3 Proposed Methodology for
4 Specifying Atrazine Levels of Concern
5 for Protection of Plant Communities
6 in Freshwater Ecosystems
7
8 Report To:
9 Environmental Fate and Effects Division
10 Office of Pesticide Programs
11 U.S. Environmental Protection Agency
12 Washington, DC
13
14 EPA/600/R-12/019
15
16 Russell Erickson
17 Mid-Continent Ecology Division
18 National Health and Ecological Effects Laboratory
19 Office of Research and Development
20 U.S. Environmental Protection Agency
21 Duluth,MN
22
23 March 5, 2012
24
-------
25 ACKNOWLEDGEMENTS
26 This document has been reviewed in accordance with U.S. Environmental Protection
27 Agency policy and approved for publication. Mention of trade names or commercial products
28 does not constitute endorsement or recommendation for use. The author would like to
29 acknowledge the valuable reviews of drafts of this document by:
30 David Mount, U.S.EPA, Mid-Continent Ecology Division
31 Dale Hoff, U.S.EPA, Mid-Continent Ecology Division
32 Mary Ann Starus, U.S.EPA, Mid-Continent Ecology Division
33 Glen Thursby, U.S.EPA, Atlantic Ecology Division
34 Joseph S. Meyer (retired), University of Wyoming, Department of Zoology and Physiology
35
36 The author would also like to acknowledge the importance to this document's development of his
37 numerous interactions over the last several years with various personnel of U.S.EPA's Office of
38 Pesticide Programs and Office of Water.
39
-------
40 1. INTRODUCTION
41 This document describes a proposed methodology for setting a level of concern (LOG)
42 for atrazine in natural freshwater systems to prevent unacceptably adverse effects on the aquatic
43 plant communities in those systems. Effects on humans and possible endocrine-disruption in
44 aquatic vertebrates are subjects of separate efforts, and certain implementation issues for aquatic
45 plant community atrazine risk assessment are also described elsewhere. This first section defines
46 the problem being addressed and describes a general framework for setting the LOG.
47 1.1 Requirements for the LOC Methodology
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
Figure 1. Examples of atrazine exposure time-series in natural
freshwater systems.
MO 01 2010
Toxic chemical risk assessment problem definition requires defining the exposure
scenarios to be addressed, specifying the assessment endpoints of concern, and identifying
measures of effect for the assessment endpoints (U.S.EPA 1998).
This LOC methodology must address the types of atrazine exposures occurring in natural
ecosystems for which risk is to be assessed. Atrazine enters natural freshwater systems primarily
in rainfall-driven runoff, resulting in highly variable and episodic exposures that depend on
rainfall distribution, atrazine application patterns, topography, and soil properties. Figure 1
provides example time-series of
atrazine exposures during 2010 in
three Missouri streams, measured as
part of a monitoring program being
conducted to satisfy risk evaluations
required under the 2003 interim
reregi strati on of atrazine (U.S.EPA
2003). These examples illustrate
substantial variation in exposure
patterns, and thus the need for the
LOC methodology to address the
relationship of effects to time,
including high concentrations with
limited durations, multiple events,
and prolonged, variable exposures at
low to moderate concentrations. The
top and bottom series have similar
average concentrations but very
different peaks, underscoring the
issue of the comparative risk of short,
intense exposures to more prolonged
exposures at lower concentrations.
o>
N
CO
150
100
50
90
60
O)
30
30
20
10
MO 07 2010
MO 02 2010
90
120
150
180
210
Julian Day
The assessment endpoint for this LOC methodology is the productivity and composition
of natural aquatic plant communities. Although atrazine has been the subject of many toxicity
tests on individual aquatic plant species and although such tests are often used as measures of
effect for aquatic plant risk assessments (e.g., Solomon et al. 1996, Giddings et al. 2000), they
will not be used directly for that purpose in this methodology. Rather, because atrazine has been
-------
Figure 2. Effects of atrazine on experimental ecosystems as a function of exposure duration and average
concentration. Closed triangles denote adverse effects, open triangles no effects.
10000 -i
N
CD
1000 -
I
O
O
0
O)
0
10-
10 20 50 100
Test Duration (days)
200
82 the subject of many experimental aquatic ecosystem studies documenting plant community
83 responses, these will be used to provide measures of effect and to serve as the foundation for
84 defining exposures causing effects of concern. Figure 2 summarizes an evaluation of such
85 studies conducted by the U.S.EPA's Office of Pesticide Programs (OPP) Environmental Fate and
86 Effects Division (EFED) (U.S.EPA 2011). In Figure 2, each experimental ecosystem treatment
87 is characterized by the duration over which effects were assessed, the average atrazine
88 concentration over this duration, and whether there were unacceptably adverse effects on the
89 plant community. For each point on Figure 2, Appendix B of this report provides more complete
90 exposure information, the effects designation, and a literature citation; other information on the
91 analyses of these studies can be found in U.S.EPA (2011). It should be emphasized that a
92 fundamental assumption in using such experimental ecosystem data is that they collectively
93 describe a relationship of effects to exposure that is relevant to the probability of effects (i.e.,
94 risk) occurring in natural freshwater systems. In other words, it is assumed that natural aquatic
95 plant communities will generally react adversely if subjected to the same atrazine exposures that
96 elicited adverse effects in the experimental ecosystem studies. This assumption is inherent in
97 any assessment that extrapolates toxicity experiments to the field, and the use of experimental
98 ecosystems arguably provides a better basis than do single-species toxicity tests.
99 Figure 2 illustrates three important requirements for the LOG methodology:
100 (1) Diversity among the experimental approaches precluded characterizing each experimental
101 ecosystem treatment with an identical, quantitative measure of effect. Therefore, LOG
102 characterizations must rely on a binary (acceptable vs. unacceptable) characterization of effect.
103 (2) Although the exposures that resulted in adverse effects are somewhat separated from those
104 that did not cause adverse effects, substantial overlap exists between these two groups, especially
105 in the 10-20 u,g/L range. This variability is presumably due to the combined effect of:
106 differences in the nature of the experimental systems; differences in the experimental design and
-------
107 the endpoints measured; and random variability of the response of any given system. The
108 methodology must address how to specify an LOG within such variability.
109 (3) The LOG methodology must address the relationship of effects to time. This is important
110 not only because of the variability of field exposures shown in Figure 1, but also because of the
111 different durations of the experimental ecosystem exposures (Figure 2) and exposure variability
112 within these durations (Appendix B). Because data in Figure 2 do not provide information on
113 the relationship of the same endpoint to different exposure time-series, this time-dependence
114 issue must be addressed in the formulation of the extrapolation methodology discussed below.
115 1.2 General Framework for the LOC methodology
116 The key issue that this LOC methodology must address is how to relate aquatic plant
117 community effects elicited in an experimental ecosystem by a particular atrazine exposure time-
118 series to markedly different time-series in other experimental studies or natural systems. If all
119 exposures of interest had the same shape (i.e., the same exposure duration and the same relative
120 changes in concentration within that duration), the LOC could be based on the relationship of
121 effects in the experimental studies to any convenient measure of exposure. However, the
122 markedly different exposure shapes discussed above preclude such a simple approach, and there
123 is thus a need for a method to translate any exposure time-series to a "common currency" that
124 integrates time and concentration into an index of the relative total severity of effects from the
125 exposure. This "effects index" serves only as a relative measure of effect because the
126 experimental ecosystem effects define the absolute levels of concern. Text Box 1 further defines
127 and discusses this concept of an effects index.
Text Box 1. The nature and purpose of the "effects index".
To further clarify the nature and purpose of the "effects index", consider a simple
hypothetical example in which the results from a single experimental ecosystem study must be
used to assess risk to the same ecosystem, but for an exposure with a different shape. For this
example, the experimental ecosystem study is specified to (a) involve constant atrazine exposure
over 30 d at several concentrations and (b) demonstrate that 20 fjg atrazine/L constitutes an LOC
based on the magnitude of effects elicited. However, this concentration-based LOC applies only to
constant, 30-d exposures, whereas the exposure of interest is specified for this example to be a
10-d exposure at 100 fjg atrazine/L. The basic question is whether this more intense (5x higher)
but more brief (3x shorter) exposure should be considered worse than the 30 d LOC concentration,
provided the effects are assessed in the same manner and over the same time period as in the
original study.
A very simple "effects index" for this would assume that effects increase linearly with both
concentration and time, so that the effects index could be the area under the exposure time-series,
measured in "ppb-days" (note: this effects index definition is provided only to illustrate the concept
- the actual methodology should consider the nonlinearity of effects versus exposure) The LOC
for this effects index would therefore by 600 ppb-days (20 /jg/L x 30 days) based on the
experimental results. This effects index-based LOC is exceeded by the effects index value of 1000
ppb-days (100 /jg/L x 10 days) for the new exposure of interest.
This effects index is a relative measure in that it has no inherent absolute meaning for risk
except when calibrated to the experimental ecosystem results. Its use is only for translating any
exposure time-series to a common scale of comparison, so that the LOC of 600 ppb-days can be
used to judge any other exposure of interest, provided the exposure is for a system to which the
experimental ecosystem is relevant.
-------
128 The effects index proposed for the LOG methodology will be described in Section 2. For
129 discussing the assessment framework here, it is only necessary to assume the existence of an
130 effects index that is suitable for comparing the relative severity of different exposure time series.
131 Figure 3 provides a schematic of an assessment framework using such an effects index.
132 The process starts (Box 1) with compiling relevant experimental ecosystem data,
133 categorizing each treatment as to whether there was an effect or not and specifying the exposure
134 time-series for the treatment. This step is not a subject of this report, but rather is addressed in
135 U.S.EPA (2011). The effects index is then calculated (Box 2) for each experimental ecosystem
136 treatment, providing the "common currency" to compare the severity of each exposure. The
137 relationship of the binary experimental ecosystem effects to this effects index is then examined
138 (Box 3) to set a level of concern for the effects index (LOCEi), based on the probability of
139 eliciting an effect (i.e., risk).
140 The LOCEi is applied to exposures in natural systems as follows. Exposure time-series
141 are compiled for the various exposures of interest in natural ecosystems (Box 4) and the effects
142 index for each exposure is computed (Box 5). Risk is characterized (Box 6) by dividing the
143 effects index by the LOCEi to compute the "effects exceedence factor" (EEF). The EEF indicates
144 whether the LOG is exceeded (i.e., EEF>1) and by how much. The EEF thus represents a risk
145 quotient approach, but this different terminology is used here to distinguish this effects-based
146 quotient from concentration-based risk quotients commonly used.
Figure 3. Assessment framework for risk of atrazine to aquatic plant communities, based on experimental
ecosystem results and an effects index for comparing different exposure time-series.
(I1 Compile Experimental Ecosystem Data:
£ *posuf * Ttm*-5enes Effect Categorization
I'll Compile Time-Series for Exposures
of interest in Natural Ecosystems
2) Compute Efftctslncl** for Each
»i'i«iini*ntii Ecosystem Treatment
(5| Compute Effects Index for Each
Exposure I»me-Seriesof Interest
) Wt level of Concern for EHstts Index ti0€|,j
B*$fd on Experimental Ecosystem Effects
C a tegor iralion Ver sus Effec is Index
LOCc
El
(6) Compute Effects E«c«*d*nee Factor or
Cc»(K«nbal!onE:««»€d«nct Factor for Each
Exposure Time-Series of Inter eft
f f) on and for
of an
-------
147 Risk can also be characterized by what is termed the "concentration exceedence factor"
148 (CEF) in Box 6. This factor is based on iterative calculations to determine the multiplicative
149 factor by which the exposure must be decreased so that the effects index exactly equals the
150 LOCEi. As for the EEF, a CEF indicates whether the LOCEi is exceeded and by how much, but
151 on a concentration scale rather than an effects scale. This could have some advantage in
152 determining remediation goals or, conversely, determining how far exposures are below levels of
153 concern. However, this is an approximate measure for such purposes, because the CEF is
154 premised on the same multiplicative factor applying to the entire concentration time-series.
155 Box 7 and the associated gray arrows in Figure 3 represent a final step in the assessment
156 framework that is not addressed in this document. It would be desirable for LOCs to be on a
157 concentration scale rather than an effects scale so that they relate more easily and directly to
158 exposure monitoring data. In Box 7, the relationship of EEFs to an average exposure
159 concentration for a large number of existing exposure time series is examined to determine an
160 LOG based on this average concentration, and which then can be applied to new exposure time-
161 series, for which the effects index need not be computed. Developing such a concentration-
162 based LOG from the effects index-based LOG is being addressed separately by EFED.
163 Finally, it should be emphasized that the only site-specific factor intended to be addressed
164 in this LOG methodology is the exposure time-series. The methodology is not intended to
165 address other site-specific factors, such as physicochemical conditions and the nature of the
166 biological community. Addressing such conditions is not feasible from a standpoint of both
167 effort/cost and knowledge of their influence on atrazine effects. Rather, this method will be
168 generic in that any site with the same atrazine concentration time-series will be assessed as
169 having the same risk.
170
-------
171 2. PLANT ASSEMBLAGE TOXICITY INDEX
172 2.1 Potential Effects Indices
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
There are various possibilities, with differing complexities, for calculating an effects
index to serve in the assessment framework of Figure 3. For illustrative purposes only, Text Box
1 assumed that effects increased linearly with both concentration and time, leading to an effects
index of ppb-days. To actually apply this simple, linear model a priori is not justified. Rather,
the effects index should consider ecotoxicological relationships.
At the other extreme of complexity are community simulation models that address not
only the immediate impact of atrazine on plant community primary production, but also consider
the ramifications of this on plant community dynamics throughout a growing season. Earlier
efforts for developing an LOG methodology considered the use of the Comprehensive Aquatic
Simulation Model (CASM) (Bartell et al. 2000, Volz et al. 2007), but determined that this model
was not suitable for the purposes here (U.S.EPA 2009, Erickson 2009). This model does not
provide any clear, validated, substantial added-value beyond describing the immediate response
of plant community growth, entails extensive data and parameterization needs that were not
completely satisfied, and involves considerable uncertainty. CASM is more suited for focused
site assessments, involving considerable resources for system-specific model development and
application, and a completely different assessment framework.
A community simulation model such as CASM applies information from atrazine toxicity
tests on individual plants species to calculate the direct (primary) impact on the plant community
being simulated, but then also considers the secondary (indirect) ramifications on plant
community dynamics. The direct, primary impact was determined to be more important for
assessing the relative impact of different atrazine exposure time-series (i.e., the purpose of the
effects index) than are the secondary impacts (U.S.EPA 2009). Thus, the approach pursued here
was to base the effects index just on this primary impact, avoiding various uncertainties and
complexities in the community model.
The need here therefore is to use the
collective information from toxicity tests on
individual plant species to provide a measure
of direct impacts of atrazine on plant
communities. To this end, past assessments
of the risk of atrazine to aquatic plant
communities (e.g., Solomon et al. 1996;
Giddings et al. 2000) have generally
summarized the results of a toxicity test as a
median effect concentration (ECso), the
concentration causing a 50% decrease in
some measure of growth over the duration of
the test. Average ECsos for each species are
then used to describe a species sensitivity
distribution (SSD) - the cumulative
percentage of species with ECsos less than a
Figure 4. Example of aquatic plant SSD based on data
from Giddings et al. (2000).
100 •
0) 80 •
(0
tr
0 60 •
Q_
•I! 40 •
1 20-
0 •
1
/ '
I
jr
f
f
/
xf
i 10 20 50 100 200 500 1000 2000
EC50 (p,g/L Atrazine)
-------
213 certain value (e.g., Figure 4). SSDs are typically applied by addressing what percentiles are
214 exceeded by an exposure. For example, in Figure 2, an exposure of 10 |J,g/L would be below the
215 ECsos of 95% of the species and an exposure of 45 |J,g/L would be below the ECsos of 80%.
216 However, such SSDs have major shortcomings, especially for addressing the types of
217 exposures in Figure 1:
218 (1) SSDs based just on ECsoS provide limited information on the overall toxic impact to the
219 assemblage of species used for the SSD. For example, the 5* percentile in Figure 4 only
220 describes the concentration at which the growth of a particular species is reduced by 50%. No
221 information is provided on how much greater effects on this species are at higher concentrations,
222 or how much smaller effects are at lower concentrations. For other species, no information is
223 given other than that their ECsoS are less than or greater than the LOG. Much more information
224 regarding effects is contained within the toxicity test data, but how should it be used?
225 (2) SSDs such as in Figure 4 also do not address the issue of time. How should effects be
226 described for longer or shorter exposures and, especially, exposure concentrations that fluctuate?
227 If an LOG based on an SSD percentile is simply applied to the peak exposure, the exposure time-
228 series in the top panel in Figure 1 would be considered of most concern, but toxic impact would
229 probably be greater for the middle time-series and perhaps as great for the lower time-series,
230 because of the more prolonged and multiple exposure periods. How should total impact be
231 assessed over an entire time-series?
232 (3) Although the ECsoS in Figure 2 all describe plant growth in some fashion, growth is measured
233 in a variety of ways (final plant biomass, net change in biomass, growth rate, oxygen evolution,
234 carbon fixation, plant length, cell numbers, changes in chlorophyll) and over a wide range of
235 exposure durations and conditions, such that these ECsoS can have greatly different meaning
236 regarding actual plant sensitivity. The spread of values in the SSD might therefore be due to
237 differences among test endpoints as well as differences among species. Such inconsistency in
238 the meaning of ECsoS will cause any LOG from the SSD to have uncertain meaning.
239 2.2 Definition of the Plant Assemblage Toxicity Index
240 To quantify the overall effect of atrazine on an assemblage of plant species of interest, the
241 effects index proposed here is the "Plant Assemblage Toxicity Index" (PATI). PATI is a simple
242 extension of the SSD concept that (a) considers the entire growth inhibition vs. concentration
243 curve ("toxicity relationship") for each plant species and (b) determines the average effect level
244 across all species (the "assemblage") at each concentration. Figure 5 illustrates this, using
245 atrazine toxicity data summarized in Appendix A. The middle panel shows overlapping toxicity
246 relationships for 20 plant genera. In the top panel, the ECsos for each genus are used to create a
247 traditional SSD - simply the cumulative percentage of the ECsoS. For the bottom panel, the
248 average magnitude of effect across all species at each concentration is used to create the PATI
249 distribution. At 50 |J,g/L, the average effect over all genera is 19%, providing the PATI value in
250 the bottom panel (arrow). Thus, rather than just providing the percentage of species that have an
251 ECso below some concentration (e.g., 50 |J,g/L corresponds roughly to the 16* percentile on the
252 SSD), PATI describes the percent reduction in plant production for the entire assemblage
253 (weighting each species equally). Although the shape of the PATI curve is similar to that of the
-------
Figure 5. Comparison of toxicity relationships for 20
plant genera (middle panel), the SSD of EC50s for
these genera (top panel), and the plant assemblage
toxicity index (bottom panel, PATI = the average of
the curves in the middle panel).
"j g
£ O
•
Atrazine Concentration (j-ig/L)
1!
254 traditional SSD curve, it provides more
255 information on the total impact on the plant
256 assemblage and allows more meaningful
257 comparisons between different exposure
258 concentrations.
259 However, the definition and
260 calculation of PATI illustrated in Figure 5 is
261 not yet complete because it does not address
262 the issue of time. For a time-series of daily
263 concentrations, there would need to be
264 separate calculations for each day to generate
265 a time-series of daily PATI values, using
266 toxicity endpoints relevant to this timeframe.
267 Because of the rapid recovery of growth rates
268 in toxicity tests when atrazine exposures are
269 terminated (e.g. Abou-Waly et al. 1991,
270 Desjardin et al. 2003), daily PATI values need
271 not consider residual toxicity from exposures
272 on previous days, but rather only the toxicity
273 for the current day's exposure.
274 Because the effects index is intended
275 to describe total toxic impact, the approach
276 here to address time is simply to sum the daily
277 PATI values to provide a "cumulative PATI".
278 This is illustrated in Figure 6. Concentrations
279 in the left panel are converted to daily PATI values (middle panel), which are then summed to
280 provide the cumulative PATI values in the right panel. The cumulative PATI can also be viewed
281 as the "area under the curve" of the daily values, this area being a measure of the total toxic
282 impact of the exposure.
Atra ine Concentration (|j.g/L)
Atrazine Concentration (|ig/L)
Figure 6. Overview of PATI calculations. A concentration time-series (left panel) is converted to expected
instantaneous or daily reductions in plant assemblage growth (middle panel), which is then integrated to provide
a cumulative PATI value for the exposure (right panel).
100
10 15
Time (d)
10
15
20
10
-------
283 The summation units of this cumulative PATI are analogous to the ppb-days discussed
284 earlier or, more familiarly, with degree-days used to describe the total heating or cooling impact
285 of seasonal weather. A fundamental aspect of such a summation is that a certain reduction in
286 growth over 1 d is treated as having equal importance as: half that reduction persisting for 2 d; a
287 quarter of that reduction persisting for 4 d; etc. Although such a general time-dependence has
288 not been demonstrated for actual aquatic ecosystems, it has been observed to approximate well
289 the cumulative effects on biomass in single-species toxicity tests that maintain a constant level of
290 effects on plant growth rate during the exposure period (e.g., Shafer et al. 1994).
291 This methodology uses a simple summation of toxic effects to provide an index for the
292 relative toxic effects of different time-series on plant communities and deliberately does not
293 address any further effects on plant community dynamics beyond short-term reductions in
294 growth across the plant assemblage. As already noted, the basic PATI calculation is similar to
295 the first step in community models such as CASM, which on each day calculates the toxic
296 impact on the growth of various species - the fundamental difference being that PATI does not
297 consider how this toxicity changes community composition through time. Because community
298 dynamics are driven on each day by the same growth reductions that are incorporated into PATI,
299 PATI does describe the primary driving force for atrazine effects on plant communities. Even if
300 community dynamics modify the relative severity of some time-series compared to that expected
301 based just on PATI, these would be secondary effects and are not understood well enough to be
302 satisfactorily addressed (U.S.EPA 2009, Erickson 2009).
303 However, this summation cannot be continued indefinitely, but rather is limited here to
304 an "assessment period" that can reflect risk management decisions about cumulative effects. For
305 example, if two short atrazine exposures were separated by 90 d, a 120 d assessment period
306 would consider them cumulative whereas a 60 d assessment period would not, this shorter period
307 instead assuming that sufficient time had passed that the second exposure should be assessed
308 independently of the first. The shorter assessment period would also avoid assigning concern to
309 prolonged low exposures of uncertain, minor impact. For exposures with finite durations less
310 than the assessment period, the summation would simply stop at the exposure duration. For
311 exposures with durations greater than the assessment period, the summation would encompass
312 the worst part of the exposure. For this report, this limit on cumulative toxicity will be
313 designated with a subscript denoting the length of the assessment period (e.g., PATIsod denotes a
314 30-d assessment period). Without a subscript, PATI will refer to daily or instantaneous values,
315 or the general PATI concept. The selection of the assessment period is addressed in Section 4.
316 2.3 Single-Species Plant Toxicity Test Data
317 Implementation of the PATI approach requires a compendium of the effects of atrazine
318 on aquatic plants or statistical distributions describing these effects. Existing compendia of plant
319 effects concentrations (ECs) (e.g., Giddings et al. 2000) have certain shortcomings regarding
320 their applicability to risk assessment, which warranted reanalysis of existing single-species
321 toxicity tests. This section describes: the shortcomings of concern; a new review and analysis of
322 toxicity data; and a new compendium of plant toxicity information more suitable for calculating
323 PATI and for conducting atrazine risk assessments.
324
11
-------
325 2.3.1. Issues in Interpreting and Applying Plant Toxicity Test Results
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
ECs from plant toxicity tests can vary widely in both value and meaning depending on
how tests are conducted and analyzed. For microalgae, tests are usually conducted on cell
suspensions under favorable (at least at test start) conditions of temperature, light, and nutrients.
These tests can involve various measurement endpoints, including (a) actual biomass; (b)
surrogates for biomass such as cell counts, cell volume, optical density, or chlorophyll content;
and (c) indicators of growth such as oxygen evolution or radioactive carbon fixation. The period
over which measurements are made can vary from several minutes to several weeks, and
measurements might be reported at multiple times or only at the end of exposure. Biomass or
biomass surrogates might be analyzed based on (a) biomass values at various times during the
exposure, (b) biomass increase (growth) at
various times, (c) the area under the growth
time-series (AUC), and/or (d) specific growth
rate(SGR)1.
The meaning of an EC can be greatly
affected by test duration and by whether it is
based on absolute biomass, growth, or SGR.
To illustrate this, Figure 7 provides a
hypothetical example comparing growth when
the control SGR (SGRC) is 1.0/d to when a
chemical exposure reduces the SGR to half of
this value. The top panel shows the actual
biomass vs. time in the control compared to
the chemical exposure, while the bottom panel
shows the percent reduction due to chemical
exposure for SGR (constant at 50%), absolute
biomass, and growth (biomass increase).
