ArDuisur

United States?

Office of Water

?

EPA-822-R-99-014

Environmental Protection?

4304

?

December 1999

Agency

&EPA 1999 Update

of Ambient

Water Quality

Criteria

for

Ammonia

Supersedes 1998 Update

---

juipoota,okt

-ri

.: ..`

7

.

44,

.

:.....'.

1;T'

---

1999 Update of

Ambient Water Quality Criteria for Ammonia

September 1999

Supersedes 1998 Update

U.S. Environmental Protection Agency

Office of Water

Office of Science and Technology

Washington, D.C.

Office of Research and Development

Mid-Continent Ecology Division

Duluth, Minnesota

---

Notices

This update provides guidance to States and Tribes authorized to establish water

quality standards under the Clean Water Act (CWA), to protect aquatic life from acute

and chronic effects of ammonia. Under the CWA, States and Tribes are to establish

water quality criteria to protect designated uses. State and tribal decision makers

retain the discretion to adopt approaches on a case-by-case basis that differ from this

guidance when appropriate. While this update constitutes EPA's scientific

recommendations regarding ambient concentrations of ammonia that protect freshwater

aquatic life, this update does not substitute for the CWA or EPA's regulations; nor is it a

regulation itself. Thus, it cannot impose legally binding requirements on EPA, States,

Tribes, or the regulated community, and might not apply to a particular situation based

upon the circumstances. EPA may change this guidance in the future..

This document has been approved for publication by the Office of Science and

Technology, Office of Water, U.S. Environmental Protection Agency. Mention of trade

names or commercial products does not constitute endorsement or recommendation for

use.

Acknowledgment

This update is a modification of the

1998 Update of Ambient Water Quality Criteria for

Ammonia.

The 1999 modifications were written by Charles Delos and Russ Erickson.

The 1998 Update, most of the text of which was carried unchanged into the 1999

Update, was written by Charles Stephan, Russ Erickson, Charles Delos, Tom

Willingham, Kent Ballentine, and Rob Pepin (with substantial input from Alex Barron of

the Virginia Department of Environmental Quality, Dave Maschwitz of the Minnesota

Pollution Control Agency, and Bob Mosher of the Illinois Environmental Protection

Agency) under the auspices of the Aquatic Life Criteria Guideline Committee. Please

submit comments or questions to: Charles Delos, U.S. EPA, Mail Code 4304,

Washington, DC 20460 (e-mail: delos.charles@epa.gov).

iv

---

Contents

Page

Notices

Introduction

and

?

Acknowledgment

?

1

Overview of Ammonia Toxicology

?

2

Temperature-Dependence of Ammonia Toxicity

?

?8

pH-Dependence of Ammonia Toxicity

?

28

Formulation of the CMC

?

36

Review and Analysis of Chronic Data

?

43

Seasonality of Chronic Toxicity Endpoints

?

68

Formulation of the CCC

?

74

Chronic Averaging Period ?

81

The National Criterion for Ammonia in Fresh Water

?

83

References

?

89

Appendix 1: Review of Some Toxicity Tests

?

101

Appendix 2: Methods for Regression Analysis of pH Data

?

104

Appendix 3: Conversion of Results of Toxicity Tests

?

107

Appendix 4: Acute Values

?

109

Appendix 5: Histopathological Effects

?

118

Appendix 6: Results of Regression Analyses of Chronic Data

?

122

Appendix 7: Acute-Chronic Ratios ?

135

Appendix 8: A Field Study Relevant to the CCC

?

139

Appendix 9: Water-Effect Ratios ?

147

---

V1

---

Introduction

Since the U.S. EPA published "Ambient Water Quality Criteria for Ammonia - 1984"

(U.S. EPA 1985a), it has issued additional information on aquatic life criteria for

ammonia (Heber and Ballentine 1992; U.S. EPA 1989,1996, 1998). Also, results of

additional toxicity tests on ammonia have been published since 1985, which could

affect the freshwater criterion for ammonia. The purpose of the 1998 Update was to

revise the 1984/1985 ammonia criteria document (U.S. EPA 1985a) and replace Heber

and Ballentine (1992) and U.S. EPA (1996) by addressing selected important issues to

the extent possible in a short-term effort without additional research.

EPA obtained public comment on the 1998 Update. In response to those comments

EPA modified its criteria recommendations and prepared the current document, the

1999 Update, which replaces and supersedes the 1998 Update. A supporting

document,

Response to Comments on. 1998 Update of Ambient Water Quality Criteria

for Ammonia (U.S.

EPA 1999) presents the comments and EPA's responses.

This 1999 Update first presents an overview of ammonia toxicology in order to provide

the background needed to explain the revisions of the freshwater ammonia criterion.

Then the equations used in the older documents to address the temperature- and pH-

dependence of ammonia toxicity in fresh water are revised to take into account newer

data, better models, and improved statistical methods. Next, a CMC (acute criterion) is

derived from the acute toxicity data in the 1984/1985 criteria document, pH-normalized

using the new equations. Then, new and old chronic toxicity data are evaluated and

used to derive a CCC (chronic criterion). Lastly, the chronic averaging period is

addressed.

The 1999 Update differs from the 1998 Update primarily in the handling of the

temperature-dependency for the CCC, and therefore the formulation of the CCC, and

the expression of the national criterion. One facet of the chronic averaging period was

also slightly changed. Neither the 1998 Update nor the 1999 Update were intended to

address (1) the CMC averaging period, or (2) the frequency of allowed exceedances.

The Updates address only the freshwater criterion and do not affect the saltwater

criterion for ammonia (U.S. EPA 1989).

Concentrations of un-ionized ammonia and total ammonia are given herein in terms of

nitrogen, i.e., as mg N/L, because most permit limits for ammonia are expressed in

terms of nitrogen. CMCs and CCCs are given to three significant figures to minimize

the effect of round-off error in the calculation of permit limits.

---

Overview of Ammonia Toxicology

The 1984/1985 ammonia criteria document reviewed data regarding the dependence of

the toxicity of ammonia to aquatic organisms on various physicochemical properties of

the test water, especially temperature, pH, and ionic composition. A key factor in these

relationships is the chemical speciation of ammonia. In aqueous solution, ammonia

primarily exists in two forms, un-ionized ammonia (NH

3

) and ammonium ion (NH4+),

which are in equilibrium with each other according to the following expressions:

NH4 NH

3

+ H +

[NH NH 1

K -

[NH3][

The equilibrium constant K depends significantly on temperature; this relationship has

been described by Emerson et al. (1975) with the following equation:

pK = 0.09018 + 2729.92

where pK = -log i

oK and T is temperature in degrees Celsius.

From Equation 2, the definition of pK, and the definition

pH = -log1o[F11, the following expressions can be derived for the fraction of total

ammonia in each of the two forms:

1 + 10pK-pH

1

1 + 1004-pK

1

?

(4)

f?

NH3?NH4

+ f • =

1

The individual fractions vary markedly with temperature and pH. The pH-dependence

of the relative amounts of un-ionized ammonia and ammonium ion at 25°C, at which

pK=9.24, is illustrated in the following graph:

273.2 + T

?

(3)

fNI-13

fNH,

2

---

Chemical Speciation of Ammonia

ro

0

▪

0.1

ro

0

E.

W

O

?

0.01

0O

tj

0.001

ro

6? 7? 8? 9? 10

pH

Ammonia speciation also depends on ionic strength, but in fresh water this effect is

much smaller than the effects of pH and temperature (Soderberg and Meade 1991) and

is sufficiently small compared to the typical uncertainty in LC5Os that it will not be

considered here as a variable affecting ammonia toxicity. (As discussed later, ionic

composition might affect ammonia toxicity in ways other than its effect on ammonia

speciation).

These speciation relationships are important to ammonia toxicity because un-ionized

ammonia is much more toxic than ammonium ion. The importance of un-ionized

ammonia was first recognized when it was observed that increased pH caused total

ammonia to appear to be much more toxic (Chipman 1934; Wuhrmann and Woker

1948). It is not surprising that un-ionized ammonia is the more toxic form, because it is

a neutral molecule and thus is able to diffuse across the epithelial membranes of

aquatic organisms much more readily than the charged ammonium ion. Ammonia is

unique among regulated pollutants because it is an endogenously produced toxicant

that organisms have developed various strategies to excrete, which is in large part by

passive diffusion of un-ionized ammonia from the gills. High external un-ionized

ammonia concentrations reduce or reverse diffusive gradients and cause the buildup of

ammonia in gill tissue and blood.

Because of the importance of un-ionized ammonia, it became a convention in the

scientific literature to express ammonia toxicity in terms of un-ionized ammonia, and

water quality criteria and standards followed this convention. However, there are

reasons to believe that ammonium ion can contribute significantly to ammonia toxicity

under some conditions. Observations that ammonia toxicity is relatively constant when

expressed in terms of un-ionized ammonia come mainly from toxicity tests conducted at

pH>7.5. At lower pH, toxicity varies considerably when expressed in terms of un-

ionized ammonia and under some conditions is relatively constant in terms of

ammonium ion (Erickson 1985). Also, studies have established that mechanisms exist

for the transport of ammonium ion across gill epithelia (Wood 1993), so this ion might

contribute significantly to ammonia exchange at gills and affect the buildup of ammonia

3

---

-P

,,(

,,.

nmonia

4

0)

0

00

-..E.X0

0.1

0.01

0.001

–

–?

Un-iou

–

,:.

....... N ......

Ammonium

N.

ion N.\

6?

7?

8 ?

9?

10

in tissues if its external concentration is sufficiently high. Thus, the very same

arguments employed for the importance of un-ionized ammonia can also be applied in

some degree to ammonium ion. This is not to say that ammonium ion is as toxic as un-

ionized ammonia, but rather that, regardless of its lower toxicity, it can still be important

because it is generally present in much greater concentrations than un-ionized

ammonia.

Also, when expressed in terms of un-ionized ammonia, ammonia toxicity is usually not

constant with temperature, on average being about four-fold greater at 5°C than at

25°C for fish (Erickson 1985). Because the relative amount of ammonium ion is also

higher at low temperatures, this raises the possibility that ammonium ion might be in

part responsible for this temperature dependence. However, temperature might also

alter ammonia toxicity by affecting membrane permeability, endogenous ammonia

production, and other physiological processes.

Various authors have evaluated models that might explain the pH and temperature

dependence of ammonia toxicity. Tabata (1962) and Armstrong et al. (1978) suggested

that the observed pH dependence is due to joint toxicity of un-ionized ammonia and

ammonium ion.

The adjacent graph shows an

idealized picture of ammonia

toxicity assuming that (a)

ammonium ion and un-

ionized ammonia jointly

0

determine toxicity and (b) un-

ionized ammonia is 100 times

more toxic than ammonium

ion. At sufficiently high pH,

the more toxic un-ionized

ammonia comprises a

sufficiently large fraction of

total ammonia to dominate

Toxicity of Ammonia

toxicity, and so toxicity is

relatively constant when

?

pH

expressed in terms of un-

ionized ammonia. As pH

decreases, the relative amount of ammonium ion increases until it contributes

significantly to toxicity, so that toxicity expressed in terms of un-ionized ammonia

increases (i.e., it appears that less un-ionized ammonia is necessary to cause toxicity

because ammonium ion is responsible for some of the toxicity). At sufficiently low pH,

ammonium ion dominates toxicity, and so toxicity is relatively constant when expressed

in terms of either ammonium ion or total ammonia.

In contrast to this theory, Lloyd and Herbert (1960) suggested that the apparent effect

of pH on un-ionized ammonia toxicity is due to the data being plotted in terms of the pH

4

---

of the bulk exposure water rather than the pH at the gill surface. The release of carbon

dioxide 'at the gill lowers pH when pH is moderately alkaline, but has less effect when

pH is already low; this results in an apparent effect of pH on toxicity when the pH of the

bulk exposure water is used even if there is no such effect if the pH at the gill surface is

used. Szumski et al. (1982) suggested that this theory explained not only much of the

pH dependence of ammonia toxicity, but also the temperature dependence.

Erickson (1985) reviewed available information concerning the effects of pH and

temperature on acute toxicity of ammonia when expressed in terms of un-ionized

ammonia and tested its adherence to these theories. He concluded that effects

associated with pH changes at the gill could not account for the effect of temperature

and only a small part of the effect of pH. In contrast, the additive joint toxicity model

explained a large part of the dependence of ammonia toxicity on pH and predicted

important features of the data, specifically a slope of zero at high pH and a slope of one .

at low pH. The joint toxicity model could also be fit to the temperature data, but led to

values of the model parameters that were questionable because they indicated that

ammonium ion is as toxic or more toxic than un-ionized ammonia. Clearly, joint toxicity

could not possibly account for both pH and temperature effects, and Erickson (1985)

concluded that joint toxicity is likely responsible for much of the pH effect, but not for

the temperature effect. In the 1984/1985 criteria document, it was noted that the one

available dataset concerning the dependence of chronic toxicity on pH (Broderius et al.

1985) also suggested joint toxicity of un-ionized ammonia and ammonium ion.

Therefore, a major consideration in deriving the aquatic life criterion for ammonia is

whether the mathematical model used to describe pH dependence should be based on

joint toxicity theory: Since the 1984/1985 criteria document was issued, several

additional studies (Sheehan and Lewis 1986; Schubauer-Berigan et al. 1995; Ankley et

al. 1995; Johnson 1995) of the pH dependence of ammonia toxicity have provided more

information regarding the relative importance of un-ionized ammonia and ammonium

ion, including indications of more diversity of pH behavior among species than was

apparent in the data reviewed by Erickson (1985).

The report of Sheehan and Lewis (1986) requires special consideration here because

they suggest that the toxicity of ammonia at low pH is due to the effect of osmotic shock

on unacclimated organisms and that this has major implications for the derivation of a

criterion for ammonia. In their investigations concerning the pH-dependence of acute

ammonia toxicity to channel catfish, Sheehan and Lewis (1986) found that LC5Os

expressed in terms of un-ionized ammonia increased with increasing pH, but less so

than reported in most studies, although Tomasso et al. (1980) also reported little effect

of pH.



7 on un-ionized ammonia toxicity to the channel catfish. Sheehan and Lewis

noted that lethal concentrations at pH=6 were associated with very high total ammonia

concentrations (2000 mg N/L) and exhibited steeper concentration-effect curves than at

higher pH. They also reported that other salts were lethal at similar concentrations and

suggested that the toxicity of ammonia at low pH was due to the effect of osmotic shock

on unacclimated organisms rather than a specific action of the ammonium ion per se.

5

---

However, the implication of this work for the ammonia criterion is doubtful for the

following reasons:

1.

Any concern that the effects of high concentrations of ammonia would be less for

acclimated organisms is really not relevant. To be adequately protective, criteria

cannot assume that acclimation takes place, because if such high ammonia

concentrations are discharged, they would create a plume of high concentrations

compared to ambient levels. Organisms entering that plume would not be

acclimated to the high concentrations.

2.

It is doubtful that the effects of high salt concentrations observed by Sheehan and

Lewis were strictly due to osmotic effects. In their experiments, potassium chloride

caused higher mortality than the physiologically balanced salt they also used. In

fact, the toxicities of such salts vary quite widely, with potassium salts generally

being more toxic (Mount et al. 1997), probably due to effects of potassium beyond

any osmotic effects. Ammonium chloride also caused higher mortality than the

physiologically balanced salt, although this might be in part due to effects of un-

ionized ammonia.

3.

As part of their evidence for supporting osmotic effects as a toxic mechanism at low

pH, Sheehan and Lewis noted that the dose-response curves were steeper at low

pH, suggestive not only of a different mechanism, but one that is less variable

among organisms within a test. However, Broderius et al. (1985) found the

opposite effect of pH on dose-response curves.

4.

The LC5Os for channel catfish at low pH are generally much higher than those for

other fishes that have been tested at low pH. When expressed in terms of total

ammonia, the LC50 for channel catfish at pH=6 is four-fold higher than any other

LC50 reported for a fish species. For many other fishes, LC5Os at pH..6.5

represent salt concentrations of only a few hundred mg/L and less than a factor of

two greater than that of control water. A role of osmotic effects in such cases is

doubtful. Of all of the fish species tested, the pH curves for channel catfish show

the least indication for an effect of ammonium ion, so it is a very questionable

species upon which to base broad conclusions.

5.

In contrast to Sheehan and Lewis, Knoph (1992) reported no mortality of Atlantic

salmon at pH=6 in KCI or in physiologically balanced salt solutions with

concentrations equivalent to ammonium chloride solutions causing 45% mortality.

Similarly, Mount et al. (1997) found acute LC5Os for fathead minnows for various

salts and combinations (except those including potassium) to be at least several-

fold higher than the total ammonia LC5Os reported at pH=6.5 by Thurston et al.

(1981 b). Although for an invertebrate, the likely role of ammonium ion other than in

association with high salt concentrations is also evident in the daphnid data of

Tabata (1962) and Mount et al. (1997).

6.

Even if a different mechanism for toxicity exists at low pH, these tests still identify

concentrations that are unacceptably toxic and this is still joint toxicity in the broad

sense of the term. Although the joint toxicity might not be strictly additive, as would

be expected if the two forms of ammonia operate by the same mechanism, it is joint

toxicity nonetheless and should exhibit a similar pH dependence and be

considered in criteria derivation.

6

---

Although there is considerable reason to consider the effects of pH on ammonia toxicity

to be largely due to the joint toxicity of ammonium ion and un-ionized ammonia, pH can

have other effects on membrane function and other physiological processes that could

also alter ammonia toxicity, especially at very low and high pHs, and these are poorly

established. The state of knowledge for the pH dependence is incomplete in terms of

understanding specific mechanisms, variation among species, and interactions with

various physicochemical processes. Lacking a definitive, thorough theoretical

approach for describing pH effects, the most reasonable approach is to adopt the best

empirical description that can be obtained from available data. However, the shape of

this empirical equation can be guided by consideration of the evidence for the role of

speciation in ammonia toxicity.

The effects of temperature on ammonia toxicity are even less well understood, and

there is no adequate theoretical basis or scientific understanding for specifying how

temperature adjustments to the ammonia criterion should be made. Therefore, an

empirical approach will also be used for temperature dependence, as developed in the

next section.

As reviewed in the 1984/1985 ammonia criteria document, ammonia toxicity can also

depend on various aspects of the ionic composition of the exposure water, but the

effects were not clear and consistent enough to warrant inclusion of other variables in

the criterion. Although Soderberg and Meade (1992), Yesaki and Iwama (1992),

Ankley et al. (1995), Johnson (1995), Borgmann and Borgmann (1997), and Iwama et

al. (1997) have provided new data concerning interactions between various ions and

ammonia toxicity and excretion, there is still insufficient understanding and information

to account for these effects in the criterion and they will have to be addressed using

water-effect ratios or other site-specific approaches.

In summary, the available evidence indicates that the toxicity of ammonia can depend

on ionic composition, pH, and temperature. The mechanisms of these effects are

poorly understood, but the pH dependence strongly suggests that joint toxicity of un-

ionized ammonia and ammonium ion is an important component. For the reasons

presented above, the following approach will be used to account for these effects.

1.

Because the effects of ionic composition on ammonia speciation in fresh water are

small and its other effects on toxicity are poorly established, the ionic composition

of the exposure water will not be considered in the derivation of the criterion.

2.

Even though temperature can strongly affect the relative amounts of un-ionized

ammonia and ammonium ion, its effect on the toxicity of ammonia is not strongly

indicative of joint toxicity and will be described strictly by an empirical approach.

3.

The effect of pH will be described by equations that include basic features of joint

toxicity of un-ionized ammonia and ammonium ion, but with an empirical

component that recognizes the incomplete knowledge of these effects.

7

---

Temperature-Dependence of Ammonia Toxicity

This section presents the temperature analysis published in the 1998 Update, followed

by the re-analysis performed for this 1999 Update.

1998 Analysis of Temperature-Dependence

The 1984/1985 ammonia criteria document identified temperature as an important

factor affecting the toxicity of ammonia. When expressed in terms of

un-ionized

ammonia, the acute toxicity of ammonia was reported in the criteria document to be

inversely related to temperature for several species of fish, whereas limited data on

acute ammonia toxicity to invertebrates showed no significant temperature .

dependence. No direct data were available concerning the temperature dependence of

chronic toxicity. It was noted, however, that the differences between chronic values for

salmonid fish species tested at low temperatures and chronic values for warmwater fish

species tested at higher temperatures paralleled differences in acute toxicity known to

be caused by temperature.

In the 1984/1985 criteria document, an average temperature relationship observed for

fish was used to adjust fish acute toxicity data to a common temperature (20°C) for

derivation of the CMC for

un-ionized

ammonia; this same relationship was used to

extrapolate this CMC to other temperatures. (Invertebrate toxicity data were not

adjusted, but invertebrates were sufficiently resistant to ammonia that adjustment of

invertebrate data was not important in the derivation of the CMC.) This temperature

relationship for fish resulted in the un-ionized ammonia CMC being higher at warm

temperatures than at cold temperatures. Additionally, because of concerns about the

validity of extrapolating the temperature relationship to high temperatures, the un-

ionized ammonia CMC was "capped" to be no higher than its value at a temperature,

called TCAP, near the upper end of the temperature range of the acute toxicity data

available for warmwater and coldwater fishes. Similarly, the CCC was capped at a

temperature near the upper end of the temperature range of the available chronic

toxicity data.

Although the un-ionized ammonia criterion is lower at low temperatures, this does not

result in more restrictive permit limits for ammonia because the ratio of ammonium ion

to un-ionized ammonia increases at low temperatures, resulting in the total ammonia

criterion being essentially constant at temperatures below TCAP. In practice, however,

the criterion at low temperatures can be more limiting for dischargers than the criterion

at high temperatures because biological treatment of ammonia is more difficult at low

temperatures. Above TCAP, the constant un-ionized ammonia criterion results in the

total ammonia criterion becoming progressively lower with increasing temperature,

which can also result in restrictive discharge limitations.

8

---

Because more data are available at moderate temperatures than at lower and higher

temperatures, the ammonia criterion is most uncertain for circumstances when

compliance can be most difficult, either because of the low total ammonia criterion at

high temperatures or because of treatment difficulties at low temperatures. This

section examines the data used in the 1984/1985 criteria document and newer data to

determine (1) whether the use of TCAPs should be continued and (2) whether a lower

un-ionized criterion at low temperature is warranted. Data used include those analyzed

by Erickson (1985), which are shown in Figure 2 of the 1984/1985 document, and more

recent data reported by Arthur et al. (1987), DeGraeve et al. (1987), Nimmo et al.

(1989), and Knoph (1992).

Data not used include those reported by the following:

1.

Bianchini et al. (1996) conducted acute tests at 12 and 25°C, but one test was in

fresh water, whereas the other was in salt water.

2.

Diamond et al. (1993) conducted acute and chronic toxicity tests on ammonia at 12

and 20°C using several vertebrate and invertebrate species. When expressed in

terms of un-ionized ammonia, they reported that vertebrates (i.e., fishes and

amphibians) were more sensitive to ammonia at 12°C than at 20°C, whereas

invertebrates were either less sensitive or no more sensitive at 12°C, compatible

with the relationships used in the 1984/1985 criteria document. However, such

factors as dilution water and test duration varied between tests at different

temperatures and possibly confounded the results (see Appendix 1), raising doubts

about the temperature comparisons for the vertebrates and invertebrates.

Arthur et al. (1987) measured the acute toxicity of ammonia to several fish and

invertebrate species at ambient temperature during different seasons of the year. For

three of the five fish species (rainbow trout, channel catfish, and white sucker), the

relationship of toxicity to temperature was similar to that used in the 1984/1985 criteria

document. When expressed in terms of

un-ionized

ammonia, no clear relationship

existed between temperature and toxicity for the other fish species (fathead minnow

and walleye). This result for the fathead minnow is different from those of three other

studies (Reinbold and Pescitelli 1982a; Thurston et al. 1983; DeGraeve et al. 1987)

reporting a significant effect of temperature on the acute toxicity of un-ionized ammonia

to the fathead minnow. For five invertebrate species, each tested over a temperature

range of at least 10°C, there was no consistent relationship between temperature and

un-ionized ammonia toxicity. An initial report of these results (West 1985) was the

basis for no temperature adjustment being used for invertebrate data in the 1984/1985

criteria document. Further analysis of the Arthur (1987) data is discussed later.

DeGraeve et al. (1987) studied the effect of temperature (from 6 to 30°C) on the toxicity

Of ammonia to juvenile fathead minnows and channel catfish using acute (4-day) and

chronic (30-day) ammonia exposures. As shown for both fish species in Figure 1,

log(96-hr un-ionized ammonia LC50) versus temperature was linear within the reported

uncertainty in the LC50s; the slopes were similar to those reported in the 1984/1985

criteria document. Problems with the channel catfish chronic tests precluded effective

use of those data and the highest tested ammonia concentrations in the fathead

9

---

minnow chronic tests at 15 and 20°C did not cause sufficient mortality to be useful.

However, sufficient mortality did occur in the fathead minnow chronic tests at 6, 10, 25,

and 30°C. Based on regression analysis of survival versus log concentration

(discussed in more detail in the section concerning the CCC below), 30-day LC2Os for

un-ionized ammonia were 0.11, 0.18, 0.48, and 0.44 mg N/L at 6, 10, 25, and 30°C,

respectively. This temperature dependence (Figure 1) is similar to that for acute

toxicity and that used in the 1984/1985 criteria document. The actual effect of

temperature on these 30-day LC2Os is probably somewhat greater, because test pH

decreased with increasing temperature.

Nimmo et al. (1989) conducted acute toxicity tests on ammonia at 6 and 20°C in a well

water using Johnny darters and in a river water using both Johnny darters and juvenile

fathead minnows. In all three sets of tests, LC5Os expressed in terms of un-ionized

ammonia were significantly higher at the warmer temperature, by factors ranging from

3.5 to 6.2.

Knoph (1992) conducted acute toxicity tests at temperatures ranging from 2 to 17°C

using Atlantic salmon parr, one series of tests at pl-k6.0 and the other at pH-6.4. In

both series of tests, LC5Os expressed in terms of un-ionized ammonia increased

substantially with temperature.

Even with these additional data, the shape of the temperature relationship is not

completely resolved, especially for chronic toxicity. Nevertheless, the acute data for

fishes overwhelmingly indicate that ammonia toxicity, expressed in terms of

un-ionized

ammonia, decreases with increasing temperature.

