatum
Diatom
0.41
SW
Chlorococcum sp.
Green algae
0.42
FW
Craticula accomoda
Diatom
0.48
SW
Fibrocapsa japonica
Raphidiophyceae
0.479
FW
Chlamydomonas intermedia
Green algae
0.5
FW
Chlorella vulgaris
Green algae
0.52
FW
Navicula forcipata
Diatom
0.59
FW
Nitzschia sp.
Diatom
0.8
SW
Ruppia maritima
Widgeon-grass
0.84
SW
Ceramium tenuicorne
Red Algae
0.96
SW
Dunaliella tertiolecta
Green algae
0.98
FW
Anabaena flosaquae
Blue-green algae
0.99
SW
Chaetoceros gracilis
Diatom
1.1
FW
Myriophyllum verticillatum
Whorl-leaf watermilfoil
1.1
SW
Zostera marina
Eelgrass
1.1
FW
Cyclotella meneghiniana*
Diatom
1.28
FW
Nitzschia palea*
Diatom
1.53
FW
Lemna gibba
Inflated duckweed
1.60
SW
Gracilaria tenuistipitata
Red algae
2
FW
Myriophyllum spicatum
Eurasian watermilfoil
2
FW
Achnanthidium minutissimum*
Diatom
2.26
FW
Pediastrum duplex
Green algae
2.4
FW
Scenedesmus subspicatus
Green algae
2.4
FW
Closterium ehrenbergii
Green algae
2.5
SW
Enteromorpha intestinalis
Green algae
2.5
FW
Staurastrum sebaldi
Desmid
2.5
SW
Eisenia bicyclis
Brown alga
2.92
FW
Pseudokirclineriella subcapitata
Green algae
3.19
FW
Mayamaea fossalis*
Diatom
3.64
FW
Scenedesmus acutus
Green algae
4.10
FW
Fragilaria rumpens*
Diatom
4.5
43
-------
Table A.9. Continued.
sw
Stuckenia pectinata
Sago Pondweed
6.76
sw
Chroococcus minor
Blue-green Algae
7.71
FW
Lemna minor
Duckweed
8.1
FW
Fragilaria capucina varvaucheriae* Diatom
17.29
FW
Craticuia accomoda*
Diatom
115
FW
Seiiaphora minima*
Diatom
174
FW
Gomphonema parvuium*
Diatom
178
FW
Asterioneiia formosa
Diatom
>253
*Larras et al. 2012
Table A.10. Data for lead.
Lowest
Geometric
Medium
Species Scientific Name
Species Category
Mean (ug/L)
SW
Skeletonema costatum
Diatom
19.5
SW
Dityium brightweilii
Diatom
40
SW
Chaetoceros sp.
Diatom
105
SW
Asterioneiia japonica
Diatom
207.2
SW
Uiva pertusa
Green Algae
654.9
SW
Nannochioropsis gaditana
Microalgae
740
SW
Rhodomonas saiina
Marine Microalga
900
FW
Anabaena flos-aquae
Blue-green Algae
990
FW
Chlorella vulgaris
Green Algae
1050
SW
isochrysis gaibana
Haptophyte
1370
SW
Tetraseimis chuii
Prasinophyte
2640
SW
Graciiaria tenuistipitata
Red Algae
4000
FW
Spirodela polyrliiza
Large Duckweed
7600
FW
Nephrocytium lunatum
Green Algae
10360
SW
Navicula incerta
Diatom
10960
SW
Gonyauiax poiyedra
Dinoflagellate
17000
SW
Cochlodinium polykrikoides*
Dinoflagellate
46700
FW
Chlorella saccharophila
Green Algae
63800
FW
Myriophyllum spicatum
Eurasian Watermilfoil
363000
*Ebenezer and Ki 2012
44
-------
Table A.ll. Data for linuron.
Lowest
Geometric
Medium
Species Scientific Name
Species Category
Mean (ug/L)
FW
Elodea nuttalli
Waterweed, ditchmoss
2.5
FW
Ceratophyllum demersum
Coon-tail
8.7
FW
Navicula pelliculosa
Diatom
13.7
FW
Potamogeton perfoliatus
Inflated duckweed
25
FW
Lemna gibba
Diatom
27.3
SW
Skeletonema costatum
Blue-green algae
35.9
FW
Anabaena flosaquae
Pondweed
38.8
FW
Clilorella vulgaris
Green algae
50
FW
Pseudokirclineriella subcapitata
Green algae
67
FW
Myriophyllum spicatum
Eurasian watermilfoil
80
Table A. 12. Data for metolachlor.