For growth, the treatment that is an
EC50 for SGR will be an EC62 at 1 d, an EC73
at 2 d, and an EC88 at 4 d if the SGRc is 1/d.
For absolute biomass, this concentration
would be an £€39, ECes, ECge, respectively, at
these times. For other values of SGRc, more
widely ranging ECs can occur. Using absolute
biomass can result in particularly misleading
ECs when growth rates are modest. For
example, when control growth is just a
doubling of biomass over the duration of the
Figure 7. Variation of plant growth effects with time
and measurement endpoint. Top panel shows
exponential biomass changes at the control SGR
(solid line) and at one-half of the control SGR (dashed
line). Bottom panel converts this to percent effect on
biomass (solid line), on biomass increase (dashed
line), and on specific growth rate (dotted line).
Control Biomass I
o
ffi
Biomass for
50% Reduction in
Specific Growth Rate
Initial Biomass
o
O
.3
•5
D
•o
0
QL
Time (d)
Biomass Increase „ — "*
Biomass
Specific Growth Rate
1 2 3
Time (d)
1 The specific growth rate (SGR) =dB(t)/dt/B(t), where B is biomass and t is time. SGR is thus the fractional rate of
change of biomass with time and has units of inverse time. If SGR is constant, the growth rate is exponential and
B(t)=B(0)-eSGR't. Thus, if SGR is 1/d, this does not mean that the biomass will double in one day; rather the
"compounding interest" of exponential growth will mean that biomass actually increases to 2.7 times the initial
value - only over short periods will fraction growth closely adhere to SGR (e.g., 1% growth over 0.01 d).
12
-------
363 test, an ECso for absolute biomass actually represents no growth. Such issues with endpoint
364 definition have been noted by others (e.g., Bergtold and Dohmen 2010) and are reflected in
365 recent OECD guidelines.
366 Therefore, ECsos reported for absolute biomass, growth, and SGR will differ from each
367 other, and these differences will vary with exposure duration and the SGRc. This is especially
368 problematic when reports for toxicity tests just provide ECs, without sufficient information on
369 absolute biomasses and/or SGRs as a function of time and concentration to calculate more
370 consistent and meaningful measures of effect. Compendia that simply transcribe reported ECsoS
371 can be describing a wide range of different effects, and assessments based on such compendia
372 will be ill-defined.
373 Other factors make the meaning of reported plant ECs even less certain. As an algal
374 suspension grows, the growth rate will decline because of nutrient depletion and self-shading.
375 This departure from exponential growth will be most pronounced in the treatments with the
376 highest growth rates (i.e., the control and low toxicant concentrations with little or no effect), so
377 that the treatments with greater toxic effects might "catch up" as exposure duration increases,
378 causing ECs for total growth to not decrease with time as much as they would without these
379 limitations, or to even increase with time. In other words, the toxicity test actually can include
380 stressors (nutrient/light limitations) in addition to the toxicant that can confound the effects of the
381 toxicant. In fact, some standard plant test protocols were originally designed to assess nutrient
382 limitations, and the durations were selected to result in nutrient depletion (e.g., Miller et al.
383 1978). When used for toxicants, this type of study design can result in complicated growth
384 dynamics and relationships that are difficult to interpret and apply. Tests can also have different
385 photoperiods, which would also need to be considered in comparing ECs for growth (although
386 ECs for SGR can be directly compared between different photoperiods).
387 Schafer et al. (1994) provide a noteworthy example of some of these problems. In a 10-d
388 test in a flow-through system in which a constant control growth rate was maintained by
389 replenishing the nutrient solution and periodically cropping biomass, they reported growth-based
390 EC50s to drop from 50 ng/L at 4 d to 20 ng/L at 7 d to 10 ng/L at 10 d. This is plausibly
391 attributable to a constant relationship of SGR to concentration during these 10 d, so that a
392 constant EC for growth rate translates into widely variant ECs for growth. These authors also
393 reported an ECso of 350 |J,g/L for a static, 3-d flask test, indicating much less sensitivity
394 compared both to the flow-through systems and to photosynthesis measurements made in the
395 first day of these static tests. This apparent lower sensitivity likely is due at least partly to a high
396 initial cell density (2-105 cells/ml), which would have resulted at 3 d in a cell density of 3-108
397 cell/ml if a SGRc similar to that in the flow-through system had been maintained for the entire 3
398 d. Such a cell density would have resulted in both self-shading and nutrient depletion in the
399 control, contributing to the apparent reduced sensitivity. Increases with time for growth-based
400 ECs are evident in other studies in the review presented later, although the opposite can also be
401 true, indicating additional complexities.
402 Changes in cell condition other than light and nutrient limitations might also affect ECs
403 and their dependence on test duration. For example, chlorophyll content per cell can increase
404 with time to compensate for reduced photosynthesis. Mayer et al. (1998) reported the
13
-------
405 chlorophyll content of algal cells to increase by 10-fold in response to exposure to 200 (ig/L
406 atrazine. Such changes in the chlorophyll content per cell make the use of chlorophyll as a
407 surrogate for plant biomass inadvisable, potentially misrepresenting toxic effects on biomass.
408 For example, van der Heever and Grobbelaar (1996) reported effect concentrations in the same
409 exposures to be about 2.5-fold higher when based on chlorophyll than when based on cell
410 numbers or dry weight. Similarly, toxicants can alter cell volume and mass (e.g., van der Heever
411 and Grobbelaar 1996), creating differences among ECs based on cell count, cell volume, and cell
412 weight, although these differences are much smaller than those due to the influence of
413 chlorophyll, test duration, nutrient depletion, and light limitations.
414 Although oxygen production and radiocarbon fixation are arguably closely linked to
415 biomass production, ECs based on these measures can also pose interpretation problems:
416 (a) They are often done over such short durations that apparent effects might be reduced because
417 of the time it takes to fully induce the effects of a toxicant, unless there is sufficient pre-exposure
418 to the toxicant before the measurements are made. Fortunately, for atrazine, effects do appear to
419 be induced quickly, such that ECsoS based on oxygen measurements with just several minutes
420 prior exposure have been reported to be similar to those based on biomass measurements (e.g.,
421 Turbaketal. 1986).
422 (b) Short-term radiocarbon fixation rates can conceivably reflect gross or net photosynthesis (or
423 a weighted combination of the two) depending on the disposition of the radioactive carbon in the
424 organism. Williams et al. (1996) determined that radiocarbon fixation over short periods
425 approximates net photosynthesis for good growing conditions (which would be expected in
426 toxicity tests); therefore, radiocarbon fixation will be assumed in this review to represent net
427 photosynthesis.
428 (c) Although oxygen production should parallel net photosynthesis, test methods using oxygen
429 evolution measurements can involve extremes of oxygen concentrations that might affect
430 photosynthesis and/or respiration - either high, supersaturated levels as oxygen increases from
431 initial levels, or low concentrations due to the methodology involving an initial purging of
432 oxygen. Studies with such extremes will not be used in this review because of uncertainty about
433 their impacts.
434 (d) Even when the test is such that oxygen production or radiocarbon fixation are arguably good
435 surrogates for biomass production, the time-scale of the measurements can affect their
436 interpretation. Short-term values for oxygen production or radiocarbon fixation for an
437 approximately constant mass of algae are analogous to the SGR, whereas measurements long
438 enough for substantial growth to occur would be analogous to net cumulative growth, creating
439 differences in the meaning of ECs similar to that for growth vs. SGR. In one study (Larsen et al.
440 1986), the situation was especially complicated because carbon-14 fixation was measured only
441 during a short period at the end of a 24-h atrazine exposure, so that the measured fixation rate
442 reflected both effects of the toxicant on the rate of carbon fixation per cell and the cumulative
443 differences in cell density due to the preceding exposure.
444 Macrophyte tests can be less susceptible to the issues of exponential growth and limiting
445 conditions discussed above. Many macrophytes grow slowly enough so that biomass increases
14
-------
446 by only a few multiples during the tests. Duckweed tests show more rapid growth, but also
447 usually do not reach biomass levels sufficient to suppress growth rates (frond crowding or
448 nutrient depletion). However, the general issues raised above for microalgae should still be
449 considered in the interpretation of macrophyte tests and the definition of their ECs. For example,
450 reduced photosynthesis can result in elongation of plant shoots with little or no biomass increase,
451 so that shoot length can be a poor surrogate for biomass changes (e.g., Fairchild et al. 1994,
452 1998). In addition, some macrophyte tests involve rhizomes, which contain resources to
453 temporarily support growth that might obscure toxic effects, again making length a questionable
454 measure and even making weight problematic if only shoot biomass is measured. Furthermore,
455 if test protocols with cuttings result in slow growth (e.g., due to the absence of rooting),
456 variability can make it difficult to quantify toxic effects and/or make such toxic effects of
457 uncertain relevance to the field. Finally, use of oxygen in interpreting growth of some vascular
458 plants might be confounded by gas exchanges to aerenchyma (air channels).
459 2.3.2. Review of Single-Species Plant Toxicity Tests
460 The inconsistency issues among single-species toxicity test ECs discussed above have not
461 been adequately addressed in past reviews of atrazine toxicity (e.g., Solomon et al. 1996;
462 Giddings et al. 2000) and might distort atrazine risk assessments. There was thus a need for
463 better analyses of single-species plant toxicity tests with atrazine to produce EC compendia
464 which are more consistent, providing a "common currency" that can be more legitimately
465 compared among tests and describe short term effects relevant to daily PATI values. The SGR
466 was selected as this "common currency" because it reduces the dependence of ECs on test
467 duration and is more directly applicable to addressing effects of time variable exposure. In
468 addition to compiling information on ECsoS, there was also a need for information on the entire
469 SGR vs. concentration curve, which is also inadequately addressed in previous compendia.
470 To this end, available single-species toxicity tests with atrazine were reviewed for
471 information regarding exposure conditions and effects by the Great Lakes Environmental Center
472 (Traverse City, MI) under support from the Office of Science and Technology of U.S.EPA's
473 Office of Water (EPA Contract 68-C-04-006, Work Assignment 4-34, Subtask 1-16). Journal
474 articles and reports identified by this review as containing potentially useful information were
475 further analyzed by U.S.EPA to compile desired information on the relationship of SGR to
476 atrazine concentration, using the following sigmoidal relationship (logistic equation):
o^n SGRc
477 SGR = — -^— —— (Equation 1)
4-Steep- loglo CATZ -loglo ECSO ^ H '
478 for which the parameters are the SGR-based EC50, the steepness of the relationship of SGR vs.
479 atrazine concentration ("Steep"}, and the control SGR (SGRc). Appendix A further discusses this
480 equation and its use in the analyses, as well as (a) guidelines and procedures used in the EPA
481 evaluations of toxicity tests and (b) a summary of each toxicity test reviewed. Table 1 provides
482 the compilation of SGR ECso, Steep, and SGRc from these analyses.
483
15
-------
484
Table
(SGR)
relatio
SGR(
onSG
1. Compiled data regarding atrazine toxicity to aquatic plants. All data pertain to the specific growth rate
of the plant. Compilation includes the EC50 for the SGR, a steepness parameter for a fitted logistic
nship of SGR to atrazine concentration (Steep=-d(SGR/SGRc)/d(logio(CATz)) at the EC50), and the control
SGRC) under the test conditions. Italicized EC50s denote values whose estimation required information
RC and/or steepness from other studies. Appendix A provides more details on these data and analyses.
Genus SGR EC,, (fig/L) Steep SGRc(d') Reference
CHLOROPHYTA (includes testedgreen algae)
Anhistrodesmus
Chlamydomonas
Chlarella
Scenedesmus
Selenastrum
Stigeoclonium
Vlothrtx
104
119
378
141
67
45
26
37
91
557
480
87
300
39
164
50
100
131
70
163
125
110
201
236
223
101
78
317
159
1.41
0.65
1.07
0.47
0.73
0.79
1.66
1.50
0.62
1.22
1.07
0.90
0.79
1.01
0.61
1.61
0.33
1.06
>1.4
0.26
1.80
1.93
1.25
1.75
1.65
1.01
0.97
Burrelletal. 1985
Laisenetal. 1986
Kallqvist and Romstad 1994
Schaferetal. 1993
Laisenetal. 1986
Hersh and Crumpton 1989
Faust etal. 1993
Hersh and Crumpton 1989
Burrelletal. 1985
Laisenetal. 1986
Strattonl984
Laisenetal. 1986
Stratton et al. 1984
Zagorc-Koncanl996
Mayer et al. 1998
Radetski et al. 1995
Cauxetal. 1996
Versteeg 1990
Hoberg 1991 A
Turbaketal. 1986
Roberts etal. 1990
Gala and Giesy 1990
Kallqvist and Romstad 1994
Kallqvist and Romstad 1994
van der Heever and Grobbelaar 1996
van der Heever and Grobbelaar 1997
Parrishl978
Laisenetal. 1986
Laisenetal. 1986
Laisenetal. 1986
CHROMALVEOLATA (includes tested diatoms, cryptomonads)
Cryptomonas
Cyclotella
Navicula
494
462
100
114
225
217
1.15
1.22
0.67
0.65
1.00
1.08
1.03
Kallqvist and Romstad 1994
Kallqvist and Romstad 1994
Millie and Heish 1987
Millie and Heish 1987
Millie and Heish 1987
Hughes etal. 1988
CYANOBACTERIA (includes tested blue-green algae)
Anabaena
Microcystis
Synechococcus
70
280
470
706
286
164
605
136
0.59
1.25
0.77
0.59
0.76
0.55
Stiattonl984
Stiattonl984
Stiattonl984
Hughes etal. 1988
Laisenetal. 1986
Pairishl978
Kallqvist and Romstad 1994
Kallqvist and Romstad 1994
ANGIOSPERMAE (includes tested vascular plants)
Ceratophyllum
Elodea
Hydrilla
Lemna
Myriophyllum
Najas sp.
Potamogeton
Vallisneria
24
65
<38
204
118
202
93
49
115
224
95
90
<150
15
63
141
0.81
0.38
0.52
0.99
1.24
1.33
1.71
0.42
1.14
1.18
1.67
0.69
0.40
0.04
0.07
0.02
0.09
0.24
0.25
0.23
0.21
0.21
0.40
0.02
Faiichild et al. 1998
Foiney and Davis 1981
Faiichild et al. 1998
Hobeig 2007
Hinman 1989
Hobeig 199 IB
Hobeig 1993B
Hobeig 1993C
Faiichild et al. 1998
Hughes etal 1988
Kiiby and Sheehan 1 994
Desjaidin 2003
Faiichild et al. 1998
Faiichild et al. 1998
Foiney and Davis 1981
Foiney and Davis 1981
16
-------
Although not included in the compilation because they were conducted in estuarine water
near 10 ppt salinity, studies on Myriophyllum spicatum and Potamogeton perfoliatus by Kemp et
al. (1985) and Jones et al. (1986) are consistent with the vascular plant results in Table 1. For
both these species, oxygen production-based reductions in photosynthesis (Kemp et al. 1985)
indicated ECSOs to be near or below 50 |J,g/L in the first two weeks of exposure (although some
lessening of these effects was apparent in the ensuing two weeks). For Potamogeton perfoliatus,
radiocarbon fixation-based reductions in photosynthesis (Jones et al. 1986) indicated the ECso to
be between 50 ng/L and 100 |ig/L.
485
486
487
488
489
490
491
492
493 2.4 Statistical Distribution of Toxicity Relationship Parameters
494 The SGR ECso data in Table 1 were logic transformed and subject to an analysis of
495 variance (ANOVA) using the general linear model (GLM) procedure of Statistica (Version 8.0,
496 StatSoft, Tulsa, OK, USA). A nested ANOVA showed no significant differences between
497 genera within the larger taxonomic groups identified in Table 1, so the analysis was simplified to
498 a one-way ANOVA on these taxonomic groups, with each test result being treated equally
499 regardless of the number of tests within a species or genus. This analysis indicated significant
500 differences among the taxonomic groups, with the mean logio(ECso) being 2.09 for green algae,
501 2.35 for diatoms/cryptomonads, 2.42 for blue-green algae, and 1 .93 for vascular plants (Table 2).
502 These log values correspond, respectively, to median ECsoS of 123, 224, 263, and 85 |ig/L.
503 However, it should be noted that these taxonomic differences are uncertain due to the limited
504 amount of data for some of the taxa - the standard errors of these mean logio(EC5o)s varied from
505 0.07 to 0. 12 (Table 2), depending on the number of observations for each group, and their 95%
506 confidence limits overlapped. The within-group variability did not differ significantly between
507 the taxonomic groups, with the within-group standard deviation ranging from 0.29 to 0.35 (Table
508 2) and the pooled value being 0.33. The overall, unweighted mean and standard deviation of all
509 logio(ECso)s were 2. 12 and 0.37 (this higher standard deviation being reflective of the intergroup
510 variability). Basing the analysis on genus means rather than individual tests produced similar
511 values for the overall mean (2.07) and standard deviation (0.35) of logio(EC5o)s.
512 The steepness parameter (Steep) data in Table 1 were also logic transformed and subject
513 to ANOVA. The ANOVAs showed no significance differences either between genera or the
Table 2. Summary statistics for SGR-based toxicity relationships from Table 1 (based on individual tests within
designated taxonomic group).
Taxonomic
Group
Green Algae
Diatoms/Cryptomonads
Blue-green Algae
Vascular Plants
Overall
log(EC50)
Mean
2.09
2.35
2.42
1.93
2.12
Std. Dev.
0.33
0.29
0.35
0.34
0.37
Std. Err.
of Mean
0.07
0.12
0.12
0.09
0.06
log(Steep)
Mean
-0.03
-0.03
-0.12
-0.07
-0.05
Std. Dev.
0.17
0.12
0.15
0.23
0.18
Std. Err.
of Mean
0.04
0.05
0.07
0.06
0.03
17
-------
514 broader taxonomic groups. The within-group means ranged from -0.03 for the green algae and
515 diatoms to -0.11 for the blue-green algae, with an overall mean of -0.05 (Table 2). The steepness
516 distribution is therefore described here based simply on this overall mean for logio(Steep)
517 (corresponding to a median value for Steep of 0.89) and the overall observed standard deviation
518 (0.18) (Table 2). Using genus means rather than individual observations resulted in a very
519 similar log mean (-0.08) and standard deviation (0.16). A correlation analysis also showed no
520 significant correlation between logio(ECso) and logio(Steep), so these parameters will be treated
521 independently in any analyses.
522 2.5 Uncertainty of PATI Relationships
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
The toxicity data analyses here provide the basis for computing an overall measure of
toxic impact on an assemblage of plant species (i.e., PATI) as a function of concentration.
However, this does involve some issues regarding data selection and processing that will be
relevant to uncertainty analyses presented in Section 4 of this document.
One issue is whether PATI should be calculated directly from the individual tests in
Table 1 (using the overall median steepness for any test without a measured steepness) or be
based on the overall distributions of logio(ECso) and logio(Steep) summarized in Table 2. For the
individual tests, calculating PATI is simply a matter of averaging the toxicity relationships across
all the tests. For the summary distributions, calculating PATI requires multiplying the level of
toxic effect expected for a particular ECso and Steep by the probability density for that
combination of ECso and Steep, and doing this for all possible combinations of ECso and Steep.
Mathematically, this can be expressed as follows, where the function "tox" (the expected toxicity
at exposure concentration C and for toxicity parameters ECso and Steep) is multiplied by the
function "dens " (the density function for the joint probability distribution of ECso and Steep), and
this product is then integrated across all values of ECso and Steep.
PATI = \ \tox(C,EC'50,Steep)-dens(EC'50,Steep)dSteep dEC50 (Equation!)
Rather than evaluating this by numerical integration, it was estimated by randomly sampling
10000 pairs of ECso and Steep from the density function (assumed to be bivariate log normal
with means and standard deviations as in Table 2), applying the toxic relationship function (Eq.
1) to these random pairs, and taking the mean
of these toxicity values. Based on repeated
tests of this process, 10000 points were
sufficient to evaluate this integral with a
relative error of <0.5%.
Figure 8 provides a comparison of
these two calculations methods, showing a
negligible difference for concentrations
>10 ng/L, with the difference growing to
about 30% at 2 |ig/L and a PATI value of
ca. 1. This calculation method issue would
thus appear not to be a significant uncertainty
source, but its impact on risk characterization
will be examined in Section 4.
Figure 8. PATI relationships based on the toxicity
relationships for individual tests (dashed line) versus
based on the overall summary distribution of the
relationship parameters EC50 and Steep (solid line).
gioo-
x
-g 50-
_c
>,
:! 20 -
x
o
5 -
n- 1
10 20 50 100 200
Atrazine Concentration (|ag/L)
500 1000
18
-------
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
Another issue is the uncertainty associated with PATI relationships because of the finite
number of toxicity relationships used in its formulation. This uncertainty is reflected in the
standard errors for the means of the toxicity relationship parameters (logio(EC50), logio(Steep))
reported in Table 2, as well as the uncertainty
in the parameter standard deviations. Figure
9 shows how PATI based on the overall
distribution in Table 2 would vary by
changing the mean and standard deviations of
the parameters to their lower and upper 95%
confidence limits. At most concentrations,
the largest effects are for the uncertainties in
the mean logio(ECSO), but the other
uncertainties become substantial at lower
concentrations, with the uncertainty in PATI
due to the mean logio(Steep) reaching a factor
of approximately 2.0 at 2 |ig/L and a PATI
value of ca. 1. The impact of this uncertainty
on risk characterizations will also be
considered in Section 4.
Figure 9. Comparison of best estimate of overall PATI
relationship (solid line), to 95% confidence limits for
the mean for logEC50 (short dash) and logSteep (dash-
dotted) and standard deviations for logEC50 (long
dash) and logSteep (dash-double dotted).
100 -i
50 -
." 20 -
10 -
5 •
2 -
10 20 50 100 200
Atrazine Concentration (ua/L)
500 1000
A third issue is that the PATI relationships in Figures 8 and 9 represent an assemblage of
plant species and tests defined by the available test data, but different assemblages are possible
by selecting or weighting particular taxa.
Figure 10 contrasts PATI relationships based
(a) on the overall distributions of logioCECso)
and logio(Steep) in Table 2, (b) the separate
distributions in Table 2 for the four major
taxa, and (c) a composite distribution based on
equal weighting of the four major taxa (in
contrast to the overall distribution, which is
unweighted across all tests regardless of the
major tax on). The PATI values for the overall
distribution and the composite distribution
have negligible differences, but the PATI
relationships for the major taxa can differ
substantially from each other due to the
apparent differences in their relative
sensitivities.
Figure 10. Comparison of PATI based on the overall
toxicity distribution (solid line) to distributions for
green algae(short dash), diatoms(long dash), blue-
green algae(dash-dotted), vascular plants(dash-double
dotted), and a composite of the four taxa (dotted).
gioo-
x
-g 50-
_c
>,
;§ 20 -
x
o
^ 10-
5 -
n- 1
10 20 50 100 200
Atrazine Concentration (|ag/L)
500 1000
Because vascular plants have the lowest estimated mean logioCECso) (i.e, the greatest
average sensitivity), they have the highest PATI values in Figure 10. At 2, 5, and 10 |ig/L
atrazine, the estimated PATI values are, respectively, 2.2-, 1.9-, and 1.7-fold larger than for the
overall distribution. Only 5.5 jig atrazine/L is needed to reach a PATI value of 5%, versus 10
|ig/L for the overall distribution.
19
-------
598 Even greater differences occur for the diatom/cryptophyte group, which has markedly
599 lower PATI values at low atrazine concentrations because of a combination of a larger-than-
600 average mean and a smaller-than-average standard deviation for logio(EC5o). At 2, 5, and 10
601 |ig/L atrazine, the estimated PATI values are, respectively, 4.4-, 3.6-, and 3.1-fold smaller than
602 for the overall distribution. Almost 25 jig/L atrazine is needed to reach a PATI value of 5%,
603 versus 10 |ig/L for the overall distribution.
604 The effects of these plant assemblage differences on risk characterization also will be
605 examined in Section 4. However, it should be noted here that, because PATI is intended to serve
606 as a relative index of the effects of different exposure concentrations, the slopes of the
607 relationships in Figure 10, not the absolute PATI values, will determine how risk
608 characterizations depend on the taxonomy of the assemblage. Although the estimated PATI
609 values for the vascular and diatom groups differ by nearly an order of magnitude at low atrazine
610 concentrations, the log slopes in Figure 10 are not very different from each other (e.g., the
611 relative changes in PATI from 10 to 20 jig atrazine/L are 1.9, 2.1, and 2.4 for the vascular plant,
612 overall, and diatom distributions, respectively). Thus, it should be anticipated that the analyses
613 in Section 4 will show limited sensitivity of risk characterizations to assemblage taxonomy.