Most importantly, the data of DeGraeve et al. (1987) show (Figure 1) that (a) a linear

relationship of log un-ionized ammonia LC50 versus temperature applies within the

reported uncertainty in the LC5Os over the range of 6 to 30°C and (b) temperature

effects on long-term mortality are similar to those on acute mortality. For invertebrates,

acute toxicity data suggest that ammonia toxicity, when expressed in terms of un-

ionized ammonia, does not decrease, and possibly even increases, with increasing

temperature. Quantifying and adjusting data for this relationship is not necessary

because even at warm temperatures invertebrates are generally more resistant to acute

ammonia toxicity than fishes and thus their precise sensitivities are of limited

importance to the criterion. At low temperatures, they are even more resistant relative

to fishes and thus their precise sensitivity is even less important to the criterion.

Based on this information, the two issues raised above were resolved as follows:

1.

TCAPs will not be used in the ammonia criterion. This does not mean that the

notion of high temperature exacerbating ammonia toxicity is wrong; rather, it

reflects the fact that such an effect is not evident in the available data, which cover

a wide temperature range.

2.

A CMC, if it were expressed as un-ionized ammonia (rather than total ammonia,

used in this document) would continue to be lower at lower temperatures,

consistent with the observed temperature dependence of ammonia toxicity to the

10

---

Fathead Minnow

96-hr LC50

4

0)

2

1

c

0

0.8

0.6

a)

0.4

0

0.2

0.1

4

2

0

1

0.8

0.6

0

a

N

)?

0.4

0

0.2

0.1

1

1

Channel

96-hr

11

Catfish

LC50

11

2

1

0.8

rn?

0.6

'E?

0.4

E

a)?

0.2

0

T

0.1

0.08

0.06

Fathead Minnow

30-day LC20

Figure 1. The effect of temperature on ammonia toxicity in terms of un-ionized ammonia

(DeGraeve et al. 1987). Symbols denote LC5Os or LC2Os and 95% confidence

limits and lines denote linear regressions of log LC versus temperature.

0?

5 10 15 20 25 30

?

0?

5 10 15 20 25 30

?

0?

5 10 15 20 25 30

Temperature (C)

?

Temperature (C)?

Temperature (C)

---

most sensitive species, i.e., fishes. Although it is possible that the temperature

relationship differs among fish species and that using the same relationship for all fish

species introduces some uncertainty, specifying a relationship for each fish species is

not possible with current data and would also introduce considerable uncertainty.

Therefore, for a criterion expressed in terms of un-ionized ammonia, available data

support the continued use of a generic temperature relationship similar to that in the

1984/1985 ammonia criteria document, but without TCAPs.

This raised a new issue, however, because the criterion expressed in terms of total

ammonia is nearly constant over all tested temperatures, and the small effect of

temperature on the total ammonia criterion in the 1984/1985 criteria document is

largely an artifact of conducting regression analyses in terms of un-ionized ammonia

and is not indicative of any established, significant trend. It was thought that the

expression and implementation of the ammonia criterion might have been simplified if

temperature were dropped as a modifying factor, which might have been possible if

ammonia toxicity is expressed in terms of total ammonia. Furthermore, permit limits

and compliance are usually expressed in terms of total ammonia nitrogen, and so

expressing the criterion in terms of total ammonia nitrogen would simplify its

implementation by eliminating conversions to and from un-ionized ammonia. Because

of such benefits and because there are no compelling scientific or practical reasons for

expressing the criterion in terms of un-ionized ammonia, the freshwater toxicity data

concerning temperature dependence were reanalyzed in terms of total ammonia

nitrogen.

The data analyzed are from the studies included in the 1984/1985 ammonia criteria

document and the studies of DeGraeve et al. (1987), Nimmo et al. (1989), and Knoph

(1992). All analyses were conducted in terms of total ammonia nitrogen, either as

reported by the authors or as converted by us from reported values for un-ionized

ammonia, pH, and temperature using the speciation relationship of Emerson et al.

(1975). The data are presented in Figure 2 and show considerable diversity, with some

datasets showing decreasing toxicity with increasing temperature, some showing

increasing toxicity, and some showing virtually no change. There are even differences

among studies using the same test species. However, in no case is the effect of

temperature particularly large, being no more than a factor of 1.5 over the range of any

dataset, except for the Johnny darter data of Nimmo et al. (1989). In some studies, test

pH was correlated with test temperature. To reduce the confounding effect of pH, the

total ammonia LC50 was adjusted to the mean pH of the data for the study using the pH

relationship discussed in the next section of the Update. These adjusted data are

shown in Figure 3 and also show neither large effects nor any clear consistency among

or within species or studies.

For each dataset containing at least three data points, a linear regression of log LC50

versus temperature was conducted (Draper and Smith 1981) and the resulting

regression lines are plotted as solid lines in Figures 2 and 3. These regressions are

significant at the 0.05 level for only one dataset (the unadjusted fathead minnow data of

Thurston et al. 1983); for this dataset, however, the regression is not significant when

12

---

Atlantic Salmon

(Knopf 1992)

Channel Catfish

?30

?

Channel Catfish

?

Rainbow Trout

?

600

—?

(Cary 1976)

?

20

?(Colt

& Tchobanoglous 1976)

40

- (Ministry of Tech 1968)

?

400

10

8

20

15

6

O

....

200150

4 -

10

8

O

100

80

A

?

15

?

30

?

300

A

?

O

6

10?

20

?

30

?

10?

20?

30

?

1

0

?20

?

30

?

10

?

20

?

30

80 -?

Rainbow Trout

(Thurston and Russo 1983) 100

60

80

150

Fathead Minnow

(Thurston et al. 1983)

100

O

80

40

60

60

30

40

40

0

20

30

30

20

20

15

Bluegill

Johnny Darter

(Reinbold & Pescitelli 1982a) 40

?

(Nimmo et al. 1

0

0

?

10

?

20?

30

Channel Catfish

(DeGraeve et al. 1987)

60 _

?

Fathead Minnow

(Nimmo et al. 1989)

40

30

•

•? •?• •?•

. . .

20 -

15

15

10

10?

20

?

30

10 —

30

I?

?

8

1

0

0?

10?

20?

30

?10

?20

(Reinbold & Pescitelli 1982a) 100 —

?

(Hazel et at. 1971)

40 -?

80 -

200

30 - 410?

60

? ■

150

20

?

..-...-2_T__

?

40?

■

100

10 —?

20

8?

1?

1?

I

?

15 ?

1?

I?

I?

40

0?

10

?

20?

30?

0?

10?

20?

30

•

60 -

?

Fathead Minnow?

Striped Bass

300

15?

•

?

30?

80

NI?

60

0

I

-

Figure 2. The effect of temperature on acute ammonia toxicity in terms of total ammonia.

Symbols denote LC50s, solid lines denote regressions for individual datasets,

and dotted lines denote pooled regressions over all datasets.

Largemouth Bass?

40

(Roseboom & Richey 1977)

?

30

20

Rainbow Trout?

40

_ (Reinbold & Pescitelli 1982a) 30

20

15

•a7

10

15

10

8

?

8

10 —

8 -

?

6

?

6

-

—.

-

10?

20

?

30

Three-Spined Stickleback

(Hazel et al. 1971)

•?

•

A

300

200

150

100

80

60

40

0

10

?

20

?

30

_

?

Fathead Minnow

(DeGraeve et al. 1987)

—?

v

1

'400

300

200

150

100

80

60

0

10

20

30

10

20

30

TEMPERATURE (C)

13

:-..I

-2

cr)

E

?

40 -?Bluegill

? 80 -?

Channel Catfish?

60

?

(Roseboom & Richey 1977)

?

(Roseboom & Richey 1977)

0

LO?

30 -?

60 -?

440

0_I

20

?

40?

30

E

< 15 -----------

?

30 ?

20

o

10 —?

20 ,■-

.-ir

15

2?

a

?

15 -

<

_J?

6 -

10

I?

1

?

1

0?

10?

20

?

30

?

0?

10

?

20

?

30

I--

10

20

30

---

Figure 3. The effect of temperature on pH-adjusted acute ammonia toxicity in terms of

total ammonia. LC5Os are adjusted to the mean pH of the dataset based on the

pooled relationship of acute toxicity to pH. Symbols denote LC50s, solid lines

denote regressions for individual datasets, and dotted lines denote pooled

regression over all datasets.

30

600

Atlantic Salmon

0 (Knopf 1992)

Rainbow Trout

(Ministry of Tech 1968)

Channel Catfish

—?

(Cary 1976)

Channel Catfish

(Colt S Tchobanoglous 1976) 40

80 –

?

Rainbow Trout

60

(Thurston and Russo 1983) 10080

Fathead Minnow

150

(Thurston et al. 1983)

20

400

••?

•

30

1 00

a

15

i■

S?V

v

300

8 0

O

40

60

z



ln.

10?

20

?

30

10

c,

?

20

200

;No•0

8

60

30

40

15

O

150

40

20

.30

30

15

6

10

8—

100

10

?

20

?

30?

6

0

?

?

10

?

20

?

30

20

15

0

4

80

I

10

20

?

30

?

0?

10?

20

20

10

30

0

1

0

20

30

Channel Catfish

?

60

(Roseboom & Richey 1977)

40

Bluegill

?

Johnny Darter

(Reinhold & Pescitelli 1982a) 40

?

(Nimmo et al. 1989)

40 –

?

Rainbow Trout

?

40

30

?

_ (Reinhold & Pescitelli 1982a)

30

Largemouth Bass

(Roseboom & Richey 1977)

Bluegill

?

80

(Roseboom & Richey 1977) 60

40

30

30

20

40

30

20

20

O

O

20

15

30

20

15

15

O

15 –

O

15

10

20

10

108

10

6

8

15

10

8

8

6

10 ?

8

6

6

10

20

30 0

10?

20

?

30

0

?

10

?

20?

30

10

?

20

?

30

10

?

20

?

30

0 10

20 30

Channel Catfish

(DeGraeve et al. 1987)

Fathead Minnow

(DeGraeve et al. 1987)

Three-Spined Stickleback 300

(Hazel et al. 1971)

Fathead Minnow

(Nimmo et al. 1989)

Striped

Base

—?

(Hazel et al. 1971)

Fathead Minnow

60

(Relnbold & Pescitelli 1982a) 100

60

40

30

40

80

200

30

60

150

•

20

40

100

20

15

30

80

15

•

60

10

8

—

?

2015

108

0

10?

20?

30?

0?

10

?

20 30

0

10

?

20?

30

0

10

?

20?

30

TEMPERATURE (C)

10?

20?

30 0

10 20

30?

0

14

---

the data are adjusted for the fact that pHs were lower in the low-temperature tests than

in the high-temperature tests. Slopes from regression analyses of datasets in Figure 3

range from -0.015 to 0.013, compared to a range from 0.015 to 0.054 when expressed

in terms of un-ionized ammonia (Erickson 1985). This narrower range of slopes in

terms of total ammonia nitrogen also argues for use of total ammonia, rather than un-

ionized ammonia, because there is less uncertainty associated with the generic

relationship. For datasets with just two points, Figures 2 and 3 also show the slopes for

comparative purposes. Based on the typical uncertainty of LC50s, these slopes also

would not be expected to be significant, except perhaps for the Johnny darter data of

Nimmo et al. (1989).

A multiple least-squares linear regression (Draper and Smith 1981) using all datasets

(with a common slope for all datasets and separate intercept for each dataset) was

conducted, both with and without pH adjustment. The results of these pooled analyses

are plotted as dotted lines in Figures 2 and 3 to show that the residual errors for the

common regression line compared to the individual regression lines are not large

relative to the typical uncertainty of LC50s. To better show the overall fit of the

common regression line, the data are also plotted together in Figure 4 by dividing each

point by the regression estimate of the LC50 at 20°C for its dataset. This normalization

is done strictly for data display purposes because it allows all of the datasets to be

overlaid without changing their temperature dependence, so that the overall scatter

around the common regression line can be better examined. The data show no

obvious trend, with the best-fit slope explaining only 1% of the sum of squares around

the means for the pH-adjusted data and 0% for the unadjusted data. The one available

chronic dataset (DeGraeve et al. 1987) also shows no significant temperature effect

when expressed in terms of total ammonia nitrogen (Figure 5) and adjusted for pH

differences among the tests. (These tests and the calculation of the LC2Os are

discussed in detail later.)

Based on the small magnitude and the variability of the effect of temperature on total

ammonia acute and chronic toxicity values for fish, the 1998 Update did not include

temperature as a modifying factor for a total ammonia criterion. For invertebrates, it

should be noted that the 1998 Update's assumption that temperature had no effect on

the toxicity of

total

ammonia differs from the 1984/1985 criteria document's assumption

that temperature has no effect on the toxicity of

un-ionized

ammonia. This

inconsistency is resolved during the 1999 re-examination of data, to be discussed

shortly, by incorporating a relationship between temperature and total ammonia toxicity

to invertebrates. That relationship, however, does not affect the CMC because

invertebrates are not among the acutely sensitive taxa.

The amount of uncertainty in this approach to the CMC can be demonstrated to be

small by considering how the criterion would differ if total ammonia toxicity was

adjusted based on the slopes in various datasets. Because the bulk of the toxicity data

used in the derivation of the criterion is within a few degrees of 20°C, the temperature

relationship used has very little effect on the criterion near this temperature, but rather

has the greatest effect on the criterion at much higher or lower temperatures. If the

15

---

3

2

(ZD

V T

0?

AV:AV)

.

•!

- --.1

.

0 4.1‘

.Aik

•

1 —• • Ow

?

a--7(4yr

• ,

•

•

0.6 -

1.5

41!>. o

z,„t

0.8?

-

L u ■

Not pH Adjusted •

O

o

Figure 4. The effect of temperature on normalized acute ammonia toxicity in terms of total

ammonia. Data were normalized by dividing measured LC5Os by regression

estimates of LC5Os at 20°C for individual datasets for Figure 2 (top plot) and

Figure 3 (bottom plot).

0.4 -

0

?5

?10

?15

?

20

?

25?

30

3

?

pH Adjusted

0.4

ChannelCatfish

Rainbow Trout

C)?

Bluegill

ChannelCatfish

LargemouthBass

-?

Rainbow Trout

Bluegill

•

Fathead Minnow

■?

Striped Bass

Stickleback

Fathead Minnow

ChannelCatfish

AtlanticSalmon

O?

Johnny Darter

•

Fathead Minnow

w

•

0

?

5?

10?

15?

20

?

25

?

30

TEMPERATURE (C)

16

---

Data Adjusted to pH 7.5

.

?

•

?

•

50

20

10

100

5

100

50

20

10

5

Data Not

pH Adjusted

Figure 5. The effect of temperature on chronic ammonia lethality to fathead minnows in

terms of total ammonia (DeGraeve et al. 1987). Symbols denote LC2Os and 95%

confidence limits and lines denote linear regressions of log LC versus

temperature. Figure on left is for estimated LC5Os at test pH and figure on

right is for LC5Os adjusted to pH=7.5 based on pooled relationship of

chronic toxicity to pH.

0?

5?

10?

15

?

20?

25?

30?

0?

5

?

10

?

15?

20?

25?

30

Temperature (C)

?

Temperature (C)

17

---

average slope for the pH-adjusted acute data from Figure 4 is used, the total ammonia

CMC at 5°C would be only about 6% higher than at 20°C. The smallest and largest

slopes from the acute regressions for individual species in Figure 3 would produce a

range from 40% lower to 68% higher at 5°C than at 20°C, but this greatly overstates

the uncertainty because effects on a CMC derived from many datasets should not be

near these extremes.

1999 Re-examination of Temperature Dependence – Acute Toxicity

The previous section, reproduced with relatively few changes from the 1998 Update,

included an analysis of available data on the temperature dependence of acute

ammonia toxicity to fish. These data (in Figures 2, 3, and 4) consisted of 20 different

data sets drawn from 11 different studies and included nine different species, four

of

these species being in more than one study. Data from Arthur et al. (1987) were not

used in the 1998 analysis because those authors reported concerns about factors

confounding temperature in their data set. Linear regression analysis of log LC50 (total

ammonia basis) versus temperature was conducted on each data set, both with and

without correcting for pH as a confounding factor. No consistent trend with temperature

was observed and only one data set showed a slope different than zero at the 0.05

level of statistical significance. Therefore, a pooled linear regression analysis was

conducted across all data to derive an average slope, which was very close to zero and

also not statistically significant. On the basis of this analysis, the 1998 Update did not

include any temperature dependence for criteria to protect fish from acute ammonia

toxicity.

In response to public comment (U.S. EPA, 1999), the 1998 analysis was re-examined.

This re-examination indicated that it is appropriate to handle the temperature

dependencies of fish and invertebrates separately. For invertebrates, the inclusion of

the Arthur et al. (1987) data in the regression analysis, yields a change in the

temperature dependency that is ultimately reflected in the difference between the 1999

CCC and the 1998 CCC.

In the 1998 Update, EPA did not use the. Arthur et al. (1987) data because of those

authors concerns that other variable factors in their tests, conducted during different

seasons, might have had a potential to confound their results. In re-examining their

data in response to comments, however, EPA found that most of the

fish

data from

Arthur et al. showed behavior similar to that from numerous other investigators: that is,

little relationship with temperature when expressed as total ammonia. Consequently, it

was concluded that the other variable factors were unlikely to be confounding the

results.

For fish, although the temperature dependency is unchanged from 1998, additional

documentation is provided here, primarily because. the apparent difference between

fish acute and chronic temperature dependencies is now used in the projection of the

invertebrate chronic temperature dependency.

18

---

First presented here will be more details on the regression analyses of the individual

data sets conducted for the Update, plus similar analyses of the data of Arthur et al. A

linear regression was conducted on each data set using the equation:

log(LC50T) = log(LC5020)

+ S • (T - 20)

?

(5)

where LC5OT

is the total ammonia LC50 at temperature T, S is the slope of log LC50

versus temperature, and LC5020

is the estimated total ammonia LC50 at 20°C. For

completeness, this effort included data sets with just two points, although the

regression analysis then provides a perfect fit and has no residual error, so that

confidence limits, significance levels, etc. cannot be evaluated using normal methods.

In such cases, the mean squared error (MSE) of data around the regression was

assumed to

.

be equal to the weighted mean residual MSE for the larger data sets, so

that approximate significance levels could be determined.

Fish acute data:

Table 1 presents the results of the regression analysis for each data set, with data

adjusted to pH=8 based on the average pH relationship used in the 1998 and 1999

Updates. Plots of these relationships (except for Arthur et al. 1987) are in Figure 3 in

the previous section.

Of the 24 entries in Table 1, nearly half (11) have very small slopes of between -0.006

and +0.006, a range which corresponds to a factor of 1.3 change or less in LC50 for a

20°C temperature change and is less than normal data variability. Of these 11, five

have positive slopes and six have negative slopes. Of the 13 entries with steeper

slopes, five have positive slopes and eight have negative slopes. Among the data sets

used in the Update, only two of the regressions are statistically significant at the 0.05

level, one with a negative slope and one with a positive slope, although two other sets

(for fathead minnows from DeGraeve et al. 1987 and Reinbold and Pescitelli 1982) are

close to being significant. (The level of significance for the Johnny darter data set

differs from what was reported in the previous section because it consists of two

different sets which were analyzed separately in the 1998 analysis, but combined here

because they were not significantly different.) Of the five data sets from Arthur et al.

(1987), only one is significant at the 0.05 level. For species with more than one entry,

slopes vary considerably. This general lack of statistical significance and consistency

precludes any reliable assessments based on these individual analyses.

The 1998 Update therefore conducted a pooled regression analysis to determine

whether the combined acute toxicity data sets indicated any significant average trends

with temperature. Table 2 summarizes the mean trends determined in various pooled

analyses. The first entry is the pooled analysis conducted for the 1998 Update, which

included all the data in Table 1 above except the fish data of Arthur et al. (1987). The

slope from this pooled analysis was very small (-0.0023), and was not statistically

significant despite the large number of data. The second entry adds the fish data of

19

---

Table 1.?

Results of regression analysis of logLC50 (mg/L total ammonia) versus temperature (°C) for

individual data sets on the temperature dependence of acute ammonia toxicity.

Reference/

Slope

logLC5020

Residual so

F„GR

Species

(95% CL)

(95% CL)

(r2)

(a)

Thurston et al. (1983)

-0.0014

1.641

0.112

0.06

Fathead Minnow

(-0.013,+0.013)

(1.582,1.700)

(<1%)

(0.81)

Thurston and Russo (1983)

+0.0059

1.350

0.121

0.30

Rainbow Trout

(-0.017,+0.029)

(1.204,1.495)

(2%)

(0.59)

Cary (1976)

+0.0028

1.676

0.093

0.32

Channel Catfish

(-0.008,+0.013)

(1.593,1.758)

(2%)

(0.58)

Colt and Tchobanoglous (1976)

+0.0004

1.604

0.016

0.02

Channel Catfish

(-0.037,+0.038)

(1.350,1.858)

(2%)

(0.91)

Ministry of Technology (1967)

+0.0008

1.231

0.051

0.03

Rainbow Trout

(-0.018,+0.019)

(1.010,1.452)

(1%)

(0.88)

Roseboom and Richey (1977)

+0.024

1.089

-

0.95

Bluegill Sunfish

(-0.025,+0.073)

(0.803,1.375)

•

?

(0.33)

Roseboom and Richey (1977)

+0.020

1.482

-

0.68

Channel Catfish

(-0.029,+0.069).

(1.196,1.768)

(0.41)

Roseboom and Richey (1977)

-0.0029

1.237

-

0.02

Largemouth Bass

(-0.040,+0.034)

(0.972,1.502)

(0.88)

Reinbold and Pescitelli (1982)

-0.0088

1.396

0.088

1.63

Rainbow Trout

(-0.028,+0.010)

(1.159,1.632)

(29%)

(0.27)

Reinbold and Pescitelli (1982)

-0.0004

1.370

0.128

0.00

Bluegill Sunfish

(-0.027,+0.026)

(1.059,1.681)

(0%)

(0.96)

Reinbold and Pescitelli (1982)

-0.0153

1.429

0.076

16.6

Fathead Minnow

(-0.031,+0.009)

(1.243,1.615)

(89%)

(0.06)

Hazel et al. (1971)

-0.0163

1.274

0.076

2.93

Striped Bass

(-0.057,+0.025)

(1.105,1.443)

(60%)

(0.23)

Hazel et al. (1971)

-0.0106

1.390

0.081

1.14

Three-Spined Stickleback

(-0.053+0.032)

(1.214,1.567)

(36%)

(0.40)

DeGraeve et al. (1987)

-0.0052

1.617

0.061

3.33

Fathead Minnow

(-0.012,+0.002)

(1.563,1.670)

(36%)

(0.12)

DeGraeve et al. (1987)

-0.0088

1.648

0.061

9.76

Channel Catfish

(-0.016,-0.002)

(1.595,1.701)

(62%)

(0.02)

Knopf (1992)

-0.0035

1.715

0.097

0.22

Atlantic Salmon

(-0.027,+0.020)

(1.406,2.025)

(7%)

(0.067)

Knopf (1992)

+0.0163

1.636

0.054

5.18

Atlantic Salmon

(-0.075,+0.108)

(0,405,2.866)

(84%)

(0.26)

Nimmo et al. (1989)

+0.021

1.463

0.072

18.1

Johnny Darter

(+0.000,+0.043)

(1.248,1.678)

(90%)

(0.05)

Nimmo et al. (1989)

+0.0070

1.568

-

0.42

Fathead Minnow

(-0.014,+0.028)

(1.353,1.782)

(0.52)

Arthur et al. (1987)

-0.032

1.762

0.105

24.8

Fathead Minnow

(-0.059,-0.004)

(1.493,2.030)

(92%)

(0.04)

Arthur et al. (1987)

-0.0100

1.348

0.158

0.56

Rainbow Trout

(-0.053,+0.033)

(0.937,1.758)

(16%)

(0.51)

Arthur et al. (1987)

-0.0058

1.558

0.030

5.15

Channel Catfish

(-0.038,+0.027)

(1.230,1.886)

(84%)

(0.26)

Arthur et al. (1987)

+0.0007

1.902

0.048

0.01

White Sucker

(-0.23,+0.25)

(1.657,2.147)

(1%)

(0.92)

Arthur et al. (1987)

-0.038

1.216

0.306

2.84

Walleye

(-0.327,+0.250)

(-1.911,4.343)

(74%)

(0.34)

20

---

Arthur et al.; it does result in a statistically significant trend. The mean slope (-0.0058)

is still small, but does amount to a 23% decrease in LC50 per 20°C increase in

temperature. However, this slope is heavily influenced by two points with high residual

(>3a) deviations. One of these points is a test at 3.4°C by Arthur et al. (1987) with

fathead minnows, which showed much greater effects of low temperature than other

studies with the same species. The other point is for a test at 22.6°C by Arthur et al.

(1987) with walleye, which showed very high sensitivity and was part of a set of three

tests which used fish from different sources, potentially confounding the temperature

effects. Without these two data, the regression has an even lower slope and is not

significant at the 0.05 level (third entry in Table 2). Overall, these analyses of the fish

acute data suggest a weak overall trend of higher LC5Os at low temperatures, with a

logLC50 versus temperature slope in the -0.002 to -0.006 range, but of questionable

statistical significance.

It is also useful to consider separately the overall trends for different fish species.

Table 1 includes multiple studies with fathead minnows, rainbow trout, channel catfish,

and bluegill sunfish. Table 2 includes the results of pooled analyses for each of these

species, both with and without data from Arthur et al. (1987). For rainbow trout,

bluegill, and channel catfish, the regressions were not statistically significant. The

bluegill data indicated virtually no temperature effect, whereas weak trends similar to

the pooled analyses over all data sets were suggested in the channel catfish data

(slope = -0.0030 without and -0.0034 with Arthur et al. data) and rainbow trout data

Table 2.?

Results of regression analysis of log LC50 (mg/L total ammonia) versus temperature (°C)

for pooled data sets on the temperature dependence of acute ammonia toxicity to fish.

Data Sets Pooled

Slope

(95% CL)

Residual SD

(r2)

FREGR

(a)

All Data excluding Arthur et al.

-0.0023

0.105

1.79

(-0.0057,+0.0011)

(2%)

(0.18)

All Data including Arthur et al.

-0.0058

0.122

10.3

(-0.0094,-0.0022)

(8%)

(<0.01)

All Data including Arthur et al. except 'Outliers"

-0.0030

0.105

3.52

(-0.0063,+0.0002))

(3%)

(0.06)

Fathead Minnow excluding Arthur et al

-0.0063

0.106

4.76

(-0.0122;0.0005)

(11%)

(0.04)

Fathead Minnow including Arthur et al.

-0.0105

0.120

13.4

(-0.0169,-0.0049)

(25%)

(<0.01)

Fathead Minnow including Arthur et al. exc "Outlier"

-0.0073

0.106

6.85

(-0.0129,-0.0017)

(15%)

(0.01)

Rainbow Trout excluding Arthur et al.