Medium
Species Scientific Name
Species Category
Lowest
Geometric
Mean (ug/L)
FW
Salvinia natans
Floating watermoss (fern)
50
SW
Skeletonema costatum
Diatom
61
FW
Ceratophyllum demersum
Coon-tail
70
FW
Pseudokirclineriella subcapitata
Green algae
86.9
FW
Lemna gibba
Inflated duckweed
120.8
FW
Chlorella vulgaris
Green algae
203
FW
Ulnaria ulna*
Diatom
220
FW
Najas sp.
Water nymph
242
FW
Navicula pelliculosa
Diatom
380
FW
Myriophyllum sibiricum
Water milfoil
579.6
FW
Scenedesmus quadricauda
Green algae
600
FW
Achnanthidium minutissimum*
Diatom
734
FW
Chlamydomonas reinhardtii
Green algae
1138
FW
Anabaena flosaquae
Blue-green algae
1200
FW
Cyclotella meneghiniana
Diatom
1790
FW
Elodea canadensis
Waterweed
2355
FW
Microcystis sp.
Blue-green algae
>3000
FW
Myriophyllum heterophyllum
Two-leaf water-milfoil
>3000
FW
Gomphonema parvulum*
Diatom
4054
FW
Anabaena cylindrica
Blue-green algae
>5000
FW
Encyonema silesiacum*
Diatom
6399
FW
Mayamaea fossalis*
Diatom
8042
FW
Chlorella pyrenoidosa
Green algae
12717.2
FW
Craticula accomoda*
Diatom
30147
FW
Eolimna minima*
Diatom
> 50000
45
-------
Table A.12. Continued
FW Fragilaria rumpens* Diatom > 50000
FW Nitzschia palea* Diatom > 50000
FW Fragilaria capuchina var vaucheriae* Diatom 310151
*Larras et al. 2012
Table A. 13. Data for metribuzin.
Medium
Species Scientific Name
Species Category
Lowest
Geometric
Mean (ug/L)
FW
Pseudokirclineriella subcapitata
Green Algae
8.09
SW
Skeletonema costatum
Diatom
8.8
FW
Navicula pelliculosa
Diatom
11.9
FW
Ceratophyllum demersum
Coon-Tail
14
FW
Lemna perpusilla
Duckweed
16
FW
Anabaena flos aquae
Blue-Green Algae
17
FW
Myriophyllum heterophyllum
Two-Leaf Water-Milfoil
17
FW
Elodea canadensis
Waterweed
21
FW
Najas sp.
Water Nymph
21.0
FW
Egeria densa
American Frog'S-Bit
22
FW
Chlamydomonas reinhardtii
Green Algae
23
FW
Chioreila vulgaris
Green Algae
31
FW
Lemna aequinoctiales
Duckweed
44.94
FW
Myriophyllum spicatum
Eurasian Watermilfoil
64
FW
Microcystis sp.
Blue-Green Algae
100
FW
Scenedesmus quadricauda
Green Algae
152
FW
Lemna gibba
Inflated Duckweed
160
FW
Oscillatoria laetevirens
Blue-Green Algae
2953.2
46
-------
Table A.14. Data for nickel.
Medium
Species Scientific Name
Species Category
Lowest
Geometric
Mean (ug/L)
SW
Chlorella vulgaris
Green Algae
150
FW
Lemna minor
Duckweed
191
FW
Pseudokirchneriella subcapitata
Green Algae
263.6
SW
Ditylum brightwellii
Diatom
300
SW
Macrocystis pyrifera
Giant Kelp
2000
SW
Isochrysis galbana
Haptophyte
2876
FW
Anacystis nidulans
Blue-Green Algae
3435.8
FW
Spirodela polyrhiza
Large Duckweed
4500
SW
Phaeodactylum tricornutum
Diatom
5780.6
SW
Gracilaria tenuistipitata
Red Algae
17000
FW
Spirulina platensis
Blue-Green Algae
23455
FW
Euglena gracilis
Flagellate Euglenoid
23476
FW
Euglena mutabilis
Flagellate Euglenoid
446044
Table A.15. Data for pentachlorophenol.
Lowest
Geometric
Medium
Species Scientific Name
Species Category
Mean (ug/L)
FW
Elodea canadensis
Waterweed
4
FW
Ranunculus peltatus
Pond Water Crowfoot
16
SW
Skeletonema costatum
Diatom
35.26
FW
Anabaena flos aquae
Blue-Green Algae
50
FW
Scenedesmus quadricauda
Green Algae
80
FW
Scenedesmus abundans
Green Algae
90
FW
Elodea nuttalli
Waterweed, Ditchmoss
109
FW
Navicula pelliculosa
Diatom
124
FW
Anabaena inaequalis
Blue-Green Algae
130
FW
Pseudokirchneriella subcapitata
Green Algae
147.8
SW
Dunaliella tertiolecta
Green Algae
170
SW
Thalassiosira pseudonana
Diatom
179
FW
Scenedesmus subspicatus
Green Algae
183
SW
Pavlova sp.