614
20
-------
615 3. USING EXPERIMENTAL ECOSYSTEM DATA TO SPECIFY THE LOC FOR PATI
616
617
618
619
620
621
622
623
624
625
626
627
628
Using the experimental ecosystem data to determine an LOC for the cumulative PATI
involves relating a binary response (yes/no effect for each experimental ecosystem treatment) to
a quantitative measure for the severity of the exposure (cumulative PATI). Before presenting
this process, it would be useful to first discuss a similar but more familiar analysis.
Mortality in a toxicity test also involves a binary response - an individual organism either
dies or not. Mortality data is often plotted as the fraction of a group of organisms that died (by
an observation time) vs. the concentration to which the group was exposed, shown in the left
panel of Figure 11. However, such data can also be plotted based on the response of each
individual organism (0 if alive, 1 if dead), shown on the right panel of Figure 11, in which offsets
are used to show points that actually have the same concentrations. Probit analysis is a common
method applied to such data to generate a sigmoidal relationship for the probability of mortality
at each concentration, this relationship being the same in the left and right panels because both
panels represent the same information and analysis.
Figure 11. Probit analysis as an example of binary data analysis. For a hypothetical toxicity test, the left panel
shows the fraction (of 10 organisms) which died at each concentration while the right panel plots individual
organism response as 0 (if survived) and 1 (if dead). Lines denote probit relationship for probability of death.
1.0 -,
CO °-8
•c
o
^ 0.6 -
it-
CD
i= 0.4
CO
"§ 0.2 -
0.0
1.0 -,
0.8 -
0.6 -
0.4 -
0.2 -
0.0
0.1 0.2
0.5 1 2
Concentration
10 0.1 0.2
0.5 1 2
Concentration
10
629
630
631
632
633
634
635
636
637
638
639
640
641
642
Because probit analysis uses the binary response of the individual organisms as the basic
observation, it is actually more directly related to the right panel of Figure 11 than to the left.
Furthermore, if individual organisms all have different exposures, the presentation format of the
left panel cannot be used (i.e., there are no groups of replicate organisms upon which to compute
fraction survival), but a plot such as in the right panel can still be done and probit analysis is still
appropriate. For example, if the offsets for the points in the right panel of Figure 11 actually
represented different concentrations, probit analysis could still be applied even without replicate
points at the same concentration.
The experimental ecosystem data provide an analysis situation analogous to the survival
data in the right panel of Figure 11. Figure 12 replots the experimental ecosystem data from
Figure 2 as binary effects (1 if there is an effect, 0 if there is not) vs. a PATIeod value. (For the
purpose of this example, the overall distribution of toxicity values in Table 2 was used as the
basis for PATI, along with the 60-d assessment period. The basis for these choices is addressed
in Section 4.)
21
-------
Figure 12. Experimental ecosystem data plotted as effect/no effect versus PATI90d, fitted to a logistic relationship
for the probability of an effect versus PATI.
o.o
LOCPATI=132
W-^?—W-ST-WW—W5F—W
10
20
50
100
200
500 1000 2000
5000
60-d Cumulative Plant Assemblage Toxicity Index (%-days)
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
Although there is a clear increase in the probability of effects as PATI6od increases in
Figure 12, there also is considerable overlap between effects and no effects with respect to
PATIeod, especially in the 100 to 200 range for PATIeod This variability/overlap issue was
already noted regarding Figure 2, and should be viewed here in terms of any particular PATIeod
value having a probability of eliciting an effect across the variety of experimental ecosystem
studies used here. That there is a probability, rather than a certainty, of having an adverse effect
at any PATI6od value is again indicative of sensitivity differences among the systems and/or
various experimental uncertainties. Across all PATIeod values, there would be an underlying
relationship for this probability, illustrated by the curve on Figure 12.
This probability relationship can be quantified using probit or similar binary analyses.
Field et al. (1999, 2002) applied binary analysis to sediment toxicity assessments of a similar
nature (i.e., relating binary effect data to an exposure concentration), but rather than the Gaussian
distribution-based relationship of probit analysis, they applied a similar, but simpler, probability
relationship based on the logistic equation. For describing the probability of effects in the
experimental ecosystem set as a function of PATIeod, this logistic probability expression can be
formulated as:
1
1
PATI5m
1 + 10"
(Equation 3)
where P is the probability (percent scale) of an adverse effect at a PATI6od value, PATI50% is the
PATI6od value at which P=0.5 (50% chance of an effect over the range of experimental
ecosystems), and S is a steepness parameter (>0) for the relationship.
Although P is the underlying probability of an actual adverse effect, this equation is not
appropriate for analyzing the data in Figure 12 because it does not reflect certain errors in the
statistical analysis regarding whether an experimental ecosystem treatment is concluded to have
22
-------
666 an adverse effect. Most importantly, Type I error (the probability of concluding a treatment has
667 an effect when it actually does not) is typically set at 0.05. This means that, although the actual
668 probability of an adverse effect approaches zero as PATI approaches zero per Equation 1, the
669 probability of stating that there is an effect does not approach zero, but rather approaches 0.05.
670 Type II error (the probability of concluding a treatment does not have an effect when it actually
671 does) will also affect the curve, but it is not possible to adjust for this without more detailed
672 information on the statistical power of the various tests. However, because Type II error will go
673 to zero as concentration increases, it will not affect the upper asymptote of the curve like Type I
674 error affects the lower asymptote, and thus will not overtly affect the basic sigmoidal shape of
675 the curve being fitted. The binary regression used in the LOG methodology will therefore use a
676 logistic model with a lower asymptote of 0.05, modifying Equation 2 as follows:
1 + 0.05- PAn™/PAn S
677 Pdata = - 7 (Equation 4)
data ,
1 _|
1 \
678 where Pdata refers to the probability of a data point being stated to have an effect, in contrast to P
679 being the actual probability of having an effect.
680 Using Equation 4, a maximum likelihood analysis was conducted on the data in Figure 10
681 to generate estimates for the equation parameters, PATIso% and S. Using these parameter
682 estimates, the curve in Figure 12 was calculated, but using Equation 3 rather than Equation 4 so
683 the curve shows the actual estimated P, not Pdata- Once estimated, this curve provides a basis for
684 making risk management decisions regarding what PATI value is considered an LOG. For
685 example, for Figure 12, a risk management decision to use P=0.5 would result in an LOCpAiieod
686 of!32%-days.
687 This LOCpAiieod of 132 %-days represents substantial reductions in growth rate for this
688 plant assemblage for short exposures (e.g., 44% for a three day exposure), but progressively
689 smaller effects for longer exposures (e.g., 10% for two weeks, 5% for four weeks). However, it
690 is important to remember these percentages do not define the level of protection; rather, it is
691 the experimental ecosystem results that define the effects of concern! PATI is only being used
692 to describe the relative effects of different exposure time-series. It is the experimental ecosystem
693 effects that define the effects of concern and what level of PATI for the selected assemblage of
694 toxicity data correlates with these effects. It is not being assumed that a certain value for PATI
695 has inherent significance, so it is not appropriate considering (under the assessment framework
696 being used here) whether reducing growth rate by 44% for three days is too restrictive or not
697 restrictive enough. This PATI-based methodology assumes only that the relative effects of
698 concentration and time on PATI are useful for extrapolating between different exposure time-
699 series for the experimental and natural plant communities being assessed.
700
23
-------
701
702
703
704
705
706
707
708
4. IMPLEMENTATION OF PATI-BASED RISK ASSESSMENT METHODOLOGY
4.1 Example Field Exposure Time-Series
Figure 1 provided three example field exposure time-series (chemographs) for use in the
problem definition. In this section, method parameterization and performance evaluations will
involve a larger set (Figure 13) of chemographs from the 2010 monitoring program to provide a
greater diversity of exposures for evaluating the methodology. EEFs for all chemographs will be
presented, but because uncertainties are most important for EEFs near 1.0, summary statistics for
sensitivity and uncertainty analyses below are based only on sites with 0.3gy evaluations.
IL 11 2010 16.
12 •
8 -
0 90 120 150 180 210 240 6
100,
KS 01 2010 80 .
0 90 120 150 180 210 240 6
100 -i
MO 02 2010 80 .
0 90 120 150 180 210 240 6
50 •
. NE 05 2010 40.
1 30-
1 1 2°'
0 90 120 150 180 210 240 6
50 -
IL 122010 40.
30 .
0 90 120 150 180 210 240 6
KS 02 2010 4°'
IIUV^L 1°'
0 90 120 150 180 210 240 6
50,
. MO04a2010 40 .
30
1 L 20-
ML.
0 90 120 150 180 210 240 6
50 -
OH 05 2010 40-
ll :::
JLIL 1°-
IL 132010
0 90 120 150 180 210 240
KS 03 2010
A*.
0 90 120 150 180 210 240
i MO 05 2010
Mu.
0 90 120 150 180 210 240
OH 06 2010
0 90 120 150 180 210 240
IL 142010
Ik
0 90 120 150 180 210 240
LA 04 2010
0 90 120 150 180 210 240
MO05B2010
kL
0 90 120 150 180 210 240
NE 09 2010
0 90 1 20 1 50 1 80 210 240 60 90 120 1 50 1 80 21 0 240 60 90 1 20 1 50 1 80 21 0 240 60 90 1 20 1 50 1 80 21 0 240
Julian Day
24
-------
709 4.2 Parameterization Issues for PATI-based LOCs
710 Implementing a PATI-based methodology requires specifying (a) the toxicity relationship
711 parameters (ECsos and Steeps) to use in daily PATI calculations and (b) the assessment period
712 over which to evaluate cumulative PATI.
713 4.2.1 Assessment Period - Issues and Options
714 Because exposure outside the assessment period is considered inconsequential by PATI,
715 this period needs to be long enough to encompass (a) exposures of significance to establishing
716 LOCpAii from the experimental ecosystems (Figure 2) and (b) effects expected from seasonal
717 field exposures (Figure 13). However, it should not be any longer than necessary, in order to
718 avoid uncertain inferences regarding (a) cumulative effects of low concentrations and (b) widely
719 separated exposures that are independent regarding ecological effects.
720 A 60-d assessment period was chosen as a provisional focus for consideration because it
721 would include all or almost all periods of significant exposure in the example chemographs of
722 Figure 13 and also encompasses the duration of all but a few of the experimental ecosystems in
723 Figure 8. A few additional considerations regarding this period relative to the experimental
724 ecosystem treatments should be noted:
725 (1) It is just slightly shorter than the longest experimental ecosystem treatment with no effect. If
726 the assessment period was significantly shorter than treatments with no effect, this would under-
727 represent how substantial exposures could be without causing effects and thus be too restrictive.
728 (2) For those treatments with effects, a shorter period would also be too restrictive by assuming
729 that less exposure was needed to elicit effects than actually was involved (e.g., an effect observed
730 over a 60-d exposure would be assumed to require less exposure than actually was required).
731 This consideration does not pertain to the few experimental ecosystems with extremely long
732 durations, because they simply verify significant effects for high PATI values. For the LOG, the
733 important treatments with effects are those whose exposures near to those without effects.
734 (3) That 60 d is longer than many experimental ecosystem treatments with effects is not an issue,
735 provided the effects from these shorter exposures would still be considered unacceptable from
736 the perspective of this longer assessment period (e.g., if a 30-d exposure showing effects had
737 been monitored for another 30 d without exposure, the effects during the first 30 d would be
738 considered unacceptable despite any recovery that occurred during the second 30 d).
739 To evaluate the suitability of 60 d as the assessment period, compared to possible
740 alternative choices, sensitivity analyses below will address how risk characterizations would
741 differ for assessment periods from 30-d to 120-d. A 30-d assessment period is included in this
742 sensitivity analysis to document the impact of a period that is arguably too short, in that it is less
743 than the duration of a substantial percentage of the experimental ecosystems treatments that
744 discriminate effects and no effects, and also inadequately covers periods of substantial exposure
745 in the example chemographs.
746
25
-------
747 4.2.2 Toxicity Relationship Parameters - Issues and Options
748 The review and analysis of single-species toxicity test data in Sections 2.2 and 2.3
749 provide the basis for specifying toxicity relationships for PATI calculations, but there are options
750 and uncertainties in applying this information, which were already discussed to some extent in
751 Section 2.4:
752 (a) Should PATI calculations be directly based on the discrete estimates for the toxicity
753 relationship parameters (ECso and Steep) in Table 1, or should the methodology follow the
754 typical assessment practice of using the data to estimate sensitivity distributions (Table 2), and
755 basing assessments on such distributions?
756 (b) Should the methodology be weighted in some manner for taxonomic groups, or follow
757 standard practice (e.g., typical SSDs) of not adjusting for the relative representation of different
758 taxa in the available data?
759 (c) Should calculations be based on average results for each species or genus, or on individual
760 tests?
761 The strategy here was to use, as a default reference, distributions based on all the
762 available, individual toxicity observations (i.e., the "overall" distributions of logio(ECso) and
763 logio(Steep) summarized in Table 2). Sensitivity analyses were conducted to determine how
764 substantially risk characterizations varied for alternatives from this default, including (a) the use
765 of discrete parameter estimates in Table 1 instead of these default distributions (as was done for
766 Figure 8), (b) different weightings of the major taxonomic groups (such as in Figure 10), and (c)
767 basing distributions on genus means instead of individual test results. Based on this sensitivity
768 analysis, decisions can be made regarding how these issues should be addressed in the final
769 methodology.
770 4.3 Sensitivity Analyses for PATI-Based LOCs
771 4.3.1 Sensitivity Analysis for Assessment Period
772 Using the overall (default) toxicity parameter distributions specified in Table 2, effects
773 assessments were made for each of the example chemographs in Figure 13, using assessment
774 periods of 30, 60, 90, and 120 d. These assessments proceeded as follows:
775 (a) The daily PATI values for each experimental ecosystem treatment were calculated. As
776 illustrated in Figure 3, this involves computing, for each daily exposure concentration, an
777 average effect across a set of toxicity relationships. Because the toxicity relationship parameters
778 are represented by distributions, this calculation was conducted as described in Section 2.5.
779 (b) The daily PATI values were used to calculate cumulative PATI values for 30-, 60-, 90-, and
780 120-d assessment periods for each experimental ecosystem treatment. When the exposure
781 duration exceeded the assessment period, the contiguous period of exposure resulting in the
782 highest cumulative PATI value was used.
26
-------
783 (c) For each assessment period, a binary logistic regression was conducted as described in
784 Section 3.2. The LOCpAii was set to the PATl5o% estimate from this regression (50% probability
785 of an effect).
786 (d) Daily PATI values were computed for each of the example chemograph in Figure 11.
787 Cumulative PATI values for each assessment period were calculated for the contiguous period of
788 exposure resulting in the highest value.
789 (e) For each assessment period and example chemograph, risk was characterized by calculating
790 the EEF and CEF (see Figure 3 and associated text for definition of these terms).
791 Figure 14 illustrates how the assessment period affects risk characterization, as
792 represented by the EEF. (CEFs showed patterns very close to the EEFs and are not included
793 here.) Relative to the proposed assessment period of 60 d, increasing the assessment period to 90
794 or 120 d resulted in small increases in the EEF, except for one site (MO 02) for which the
795 increases were 28-29%. For the other sites, EEFs increased by an average of 5.6% (range 2.7%-
796 11.0%) for the 90 d assessment period and 8.6% (range 1.6-17.4%) for the 120 d assessment
797 period. In contrast, using a 30-d assessment period reduced the EEF, relative to 60 d, by a mean
798 of 24% (range 2-40%), the larger reductions being associated with sites with substantial
799 exposures for more than 30 d. Using such a short averaging period poorly addresses
800 experimental ecosystem treatment effects, but more importantly assumes that major portions of
801 many field exposures should be ignored.
Figure 14. Sensitivity of risk characterization to assessment period length, based on effects exceedence factors at
20 selected sites monitored in 2010.
10 n
5 -
2 -
1 •
o
"o
CD
LJ_
8
-------
805
806
807
808
809
810
(1) The overall distributions for logioCECso) and logio(Steep) reported in Table 2 (default).
(2) The individual logioCECso) distributions for the four major taxonomic groups in Table 2
(using the overall distribution for logio(Steep)).
(3) An equal -weighted composite of the logio(ECso) distributions for the four taxonomic groups.
(4) The individual tests in Table 1 (using the average value of -0.05 for logio(Steep) for tests in
which this was not determined).
811 (5) The overall distribution using genera means rather than individual tests (Section 2.3).
812 These evaluations were conducted in accord with the protocol described above for the
813 assessment period evaluations and are summarized in Figure 15. For most options (green algae,
814 bluegreen algae, individual tests, composite taxa, genus means), the EEF deviations from the
815 default option were generally negligible, averaging <3.5% and never exceeding 13%. For the
816 diatom distribution (the least sensitive group at low atrazine concentrations per Figure 10), EEFs
817 usually are lower than for the default option - averaging 14% lower and ranging from 37% lower
818 to 22% higher. For the vascular plant distribution (the most sensitive group), EEFs usually are
819 higher than for the default option - averaging 12% higher and ranging from 3% lower to 33%
820 higher. Given the magnitude of the differences in mean logio(ECso), these differences are rather
821 small, and also are not statistically significant given the uncertainties in the toxicity data.
822 This small sensitivity of EEFs to changes in the toxicity information used in PATI might
823 seem surprising given the large sensitivity of PATI itself to these changes (Figure 10), but this is
824 because the experimental ecosystems, not the toxicity distributions, determine the level of
Figure 15. Sensitivity of risk characterization to selection of toxicity data, based on effects exceedence factors at
20 selected sites monitored in 2010
10 n
Green Algae
Diatoms
Bluegreen Algae
Vascular Plants
Composite Taxa
Individual Tests
Genus Means
Oveerall Distribution
28
-------
825 concern. PATI is only being used to assess the relative effects between different exposure times-
826 series, and these relative effects are similar whether the plant assemblage is sensitive or tolerant.
827 As noted in Section 2.5, these relative effects are related to the slopes in Figure 10, which differ
828 little among the various taxonomic assemblage definitions compared to the large variation in the
829 absolute PATI values. From another perspective, using a more sensitive set of toxicity data will
830 result in higher PATI values for both the experimental ecosystem treatments and the field
831 exposures, so that the net effect of taxonomy on the EEFs is much less than that on PATI itself.
832 However, there are still some effects of taxonomy on EEFs because PATI is not linear
833 with concentration. The smaller slopes in Figure 10 for the vascular plants than the diatoms
834 mean that the lower atrazine concentrations will contribute relatively more to the vascular plant-
835 based PATI than the diatom-based PATI. And because periods of relatively lower concentration
836 are more prevalent in most field exposures than in most experimental ecosystem treatments, this
837 results in slightly higher EEFs for the vascular plant-based PATI than the diatom-based PATI.
838 However, these differences are small for any field exposure with an EEF near 1.0 and thus have
839 negligible effect on risk characterizations (Figure 15) despite the substantial differences in
840 absolute PATI values.
841 Because this sensitivity analysis shows such small effects from even extreme choices for
842 the taxonomic composition of the plant assemblage and because of the statistical uncertainties of
843 these effects, the recommendation here is to use the overall toxicity distribution in Table 2 that
844 was used as the default for this analysis. Using all the data, rather than a subset, is also more in
845 keeping with how aquatic risk assessments generally reflect a broad assemblage of organisms.
846 4.4 Contribution of Toxicity Distribution Uncertainty to Overall Assessment Uncertainty
847 Although varying the assemblage taxonomy in Section 4.3.2 did not affect risk
848 characterizations enough to support using something other than the overall parameter
849 distributions, this does not mean that uncertainty in these distributions is negligible. More
850 evaluation was needed of the uncertainty of EEFs as a function of the uncertainties of all the
851 parameters for the toxicity relationships used to calculate PATI.
852 To this end, an uncertainty assessment was conducted that involved (a) generating 10000
853 sets of toxicity parameter distributions (means and standard deviations for both logio(EC5o) and
854 logio(Steep)), (b) determining the LOCpAii for each parameter distribution set, and (c)
855 determining the EEF for each example chemograph for each parameter distribution set. The
856 means of the 10000 distributions for logio(EC5o) and logio(Steep) were generated by random
857 sampling from normal distributions with the overall distribution means and standard errors for
858 these parameters in Table 2. The standard deviations of the 10000 distributions for logio(ECso)
859 and logio(Steep) were generated by random sampling from chi-square distributions based on the
860 overall distribution standard deviations for these parameters in Table 2, using a degree of
861 freedom based on the number of data in Table 1. Due to the observed lack of correlation
862 between ECso and Steep, the sampling for these two parameteers was done independently.
863 Figure 16 summarizes this uncertainty analysis, comparing the 10* and 90 uncertainty
864 percentiles to the median results. The lower bound for the EEF varies from 85% to 98% of the
29
-------
Figure 16. Uncertainty analysis for risk characterizations due to uncertainties intoxicity distributions used to
parameterize PATI. Solid line denotes EEFs based on best estimates of toxicity parameter distributions. Dashed
lines denote 10th and 90th percentiles due to uncertainty of these distributions.
10-1
03
LL
0
O
5-
2-
1 •
0
T3
0
S0.5
X
LLJ
t) 0.2-
!t
LJJ 0.1
865
866
867
868
869
870
871
872
873
median among the chemographs, with an average of 95%, while the upper bound varies from
102% to 120% of the median, with an average of 107%.
Although this demonstrates that uncertainties in the toxicity data used to parameterize
PATI result in very little uncertainty in the final risk characterizations, this is only one
component of the uncertainty for the total methodology. If uncertainty estimates are to be
provided, they would need to reflect all important sources2, compared to which these
uncertainties for the toxicity distributions used by PATI should be relatively minor.
2An example of another source of error in the overall methodology is the uncertainty in the log(LOCPATi) from the
logistic regression. When the best estimates of the overall toxicity distributions are used in calculating PATI, the
standard error for log(PATI50o/0) is 0.16 from the binary regression analysis, which produces a 10th to 90th percentile
range for the CEF of 55-183% of the median. Other sources of uncertainty include the characterization of field
exposures and of experimental ecosystem effects.
30
-------
874 5. SUMMARY AND RECOMMENDATIONS REGARDING LOC METHODOLOGY
875 As noted in Section 1, this LOC methodology starts with experimental ecosystem studies
876 regarding effects of atrazine on aquatic plant communities. Each experimental ecosystem
877 treatment must be characterized regarding (a) whether there is an unacceptably adverse effect
878 and (b) the atrazine concentration time series. This characterization was provided in U.S.EPA
879 (2011) and summarized in Appendix B. The basic problem addressed here is the issue of
880 comparing effects across different exposure time series, both among the experimental
881 ecosystems and between the experimental ecosystems and exposures of interest in natural
882 systems. This is done with an effects index that specifies the relative toxic severity of different
883 time-series. The proposal here is that this index be the 60-d cumulative PATI value. This index
884 is applied as follows:
885 (1) Based on available toxicity tests with individual aquatic plant species, relationships of SGR
886 versus atrazine concentration are developed and used to specify statistical distributions for the
887 relationship parameters (ECso, Steep). For this report, the tests were described using a logistic
888 relationship of SGR versus log atrazine concentration, and the distributional recommendations
889 were for the logio(ECso) to have a mean 2.12 and standard deviation 0.37 and for the logio(Steep)
890 to have a mean of -0.05 and a standard deviation of 0.18, based on an unweighted analysis across
891 all tests. Although some differences among major taxa were indicated, alternative distributions
892 using different taxonomic weightings had small and uncertain effects on assessment results. The
893 distributions recommended here merit additional evaluation regarding the toxicity test data set
894 used and the distributional shape and composition.
895 (2) The relationship of daily PATI values to atrazine concentration should be developed for the
896 assemblage of species described by the distributions for the toxicity relationship parameters
897 (ECso, Steep). This requires integrating the expected toxic response across the joint distribution
898 of the parameters; this integration is best done by randomly selecting a large number (e.g.,
899 10000) of EC50/Steep pairs from these distributions, determining the toxicity relationship for
900 each parameter pair, and averaging across all these relationships (note: this numerical method for
901 integrating across the distributions need only be done once and then applied to all subsequent
902 PATI calculations). For the distributions specified here, this results in the following relationship
Figure 17. Relationship of PATI to atrazine concentration.
10 20 50 100 200
Atrazine Concentration (i-ig/L)
500 1000
31
-------
903 of daily PATI values to atrazine concentration (Figure 17):
904
905
906
907
908
909
910
911
912
(3) Based on this relationship of daily PATI to atrazine concentration, a cumulative PATI value
(=the sum of the daily PATI values) is calculated for each experimental ecosystem exposure to
provide a measure for the total relative toxic impact of that exposure. This cumulative PATI
value must be limited to a time frame (assessment period) consistent with risk management goals
and the experimental ecosystem data, for which a provisional period of 60 d is proposed here.