-0.0013

0.109

0.06

(0.0122,+0.0096)

(<1%)

(0.80)

Rainbow Trout including Arthur et al.

-0.0034

0.115

0.51

(-0.0133,+0.0064)

(2%)

(0.48)

Channel Catfish excluding Arthur et al.

-0.0030

0.088

1.05

(-0.0091,+0.0031)

(4%)

(0.32)

Channel Catfish including Arthur et al.

-0.0034

0.085

1.64

(-0.088,+0.021)

(6%)

0.21)

Bluegill Sunfish

+0.0006

0.120

0.01

(-0.0172,+0.0184)

(<1%)

(0.92)

21

---

(slope = -0.0014 without and -0.0034 with Arthur et al. data). For fathead minnow, the

pooled analyses were statistically significant and stronger, with slopes ranging from -

0.0063 to -0.0105 depending on the treatment of data from Arthur et al. Such slopes

for fathead minnow would result in moderate effects over a broad temperature range: a

20°C decrease in temperature would result in a 33% to 62% increase in LC50.

However, this species is not sensitive enough that this would affect the acute criterion

values. For the species used in the acute criterion calculations, no temperature

correction for acute toxicity is appropriate due to the lack of any significant trend over

all data sets.

Invertebrate acute data:

Unlike fish, available acute toxicity data for invertebrates indicates that their acute

sensitivity to ammonia decreases substantially with decreasing temperature. The 1998

Update noted this temperature dependence, but did not present any analysis of it

because tested invertebrates were sufficiently tolerant to acute ammonia exposures

that this dependence would not affect the acute ammonia criterion. The 1998 Update

also noted that this temperature dependence should be a consideration in setting low

temperature chronic criterion, but did not provide any specific analysis regarding this.

This section will provide an analysis of available information on the temperature-

dependence of invertebrate

acute

ammonia toxicity, to be used later for estimating the

temperature-dependence of

chronic

ammonia toxicity.

Arthur et al. (1987) provide the only available data on the temperature dependence of

acute ammonia toxicity to invertebrates. As noted earlier, these toxicity tests did not

specifically test temperature effects, but rather were seasonal tests in which various

chemical characteristics of the tests water varied as well as temperature. Test

organisms were whatever were available in outdoor experimental streams at the time of

the test, so the size, life stage, and condition of the organisms also varied. The authors

of this study expressed some doubt as to how much of the effects they observed were

actually due to temperature. However, for invertebrates, they did observe strong

correlations of total ammonia toxicity with temperature. Confounding factors might

contribute somewhat to this correlation, but temperature is still likely the primary

underlying cause. If other factors were largely responsible for the apparent effects of

temperature, it would be expected that strong correlations with temperature would also

be evident in their fish data. However, as discussed above, the fish data usually

showed much weaker effects of temperature, similar to other studies with fewer

confounding factors.

These invertebrate acute data were analyzed using the same regression model and

techniques as discussed above for fish. The study of Arthur et al. (1987) included data

sets for nine invertebrate species, but two of these sets were not included in the

analysis because they consisted of two tests at temperatures only 3°C apart. For the

other species, the number of tests ranged from 2 to 6, with temperature ranges of from

9°C to 21 °C. Table 3 summarizes the regression results for the data sets of each

species and for pooled analyses conducted on (a) all seven species, and (b) three

22

---

species that had more than two tests and a temperature range of at least 15°C. All

data were corrected to pH 8 based on the average acute pH relationship (described

later). All species show substantially greater tolerance to ammonia at lower

temperatures, and in most cases the significance level of the regression is better than

0.05. (As for the analysis of the fish data, when there were just two tests for a species,

the significance level for the individual analysis is based on the MSE from the pooled

analysis.) The slope of log LC50 versus temperature does not vary widely, ranging

from -0.028 to -0.046 and being -0.036 for both pooled analyses. Figure 6 provides

plots of this data and the regression lines comparable to those for fish previously

shown in Figures 3 and 4.

Again, because invertebrates are not among the species acutely sensitive to ammonia,

the invertebrate acute temperature slope does not affect the formulation of the acute

criterion. It will be used subsequently, however, in formulating the invertebrate

chronic

temperature slope, which ultimately will affect the formulation of the chronic criterion.

1999 Re-examination of Temperature Dependence – Chronic Toxicity

Fish chronic data

As in the 1998 Update, the only available data on the temperature dependence of

chronic ammonia toxicity are from the study by DeGraeve et al. (1987) on survival of

juvenile fathead minnows during 30-day exposures to ammonia at temperatures

ranging from 6°C to 30°C. In contrast to acute toxicity, which for fathead minnows

showed sensitivity to be slightly

reduced at

low temperatures, this data on chronic

toxicity suggested

greater

sensitivity at low temperatures. However, this trend was

small, at least once the confounding effect of pH was corrected for, and not statistically

significant. Based on this analysis, the 1998 Update treated effect concentrations for

chronic ammonia toxicity to fish as it did for acute toxicity: as being invariant with

temperature. However, the 1998 Update also noted that, if seasonal variations in

temperature cause a shift in what endpoints the criterion should be based on, the

chronic criterion could have a seasonal temperature dependence even if effect

concentrations for specific chronic endpoints do not vary with temperature. (This is

discussed in a later section on seasonality.)

This section will provide more details regarding the analysis of the chronic toxicity data

from DeGraeve et al. (1987), and a comparison of its temperature dependence to that

of acute toxicity in the same study. Figure 5 showed the temperature dependence of

acute and chronic effect concentrations from this study.

An important issue in this analysis is the confounding effect of pH on the apparent

effect of temperature, because pH increased with decreasing temperature in these

chronic exposures. To examine what the effect of temperature is, the effect

23

---

Physa gyrina

500 •

?

200 —

Crangonyx pseudogracilis

?

Musculium transversum

100

50

100

20

10

20

200

50

2000

Philarctus quaeris

0

10

20

30

Helisoma trivolvis

Asellus recovitzai

100

10?

20?

30

10

I

?

?

20

I

?

30

1000

1000

500

200

100

50

500

200

Temperature Dependence of Acute Ammonia Toxicity

to Invertebrate Organisms from Arthur et al. (1987).

(Solid line denotes species slope; Dotted line denotes pooled slope)

0

?

10

?

20

?

30

?

0

?

10

?

20

?

30

?

10

?

20

?

30

Orconectes immunis

CNI

10—

All Species

0.1

0

?

10-Physa,Crangonyx,

Musculium

0

U,

5

Lo°

5

O

_i

?

.A.?

0

_.,

-FD■

co

4 •

ID?

.

1

i

2

w

O

Lil?

lit

?

Lu

8

1.0

7)

2

-IO13

1

?

0

1

.1.

•

?...,

-a

a)?

.

?

.:13

-2

O

0.5?

i?

i ■ I.

02

0.5

10?

20?

30?

0?

10?

20?

30

TEMPERATURE (C)

0

■

•

•1*

I?

I?

■ 1.

10?

20?

30

Figure 6.

500

200

100

i

cc

cis

50

'E

20

E

E

500

I

0-

200

E

0

100

o

50

2

20

0F-

0LU

5000

2

0

2000

1000

500

200

24

---

concentrations should be adjusted to a common pH using an equation that accounts for

the effect of pH. A critical question then is what pH equation to use, because no study

exists for the effect of pH on this particular chronic endpoint (juvenile 30-day survival),

or on the interaction of pH and temperature effects. The 1998 Update used the pH

relationship derived for the chronic criterion. Of the pH relationships available, that one

is probably most appropriate, but entails some uncertainty. To evaluate how

conclusions about temperature effects will vary if the true pH relationship is different,

this analysis will also use the pH relationship for acute toxicity to fathead minnows from

Thurston et al. (1981). This relationship likely represents an extreme possibility; i.e., it

assumes that chronic toxicity pH relationships are the same acute ones, contrary to

what is indicated by available chronic studies.

Using the pooled chronic pH relationship (presented later in this document),

slope=0.010, significance=0.13, and r 2=0.76.

Using the fathead minnow acute pH

relationship, slope=0.0053, significance=0.32, and r2=0.45.

In neither case is the

regression statistically significant at the 5 percent level, due to the amount and

variability of the data. Nevertheless, it should be noted that, in both cases, the chronic

data show an upward trend with temperature, in contrast to that observed for acute

toxicity. Even under the extreme assumption that these data have a pH relationship

similar to acute toxicity, the slope is 0.005, and is twice this under the assumption that

these data follow the chronic pH relationship. Thus, even if fathead minnows show

increased

acute

tolerance to ammonia at low temperatures, a similar assumption for

chronic

toxicity is contraindicated.

The difference between acute and chronic temperature relationships can be better

assessed by looking at acute-chronic ratios. Figure 7 shows the temperature

dependence of the ratio

of the acute LC50 to the chronic LC20. The chronic LC2Os

used for the ACRs were normalized via the above two alternative pH relationships,

while the acute LC5Os were normalized only using the acute pH relationship. The

results show that for either pH normalization alternative, the ACRs are substantially

higher at lower temperatures than at higher temperatures. If the chronic data are pH-

normalized using the chronic pH relationship, the regression is significant at the 0.05

level, with a slope of -0.0155.

If normalized using the acute pH relationship, the slope

is less (-0.0110), but even with this extreme assumption, there is only a 13 percent

probability that the regression slope arose by chance.

It is not surprising that acute-chronic ratios are higher at low temperature. Temperature

can affect toxicity in a variety of ways, one of which is simply to slow down responses.

This is evident in some reports on the effect of temperature on ammonia toxicity. For

example, for the rainbow trout data from Ministry of Technology (1967), there was little

effect of temperature on total ammonia LC5Os at 96 hours, but at shorter durations

LC5Os increased with decreasing temperature. The overall impact on the temperature

dependence of LC5Os and ACRs will depend on the duration of the acute toxicity test

and on the speed of action of

acute ammonia toxicity in the species of concern.

However, temperature is likely to affect ammonia toxicity in multiple ways, some of

25

---

Figure 7. Temperature-dependence of ammonia ACRs for fathead minnows.

(The choice of reference condition, pH=8 here versus pH=7.5 in Figure 5,

has no effect on slope or significance.)

Temperature-dependence of ammonia acute:chronic ratios for fathead minnows

10 —

10 —

Chronic Data Adjusted to pH=8

Using Update Average

Chronic pH Relationship

Chronic Data Adjusted to pH=8

Using Fathead Minnow

0 7

7

Acute pH Relationship

7

5

ch

4

0

LU

3

5

4

3

•

•

2

O

_c

co

2

0

0

Slope = -0.0155

r2 = 90%

Siglevel = 0.05

Slope = -0.0111

r2

= 80%

Siglevel = 0.11

1

0

10?

15

?

20?

25?

30

Temperature (C)

5?

10

?

15

?

20

?

25?

30

Temperature (C)

which would alter acute and chronic toxicity similarly. Nonetheless, to some degree the

ratio of effect concentrations at different durations is expected to increase at lower

temperatures. This expectation, as well as the empirical evidence, argues against the

direct application of acute temperature relationships to chronic toxicity.

Invertebrate chronic projections

No data are available on the effect of temperature .on chronic ammonia toxicity to

invertebrates. Because invertebrates are much more acutely tolerant at low

temperatures than at high temperatures, it is likely that their chronic toxicity would also

show some temperature dependence. However, as discussed above, there is reason

to expect acute and chronic toxicity to vary somewhat differently with temperature, with

acute-chronic ratios increasing at low temperature, especially for organisms for which

acute ammonia toxicity is not especially fast, which is the case for invertebrates

(Thurston et al. 1984). The observed trend in the fathead minnow ACRs provides

support for this expectation.

The critical question then becomes, how much of the acute temperature slope for

invertebrates should be assumed to apply to chronic toxicity? If this slope is

predominantly due to temperature-induced delays of acute toxicity, chronic toxicity

26

---

might have very little slope. If this slope is not at all due to such delays, then all the

slope should be applied to chronic toxicity.

One option for an objective mathematical prediction of the invertebrate chronic slope is

to assume that the difference between acute and chronic slopes will be the same for

fish and invertebrates, potentially implying that the effect of temperature on the kinetics

of toxicity is roughly the same for fish and invertebrates. In this case the invertebrate

chronic slope would be the difference between -0.036, the average invertebrate acute

slope, and -0.016, the observed slope for fish acute-chronic ratios. This would yield an

invertebrate chronic slope of -0.020. This correction still applies most of the acute

slope to chronic toxicity, but recognizes that the chronic slope should probably be less

steep.

It is recognized that few data are available to define the Figure 7 fish ACR slope, and

that the assumption the invertebrate ACR slope would equal the fish ACR slope is quite

uncertain despite having some theoretical underpinning in the kinetics of toxicity.

Consequently a second option is to equate the invertebrate chronic slope to the

invertebrate acute slope (-0.036) minus one-half the fish ACR slope (-0.016/2). This

splits the difference between no correction and full correction for the fish ACR slope,

resulting in an invertebrate chronic slope of -0.028.

A third, related option is suggested from the appearance of data in the last two plots in

Figure 6, plots of "All Species" and "Physa, Crangonyx, Musculium". These plots

suggest a steeper invertebrate acute slope at higher temperatures than at very low

temperature. At greater than 10°C, these data also comfortably fit a slope of -0.044. If

such a slope were used to fit those data, however, a concentration plateau would need

to be imposed between 5 and 10°C to avoid over-estimating the acute effect

concentrations measured near 5°C. If the invertebrate chronic slope is obtained by

subtracting the full value of the fish ACR slope (-0.016), this would yield the same

invertebrate chronic slope, -0.028, as the option in the previous paragraph. In this

case, however, concentrations would be capped between 5 and 10°C in order to reflect

the implied attenuation of slope at low temperature relative to higher temperatures.

EPA selected this third option, a compromise between the first two options, for defining

the invertebrate chronic temperature slope in formulating the CCC, discussed later.

This provides a good fit to the available information.

27

---

pH-Dependence of Ammonia Toxicity

The 1984/1985 ammonia criteria document identified pH as an important factor

affecting the toxicity of ammonia and used an empirical model to describe the pH-

dependence of ammonia toxicity when expressed in terms of un-ionized ammonia. The

major features of this empirical model were a slope for logLC50 versus pH which was

approximately 1 at low pH and decreased as pH increased until

?

above which the

slope was 0. Such a model closely mimics a joint toxicity model, which also has a

slope of 1 at low pH and a slope of 0 at high pH when ammonia toxicity is expressed in

terms of un-ionized ammonia. The empirical model was parameterized based on a

pooled analysis of four datasets concerning the effect of pH on the acute toxicity of

ammonia. This effect of pH was generally supported by several additional datasets

reviewed by Erickson (1985), although some variation among species was evident,

especially for channel catfish. A dataset concerning chronic ammonia toxicity

(Broderius et al. 1985) indicated a somewhat greater effect of pH than for acute toxicity

and was used as the principal basis for the pH-dependence of the CCC.

As explained in the overview of this update, the effect of pH on the toxicity of ammonia

will be described here largely in terms of the joint (combined) toxicity of un-ionized

ammonia and ammonium ion. However, there is some dispute about whether ammonia

toxicity merely involves such joint toxicity. Also, a variety of factors might affect the

combined toxicity of the two forms. Therefore, use of a simple, mechanistic joint toxicity

model is inadvisable, and the following "S-shaped" model will be used to describe the

pH dependence of total ammonia toxicity:

LC50 - ?

+

LIM

10PHT-PH

H?LIML

1 + 1 OPH-pHT

where the subscript t denotes total ammonia, LIMH

and LIML

are asymptotic (limiting)

LC5Os at high and low pH respectively, and pHT

is the transition pH at which the LC50

is the arithmetic average of UK., and LIM

L

. This model is justified by various data (see

the overview) and is consistent with joint toxicity of un-ionized ammonia and ammonium

ion. However, the model treats pH

T

as a fitted parameter, whereas if joint toxicity were

assumed it would be dictated by the pK of ammonia (see Equation 4) and the relative

toxicity of the two forms.

Use of LIM H

and LIM

L

as model parameters results in a simple equation, but is

inconvenient for data analysis for two reasons. First, when analyzing toxicological

variables across multiple datasets, an important issue is whether the shapes of the

curves are similar among the datasets. For making such comparisons and for

estimating the best average shape, it is necessary that each parameter of the equation

either is related only to the shape or is not related to the shape at all. For example, in

linear regression, the equation is generally expressed in terms of a slope and an

(6)

28

---

LC5Ot

=

?

LC50L8

1 + 1 OPHT-8

R

1

1 +1 08-PHT

/

intercept (i.e., the value of y at a specified value of x, such as x=0). The slope

completely defines the shape of the relationship, whereas the intercept anchors the

relationship at a particular point and has no effect on the shape. For the nonlinear

regression used here, there needs to be one, and only one, "intercept" parameter that

specifies the LC50 at a particular pH, independent of the shape, whereas the other

parameters must describe aspects of the shape and not affect the intercept. In the

above equation, LIM

H

and LIM L

are both "intercepts" (at high and low pH, respectively),

and they also in part dictate the shape of the curve because the shape partly depends

on the difference between the two intercepts. Thus, it is not possible to completely

separate the shape from the intercepts. To eliminate this problem, the equation was

reformulated so that LIM

L

is the only intercept parameter. This was accomplished by

using the parameter R = LIM

H/LIML

, which, along with pHT

, defines the shape of the

curve:

LC5Ot = (LIM

L) I ?

1 + 1

R

OPHT-PH

?

1

+ OPH-PHT)

1

The second shortcoming of the use of LIM

H and/or LIM

L

is that they are LC5Os at

extreme pHs which are not observed and are largely hypothetical; it is preferable to

have an "intercept" parameter that lies in the range of the observed data. Therefore,

the equation was reformulated to use the LC5O

t

at pH=8 (LC50L8)

as the intercept

parameter instead of LIM

L

. Switching from LIM

L to LC50L8

requires use of a term that is

the ratio between LC50L8

and LIML:

(7)

(

1

R?

1

+

?

1 OPHT-PH

+ 1 + 1 OPH-PHT

(8)

All three of the above model equations are equivalent, differing only in the way in which

the parameters are formulated.

Unfortunately, analyses based on any of these three model equations can be subject to

serious problems with some datasets, especially for estimation of LI

M H or R. This is

because LC5Ot

is generally much greater than LIM

H

even at the highest pH in most

datasets (pH=8 to 9), so that the approach to this asymptotic value is very uncertain.

However, the pH is usually sufficiently high that un-ionized ammonia, although only a

small fraction of total ammonia, dominates toxicity and provides information about LIMH

and R that is not apparent when only total ammonia is examined. To address this

problem, the formulation of the model was changed by splitting the equation into two

parts:

29

---

LC5Ou

LC50

;

-

LC500

R

(9)

(10)

1

+ 1 OPHT-PH

1

(

+ I OPHT-8?

I + 1 08-PHT

LC50t,8

R? 1

+

1?

+ 1 OPH-PHT)

1 + 1 OPFIT-'?

1 + 108 PHT

where LC50„ and LC50

;

are the LC50s expressed in terms of un-ionized ammonia and

ammonium ion, respectively, and LC50„ + LC50

; = LC50t

. This approach more strongly

emphasizes the notion of joint toxicity, but still is somewhat empirical because pH

T is a

fitted parameter. Regression methods for multiple response variables (see Appendix 2)

were used to fit this model to the available datasets.

Acute datasets evaluated included those cited in the 1984/1985 ammonia criteria

document and Erickson (1985), as well as more recent studies by Sheehan and Lewis

(1986), Schubauer-Berigan et al. (1995), Ankley et al. (1995), and Johnson (1995).

1.

Sheehan and Lewis (1986) investigated the pH-dependence of acute ammonia

toxicity to channel catfish. LC5Os expressed in terms of un-ionized ammonia

increased with increasing pH, but less so than reported in most studies, although

Tomasso et al. (1980) also reported little effect of pF1



7 on un-ionized ammonia

toxicity to the channel catfish.

2.

Schubauer-Berigan et al. (1995) evaluated the effect of pH on the toxicity of

ammonia to the oligochaete

Lumbriculus variegatus

and to larvae of the dipteran

Chironomus tentans.

Both species exhibited increases in 10-day un-ionized

ammonia LC5Os with increasing pH, but the increase for C.

tentans

was somewhat

larger than those for other species for which data are available, whereas those of

L. variegatus

were smaller. Such interspecies differences would be of concern in

the derivation of the criterion if they substantially altered relationships for sensitive

species; these particular species, however, are sufficiently resistant to ammonia

that the pH relationship used for them has no impact on the criterion.

3.

Ankley et al. (1995) tested the effect of pH on the toxicity of ammonia to the

amphipod

Hyalella azteca

in waters of three different ionic compositions. In all

three waters, 96-hr LC5Os expressed in terms of un-ionized ammonia increased

with pH, but the amount of increase was greater in waters with low ion

concentrations. These waters differed with respect to a variety of ions, so it is

uncertain which constituent is responsible for the difference in the effect of pH,

although recent work by Borgmann and Borgmann (1997) suggests that the

concentration of sodium is a major factor. These results not only indicate some

effect of the ionic composition of the test water on ammonia toxicity, but also

suggest that this composition might differentially affect the relative toxicity of un-

ionized ammonia and ammonium ion. In the low ion concentration test water,

30

---

H. azteca

was one of the most sensitive species tested at low pH and

consequences for the criterion will be considered later.

4. Johnson (1995)

investigated the effect of

pH

on the chronic toxicity of ammonia to

Ceriodaphnia dubia

in test waters of three different ionic compositions. In all three

waters, LC5Os

expressed in terms of un-ionized ammonia increased with increasing

pH, but, unlike Ankley et al.

(1995), the pH

dependence was greater in waters with

higher, rather than lower, hardness.

Acute total ammonia LC50s

versus pH are presented in Figure 8 for all studies

analyzed; for the study of Ankley et al.

(1995) with

H. azteca,

the small, medium, and

large symbols denote low, medium, and high ion concentrations in test waters. All

analyses were conducted in terms of total ammonia nitrogen, either as reported by the

authors or as converted by us from the reported un-ionized ammonia

LC50, pH, and

temperature using the speciation relationship of Emerson et al.

(1975). All of the

datasets show a strong trend of total ammonia

LC5Os

decreasing with increasing pH,

except that of

H. azteca

at low ion concentrations. There are, however, differences

among the datasets in the magnitude and shape of the trend. Some datasets show an

approach to an asymptote at low

pH

whereas others do not. In addition, C.

tentans

and

H. azteca

show lower slopes than other species. Nevertheless, it would be speculative

to assign different relationships to different taxa, especially because the same or

closely related species show some variation. Consequently, all of the datasets were

used to determine an average, generic shape for the

pH

dependence.

Regression analyses were conducted individually on each dataset, and on the pooled

datasets assuming that only LC500

varied among datasets. The pooled analysis

estimated pHT to be

7.204 (95% confidence limits =

7.111 and 7.297) and R to be

0.00704 (95%

confidence limits = 0.00548 and 0.00904).

The individual regression

results are plotted as solid lines and the pooled analysis as dotted lines in Figure

8.

The data points and the common regression line from the pooled analysis are also

plotted together in Figure 9 by dividing each point by the

LC500 for its dataset (this

normalized plot allows a different, combined perspective of the overall scatter of data

from the shape of the generic relationship not possible in Figure 8). Except for the

datasets for

L. variegatus

and

H. azteca at

low ion concentrations, the deviation of data

from this generic relationship at

pH>7 is

rather small and consistent with the typical

uncertainty of LC50s. At pH<7,

however, some of the deviations are substantial; some

species, most notably channel catfish and

L. variegatus,

have higher than expected

total ammonia LC50s,

whereas others, such as

Daphnia

sp. and

H. azteca

have lower

than expected

LC50s.

Fortunately, these species are generally sufficiently resistant

that more accurately describing their pH

dependence is unimportant for deriving a

CMC. Despite the variation among species at low

pH, this generic relationship is

appropriate for criteria derivation, because it provides significantly higher values at low

pH,

but not higher than those for fish species that are relatively sensitive at low pH, a

suitably conservative assumption for sensitive species for which data do not exist at

low pH.

31

---

100

40

20

10

Fathead Minnow

4

_ (Thurston et al. 1981h)

I?...,1?

1

Coho Salmon

(Robinson-Wilson & Seim 1975)

490

200

100

40

20

V

1 0

Smallmouth Bass

V

4

2

? 2 ?

4

?

Rainbow Trout

2

_ (Thurston et al. 19811))

1?

1?

.?

.?

1?

.?

.?

.?

.?

1?r .

?

.■■111

?

.

200

Daphnia sp.

(Tabata 1962)

(Broderius et al. 1985)

400

200

100

40

20

10

4

2

400

200

100

40

20

10

200

100

100

400

200

100

200

40

20

10

40

20

10

10

40

20

10

—

Green Sunfish

4

(McCormick et al. 1984)

2

4 —

Rainbow Trout

2

_ (Lloyd & Herbert 1960)

Macrobrachium rosenbergii

4

(Armstrong et al. 1978)

9?

1?

6

1

?

?

,...1,..,1

7?

8

?...,1

I?I?I?I?I?

1?

1?

1?1?1

2

9

2?

6

1?

?

•■■■■

7?

8

9?

6 7?

8 9 6?

7?

8

400

200

100

2000

400

200

100

1 0

00

400

200

100

400

200

100

400

200

100?

40

20

1 0

Channel Catfish

?

—Lumbriculus variegatus

(Sheehan & Lewis 1986)

4

?(Schubauer-Berigan et al. 1

2 l '"'

1 "" 1 "" 1?2

1 1 1 1 1 1.,■.1

40

40

•

20

40

•

20

10

•

40

20

10

10

10

20

Hyalella azteca?

•?

•

4

(Ankley et al. 1995)

2 ?

9

Chironomus lentans

— (Schubauer-Berigan et al 1995)

Guppy

(Tabata 1962)

2

?

6

1?

?

1.1

?

7

1?

1

?

1?

1?,.1

4

4

It...IIIIfltil

l?

I

9

6?7?

8

4

8

9?

6?

7?

8 9?

6?

7?

8 9?

6?

7?

8

400

200

100

40

20

Channel Catfish

4

_ (Tomasso et al.

1980)

Figure 8. The effect of pH on acute ammonia toxicity in terms of total ammonia. Symbols

denote LC50s, solid lines denote regressions for individual datasets, and

dotted lines denote pooled regression over all datasets.

6?7?

8

?

9?

6?

7?

8?

9?

6?

7?

8

?

6?

7?

8?

9

?

6?7?

8?

9

pH

32

---

p H

A

1

0.8

0.6

0.4

0.3

0.2

0.1

6

?7

?8

?

9

Figure 9. The effect of pH on normalized acute ammonia toxicity in terms of total

ammonia. Data were normalized by dividing measured LC5Os by regression

estimates of LC5Os at pH=8 for individual datasets from Figure 8. Data were

not normalized in any way for temperature.

