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

-------