The binary effects determinations for each exposure are plotted against the cumulative PATIeod
values, and a regression is performed to describe the probability of effect versus PATI. For the
daily PATI relationship and the experimental ecosystem dataset used here, this results in the
relationship already shown in Figure 12 and repeated in Figure 18:
Figure 18. Experimental ecosystem data plotted as effect/no effect versus PATI90d, fitted to a logistic relationship
for the probability of an effect versus PATI.
1.0 H
5 10 20 50 100 200 500 1000 2000
5000
60-d Cumulative Plant Assemblage Toxicity Index (%-days)
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
The above relationship describes the probability of effect versus PATI6od, using the logistic
equation, with equation parameters logio(ECso)=132 %-days and a steepness=2.03. If the ECso is
the designated level of concern, the LOCpAii is thus 132 %-days for a 60 d assessment period.
These particular values are contingent on the toxicity data set used for PATI, the experimental
ecosystem dataset, and a risk management decision regarding what probability of effect is of
concern, and thus would change if any of these factors is modified.
(4) This level of concern for PATI is applied to environmental data by calculating the cumulative
PATI for each environmental exposure time-series of interest. The effects exceedence factor
(EEF) (=ratio of PATIeodS calculated for field exposures of interest to the LOCpAii) is used to
determine whether the exposures exceed a level of concern. If desired, iterative calculations can
be used to determine the concentration exceedence factor (CEF) by which the exposure exceeds
a level of concern. FORTRAN-based computer programs and associated input files for this
implementation have been developed and are separately available from the author. PATI-based
EEFs for a suitable set of field exposures can be used to develop a concentration-based LOG to
apply to future exposures without needing to make actual PATI-based calculations, and this is a
subject of a separate effort.
32
-------
930 6. REFERENCES
931 Abou-Waly H, Abou-Setta MM, Nigg HN, Mallory LL. 1991. Growth response of freshwater
932 algae, Anabaena flos-aquae and Selenastrum capricornutum to atrazine and hexazinone
933 herbicides. Bull Environ Contam Toxicol 46:223-229.
934 Bartell SM, Campbell KR, Lovelock CM, Nair SK, Shaw JL. 2000. Characterizing aquatic
935 ecological risk from pesticides using a diquat dibromide case study. III. Ecological Process
936 Models. Environ. Toxicol. Chem. 19(5): 1441-1453.
937 Berard A, Benninghoff C. 2001. Pollution-induced community tolerance (PICT) and seasonal
938 variations in the sensitivity of phytoplankton to atrazine in nanocosms. Chemosphere 45:427-
939 437.
940 Berard A, Leboulanger C, Pelte T. 1999. Tolerance of Oscillatoria limnetica (Lemmermann) to
941 atrazine in natural phytoplankton populations and in pure culture: Influence of season and
942 temperature. Arch Environ Contam Toxicol 37:472-479.
943 Berard A, Pelte T, Druart J. 1999. Seasonal variations in the sensitivity of Lake Geneva
944 phytoplankton community structure to atrazine. Arch Hydrobiol 145:277-295.
945 Boger P, Schlue U. 1976. Long-term effects of herbicides on the photosynthetic apparatus I.
946 Influence of diuron, triazines and pyridazinones. Weed Res 16:149-154.
947 Boone MD, James SM. 2003. Interactions of an insecticide, herbicide, and natural stressors in
948 amphibian community mesocosms. Ecol Appl 13:829-841.
949 Brockway DL, Smith PD, Stancil FE. 1984. Fate and effects of atrazine on small aquatic
950 microcosms. Bull. Environ. Contam. Toxicol 32:345-353.
951 Burrell RE, Inniss WE, Mayfield CI. 1985. Detection and analysis of interactions between
952 atrazine and sodium pentachlorophenate with single and multiple algal-bacterial populations.
953 Arch Environ Contam Toxicol 14:167-177.
954 Carney EC. 1983. The effects of atrazine and grass carp on freshwater communities. Thesis.
955 University of Kansas, Lawrence, Kansas, USA.
956 Caux P-Y, Menard L, Kent RA. 1996. Comparative study on the effects of MCPA, butylate,
957 atrazine, and cyanazine on Selenastrum capricornutum. Environ Poll 92:219-225.
958 Davis DE. 1980. Effects of herbicides on submerged seed plants. Completion Report, Project A-
959 067-A1-A, Office of Water Research and Technology, Washington, DC, USA. NTIS-PB81-
960 103103.
961 Desjardin D, Krueger HO, Kendall TZ. 2003. Atrazine technical: A 14-Day static-renewal
962 toxicity test with duckweed (Lemna gibba G3) including a recovery phase. Report 528A-131 A.
963 Wildlife International, Ltd., Easton, MD, USA.
33
-------
964 deNoyelles F, Dewey SL, Huggins DG, Kettle WD. 1994. Aquatic mesocosms in ecological
965 effects testing: Detecting direct and indirect effects of pesticides. In: Aquatic mesocosm studies
966 in ecological risk assessment. Graney, RL, Kennedy JH, Rodgers JH (eds.). Lewis Publ., Boca
967 Raton, FL, USA. pp. 577-603.
968 deNoyelles F, Kettle WD. 1980. Herbicides in Kansas waters - evaluations of effects of
969 agricultural runoff and aquatic weed control on aquatic food chains. Contribution Number 219,
970 Kansas Water Resources Research Institute, University of Kansas, Lawrence, Kansas, USA.
971 deNoyelles F, Kettle WD. 1983. Site studies to determine the extent and potential impact of
972 herbicide contamination in Kansas waters. Contribution Number 239, Kansas Water Resources
973 Research Institute, University of Kansas, Lawrence, Kansas, USA.
974 deNoyelles F, Kettle WD. 1985. Experimental ponds for evaluating bioassay predictions. In:
975 Validation and predictability of laboratory methods for assessing the fate and effects of
976 contaminants in aquatic ecosystems. Boyle, TP (ed.). ASTM STP 865, American Society for
977 Testing and Materials, Philadelphia, PA, USA. pp. 91-103.
978 deNoyelles F, Kettle WD, Fromm CH, Moffett MF, Dewey SL. 1989. Use of experimental
979 ponds to assess the effects of a pesticide on the aquatic environment. In: Using mesocosms to
980 assess the aquatic ecological risk of pesticides: Theory and practice. Voshell, JR (ed.). Misc.
981 Publ. No. 75. Entomological Society of America, Lanham, MD, USA.
982 deNoyelles F, Kettle WD, Sinn DE. 1982. The responses of plankton communities in
983 experimental ponds to atrazine, the most heavily used pesticide in the United States. Ecol
984 63:1285-1293.
985 Detenbeck NE, Hermanutz R, Allen K, Swift MC. 1996. Fate and effects of the herbicide
986 atrazine in flow-through wetland mesocosms. Environ Toxicol Chem 15:937-946.
987 Dewey SL. 1986. Effects of the herbicide atrazine on aquatic insect community structure and
988 emergence. Ecol 67:148-162.
989 Diana SG, Resetarits WJ, Schaeffer DJ, Beckmen KB, Beasley VR. 2000. Effects of atrzine on
990 amphibian growth and survival in artificial aquatic communities. Environ Toxicol Chem
991 19:2961-2967.
992 Downing HF, Delorenzo ME, Fulton MH, Scott GI, Madden CJ, Kucklick JR. 2004. Effects of
993 the agricultural pesticides atrazine, chlorothalonil, and endosulfan on South Florida microbial
994 assemblages. Ecotoxicology 13:245-260.
995 Erickson RJ. 2009. Critique of Syngenta Corporation's Probabilistic Implementation of
996 CASMAiz for Ecological Risk Assessment of Atrazine. Internal Report, U.S. Environmental
997 Protection Agency, Mid-Continent Ecology Division, Duluth, MN, USA.
34
-------
998 Fairchild JF, Ruessler DS, Carlson AR. 1998. Comparative sensitivity of five species of
999 macrophytes and six species of algae to atrazine, metribuzin, alachlor, and metolachlor. Environ
1000 Toxicol Chem 17:1830-1834.
1001 Fairchild J, Ruessler S, Nelson M, Haveland P. 1994. An aquatic hazard assessment of four
1002 herbicides using six species of algae and five species of aquatic macrophytes. Report, Midwest
1003 Science Center, National Biological Service, Columbia, MO, USA.
1004 Fairchild JF, Ruessler DS, Haverland PS, Carlson AR. 1997. Comparative sensitivity of
1005 Selenastrum capricornutum and Lemna minor to sixteen herbicides. Arch Environ Contam
1006 Toxicol 32:353-357.
1007 Fairchild, JF, Ruessler SD, Lovely PA, Whites DA, Heine PR. 1995. An aquatic plant risk
1008 assessment of sixteen herbicides using toxicity tests with Selenastrum capricornutum and Lemna
1009 minor. Report, Midwest Science Center, National Biological Service, Columbia, MO, USA.
1010 Faust M, Altenburger R, Boedeker W, Grimme LH. 1993. Additive effects of herbicide
1011 combinations on aquatic non-target organisms. Sci Total Environ 113/114:941-951.
1012 Field LJ, MacDonald DD, Norton SB, Severn CG, Ingersoll CG. 1999. Evaluating sediment
1013 chemistry and toxicity data using logistic regression modeling. Environ Toxicol Chem 18:1311-
1014 1322.
1015 Field LJ, MacDonald DD, Norton, SB, Ingersoll CG, Severn CG, Smorong D, Lindskoof R.
1016 2002. Predicting amphipod toxicity from sediment chemistry using logistic regression models.
1017 Environ Toxicol Chem 21:1993 -2005.
1018 Forney DR. 1980. Effects of atrazine on Chesapeake aquatic plants. M.S. Thesis, Auburn
1019 University, Auburn, AL, USA.
1020 Hinman ML. 1989. Utility of rooted aquatic vascular plants for aquatic sediment hazard
1021 evaluation. Ph.D. Thesis, Memphis State University, Memphis, TN, USA.
1022 Forney DR, Davis DE. 1981. Effects of low concentrations of herbicides on submersed aquatic
1023 plants. Weed Sci 29:677-685.
1024 Gala WR, Giesy JP. 1990. Flow cytometric techniques to assess toxicity to algae. In: Landis
1025 WG, van der Schalie WH (eds). Aquatic Toxicology and Risk Assessment: 13th Volume.
1026 American Society for Testing and Materials, Philadelphia, PA, USA.
1027 Geyer H, Scheunert I, Korte F. 1985. The effects of organic environmental chemicals on the
1028 growth of the alga Scenedesmus subspicatus: a contribution to environmental biology.
1029 Chemosphere 14:1355-1369.
1030 Giddings, JM, Anderson TA, Hall Jr LW, Kendall RJ, Richards RP, Solomon KR, Williams
1031 WM. 2000. Aquatic Ecological Risk Assessment of Atrazine: A Tiered, Probabilistic Approach.
1032 Prepared by the Atrazine Ecological Risk Assessment Panel, ECORISK, Inc. Report 709-00,
1033 Novartis Crop Protection, Greensboro, NC, USA.
35
-------
1034 Gramlich JV, Grams RE. 1964. Kinetics ofChlorella inhibition by herbicides. Weeds 12:184-
1035 189.
1036 Gruessner B, Watzin MC. 1996. Response of aquatic communities from a Vermont stream to
1037 environmentally realistic atrazine exposure in laboratory microcosms. Environ Toxicol Chem
1038 15:410-419.
1039 Gustavson K, Wangberg SA. 1995. Tolerance induction and succession in microalgae
1040 communities exposed to copper and atrazine. Aquat Toxicol 32:283-302.
1041 Hamala JA, Kollig HP. 1985. The effects of atrazine on periphyton communities in controlled
1042 laboratory ecosystems. Chemosphere 14:1391-1408.
1043 Hamilton PB, Jackson GS, Kaushik NK, Solomon KR. 1987. The impact of atrazine on lake
1044 periphyton communities, including carbon uptake dynamics using track autoradiography.
1045 Environ Poll 46:83-103.
1046 Hamilton PB, Jackson GS, Kaushik NK, Solomon KR, Stephenson GL. 1988. The impact of two
1047 applications of atrazine on the plankton communities of in situ enclosures. Aquat Toxicol
1048 13:123-140.
1049 Hamilton PB, Lean DRS, Jackson GS, Kaushik NK, Solomon KR. 1989. The effect of two
1050 applications of atrazine on the water quality of freshwater enclosures. Environ Pollut 60:291-
1051 304.
1052 Hersh CM, Crumpton WG. 1989. Atrazine tolerance of algae isolated from two agricultural
1053 streams. Environ Toxicol Chem 8:327-332.
1054 Hinman ML. 1989. Utility of rooted aquatic vascular plants for aquatic sediment hazard
1055 evaluation. Ph.D. Thesis. Memphis State University, Memphis, TN, USA.
1056 Hoberg JR. 199la. Atrazine technical toxicity to the freshwater green algae, Selenastrum
1057 capricornutum. SLI Report #91-1-3600. Springborn Laboratories, Inc., Wareham, MA, USA.
1058 Hoberg JR. 1991b. Atrazine technical toxicity to the duckweed Lemna gibba. SLI Report #91-1-
1059 3613. Springborn Laboratories, Inc., Wareham, MA, USA.
1060 Hoberg JR. 1993a. Atrazine technical - toxicity to the freshwater green algae, Selenastrum
1061 capricornutum. SLI Report # 93-4-4751. Springborn Laboratories, Inc., Wareham, MA, USA.
1062 Hoberg JR. 1993b. Atrazine technical - toxicity to duckweed (Lemna gibba). SLI Report #93-
1063 4-4755. Springborn Laboratories, Inc., Wareham, MA, USA.
1064 Hoberg JR. 1993c. Atrazine technical - toxicity to duckweed (Lemna gibba). SLI Report #93-
1065 11-5053. Springborn Laboratories, Inc., Wareham, MA, USA.
36
-------
1066 Hoberg JR. 2007. The toxicity of atrazine to the freshwater macrophyte Elodea canadensis at
1067 three light intensities for 14 days. Report 1781.6691. Springborn Smithers Laboratories, Inc.,
1068 Wareham, MA, USA.
1069 Hughes JS, Alexander MM, Balu K. 1988. An evaluation of appropriate expressions of toxicity
1070 in aquatic plant bioassays as demonstrated by the effects of atrazine on algae and duckweed. In:
1071 Adams W, Chapman GA, Landis WG (eds). Aquatic Toxicology and Hazard Assessment: 10th
1072 Volume. Philadelphia, PA: American Society for Testing and Materials.
1073 Hughes JS. 1986. The toxicity of atrazine, lot no. FL-850612, to four species of aquatic plants.
1074 Report, Study 267-29-1100, Malcolm Pirnie, White Plans, NY, USA.
1075 Johnson BT. 1986. Potential impact of selected agricultural chemical contaminants on a northern
1076 prairie wetland: A microcosm evaluation. Environ Toxicol Chem 5:473-485.
1077 Jones TW, Kemp WM, Estes PS, Stevenson JC. 1986. Atrazine uptake, photosynthetic
1078 inhibition, and short-term recovery for the submersed vascular plant, Potamogetonperfoliatus L.
1079 Arch Environ Contam Toxicol 15:277-283.
1080 Jurgensen TA,. Hoagland KD. 1990. Effects of short-term pulses of atrazine on attached algal
1081 communities in a small stream. Arch Environ Contam Toxicol 19:617-623.
1082 Juttner I, Peither A, Lay JP, Kettrup A, Ormerod SJ. 1995. An outdoor mesocosm study to assess
1083 ecotoxicological effects of atrazine on a natural plankton community. Arch Environ Contam
1084 Toxicol 29:43 5-441.
1085 Kallqvist T, Romstad R. 1994. Effects of agricultural pesticides on planktonic algae and
1086 cyanobacteria - examples of interspecies sensitivity variations. Norw J Agric Sci 13:117-131.
1087 Kemp WM, Boynton WR, Cunningham JJ, Stevenson JC, Jones TW, Means JC. 1985. Effects of
1088 atrazine and linuron on photosynthesis and growth of the macrophytes, Potamogeton perfoliatus
1089 and Myriophyllum spicatum in an estuarine environment. Mar Environ Res 16:255-280.
1090 Kettle WD, deNoyelles F, Heacock BD, Kadoum AM. 1987. Diet and reproductive success of
1091 bluegill recovered from experimental ponds treated with atrazine. Bull Environ Contam Toxicol
1092 38:47-52.
1093 Kettle WD. 1982. Description and analysis of toxicant-induced responses of aquatic
1094 communities in replicated experimental ponds. Ph.D. Thesis. University of Kansas, Lawrence,
1095 KS, USA.
1096 Kirby MF, Sheahan DA. 1994. Effects of atrazine, isoproturon, and mecoprop on the macrophyte
1097 Lemna minor and the alga Scenedesmus subspicatus. Bull Environ Contam Toxicol 53:120-126.
1098 Knauert S, Dawo U, Hollender J, Hommen U, Knauer K. 2009. Effects of photosystem II
1099 inhibitors and their mixture on freshwater phytoplankton succession in outdoor mesocosms.
1100 Environ Toxicol Chem 28:836-845.
37
-------
1101 Knauert S, Escher B, Singer H, Hollender J, Knauer K. 2008. Mixture toxicity of three
1102 photosystem II inhibitors (atrazine, isoproturon, and diuron) toward photosynthesis of freshwater
1103 phytoplankton studied in outdoor mesocosms. Environ Sci Tech 42:6424-6430.
1104 Kosinski RJ. 1984. The effect of terrestrial herbicides on the community structure of stream
1105 periphyton. Environ Pollut (Series A) 36:165-189.
1106 Kosinski RJ, Merkle MG. 1984. The effect of four terrestrial herbicides on the productivity of
1107 artificial stream algal communities. J Environ Qual 13:75-82.
1108 Krieger KA, Baker DB, Kramer JW. 1988. Effects of herbicides on stream aufwuchs
1109 productivity and nutrient uptake. Arch Environ Contam Toxicol 17:299-306.
1110 Lampert W, Fleckner W, Pott E, Schober U, Storkel KU. 1989. Herbicide effects on planktonic
1111 systems of different complexity. Hydrobiologia 188/189:415-424.
1112 Larsen DP, DeNoyelles Jr. F, Stay F, Shiroyama T. 1986. Comparisons of single-species,
1113 microcosm and experimental pond responses to atrazine exposure. Environ Toxicol Chem 5:179-
1114 190.
1115 Leboulanger C, Rimet F, de Lacotte MH, Berard A. 2001. Effects of atrazine and nicosulfuron
1116 on freshwater microalgae. Environ Internal 26:131-135
1117 Lynch TR, Johnson HE, Adams WJ. 1985. Impact of atrazine and hexachlorobiphenyl on the
1118 structure and function of model stream ecosystems. Environ Toxicol Chem 4:399-413.
1119 Mayer P, Frickmann J, Christensen ER, Nyholm N. 1998. Influence of growth conditions on the
1120 results obtained in algal toxicity tests. Environ Toxicol Chem 17:1091-1098.
1121 McGregor EB, Solomon KR, Hanson ML. 2008. Effects of planting system design on the
1122 toxicological sensitivity of Myriophyllum spicatum and Elodea canadensis to atrazine.
1123 Chemosphere 73:249-260.
1124 Millie DF, Hersh CM. 1987. Statistical characterizations of the atrazine-induced photosynthetic
1125 inhibition of Cyclotella meneghiniana. Aquatic Toxicol 10:239-249.
1126 Miller WE, Greene, JC, T Shiroyama. 1978. The Selenastrum capricornutum (Printz) algal assay
1127 bottle test. EPA/600/9-78/018. U.S. Environmental Protection Agency, Corvallis, OR, USA.
1128 Moorhead DL, Kosinski RJ. 1986. Effect of atrazine on the productivity of artificial stream algal
1129 communities. Bull. Environ Contam Toxicol 37:330-336.
1130 Parrish R. 1978. Effects of atrazine on two freshwater and five marine algae. Ciba-Geigy
1131 Corporation, Greensboro, NC, USA.
1132 Pratt JR, Bowers NJ, Niederlehrer BR, Cairns Jr. J. 1988. Effects of atrazine on freshwater
1133 microbial communities. Arch Environ Contam Toxicol 17:449-457.
38
-------
1134 Radetski CM, Ferard JF, Blaise C. 1995. A semistatic microplate-based phytotoxicity test.
1135 Environ Toxicol Chem 14:299-302.
1136 Relyea RA. 2009. A cocktail of contaminants: how mixtures of pesticides at low concentrations
1137 affect aquatic communities. Oecologia 159:363-376.
1138 Roberts SP, Vasseur P, Dive D. 1990. Combined effects between atrazine, copper and pH on
1139 target and non-target species. Water Res 24:485-491.
1140 Rohr JR, Crumrine PW. 1005. Effects of an herbicide and an insecticide on pond community
1141 structure and processes. Ecological Applications 15:1135-1147.
1142 Rohr JR, Schotthoefer AM, Raffel TR, Carrick HG, Halstead N, Hoverman JT, Johnson CM,
1143 Johnson LB, Lieske C, Piwoni MD, Schoff P, Beasley VR.2008. Agrochemicals increase
1144 trematode infections in a declining amphibian species. Nature 455:1235-1239.
1145 Saenz ME, Alberi JL, DiMarzio WD, Accorinti J, Tortorelli MC. 1997. Paraquat toxicity to
1146 different green algae. Bull Environ Contam Toxicol 58:922-928.
1147 Schafer H, Hettler H, Fritsche U, Pitzen G, Roderer G, Wenzel A. 1994. Biotests using
1148 unicellular algae and ciliates for predicting long-term effects of toxicants. Ecotox Environ Saf
1149 27:64-81.
1150 Schafer H, Wenzel A, Fritsche U, Roderer G, Traunspurger W. 1993. Long-term effects of
1151 selected xenobiotica on freshwater green algae: development of a flow-through test system. Sci
1152 Total Environ 113/114:735-740.
1153 Seguin F, Druart J-C, Le Cohu R. 2001. Effects of atrazine and nicosulfuron on periphytic
1154 diatom communities in freshwater outdoor lentic mesocosms. Ann Limnol - Internal J Limnol
1155 37:3-8.
1156 Seguin F, Le Bihan F, Leboulanger C, Berard A. 2002. A risk assessment of polution: induction
1157 of atrazine tolerance in phytoplankton communities in freshwater outdoor mesocosms, using
1158 chlorophyll fluorescence as an endpoint. Water Research 36:3227-3236.
1159 Seguin F, Leboulanger C, Rimet F, Druart J-C, Berard A. 2001. Effects of atrazine and
1160 nicosulfuron on phytoplankton in systems of increasing complexity. Arch Environ Contam
1161 Toxicol 40:198-208.
1162 Solomon KR, Baker DB, Richards P, Dixon KR, Klaine SJ, LaPoint TW Kendall RJ, Giddings
1163 JM, Diesy JP, Hall Jr. LW, Weisskopf C, Williams M. 1996. Ecological risk assessment of
1164 atrazine in North American surface waters. Environ Toxicol Chem 15:31-76.
1165 Stay EF, Katko A, Rohm CM, Fix MA, Larsen DP. 1989. The effects of atrazine on microcosms
1166 developed from four natural plankton communities. Arch Environ Contam Toxicol 18:866-875.
1167 Stay FS, Larsen DP, Katko A, Rohm CM. 1985. Effects of atrazine on community level
1168 responses in Taub microcosms. In: Validation and predictability of laboratory methods for
39
-------
1169 assessing the fate and effects of contaminants in aquatic ecosystems. Boyle, TP (ed.). ASTM
1170 STP 865. American Society for Testing and Materials, Philadelphia, PA, USA.
1171 Stratton, GW. 1981. The effects of selected pesticides and their degradation products on
1172 microorganisms and Daphnia magna. Ph.D. dissertation, University of Guelph, Guelph, Ontario,
1173 Canada.
1174 Stratton GW. 1984. Effects of the herbicide atrazine and its degradation products, alone and in
1175 combination, on phototrophic microorganisms. Arch Environ Contam Toxicol 13:35-42.
1176 Stratton GW, Giles J. 1990. Importance of bioassay volume in toxicity tests using algae and
1177 aquatic invertebrates. Bull Environ Contam Toxicol 44:420-427.
1178 Tang J-X, Hoagland KD, Siegfried BD. 1997. Differential toxicity of atrazine to selected
1179 freshwater algae. Bull Environ Contam Toxicol 59:631-637.
1180 Turbak SC, Olson SB, McFeters GS. 1986. Comparison of algal assay systems for detecting
1181 waterborne herbicides and metals. Water Res 20:91-96.
1182 University of Mississippi. 1991. Effects of atrazine on Selenastrum capricornutum, Lemna minor
1183 and Elodea canademis. Ciba-Geigy Corporation, Greensboro, NC, USA.