O

Fathead Minnow

q

Rainbow Trout

Coho Salmon

V?

Daphnia sp.

•

SmallmouthBass

O

Green Sunfish

Rainbow Trout

•

Prawn Larvae

•

ChannelCatfish

0?

White Perch

•

Guppy

ChannelCatfish

Lumbriculus

Chironomus

Hyalella

33

---

For chronic toxicity, the data of Broderius et al. (1985) and Johnson (1995) were

analyzed in terms of total ammonia nitrogen using the same pH model (Figure 10). The

data used were EC25s reported by Johnson (1995) and EC2Os calculated from the

data of Broderius et al. (1985) by regression analyses discussed later. (Because

Johnson's raw data were not available, EC2Os could not be calculated, but

the

shape

of the curve should be the same for EC2Os and EC25s.) Because the uncertainty of

the EC25s from Johnson (1995) was greater than that of Broderius et al. (1985) and to

prevent the greater number of data points for the invertebrate from overwhelming the

data for the fish, data points from Johnson (1995) were given a weighting factor of 0.5

in this analysis. These chronic data had a higher transition pH (7.688; 95% confidence

limits = 7.554 and 7.821) and a higher R (0.0232; 95% confidence limits = 0.0160 and

0.0334) than the acute data. The higher pH

T is in accordance with differences

previously noted in the 1984/1985 criteria document regarding the pH dependence of

acute and chronic toxicity. Tests. by Borgmann (1994) on the chronic toxicity of

ammonia to

Hyalella azteca

and by Armstrong et al. (1978) on the 6-day toxicity of

ammonia to

Macrobrachium rosenbergii

also support a lower slope for total ammonia

chronic toxicity versus pH at pH<8. The dependence of chronic ammonia toxicity on pH

appears to be sufficiently different from the dependence of acute ammonia toxicity to

justify use of two equations.

By substituting the values for R and pH-

1

- into Equation 8, the following equations are

obtained for describing the pH-dependence of acute values (AVs) and chronic values

(CVs) expressed in terms of total ammonia nitrogen:

AV

(AVt,8)

0.0489

6.95

+?

07.204-pH

1+ 1 0pH-7.204

c

vt = (

0.0676

2.91

1 + 10PH-7.688

+ 1 07.688-pH

(12)

The range of the data used to derive these equations indicates that they should be

applicable from pH=6 to 9, although considerable error might exist at the lower end of

this range for certain species. Extrapolation below pH=6 is not advisable because of

the increasing scatter of the data from the common regression line at lower pH, and

extrapolation above pH=9 is not advisable because of inadequate knowledge about the

effect of the inhibition of ammonia excretion at high pH on results of toxicity tests

(Russo et al. 1988).

34

---

Ceriodaphnia dubia

(Johnson 1995)

40

A

_J

-2 20

0)

E

o

10

w-

c.1

04

2

2<

2

O1

678

1

6

Smallmouth Bass

(Broderius et al. 1985)

100

"2 40

0)

Lo

(.1

20

< 10

2

<4

O2

Figure 10. The effect of pH on chronic ammonia toxicity in terms of total ammonia. '

Symbols denote chronic effect concentrations and lines denote regressions of

effect concentrations versus pH. (For C.

dubia,

different symbols denote

different test water formulations. It is important to note the triangle, partially

obscured by the square at pH=6.5.)

,?

.?

I?

,?

,?

,?

.?

I?

,?

,?

i?

.?

I

7? 8? 9

pH

pH

35

---

Formulation of the CMC

The scope of this project included a re-examination of the temperature and pH

relationships underlying the 1984/1985 Criterion Maximum Concentration (CMC).

Because the acute toxicity dataset contained in the 1984/1985 criteria document (U.S.

EPA 1985a) is relatively large, with tests involving species in 34 genera, the scope of

this project did not include a comprehensive literature search and critical review of all

of the acute toxicity data now available. Thus, the derivation here relies solely on

acute tests reported in Table 1 in the 1984/1985 criteria document, supplemented by

some newer studies relevant to the revised pH relationship.

However, some other newer studies of acute toxicity known by EPA were examined to

determine whether new data might materially affect the CMC. These studies include

Ankley et al. 1995; Arthur et al. 1987; Bailey et al. 1985; Bergerhouse 1992,1993;

Dabrowska and Sikora 1986; DeGraeve et al. 1987; Diamond et al. 1993 (see

Appendix 1); Gersich and Hopkins 1986; Goudreau et al. 1993; Gulyas and Fleit 1990;

Hasan and Macintosh 1986; Henderson et al. 1961; Lee 1976; Mayes et al. 1986;

Monda et al. 1995; Nimmo et al. 1989; Russo et al. 1988; Sheehan and Lewis 1986;

Snell and Persoone 1989; Thomas et al. 1991; Tomasso and Carmichael 1986; Wade

1992; and Williams et al. 1986. These studies would add few new genera to the

dataset and their data are generally in the range already observed and would have little

impact on the four lowest Genus Mean Acute Values (GMAVs). The most significant

result of these studies is that some invertebrates are acutely sensitive to ammonia at

low pH and low ion concentration (Borgmann 1994; Ankley et al. 1995). Although new

data are not used in the derivation of the new CMC, they are compared to the new

CMC below.

All of the un-ionized ammonia acute values (LC5Os and EC50s) in Table 1 of the

1984/1985 criteria document were converted to total ammonia nitrogen acute values,

using the reported temperatures and pHs and using the pK relationship from Emerson

et al. (1975). These total ammonia nitrogen acute values were then adjusted (see

Appendix 3) to pH=8 using the pH relationship developed above, with no adjustment for

temperature. These adjusted total ammonia nitrogen acute values (see Appendix 4)

were then averaged to determine Species Mean Acute Values (SMAVs) and GMAVs at

pH=8 using the procedure described in the 1985 Guidelines (U.S. EPA 1985b). These

SMAVs and GMAVs are presented in Table 4 (which is the same as Table 1 of the

1998 Update). Both are based on the test results of in Table 1 of the 1984/1985

criteria document. The GMAVs in the 1984/1985 and the 1998/1999 documents differ

because (a) pH and temperature are addressed differently in the two sets of

calculations, (b) the golden trout, cutthroat trout, and rainbow trout are now in a

different genus, and (c) and the new GMAVs are expressed in terms of total ammonia

nitrogen; the order of the genera is different mostly because the absence of an

invertebrate temperature normalization in either the 1984/1985 criteria document or the

1998 and 1999 Updates affects the 1984/1985 un-ionized LC5Os differently than the

1998 and 1999 total ammonia LC50s. As noted earlier in the document, a temperature

36

---

Table 4. Ranked Genus Mean Acute Values

Genus Mean

?

Species Mean

Acute Value

?

Acute Value

Rank?

(mq N/La)?

Species

?

(mq N/La)

34?388.8

?

Caddisfly,?

388.8

Philarctus quaeris

33?246.0?

Crayfish,?

1466.

Orconectes immunis

Crayfish,?

41.27

Orconectes nais

32?210.6

?

Isopod,?

210.6

Asellus racovitzai

31?189.2?

Mayfly,?

189.2

Ephemerella grandis

30?

115.5?

Mayfly,

?

175.6

Callibaetis skokianus

Mayfly,?

75.93

Callibaetis sp.

29?

• 113.2?

Beetle,

?

113.2

Stenelmis sexlineata

28?108.3?

Amphipod,?

108.3

Crangonyx pseudogracilis

27

?

97.82

?

Tubificid worm,?

97.82

Tubifex tubifex

26?93.52?

Snail,?

93.52

Helisoma trivolvis

25?77.10?

Stonefly,

?

77.10

Arcynopteryx parallela

24?73.69?

Snail,?

73.69

Physa gyrina

23

?

51.73

?

Mottled sculpin,?

51.73

Cottus bairdi

22?51.06?

Mosquitofish,

?

51.06

Gambusia affinis

21?43.55?

Fathead minnow,?

43.55

Pimephales promelas

20?38.11?

White sucker,?

45.82

Catostomus commersoni

37

---

Genus Mean

Acute Value

Rank

?

(mq N/12)

Species

Species Mean

Acute Value

(mq N/12)

Mountain sucker,

31.70

Catostomus platyrhynchus

19?

36.82

?

Cladoceran,?

35.76

Daphnia magna

Cladoceran,

?

37.91

Daphnia pulicaria

18?

36.39

?

Brook trout,

?

36.39

Salvelinus fontinalis

17?

35.65?

Clam,?

35.65

Musculium transversum

16?

34.44

?

Channel catfish,

?

34.44

Ictalurus punctatus

15?33.99?

Cladoceran,

?

33.99

Simocephalus vetulus

14?

33.14

?

Guppy,?

33.14

Poecilia reticulata

13?

32.82?

Flatworm,

?

32.82

Dendrocoelum lacteum

12?30.89

?

White perch,?

30.89

Morone americana

11?

26.97?

Stoneroller,

?

26.97

Campostoma anomalum

10?26.50

?

Smallmouth bass,?

35.07

Micropterus dolomieu

Largemouth bass,?

20.03

Micropterus salmoides

9?

26.11

?

Walleye,

?

26.11

Stizostedion vitreum

8?

25.78

?

Cladoceran,?

25.78

Ceriodaphnia acanthina

?

25.60

?

Red shiner,?

45.65

Notropis lutrensis

Spotfin shiner,?

19.51

Notropis spilopterus

Steelcolor shiner,?

18.83

Notropis whipplei

38

---

Genus Mean

?

Species Mean

Acute Value

?

Acute Value

Rank?

(mq

N/La)

?

Species

?

(mq N/La)

6?

23.74

?

Brown trout,

?

23.74

Salmo trutta

5? 23.61

?

Green sunfish,?

30.27

Lepomis cyanellus

Pumpkinseed,?

18.05

Lepomis gibbosus

Bluegill,

?

24.09

Lepomis macrochirus

4? 21.95?

Golden trout,?

26.10

Oncorhynchus aquabonita

Cutthroat trout,

?

25.80

Oncorhynchus clarki

Pink salmon,?

42.07

Oncorhynchus gorbuscha

Coho salmon,

?

20.26

Oncorhynchus kisutch

Rainbow trout,

?

11.23b

Oncorhynchus mykiss

Chinook salmon,?

17.34

Oncorhynchus tshawytscha

3

?

17.96?

Orangethroat darter,

?

17.96

Etheostoma spectabile

2? 14.67?

Golden shiner,?

14.67

Notemigonus crysoleucas

1? 12.11?

Mountain whitefish,

?

12.11

Prosopium williamsoni

All values are total ammonia nitrogen at pH=8.

Thurston and Russo (1983) conducted numerous acute toxicity tests with larval, juvenile, yearling,

and larger rainbow trout and demonstrated that large rainbow trout were measurably more sensitive

than other life stages. The average adjusted total ammonia nitrogen acute value for large rainbow

trout was 11.23 mg N/L. Therefore, this SMAV was lowered to 11.23 mg N/L in order to protect large

rainbow trout, as per the 1985 Guidelines (U.S. EPA 1985b).

39

---

CMC - ?

00.275

+ 1 0

7'204-pH

39.0

+ 1 0pH-7.204

(13)

normalization for the invertebrates, although technically justifiable, is irrelevant to the

CMC formulation, due to the insensitivity of all invertebrate taxa to ammonia acute

toxicity.

The Final Acute Value (i.e., the fifth percentile) at pH=8 was calculated from this set of

adjusted total ammonia GMAVs to be 14.32 mg N/L. The SMAV for rainbow trout is

11.23 mg N/L, and so the FAV is lowered to this value, as per the 1985 Guidelines

(U.S. EPA 1985b), comparable to what was done in the 1984/1985 ammonia criteria

document. The CMC at pH=8 equals one-half of this FAV. Substitution of this CMC at

pH=8 for AVo in Equation 11 results in the following equation for expressing the CMC

as a function of pH:

If the four genera

(Oncorhynchus, Prosopium, Salmo,

and

Salvelinus)

in the family

Salmonidae are excluded from the dataset in Table 4, the fifth percentile FAV with

salmonids absent is 16.8 mg N/L and the CMC is 8.4 mg N/L at pH=8; substitution into

Equation 11 gives the CMC as a function of pH:

CMC -

+

0.4111

07.204-pH

?

?

1 + 1

58.4

0pH-7.204

Figure 11 shows the ranked GMAVs, the CMC with salmonids present, and the CMC

with salmonids absent, all at pH=8. The GMAVs represent LC50s, whereas the CMCs

represent concentrations that are lethal to a minimal percentage of the individuals in

either the fifth percentile genus or a sensitive important species.

FAVs and CMCs are plotted in Figure 12, along with all of the individual total ammonia

acute values, unadjusted for pH or temperature, used in the calculations. The FAVs

show good correspondence with the lower range of the acute values. As discussed

above, more recent acute data are also in general accordance with the FAVs, except

that the

Hyalella azteca

LC50 from Ankley et al. (1995) at low ion concentration and

pH=6.5 is more than a factor of two below the FAV. Although some toxicity data are

expected to be below the FAV, inclusion of this genus in the calculation would have

resulted in a lower CMC, but only under these extreme water quality conditions and

only if the effects of both pH and ionic composition were described for each individual

genus, which is not possible with the data that are currently available.

(14)

40

---

Figure 11. Ranked Genus Mean Acute Values (GMAVs) with Criterion

Maximum Concentrations (CMCs).

1000

co

?

100

?

.0111•

NM

■■

-2

E

O

111•••111111111111011."1111.1111111M

2?

10

•

■

? CMC salmonids absent

?

CMC salmonids present

1

Ranked Genera

---

■

•

?

■

FAV - 5th Pct

w/o Salmonide

FAV

Rainbow

- AdultTrour■....

?

• II

New CMC

w/o Salmonids

New CMC

w/ Salmonids

Old CMC :

(10, 20 C)

O

Molluscs

•

Other Invertebrates

•

Salmonid Fishes

■

Nonsalmonid Fishes

•

•

Elm

Figure 12. Acute LC5Os used in criteria derivation in

relationship to Final Acute Values (FAVs) and

Criterion Maximum Concentrations (CMCs).

1000

0

■

1

100 —

10 —

6789

pH

42

---

Review and Analysis of Chronic Data

Due to the magnitudes of the acute-chronic ratios (ACRs) for ammonia, the ammonia

CCC is sufficiently low relative to the CMC that the CCC generally will be the

determining factor for permit limits. In the 1984/1985 ammonia criteria document, the

CCC is more uncertain than the CMC because (1) the CCC was calculated by dividing

the FAV by an ACR (thus including the uncertainties of both the FAV and the ACR) and

(2) fewer acceptable chronic toxicity tests were available and not all of them could be

used to derive ACRs. Additionally, depending on how they were derived, the individual

chronic values could differ with respect to the nature and degree of the toxic effects

they represented. To reduce this variability, EPA reviewed and analyzed all of the

chronic data used in the 1984/1985 criteria document and newer chronic data known to

the authors or suggested by peer reviewers to produce a more extensive and

consistent set of Chronic Values (CVs) that could be used to directly calculate a CCC

rather than to calculate it using ACRs. This procedure also has some limitations

because (a) the criterion usually decreases as the number of genera used in the

calculation of the 95th percentile decreases and (2) chronic tests have been conducted

with a larger proportion of the species that are acutely sensitive to ammonia than those

that are acutely resistant to ammonia.

The first two parts of this section describe how the chronic tests on ammonia were

reviewed and how the CVs were calculated. The third part discusses each chronic test

of which EPA was aware and presents the relevant results.

Review of Chronic Data

Each chronic dataset was subjected to the following two-step review process. The first

step was to determine whether the test methodology was acceptable for providing

information about a CV. A test was considered acceptable if the dilution water, control

mortality, experimental design, loading, etc., were consistent with ASTM Standards

E1193, E1241, and E1295 (ASTM 1997a,b,c). The concentration of dissolved oxygen

was also reviewed on the basis of the U.S. EPA (1986) dissolved oxygen criteria

document.

Reviewing the concentration of dissolved oxygen (DO) was difficult because (a) ASTM

Standards E1193, E1241, and E1295 (ASTM 1997a,b,c) express limits on high and low

concentrations of DO in terms of percent saturation, whereas U.S. EPA (1986)

expresses limits on low concentrations of DO in terms of the concentration itself, and

(b) neither specifies the limits in a way that can be used directly to interpret the kinds of

information that are given in most reports of the results of toxicity tests. Therefore, the

following rationale was used. The mean DO concentration needs to be within an

acceptable range, but limits expressed as long-term averages can allow excessively

low or high concentrations for too long a period. Conversely, a limit that must be

satisfied at all times can unnecessarily penalize investigators who make more than the

minimum number of measurements and ignores the fact that organisms can tolerate

43

---

extreme concentrations for brief periods of time. Therefore, limits were placed on the

mean and the fifth and ninety-fifth percentiles of the DO concentrations. Use of limits

that are expressed in terms of the mean and the fifth and ninety-fifth percentiles is

straightforward when the mean and standard deviation are reported or when all of the

individual measurements are reported, but not when only the range is reported. If the

measured concentration of DO during a chronic test was reported as a range, the

lowest and highest values were considered to be concentrations that existed for at least

5 percent of the time during the test.

The limits used were:

1.

A chronic test was considered questionable if either (a) the mean DO concentration

was below 60 or above 100 percent of saturation or (b) the concentration of DO

was below 50 or above 105 percent of saturation more than 5 percent of the time

during the test. These limits are similar to, but different from, the limits given in

ASTM Standards E1193, E1241, and E1295 (ASTM 1997a,b,c).

It is clear that 60 percent of saturation is the desirable lower limit in Section 11.2.1

of ASTM Standard E729 (ASTM 1997d); for practical reasons, this section allows

the concentration of DO to be between 40 and 60 percent of saturation during the

last 48 hours of 96-hr static acute tests. Because test organisms and BOD utilize

oxygen, when the concentration of DO is above 100 percent of saturation, it is quite

possible that the concentration of dissolved nitrogen is even more supersaturated,

which increases the possibility of gas bubble disease.

2.

A chronic test was considered questionable if either (a) the mean measured DO

concentration was below the mean given below or (b) the DO concentration was

below the lower limit given below for more than 5 percent of the time during the

test:

Mean (mg/L)

Lower Limit (mg/L)

Salmonids:

6.5

5.0

Warmwater fishes

Early life stages

6.0

5.0

Other life stages

5.5

4.0

Invertebrates

6.0

5.0

The first three means are presented on page 34 of U.S. EPA (1986) and are 0.5

mg/L above the concentrations given for "slight production impairment" on page 31.

U.S. EPA (1986) does not give a "mean" for invertebrates on page 34 and so the

last mean given above is 1 mg/L higher than the concentration given for "some

production impairment" on page 31. The lower limits are concentrations given on

page 31 for "moderate production impairment" or "some production impairment".

Regardless of how limits on the DO concentration are expressed, it is sometimes

difficult to apply them to the information that is reported concerning toxicity tests.

If there was no reason to believe that the test methodology was unacceptable, the

second step of the review process was to determine whether the test satisfied one of

the definitions given in the 1985 Guidelines for life-cycle, partial life-cycle, and early

44

---

life-stage test. By definition, life-cycle tests can be conducted with either a fish species

or an invertebrate species, but partial life-cycle and early life-stage tests can only be

conducted with a fish species. The considerations that excluded the most tests were

that (a) tests that did not include the newly hatched life stage cannot be acceptable life-

cycle, partial life-cycle, or early life-stage tests, and (b) tests that did not study

reproduction cannot be acceptable life-cycle or partial life-cycle tests. Each test that

satisfied one of the definitions could provide one of three kinds of information:

1.

If all of the tested concentrations of the toxicant were so high that all of them

caused unacceptable effects, the test will probably provide an upper limit on a CV,

i.e., the CV will be lower than the lowest tested concentration.

2. If all of the tested concentrations were so low that none of them caused an

unacceptable effect, the test will probably provide a lower limit on a CV, i.e., the CV

will be higher than the highest tested concentration.

3. If the low tested concentrations did not cause unacceptable effects but the high

tested concentrations did, the test will probably provide a CV.

If the test did not satisfy the requirements for any of the three kinds of tests, it was

necessary to determine whether the toxicant caused an unacceptable reduction in (a)

survival, reproduction, and/or hatchability over any period of at least seven days, or (b)

growth over a period of at least 90 days. If it caused either kind of unacceptable

reduction, the test will probably provide an upper limit on a CV or it might lower a CV

from an early life-stage test. If it did not cause either kind of unacceptable reduction,

the test cannot provide a CV or an upper or lower limit on a CV, but the test might

provide other useful information. Because the test is not an acceptable life-cycle,

partial life-cycle, or early life-stage test, an upper limit on a CV can be based on a

reduction in survival, reproduction, and/or hatchability over any period of at least seven

days, but it cannot be based on a reduction in weight gain for fewer than 90 days

because such a reduction might be temporary; such a test cannot provide a lower limit

on a CV because some other life stage might be more sensitive. Although some CVs

were based on histopathological effects in the 1984/1985 ammonia criteria document,

this current effort could find no justification for equating histopathological effects with

effects on survival, growth, and reproduction (see Appendix 5).

Calculation of Chronic Values

Chronic values used in aquatic life criteria documents have traditionally been based on

analysis of data to determine the highest tested concentration at which no relevant

toxicological variable had a value that was statistically significantly different from the

value for the control treatment (highest no observed 'adverse effect concentration,

HNOAEC) and the lowest concentration at which the value for at least one of the

relevant toxicological variables was significantly different from the value for the control

treatment (lowest observed adverse effect concentration, or LOAEC). When endpoints

are defined on the basis of such hypothesis testing of each tested concentration

against the control treatment, the CV is set equal to the geometric mean of the

HNOAEC and the LOAEC. Such a procedure has the disadvantage of resulting in

marked differences between the magnitudes of the effects corresponding to the

45

---

individual CVs, due to variation in the power of the statistical tests used, the

concentrations tested, and the size and variability of the samples used (Stephan and

Rogers 1985). For example, the CVs reported in the 1984/1985 ammonia criteria

document corresponded to reductions from the control treatment of just a few percent to

more than fifty percent.

To make CVs reflect a uniform level of effect, regression analysis was used here both

to demonstrate that a significant concentration-effect relationship was present and to

estimate CVs with a consistent level of effect. Use of regression analysis is provided

for on page 39 of the 1985 Guidelines (U.S. EPA 1985b). The most precise estimates

of effect concentrations can generally be made for 50 percent reduction (EC50);

however, such a major reduction is not necessarily consistent with criteria providing

adequate protection. In contrast, a concentration that caused a low level of reduction,

such as an EC5 or EC10, is rarely statistically significantly different from the control

treatment. As a compromise, the EC20 is used here as representing a low level of

effect that is generally significantly different from the control treatment across the useful

chronic datasets that are available for ammonia.

Regression analysis was performed on a chronic dataset only if the dataset met the

following conditions: (1) it contained a control treatment to anchor the curve at the low

end, (2) it contained at least four concentrations of ammonia to provide at least two

error degrees of freedom when the three-parameter equation is fit to a set of data, (3)

the highest tested concentration of ammonia caused >50 percent reduction relative to

the control treatment to anchor the curve at the high end, and (4) at least one tested

concentration of ammonia caused <20 percent reduction relative to the control

treatment to ensure that the EC20 was bracketed by tested concentrations of ammonia.

For life-cycle and partial life-cycle tests, the toxicological variables used in these

regression analyses were survival, embryo production, and embryo hatchability. For

early life-stage tests, the variables used were embryo hatchability, fry survival, and fry

growth; if ammonia apparently reduced both survival and growth, the product of these

variables (biomass) was analyzed, rather than analyzing them separately. For other

acceptable chronic tests, the toxicological variable analyzed was survival, reproduction,

hatchability, and/or growth as appropriate, based on the requirements stated above

concerning acceptability of chronic tests.

The regression model used was based on the logistic equation:

T - ?

To

1 + A•C

This equation produces an "S-shaped" curve, with the toxicological variable of interest

(T) being at a control value (T0)

at low concentrations, zero at high concentrations, and

declining at intermediate concentrations; the location and steepness of this decline are

determined by the parameters A and B, respectively. It is not argued that this equation

(15)

46

---

embodies a mechanistic description of chronic toxicity, but rather that this is a useful

equation that incorporates the major features commonly observed in concentration-

effect relationships. Application of various forms and extensions of this equation to

toxicological data have been discussed by various authors, most recently by Moore and

Caux (1997).

To make the equation more directly interpretable with respect to effect concentrations

and to assist in determining confidence limits for such effect concentrations as the

EC20, the equation was reformulated to:

T-

To

I +1 ?

)(10B

(log C - log ECp

100 -p

where log ECp (i.e., the logarithm of the concentration causing T to be reduced by p

percent from To)

is a parameter rather than A. This equation was applied to each

dataset using nonlinear least-squares regression analysis (Draper and Smith 1981),

with p=20%. Software used for determining the least-squares solution was written in

FORTRAN using nonlinear search routines based on the Newton-Raphson method

(Dahlquist and Bjorck 1974).

Either transformation or weighting was applied to each dataset to improve the

homogeneity of the variance:

1.

When T was a percentage, the regression analysis was conducted on a

transformation VI

of each data point T;

asfollows (Draper and Smith 1981):

Ti* = arcsin(JTi /100)

?

(17)

The regression equation was similarly transformed and the parameter T

o

was

formulated to be the transformed effect.

2.

When T was count data, the regression analysis was conducted on the square root

transformation of T, and the regression equation was similarly transformed (Draper

and Smith 1981).

3.

When T was weight or biomass, no transformation was used, but each datum was

weighted by the inverse of its variance (Draper and Smith 1981). For weight data,

these weighting factors were based on standard errors (SEs) or standard

deviations (SDs) divided by N

Y'

as reported by the authors. For biomass [B =

product of proportion survival (P) and weight (W) in early life-stage tests], the

variance was estimated as follows:

?

where SEp is the SE of

VAR(B)

P as reported

W 2

-SE:

by the

+

authors

P SEor

2

?(18)calculated

as (P(1-P)/N)Y2,

and SEW

is the SE of W as reported by the authors or calculated from their data.

(16)

47

---

In addition to the dataset-specific transformation or weighting described above, all

regression analyses used a general weighting scheme to make the analyses more

appropriate for calculating EC20s. When this type of regression analysis is used to

calculate such low-effect concentrations as an EC20, lack of fit of the model at high-

effect concentrations can perturb the fit of the model at low-effect concentrations. If the

form of the regression equation is known to be completely accurate, such perturbation

is appropriate; in this case, however, the equation is not expected to describe the exact

form of the concentration-effect curve over the whole range of T. Because high effect

concentrations contain useful information about the nature of the curve, they should not

be excluded, but they should not be allowed to unduly influence the fit in the range from

0 to 50 percent reduction. Consequently, normal weights were given to data points up

to the first concentration with a 50% or greater reduction relative to the control

treatment and points at higher concentrations were weighted by half. An alternative

was to use a more complicated form of the logistic equation (e.g., Moore and Caux

1997), but such equations introduce their own uncertainties, especially for small

datasets, and their main effect on calculation of the EC20 is to reduce the influence of

data points at high effects, with much the same results as the weighting scheme used

here.