Chrysophyte
200
FW
Myriophyllum spicatum
Eurasian Watermilfoil
236
FW
Lemna gibba
Inflated Duckweed
250
FW
Chlamydomonas reinhardtii
Green Algae
260.8
SW
Macrocystis pyrifera
Giant Kelp
300
FW
Potamogeton crispus
Curled Pondweed
338
FW
Ranunculus longirostris
Longbeak Buttercup
341
FW
Lemna minor
Duckweed
1250
47
-------
Table A.15. Continued.
FW
Lemna trisulca
Duckweed
1282
SW
Phaeodactylum tricornutum
Diatom
3000
FW
Callitriche platycarpa
Macrophyte
>3300
FW
Chlorella pyrenoidosa
Green Algae
6205
FW
Chlorella vulgaris var. viridis
Green Algae
7714
FW
Chlorella kessleri
Green Algae
34300
Table A.16. Data for prometryn.
Lowest
Geometric
Medium
Species Scientific Name
Species Category
Mean (ug/L)
FW
Navicula pelliculosa
Diatom
1
FW
Scenedesmus acutus var. acutus
Green Algae
1.65
SW
Skeletonema costatum
Diatom
7.6
FW
Scenedesmus quadricauda
Green Algae
9.7
FW
Lemna gibba
Inflated Duckweed
11.8
FW
Chlorella pyrenoidosa
Green Algae
12
FW
Pseudokirclineriella subcapitata
Green Algae
12
FW
Lemna perpusilla
Duckweed
13
FW
Anabaena flosaquae
Blue-Green Algae
40.1
FW
Lemna aequinoctiales
Duckweed
40.97
SW
Dunaliella tertiolecta
Green Algae
53
FW
Chlorella vulgaris
Green Algae
53.6
FW
Spirodela polyrhiza
Large Duckweed
85
able A.17. Data for selenium.
Lowest
Geometric
Medium
Species Scientific Name
Species Category
Mean (ug/L)
FW
Scenedesmus acutus var. acutus
Green Algae
89
FW
Monoraphidium griffithii
Green Algae
622
FW
Pseudokirchneriella subcapitata
Green Algae
820
FW
Monoraphidium convolutum
Green Algae
1068
FW
Oscillatoria agardhii
Blue-Green Algae
1285
FW
Monoraphidium contortum
Green Algae
1616
FW
Lemna minor
Duckweed
2903
FW
Anabaena flosaquae
Blue-Green Algae
4899
FW
Microcystis aeruginosa
Blue-Green Algae
6788
SW
Skeletonema costatum
Diatom
7664
FW
Hymenomonas elongata
Coccolithophorid
26500
FW
Anabaena constricta
Blue-Green Algae
104499
FW
Chlorella ellipsoidea
Green Algae
112250
48
-------
Table A.18. Data for terbuthylazine.
Lowest
Geometric
Medium
Species Scientific Name
Species Category
Mean (ug/L)
FW
Pseudokirclineriella subcapitata
Green Algae
7.88
FW
Navicula pelliculosa
Diatom
11
FW
Lemna gibba
Inflated Duckweed
16
FW
Scenedesmus subspicatus
Green Algae
16
SW
Skeletonema costatum
Diatom
31
FW
Myriophyllum spicatum
Eurasian Watermilfoil
55
FW
Ceratophyllum demersum
Coon-Tail
61.26
FW
Lemna minor
Duckweed
66.63
FW
Anabaena flosaquae
Blue-Green Algae
99
FW
Callitriche platycarpa
Macrophyte
137.1
FW
Potamogeton crispus
Curled Pondweed
147.3
FW
Elodea canadensis
Waterweed
172.9
FW
Spirodela polyrhiza
Large Duckweed
182.4
FW
Lemna trisulca
Duckweed
254
Table A.19. Data for triclosan.
Medium
Species Scientific Name
Species Category
Lowest
Geometric
Mean (ug/L)
FW
Scenedesmus subspicatus
Green Algae
0.990
FW
Anabaena flosaquae
Blue-Green Algae
1.25
FW
Pseudokirclineriella subcapitata
Green Algae
2.23
SW
Dunaliella tertiolecta
Green Algae
3.55
FW
Navicula pelliculosa
Diatom
19.1
FW
Lemna minor
Duckweed
26.3
FW
Lemna gibba
Inflated Duckweed
57.14
SW
Skeletonema costatum
Diatom
>66
FW
Nitzschia palea
Diatom
370
FW
Closterium ehrenbergii
Green Algae
620
49
-------
Table A.20. Data for zinc.