1184 U.S. Environmental Protection Agency. 1998. Guidelines for Ecological Risk Assessment.
1185 EPA/630/R-95/002F, U.S.EPA Office of Research and Development, Washington, DC, USA.
1186 U.S. Environmental Protection Agency. 2003. Interim Reregi strati on Eligibility Decision for
1187 Atrazine. Case No. 0062. U.S.EPA Office of Pesticide Programs, Washington, DC, USA.
1188 U.S. Environmental Protection Agency. 2009. The ecological significance of atrazine effects on
1189 primary producers in surface water streams in the corn and sorghum growing region of the
1190 United States (Part II). Briefing document to the FIFRA Scientific Advisory Panel, U.S. EPA
1191 Office of Pesticide Programs, Washington, DC, USA.
1192 U.S. Environmental Protection Agency. 2011. Bibliography of Microcosm and Mesocosm
1193 Studies and Criteria Used to Screen Studies for Analysis of Atrazine Risks to Aquatic Plant
1194 Communities. Memorandum from Michael Lowit, Anita Pease, Dana Spatz, and Brian
1195 Anderson, Environmental Fate and Effects Division, to Melanie Biscoe, Anne Overstreet,
1196 Pesticide Reevaluation Division, Office of Pesticide Programs, U.S. EPA.
1197 Van den Brink PJ, van Donk E, Gylstra R, Crum SJH, Brock TCM. 1995. Effects of chronic low
1198 concentrations of the pesticides chlorpyrifos and atrazine in indoor freshwater microcosms.
1199 Chemosphere 31:3181-3200.
1200 van der Heever JA, Grobbelaar JU. 1996. The use of Selenastrum capricornutum growth
1201 potential as a measure of toxicity of a few selected compounds. Water SA 22: 183-191.
40
-------
1202 Versteeg DJ. 1990. Comparison of short- and long-term toxicity test results for the green alga,
1203 Selenastrum capricornutum. In: Wang W, Gorsuch JW, Lower WR (eds). Plants for Toxicity
1204 Assessment. Philadelphia, PA: American Society for Testing and Materials.
1205 Volz DC, Bartell SM, Nair SK, Hendley P. 2007. Modeling the Potential for Atrazine-Induced
1206 Changes in Midwestern Stream Ecosystems Using the Comprehensive Aquatic System Model
1207 (CASM). Final Report, T001403-07, Syngenta Crop Protection, Greensboro, NC.
1208 Williams PJB, Robinson C, Sondergaard M, Jespersen A-M, Bentley TL, Lefevre D, Richardson
1209 K, Riemann B. 1996. Algal 14C and total carbon metabolisms. 2. Experimental observations
1210 with the diatom Skeletonema costatum. J Plank Res 18:1961-1975.
1211 Zagorc-Koncan J. 1996. Effects of atrazine and alachlor on self-purification processes in
1212 receiving streams. Water Sci Technol 33:181-187.
1213
41
-------
1214
42
-------
1215 APPENDIX A
1216 SINGLE-SPECIES PLANT TOXICITY TEST REVIEW
1217 This appendix provides a summary for each report and journal article reviewed for
1218 developing the compilation of ECsos and steepness values for the relationship of plant specific
1219 growth rate (SGR) to atrazine concentration. Bold numbers in the tables or text denote values
1220 from each study selected for inclusion in the compilation.
1221 A.I Protocol for Application of Toxicity Test Data
1222 A.1.1 Acceptability of measurement variables
1223 (1) The preferred measurement variable for assessing atrazine effects was plant biomass (dry
1224 weight, or wet weight if procedures provided consistent removal of adhering water), but
1225 measures that are approximately proportional to biomass (algal cell count or cell volume,
1226 duckweed frond count) were also accepted.
1227 (2) If measures outlined in (1) were not available, C>2 evolution or 14C fixation measurements
1228 were accepted provided that they were not significantly compromised by any lag in inducing
1229 effects and their relationship to SGR could be defined.
1230 (3) Data based just on chlorophyll content were not used because the chlorophyll content per cell
1231 can change markedly in response to atrazine, leading to markedly different ECsoS for chlorophyll
1232 than for actual biomass (see discussion in Section 2.2.1 in main report text). Similarly, optical
1233 density was not accepted because it also is affected by chlorophyll content, often being measured
1234 near a chlorophyll absorbance maximum.
1235 A.1.2 Translating reported data into SGR ECso and steepness parameter values
1236 The nature of the data and the level of detail provided in the reviewed reports/papers varied
1237 widely, requiring several different procedures for translating the reported data into the elements
1238 of the data compilation: the SGR ECso, a steepness for the SGR vs. atrazine concentration
1239 relationship, and the SGRC.
1240 A. 1.2.1 Initial and final biomasses (or surrogate) were reported for a concentration series.
1241 The preferred data were reported initial and final biomasses (or acceptable surrogates) for all
1242 treatment concentrations, from which SGRs would then be computed. A regression analysis of
1243 SGR vs. atrazine concentration (CATZ) was then conducted, resulting in characterizing both the
1244 ECso and the steepness for the relationship based on the basic measurements in the study. The
1245 analyses were by least-square, nonlinear regression using Version 1.2 of the software package
1246 TRAP (Toxicity Relationship Analysis Program) (U.S.EPA Mid-Continent Ecology Division,
1247 Duluth, MN, http://www.epa.gov/medatwrk/Prods_Pubs/trap.htm), using the "logistic equation"
1248 model option and the log-transform option for CATZ. This model option uses the logistic
1249 equation to provide a sigmoidal regression function shape, but is a regression of a continuous
1250 variable, not binary logistic analysis:
43
-------
1251 SGR = -
4-Steep- Iog10 CATZ -Iog10 EC50
1252 The defining parameters for this function are the control SGR (SGRc), the logio(ECso) for the
1253 SGR, and a measure of relative steepness ("Steep") defined as I d(SGR/SGRc)/d(logio(CAiz)) at
1254 the EC50.
1255 A. 1.2.2 SGRs or relative SGRs were reported for a concentration series.
1256 If the author reported SGRs (based on biomass or acceptable biomass surrogates) for all
1257 treatment concentrations, but not the actual initial and final biomasses, these SGRs were used
1258 directly in regression analysis as described in (Al) above to obtain the SGRc, SGR ECso, and
1259 steepness parameter. If the reported SGRs were relative (fraction of the control), the regression
1260 was conducted to obtain an ECso and steepness to include in the compilation, but not an SGRc,
1261 although in some cases the latter was specified separately by the author(s).
1262 A. 1.2.3 ECsofor the SGR was reported with or without slope.
1263 If the author computed SGRs, but only reported an SGR-based ECso without SGRs for individual
1264 treatment concentrations, the author-calculated SGR ECso was included in the compilation. If
1265 the author also specified the type of relationship used in the ECso estimation and a slope for that
1266 relationship, this information was converted to the steepness parameter of the relationship used
1267 in EPA's regressions; otherwise no steepness was compiled. If the author separately provided
1268 information on the SGRc, this also was included in the compilation.
1269 A. 1.2.4 Multiple ECpSfor growth reported; SGRC reported.
1270 (a) If multiple ECps for growth over a specified duration (t) and the SGRc for that duration were
1271 reported, SGRs corresponding to these biomass-based ECps were calculated using the equation:
1272 SGR =-In 1-^ • eSGRc'< +1
1273 In other words, this is the value for the SGR at the concentration causing a p% decrease in
1274 growth. The resultant SGRs (and their associated concentrations) were then subject to regression
1275 analysis to provide estimates for the SGR ECso and steepness. This provided a SGR ECso, a
1276 steepness, and a SGRc for the compilation.
1277 (b) If the author did not specify multiple ECps for growth, but did provide the growth-based
1278 ECso, the type of relationship used in this ECso estimation, and the slope for that relationship,
1279 additional ECps for growth (p<90%) were calculated for this author-reported curve and also
1280 converted to SGRs. These were then subject to regression analysis to provide estimates for the
1281 SGR ECso and steepness, although any confidence limits on these estimates would not be valid
1282 given that the data points were not independent. Rather, this was simply a mechanism to convert
1283 the the author-reported curve for biomass-based ECs to the equivalent curve for SGR-based ECs.
44
-------
1284 (c) If the smallest SGR was more than 75% of the SGRc for either of the above options, the
1285 regression analysis was not conducted because this would involve too much extrapolation to
1286 estimate the SGR ECso. However, the possibility of extrapolating this SGR to the SGR EC50 per
1287 A. 1.2.6 below was then considered.
1288 A.I.2.5 Multiple ECpSfor growth reported; SGRC not reported.
1289 If multiple ECps or an ECso/slope combination for algal growth were reported, but an SGRc was
1290 not reported, the process in A.I.2.4 above was still used, but using SGRcs reported for other
1291 studies on test species in the same taxonomic group. Because this involves using data from other
1292 experimental systems and test species, three separate analyses were conducted using median
1293 (low-high) estimates for the SGRC of 1.35 (1.05-1.74) for green algae, 1.03 (0.80-1.32) for
1294 diatoms, and 0.65 (0.50-0.83) for blue-green algae. The SGR ECso and steepness from the
1295 regression analysis using the median SGRc estimate were included in the compilation, provided
1296 the SGR ECsos derived using the low and high SGRc estimates differed by no more than a factor
1297 of2.0.
1298 [The low/mid/high SGRC estimates were based on ANOVA of logSGRcs from algal studies in which SGRC
1299 was reported (see Table 1 in Section 2.2). Analyses using Statistica (Version 8.0, StatSoft, Tulsa, OK)
1300 provided a log mean for each major algal taxonomic group (0.135 for green algae, 0.013 for diatoms, -
1301 0.189 for blue-green algae) and a pooled standard deviation (0.122). The low/mid/high estimates for
1302 SGRC were based on calculating the mean ± 1 std.dev. of these log values and then taking antilogarithms.
1303 Separate SGRC values for species within a taxonomic group were not justified because of large within-
1304 species variability relative to between-species variability, as evidenced in Table 1 and other sources (e.g.,
1305 Saenz etal. 1997).}
1306 A. 1.2.6 ECso only for growth reported; SGRc reported.
1307 If the ECsos for growth over a specified duration (t) and the SGRc for that duration were
1308 reported, this biomass-based ECso was equated to an SGR ECP using the following equation to
1309 determine p:
SGR =-In 0.5- eSGRc''+l
1310 t
1__SGR_
¥ 1 SGRC
1311 When only the SGRc and this single SGR are available, no regression analysis is possible.
1312 Rather, this SGR ECP was extrapolated to an SGR EC50 using the equation EC50 = ECP • 10,
1313 where S is based on regression curve steepnesses from other studies. Because this involves using
1314 data from other experimental systems and test species, three estimates of the SGR ECso were
1315 made using low, middle, and high estimates for the steepness of 0.68, 0.95, and 1.31. The
1316 estimate for the SGR ECso from the middle steepness estimate was included in the compilation,
1317 but only if the estimates based on low and high steepness differed by less than a factor of 2. This
1318 factor of 2 requirement was met if p>l6 for the estimated SGR ECP.
45
-------
1319 [An ANOVA of all the log steepness determined in all studies indicated no significant differences
1 320 among species or broader taxonomic groups, so the overall mean and standard deviation of the
1321 log steepness w ere used to se t low /mid/high estimates. ]
1322 A. 1.2. 7 EC 50 only for growth reported; SGRc not reported.
1323 When only an ECso for growth was reported and a study-specific SGRc was not reported, the
1324 biomass-based ECso was equated to SGR-based ECps per section A. 1.2. 5 using low, middle, and
1325 high estimates for SGRc. Then, each of these SGR-based ECP estimates was extrapolated to
1326 SGR ECso estimates per section A.I. 2. 6 using low, middle, and high steepness estimates. The
1327 SGR ECso estimate based on the middle SGRc and steepness estimates was included in the
1328 compilation, provided the extremes of the estimates varied by less than a factor of 2. This factor
1329 of 2 requirement resulted in this procedure being applicable for green algae tests of up to 2 d
1330 long, but tests could be up to 4-d long for blue-green algae and up to 3-d long for diatoms.
1331 Extrapolating ECsos for net growth to SGR ECsoS were just too uncertain for tests longer than
1332 this.
1333 A. 1.2. 8 Oxygen evolution or 14C fixation reported
1334 (a) If the exposure and measurement periods were short enough so that biomass did not change
1335 appreciably during these periods, and if initial biomasses were either measured or could be
1336 treated as approximately the same among treatments, oxygen evolution and radiocarbon fixation
1337 rates were treated as proportional to SGR and ECps for these rates were treated as comparable to
1338 SGR-based ECps. However, this also required consideration of whether these periods were so
1339 short that any lag in the induction of toxicity would significantly perturb the measurement.
1340 Hersh and Crumpton (1989) and Millie and Hersh (1987) reported effects on oxygen evolution
1341 that were >50% within several minutes of exposure to atrazine concentrations that caused similar
1342 effects on biomass-based SGRs. Thus, data were accepted provided an induction lag of 5 min
1343 would not significantly confound results.
1344 (b) When the exposure and measurement periods were the same and biomass changed enough
1345 over the period to substantially affect estimated ECps, oxygen evolution and radiocarbon fixation
1346 were treated as being proportional to net growth (e GR'f), and ECs were converted to an SGR
1347 basis analogously to procedures described above for biomass-based ECs.
1348 (c) If a substantial exposure period of duration "t" preceded a short measurement period, so that
1349 the treatments would start with significantly different initial biomasses for the oxygen
1350 evolution/radiocarbon fixation measurement period, these measures were treated as being
1351 proportional to SGR-e8011'*; i.e., the biomass accretion in the exposure period prior to the start of
1352 measurement is e and the oxygen evolution/radiocarbon fixation rate is proportional to the
1353 SGR times that biomass accretion. This required converting ECs to an SGR-basis using
1354 approaches analogous to that described above for biomass.
1355 A.1.3 Issues regarding biomass surrogates and variability.
1356 One uncertainty issue occurred when the biomass surrogate was cell counts made manually using
1357 a hemocytometer or similar device. In some cases, cell density estimates were based on <100
46
-------
1358 cells counted in total for the control treatment and just several cells for atrazine treatments with
1359 large effects. Even 100 cells represents about +/-20% uncertainty in the cell density. Therefore,
1360 it was desired to have >200 cells counted in the control treatment in order to have reasonable
1361 discrimination between the control and treatments with 25-50% reduced growth. Another area of
1362 concern was frond counts for duckweed, and how closely such counts mirror biomass when
1363 growth is limited and thus might have a greater percentage of newer, small fronds. Where
1364 possible, it was desired to have at least a 4-fold increase in the number of control fronds so that
1365 the counts were not excessively dominated by new, small fronds. A final area of concern was
1366 macrophyte shoot tests at times when controls had not increased by at least 50%, especially if
1367 this was measured by shoot length, which can change disproportionately to shoot weight when
1368 photosynthesis is inhibited. No firm rules were imposed with regard to any of these concerns,
1369 because any uncertainty depends on the number of replicates in a test, the specific times, the
1370 variability among replicates, etc. How these concerns are addressed in the summaries for each
1371 study in Appendix A.
1372 A.1.4 Treatment of data at multiple times
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
When biomasses or biomass surrogates were reported at multiple times within a test's duration,
analyses were conducted for each time; however, the compilation selected results from only one
of these times. The time was selected to be long enough to avoid problems with uncertain
measurements of biomass early in some tests (e.g., the hemocytometer count issue discussed
above), but short enough to avoid potential biases associated with declining SGRc discussed
earlier. Again, no firm rules could be adopted for this because of various study-specific factors
and because it involved balancing uncertainties at early times with those at later times. The
decision process regarding this is provided in the summaries for each study below.
A.2 Review Summaries
A.2.1 Algae
(1) Gala and Giesy 1990
The authors conducted a 96-h flask test of Selenastrum capricornutum growth at multiple
atrazine concentrations, enumerating cell density based on hemocytometer cell counts.
Concentrations were measured. Illumination was continuous at 40 |jE/m2/s, temperature was 24
C. They reported average SGRs over 96 h at each treatment concentration, which were directly
used in EPA regression analyses. Data for earlier times were not reported, but authors noted the
use of extra nutrients to maintain exponential growth. Due to the duration and growth rates, cell
densities would have been high enough to avoid concerns about low numbers of individuals
manually counted.
Measured (Target)
Concentration (ng/L)
Control
64 (60)
121 (120)
261 (250)
Author Measured
SGR (1/d)
1.007
0.773
0.508
0.244
47
-------
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
499 (500)
EC50 (ng/L)
Steepness
0.013
125
(80-194)
1.07
(0.46-1.77)
(2) van der Heever and Grobbelaar 1996
The authors conducted a 72-h flask test of Selenastrum capricornutum growth at multiple
atrazine concentrations, determining biomass (dry weight), cell density (electronic particle
counter), and chlorophyll (by both spectrometry and fluorometry) at 0, 24, 48, and 72 h.
Concentrations were nominal. Illumination was continuous at 300 |jE/m2/s and temperature was
23 C. The authors graphically reported relative (to control) SGRs based on all these measures.
Author-reported ECs based on chlorophyll were substantially (almost 3X) higher than for cell
density and biomass, and were not used in accordance with the review guidelines. Relative
SGRs for cell density and biomass were estimated from the figures, reported in the table below,
and used in EPA regression analyses to determine ECso and steepness. The results based on dry
weight were selected for use because ECsos were modestly higher for cell density (average LCso
= 406 by cell density, 311 by weight) indicative of decreases in mass per cell at higher atrazine
concentrations, so that using cell density would slightly reduce the apparent sensitivity of
biomass to atrazine. The results at 1 d were selected for use because it was unknown whether
control growth rates declined with time, given that only relative SGRs were reported, and
because use of an electronic particle counter should have avoided the problems with low manual
cell counts at early times.
Nominal
Cone (ng/L)
1
5
10
50
100
500
1000
5000
EC50 (ng/L)
Steepness
Author Relative SGR, Cell Counts
Id
1.13
0.98
0.98
0.97
0.95
0.35
0.37
0.20
439
0.56
2d
1.30
1.00
1.11
0.97
1.10
0.30
0.34
0.12
370
0.79
3d
1.22
0.95
1.07
0.97
1.08
0.30
0.37
0.10
401
0.78
Author Relative SGR, Dry Weight
Id
1.06
1.00
0.84
0.88
0.83
0.18
0.10
0.00
236
(149-376)
1.01
(0.52-1.50)
2d
.10
.18
.02
.00
.06
0.30
0.10
0.00
352
1.44
3d
1.00
1.02
0.91
0.93
0.91
0.33
0.10
0.00
352
1.14
(3) van der Heever and Grobbelaar 1997
The authors conducted a 30-min oxygen evolution assay for Selenastrum capricornutum
exposed to multiple atrazine concentrations. Concentrations were nominal. Illumination was
continuous at 300 |jE/m /s and temperature was 23 C. Oxygen evolution rates relative to the
48
-------
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
control were reported graphically and the values in the table below were estimated from the
figure. Because of negative responses at high concentrations, the regression in this review
included a non-zero asymptote at high concentrations, but the ECso is still defined relative to zero
oxygen evolution, not this negative asymptote, so that this would best reflect net production.
Although there was no prior exposure before oxygen evolution measurements were made, the
measurement period was long enough relative to the 5-min induction standard that these results
were accepted. It should be noted that the results are consistent with those for a flask test by the
same authors discussed above.
Nominal
Cone (ng/L)
5
50
500
1000
5000
10000
EC50 (ng/L)
Steepness
Author Relative
Oxygen Evolution
100
84
27
0
-14
-25
223
(144-346)
0.61
(0.42-0.80)
(4) Kallqvist and Romstad 1994
The authors conducted a 72-h flask test of Selenastrum capricornutum growth at multiple
atrazine concentrations, enumerating cell density using an electronic particle counter.
Concentrations were nominal. Illumination was continuous at 70 |jE/m /s and temperature was
not reported but followed OECD standards of 23±2 C. The authors conducted a regression
analysis of probit-transformed relative SGRs, reporting an SGR ECso of 110 |J,g/L (95% cl = 99-
121) and an ECio of 27 ng/L. Individual SGRs were not reported, but these two ECs allow
estimating a steepness of 0.90 for the sigmoidal function used in this review.
The authors also conducted 3- to 6-d microplate exposures of several algal species to atrazine.
The duration of the test varied with species in order to be within the period of exponential
growth. Illumination was continuous at 70 |jE/m2/s for green algae and 30 |jE/m2/s for others.
For these exposures, relative SGRs for each treatment were reported graphically. Values
estimated from the figures are provided in the following table, along with ECsos and steepnesses
estimated from regression analysis of this data. The ECso for Selenastrum was higher for the
microplate exposures than for the flask tests (although by less than 2-fold), suggesting that the
microplate exposure methodology might involve factors that lead to decreased apparent
sensitivity (e.g., nutrient or atrazine reductions, although the former would not be expected if
exponential growth was maintained). These microplate-based numbers were still compiled for
use in subsequent analyses because the Selenastrum ECsos was well within the reported range of
results for this species from other studies; however, this possible source of uncertainty was
recognized in applications of these data.
49
-------
1454
Nominal
Concentration
0
3.2
10
20
32
60
100
200
320
600
1000
2000
3200
6000
10000
ECso
Steepness
Relative SGR (% of Control)
Selenastmm
capricornutum
100
100
93
73
34
12
0
201
(177-227)
0.79
(0.68-0.90)
Chlamydomonas
noctigama
100
100
97
84
53
28
7
0
378
(313-456)
0.65
(0.53-0.77)
Cyclotella
sp.
100
100
100
96
95
61
40
17
0
462
(383-556)
1.22
(0.80-1.64)
Cryptomonas
pyrinoidifera
100
95
99
99
91
85
69
5
0
494
(415-587)
1.15
(0.85-1.45)
Microcystis
aemginosa
100
110
102
95
88
69
58
33
o
J
0
603
(443-820)
0.77
(0.43-1.11)
Synechococcus
leopoliensis
100
91
80
70
57
30
16
13
0
0
136
(116-159)
0.59
(0.52-0.66)
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
(5) Hoberg 1991a
The author conducted a 96-h flask test of Selenastrum capricornutum growth at multiple atrazine
concentrations, enumerating cell density based on hemocytometer cell counts. The author
provided a data table of cell counts at 1, 2, 3, 4 d at multiple concentrations; initial cell counts
were reported to be 1 • 104. Concentrations were measured and were stable for 4 d
(concentrations were 2X higher than target due to diluting error). Light was continuous at 450-
500 ft-c and temperature was 24-25 C. SGRs were calculated by EPA for each duration and
concentration and used in regression analyses to estimate ECso and steepness. Substantial and
continuing declines in control SGRs were observed, so that the growth rate over 2 d was 24%
less than that over the first day. However, cell counts over the first day were lower than desired
for good quantification and the drop in SGR could be partly due to uncertainty in both the initial
and day 1 cell counts. Therefore, day 2 values were selected for the data compilation.
Cone (ng/L)
Target
0
32
63
120
Measured
-
76
130
250
Author Cell Counts (/104)
Id
10.0
5.0
2.3
1.7
2d
33.0
9.3
5.0
4.0
3d'
71.7
49.7
31.7
1.7
4d
105.0
101.7
27.7
2.0
Calculated SGR (1/d)
Id
2.30
1.61
0.83
0.53
2d
1.75
1.12
0.80
0.69
3d
1.42
1.30
1.15
0.18
4d
1.16
1.16
0.83
0.17
50
-------
240
490
EC50
(Hi/y
Steepness
510
970
0.7
0
2.3
0
2.0
0
1.0
0
O.OO
-
109
1.13
0.42
-
131
(59-290)
0.62
(0.18-1.10)
0.23
-
180
2.61
0.00
-
161
2.42
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
(6) Hoberg 1993a
The author conducted a 96-h flask test of Selenastrum capricornutum growth at multiple atrazine
concentrations, enumerating cell density based on hemocytometer cell counts. The author
provided a data table of cell counts at 1, 2, 3, 4 d at multiple concentrations; initial cell counts
were reported to be 0.3 «104. Concentrations were measured and were stable for 4 d. Light was
continuous at 300-450 ft-c and temperature was 24 C. SGRs were calculated by EPA for each
duration and concentration from these counts. The control SGR during the first day was
exceptionally high (3.32/d) and dropped to more typical levels during subsequent days. In
addition, SGRs were high during the first day even at the highest atrazine concentration (2.30/d
at 450 ng/L), and also dropped to more typical values during subsequent days (<0.1/d). These
atypical results might represent an error in the initial cell density, the reported value of which
was atypically low and could not be verified. These data were therefore not used.