SEs of the regression parameters were calculated based on the variance/covariance

matrix of the linearized model at the least-squares solution (Draper and Smith 1981)

and 95% confidence limits for the parameters were calculated by multiplying these SEs

by the applicable t-statistic. Simulations showed that this procedure produces

confidence levels that are near or greater than 95%. The

EC20

and its confidence

limits were computed by taking the antilog of the calculated logEC20 and its confidence

limits. Confidence limits on effect concentrations for percentages other than 20 and on

values for T at concentrations other than 0 were estimated by reformulating the

regression equation to use these values rather than EC20 and T

o as parameters, and

then recomputing the variance/covariance matrix at the least-squares solution to

determine the SEs of the new parameters.

Evaluation of the Chronic Data Available for Each Species

The following presents a species-by-species discussion of each chronic test on

ammonia evaluated by this project. For each species, the available chronic tests are

discussed in the following order: life-cycle tests, partial life-cycle tests, early life-stage

tests, other laboratory tests, and then results from a field study. Also presented are the

results of regression analysis of each dataset that was from an acceptable chronic test

and contained sufficient acceptable data. For each such dataset, Appendix 6 contains

a figure that presents the data and regression line, and a table with the regression

parameters. All analyses were conducted in terms of total ammonia nitrogen, either as

reported by the authors or as converted by us from the reported values for un-ionized

ammonia, pH, and temperature using the speciation relationship of Emerson et al.

(1975). When an EC20 could be determined, it is first reported as calculated by

regression analysis of the data at the pH and temperature of the test. Then, to facilitate

comparisons of sensitivities within and between species, each EC20 is adjusted to

48

---

pH=8 using Equation 12, and adjusted to a temperature of 25°C using Equation 5 with

a slope of -0.028, per the discussions of the temperature- and pH dependency

sections. Species Mean Chronic Values (SMCVs) were derived when justified by the

data, and then Genus Mean Chronic Values (GMCVs) were derived when justified by

the SMCVs. All of the EC20s, SMCVs, and GMCVs that were derived are tabulated in

Table 5, which is located at the end of this section.

Musculium transversum (Sphaerium transversum)

(Fingernail clam)

Anderson et al. (1978) conducted two 42-day tests of the effect of ammonia on

survival of field-collected juvenile clams whose length averaged 2.2 mm. The

results of the two tests were so similar that the data were pooled for analysis. The

lowest mean measured DO concentration in any treatment was 6.5 mg/L (77

percent of saturation) and the lowest individual measured concentration was 5

mg/L (60 percent of saturation). Survival in the control treatment and low ammonia

concentrations (<5.1 mg N/L) ranged from 79 to 90%, but decreased to zero at 18

mg N/L. Regression analysis of the survival data using an arcsine transformation

resulted in a calculated EC20 of 5.82 mg N/L at 23.5°C and pH=8.15. The EC20 is

6.63 mg N/L when adjusted to pH=8 and 25°C.

Sparks and Sandusky (1981) conducted a test similar to that of Anderson et al.

(1978) with field-collected juvenile clams whose average length was 2.1 mm.

Although this test used a better food, the test was conducted in the same laboratory

and used test organisms from the same pool in the Mississippi River as Anderson

et al. (1978); Sparks participated in both studies. The lowest mean measured DO

concentration in any treatment was 6.4 mg/L (73 percent of saturation) and the

lowest individual measured concentration was 5.0 mg/L (57 percent of saturation).

Survival in the control treatment was 92% and decreased with increasing

concentration of ammonia to 17% at 18 mg N/L. Effects on survival were evident at

lower concentrations, resulting in an EC20 of 1.23 mg N/L at 21.8°C and pH=7.80.

The EC20 adjusted to pH=8 and 25°C is 0.77 mg N/L. Although this EC20 is

substantially lower than that obtained by Anderson et al. (1978), the difference is

less than a factor of 10.

Zischke and Arthur (1987) studied fingernail clam growth, survival, and

reproduction in enclosures placed in experimental streams for periods

,

of 4 to 10

weeks during a 16-month field study of the effects of ammonia (Hermanutz et al.

1987). Experiments during the first year showed reductions in survival of clams in

a stream in which the concentration of total ammonia nitrogen was approximately 2

mg N/L during the test period (Hermanutz et al. 1987), but not in a stream in which

the concentration was 0.7 mg N/L. The daily mean stream temperature ranged

from 20 to 25°C and pH ranged from 7.4 to 7.8 during this test period.. During the

second year of the study, substantial effects occurred on reproduction of clams at 1

mg N/L (the lowest tested concentration of ammonia) at 24 to 26°C and pH=7.8 to

8.2 during the test period. Adjusted to pH=8, both years showed effects at about 1

mg N/L. These results are not included in Table 5 because results of field tests are

not used in the derivation of Final Chronic Values (U.S. EPA 1985b).

49

---

The SMCV at pH=8 and 25°C is s2.26 mg N/L. This concentration is the geometric

mean of the adjusted EC2Os for the two laboratory studies and is an upper limit on

the SMCV because the EC2Os are based on survival of juveniles, which might not

be as sensitive to ammonia toxicity as early life stages. This SMCV is uncertain

due to the difference between the results of the two chronic tests. However, the

experimental stream data suggest that the SMCV should be close to 1 mg N/L. The

GMCV is also s2.26 mg N/L.

Ceriodaphnia acanthina

Mount (1982) conducted a life-cycle test that started with <1-day-old organisms and

proceeded until most of the control organisms produced three broods. The DO

concentration ranged from 5.7 to 6.4 mg/L (68 to 77 percent of saturation). Total

offspring production per treatment was unaffected at concentrations s21 mg N/L,

but reproduction was virtually absent at concentrations



77 mg N/L. Regression

analysis using a square root transformation resulted in an EC20 of 44.9 mg N/L at

pH=7.15 and 24.5°C. The EC20 adjusted to pH=8 and 25°C is 19.1 mg N/L, which

is the SMCV.

Ceriodaphnia dubia

Willingham (1987) conducted a 7-day life-cycle test starting with <1-day-old

organisms. The lowest mean measured DO concentration in any treatment was

6.04 mg/L (74 percent of saturation) and the lowest calculated fifth percentile of the

DO concentrations was 5.62 mg/L (69 percent of saturation). Production of young

during the third brood was unaffected at concentrations up to 2.8 mg N/L, but was

reduced at higher concentrations and was absent at 43 mg N/L. The EC20

calculated using regression analysis was 5.80 mg N/L at pH=8.57 and 26.0°C.

Adjusted to pH=8 and 25°C, the EC20 is 15.6 mg N/L.

Nimmo et al. (1989) conducted a 7-day life-cycle test at 25°C and pH=7.8 in water

from the St. Vrain River. The DO concentration was reported to be low in some

other tests that were conducted during this study, but it was not reported to be low

in this test. Based on the average number of neonates per original female, the

EC20 calculated using regression analysis and a square root transformation was

15.2 mg N/L. Adjusted to pH=8 and 25°C, the EC20 is 11.6 mg N/L.

As stated above in the discussion of the effect of pH on the toxicity of ammonia,

Johnson (1995) conducted twelve chronic tests on ammonia with C.

dubia

at four

pHs and three hardnesses. The lowest reported mean concentration of DO was

6.9 mg/L (82 percent of saturation). When adjusted to pH=8, the mean EC25s are

9.03, 7.46, and 17.1 mg N/L at average hardnesses of 42, 86, and 170 mg/L,

respectively. These mean adjusted EC25s are similar to the adjusted EC2Os

obtained by Willingham (1987) and Nimmo et al. (1989). These EC25s are not

included in Table 5 because they are not EC2Os and were calculated using a

different regression-type approach.

50

---

Adjusted to pH=8 and 25°C, the two EC2Os for C.

dubia

are 15.6 and 11.6 mg N/L,

which gives a SMCV of 13.5 mg N/L. For C.

acanthina

at pH=8 and 25°C, the

SMCV is 19.1 mg N/L, which gives a GMCV of 16.1 mg N/L.

Daphnia magna

Gersich et al. (1985) and Gersich and Hopkins (1986) reported results of a life-

cycle test that was conducted in water from the Tittabawassee River. This water

was probably an acceptable dilution water because it was apparently collected

upstream of all known point discharges (Alexander et al. 1986; James Grant,

Michigan Department of Environmental Quality, personal communication). The

lowest and highest measured DO concentrations were 8.8 and 9.2 mg/L (96 and

101 percent of saturation). No significant effects were found at concentrations up

to 4.2 mg N/L at pH=8.45 and 19.8°C, but progressively larger reductions were

found at concentrations of 9 to 36 mg N/L. The EC20 calculated from regression

analysis was 7.37 mg N/L.

In another life-cycle test, Reinbold and Pescitelli (1982a) found little reduction in

reproduction at 20 mg N/L, but a large reduction at 33 mg N/L. The measured DO

concentrations averaged 88 to 91 percent of saturation. The EC20 is 21.7 mg N/L

at pH=7.92 and 20.1 °C.

Gulyas and Fleit (1990) conducted a 9-day chronic test to study the effect of

ammonia on development and growth. Concentrations that caused more than fifty

percent reduction compared to the controls were considered toxic. The "no effect

level" was reported to be 0.1 mg/L. No results from this test are included in Table 5

because neither survival nor reproduction was studied.

Adjusted to pH=8 and 25°C, the respective EC2Os are 10.8 and 14.1 mg NIL. The

SMCV for this species is 12.3 mg N/L, which is the geometric mean of the two

adjusted EC20s; this is also the GMCV.

Crangonyx

spp. (amphipod)

The available data for this species are not used for the reason(s) given in

Appendix 1.

Hyalella azteca

(amphipod)

Borgmann (1994) conducted three tests that began with <1-week-old organisms, all

of which utilized weekly renewals and dechlorinated tap water originating from Lake

Ontario. One of the three tests lasted four weeks, but the other two lasted ten

weeks and produced data concerning both survival and reproduction. The results

of these last two tests were sufficiently similar that the results were analyzed

together. No information was reported concerning the DO concentration. Sufficient

raw data were obtained from the author so that each test chamber could be plotted

as a separate point for the combined regression analysis. Survival over the ten

weeks in the control treatment averaged 66.3 percent and reproduction per

chamber averaged 48 offspring. The 33.7% mortality in the control treatment is

51

---

considered acceptable in a 10-week test because ASTM Standard E1706 (ASTM

1997e, Tables 10, 11, and 15) allows 10% mortality of

H. azteca

in a 4-day test and

allows 20% mortality in a 10-day test. In addition, although ASTM Standard E1706

allows 20% mortality in a 10-day test, Table 3 in Borgmann (1994) indicates that

only 11.6% of the controls died in four weeks.

At the lowest tested concentration, survival was reduced 25 percent relative to the

control treatment and reproduction was reduced 55 percent. Regression analysis

produced an EC20 of 0.88 mg N/L based on reproduction, but this EC20 is below

the lowest tested concentration because the dataset does not contain a

concentration that caused <20 percent reduction relative to the control treatment.

However, the confidence limits on the regression analysis indicate that the 55

percent reduction in reproduction caused by the lowest tested concentration is

statistically significant. Based on the raw data, the concentration of ammonia in the

lowest tested concentration was 1.58 mg N/L and the mean pH of this treatment

was 7.94. Therefore; the EC20 is <1.58 mg N/L at pH=7.94 and 25°C. Adjusted to

pH=8 and 25°C, the EC20 is <1.45 mg N/L. Even though chronic survival

appeared to be less sensitive than reproduction in this test, slightly more than 20%

mortality occurred at the lowest tested concentration; therefore, the LC50 for

chronic survival is <,-

.

1.45 mg N/L.

Because the test solutions were renewed once a week, the pH dropped and the

concentration of total ammonia increased between renewals; the average of the

weekly measured initial and final values was used for both pH and total ammonia.

The pH measured at the end of each week averaged 0.54 lower than the pH

measured at the beginning of each week in the control test chambers, and

averaged 0.78 lower in the two test chambers at the lowest tested concentration of

ammonia. Even though the average pH drop in the control test chambers for the

second test was 0.21 and was 0.87 in the control test chambers for the third test,

survival and reproduction were both higher in the control test chambers for the third

test; therefore, the pH variation probably did not reduce survival or reproduction.

The pH-adjustment was based on the average measured pH

in the lowest tested

concentration of ammonia. The SMCV and the GMCV are <1.45 mg N/L.

Procambarus clarkii

(crayfish)

The available data for this species are not used for the reason(s) given in

Appendix 1.

Pteronarcella badia

(stonefly)

Thurston et al. (1984a) studied the effect of ammonia on the survival and

emergence of nymphs from two sources for 30 and 24 days. When expressed in

terms of total ammonia nitrogen adjusted to pH=8, the 30-day LC50 for nymphs

from the Gallatin River was about 170 mg N/L, whereas the 24-day LC50 for

nymphs from Rocky Creek was about 70 mg N/L. The degree of development of

the nymphs at the beginning of each test was not determined and there is no

reason to believe that the tested life stage is the one that is most sensitive to

52

---

ammonia. In addition, it is not possible to interpret the data concerning emergence

from either test. The test with nymphs from the Gallatin River might have been

ended before emergence was complete in the control or any other treatment. In the

test with nymphs from Rocky Creek, 25 percent of the nymphs in the control

treatment neither died nor emerged, whereas this percentage was 5 to 15 in the

treatments that contained ammonia. These tests do not allow derivation of a SMCV

for this species, but they imply that this species is resistant to ammonia.

Carassius auratus

(goldfish)

Marchetti (1960) exposed fish for 90 minutes and then observed mortality and

histological effects for up to 42 days, whereas Reichenbach-Klinke (1967) studied

the effects of a one-week exposure on gills and blood. Neither study provided

useful information concerning the SMCV for the goldfish.

Pimephales promelas

(fathead minnow)

Thurston et al. (1986) reported similar results from two life-cycle tests that started

with 3 to 5-day-old fry and ended with 60-day-old offspring. The lowest mean

measured DO concentration in any treatment was 6.08 mg/L (72 percent of

saturation) and the lowest calculated fifth percentile of the DO concentrations was

5.16 mg/L (61 percent of saturation). At the highest tested un-ionized ammonia

concentration of 0.93 mg NH 3/L,

significant mortality occurred throughout the

development of the parental generation. The most sensitive effect was reduction in

egg hatching and the highest concentration that reportedly did not cause a

significant reduction in egg hatching was 0.19 mg NH 3/L, but this concentration

caused 33 and 55% reductions in percent hatch. For the purpose of regression

analysis of percent hatch, the tested concentrations and results were so similar in

the two tests that the data were analyzed as replicates of the test concentrations.

In terms of total ammonia nitrogen, the EC20 based on percent hatch was 1.97 mg

N/L at 24.2°C and pH=8.0. However, there are concerns about this test:

1. Effects on survival and weight of Fl fry were uncertain due to high mortality

attributed to handling during cleaning.

2. The eggs were dipped in malachite green daily.

3. Hatchability of the controls was about 50 percent.

4. There was a large difference between the replicate test chambers in the

control-adjusted percent hatch at 0.09 mg NH3/L.

Swigert and Spacie (1983) conducted a 30-day early life-stage test starting with 10

to 18-hour-old embryos. The fifth percentile of the measured DO concentrations

was 6.5 mg/L (79 percent of saturation) and the highest measured DO

concentration was 7.96 mg/L (97 percent of saturation). Both survival and weight

gain were reduced at 30 days and the product of these two (i.e., biomass) was

analyzed using regression analysis. The resulting EC20 was 3.73 mg N/L at

25.1 °C and pH=7.82, which would be 2.92 mg N/L at pH=8.

Mayes et al. (1986) conducted a 28-day early life-stage test in water from the

Tittabawassee River. This water was probably an acceptable dilution water

53

---

because it was apparently collected upstream of all known point discharges

(Alexander et al. 1986; James Grant, Michigan Department of Environmental

Quality, personal communication). The lowest and highest measured DO

concentrations were 5.0 and 8.5 mg/L (59 and 101 percent of saturation): Adverse

effects were observed on 28-day survival, but only the highest tested concentration

reduced weight. Regression analysis of the survival data resulted in an EC20 of

5.12 mg N/L at 24.8°C and pH=8.0.

As stated above in the discussion of the effect of temperature on the toxicity of

ammonia, DeGraeve et al. (1987) studied the effect of ammonia on 30-day survival

of juvenile fathead minnows at several temperatures. The tests at 15 and 20°C did

not have concentrations sufficiently high to cause effects, but survival was

significantly decreased at the higher concentrations of ammonia in the tests run at

6, 10, 25, and 30°C. At 30°C, the mean measured DO concentration in most of the

treatments was below 5.5 mg/L, but it was above 60% of saturation in all

treatments. EC20s based on survival were calculated to be 11.9, 13.8, 39, and 39

mg N/L at temperatures of 6.0, 10.0, 25.4, and 30.2°C and pHs of 7.83, 7.73, 7.35,

and 7.19, respectively. When adjusted to pH=8, the EC2Os are 9.45, 9.72, 19.35,

and 17.54 mg N/L, respectively. Although these EC2Os were used to assess the

effect of temperature on the chronic toxicity of ammonia, they are not included in

Table 5 and are not used in the derivation of the SMCV because they indicate that

30-day survival of juveniles is not as sensitive to ammonia as the life-cycle and

early life-stage tests discussed above.

The study of Smith (1984) concerned histopathological examination of lesions on

the test fish and cannot be used to calculate an EC20.

Hermanutz et al. (1987) studied the survival, growth, and reproduction of fathead

minnows in experimental streams. (See Appendix 8, titled "A Field Study Relevant

to the CCC.") Two generations were each exposed for periods of approximately

two months, during which pH averaged 7.5 to 7.7 and temperature averaged

19.6°C. Deleterious effects on biomass were not apparent at or below the highest

tested concentration of ammonia, which was 3.92 mg N/L when adjusted to pH=8.

These results are not included in Table 5 because they are from a field study.

In the 1985 Guidelines (U.S. EPA 1985b), results of early life-stage tests are used

as predictors of results of life-cycle and partial life-cycle tests; comparisons of

these kinds of chronic tests had been reported by McKim (1977) and Macek and

Sleight(1977). Because early life-stage tests are only predictors, results of such

tests are not used when results of life-cycle or partial life-cycle tests are available.

In the present case, however, because of the concerns about the life-cycle test, the

SMCV for the fathead minnow at pH=8 is set equal to 3.09 mg N/L, which is the

geometric mean of the three EC2Os from Thurston et al. (1986), Swigert and

Spacie (1983), and Mayes et al. (1986); the range of the three EC2Os is only a

factor of 2.6.

54

---

Catostomus commersoni

(white sucker)

Reinbold and Pescitelli (1982a) conducted a 31-day early life-stage test starting

with 3-day-old embryos. The concentration of DO averaged 68 to 74 percent of

saturation (6.3 to 6.9 mg/L). No effect on growth or survival was observed at

concentrations of total ammonia nitrogen up to 2.9 mg N/L at pH=8.32 and 18.6°C,

which is equivalent to 4.79 mg N/L at pH=8. As measured by time-to-swimup,

development of larvae was delayed, suggesting that slightly higher concentrations

would have affected growth and/or survival. The results of this test do not provide

sufficient data to allow regression analysis, but the data indicate that the EC20

would be greater than 4.79 mg NIL if an EC20 could be calculated.

Hermanutz et al. (1987) studied survival and growth of juvenile white suckers in

experimental streams. (See Appendix 8.) Two separate tests were started with

individuals whose average weight was 10 g and lasted 88 and 183 days. The

average temperatures in the two tests were 18 and 21°C. The two highest tested

concentrations'caused a slight reduction in biomass. However, juveniles might not

be as sensitive to ammonia toxicity as early life stages. These results are not

included in Table 5 because they are from a field study.

The value of ">4.79 mg N/L" is included in Table 5 and is the GMCV; even though it

is a "greater than" value, it can be used in the calculation of the FCV because it is

not one of the four lowest GMCVs.

lctalurus punctatus

(channel catfish)

Swigert and Spacie (1983) conducted a 30-day exposure starting with newly

hatched larvae that were fewer than 3 hours old: The mean measured DO

concentration was 5.66 mg/L (70 percent of saturation) but the lowest individual

measured concentration was 3.5 mg/L (45 percent of saturation). Reduced growth

was found at total ammonia concentrations of 5.8 mg N/L and above and reduced

survival at concentrations of 21 to 22 mg N/L. In separate tests, they determined

that survival and hatching of embryos were more resistant than survival and growth

of fry. Regression analysis of biomass at the end of the 30-day exposure produced

an EC20 of 11.5 mg N/L at pH=7.76 and 26.9°C. The EC20 adjusted to pH=8 is

8.38 mg N/L. This EC20 is questionable because the lowest measured DO

concentration was below 5.0 mg/L and was below 50 percent of saturation.

Reinbold and Pescitelli (1982a) conducted a 30-day exposure starting with <36-

hour old embryos. The concentration of DO averaged 70 to 76 percent of

saturation (5.7 to 6.2 mg/L). No effect on either percent hatch or fry survival was

found at concentrations up to 11 mg N/L, but reduced growth was found at 5.2 mg

N/L and above, as well as a delay in swimup at concentrations as low as 1 mg N/L.

The EC20 for growth is 12.2 mg N/L at pH=7.80 and 25.8°C. Adjusted to pH=8,

this EC20 is 9.33 mg N/L. However, the percent reduction at the highest tested

concentration was less than 50%, as specified above in the data requirements.

55

---

Colt and Tchobanoglous (1978) and Colt (1978) exposed juveniles for 31 days to

total ammonia nitrogen concentrations ranging from 1.6 to 14.4 mg N/L. The mean

measured DO concentration was 7.6 mg/L (97 percent of saturation) and the

calculated fifth percentile of the DO concentrations was 7.27 mg/L (93 percent of

saturation); the calculated 95th percentile of the DO concentrations was 7.93 mg/L

(101 percent of saturation). Biomass in the control treatment increased tenfold

during the test, but the increases were smaller at ammonia concentrations as low

as 1.6 mg N/L. Because this was a test with juveniles that lasted only 31 days, only

the data concerning mortality will be used. The concentration of 6.81 mg N/L killed

83%, whereas the higher concentration killed 100%. A range is reported for the

concentration of 5.71 mg N/L and so the mean percent mortality is between 28 and

45%. It was reported that the lower concentrations killed 9 of 400 organisms, and

so it is likely that the concentration of 5.02 mg N/L killed no more than 5%.

Therefore, the EC20 at pH=8.35 and 27.9°C is between 5.02 and 5.71 mg N/L;

adjusted to pH=8, the EC20 is between 8.7 and 9.9 mg N/L. Although this EC20 is

included in Table 5, it is not used in the derivation of the SMCV and GMCV

because it is based on survival of juveniles in a 31-day test and therefore is an

upper limit on the SMCV because juveniles might not be as sensitive to ammonia

toxicity as early life stages.

In several tests, each of which consisted of one concentration of ammonia and a

control, Robinette (1976) studied the effect of ammonia on growth of 25 to 30-g

channel catfish for about thirty days at 23 to 26°C. No information was reported

concerning survival of the test fish. A concentration of total ammonia nitrogen of

2.7 mg N/L at pH=7.6 caused fish to gain weight faster than the control fish. In

contrast, concentrations of 3.5 and 3.6 mg N/L at pH=7.8 caused fish to lose weight

while the controls were gaining weight. Adjusted to pH=8, these concentrations

would be 1.7, 2.7, and 2.8 mg N/L, respectively. Because these tests studied

growth of juveniles for only 30 days, the results are not included in Table 5.

Bader (1990) and Bader and Grizzle (1992) reported that ammonia reduced

growth, but the concentration of ammonia in the controls was substantial.

DeGraeve et al. (1987) studied the effect of ammonia on survival and growth of

juveniles for thirty days. Some of the test organisms were treated with acriflavine

up to two days prior to the beginning of the test. In addition, the mean measured

DO concentration was below 5.5 mg/L and below 60 percent of saturation in some

of the treatments. Mitchell and Cech (1983) reported that ammonia did not damage

gills unless residual chlorine was present. Soderberg et al. (1984) studied the

culture of channel catfish in ponds and found that the ambient concentration of

ammonia caused gill lesions, but did not affect survival or growth. Results of these

tests are not included in Table 5.

Hermanutz et al. (1987) studied survival and growth of juvenile channel catfish in

experimental streams. (See Appendix 8.) Three separate tests lasted from 36 to

177 days and were started with individuals whose average weights ranged from 6

to 19 g. Average temperatures in the three tests were 17 to 21°C. Both of the

56

---

longer tests showed monotonic, substantial reductions in biomass; these results

are in reasonable agreement with the results of the laboratory tests. However,

juveniles might not be as sensitive to ammonia toxicity as early life stages are.

These results are not included in Table 5 because they are from a field study.

Although there are problems with the early life-stage tests by Swigert and Spacie

(1983) and Reinbold and Pescitelli (1982a), the EC2Os are similar. Therefore, the

channel catfish SMCV at pH=8 is 8.84 mg N/L, which is the geometric mean of the

two EC20s. The data of Colt and Tchobanoglous (1978) and Robinette (1976)

support a SMCV of this magnitude. The GMCV is also 8.84 mg N/L.

Oncorhynchus clarki

(cutthroat trout)

Thurston et al. (1978) obtained 29-day LC5Os of 16.4 and 15.9 mg N/L with fish

whose average weights were 3.3 and 3.4 g, respectively; the 96-hr LC5Os were 1.2

and 1.7 times higher than the 29-day LC50s. In two other tests they obtained 36-

day LC5Os of 23.7 and 24.4 mg N/L with fish whose average weight was 1.0 g; no

fish died after day 29. The tests were conducted at 12.2 to 13.1°C and all four of

the LC5Os are expressed as total ammonia nitrogen at pH=8.0. The mean

measured DO concentrations for the various tests ranged from 8.2 to 8.6 mg/L (77

to 82 percent of saturation). The lowest and highest measured DO concentrations

were 7.4 and 9.2 mg/L (70 and 87 percent of saturation). EC2Os cannot be

calculated, but would be lower than the geometric mean of 19.7 mg N/L. The

SMCV might be substantially lower than 19.7 mg N/L because this test was not

conducted with an early life stage. In all four of the tests, there was a negative

correlation between the concentration of ammonia and weight gain, but this might

have been a temporary effect. Histological examinations were performed at the

end of the tests. The EC20 of <19.7 mg N/L is included in Table 5, but this value

cannot be used in the calculation of a SMCV.