Medium
Species Scientific Name
Species Category
Lowest
Geometric
Mean (ug/L)
FW
Synecliococcus leopoliensis
Blue-Green Algae
4.09
FW
Clilorella vulgaris
Green Algae
34
FW
Pseudokirchneriella subcapitata
Green Algae
52.85
SW
Asterionella japonica
Diatom
58.85
SW
Ditylum brightwellii
Diatom
100
SW
Nitzschia closterium
Diatom
132.72
FW
Stichococcus bacillaris
Green Algae
293.4
FW
Scenedesmus quadricauda
Green Algae
319.9
FW
Desmodesmus subspicatus
Green Algae
326.3
SW
Ulva pertusa
Green Algae
649.7
FW
Chlamydomonas reinhardtii
Green Algae
879.5
FW
Parachlorella kessleri
Green Algae
933.6
FW
Azolla pinnata
Water Velvet
948.2
FW
Navicula seminulum
Diatom
2523.9
SW
Chaetoceros gracilis
Diatom
4000
SW
Dunaliella tertiolecta
Green Algae
6000
SW
Isochrysis galbana
Haptophyte
6408
FW
Chlorella saccharophila
Green Algae
7050
FW
Elodea canadensis
Waterweed
8100
FW
Lemna minor
Duckweed
9600
SW
Macrocystis pyrifera
Giant Kelp
10000
FW
Navicula incerta
Diatom
10100
SW
Hormosira banksii
Neptune'S Necklace
19975
FW
Myriophyllum spicatum
Eurasian Watermilfoil
20900
SW
Phaeodactylum tricornutum
Diatom
37100
FW
Spirodela polyrhiza
Large Duckweed
48600
FW
Euglena gracilis
Flagellate Euglenoid
57543
FW
Chlamydomonas acidophila
Green Algae
75852
50
-------
Appendix B. Lognormal Probability Plots for Species Sensitivity
Distributions (Equation 3).
Open markers are data for the full data set. Close markers are the FIFRA-5 subset. Triangle data
points are for data that were greater than. Solid (full dataset) and dashed lines (FIFRA) are the
regression best fit for each data set. Vertical red dashed line intersects the regressions at the log of
the HC5 value.
51
-------
Arsenic V
9.0
8.0
7.0
6.0
o 5.0
W
Jj 4.0
3.0
2.0
0.0
-2.0
¦1.5
¦1.0
-0.5
0.0
0.5
1.0
1.5
2.0
-------
Cadmium
12.0 n
10.0 -
8.0 -
o
W
4.0 -
2.0 -
0.0
/o
o
Y
xgr •
°A-
er
v
y
—i—
—i 1 1 1 1 1 1 1 1
-2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5
-------
Copper
11.0
9.0
7.0
W
5.0
3.0
¦2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5
-------
Diuron
10.0
8.0
OO O
qOO
6.0
io
4.0
2.0
0.0
-2.0
-2.0
¦1.5
1.0
-0.5
0.0
0.5
1.0
1.5
2.0
-------
7.0
Irgarol
o o
5.0
3.0
o
W
O
LU
C
_l
-3.0
-5.0
¦2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5
-1F(c)
0.5
2.0
56
-------
6.0
Linuron
5.0
4.0
3.0
2.0
0
0.0
-2.0
-0.5
0.0
0.5
2.0
-------
Metribuzin
9.0
8.0
7.0
6.0
o
W
O
LU
C
_l
3.0
2.0
0.0
-2.0
1.5
¦1.0
-0.5
0.0
0.5
1.0
1.5
2.0
-------
Pentachlorophenol
12.0
10.0
8.0
oo
o
io
w 6.0
,00
,ooo'
c
_i
4.0
2.0
0.0
-2.0
¦1.5
¦1.0
-0.5
0.0
0.5
1.0
1.5
2.0
-------
Selenium
14.0
12.0
10.0
O
If)
w 8.0
c
_i
6.0
4.0
2.0
-2.0
¦1.5
1.0
-0.5
0.0
0.5
1.0
1.5
2.0
-------
Triclosan
7.0
6.0
5.0
4.0
3.0
O
in
O
!±L 2.0
c
_l
0.0
-2.0
-2.0
¦1.5
¦1.0
-0.5
0.0
0.5
1.0
1.5
2.0
------- |