(7) Caux et al. 1996
The authors conducted a 4-d microplate test of Selenastrum capricornutum growth at multiple
atrazine concentrations, enumerating cell density using an electronic particle counter. Light was
continuous at 60 |jE/m2/s and temperature was 24 C. The authors only provided a 4-d ECso for
cell density (26 ng/L), with no data on actual cell counts at test termination for atrazine
treatments. No information was provided on actual treatment concentrations. However, they did
report an initial cell density of I'lO4 and a final control cell density of 1-2«106, corresponding to
an SGRc of 1.15-1.32/d, a relatively narrow range. Based on the midrange of the reported final
control cell counts, an SGRc of 1.25/d was used for adjusting the cell density-based ECso to the
SGR (1.08/d) that would result in half the final control density. The authors also reported a
probit slope of 4.95 for the cell density vs. logioC relationship, which allowed calculation of
other ECpS for cell density (e.g., ECie and ECg4 corresponding to ±1 standard deviation in probit
equation) and their corresponding SGRs. Per item A.1.2.4(b) in the protocol, these estimated
SGRs were subject to regression analysis to estimate the SGR ECso and steepness. Confidence
limits are not reported because this regression was not based on independent data points, but on a
conversion of the reported relationship for the cell density ECs.
p
(% reduction in cell counts)
0
16
50
84
ECp
(Hi/y
16.4
26
41
4-d Cell Density
(104 cell/ml)
1.50
1.26
0.75
0.24
Estimated SGR
(1/d)
1.25
1.21
1.08
0.795
51
-------
EC50(^g/L)
Steepness
50
1.66
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
(8) Versteeg 1990
The author compared three assays of atrazine effects on Selenastrum capricornutum growth: a 4-
d flask test enumerating cell density based on hemocytometer cell counts, 5-min 14C fixation
after 30-min exposure, and 30-min oxygen evolution. Light was continuous at 86 |jE/m2/s for
the flask test, 350 |jE/m2/s for the 14C fixation, and 250 |jE/m2/s for the oxygen evolution;
temperature was 24 C. Reported ECsos were 50 ng/L for 4-d cell density, 100 |J,g/L for 14C
fixation, and 380 |J,g/L for oxygen evolution. Data for individual treatments were not reported
for atrazine, but were for simazine, another triazine herbicide. Measurement variables (cell
densities, 14C fixation rate, oxygen evolution rate) relative to the control are provided in the
following table for simazine. Simazine showed differences among the ECsos based on cell
densities, 14C fixation rate, and oxygen evolution similar to atrazine. SGRs based on cell density
effects were also estimated per item A.I.2.5 of the protocol, resulting in an SGR-based ECso
similar to that for 14C fixation. This simazine analysis also resulted in a slope for SGR-based
ECs that was included in the compilation.
Analysis of Versteeg 1990 Results for Simazine
Concentration
(Hi/y
0
25
50
100
150
175
200
225
300
500
ECsoCl^g/L)
Steepness
Cell Density
(% of Control)
100
78
47
23
10
95
1.58
SGR
(% of Control)
100
95
86
73
58
180
1.50
14C Fixation Rate
(% of Control)
100
104
103
59
38
215
1.19
Oxygen Evolution
(% of Control)
100
93
80
70
43
437
1.26
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
Based on the experimental procedures and the results for both atrazine and simazine, this study
was applied as follows regarding ECsos:
(a) Because the oxygen evolution assay involved purging oxygen, with uncertain effects on
photosynthesis rates and sensitivity to atrazine, these data were not used.
(b) Because the 14C fixation assay included prior exposure, the results will be used. Because of
the short exposure and measurement periods, the ECso (100 jig/L) for 14C fixation will be
treated as being equivalent to those for SGRs.
(c) The smaller ECso for the flask test cell density is likely due to it being for cumulative growth
over 4 d. Per item A. 1.2.7 in the protocol, this had too long a duration to extrapolate the cell
density-based ECso to an SGR-based ECso given the range of estimates for the unknown SGRc
52
-------
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
and steepness. However, if the steepness for simazine was used, the procedure would result in
14,
the estimates for the SGR ECso from 95-115 |J,g/L, consistent with that for C fixation.
(9) Larsen et al. 1986
The authors reported ECsos for 14C fixation rates of several algal species, measured over 2 h after
24 h prior exposure to atrazine. Light was continuous at 400 ft-c and temperature was 24 C.
Because the 24-h prior exposure would result in substantially different biomasses among
treatments, this measure is not proportional to the SGR, and because fixation was not cumulative
over the entire period (26 h), it is also not proportional to net growth. Assuming that SGR is
approximately constant within each treatment, the biomass at 24 h would be eSGR and the carbon
fixation over the 2-h measurement period would be proportional to SGR«eSGR, ignoring the small
amount of growth over that 2 h and assuming that the measured fixation over the 2 h is
approximately proportional to the SGR. Given this relationship, per item A. 1.2.7 of the
protocol, an ECso for the SGR can still be calculated from this information, if an SGRc and
steepness can be estimated for use in the following calculations:
(a) Solve for SGKP (p = percent reduction in SGR relative to control)
14,
corresponding to the ECso for C fixation using the equation
~SGR> = 0.5 • SGRceSGRc (i.e., this equation describes what the SGR would
SGRPe"
SGR •
have to be so that the function SGR-e is at half of its control value).
(b) Calculate/? as 100-(l-SGRp/SGRc).
(c) Use the estimated steepness for the toxicity relationship to extrapolate the
known SGR ECP (=EC50 for 14C fixation) to the SGR EC50.
For Selenastrum capricornutum, the authors reported ECsos for 14C fixation of 34-53 (ig/L (three
tests, average 43). Using this average ECso, the procedure described above was conducted
multiple times using the low, middle, and high estimates for SGRc and steepness identified in the
protocol for this review. The range of the resultant SGR ECsos was 66-114 (J,g/L, narrow enough
to include the median SGR ECso (78 (ig/L) in the data compilation. For the other species, the
following table summarizes comparable calculations. For green algae, the same ratio (1.88)
between the carbon fixation and SGR ECSOs was used as for Selenastrum. For blue-green algae,
the ratio used was 1.43 based on the estimates for SGRc for blue-green algae specified in the
review guidelines.
Test Species
Selenastrum capricornutum
Ankistrodesmus sp.
Chlamydomonas reinhardi
Scenedesmus obliquus
Chlorella vulgaris
Stigeoclonium tenue
14CEC50
(Hi/y
43
66
37
48
308
175
SGRECso
(Hi/y
78
119
67
87
557
317
53
-------
Ulothrix subconstricta
Anabaena cylindrica
88
204
159
286
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
(10) Mayer et al. 1998
The authors provided an ECio, ECso, and ECgo for SGRs from a standard ISO 8692 toxicity flask
test (3 d) with Selenastrum capricornutum. The actual temperature and light intensity was not
reported, but the cited test protocol specified 60-120 (jE/m2/s and 23±2 C. The author-reported
SGR ECso of 164 |ag/L will be used, but the multiple ECs can also be used to estimate the
steepness parameter for the sigmoidal relationship used in this review. The author also reported
information on effects of light, temperature, pH, and nitrogen source on both control growth and
toxic effects. This information indicated the SGRc for this study under standard conditions was
about 1.8/d, but insufficient information was available to use other toxicity information for the
present analysis. This study did document a 10-fold increase in chlorophyll content per cell due
to atrazine exposure (200 ng/L), which provides some of the basis for not accepting this as a
surrogate for biomass.
p
(% reduction in control
SGR)
0
10
50
90
ECsoCl-ig/L)
Steepness
ECp
(Hg/L)
17.2
164
688
Relative
SGR
1.0
0.90
0.50
0.10
164
0.79
(11) Roberts et al. 1990
The authors conducted a 7-d flask test of Selenastrum capricornutum growth at multiple atrazine
concentrations, enumerating cell density based on hemocytometer cell counts. Concentrations
were nominal. Light was continuous at 2300 ft-c and temperature was 24 C. The authors
reported the number for the doublings (cell count basis) over 3 d. This number of doublings was
converted to a factor increase, which was converted to an SGR and subject to regression
analysis.
Nominal Concentration
o±g/y
0
50
100
150
EC50(^g/L)
Steepness
Number of
Doublings
7.13
6.64
5.08
4.10
Relative Growth
(Factor increase)
140
100
33.8
17.2
Calculated SGR
(1/d)
1.65
1.53
1.17
0.95
163
1.22
1593
54
-------
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
(12) Parrish, 1978
The author conducted 5-d flask tests of Selenastrum capricornutum and Microcystis aeruginosa
growth at multiple atrazine concentrations, enumerating cell density based on hemocytometer
cell counts. Concentrations were nominal. Light was continuous at 400 ft-c and temperature
was 24 C. The author provided a data table of cell counts at 3 and 5 d at multiple concentrations;
initial cell counts were 2« 104 for Selenastrum and 5« 104 for Microcystis. SGRs were calculated
from the counts for each duration and concentration. Results for Selenastrum are in the
following table. Because there was not a substantial decline in the SGRc and results agreed
between the two durations, the 5-d results were selected for use.
Cone (ng/L)
(nominal)
0
32
54
90
150
250
EC50
Steepness
Author Cell Counts
(/104)
3d
55.8
50.6
34.5
14.6
8.9
0.7
5d
249.6
207.3
130.3
28.2
8.9
0.7
Calculated SGR
(1/d)
3d
1.110
1.077
0.949
0.663
0.498
<0
115
1.47
5d
0.965
0.928
0.835
0.529
0.300
<0
101
(79-130)
1.61
(0.67-2.55)
Results for Microcystis are in the following table. Control growth actually increased later in the
test and ECsos were similar for both durations, so the 5-d results were selected for use.
Cone (ng/L)
(nominal)
0
65
108
180
300
500
EC50
Steepness
Author Cell Counts
(/104)
3d
14.3
13.2
12.9
6.5
5.1
4.7
5d
77.1
71.6
26.1
21.5
9.6
4.0
Calculated SGR
(1/d)
3d
0.350
0.324
0.316
0.087
0.007
0.000
154
4.2
5d
0.547
0.532
0.330
0.292
0.130
0.000
164
(95-285)
1.25
(0.24-2.46)
(13) Turbak et al. 1986
55
-------
1612 The authors reported an ECso of 70 |lg/L based on a 30-min oxygen evolution assay with
1613 Selenastrum capricornutum, with no additional information to determine the steepness of the
1614 relationship. The actual temperature and light intensity was not reported, but the test protocol
1615 specified 400 ft-c and 24 C. The methods description did indicate that there was some exposure
1616 prior to oxygen measurements, and 30 min is long enough not to be greatly perturbed by
1617 induction lags of several minutes. Therefore, this ECso based on rate of oxygen evolution was
1618 accepted as informative of an SGR ECso. They also reported a 59 |J,g/L SGR ECso based on a 2-
1619 3 week bottle test. Because of the length of this test and the lack of specifics regarding it, this
1620 ECso was not used, but this result does not contradict the ECso based on oxygen evolution.
1621
1622 (14) Radetski et al. 1995
1623
1624 The authors reported a 72-h ECso of 118 |J,g/L for Selenastrum capricornutum based on cell
1625 counts (Coulter counter) in a semistatic microplate well test. The actual temperature and light
1626 intensity was not reported, but the cited test protocol specified 60-120 |iE/m2/s and 23±2 C.
1627 They also reported an initial cell count of 2«104and a final control cell count of 6.6«106,
1628 corresponding to an SGRc of 1.93/d. At the reported ECso, the final cell count would thus have
1629 been 3.3«106, equivalent to an SGR of 1.70, corresponding to a 12% reduction from the control
1630 value (i.e., the growth ECso is an SGR ECi2). Per protocol item A.I.2.6, this is too long of an
1631 extrapolation to estimate an SGR ECso given the uncertainty in the steepness of the relationship,
1632 so an SGR ECso was not computed. However, the SGRc was used in the compilation.
1633
1634 (15) Abou-Waly et al. 1991
1635
1636 The authors conducted 7-d flask tests of Selenastrum capricornutum andAnabaenaflos-aquae
1637 aeruginosa growth at multiple atrazine concentrations, measuring weights and chlorophyll
1638 concentrations. Concentrations were nominal. The authors reported SGRs for multiple durations
1639 and concentrations, but only for chlorophyll measurements. Therefore, these data were not used
1640 in accordance with item (A3) of the protocol. Reported chlorophyll-based growth rates and
1641 ECsoS had complex relationships to time and exposure concentration, thereby substantiating
1642 concerns about using chlorophyll measurements. For Anabaena, transferring organisms to
1643 control media after the end of the exposure test showed rapid recovery of growth rates.
1644
1645 (16) Hughes et al. 1988, Hughes 1986
1646
1647 The authors conducted 5-d flask tests of the growth of two algal species, Anabaena flos-aquae
1648 and Naviculapelliculosa, at multiple atrazine concentrations, enumerating cell density by
1649 electronic particle counting. Concentrations were not measured. Light was continuous, and light
1650 intensity/temperatures were 200 ft-c/24 C for Anabaena and 400 ft-c/20 C for Navicula. The
1651 author provided data tables of algal cell densities at 3 and 5 d. SGRs were calculated for each
1652 duration and concentration from these counts, based on the reported initial algal cell densities of
1653 2-104 cells/ml.
1654
1655 The following table provides results for Anabaena flos-aquae. Because no significant effects of
1656 duration are evident on either control growth rates or the ECso, the 5-d results were selected for
1657 further use.
56
-------
1658
Cone (ng/L)
(nominal)
0
100
200
400
800
1600
3200
EC5o
Steepness
Author Cell Counts
(/104)
3d
23.4
16.9
16.1
8.4
6.7
3.9
4.5
5d
88.0
68.4
47.5
24.7
10.2
5.6
5.5
Calculated SGR
(1/d)
3d
0.82
0.71
0.69
0.48
0.40
0.22
0.27
736
0.48
5d
0.76
0.71
0.63
0.50
0.33
0.21
0.20
706
(440-1131)
0.59
(0.35-0.83)
1659
1660
1661
1662
The following table provides results for Naviculapellculosa. Because control growth was
maintained or even increased through 5 d, the 5-d results were selected for further use.
Cone (ng/L)
(nominal)
0
100
200
400
800
1600
3200
ECso
Steepness
Author Cell Counts
(/104)
3d
26.2
9.4
6.0
3.6
2.3
1.9
2.1
5d
347
132
29.3
7.7
2.8
1.9
1.8
Calculated SGR
(1/d)
3d
0.86
0.53
0.37
0.20
0.05
0.00
153
0.80
5d
1.03
0.84
0.54
0.27
0.07
0.00
217
(189-248)
1.08
(0.87-1.29)
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
(17) Fairchild et al. 1994,1998
The authors assessed the effects of four herbicides on plant growth using 4-d tests with six algal
species. Concentrations were not measured in exposure chambers, but the stock concentrations
were verified. Because chlorophyll was used to quantify algal biomass, these data were not used
here per item (A3) of the protocol.
(18) Fairchild et al. 1995,1997
The authors conducted 4-d tests of Selenastrum capricornutum at multiple atrazine
concentrations (as well as 15 other herbicides). Concentrations were not measured. Because
57
-------
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
chlorophyll was used to quantify Selenastrum biomass, these data were not used here per item
(A3) of the protocol.
(19) Burrell et al. 1985
The authors conducted an 11-d flask tests of the growth ofCMorella vulgaris and
Ankistrodesmus braunii at multiple atrazine concentrations, enumerating cell density based on
optical density and hemocytometer cell counts. Concentrations were not measured. Illumination
was continuous at 30 |jE/m2/s and temperature was 24 C. Initial cell densities were 1«105 and
exponential cell growth was reported to be maintained for the test duration, culminating in a final
cell density of 1.7-106 (SGRc=0.26/d) in the Chlorella test and 3.8-106 (SGRc=0.33/d) in the
Ankistrodesmus test. The authors graphically reported the percent reduction in the final cell
density at each atrazine concentration, which were estimated from the figure and reported in the
table below. Based on the final cell densities in the control and the test durations, these percent
reductions in cell density were converted to SGRs at each atrazine concentration and subject to
regression analyses to determine the SGR ECso and steepness. Although this test was longer
than would typically be used for this compilation, the SGRc were low enough (at least in part
due to low light intensities) that total cell densities were not so high as to confound results or to
doubt the authors' statement that exponential growth was maintained. However, because these
SGRcs were so low they were not used for estimating SGRcs for other studies.
Ankistrodesmus
Nominal
Atrazine Cone
(Hi/y
Control
40
60
70
100
EC50(^g/L)
Steepness
% Reduction
in Growth
0
19
49
66
81
SGR
(1/d)
.331
.312
.269
.232
.180
104
(83-131)
1.41
(0.56-2.36)
Chlorella
Nominal
Atrazine Cone
(Hi/y
Control
10
30
50
70
100
EC50(^g/L)
Steepness
% Reduction
in Growth
0
27
55
67
72
75
SGR
(1/d)
.258
.229
.185
.157
.142
.131
91
(70-118)
0.47
(0.32-0.63)
(20) Kirby and Sheahan 1994
The authors conducted a 4-d flask test of the growth of Scenedesmus subspicatus at multiple
atrazine concentrations; concentrations were measured. Illumination was continuous at 3500 lux
and temperature was 25 C. The authors only reported ECsoS based on final biomass, without any
information on specific treatments, growth rates, etc. Initial cell density was MO4 cell/ml and
growth was quantified by spectrophotometric absorbance calibrated to cell density. The ECso
based on final cell density was 21 |j,g/L. Because only an ECso was reported and an SGRc was
not reported, estimation of the SGR ECso would be per item A.I.2.7 of the protocol, but this was
58
-------
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
not done because the extrapolation would be too great (the extrapolated value would be 80 |j,g/L
with a range of 50 to 150 |J,g/L). In addition, this study used optical density near the chlorophyll
a maximum, and so would not be used per the review guidelines.
(21) Millie and Hersh 1987
The authors determined oxygen evolution rates in an electrode chamber for three geographical
races ofCyclotella meneghiana exposed to different atrazine concentrations (unmeasured).
Illumination was at 300 |jE/m /s and temperature was 25 C. The authors graphically reported the
percent inhibition of oxygen evolution rate relative to controls at each concentration, and these
percentages were determined from the graph and subject to regression analysis to determine
oxygen evolution ECso and steepness. Because these were based on a short-term (1 min) oxygen
evolution and because there was prior exposure to each atrazine concentration of several minutes
before oxygen evolution was measured, ECs from these oxygen evolution rates were accepted as
being comparable to SGR ECs.
Nominal
Atrazine Cone
(Hi/y
i
6
31
64
95
143
213
277
338
EC50(^g/L)
Steepness
Oxygen Evolution Rate - % of Control
Minnesota Race
89
78
71
53
40
32
225
(202-251)
1.00
(0.79-1.20)
Arizona Race
94
95
80
58
51
39
31
25
15
100
(86-116)
0.67
(0.56-0.79
Iowa Race
92
85
77
62
54
40
34
21
22
114
(93-141)
0.65
(0.49-0.81)
(22) Hersh and Crumpton 1989
The authors determined oxygen evolution rates in an electrode chamber of a commercial strain of
Chlamydomonas reinhardii and of three isolates ofCMorella sp. obtained from an
uncontaminated natural system exposed to different atrazine concentrations (unmeasured).
Illumination was at 300 |jE/m /s and temperature was 25 C. Only the ECso for the reduction in
oxygen evolution rates relative to control were reported (no data on actual oxygen evolution vs.
concentration), but because these were based on a short-term (1 min) oxygen evolution and
because there was prior exposure to each atrazine concentration of several minutes before
oxygen evolution was measured, these oxygen evolution ECsos were accepted as being
comparable to SGR ECsos. For Chlamydomonas, the EC50 was 45 jlg/L and for Chlorella it
averaged 37 jig/L across the three isolates (range=36-41).
59
-------
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
(23) Stratton 1981,1984
The author measured 14C fixation over 3 h and cell growth rate (by optical density) over 12-14 d
for five algal species exposed to various atrazine and atrazine metabolite concentrations.
Concentrations were unmeasured. For the 14C fixation tests, light intensity was 7000 lux and
temperature was 20 C; these were not specified for the growth test, but presumably were the
same because these were also the culture conditions. For the growth tests, data other than ECsos
at the end of the test were not provided, except for A. inaequalis, and this showed non-
exponential growth throughout the last 10 d of the test and indicated the EC50 was lower at 4-5 d
than later in the text, although the plotted data were insufficient to quantify this. In addition,
optical density was measured at wavelengths with substantial chlorophyll absorption for at least
three of the species. For these reasons, the ECs from the long growth test were not used, and
only the 14C fixation ECSOs were compiled:
14C fixation
ECsoCl-ig/L)
Anabaena
inaequalis
280
Anabaena
cylindrica
470
Anabaena
variabilis
70
Chlorella
pyrenoidosa
480
Scenedesmus
quadricauda
300
(24) Schafer et al. 1994
The authors conducted a 10-d test of the growth of Chlamydomonas reinhardi in a flow-through
apparatus that maintained exponential cell growth, and reported ECsos and ECios for growth at 4,
7, and 10 d. Concentrations were measured. The light intensity was 7000 lux with a 14/10
photoperiod and the temperature was 24 C. Information was also provided to allow estimation of
the SGRc to be 1.06/d, but no additional information on actual or relative cell counts at different
concentrations and times, etc. was given. These ECs were reported to be for growth (not growth
rate) and to be derived per OECD method 201, so presumably were based on "area under the
curve" (AUC). They thus do not represent the difference between the biomass at the stated time
and the biomass at test start, but rather the sum of these differences across the whole time
interval (and thus a measure of the average increase). Because this system maintained an
exponential growth and because the SGRc is known, the ECsos can be used to estimate SGRs for
those concentrations, as summarized in the following table. The magnitudes of these estimated
effects on the SGR are insufficient to support a regression analysis to estimate the SGR ECso and
steepness (due to the large extrapolation from 16% effect to 50% effect). However, per item
A.I.2.6 in the protocol, this SGR ECie of 51 (ig/L can be extrapolated to an estimate of 141
Hg/L for the SGR EC50.
Concentration
o±g/y
Control
10.2
21
51
Duration (d) for which
concentration is AUC EC50
N/A
10
7
4
SGR (1/d)
1.060
0.99
0.96
0.89
60
-------
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
The authors also conducted 3-d flask tests of the growth of Chlamydomonas reinhardii and
Scenedesumus subspicatus at different atrazine concentrations, measuring cell densities at 1, 2,
and 3 d with an electronic particle counter. Illumination was continuous at 8000 lux and the
temperature was 20 C. The authors reported 3-d ECsos and ECios from these tests, but without
any other effect information (e.g., actual or relative cell counts at different concentrations and
times, growth rates). Because of high initial cell densities (2-105 cell/ml) that would have led to
growth-inhibiting densities based on the SGRc from the flow-through test, the growth ECso for
Chlamydomonas (350 |J,g/L) cannot be converted to information on an SGR EC. For
Scenedesmus, initial cell densities were low enough (5-104 cell/ml) to make converting the
growth ECso (72 ng/L) reasonable; however, this would follow item A. 1.2.7 of the protocol, and
the duration of the test is too long for this extrapolation given uncertainties in both SGRc and
steepness.
(25) Faust et al. 1993
The authors conducted 1-d tests of Chlorellafusca growth at multiple atrazine concentrations.
This was a synchronized culture of 1 generation per day, in which a cell grows during the light
period (14 h) and releases a set of daughter cells in the subsequent dark period (10 h); cell counts
were by Coulter counter. The SGRc for cell number would be ln(# of daughter cells) for the
control treatment, but this number was not reported. This number can be as low as 4
(SGRc=1.4/d), but in a related paper by Altenburger et al. (1990), a value of 12 was indicated
(SGRc=2.5/d). The authors reported a probit equation for cell reproduction over 24 h. The
points on this probit equation corresponding to -2, -1, 0, 1, and 2 probit units from the median
were calculated to provide ECps for cell "reproduction" (table below). Then, two sets of SGR
estimates corresponding to these ECps were calculated based on the two alternatives for the
SGRc, and regression analyses were conducted on each of these sets of SGRs. The resultant
SGR EC50 estimates did not differ markedly (table below), so the average of these were
included in the data compilation.
Concentration
(Hg/L)
Control
2.45
6.1
15.1
37.2
92
EC50(^g/L)
Steepness
Percent of Control
Reproduction
100
97.5
84
50
16
2.5
SGR (1/d)
1.4
1.381
1.272
0.927
0.398
0.074
22
1.08
2.4
2.377
2.243
1.794
0.957
0.224
29
1.06
(26) Geyer et al. 1985
The authors conducted 4-d flask tests of Scenedesmus subspicatus growth at multiple atrazine
concentrations. The AUC ECso was reported to be 110 ng/L, but other information (effects at
61
-------
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
higher concentrations, control SGR) were not reported. This test does not meet the protocols
stated earlier for extrapolating such an ECso to one for the SGR.