Oncorhynchus gorbuscha

(pink salmon)

Rice and Bailey (1980) exposed embryos and alevins of pink salmon for 61 days to

concentrations of total ammonia nitrogen ranging from 0.07 to 13.6 mg/L at pH=6.4

and 4°C. The only chronic test began sometime after hatch and ended when the

alevins emerged (i.e., at the beginning of swimup); therefore the test did not

include effects of ammonia on the growth and survival of fry after feeding started.

In addition, no information was given concerning survival to the end of the test in

the control or any other treatment. At the higher tested concentrations, the weight

of emerging alevins was significantly reduced, relative to the controls, by as much

as 22% at 11.2 mg/L. This would be equivalent to about 4.1 mg N/L at pH=8. Size

at emergence was said to be important because smaller fry are less capable of

surviving in the environment because they have less swimming endurance and are

selectively preyed upon by larger predators. This test did not provide data

concerning survival and is not an early life-stage test because it began after hatch;

therefore, this test did not provide a useful EC20 and is not included in Table 5.

57

---

Oncorhynchus kisutch

(coho salmon)

Buckley et al. (1979) exposed fish whose average wet weight was 3.4 g for 91 days

to study effects of ammonia on blood. The highest tested concentration of 47 mg

N/L killed only three percent of the fish. The EC20 is >47 mg N/L, but this not

useful information about the SMCV because there is no reason to believe that the

tested life stage is the one that is most sensitive to ammonia. This test is not

included in Table 5 because it does not provide useful information concerning the

SMCV for this species.

Oncorhynchus mykiss (Salmo gairdneri)

(rainbow trout)

Many investigators have reported results of chronic tests conducted on ammonia

with rainbow trout, but the most ambitious chronic test was the five-year test

conducted by Thurston et al. (1984b). In this test the initial fish were exposed

through growth, maturation and reproduction, the next generation through hatch,

growth, maturation, and reproduction, and the third generation through hatch and

survival of the young. The mean measured DO concentration was 7.43 mg/L (65

percent of saturation) and the lowest calculated fifth percentile of the measured DO

concentrations in the various treatments was 5.9 mg/L (51 percent of saturation).

Measured temperatures ranged from 7.5 to 10.5°C and the tested concentrations of

total ammonia nitrogen ranged from 1.1 to 8.0 mg N/L at pH=7.7. When adjusted

to pH=8, the range is 0.77 to 5.4 mg N/L. All of the fish used to start the test came

from one pair of adults of the Ennis strain. In addition, the important data for each

life stage are so variable that it is not possible to discern whether there is a

concentration-effect curve. Despite the variability, it can be inferred that the EC20

cannot be much lower than the highest tested concentration because severe

effects were not apparent at any tested concentration; if the EC20 was much lower

than the highest tested concentration, this concentration would have caused severe

effects.

Also using fish from the Ennis strain, Burkhalter (1975) and Burkhalter and Kaya

(1977) reported a 21-day LC50 of 39.6 mg N/L for embryos and sac fry and

interpolation off a graph indicates a 42-day LC50 of 33.6 mg N/L, based on total

ammonia nitrogen, at 9.5 to 12.5°C and pH=7.5, assuming either no control

mortality or adjustment for control mortality. When adjusted to pH=8, the LC5Os

would be 22.0 and 18.7 mg N/L, respectively, but LC2Os would be lower than

LC50s. The measured DO concentrations were all above 8 mg/L (72 percent of

saturation). The test began within 24 hours of fertilization, continued to the

beginning of feeding, and found retardation of development and growth of very

young fish, similar to the tests discussed above with the pink salmon (Rice and

Bailey 1980). Thurston et al. (1984b) speculated that they did not observe the

reduced growth reported by Burkhalter and Kaya (1977) because of compensation

during the next several months of the longer exposure. Indeed, Burkhalter and

Kaya (1977) reported compensation at the lowest tested concentration.

Contrasting information concerning EC2Os is provided by the early life-stage tests

conducted by Solbe and Shurben (1989) and Calamari et al. (1977,1981). Both

58

---

tests began within 24 hours after fertilization and lasted for 72 to 73 days until the

fry had been feeding for about 30 days.

1.

Solbe and Shurben (1989) reported that the dry weight of the test organisms

varied little between treatments. The test was conducted at pH=7.52 and an

average temperature of 14.9°C. The DO concentration equaled or exceeded

76 to 95 percent of saturation during various portions of the test. The four

highest concentrations of ammonia killed 78 to 99 percent. The fifth and lowest

tested concentration of total ammonia nitrogen was 2.55 mg N/L and it reduced

survival by 67 percent; this would correspond to 1.44 mg N/L at pH=8, and the

LC20 would be lower. These authors demonstrated that exposure to ammonia

should begin soon after fertilization. When exposure began within 24 hours

after fertilization, 26 mg N/L killed 98 percent of the embryos, whereas when

exposure began 24 days after fertilization, 26 mg N/L killed only 3 percent of

the embryos and killed only 40 percent in a 49-day exposure.

2.

Calamari et al. (1977,1981) conducted an early life-stage test, but did not

report any information concerning weight, although, as stated above, Solbe and

Shurben (1989) reported no effect on weight during their early life-stage test.

The DO concentration was over 80 percent of saturation. For total ammonia

nitrogen at pH=7.4, Calamari et al. (1977,1981) obtained a 72-day LC50 of 8.2

mg N/L at 14.5°C. They also reported that adjusted mortalities were 15 and 23

percent at 1.5 and 3.7 mg N/L, respectively, and that higher tested

concentrations killed more than 50 percent of the test organisms. Because

Calamari et al. did not report the actual percentage killed at the higher tested

concentrations, regression analysis could not be applied; semilog interpolation

between 1.5 and 3.7 mg N/L produced an LC20 of 2.6 mg N/L, which would

correspond to 1.34 mg N/L at pH=8.

Both Calamari et al. (1977,1981) and Solbe and Shurben (1989) found that longer

exposures of embryos and fry resulted in much lower LC5Os than 96-hour

exposures.

Several investigators reported results concerning the effect of total ammonia

nitrogen on long-term survival:

1. Thurston and Russo (1983) reported five 35-day LC5Os that were determined

using fish whose average initial weights were 0.7 to 10 g. The 35-day LC5Os

were 27.9 and 36.1 mg N/L for fish whose average weights were 3.7 and 9.7 g,

respectively. The 35-day LC5Os were 32.4, 34.5, and 37.0 mg N/L for fish

whose average weights were 0.7 to 3.3 g; when adjusted to pH=8, the

geometric mean of these three 35-day LC5Os with the smaller fish was 26.4 mg

N/L.

2.

Broderius and Smith (1979) reported that 16.2 mg N/L killed 30 percent of fry in

30 days at 10°C and pH=7.95, which corresponds to 15.1 mg N/L at pH=8.

3.

Daoust and Ferguson (1984) reported that 23.3 mg N/L did not kill any

fingerlings in 90 days at pH=7.93, which would correspond to 21.1 mg N/L at

pH=8. However, some of the fish that exhibited clinical signs during the

exposure were removed for examination during the test. The swimming and

feeding of some fish were affected for a while, but the fish recovered.

59

---

This variety of results might be due to differences in the size or age of the test

organisms.

Several other chronic tests did not provide information that could be used in the

derivation of a SMCV. Fromm (1970), Reichenback-Klinke (1967), and Smart

(1976) exposed fish to study the effects of ammonia on gills and blood. In a test

reported by Smith and Piper (1975), exposed fish had abnormal tissues, but fish

placed in clean water for 45 days at the end of the test had normal tissues. When

Soderberg et al. (1983) studied the culture of rainbow trout in ponds, parasitic

epizootics caused mortalities. The Ministry of Technology (1967) reported the

effect of ammonia on percent survival in a 90-day test, but did not report the age or

size of the fish or the temperature or the pH of the water. Samylin (1969)

conducted tests in water from the Vyg River, with some of the exposures being

conducted in Petri dishes. Schulze-Wiehenbrauck (1976) found that growth of

juveniles at 10°C and pH=8 was reduced during two-week exposures to a total

ammonia nitrogen concentration of 2.26 mg N/L, but the decrease was completely

compensated for during the next three or four weeks. Smith (1972) reported that as

long as the DO concentration was maintained at 5 mg/L or greater, growth of

rainbow trout was not significantly reduced until average total ammonia

concentrations reached 1.6 mg/L.

Hermanutz et al. (1987) studied survival and growth of juvenile rainbow trout in

experimental streams. (See Appendix 8.) Three separate tests were conducted

with individuals whose average initial weights were 7 to 11 g. The tests lasted from

28 to 237 days, with the 237-day test including an entire winter. Average

temperatures in the three tests ranged from 5.9 to 10.6°C, whereas pH averaged

7.7 to 8.4. Reductions in biomass were consistently observed at concentrations

greater than or equal to 2.29 mg N/L when adjusted to pH=8. However, juveniles

might not be as sensitive to ammonia toxicity as early life stages. These results

are not included in Table 5 because they are from a field study.

The early life-stage test by Calamari et al. (1977,1981) produced a total ammonia

nitrogen LC20 of 1.34 mg N/L at pH=8, whereas Solbe and Shurben (1989)

indicate that the LC20 might be lower. In contrast, both Thurston et al. (1984a) and

Burkhalter and Kaya (1977) found no indication of severe mortality of young fish at

higher concentrations. Exposure was continuous for several generations in the test

of Thurston et al. (1984b), whereas exposure began within 24 hours of fertilization

in the other three tests. Because of the concerns about some of the tests, the

differences among the results, and the fact that some of the results are either

"greater than" or "less than" values, even though the various results are included in

Table 5, a SMCV is not derived for rainbow trout; instead, the results of the chronic

tests will be used to assess the appropriateness of the CCC.

Oncorhynchus nerka

(sockeye salmon)

Rankin (1979) exposed embryos of sockeye salmon for 62 days from fertilization to

hatch; the tested concentrations of total ammonia nitrogen ranged from 2.13 to 87

60

---

mg N/L at 10°C. The DO concentration was reported to be at saturation. This test

ended as soon as the embryos hatched, and so hatchability was the only

toxicological variable studied. The percentage of the embryos that hatched was

63.3% in the controls, but was 49% at the lowest tested concentration (2.13 mg

N/L) and was 0% at 8.1 mg N/L and above. The concentration of 2.13 mg N/L at

pH=8.42 corresponds to 4.16 mg N/L at pH=8. Thus the EC20 at pH=8 is less than

4.16 mg N/L. Because the effects on newly hatched fish were not studied, the

SMCV is <4.16 mg N/L.

Oncorhynchus tshawytscha

(chinook salmon)

Burrows (1964) exposed fingerlings for six weeks at 6 and 14°C to three

concentrations of ammonia and a control treatment to study effects on gills at

pH=7.8. There was no recovery in three weeks in clean water at 6°C, but there

was recovery at 14°C. At both temperatures, no significant mortality occurred

during exposure to the highest tested concentration of 0.57 mg N/L or for three

• weeks afterward in clean water. No information is given concerning the DO

concentration during the exposures, and there is no reason to believe that the

tested life stage is the one that is most sensitive to ammonia.

Tests conducted by Sousa et al. (1974) suggest that chinook salmon tolerate

higher concentrations of ammonia when pH is decreased and salinity is increased.

However, there was no control treatment, no information was given concerning the

DO concentration, temperature was not controlled, and the fish were given an

antibiotic.

These tests are not included in Table 5 because they do not provide useful

information concerning the SMCV for this species.

A GMCV is not derived for

Oncorhynchus

because the available data do not

provide an adequate basis for a useful conclusion concerning the GMCV.

Salmo trutta

(brown trout)

Carline et al. (1987) exposed brown trout for twelve months to dilutions of effluent

from a sewage treatment plant. Survival, growth, swimming performance, and

degree of damage to gills were studied, but no information was obtained

concerning effects on embryos, newly hatched fish, or reproduction. No data from

this test are included in Table 5 because this test does not provide useful

information concerning the SMCV for this species.

Lepomis cyanellus

(green sunfish)

Reinbold and Pescitelli (1982a) conducted a 31-day early life-stage test that

started with <24-hour-old embryos. No information was reported concerning the

DO concentration but it averaged 70 to 76 percent of saturation (5.7 to 6.2 mg/L) in

a similar test in the same report with another fish species at about the same

temperature. The weight data were not used in the calculation of an EC20 because

the fish were heavier in chambers containing fewer fish, which indicated that weight

61

---

was density-dependent. Although overflows resulted in loss of fish from some

chambers, survival was 96 percent in one of the chambers affected by overflow,

indicating that the survival data were either adjusted or not affected by the

overflows. Survival to the end of the test was reduced at total ammonia nitrogen

concentrations of 6.3 mg N/L and above and regression analysis of the survival

data calculated an EC20 of 5.84 mg N/L at pH=8.16 and 25.4°C. Adjusted to

pH=8, the EC20 is 7.44 mg N/L.

McCormick et al. (1984) conducted a 44-day early life-stage test, starting with <24-

hour-old embryos. The mean measured DO concentration was 7.9 mg/L (91

percent of saturation) and the calculated fifth percentile of the measured DO

concentrations was 7.7 mg/L (88 percent of saturation). No effect was found on

percent hatch, but reduced survival and growth occurred at concentrations of 14

mg N/L and above. Although survival in one control test chamber and in the low

concentrations of ammonia averaged about 40 percent and was only 10 percent in

the other control chamber, the concentration-effect curve was well defined.

Regression analysis of biomass calculated an EC20 of 5.61 mg N/L at pH=7.9 and

22.0°C. This EC20 was obtained with the 10 percent used in the regression

analysis. An EC20 of 5.51 mg N/L was obtained if the 10 percent was not used; the

two EC2Os are similar partly because the weight given to each treatment was

inversely related to the variance for the treatment, which meant that the control

treatment was given a low weight in the regression analysis. Adjusted to pH=8, the

EC20 calculated using all of the data is 4.88 mg N/L.

Jude (1973) found that growth of juveniles weighing 4 to 16 g each for 40 days was

proportional to temperature at 13, 22, and 28°C. In a second test, the effect of

ammonia on survival and growth of 10 to 14-g juveniles was studied for 20 days.

Too few fish died to allow calculation of an EC20. Neither of these tests provided

results that can be included in Table 5.

Adjusted to pH=8, the EC20 of 7.44 mg N/L from Reinbold and Pescitelli (1982a)

agrees quite well with the EC20 of 4.88 mg N/L from McCormick et al. (1984). It is

possible that the second value is lower because it was based on survival and

growth, whereas the first value was based only on survival. Even though there

were experimental problems with both tests, the results of the tests agree well and

therefore the geometric mean (6.03 mg N/L) of the two EC2Os is used as the

SMCV.

Lepomis macrochirus

(bluegill)

Smith et al. (1984) conducted a 30-day early life-stage test, starting with <28-hour-

old embryos. No information was reported concerning the DO concentration, but

the flow-rate was high. The values reported in their Table 1 as standard deviations

on the pH appear excessively large; it is likely that they were not calculated

correctly, because, as explained in footnote d, the mean pH was calculated by

conversion of pH to H

+

(i.e., hydrogen ion) concentration. Other tests conducted

on ammonia in the same laboratory at about the same time reported much less

62

---

variation in pH. For example, McCormick et al. (1984) reported that the 95%

confidence interval on the experiment-wide pH was 7.8 to 8.0. Broderius et al.

(1985) calculated average pH by converting to hydrogen ion concentration, but

reported small standard deviations and ranges for four acute tests and four chronic

tests.

Smith et al. (1984) found no significant reduction in percent hatch up to a total

ammonia nitrogen concentration of 37 mg N/L, but hatched larvae were deformed

at this concentration and died within six days. At the end of the test, survival and

growth at 1.64 mg N/L were near values for the controls, but were greatly reduced

at 3.75 to 18 mg N/L. Regression analysis of biomass calculated an EC20 of 1.85

mg N/L at pH=7.76 and 22.5°C. The EC20 adjusted to pH=8 is 1.35 mg N/L.

Diamond et al. (1993) conducted two chronic tests. The test at 12°C is discussed

in Appendix 1. The data sheets for the test at 20°C indicate that this test studied

the effect of ammonia on survival and growth of bluegills for 21'days. (The

durations of the chronic tests with the bluegill at 12 and 20°C are switched in Table

1 in the publication.) The test at 20°C was started with bluegills that were less than

98-days old, were less than 1 inch (2.5 cm), and averaged 0.11 to 0.15 g. The

highest tested concentration of total ammonia nitrogen was 64 mg N/L, which

caused 30% mortality at the test pH of 7.3; most of the deaths occurred in the last

two days

.

of the test. Adjusted to pH=8, the highest tested concentration was 31 mg

N/L as total ammonia nitrogen, which is in the range of the adjusted 96-hr LC5Os

reported in Table 1 of the 1984/1985 ammonia criteria document. This test is not

very useful because it lasted for only 21 days and mortality began occurring near

the end of the test. Neither of these tests provides results that can be included in

Table 5.

Hermanutz et al. (1987) studied survival and growth of the juvenile bluegills in

experimental streams. (See Appendix 8.) The individual weights averaged 2.2 g at

the beginning and the test duration was 90 days. The mean pH and temperature

were 8.2 and 21.1 °C, respectively. A substantial effect on biomass was apparent

only at the highest concentration, which was 9.5 mg N/L when adjusted to pH=8.

These juvenile bluegills were not particularly sensitive compared to older life

stages of other species tested during this study. However, juveniles apparently are

not as sensitive to ammonia toxicity as the early life stages tested by Smith et al.

(1984). These results are not included in Table 5 because they are from a field

study.

The SMCV for the bluegill is 1.35 mg N/L, and the GMCV of 2.85 mg N/L for

Lepomis

is calculated as the geometric mean of the two SMCVs (6.03 and 1.35 mg

N/L).

Micropterus dolomieu

(smallmouth bass)

As stated above in the discussion of the effect of pH on the toxicity of ammonia,

Broderius et al. (1985) conducted 32-day early life-stage tests at four pHs at

63

---

22.3°C, starting with embryos near hatch. The mean measured DO concentration

was 7.72 mg/L (89 percent of saturation); the lowest and highest measured DO

concentrations were 7.1 and 8.3 mg/L (81 and 96 percent of saturation). Survival

of embryos and fry within the first week was not affected by ammonia, except at the

highest concentration at the highest pH, although effects on these life stages might

have been reduced due to the exposure not starting until just prior to hatch. In all

tests, growth and survival of older fry were reduced at higher concentrations and

regressions of biomass resulted in EC2Os of 9.61, 8.62, 8.18, and 1.54 mg N/L at

pHs of 6.60, 7.25, 7.83, and 8.68, respectively. Adjusted to pH=8, these EC2Os are

3.57, 4.01, 6.50, and 4.65 mg N/L, with a geometric mean of 4.56 mg N/L, which is

the SMCV and the GMCV.

Stizostedion vitreum

(walleye)

Reinbold and Pescitelli (1982a) could not conduct a successful early life-stage test

because only 20% of the newly hatched fish survived.

Hermanutz et al. (1987) studied survival and growth of juvenile walleyes in

experimental streams. (See Appendix 8.) A 46-day test was conducted at an

average temperature of 24°C and was started with yearlings averaging 100 g initial

weight. A second test at an average temperature of 17°C was started with young-

of-year averaging 19 g initial weight and lasted 43 days. Adjusted to pH=8,

concentrations of 2.0 to 3.7 mg N/L somewhat reduced walleye biomass, whereas

concentrations of 9.5 to 13.3 mg N/L completely eliminated walleye from the

streams. However, juveniles might not be as sensitive to ammonia toxicity as early

life stages. These results are not included in Table 5 because they are from a field

study.

Rana pipiens

(leopard frog)

The available data for this species are not used for the reason(s) given in

Appendix 1.

Hyla crucifer

(spring peeper)

The available data for this species are not used for the reason(s) given in

Appendix 1.

64

---

Table 5. EC2Os from Acceptable Chronic Testsa

Species

Reference

Test and

Effectb

Temp.

(C)

pH

EC20c at

test pH

& Temp.

(mg N/L)

EC20b

at pH=8

& 25°C

(mg N/L)

SMCV

at pH=8

& 25°C

(mg N/L)

GMCV

at pH=8

& 25°C

(mg N/L)

Musculium

transversum

Anderson et al.

1978

42-d Juv

Survival

23.5

8.15

5.82

6.63

s2.26

s2.26

Sparks and

Sandusky 1981

42-d Juv

Survival

21.8

7.80

1.23

0.77

Ceriodaphnia

acanthina

Mount 1982

LC

Reproduction

24.5

7.15

44.9

19.1

19.1

Ceriodaphnia

16.1

dubia

Willingham

1987

7-d LC

Reproduction

26.0

8.57

5.80

15.6

13.5

Nimmo et al.

1989

7-d LC

Reproduction

25.

7.8

15.2

11.6

Daphnia magna

Gersich et

al. 1985

21-d LC

Reproduction

19.8

8.45

7.37

10.8

12.3

12.3

Reinbold and Pescitelli

1982a

21-d LC

Reproduction

20.1

7.92

21.7

14.1

Hyalella

azteca

Borgmann 1994

10-wk LC

Reproduction

25.

7.94

<1.58

(EC50)

<1.45

<1.45

<1.45

Pimephales

promelas

Thurston et al.

1986

LC

Hatchability

24.2

8.0

1.97

1.97

Swigert and

3.09

3.09

Spacie 1983

30-d ELS

Biomass

25.1

7.82

3.73

2.92

Mayes et al.

1986

28-d ELS

Survival

24.8

8.0

5.12

5.12

65

---

Species

Reference

Test and Effectb

Temp.

(C)

pH

EC20b at

test pH

& Temp.

(mg N/L)

EC2Oc

at pH=8

& 25°C

(mg N/L)

SMCV

at pH=8

& 25°C

(mg N/L)

GMCVb

at pH=8

& 25°C

(mg N/L)

Catostomus

commersoni

Reinbold and Pescitelli

1982a

30-d ELS

Biomass

18.6

8.32

>2.9

>4.79

>4.79

>4.79

Ictalurus

punctatus

Swigert and Spacie

1983

30-d ELS

Biomass

26.9

7.76

11.5

8.38

Reinbold and Pescitelli

8.84

8.84

1982a

30-d ELS

Weight

25.8

7.80

12.2

9.33

Colt and

Tchobanoglous 1978

30-d Juv

Survival

27 . 9

8.35

s5.02-

s5.71

s8.7-

s9.9d

Oncorhynchus

clarki

Thurston et al.

1978

29-d Juv

Survival

12.2-

13.1

8.0

<19.7

<19.7d

Oncorhynchus

mykiss

Thurston et al. 1984b

5-year LC

7.5-

10.5

7.7

>.8.0

>--.5.4d

Burkhalter and

Kaya 1977

42-d ELS

Survival

9.5-

12.5

7.5

<33.6

<18.7d

Solbe and

Shurben 1989

73-d ELS

Survival

14.9

7.52

<2.55?

.

<1.44d

Calamari et

al. 1977,1981

72-d ELS

Survival

14.5

7.4

2.6

1.344

Oncorhynchus

nerka

Rankin 1979

62-d Embryos

Hatchability

10.

8.42

<2.13

<4.16

<4.16e

66

---

Species

Reference

Test and Effectb

Temp.

(C)

pH

EC20a at

test pH

& Temp.

(mg N/L)

EC20a

at pH=8

& 25°C

(mg N/L)

SMCV`

at pH=8

& 25°C

(mg N/L)

GMCVC

at pH=8

& 25°C

(mg N/L)

Lepomis

cyanellus

Reinbold and Pescitelli

1982a

30-d ELS

Survival

25.4

8.16

5.84

7.44

6.03

McCormick et al. 1984

30-d ELS

2.85

Biomass

22.0

7.9

5.61

4.88

Lepomis

macrochirus

Smith et al.

1984

30-d ELS

Biomass

22.5

7.76

1.85

1.35

1.35

Micropterus

dolomieu

Broderius et

al. 1985

32-d ELS

Biomass

22.3

6.60

9.61

3.57

4.56

4.56

Broderius et

al. 1985

32-d ELS

Biomass

22.3

7.25

8.62

4.01

Broderius et

al. 1985

32-d ELS

Biomass

22.3

7.83

8.18

6.50

Broderius et

al. 1985

32-d ELS

Biomass

22.3

8.68

1.54

4.65

a

An EC20 is assumed for a stonefly but is not given in this table (see text concerning calculation of the

b

CCC).Juv

= juvenile; LC = life cycle; ELS = early life stage.

a

Total ammonia nitrogen.

Not used in the derivation of a SMCV (see text).

e

Not used in the derivation of a GMCV (see text).

67

---

Seasonality of Chronic Toxicity Endpoints

Dischargers that use biological treatment of ammonia are likely to find it most difficult to

meet water quality-based effluent limits for ammonia when the temperature is low. This

has raised questions about whether criteria based on toxicity tests conducted mainly at

warm temperatures appropriately define concentrations that are needed for aquatic life

protection under cold-weather conditions. The effect of temperature on chronic EC2Os

was discussed in an earlier section of this document. This section will consider the

relevance of particular chronic endpoints at various times of the year.

The CMC is appropriate during all portions of the year because the organisms (i.e.,

juvenile and adult fish) and effects (i.e., survival) on which it is based are relevant

during all portions of the year and because available data indicate that these endpoints

are largely independent of temperature. The CCC, however, is based in part on

endpoints that might not be of concern during cold-season conditions (fish early life

stages,

Hyalella

reproduction). Consequently, it is also necessary to consider the

effect of seasonality on the chronic endpoint selection.

Endpoint Selection for Fish

Two of the four most sensitive GMCVs used to quantify the chronic criterion are for fish

genera --

Pimephales

and

Lepomis.

The next most sensitive genera were also fish

(Micropterus, lctalurus, Catostomus),

which could become more important in the criteria

derivation if one or more of the four most sensitive genera tested at 20-25°C are

sufficiently less sensitive at low temperatures. All of these GMCVs were based on

early-life stage or life-cycle tests. Because the sensitive test endpoints involved

survival and/or growth of very young fish that are present only at temperatures at which

reproduction occurs, these GMCVs are not necessarily appropriate for lower

temperatures. Seasonal criteria should consider the likely response of life stages

present during particular seasons.