(27) Zagorc-Koncan 1996
The author determined the net production of oxygen over 24 h (by liberated gas via Warburg-
type apparatus) and increased biomass as measured by chlorophyll over 72 h of Scenedesmus
subspicatus exposed to multiple atrazine concentrations. Light was continuous at 800 lux and
temperature was 20 C. As noted in the protocol, chlorophyll is not an acceptable surrogate for
biomass. Regarding oxygen evolution, the authors reported an EC50 of 25 ng/L, but because of
the lengthy incubation this should be proportional to net biomass gain and not directly related to
effects on SGR. To convert to an SGR-basis requires estimating SGRs based on the oxygen
production and assumptions regarding SGRc. Such estimates based on the range of SGRc for
green algae observed in other studies are included in the table below and subject to regression
analysis. Variation in the assumed SGRc did not cause great variation in the estimated SGR
EC50; because of the low temperature and light intensity, the compilation used the value from
the lowest SGRc value.
Nominal
Atrazine Cone
(Hi/y
Control
0.1
1.0
5.0
10
50
ECsoC^g/L)
Steepness
Estimated SGR (1/d)
SGRC=1.05
1.050
1.038
1.004
0.926
0.896
0.431
39
(27-56)
0.73
(0.45-1.01)
SGRC=1.35
1.350
1.336
1.297
1.208
1.173
0.604
44
0.72
SGRC=1.74
1.740
1.724
1.681
1.580
1.54
0.86
51
0.70
(28) Tang et al. 1997
The authors conducted 28 d tests with several algal species. Growth was measured based on
chlorophyll measurements and optical density near the chlorophyll a maximum. Due to both the
length and the type of measurement, these data were not used.
(29) Gramlich and Frans 1964
The authors conducted a 5-d flask test with Chlorellapyrenoidosa at several atrazine
concentrations. Because biomass was measured by optical density and because initial values for
biomass were not given, useful results for the compilation could not be obtained from this study.
62
-------
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
(30) Stratton and Giles 1990
The authors examined the effect of volume and initial cell density on the toxicity of atrazine to
Chlorellapyrenoidosa, measured by radiocarbon uptake over 24 h. Although these experiments
demonstrated inhibition relative to the control and did include some treatments with
approximately 50% inhibition, only one concentration was tested, absolute fixation rates were
not tested, and a variety of processes might be affecting the observed inhibition. This precluded
applying these data to the data compilation of interest here.
(31) Boger and Schlue 1976
The authors evaluated photosynthesis based on oxygen evolution rate after several days of
exposure to atrazine and the recovery of photosynthesis upon transfer of exposed algae to clean
medium and control algae to contaminated medium. However, only one concentration was
tested and results could not be related to the effect concentrations desired in this review.
(32) University of Mississippi 1991
The authors evaluated growth of Selenastrum capricornutum (4 d) at multiple atrazine
concentrations. This test involved methodological and performance problems that precluded its
use, especially for determining SGR-based ECs. Chlorophyll measurements were made, but
were erratic in addition to being not accepted in the protocol used here. Both cell densities and
weights were also measured, but no initial cell density was specified, final densities were based
on inadequate numbers of cells, and many of the measurements of final weight were negative.
Atrazine effects were evident at 100 |j,g/L, but the next lower and higher concentration was 10-
fold different (10 and 1000 |J,g/L) , precluding any good characterization of dose-response.
A.2.2 Vascular plants
(1) Hughes et al. 1988, Hughes 1986
The authors conducted a 5-d test with the duckweed, Lemna gibba, at multiple atrazine
concentrations, assessing growth by frond count. Concentrations were not measured. Light was
at 500 ft-c and temperature was 25 C. The authors provided data tables of duckweed frond
counts at 3 and 5 d. SGRs were calculated for each duration and concentration from these
counts, based on an initial frond count of 16. The following table summarizes observations and
the estimated SGRs. Because control growth was less than a factor of two at 3 d, the 5-d results
were selected for further use.
Cone (ng/L)
(nominal)
0
100
200
Average Frond
Counts
3d
29.0
27.0
19.7
5d
49.3
40.0
29.7
SGR (1/d)
3d
0.198
0.174
0.069
5d
0.225
0.183
0.124
63
-------
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
400
800
1600
3200
EC50
Steepness
16.3
16.0
1.9
2.1
21.7
16.3
1.9
1.8
0.006
0.000
169
2.17
0.061
0.004
224
(151-332)
1.14
(0.43-1.85)
(2) Hoberg 2007
The author conducted growth tests with isolated shoots ofElodea canadensis at multiple atrazine
concentrations and at zero, dim (500 lux), and optimal (6000 lux) light levels (only the higher
light level is appropriate for this review). Concentrations were measured and temperature was
20-25C. Data tables were provided for individual shoot lengths at 0 and 14 d and individual
shoot dry weights at 14 d for multiple concentrations. Only dry weight is considered here (shoot
lengths were a poor surrogate for growth because substantial shoot elongation was observed in
low light and at high atrazine concentrations were no growth in weight was observed). This
requires having an estimate of the initial dry weight, which the author reported for a separate
initial sample of shoots as being 0.1346 g/shoot. It was assumed that this weight applied to the
average initial shoot length (8.3 cm/shoot) so that the initial weight per cm 0.0162 g/cm. This
factor was used to estimate the initial weights for each replicate tanks based on the initial shoot
lengths within that tank, allowing SGRs to be computed for each tank. The following table lists
the reported final weights, the estimated initial weights, and the resultant shoot weight SGRs,
along with the EC50 and steepness parameter estimated by regression analysis. This regression
analysis is relatively uncertain because the lowest treatment concentration corresponds to an
EC68, leaving an absence of data at low to moderate effect. However, the estimated steepness is
similar to others reported for this species (Table XX) so the EC50 estimate was still deemed
acceptable for us.
Measured Concentration
(Hg/L)
0
464
853
1761
Regression EC50 (ng/L)
Regression Steepness
Estimated Initial Average
Shoot Weight (g dwt)
0.133,0.120,0.129,0.121
0.126,0.131,0.141,0.129
0.137,0.139,0.136,0.153
0.131,0.133,0.149,0.136
Reported Final Average
Shoot Weight (g dwt)
0.420,0.415,0.420,0.471
0.166,0.218,0.225,0.178
0.213,0.179,0.184,0.185
0.128,0.166,0.214,0.126
Shoot Weight SGR
(1/d)
0.082,0.089,0.084,0.097
0.020,0.036,0.034,0.023
0.031,0.018,0.022,0.009
-0.001,0.016,0.026,-0.005
204
(59-600)
0.52
(0.15-0.98)
1898
1899
1900
(3) Hoberg 1991b
64
-------
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
The author conducted a 7-d test ofLemna gibba growth at multiple atrazine concentrations;
concentrations were measured. Light was continuous and temperature was 24 C. The author
provided a data table of frond counts at 3, 6, and 7 d at multiple concentrations; initial frond
counts were 15. SGRs were calculated for each duration and concentration from the counts and
regression analyses were conducted on these SGRs. Because of the absence of growth on day 7,
the 6-d values were compiled.
Measured Atrazine
Concentration (ug/L)
0
15
28
57
120
220
390
EC50
Steepness
Average Frond Counts
3d
34.0
32.0
31.0
33.0
28.3
21.7
19.0
6d
78.0
84.0
78.0
68.0
52.0
34.0
19.7
7d
80.7
85.3
77.0
68.3
51.3
31.3
19.3
SGR (1/d)
3d
0.273
0.253
0.242
0.263
0.212
0.123
0.079
230
1.14
6d
0.275
0.287
0.275
0.252
0.207
0.136
0.045
202
(174-234)
1.24
(0.85-1.62)
7d
0.240
0.248
0.234
0.217
0.176
0.105
0.036
189
1.24
(4) Hoberg 1993b
The author conducted a 14-d test ofLemna gibba growth at multiple atrazine concentrations.
Concentrations were measured. Light was at 400 ft-c and temperature at 24 C. The author
provided a data table of frond counts at 3, 6, 9, 12, and 14 d and dry weight at 14 d. Initial frond
counts were 15. Initial dry weight was unreported but it was assumed for this analysis that the
initial dry weight per frond was equal to that in the control at the end (=110 mg/529=0.208
mg/frond), so that the initial dry weight would be 3.12 mg. SGRs were calculated for each
duration and concentration from the counts and dry weights and regression analyses were
conducted on these SGRs. Based on frond count, some reduction in control growth rate occurred
after 9 d, but did not appreciably affect estimated SGR ECsos. For the 14-d data, dry weights
resulted in an ECso 29% lower than that based on frond count. This is likely attributable to the
lower dry weight/frond at higher atrazine concentrations (i.e., smaller fronds due to atrazine
effects), but also could be contributed to by overestimation of the initial dry weight if control
fronds at the end were on average larger than those at the beginning. This illustrates a possible
weakness in the use of frond counts for duckweed tests, but also a weakness in most tests
regarding measuring initial weights. Due to it being a direct measure of biomass rather than an
indicator, the dry weight-based results were compiled.
Measured
Atrazine
Concen.
0
3.4
Average Frond Count
3d
37.0
35.3
6d
99.0
91.0
9d
255
244
12d
424
426
14d
529
440
Avgdwt
(mg)
14d
110
96
Frond Count
SGR (1/d)
3d
.301
.285
6d
.314
.300
9d
.315
.310
12d
.278
.279
14d
.254
.241
Dwt
SGR
14d
.254
.245
65
-------
7.2
17
47
92
240
FC
-H^50
Steepness
36.0
36.3
32.3
26.7
20.7
89.0
76.0
71.7
45.0
25.7
253
202
163
79
35
475
334
303
117
36
470
364
310
117
43
117
77
17
16
5
.292
,295
.256
.192
.107
156
0.87
.297
.270
.261
.183
.090
133
0.85
.313
.289
.265
.185
.094
130
0.85
.288
.259
.250
.171
.073
129
1.09
.246
.228
.216
.147
.075
134
0.90
.259
.229
.222
.116
.036
93
(72-120)
1.33
(.58-2.07)
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
(5) Hoberg 1993c
The author conducted a 14-d test ofLemna gibba growth at multiple atrazine concentrations.
Concentrations were measured. Light was continuous at 450-500 ft-c and temperature was 25 C.
The author provided a data table of frond counts at 3, 6, 9, 12, and 14 d and dry weight at 14 d.
Initial frond counts were 15. Initial dry weight was unreported but it was assumed for this
analysis that the initial dry weight per frond was equal to that in the control at the end, resulting
in a estimated initial dry weight of 3.7 mg. SGRs were calculated for each duration and
concentration from the counts and dry weights and regression analyses were conducted on these
SGRs. As for Hoberg 1993b, dry weight-based SGRs showed a lower ECso and higher steepness
than frond count-basis, and were selected for the compilation.
Measured
Atrazine
Concen
0
0.53
1.3
3.0
8.3
18
44
100
EC50
Steepness
Average Frond Count
3d
37.2
37.3
37.0
36.7
34.3
32.3
26.0
20.3
6d
88.7
84.7
85.7
89.7
83.3
71.0
46.3
26.7
9d
191
187
185
178
162
136
81
35
12d
277
257
241
284
278
204
132
48
14d
356
364
327
298
321
258
147
53
Avgdwt
(mg)
14d
88
82
94
90
72
58
24
4.2
Frond Count
SGR (1/d)
3d
0.303
0.304
0.301
0.298
0.276
0.255
0.183
0.101
61
0.78
6d
0.296
0.288
0.290
0.298
0.286
0.259
0.188
0.096
63
0.95
9d
0.283
0.280
0.278
0.275
0.264
0.245
0.187
0.094
67
0.91
12d
0.243
0.237
0.231
0.245
0.243
0.218
0.181
0.097
82
0.099
14d
0.226
0.228
0.220
0.214
0.219
0.203
0.163
0.090
81
0.96
Dwt
SGR
14d
0.226
0.221
0.231
0.228
0.212
0.197
0.134
0.009
49
(42-58)
1.71
(.82-2.60)
1942
1943
1944
1945
1946
1947
1948
(6) Desjardin et al., 2003
The authors conducted tests on Lemna gibba growth at multiple atrazine concentrations and for
multiple durations (1-14 d) followed by examination of recovery. Concentrations were
measured. Temperature was 24-25 C and light intensity 4250-5750 lux. Rapid recovery was
demonstrated, but the analyses here are concerned with effects during the exposure period.
66
-------
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
Furthermore, this analysis will be restricted to a 7-d test, because both the longer tests (9-14 d)
produced less than a 20% reduction in the SGR and the 1-3 d tests provided uncertain results due
to the short duration and limited concentration range. The authors provided data at day 2, 4, and
7 d and dry weight at 7 d at multiple concentrations. Initial frond counts were 15 at day -1 and
were 20-21 at the start of exposure (this 1 d period of growth was done to identify/discard
chambers that showed little or no growth; despite this precaution, one control replicate had poor
enough growth to be excluded as an outlier). The initial dry weight was estimated to be 2.8 mg
based on the average dry weight/frond in the no-effect concentrations at the end of the exposure.
SGRs were calculated for each duration and concentration from the counts and dry weights.
Measured
Atrazine
Concen
0.0
4.7
9.4
19.0
38.0
77.0
157
EC50
Steepness
Average Frond Count
2d
40
42
41
41
43
32
31
4d
76
93
96
95
88
60
47
7d
321
349
340
294
262
121
61
Avgdwt
(mg)
7d
37.1
46.0
46.2
38.1
30.8
12.0
5.7
Frond Count
SGR (1/d)
2d
0.347
0.347
0.359
0.359
0.383
0.235
0.195
159
1.09
4d
0.334
0.372
0.392
0.390
0.370
0.275
0.201
165
1.05
7d
0.397
0.402
0.405
0.384
0.368
0.257
0.152
116
1.06
Dwt
SGR
7d
0.381
0.405
0.412
0.385
0.354
0.220
0.106
90
(75-108)
1.18
(.75-1.62)
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
(7) Fairchild et al. 1994,1998
The authors assessed the effects of four herbicides on plant growth using 4-d tests with Lemna
minor and 14-d tests with Ceratophyllum dermersum, Elodea canadensis, Myriophyllum
heterophyllum, and Najas sp. Temperature was 25 C and light was 60 |j,E/m2/sConcentrations
were not measured in exposure chambers, but the stock concentrations were verified. The 1994
report provided detailed biomass measurements absent in the 1998 journal article.
Lemna Initial frond counts were 12 in each replicate and final frond counts are listed in the
following table. The limited duration resulted in limited growth (barely 2-fold in the control)
that makes these results rather uncertain, particularly based on frond counts.
Nominal
Atrazine Cone
(Hi/y
0
37.5
75
150
300
Final frond counts
in replicates
34,26,23
25,25,19
19,20,15
15,17,20
16,18,22
SGRs
(1/d)
0.260,0.193,0.163
0.184,0.115,0.163
0.128,0.056,0.101
0.087,0.128,0.092
0.101,0.152,0.110
67
-------
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
600
ECsod-ig/L)
Steepness
12,14,14
0.000,0.038,0.026
114
(34-390)
0.42
(0.06-0.79)
Najas: Replicates were created by placing natural pond sediments from Najas beds in beakers,
from which plants germinated. Plants were grown for approximately 2 weeks to approximately 3
cm in height, at which time the 14-d chemical exposure began. After the exposure, plants were
sieved and wet weights were determined. Initial wet weights were not determined, but based on
the similarity in the average weights in the highest three treatments (following table) it was
assumed that these treatments had zero net growth and SGRs were estimated based on an initial
wet weight of 69.5 mg, the overall average final weight of these treatments. Given the number
of replicates with lower final weights, the initial weights obviously varied considerably across
replicates, but by basing SGR on the mean weight across replicates, this variability is reduced
enough to produce a clear dose-response. To the extent that the highest three treatments did not
have zero net growth the estimated EC50 will be biased, but substantial bias would be unlikely
because (a) if substantial positive growth was occurring a concentration effect should be evident
and (b) if substantial negative growth was occurring this would imply a high initial weight
incompatible with the information on control growth (i.e. a disproportionate amount of control
growth in the two weeks prior to exposure compared to the 2 weeks of exposure).
Nominal
Atrazine Cone
(Hi/y
Control
Solvent Control
8.4
18.8
37.5
75
150
ECsod^g/L)
Steepness
Final wwt
for replicates
(mg)
306,111,122
285,168,57
66,170,185
164,68,57
57,91,55
65,7,137
49,75,90
Final mean wwt
for treatment
(mg)
180
170
140
96
68
70
71
SGRs
(1/d)
0.068
0.064
0.050
0.023
-0.001
+0.001
+0.002
14.5
(12.3-17.2)
1.67
(1.00-2.33)
Ceratophyllum: The authors provided wet weights for each replicate at 0, 7, and 14 d, allowing
calculation of SGRs and regression analysis of these SGRs to determine the ECso and steepness
of the SGR vs concentration relationship. There was nearly a doubling of weight in the controls
over the 14-d, allowing sufficient growth so that effects were apparent and could be quantified.
68
-------
Nominal
Atrazine Cone
(Hi/y
Control
Solvent Control
18.8
37.5
75
150
300
ECsoC^g/L)
Steepness
Initial wwt
for replicates
(mg)
1578,1202,1730
1310,1746,1622
1209,937,1232
1960,1777,1089
2649,1062,2420
1362,1322,1482
1166,1516,878
Final (14 d) wwt
for replicates
(mg)
2292,2409,2735
2010,2965,2477
1476,1262,1798
2281,2076,1378
2410,1078,2434
1454,1446,1415
1102,1563,1023
SGR
for replicates
(1/d)
0.027,0.050,0.033
0.031,0.038,0.030
0.014,0.021,0.027
0.011,0.011,0.017
-0.007,0.001,0.000
0.005,0.006,-0.003
-0.004,0.002,0.010
24
(14-42)
0.81
(0.12-1.50)
1998
1999 Myriophyllum: The authors provided wet weights for each replicate at 0, 7, and 14 d, allowing
2000 calculation of the SGR for each replicate. However, the growth in controls and in NOECs was
2001 too small and variable for good quantification of effects on SGR. At day 14 (table below), the
2002 weight gain of individual replicates varied from -4-16% (average 8%) in the control, 1-31%
2003 (13%) in the solvent control, from 11-16% (15%) at 37.5 |ig/L, and 2-26% (15%) at 75 |ig/L. In
2004 addition, at day 7, the weight gains were 12-17% (15%) in the controls, 25-33% (28%) in the
2005 solvent controls, 13-16% (13%) at 37.5 ng/L, and 6-21% (11%) at 75 ng/L. These data illustrate
2006 not just a small amount of growth and great variability relative to the average net growth, but
2007 also no or negative growth in most replicates during the second week, which the authors also
2008 noted in other experiments. In addition, there is an inconsistency between the 7- and 14-d data in
2009 that the 14-d data show no difference among the controls and the two lowest concentrations,
2010 whereas the 7-d data indicate better growth in the solvent controls relative to the control without
2011 solvent and the two lowest concentrations. Although there are clear effects at 150 |j,g/L and
2012 above, there is not a good reference against which to quantify effects on the SGR. This
2013 underscores the requirement in the protocol that control growth be large and consistent enough to
2014 quantify ECs with reasonable precision. The most that can be inferred from this test is that 37.5
2015 and 75|ig/L are apparently NOECs and the SGR EC50 is probably ~<150
2016
Nominal
Atrazine Cone
(Hi/y
Control
Solvent Control
37.5
75
150
300
600
ECsod-ig/L)
Initial wwt
for replicates
(mg)
3330,4547,3200
3137,3767,3817
2600,3077,3084
3046,2872,4122
3262,3854,4414
3559,3039,2756
2812,3748,3341
Final (14 d) wwt
for replicates
(mg)
3696,4379,3712
3184,3981,5017
3021,3402,3603
3895,3382,4197
3782,3726,4454
3359,2074,2829
1877,3363,2992
SGR
for replicates
(1/d)
0.007,-0.003,0.011
0.001,0.004,0.020
0.011,0.007,0.011
0.018,0.012,0.001
0.011,-0.002,0.001
-0.004,-0.027,0.002
-0.029,-0.008,-0.008
SGR
for treatment
(1/d)
0.005
0.008
0.010
0.010
0.003
-0.010
-0.015
<«150
69
-------
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
Steepness | | | |
Elodea: The authors provided both wet weights for each replicate at 0, 7, and 14 d, allowing
calculation of the SGR for each replicate. However, as for Myriophyllum, the control growth
was very small, averaging only about 15% over the two weeks. Although, this growth was not as
variable as for Myriophyllum, it still is a questionable reference against which to quantify effects
on SGRs. In addition, the lowest treatment concentration produced no growth on average, and
negative growth became progressively greater at higher concentrations, so that ECs for SGR
could not be quantified even if the controls were good references for quantifying the SGR. The
most that can be inferred from this test is that the SGR ECso is <38 ng/L, although even this
might be confounded by the low control growth.
Nominal
Atrazine Cone
(Hi/y
Control
Solvent Control
37.5
75
150
300
600
ECsoC^g/L)
Steepness
Initial wwt
for replicates
(mg)
4820,5564,6866
5554,5672,6624
7146,3370,5500
6028,5477,6477
4941,4929,4992
6080,5937,5398
6902,7160,6200
Final (14 d) wwt
for replicates
(mg)
5949,6345,7802
6336,6140,7016
7258,3232,5556
5435,5178,6478
4778,4851,5554
5575,5543,5087
3960,6302,5605
SGR
for replicates
(1/d)
0.015,0.009,0.009
0.009,0.006,0.004
0.001,-0.003,0.001
-0.007,-0.004,0.000
-0.002,-0.001,0.007
-0.006,-0.005,-0.004
-0.040,-0.009,-0.007
SGR
for treatment
0.014
0.008
0.001
-0.002
-0.002
-0.004
-0.018
<37.5
(8) Fairchild et al. 1995,1997
The authors conducted 4-d tests ofLemna minor growth at multiple atrazine concentrations (as
well as 15 other herbicides). Concentrations were not measured. ForLemna, the reported ECso
of 153 |j,g/L was based on growth (frond count basis), and insufficient information was provided
to convert this to a growth rate basis. Based on a control growth rate of 0.21/d for identical
methodology used above by Fairchild et al. (1994, 1998) this ECso would correspond to an ECg2-
Because this extrapolation was greater than allowed in the protocol, this data just indicate that
the SGR EC50 is <153 ng/L, which does not contradict the results of Fairchild et al. (1994, 1998).
(9) Kirby and Sheahan 1994
The authors conducted a 10-d test of the growth ofLemna minor at multiple atrazine
concentrations; concentrations were measured. Temperature was 25 C and light intensity was
3500 lux. The authors only reported ECSOs based on final biomass, without any information on
specific treatments, growth rates, etc. The initial biomass was 10 fronds and growth was
quantified by chlorophyll, frond count, and fresh weight, with the respective ECsos being 56, 60,
and 62 |j,g/L. Using the average SGRc from other studies with Lemna (0.27/d, range 0.21-
0.38/d), the ECso for frond count would correspond to an £€25 for SGR. Using the average
70
-------
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
steepness for SGR vs. concentration from other studies with Lemna (1.0 for frond count increase,
1.4 for weight increase), the SGR ECso would then be 105 jig/L based on frond count and 95
jig/L based on weight.
(10) University of Mississippi 1991
The authors evaluated growth of Lemna gibba (14 d), and Elodea canadensis (10 d) at multiple
atrazine concentrations. These assays entailed methodological and performance problems that
precluded their use, especially for determining SGR-based ECs. Chlorophyll measurements were
erratic in addition to being not accepted in the protocol used here. For Lemna, both frond counts
and weights were measured, but frond counts indicated poor control growth (an SGR of 0.1/d,
compared to 0.2-0.4/d in other studies), no initial weights were given, and final weights had poor
precision. For Elodea, final dry weights did show a substantial effect of atrazine, but initial
weights were not given, so that growth could not be assessed either as a rate or an absolute
amount. For both species, atrazine effects were evident at 100 |ig/L, but the next lower and
higher concentration was 10-fold different (10 and 1000 ng/L) , precluding any good
characterization of dose-response.
(11) Forney and Davis 1981; Davis 1980, Forney 1980
The authors evaluated growth of Elodea canadensis, Myriophyllum spicatum, Potamogeton
perfoliatus, and Vallisneria americana in exposures of 3-9 weeks to multiple atrazine
concentrations. Depending on the experiment and test species, light varied from 3 to 170
|iE/m2/s (14/10 h photoperiod) and temperature was 20-30 C. Unfortunately, most of the
evaluations were of shoot length increase, which as discussed above is a questionable surrogate
for growth. In three instances, useful information regarding the SGR ECso could be obtained:
For Potamogeton, in one experiment, dry weight was measured in addition to shoot length.