The appropriate tests for deriving criteria when early life stages are not present would

be long chronic tests of survival and growth with older life stages, conducted at a wide

range of temperatures. Such data are not available. Life cycle tests with fathead

minnows conducted by Thurston et al. 1986 provide survival and growth data for the

parental generation from a few days of age through spawning (150-200 days), but only

at one temperature (24°C). DeGraeve et al. (1987) conducted chronic exposures with

fathead minnows at four different temperatures, but the duration was only 30 days.

Other chronic exposures with juvenile warm-water fish are limited by both being only at

high temperature and having relatively short durations.

Thus, available chronic data for juvenile and adult fish are somewhat limited with

respect both to the duration of the tests and the temperature range of the tests. With

regard to the limited temperature range, the importance and likely magnitude of

68

---

temperature effects was discussed above. Despite various uncertainties, available

information can serve as the basis for reasonable assumptions on how tests at warm

temperatures can be used to derive criteria applicable to lower temperatures. For the

calculations in this document, it will be assumed that for fish the EC20 for a particular

type of effect does not vary with temperature.

With regard to the duration issue, it is not justified to simply assume that 30-day

exposures are long enough to assess chronic effect concentrations for juvenile and

adult fish, even though this is the duration accepted for early life stage tests for warm

water fish. For such early life stage tests, a test duration of 30 days is considered

adequately long because it encompasses a wide range of developmental stages that

likely include the most sensitive endpoints, and because empirical comparisons have

suggested these tests are often nearly as sensitive as longer tests. However, this is

not necessarily true if the tests of interest address less sensitive endpoints involving

juvenile and adult survival and growth. For example, in the fingernail clam tests the

effect concentrations (juvenile survival) at 6 weeks were significantly lower than those '

at 4 weeks. Also, for 90-day ammonia exposures to rainbow trout reported by Ministry

of Technology (1967), most of the observed mortality occurred after 30 days.

With regard to the warm water fish of concern here, the effect of duration is most

evident in the fathead minnow life cycle test by Thurston et al. (1986). Starting with

larvae a few days old, these investigators reported survival after 30 days, after 60 days,

and just prior to spawning (155 days in one test and 200 days in another). The first 30

days falls under the realm of early life stage tests and will be ignored here. The

mortality between 30 and 60 days covers a fish age and exposure duration similar to

short chronic tests with juveniles, such as that of DeGraeve et al. (1987). During that

time, mortality was approximately 25% at a total ammonia concentration of 18 mg N/L

and averaged <2% at lower concentrations. This suggests an LC25 of about 18 mg

N/L, which is consistent with the results of DeGraeve et al. at temperatures of 25 and

30°C. But mortality continued after this 30 day period. By spawning time, total

mortality at 18 mg N/L reached about 67% and a mortality of about 10% was observed

at 9 mg N/L, in contrast to an average mortality of <2% at lower concentrations. The

estimated LC25 for this longer exposure was 12.3 mg N/L. This is a factor of 1.5 less

than the LC25 at the shorter exposure. This factor will be used to adjust 30-day effect

concentrations from other tests with juvenile fish to be more reflective of long chronic

exposures.

With the issues of test temperature and test duration so addressed, chronic, "non-

early-life-stage" effect concentrations for different species of warm-water fish can be

assessed as follows:

Fathead minnow

DeGraeve et al. (1987) exposed juvenile fathead minnows to ammonia for 30 days

at various temperatures, as previously discussed. When corrected to pH=8 using

the chronic pH

relationship and divided by the factor of 1.5 to adjust to longer

exposures, chronic LC2Os from this study at pH=8 are estimated to be 6.5, 8.5,

69

---

12.9, and 10.6 mg N/L at 6, 10, 25, and 30°C, respectively. If temperature effects

are assumed to be negligible, the geometric average of these four values, 9.3 mg

N/L, can be used as a representative value for long-term LC2Os for juveniles of this

species. If there were concern about better representing low temperature

conditions, the average of the two values at lower temperatures, 7.5 mg N/L, could

be used.

As discussed above, Thurston et al. (1986) conducted two life-cycle exposures of

fathead minnows to ammonia at 24°C and pH.--8, starting with 3- to 5-day-old

larvae. For the exposure of fish from 30 day old through spawning, the data and

regression indicate a long-term chronic LC20 of 11.4 mg N/L, similar to the higher

temperature results from DeGraeve et al. (1987).

While the effects concentrations summarized here can be directly used to set a

GMCV for fathead minnows in the absence of early life stages, they can also be

used to develop ratios to other endpoints. Such ratios might be useful for

estimating GMCVs for other species for which chronic tests with juveniles and

adults have not been conducted. One useful comparison would be to the early life

stage and life cycle test results. For fathead minnows, Table 5 listed the GMCV at

pH=8 as 3.09 mg N/L based on the results of two early life stage tests and one life

cycle test. Using the estimate of 9.3 mg N/L from above as representative of what

the GMCV should be when early life stages are absent, this represents a factor of 3

increase in effect concentration. A second comparison that is useful is to the acute

toxicity of this species. The GMAV for fathead minnow at pH=8 is listed as 43.6 mg

N/L in Table 4. This is 4.7-fold higher than the 9.3 mg N/L derived here for chronic

toxicity of older life stages from the DeGraeve et al. (1987) data.

Channel catfish

One of the chronic tests with channel catfish was a 31-day test of juvenile growth

and survival by Colt and Tchobanoglous (1976). These authors did not provide

complete data on survival at each concentration, but the information provided

indicated that (a) at 5.0 mg N/L (total ammonia) and below, mortality averaged

about 2%, (b) at 5.7 mg N/L mortality was between 28 and 45% (based on a

reported range of 11 to 62% mortality in three replicates), (c) at 6.8 mg N/L

mortality was 83%, and (d) at 9.5 mg N/L and above mortality was 100%. This

indicates that the LC20 was between 5.0 and 5.7 mg N/L (at pH=8.35 and T=28°C),

or between 8.7 and 9.9 mg N/L when adjusted to pH=8 based on the chronic pH

relationship. While this test was not used in deriving the GMCV, it did indicate that

juvenile channel catfish were as sensitive as early life stages and was included as

support that the GMCV derived from the early life stage tests was reasonable. In

fact, if the factor of 1.5 is applied to this number to make it more applicable to

longer chronic exposures, the GMCV would be around 6.2 mg N/L -- below the

GMCV (8.84 mg N/L) derived from the early life stage tests. Because the 1.5

adjustment factor has questionable applicability to channel catfish, the GMCV from

Table 5 is the most appropriate number to use for both early-life-stage and juvenile

sensitivity.

70

---

Juvenile channel catfish show effects on growth at even lower concentrations. The

EC20 for growth from the study of Colt and Tchobanoglous was 2.39 mg N/L at

pH=8.35, or 4.2 mg N/L when adjusted to pH=8. Robinette (1976) also found

growth effects at similar concentrations. In separate 27- and 29-day exposures

with 3.5-3.6 mg N/L total ammonia (pH=7.8), he found fingerling channel catfish to

lose weight, while control fish gained weight. These results were not used in the

criterion derivation, because growth effects on juvenile fish are sometimes

transient, so that short chronic tests with such fish might overestimate risk.

Nonetheless, this does indicate effects of ammonia at rather low concentrations.

The relationship of these effects on juvenile channel catfish to effects observed in

early life stages are quite different than in fathead minnow. Whereas fathead

minnow juvenile tests show about three-fold less sensitivity than early life stage

tests, channel catfish juvenile tests seem to be as, or more, sensitive than the

available early life stage tests. This might be in part due to comparisons across

different studies, but the differences are so great that it does suggest some

fundamental interspecies differences in this regard. It also makes uncertain the

development of a ratio which can be applied to early life stage tests of other

species to estimate what GMCVs should be for later life stages. In contrast, the

relationship of the juvenile chronic test results to acutely lethal concentrations

seem more consistent. In Table 4, the channel catfish SMAV at pH=8 is listed as

34.4 mg N/L total ammonia. This is a factor of 5.5 greater than the long-term

juvenile LC20 estimated above and a factor of 3.9 greater than the GMCV for

catfish in Table 5, both similar to the factor of 4.7 for fathead minnow.

Bluegill sunfish

Bluegills are the only other nonsalmonid species for which juveniles have been

chronically exposed in the laboratory. Diamond et al. (1993) conducted a 21-day

exposure with young bluegill (ca. 150 mg) and reported significant mortality (30%)

and reduced growth at 64 mg N/L total ammonia (pH=7.4, T=22°C). A comparison

of this value to effect concentrations discussed above requires adjustment for (a)

the low pH, (b) the short duration, and (c) mortality>20%. Using the chronic pH

relationship, 64 mg N/L at pH=7.4 corresponds to 33 mg N/L at pH=8. The LC20

can be estimated to be 28 mg N/L based on the average ratio (0.82) between an

LC20 and LC30 in the fathead minnow chronic tests of DeGraeve et al. Regarding

duration, the heaviest mortality in this study was in the last few days, so LC2Os at

even slightly longer durations would be less, but it is uncertain by how much. Even

if the 30-day LC20 is just 25% less than the 21-day LC20, when combined with the

factor of 1.5 used above for adjusting 30-day tests to long term mortality, the long-

term chronic value would be estimated to be no higher than 14 mg N/L.

Diamond et al. (1993) also conducted a 14-day exposure at 13°C and for this

exposure they reported 80% mortality at 53 mg N/L total ammonia (pH=7.7), but

none at 21 mg N/L and below. Linear interpolation of mortality versus log

concentration would suggest an LC20 of 29 mg N/L. Adjustment to pH=8 would

lower this value to 20 mg N/L. This should be lowered by at least a factor of 2 to

71

---

account for the short duration of this test, suggesting the chronic value should be

no higher than 10 mg N/L. However, this test is made uncertain by a variety of

factors. First, the mortality was mainly in the first four days so that the effect

concentration is actually mainly determined by acute mortality, although some

mortality at 53 mg N/L occurred throughout the two weeks. Increased temperatures

and decreased pHs occurred on the fifth through seventh days. Dissolved oxygen

concentrations were not measured for the first five days of exposure and were low

on the sixth day, although they were acceptably high through the rest of the

exposure.

In the experimental stream study of Hermanutz et al. (1987), bluegills with an initial

average weight of 2.2 g grew during a 90-day exposure to an average of 26.9 g,

28.1 g, and 26.0 g in the control and lowest two exposure concentrations,

respectively. At the highest exposure concentration (9.4 mg N/L total ammonia,

pH=8.2, T=21 °C), the average final weight was 14.6 g, about a 50% reduction in

growth. Adjusted to pH=8, this corresponds to an EC50 of 12.7 mg N/L. .This

would be reduced to a value of 7.6 by applying an EC20:EC50 ratio (=0.6) from the

channel catfish chronic growth study of Colt and Tchobanogous. Again, while this

study might not be applied directly to criterion calculations, it does support the

need for a GMCV<10 mg N/L for older bluegills.

Unlike for fathead minnow and channel catfish, this information does not allow

good quantification of a chronic effect concentration for non-early-life-stage

bluegills. The concentration for bluegill will be quantified using procedures

proposed in the next subsection.

Other fish

Although chronic ammonia toxicity to older life stages of other warm water species

have not been tested in the laboratory, chronic tests with juvenile salmonids (as

summarized previously) also suggest that GMCVs should be below 15 mg N/L (at

pH=8). If the limited duration of these tests is considered, concentrations should

probably be restricted to below 10 mg N/L. In the experimental stream study of

Hermanutz et al. (1987), rainbow trout showed significantly reduced survival and/or

growth in three separate exposures at mean total ammonia concentrations of from

4.4 to 8.0 mg N/L (adjusted to pH=8). Also in this stream study, in two six-week

exposures using walleye pike, complete mortality was observed at concentrations

of 14 to 21 mg N/L and reduced growth was observed at 3.8-7.0 mg N/L (adjusted

to pH=8).

While such data are not directly used in criteria derivation, they do support an

expectation that juvenile fish chronic values should be <10 mg N/L. But this

doesn't allow actual computation of "non-early-life-stage" (non-ELS) GMCVs for

warm water fish species for which there are no juvenile chronic tests.

To quantify this concentration, one approach would be to apply the ratios among

endpoints discussed above for fathead minnow and channel catfish. However, the

72

---

ratio for the non-ELS to ELS chronic concentrations is variable: about 1 for the

catfish, 3 for the fathead minnow and maybe even higher given the information

presented above on the bluegill. The ratio for the SMAV to non-ELS chronic

concentration is more constant: 4.4 for the fathead and 5.5 for the catfish. This

latter ratio is analogous to the acute-chronic ratios often used in criteria, except

estimated here using non-ELS chronic effects. However, the average of the above

two values (approximately 5) is too high to apply to some other species of concern,

because it would place the non-ELS chronic values at or below the ELS chronic

values. Furthermore, this approach does not take into account how close acute

and chronic toxic effects concentrations are for a particular species.

Another approach to quantifying this concentration is to note that, for both fathead

minnow and channel catfish, the non-ELS chronic effect concentration is closer, on

a relative scale, to the early-life-stage effect concentration than it is to the SMAV

for these species (i.e., the ratio of non-ELS to ELS chronic is less than the ratio of

acute to non-ELS chronic). In the absence of good non-ELS information, a

reasonable default assumption would be to set the non-ELS SMCV as the

geometric mean of the ELS SMCV and the SMAV. This makes the non-ELS

concentration have the same ratio to both of the other concentrations.

Using this procedure, the non-ELS GMCVs will be 8.78 mg N/L for

Lepomis

and

9.55 mg N/L for

Micropterus.

For

Lepomis,

this number is consistent with the

information presented above which suggested concentrations below 10 mg N/L

might be needed. It is several times the SMCV for bluegill and 50% greater than

the SMCV for green sunfish. It is about three-fold lower than the SMAV for both

species, making it a reasonable factor below the acute concentrations. For

Micropterus,

this number is slightly more than double the SMCV and slightly less

than half the SMAV.

Endpoint Selection for Invertebrates

Two of the four most sensitive GMCVs used to quantify the chronic criterion were for

invertebrates: the fingernail clam

Musculium

and the amphipod

Hyalella.

The endpoint

for the fingernail clam toxicity tests was 6-week survival of juvenile clams, and would be

relevant to all temperatures. The endpoint for the amphipod toxicity tests was

reproduction, which would not be relevant for low temperatures. However, the LC20 for

juvenile survival in these ten-week tests was the same as the effects concentration

used for reproduction, the difference being that for reproduction this concentration

actually was for >20% effect and was reported as a "less than" value, because the IC20

could not be reliably calculated. In any event, the same concentration can be used for

lower temperatures because it applies to juvenile and adult mortality as well as to

reproduction.

73

---

Formulation of the CCC

Nine Genus Mean Chronic Values (GMCVs) are presented in Table 5. Although

Table 5 contains chronic data for the genus

Oncorhynchus,

no GMCV is derived

because of the large range in the EC20s; rather these chronic data will be used to

evaluate whether the CCC poses a risk to this genus.

Although Table 5 does not contain data for an insect genus, available information

concerning a stonefly (Thurston et al. 1984a) indicates that at least one species is

relatively resistant to ammonia. Therefore, calculation of the fifth percentile directly

from the GMCVs in Table 5 should adequately reflect the intent of the 1985 Guidelines.

For this calculation the number of tested species was taken to be 10, because a GMCV

for an insect is assumed to be

.

greater than the four lowest GMCVs. The fifth percentile

value calculated by this procedure could be considered to be a "less than" value

because the lowest two GMCVs are "less than" values.

The relevant relationships for formulating a seasonal, pH- and temperature-dependent

chronic criterion are summarized below.

pH-dependence of chronic toxicity

The chronic pH dependence derived from tests with

Micropterus

and

Daphnia

are

used here. The acute pH dependence is not applied to chronic toxicity because

the measured acute-chronic ratios change substantially with pH. Equation 12,

presented earlier in the document, describes the shape of chronic pH dependence.

Temperature-dependence of chronic toxicity - fish

Available data suggest minimal dependence of fish ammonia toxicity on

temperature. Even where some acute toxicity data indicate higher LC5Os at low

temperatures (i.e., for fathead minnows), applying this to chronic toxicity is

contraindicated because available data, as well as theoretical considerations,

suggest that acute-chronic ratios increase at lower temperatures. Although limited

available chronic data suggest LC20s might be lower at low temperatures, the

effect is small and uncertain. This criteria formulation assumes no temperature

dependence for fish endpoints.

Temperature-dependence of chronic toxicity - invertebrates

Available data suggest a strong dependence of invertebrate

acute

ammonia toxicity

on temperature, with a mean log LC50 versus temperature slope of -0.036 over the

entire temperature range, or slightly higher, around -0.044 above approximately

10°C. No data are available regarding the temperature dependence of

invertebrate chronic toxicity, but both the fish data and theoretical expectations

argue against assuming that the chronic slope is as steep as the acute slope.

Alternative approaches for the chronic slope were discussed at the end of the

chronic temperature relationship section earlier in the document. The approach

selected for calculating the criterion is to set the invertebrate chronic slope equal to

74

---

an invertebrate acute slope of -0.044 above 7°C, minus the observed fish ACR

slope of -0.016. This yields an invertebrate chronic slope of -0.028 above 7°C.

The slope is then set to zero below 7°C.

Endpoint selection - fish

The GMCVs appropriate for fish depend on whether early life stages of fish are

present. Determining when or whether time periods exist when such life stages are

present requires some information about the site or eco-region. The national

criterion will specify two sets of numbers between which the applied criterion can

select depending on the presence or absence of fish early life stages.

When fish early life stages are present, the GMCVs are as presented in Table 5,

that is, 2.85 mg N/L for

Lepomis,

3.09 mg N/L for

Pimephales,

4.56 mg N/L for

Micropterus,

and 8.84 mg N/L for

lctalurus).

When fish early life stages are not

present, the assigned GMCVs are 9.30 mg N/L for

Pimephales,

8.84 mg N/L for

ictalurus,

8.78 mg N/L for

Lepomis,

and 9.55 mg N/L for

Micropterus.

Although

these GMCVs are uncertain, the calculation of the criterion ends up being

insensitive to these uncertainties. That is, because invertebrate GMCVs control

the value of the criterion, the fish non-ELS GMCVs would need to be reduced by

nearly 50% before they would affect the criterion.

Endpoint selection - invertebrates

The GMCV for

Musculium

is applicable to all temperatures because it is for long-

term juvenile survival. The GMCV for

Hyalella

is applicable to all temperatures

because, although it was derived for reproductive effects, the same effects

concentration would be obtained based on juvenile survival. Therefore, the

temperature-normalized GMCVs in Table 5 are used: 2.26 mg N/L for

Musculium

and 1.45 mg N/L for

Hyalella.

Temperature- and pH-dependent Criteria Calculation

Part of a criterion derivation is the estimation of the fifth percentile of sensitivity based

on the set of toxicity values available for different genera. This fifth percentile estimate

is intended to be what would be obtained by simple inspection if many genera had

been tested. When the number of tested genera is less than 19, this will ordinarily

entail an extrapolation below the lowest value, because, in such small sets, the lowest

value would not be expected to be as low as the fifth percentile (i.e., if many more

genera were tested, it is likely that the fifth percentile genus would be more sensitive

than any in the small data set). One characteristic of such extrapolations is that, if the

genera vary widely in sensitivity, the extrapolated criterion value will be further below

the lowest value than if the genera are tightly grouped. This is statistically appropriate

because if the variance is high, the fifth percentile of a distribution would be expected

to lie further below the lowest datum of a small data set than if the variance was low.

75

---

However, while this characteristic of extrapolations is desirable for criteria derivations

across chemicals with different variances for genus sensitivities, it is not necessarily

appropriate when the genera are following different temperature or seasonal

dependencies, as is the case here. As sensitivities change with temperature or life

stage, the spread of the lowest four GMCVs changes, and thus the degree of

extrapolation also changes, such that the criterion does not end up increasing in

concert with reduced sensitivities as the temperatures decrease. That is, if the fifth

percentile genus value is re-extrapolated at each different temperature, the resulting

temperature dependence of the criterion would be found not to accord with any

temperature dependence that could be reasonably assumed for untested sensitive

species. Therefore, while temperature dependent criteria could be calculated by

developing sets of GMCVs for each temperature and computing the CCC from the four

most sensitive GMCVs at each temperature, this approach will not be used here.

Rather, the derivation here will start with a chronic criterion calculated at 25°C. This

reference temperature was selected because it is the temperature for the chronic test

with the most sensitive genus (Hyalella), as well as being within 3°C of all the other

tests with the four most sensitive genera. Consequently, temperature extrapolation

uncertainties are minimized. Table 5 presented the GMCVs normalized to 25°C. The

GMCVs for the four most sensitive species are thus as follows:

1.45 mg N/L for Hyalella

2.26 mg N/L for Musculium

2.85 mg N/L for Lepomis

3.09 mg N/L for Pimephales.

With N=10, this results in a CCC of 1.24 mg N/L at 25°C and pH=8. Figure 13 shows

the ranked GMCVs and the CCC, all at pH=8 and 25°C.

At 25°C, the 1999 CCC is two percent lower than the 1998 CCC, due to the

temperature adjustment of the

Musculium

EC2Os from the tested temperatures to the

25°C reference temperature. (In contrast to the 1999 CCC, the 1998 CCC did not

change with temperature.)

The CCC of 1.24 mg/L is 15 percent lower than the lowest GMCV (1.45 mg N/L for

Hyalella).

This degree of extrapolation below the lowest GMCV is modest and

reasonable given the low number of tested genera, and will be used for criteria

calculations for all other temperatures and for conditions of fish early life stages

present or absent. As discussed above, the alternative of calculating the CCC directly

from new sets of GMCVs for each condition results in different degrees of extrapolation,

ranging up to 50% below the lowest GMCV, an extrapolation that would not be

reasonable here.

Because the most sensitive genera are invertebrates, the criterion will vary with

temperature according to the invertebrate chronic temperature relationship, but cannot

exceed 85.4% of the lowest fish GMCV. The lowest seasonal GMCVs for fish are

76

---

■

■

■

■■

■■

■

■

CCC

100

10

1

2.85 mg N/L for

Lepomis

early life stages, and 8.78 mg N/L for

Lepomis

juveniles and

adults. When fish early life stages are present, the upper limit would be 2.43 mg N/L

(that is, 0.854 • 2.85 mg N/L lowest fish GMCV). When fish early life stages are

absent, the upper limit (0.854 • 8.78 mg N/L lowest juvenile and adult fish GMCV) is

never reached under any temperature condition. Therefore, the CCC equals 85.4

percent of lower of (a) the temperature-adjusted

Hyalella

GMCV, or (b) the lowest fish

GMCV.

With fish early life stages present,

at pH=8 the criterion would be expressed

as:

CCC = 0.854 • MIN ( 2.85 , 1.45

. 100.028.(25-T)

)

?

(19)

This function increases steadily with decreasing temperature, T, until it reaches its

maximum (0.854 • 2.85).at 14.5°C, below which it remains constant.

Figure 13. Ranked Genus Mean Chronic Values (GMCVs) with the

Criterion Continuous Concentration (CCC).

Ranked Genera

77

---

With early life stages absent,

at pH=8 the 2.85 mg N/L GMCV for early life stages of

Lepomis

would have been replaced by the 8.78 mg N/L GMCV for juvenile and adult

Lepomis.

However, since the latter GMCV is so high that it is never limiting, it can be

dropped. In this case the invertebrate slope plateau will become limiting below 7°C,

and this condition is therefore incorporated into the equation for early life stages absent

(at pH=8):

CCC = 0.854 1.45 • 10

0.028

.

(25 - MAX(T, 7))

?

(20)

This function increases steadily with decreasing temperature, T, until it reaches its

plateau at 7°C.

The pH dependency is incorporated through Equation 12, which converts chronic effect

concentrations from pH=8 to any other pH. Consequently, the seasonally varying, pH-

and temperature-dependent CCC would be expressed as follows.

For fish early life stages present:

CCC =

0.854 •(

0.0676

?

2.912 ?

) MIN (2.85 , 1.45-

1

0"2"25-T)

)

?

(21)

1 +1

07'688-P"

1

+ 0pH-7.688

For fish early life stages absent,

CCC = 0.854 • ( ?

0.0676 ?

+

? ?

2.912

?

)

1.45. 1

00'028.(25-MAXM7))

1 + 07.688-pH

?

+ 0PH-7.688

In the final criteria statement, these equations can be shortened by multiplying

coefficients together.

The 1999 CCC with fish early life stages present is plotted as the "new" CCC in

Figure 14, along\ with the 1984/1985 "old" CCC and the EC2Os from Table 5. The 1999

CCC is near the 1984/1985 CCC in the range of pH from about 7.5 to 8, but is

increasingly higher than the old CCC at lower and higher values of pH. At pH=8 and

25°C, the new CCC corresponds to acute-chronic ratios of (14.4 mg N/L)/(1.24 mg N/L)

= 11.6 using the calculated FAV when salmonids are present (but not lowered to

protect large rainbow trout) and (16.8 mg N/L)/(1.24 mg N/L) = 13.5 using the FAV

when salmonids are absent. These are in the range of the ACRs that can be derived

from the EC2Os in Table 5 (see Appendix 7). The ACR used to calculate the old CCC

was 13.5 (Heber and Ballentine 1992).

(22)

78

---

50

0

20

10

■

•

•

5

New CCC

Old CCC, 10 C

2

Old CCC,

20 C

1

0

Musculium

q

Ceriodaphnia

0

Daphnia

0.5

Hyalella

•

Pimephales

Catostomus

lctalurus

■

Oncorhynchus

0.2

♦ Lepomis

•

Micropterus

■

?

0

•

■

♦

•0►

v

q

■

•

Figure 14. Chronic EC2Os used in criteria derivation in relationship to Criterion

Continuous Concentrations (CCCs). The curve labeled "New CCC" is the

1999 CCC at a temperature of 24.6°C, at which it equals the 1998 CCC.

The "Old CCC" is from 1984/1985.

0.1

67

?

8

?9

pH

79

---

Several points should be noted concerning the CCC:

a.

The two lowest GMCVs are "less than" values. The CCC would be lower if a point

estimate, rather than a "less than" value, could have been derived from the

Borgmann (1994) study with

Hyalella,

the most sensitive genus. The CCC also

might be lower if a point estimate, rather than a "less than" value, could have been

derived from the studies with the fingernail clam.

b.

At 25°C and pH=8 any substantial increase in the CCC derived with the procedures

in this 1999 Update would require a higher GMCV for

Hyalella

and possibly a

higher SMCV for the recreationally important bluegill.

c.

Because acutely resistant taxa are under-represented in the chronic dataset in

Table 5, it could be argued that N, the number of genera used in the calculation of

the CCC, should be increased from 10 to a higher value. A reasonable increase in

N would not have a large effect, however. For example, adding three resistant

genera would raise the CCC less than 10 percent.

d.