However, the nature of the weight measurements was unclear (gross weight vs. growth, how
much of plant included) and the authors noted that food reserves in the tuber used to sprout
Potamogeton would partially mask herbicide effects, so that these weight measurements would
overestimate ECs. This experiment also showed atrazine-dependent mortality at concentrations
of 32 ng/L and above. The following table shows the average dry weight of plants (at death or
end of test for survivors), the percent survival, and the product of dry weight and survival as an
estimate of live biomass at the end of the study. For issues regarding weight effects already
noted, this product might still underestimate biomass production, but was considered adequately
informative of atrazine effects on the SGR of a population of this plant. A regression analysis
was thus conducted on this product and used for the compilation.
Nominal
Atrazine Cone
(|ig/L)
0
10
32
% of Control
Dry Weight
100
86
86
% Survival
100
100
73
% of Control
Biomass
100
86
63
71
-------
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
100
320
EC50(^g/L)
Steepness
74
55
62
0
46
0
63
0.69
For Vallisneria, leaf length was measured and was used as a surrogate for growth because it
would be less susceptible than shoot length to elongation with little or no weight increase. Even
with this acceptance, most data could not be used because the authors noted that effects of
atrazine were not evident early in the experiments, likely due to food reserves in the tubers, and
that some experiments had light intensities high enough to inhibit leaf growth in favor of tuber
and lateral shoot development. Thus, analysis here was restricted to the latter part of one test
that the authors reported as being most informative about atrazine effects. The following table
provides the percentage increase in leaf length during the last week of this experiment, which
should be approximately proportional to the SGR. In another experiment with insufficient data
for analysis here, there was information on the ratio of plant weight to leaf length as a function of
atrazine, which did indicate some thinning of the leaves due to atrazine. The following table
includes those ratios, which provided a basis for estimating weight based on leaf length (only
three measured values - so interpolated value used for 32 ng/L and possible extrapolated values
for 1000 ng/L). This resulted in a decrease in the SGR EC50 of about 28%.
Nominal
Atrazine Cone
(Hi/y
0
32
100
320
1000
EC50(^g/L)
Steepness
% Increase in
Leaf Length
in Week 6
14.3
9.8
10.2
5.9
3.6
195
0.36
Dry Weight/
Leaf Length
(fraction of control)
1.00
0.97
0.94
0.82
0.7-0.8
Estimated
% Increase
in Weight
14.3
9.5
9.6
4.8
2.5-2.9
140-141
0.39-0.41
For Elodea, in one experiment dry weight increase was measured. The following table provides
these data. Because initial and final dry weights weren't provided, SGRs cannot be calculated,
but the slow growth rates of these plants should make the net increase proportional to SGR.
Because of the widely space concentrations, the estimated parameters are uncertain, but clearly
indicate the SGR EC50 to be less than 100 (ig/L.
Nominal
Atrazine Cone
(Hi/y
0
10
100
1000
Average Increase
in Plant Dry Wt.
(mg)
37
28
17
11
72
-------
EC50(^g/L)
Steepness
65
0.28
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
(10) Hinman 1989
The author tested the effects of atrazine on both root and shoot growth ofHydrilla verticillata in
both water and sediment exposures (14 d). Concentrations were nominal, light was 40-50
|jE/m2/s, and temperature was 25 C. Both shoot and root growth was monitored by increase in
length. Increases in shoot length are subject to questions about elongation without increasing
weight, but this is not true for root growth, which should still be an indicator of atrazine effects
on primary production. The following table compares the data on root and shoot growth for the
water-based exposures. Shoot lengths do indicate a higher threshold for effects, but then a
steeper decline, with the EC50 being about 80% higher than for root length.
Nominal
Atrazine Cone
(Hi/y
0
16
80
160
800
1600
EC50(^g/L)
Steepness
Shoot Length
Increase
(% of Control)
100
97
127
83
5
5
222
2.26
Root Length
Increase
(% of Control)
100
98
71
25
25
8
118
0.6
2124
2125
73
-------
2126
74
-------
APPENDIX B.
EXPERIMENTAL ECOSYSTEM DATA
2130
75
-------
T
n
c
able Bl. Summary of experimental ecosystem studies used in development of PATlLoc. ID# identifies treatment and cross-
rferences exposure time-series provided in Table B2. Effect is binary (yes/no) regarding whether substantial impact on plant
3mmunity occurred.
ID#
1
2
3
4
5
7
8
9
10
13
14
15
17
18
19
22
23
24
25
26
27
28
29
30
31
32
33
Duration
(d)
365
365
63
365
340
56
56
96
96
53
53
53
7
12
12
15
43
32
17
14
30
21
21
21
12
12
7
Initial Cone.
(ng/L Atrazine)
500
20
500
100
200
80
140
100
100
430
820
3980
100
500
5000
15
25
50
79
100
1000
10
1000
10000
24
134
10000
Significant
Effect?
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Reference
Carney 1983; Kettle et al. 1987; deNoyelles et al. 1989; deNoyelles et al. 1994
Carney 1983; Kettle et al. 1987; deNoyelles et al. 1989; deNoyelles et al. 1994,
deNoyelles & Kettle 1980, Dewey 1986
deNoyelles et al. 1982; deNoyelles et al. 1989
deNoyelles et al. 1989 Carney 1983
deNoyelles et al. 1989 Carney 1983
Hamilton et al. 1987
Hamilton et al. 1987
Hamilton et al. 1988
Herman et al. 1986; Hamilton et al. 1989
Stayetal. 1985
Stayetal. 1985
Stayetal. 1985
Brockway et al. 1984
Brockway et al. 1984
Brockway et al. 1984
Detenback et al. 1996
Detenback et al. 1996
Detenback et al. 1996
Detenback et al. 1996
Hamala and Kollig 1985
Johnson 1986
Kosinski 1984; Kosinski and Merkle 1984
Kosinski 1984; Kosinski and Merkle 1984
Kosinski 1984; Kosinski and Merkle 1984
Kriegeretal. 1988
Kriegeretal. 1988
Moorhead and Kosinski 1986
76
-------
T
ableBl (continued).
ID#
34
35
36
37
38
39
40
41
42
44
45
46
47
48
49
50
51
52
53
54
58
58b
59
60
61
62
63
64
65
66
Duration
(d)
21
42
42
42
42
55
15
360
360
21
7
7
53
53
53
42
12
63
30
30
18
42
21
21
42
35
7
7
29
70
Initial Cone.
(ng/L Atrazine)
337
204
500
1000
5000
50
100
100
200
100
100
1000
53
84
170
100
50
20
10
100
1
0.1
32
110
20
5
0.5
5
0.5
5
Significant
Effect?
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
N
N
Y
Y
Y
Y
N
N
N
N
N
N
Reference
Pratt etal. 1988
Stayetal. 1989
Stay etal. 1989
Stayetal. 1989
Stayetal. 1989
Brockway et al. 1984
Brockway et al. 1984
deNoyelles et al. 1989
deNoyelles et al. 1989
Kosinski 1984; Kosinski and Merkle 1984
Moorhead and Kosinski 1986
Moorhead and Kosinski 1986
Stayetal. 1985
Stayetal. 1985
Stayetal. 1985
Stayetal. 1989
Brockway et al. 1984
deNoyelles et al. 1982; deNoyelles et al. 1989
Johnson 1986
Johnson 1986
Lampertetal 1989
Lampertetal 1989
Pratt etal. 1988
Pratt etal. 1988
Stayetal. 1989
van den Brink et al. 1 995
Brockway et al. 1984
Brockway et al. 1984
Brockway et al. 1984
Brockway et al. 1984
77
-------
T
ableBl (continued).
ID#
67
68
69
70
71
72
73
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
95
96
97
98
99
100
101
Duration
(d)
14
20
20
20
28
28
28
30
21
21
30
28
36
173
23
40
40
40
40
25
7
7
25
51
51
51
42
42
42
42
Initial Cone.
(ng/L Atrazine)
5
1
20
10
2
30
100
25
3.2
10
25
117
6.4
84
10
30
2
30
2
30
148
24.3
207
20
196
2036
25
50
100
250
Significant
Effect?
N
N
N
N
N
N
N
N
N
N
Y
Y
N
Y
Y
N
N
Y
Y
Y
Y
Y
N
N
Y
Y
N
N
Y
Y
Reference
Gruessnerand Watzin 1996
Gustavson and Wangberg 1995
Gustavson and Wangberg 1995
Gustavson and Wangberg 1995
Jurgensen and Hoagland 1990
Jurgensen and Hoagland 1990
Jurgensen and Hoagland 1990
Lynch et al. 1985
Pratt etal. 1988
Pratt etal. 1988
Rohrand Crumrine, 2005
Rohret al., 2008
Relyea, 2009
Knauert et al., 2008; Knauert et al., 2009
Berard et al. 1999a, Berard et al. 1999b, Berard and Benninghoff 2001, Sequin
et al. 2001 b, Leboulanger et al. 2001
Seguin etal. 2001 a
Seguin etal. 2001 a
Seguin etal. 2001 b
Seguin etal. 2001 b
Seguin et al. 2002
Downing et al. 2004
Downing et al. 2004
Boone and James 2003
Diana et al. 2000
Diana et al. 2000
Diana et al. 2000
McGregor et al. 2008
McGregor et al. 2008
McGregor et al. 2008
McGregor et al. 2008
78
-------
Table
B2. Atrazine exposure time-series for experimental ecosystem treatments, with ID# as specified in Table Bl.
ID#1
Time
(d)
0
10
20
40
70
100
130
180
285
330
365
Cone
(ng/L)
500
525
490
350
490
400
400
375
250
200
160
ID#2
Time
(d)
0
10
20
40
70
100
130
180
285
330
365
Cone
(ng/L)
20.0
16.0
16.0
16.0
15.0
12.0
14.0
15.0
7.0
5.0
4.0
ID#3
Time
(d)
0
2
25
30
55
63
Cone
(ng/L)
500
490
465
453
390
360
ID#4
Time
(d)
0
10
20
40
70
100
130
180
285
330
365
Cone
(W3/L)
100
90
85
90
80
75
70
70
35
30
25
ID#5
Time
(d)
0
20
40
60
70
80
105
130
160
190
220
250
290
340
Cone
(ng/L)
200
190
120
160
140
150
120
120
110
140
120
100
90
50
ID#7
Time
(d)
1
3
5
7
9
11
13
15
17
19
21
23
25
27
29
31
33
35
37
39
41
43
45
47
49
51
53
55
Cone
(ng/L)
80
79
78
78
77
76
76
75
75
74
73
73
72
71
71
70
70
69
69
68
67
67
66
66
65
65
64
64
ID#8
Time
(d)
1
56
Cone
(ng/L)
140
110
ID#9
Time
(d)
1
5
14
20
24
34
37
42
54
68
96
Cone
(W3/L)
100
117
108
107
87
105
142
148
132
115
53
79
-------
2135
Table
B2, Page 2.
ID#10
Time
(d)
1
5
14
20
24
34
37
42
54
68
96
Cone
(W3/L)
100
117
108
107
87
105
142
148
132
115
53
ID#13
Time
(d)
0
21
46
53
Cone
(W3/L)
430
264
223
198
ID#14
Time
(d)
0
21
46
53
Cone
(W3/L)
820
505
443
417
ID#15
Time
(d)
0
21
46
53
Cone
(W3/L)
3980
1890
1390
1540
ID#17
Time
(d)
0
1
2
3
4
5
6
7
Cone
(W3/L)
100
100
99
99
98
98
98
97
ID#18
Time
(d)
0
1
2
3
4
5
6
7
8
9
10
11
12
Cone
(W3/L)
500
498
496
494
492
490
488
486
484
481
479
477
475
ID#19
Time
(d)
0
1
2
3
4
5
6
7
8
9
10
11
12
Cone
(W3/L)
5000
4979
4958
4937
4917
4896
4876
4855
4835
4815
4794
4774
4754
ID#22
Time
(d)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Cone
(W3/L)
15.0
13.6
12.9
12.3
11.7
11.1
10.6
10.1
9.6
9.1
8.7
8.3
7.9
7.5
7.1
80
-------
Table
B2, Page3.
ID #23
Time
(d)
1
3
5
7
9
11
13
15
17
19
21
23
25
27
29
31
33
35
37
39
41
43
Cone
(M9/L)
25.1
21.6
19.6
17.7
16.1
14.6
13.2
11.9
10.8
9.8
8.9
8.0
7.3
6.6
6.0
5.4
4.9
4.4
4.0
3.6
3.3
3.0
ID#24
Time
(d)
1
3
5
7
9
11
13
15
17
19
21
23
25
27
29
31
Cone
(M9/L)
50
43
39
35
32
29
26
24
21
19
18
16
14
13
12
11
ID#25
Time
(d)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
Cone
(M9/L)
79
72
68
65
62
59
56
53
51
48
46
44
42
40
38
36
34
ID#26
Time
(d)
0
14
Cone
(M9/L)
100
100
ID#27
Time
(d)
0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
Cone
(M9/L)
1000
992
983
975
967
959
951
943
935
927
919
912
904
897
889
882
ID#28
Time
(d)
1
21
Cone
(M9/L)
10.0
10.0
ID#29
Time
(d)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Cone
(M9/L)
1000
648
522
420
339
273
220
177
142
115
92
74
60
48
39
31
25
20
16
13
11
ID#30
Time
(d)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Cone
(M9/L)
10000
6484
5221
4205
3386
2726
2195
1768
1424
1146
923
743
599
482
388
313
252
203
163
131
106
81
-------
Table
B2, Page 4.
ID #31
Time
(d)
0
12
Cone
(M9/L)
24.0
24.0
ID#32
Time
(d)
0
12
Cone
(M9/L)
134
134
ID#33
Time
(d)
0
1
2
3
4
5
6
7
Cone
(M9/L)
10000
9958
9916
9875
9833
9792
9751
9710
ID#34
Time
(d)
0
21
Cone
(M9/L)
337
337
ID#35
Time
(d)
1
3
5
7
9
11
13
15
17
19
21
23
25
27
29
31
33
35
37
39
41
Cone
(M9/L)
204
199
196
193
190
187
184
181
178
175
172
169
167
164
161
159
156
154
151
149
146
ID#36
Time
(d)
1
3
5
7
9
11
13
15
17
19
21
23
25
27
29
31
33
35
37
39
41
Cone
(M9/L)
492
474
463
452
441
430
420
410
400
390
381
372
363
354
346
337
329
321
314
306
299
ID#37
Time
(d)
1
3
5
7
9
11
13
15
17
19
21
23
25
27
29
31
33
35
37
39
41
Cone
(M9/L)
961
931
918
907
895
883
872
860
849
838
827
816
806
795
785
775
765
755
745
735
726
ID#38
Time
(d)
1
3
5
7
9
11
13
15
17
19
21
23
25
27
29
31
33
35
37
39
41
Cone
(M9/L)
4929
4806
4758
4710
4662
4615
4569
4523
4477
4432
4388
4344
4300
4257
4214
4171
4129
4088
4047
4006
3966
82
-------
Table
B2, Page 5.
ID #39
Time
(d)
0
55
Cone
(M9/L)
50
50
ID#40
Time
(d)
0
15
Cone
(M9/L)
100
100
ID#41
Time
(d)
0
180
360
Cone
(M9/L)
100
70
25
ID#42
Time
(d)
0
180
360
Cone
(M9/L)
200
140
50
ID#44
Time
(d)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Cone
(M9/L)
100
65
52
42
34
27
22
18
14
12
9
7
6
5
4
3
3
2
2
1
ID#45
Time
(d)
1
2
3
4
5
6
7
Cone
(M9/L)
100
99
99
98
98
98
97
ID#46
Time
(d)
1
2
3
4
5
6
7
Cone
(M9/L)
1000
992
988
983
979
975
971
ID#47
Time
(d)
0
21
46
53
Cone
(M9/L)
52
48
41
34
83
-------
Table
B2, Page 6.
ID #48
Time
(d)
0
21
46
53
Cone
(M9/L)
84
63
60
51
ID#49
Time
(d)
0
21
46
53
Cone
(M9/L)
169
114
95
98
ID#50
Time
(d)
1
3
5
7
9
11
13
15
17
19
21
23
25
27
29
31
33
35
37
39
41
Cone
(M9/L)
100
97
96
94
92
91
89
88
86
85
83
82
80
79
78
76
75
74
72
71
70
ID#51
Time
(d)
0
1
2
3
4
5
6
7
8
9
10
11
12
Cone
(M9/L)
50
50
50
49
49
49
49
49
48
48
48
48
48
ID#52
Time
(d)
1
2
25
30
55
63
Cone
(M9/L)
20.0
19.5
18.0
17.0
15.0
14.5
ID#53
Time
(d)
0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
Cone
(M9/L)
10.0
9.9
9.8
9.8
9.7
9.6
9.5
9.4
9.3
9.3
9.2
9.1
9.0
9.0
8.9
8.8
ID#54
Time
(d)
0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
Cone
(M9/L)
100
99
98
98
97
96
95
94
94
93
92
91
90
90
89
88
ID#58
Time
(d)
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Cone
(M9/L)
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
0.9
0.9
0.9
0.9
0.9
0.9
84
-------
2140
Table
B2, Page 7.
ID #58b
Time
(d)
0
42
Cone
(M9/L)
0.1
0.1
ID#59
Time
(d)
0
10
21
Cone
(M9/L)
32
32
32
ID#60
Time
(d)
0
10
21
Cone
(M9/L)
110
110
110
ID#61
Time
(d)
1
3
5
7
9
11
13
15
17
19
21
23
25
27
29
31
33
35
37
39
41
Cone
(M9/L)
17.7
17.4
17.1
16.9
16.7
16.5
16.3
16.1
15.9
15.7
15.5
15.3
15.1
14.9
14.7
14.5
14.3
14.1
14.0
13.8
13.6
ID#62
Time
(d)
0
35
Cone
(M9/L)
5.0
5.0
ID#63
Time
(d)
1
2
3
4
5
6
7
Cone
(M9/L)
0.5
0.5
0.5
0.5
0.5
0.5
0.5
ID#64
Time
(d)
1
2
3
4
5
6
7
Cone
(M9/L)
5.0
5.0
4.9
4.9
4.9
4.9
4.9
ID#65
Time
(d)
0
29
Cone
(M9/L)
0.5
0.5
85
-------
Table
B2, Page 8.
ID #66
Time
(d)
0
70
Cone
(M9/L)
5.0
5.0
ID#67
Time
(d)
1
5
10
14
Cone
(M9/L)
4.7
3.6
1.2
1.2
ID#68
Time
(d)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Cone
(M9/L)
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
0.9
0.9
0.9
0.9
0.9
0.9
0.9
0.9
0.9
0.9
ID#69
Time
(d)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Cone
(W3/L)
20.0
19.8
19.7
19.7
19.6
19.5
19.4
19.3
19.3
19.2
19.1
19.0
18.9
18.9
18.8
18.7
18.6
18.5
18.5
18.4
ID#70
Time
(d)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Cone
(W3/L)
10.0
9.9
9.9
9.8
9.8
9.8
9.7
9.7
9.6
9.6
9.5
9.5
9.5
9.4
9.4
9.3
9.3
9.3
9.2
9.2
ID#71
Time
(d)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
Cone
(W3/L)
2.0
1.6
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
2.0
1.6
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
ID#72
Time
(d)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
Cone
(W3/L)
30
23
0
0
0
0
0
0
0
0
0
0
0
30
23
0
0
0
0
0
0
0
0
0
0
0
0
0
ID#73
Time
(d)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
Cone
(W3/L)
100
78
0
0
0
0
0
0
0
0
0
0
0
100
78
0
0
0
0
0
0
0
0
0
0
0
0
0
86
-------
Table
B2, Page 9.
ID #75
Time
(d)
0
30
Cone
(M9/L)
25.0
25.0
ID#76
Time
(d)
0
21
Cone
(M9/L)
3.2
3.2
ID#77
Time
(d)
0
21
Cone
(M9/L)
10.0
10.0
ID#78
Time
(d)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Cone
(M9/L)
25.0
24.8
24.7
24.6
24.5
24.4
24.3
24.2
24.1
24.0
23.9
23.8
23.7
23.6
48.5
48.3
48.1
47.9
47.7
47.5
47.3
47.1
46.9
46.7
46.5
46.3
46.1
45.9
45.7
45.5
ID#79
Time
(d)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
Cone
(M9/L)
117
116
116
115
115
114
114
113
113
112
112
111
111
110
110
109
109
109
108
108
107
107
106
106
105
105
105
104
ID#80
Time
(d)
1
3
5
7
9
11
13
15
17
19
21
23
25
27
29
31
33
35
Cone
(M9/L)
6.4
6.3
6.3
6.2
6.2
6.1
6.1
6.0
6.0
5.9
5.9
5.8
5.8
5.7
5.7
5.6
5.6
5.5
ID#81
Time
(d)
1
7
13
19
25
31
37
43
49
55
61
67
73
79
85
91
97
103
109
115
121
127
133
139
145
151
157
163
169
Cone
(M9/L)
84
80
77
74
78
75
72
69
66
64
61
59
57
55
53
51
49
47
45
43
42
40
39
37
36
34
33
32
31
ID#82
Time
(d)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Cone
(M9/L)
10.0
9.9
9.9
9.8
9.8
9.8
9.7
9.7
9.6
9.6
9.5
9.5
9.5
9.4
9.4
9.3
9.3
9.3
9.2
9.2
9.2
9.2
9.2
87
-------
Table
B2, Page 10.
ID #83
Time
(d)
1
3
5
7
9
11
13
15
17
19
21
23
25
27
29
31
33
35
37
39
Cone
(M9/L)
30
30
29
29
29
29
28
28
28
28
28
27
27
27
27
26
26
26
26
26
ID#84
Time
(d)
1
3
5
7
9
11
13
15
17
19
21
23
25
27
29
31
33
35
37
39
Cone
(M9/L)
2.0
2.0
2.0
1.9
1.9
1.9
1.9
1.9
1.9
1.8
1.8
1.8
1.8
1.8
1.8
1.8
1.7
1.7
1.7
1.7
ID#85
Time
(d)
1
3
5
7
9
11
13
15
17
19
21
23
25
27
29
31
33
35
37
39
Cone
(W3/L)
30
30
29
29
29
29
28
28
28
28
28
27
27
27
27
26
26
26
26
26
ID#86
Time
(d)
1
3
5
7
9
11
13
15
17
19
21
23
25
27
29
31
33
35
37
39
Cone
(W3/L)
2.0
2.0
2.0
1.9
1.9
1.9
1.9
1.9
1.9
1.8
1.8
1.8
1.8
1.8
1.8
1.8
1.7
1.7
1.7
1.7
ID#87
Time
(d)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Cone
(W3/L)
30
30
30
30
29
29
29
29
29
29
29
29
28
28
28
28
28
28
28
28
28
27
27
27
27
ID#87
Time
(d)
1
2
3
4
5
6
7
Cone
(W3/L)
148
127
120
112
105
98
88
ID#89
Time
(d)
1
2
3
4
5
6
7
Cone
(W3/L)
24.3
18.3
20.7
19.6
18.6
17.6
15.4
ID#90
Time
(d)
1
3
5
7
9
11
13
15
17
19
21
23
25
27
29
31
33
35
37
39
41
43
45
47
49
51
53
55
Cone
(W3/L)
207
170
148
130
114
99
87
76
67
58
51
45
39
34
30
26
23
20
18
15
14
12
10
9
8
7
6
5
-------
Table
B2, Page 11.
ID #95
Time
(d)
1
3
5
7
9
11
13
15
17
19
21
23
25
27
29
31
33
35
37
39
41
43
45
47
49
51
Cone
(M9/L)
20.1
19.5
19.0
18.6
18.2
17.8
17.4
17.0
16.6
16.3
15.9
15.6
15.2
14.9
14.6
14.2
13.9
13.6
13.3
13.0
12.7
12.4
12.2
11.9
11.6
11.4
ID#96
Time
(d)
1
3
5
7
9
11
13
15
17
19
21
23
25
27
29
31
33
35
37
39
41
43
45
47
49
51
Cone
(W3/L)
196
193
191
189
188
186
184
183
181
180
178
176
175
173
172
170
169
167
166
164
163
161
160
158
157
155
ID#97
Time
(d)
1
3
5
7
9
11
13
15
17
19
21
23
25
27
29
31
33
35
37
39
41
43
45
47
49
51
Cone
(W3/L)
2036
1986
1954
1922
1890
1859
1829
1799
1769
1740
1712
1684
1656
1629
1603
1576
1551
1525
1500
1476
1452
1428
1404
1381
1359
1337
ID#98
Time
(d)
1
42
Cone
(W3/L)
24.5
24.5
ID#99
Time
(d)
1
42
Cone
(W3/L)
50
50
ID#100
Time
(d)
1
42
Cone
(W3/L)
104
104
ID#101
Time
(d)
1
42
Cone
(W3/L)
248
248
89
-------
2145
90
------- |