The central tendency of the available chronic EC2Os for salmonids, even though

not used directly in the calculation of the CCC, indicates that these species would

probably be protected by the CCC. Nevertheless, in some tests, effects were

observed at concentrations below the lower-temperature values of the CCC. The

data suggest that there might be important differences between strains of rainbow

trout.

e.

Some of the laboratory and field data for the fingernail clam, which might be

considered to have special ecological importance at some sites, indicate that this

species would be affected at concentrations below the CCC. Other data indicate

that it would not be affected by such concentrations. At most sites the intermittency

of exposures would probably reduce risk.

f.

When a threatened or endangered species occurs at a site and sufficient data

indicate that it is sensitive at concentrations below the CCC, it is appropriate to

consider deriving a site-specific criterion.

g.

Partly for statistical reasons, the CCC is based on a 20 percent reduction in

survival, growth, and/or reproduction. Whether the maximum acceptable percent

reduction should be lower or higher than 20 percent under a set of conditions is a

risk management decision. Consult Appendix 6 for regression parameters to

calculate ECs corresponding to other percentages.

h.

If it had been derived using available acute-chronic ratios (see Appendix 7), the

CCC would be greater than 2 mg N/L, which would be inappropriate because (1) it

would be above one of the GMCVs in a dataset for which N is only 10, (2) it would

not appear to protect early life stages of the recreationally important bluegill, and

(3) it might not protect the fingernail clam.

80

---

Chronic Averaging Period

The averaging period for a CCC often needs to be shorter than the length of the tests

upon which it is based for two main reasons. First, concentrations in the field are

typically much more variable than concentrations in laboratory tests, and variable

concentrations of ammonia have been shown to be more toxic than constant

concentrations when the comparisons are based on average concentrations during the

exposure (Thurston et al. 1981 a). By shortening the averaging period to which the

CCC applies, the average concentration over the entire exposure will be below the

CCC, increasingly so as the variability of the concentration increases. Second, chronic

tests generally encompass different life stages, which might have different sensitivities,

so that effects might be elicited only, or disproportionately, during the fraction of the

test in which a sensitive life stage is present, rather than cumulatively over the whole

test. The 1984/1985 ammonia criteria document specified a CCC averaging period of 4

days as recommended in the 1985 Guidelines (U.S. EPA 1985b), except that an

averaging period of 30 days could be used when exposure concentrations were shown

to have "limited variability". The purpose of this section is to better define when a 30-

day averaging period is acceptable.

Tests having different durations and/or starting with organisms of different ages can

indicate how restrictive the averaging period needs to be. The best information

available is for the fathead minnow. Based on 7-day tests, EC2Os of 7.08 mg N/L at

pH=8.34 and 5.25 mg N/L at pH=8.42 were calculated from the data of Willingham

(1987) and CVs of 8.37 mg N/L at pH=8 and 3.87 mg N/L at pH=8.5 were reported by

Camp Dresser and McKee (1997). Adjusted to pH=8, these concentrations are 12.1,

10.25, 8.37, and 8.65 mg N/L, respectively, with a geometric mean of 9.7 mg N/L. This

is approximately 2.5 times the geometric mean EC20 for the 30-day early life-stage

tests conducted by Swigert and Spade (1983) and Mayes et al. (1986) as discussed

above. This suggests that the CCC averaging period could be 30 days, as long as

excursions above the CCC are restricted sufficiently to not exceed the mean EC20 from

the 7-day tests. A rigorous definition of this excursion restriction is not possible with

the limited data available, especially because no information is available concerning

the effects of variations within the 7-day period. It is convenient, however, to base the

excursion restriction on a 4-day period, because this period is the default that already

has to be considered in calculations of water quality-based effluent limits, and because

it provides a substantial limitation of variability relative to the 7-day EC20s. It is

uncertain how much higher than the CCC the 4-day average can be, but based on

these fathead minnow test results, 2.5-fold higher concentrations should be acceptable.

Some other data support the use of a longer averaging period. For example, the

studies of Anderson et al. (1978) and Sparks and Sandusky (1981) with fingernail

clams showed that effects gradually accumulated during exposures, suggesting that

longer averaging periods are acceptable. Also, in the field study at Monticello, time

variations in pH yielded time variations in the applicable CCC. Analysis of the data

presented by Zischke and Arthur (1987) for the fingernail clam indicated that limiting

81

---

the highest 4-day average concentration to 2.0-2.5 times the CCC would protect this

species, whereas application of a 30-day average without this stipulation would allow

substantial effects on this species. In addition, Calamari et al. (1977,1981) and Solbe

and Shurben (1989) found that longer exposures of embryos and fry resulted in much

lower LC5Os than 96-hr exposures.

In contrast, some other studies suggest possible risks from longer averaging periods

under variable concentrations. For channel catfish, Bader (1990) reported a 24%

reduction in growth at 2.4 mg N/L in 7-day tests with young fry at pH=8.2; this

corresponds to just 3.3 mg N/L at pH=8, which is lower than the adjusted EC2Os

reported from longer early-life stage tests and juvenile tests in Table 5. This suggests

that a short averaging period is advisable, but such a conclusion is very uncertain

because it involves interlaboratory comparisons with very few data and because Bader

(1990) also found similar sensitivity with older fry, so his results might represent a high

sensitivity of the test stock rather than factors relevant to the averaging period. A short

averaging period might also be inferred by the fact that the fathead minnow life-cycle

test (Thurston et al. 1986) showed an EC20 of 2.0 mg N/L for embryo hatchability,

substantially lower than for early life-stage tests. It is possible that this greater

sensitivity might be due to exposures starting earlier in the life-cycle tests than in the

early life-stage tests. The importance of early exposure to embryos was demonstrated

by Solbe and Shurben (1989) for rainbow trout. However, they dealt with a one-week

delay in exposures rather than <1 day and there are other possibilities for the more

sensitive results of Thurston et al. (1986).

Based on the fathead minnow early life-stage data, a 30-day averaging period is

justified with the restriction that the highest 4-day average within the 30 days is no

greater than twice the CCC. The data of Bader (1990) and Thurston et al. (1986)

suggest a potential risk from long averaging periods during fish spawning season, but

the evidence is weak and, even if variability within long averaging periods produces

short exposures that are sufficiently high to affect young embryos, only a small fraction

of total reproduction would generally be affected. A high priority should be given to

research to resolve how to better address different time-series of exposure.

82

---

The National Criterion For Ammonia in Fresh Water

The available data for ammonia, evaluated using the procedures described in the

"Guidelines for Deriving Numerical National Water Quality Criteria for the Protection of

Aquatic Organisms and Their Uses", indicate that, except possibly where an unusually

sensitive species is important at a site, freshwater aquatic life should be protected if

both of the following conditions are satisfied for the temperature, T, and pH of the

waterbody:

1. The one-hour average concentration of total ammonia nitrogen (in mg N/L) does

not exceed, more than once every three years on the average, the CMC (acute

criterion) calculated using the following equations. Where salmonid fish are

present:

?

CMC -

?

0.275

? 39.0

?

1

+

10

7.204-pH

?

+ 10 pH-7.204

Or where salmonid fish are not present:

?

CMC -

?0.411

?

58.4

?

1 + 10

7.204-pH

?

1

+ 10 pH-7.204

2A.

The thirty-day average concentration of total ammonia nitrogen (in mg N/L) does

not exceed, more than once every three years on the average, the CCC (chronic

criterion) calculated using the following equations.

When fish early life stages are present:

CCC - ?

0.0577 ?

+

? ?

2.487

?

)

• MIN(2.85, 1.45•10"28-

(25-T))

1 + 107.688-pH

?

+ 10pH -7.688

When fish early life stages are absent:

=

?

CCC

+107.688-pH0.0577 ?

?

+

?

+ 0pH

2.487

-7.688

?

) .

1.45 .

1

0 0.028- (25 -MAX (T, 7))

2B.

In addition, the highest four-day average within the 30-day period should not

exceed 2.5 times the CCC.

83

---

Several points should be noted concerning the criterion:

1.

The two lowest GMCVs are "less than" values. The CCC would be lower if a point

estimate, rather than a "less than" value, could have been derived from the

Borgmann (1994) study with

Hyalella,

the most sensitive genus. The CCC also

might be lower if a point estimate, rather than a "less than" value, could have been

derived from the studies with the fingernail clam.

2.

The central tendency of the available chronic EC2Os for salmonids, even though

not used directly in the calculation of the CCC, indicate that these species would

probably be protected by the CCC, although the data suggest that there might be

important differences between strains of rainbow trout, and some tests indicated

effects at concentrations below the CCC.

3.

Some of the laboratory and field data for the fingernail clam, which might be

considered to have special ecological importance at some sites, indicate that this

species would be affected at concentrations below the CCC. Other data indicate

that it would not be affected by such concentrations. At most sites the intermittency

of exposures would probably reduce risk.

4.

When a threatened or endangered species occurs at a site and sufficient data

indicate that it is sensitive at concentrations below the CCC, it is appropriate to

consider deriving a site-specific criterion.

5.

Partly for statistical reasons, the CCC is based on a 20 percent reduction in

survival, growth, and/or reproduction. Whether the maximum acceptable percent

reduction should be lower or higher than 20 percent under a set of conditions is a

risk management decision. ECs corresponding to other percentage reductions can

be calculated using the parameter values presented in Appendix 6.

The Recalculation Procedure, the WER Procedure, and the Resident Species

Procedure may be used to derive site-specific criteria for ammonia, but most WERs

that have been determined for ammonia are close to 1.

The CMC, CCC, and chronic averaging period presented above supersede those given

in previous guidance on the aquatic life criterion for ammonia in fresh water. The 1998

and 1999 Updates do not address or alter the past recommendation of a one-hour

averaging period for the CMC or the past recommendation of a once-in-three years on

the average allowable frequency for exceeding the CMC or CCC. Many issues

concerning the implementation of aquatic life criteria are discussed in the "Technical

Support Document for Water Quality-based Toxics Control" (U.S. EPA 1991).

If concentrations in 95 percent of grab or 24-hour composite samples do not exceed the

CCC, then the 30-day average concentrations are unlikely to exceed the CCC more

then once in three years (Delos 1999). This assumes that concentrations are log

normally distributed, with first-order log serial correlation coefficient between 24-hour

composite samples approximately 0.86 - 0.94 (or less), and a log standard deviation of

0.5 - 0.8 (or less).

Because the ammonia criterion is a function of pH and temperature, calculation of the

appropriate weighted average temperature or pH is complicated. For some purposes,

84

---

calculation of an average pH and temperature can be avoided. For example, if

samples are obtained from a receiving water over a period of time during which pH

and/or temperature is not constant, the pH, temperature, and the concentration of total

ammonia in each sample should be determined. For each sample, the criterion should

be determined at the pH and temperature of the sample, and then the concentration of

total ammonia nitrogen in the sample should be divided by the criterion to determine a

quotient. The criterion is attained if the mean of the quotients is less than 1 over the

duration of the averaging period.

85

---

pH-Dependent Values of the CMC (Acute Criterion)

CMC, mg N/L

PH

Salmonids

Present

Salmonids

Absent

6.5

32.6

48.8

6.6

31.3

46.8

6.7

29.8

44.6

6.8

28.1

42.0

6.9

26.2

39.1

7.0

-?

24.1

36.1

7.1

22.0

32.8

7.2

19.7

29.5

7.3

17.5

26.2

7.4

15.4

23.0

7.5

13.3

19.9

7.6

11.4

17.0

7.7

9.65

14.4

7.8

8.11

12.1

7.9

6.77

10.1

8.0

5.62

8.40

8.1

4.64

6.95

8.2

3.83

5.72

8.3

3.15

4.71

8.4

2.59

3.88

8.5

2.14

3.20

8.6

1.77

2.65

8.7

1.47

2.20

8.8

1.23

1.84

8.9

1.04

1.56

9.0

0.885

1.32

86

---

Temperature and pH-Dependent Values of the CCC (Chronic Criterion)

for Fish Early Life Stages Present

CCC for Fish Early Life Stages Present, mg N/L

Temperature, C

p1-1

0

14

16

18

20

22

24

26

28

30

6.5

6.67

6.67 6.06

5.33 4.68

4.12

3.62 3.18

2.80 2.46

6.6

6.57 6.57

5.97

5.25

4.61

4.05

3.56

3.13 2.75 2.42

6.7

6.44 6.44

5.86

5.15

4.52

3.98

3.50 3.07

2.70

2.37

6.8

6.29

6.29 5.72

5.03

4.42 3.89

3.42

3.0G 2.64

2.32

6.9

6.12 6.12

5.56

4.89 4.30

3.78

3.32 2.92

2.57 2.25

7.0

5.91

5.91

5.37

4.72

4.15

3.65

3.21

2.82 2.48 2.18

7.1

5.67 5.67

5.15

4.53

3.98 3.50

3.08

2.70 2.38 2.09

7.2

5.39 5.39 4.90

4.31

3.78

3.33 2.92

2.57

2.26 1.99

7.3

5.08

5.08 4.61

4.06

3.57 3.13 2.76

2.42 2.13 1.87

7.4

4.73 4.73

4.30

3.78

3.32 2.92 2.57

2.26

1.98

1.74

7.5

4.36 4.36

3.97

3.49

3.06

2.69

2.37

2.08

1.83 1.61

7.6

3.98 3.98

3.61

3.18

2.79

2.45 2.16

1.90

1.67 1.47

7.7

3.58 3.58

3.25

2.86 2.51

2.21

1.94

1.71

1.50 1.32

7.8

3.18 3.18 2.89

2.54

2.23

1.96

1.73

1.52

1.33 1.17

7.9

2.80 2.80

2.54

2.24

1.96

1.73

1.52

1.33

1.17 1.03

8.0

2.43 2.43 2.21

1.94

1.71

1.50

1.32

1.16

1.02 0.897

8.1

2.10 2.10

1.91

1.68

1.47

1.29

1.14

1.00

0.879

0.773

8.2

1.79

1.79

1.63

1.43

1.26

1.11

0.973 0.855

0.752 0.661

8.3

1.52

1.52

1.39

1.22

1.07

0.941 0.827

0.727

0.639 0.562

8.4

1.29

1.29 1.17

1.03

0.906 0.796

0.700 0.615 0.541

0.475

8.5

1.09

1.00

0.990 0.870 0.765

0.672 0.591 0.520

0.457 0.401

8.6

0.920 0.920

0.836 0.735 0.646

0.568 0.499 0.439

0.386

0.339

8.7

0.778

0.778 0.707 0.622

0.547 0.480

0.422 0.371 0.326

0.287

8.8

0.661 0.661 0.601

0.528 0.464 0.408

0.359 0.315 0.277

0.244

8.9

0.565

0.565 0.513 0.451

0.397 0.349 0.306

0.269

0.237 0.208

9.0

0.486 0.486

0.442 0.389

0.342 0.300 0.264

0.232 0.204

0.179

87

---

Temperature and pH-Dependent Values of the CCC (Chronic Criterion)

for Fish Early Life Stages Absent

CCC for Fish Early Life Stages Absent, mg N/L

pH

Temperature

0-7

8

9

10

11

12

13

14

15*

16*

6.5

10.8 10.1

9.51

8.92

8.36

7.84 7.35

6.89 6.46

6.06

6.6

10.7

9.99

9.37 8.79

8.24

7.72

7.24 6.79

6.36

5.97

6.7

10.5

9.81

9.20

8.62

8.08

7.58

7.11

6.66

6.25

5.86

6.8

10.2

9.58 8.98 8.42

7.90

7.40

6.94 6.51

6.10

5.72

6.9

9.93

9.31

8.73

8.19

7.68

7.20 6.75

6.33 5.93

5.56

7.0

9.60

9.00

8.43 7.91

7.41

6.95

6.52

-6.11

5.73

5.37

7.1

9.20

8.63

8.09

7.58

7.11

6.67 6.25 5.86.

5.49

5.15

7.2

8.75

8.20

7.69

7.21

6.76

6.34 5.94 5.57

5.22

4.90

7.3

8.24

7.73 7.25 6.79

6.37

5.97

5.60 5.25 4.92

4.61

7.4

7.69

7.21

6.76

6.33

5.94

5.57 5.22

4.89

4.59 4.30

7.5

7.09

6.64

6.23

5.84

5.48

5.13 4.81

4.51

4.23

3.97

7.6

6.46

6.05

5.67 5.32

4.99

4.68 4.38

4.11

3.85

3.61

7:7

5.81

5.45

5.11

4.79

4.49

4.21

3.95 3.70

3.47

3.25

7.8

5.17

4.84 4.54 4.26

3.99

3.74

3.51

3.29 3.09

2.89

7.9

4.54

4.26

3.99

3.74

3.51

3.29 3.09 2.89

2.71

2.54

8.0

3.95

3.70

3.47

3.26

3.05

2.86

2.68 2.52

2.36

2.21

8.1

3.41

3.19

2.99

2.81

2.63

2.47

2.31

2.17

2.03

1.91

8.2

2.91

2.73

2.56

2.40

2.25

2.11

1.98

1.85

1.74

1.63

8.3

2.47

2.32

2.18

2.04

1.91

1.79

1.68

1.58

1.48

1.39

8.4

2.09

1.96

1.84

1.73

1.62

1.52

1.42

1.33

1.25

1.17

8.5

1.77

1.66

1.55

1.46

1.37

1.28

1.20

1.13

1.06

0.990

8.6

1.49

1.40

1.31

1.23

1.15

1.08

1.01

0.951 0.892

0.836

8.7

1.26

1.18

1.11

1.04

0.976 0.915

0.858 0.805 0.754

0.707

8.8

1.07

1.01

0.944 0.885

0.829 0.778 0.729

0.684 0.641

0.601

8.9

0.917 0.860 0.806

0.756 0.709 0.664

0.623

0.584

0.548 0.513

9.0

0.790 0.740 0.694

0.651 0.610

0.572 0.536 0.503

0.471 0.442

* At 15 C and above, the criterion for fish ELS absent is the same as the criterion for fish ELS

present.

88

---

References

Note: Three unpublished manuscripts that were cited in the 1984/1985 criteria

document have been published as Broderius et al. (1985), Erickson (1985), and

Thurston et al. (1986). West (1985) was published as Arthur et al. (1987).

Alexander, H.C., P.B. Latvaitis, and D.L. Hopkins. 1986: Site-Specific Toxicity of Un-

ionized Ammonia in the Tittabawassee River at Midland, Michigan: Overview. Environ.

Toxicol. Chem. 5:427-435.

Anderson, K.B., R.E. Sparks, and A.A. Paparo. 1978. Rapid Assessment of Water

Quality, using the Fingernail Clam, Musculium transversum. WRC Research Report

No. 133, University of Illinois, Water Resources Center, Urbana, IL.

Ankley, G.T., M.K. Schubauer-Berigan, and P.D. Monson. 1995. Influence of pH and

Hardness on Toxicity of Ammonia to the Amphipod

Hyalella azteca.

Can. J. Fish.

Aquat. Sci. 52:2078-2083.

Armstrong, D.A., D. Chippendale, A.W. Knight, and J.E. Colt. 1978. Interaction of

Ionized And Un-ionized Ammonia on Short-term Survival and Growth of Prawn Larvae,

Macrobrachium rosenbergii.

Biol. Bull. 154:15-31.

Arthur, J.W., C.W. West, K.N. Allen, and S.F. Hedtke. 1987. Seasonal Toxicity of

Ammonia to Five Fish and Nine Invertebrate Species. Bull. Environ. Contam. Toxicol.

38:324-331.

ASTM. 1997a. Standard Guide for Conducting Renewal Life-Cycle Toxicity Tests with

Daphnia magna.

Standard E1193 in Vol. 11.05 of the Annual Book of ASTM

Standards. American Society for Testing and Materials, West Conshohocken, PA.

ASTM. 1997b. Standard Guide for Conducting Early Life-Stage Toxicity Tests with

Fishes. Standard El 241 in Vol. 11.05 of the Annual Book of ASTM Standards.

American Society for Testing and Materials, West Conshohocken, PA.

ASTM. 1997c. Standard Guide for Conducting Three-Brood, Renewal Toxicity Tests

with

Ceriodaphnia dubia.

Standard El 295 in Vol. 11.05 of the Annual Book of ASTM

Standards. American Society for Testing and Materials, West Conshohocken, PA.

ASTM. 1997d. Standard Guide for Conducting Acute Toxicity Tests on Test Materials

with Fishes, Macroinvertebrates, and Amphibians. Standard E729 in Vol. 11.05 of the

Annual Book of ASTM Standards. American Society for Testing and Materials, West

Conshohocken, PA.

89

---

ASTM. 1997e. Standard Test Methods for Measuring the Toxicity of Sediment-

Associated Contaminants with Fresh Water Invertebrates. Standard E1706 in Vol.

11.05 of the Annual Book of ASTM Standards. American Society for Testing and

Materials, West Conshohocken, PA.

Bader, J.A. 1990. Growth-Inhibiting Effects and Lethal Concentrations of Un-ionized

Ammonia for Larval and Newly Transformed Juvenile Channel Catfish (Ictalurus

punctatus). M.S. Thesis, Auburn University, Auburn, AL.

Bader, J.A., and J.M. Grizzle. 1992. Effects of Ammonia on Growth and Survival of

Recently Hatched Channel Catfish. J. Aquatic Animal Health 4:17-23.

Bailey, H.C., D.H.W. Liu, and H.A. Javitz. 1985. Time/Toxicity Relationships in Short-

Term Static, Dynamic, and Plug-Flow Bioassays. In: Aquatic Toxicology and Hazard

Assessment: Eighth Symposium. R.C. Bahner and D.J. Hansen, Eds. ASTM STP 891.

American Society for Testing and Materials, Philadelphia, PA. pp. 193-212.

Bergerhouse, D.L. 1992. Lethal Effects of Elevated pH and Ammonia on Early Life

Stages of Walleye. N. Amer. J. Fish. Manage. 12:356-366.

Bergerhouse, D.L. 1993. Lethal Effects of Elevated pH and Ammonia on Early Life

Stages of Hybrid Striped Bass. J. Appl. Aquacult. 2:81-100.

Bianchini, A., W. Wasielesky, Jr., and K.C.M. Filho. 1996. Toxicity of Nitrogenous

Compounds to Juveniles of Flatfish

Paralichthys orbignyanus.

Bull. Environ. Contam.

Toxicol. 56:453-459.

Borgmann, U. 1994. Chronic Toxicity of Ammonia to the Amphipod

Hyalella azteca;

Importance of Ammonium Ion and Water Hardness. Environ. Pollut. 86:329-335.

Borgmann, U., and A.I. Borgmann. 1997. Control of Ammonia Toxicity to

Hyalella

azteca

by Sodium, Potassium and pH. Environ. Pollut. 95:325-331.

Box, G.E.P., and N.R. Draper. 1965. The Bayesian Estimation of Common Parameters

from Several Responses. Biometrika 52:355-365.

Box, G.E.P., W.G. Hunter, J.F. MacGregor, and J. Erjavec. 1973. Some Problems

Associated with the Analysis of Multiresponse Data. Technometrics 15:33-51.

Broderius, S.J., and L.L. Smith, Jr. 1979. Lethal and Sublethal Effects of Binary

Mixtures of Cyanide and Hexavalent Chromium, Zinc, or Ammonia to the Fathead

Minnow

(Pimephales promelas)

and Rainbow Trout

(Salmo gairdneri).

J. Fish. Res. Bd.

Can. 36:164-172.

90

---

Broderius, S., R. Drummond, J. Fiandt, and C. Russom. 1985. Toxicity of Ammonia to

Early Life Stages of the Smalimouth Bass at Four pH Values. Environ. Toxicol. Chem.

4:87-96.

Buckley, J.A., C.M. Whitmore, and B.D. Liming. 1979.. Effects of Prolonged Exposure

to Ammonia on the Blood and Liver Glycogen of Coho Salmon

(Oncorhynchus kisutch).

Comp. Biochem. Physiol. 63C:297-303.

Burkhalter, D.E. 1975. Effects of Prolonged Exposure to Ammonia on Rainbow Trout

(Salmo gairdnen)

Eggs and Sac Fry. M.S. Thesis, Montana State University, Boseman,

MT. 67 pp.

Burkhalter, D.E., and C.M. Kaya. 1977. Effects of Prolonged Exposure to Ammonia on

Fertilized Eggs and Sac Fry of Rainbow Trout

(Salmo gairdneri).

Trans. Amer. Fish.

Soc. 106:470-475.

Burrows, R.E. 1964. Effects of Accumulated Excretory Products on Hatchery-Reared

Salmonids. Research Report 66, U.S. Fish and Wildlife Service, Washington, DC.

Calamari, D., R. Marchetti, and G. Vailati. 1977. Effects of Prolonged Treatments with

Ammonia on Stages of Development of

Salmo gairdneri.

(English translation used.)

Nuovi Ann. lg. Microbiol. 28:333-345.

Calamari, D., R. Marchetti, and G. Vailati. 1981. Effects of Long-term Exposure to

Ammonia on the Developmental Stages of Rainbow Trout

(Salmo gairdneri

Richardson). Rapp. P.-v. Reun. Cons. int. Explor. Mer. 178:81-86.

Camp Dresser and McKee. 1997. Site-Specific Limits for Un-ionized Ammonia - Red

River of the North. Report to the City of Moorhead, MN.

Carline, R.F., A.J. Benson, and H. Rothenbacher. 1987. Long-term Effects of Treated

Domestic Wastewater on Brown Trout. Wat. Res. 21:1409-1415.

Cary, G.A. 1976. A Report on the Assessment of Aquatic Environmental Impact of

Union Carbide's Uravan Operations. Aquatic Environmental Sciences, Tarrytown, NY.

67 pp.

Chipman, W.A., Jr. 1934. The Role of pH in Determining the Toxicity of Ammonium

Compounds. Ph.D. Thesis. University of Missouri, Columbia, MO. 153 pp.

Colt, J.E. 1978. The Effects of Ammonia on the Growth of Channel Catfish,

Ictalurus

punctatus.

Ph.D. Thesis, University of California, Davis, CA. 185 pp.

Colt, J., and G. Tchobanoglous. 1976. Evaluation of the Short-term Toxicity of

Nitrogenous Compounds to Channel Catfish,

Ictalurus punctatus.

Aquaculture 8:209-

224.

91

---

Colt, J., and G. Tchobanoglous. 1978. Chronic Exposure of Channel Catfish,

lctalurus

punctatus,

to Ammonia: Effects on Growth and Survival. Aquaculture 15:353-372.

Dabrowska, H., and H. Sikora. 1986. Acute Toxicity of Ammonia to Common Carp

(Cyprinus carpio

L.) Pol. Arch. Hydrobiol. 33:121-128.

Dahlquist, G., and A. Bjorck. 1974. Numerical Methods. Translated by N. Anderson.

Prentice-Hall, Englewood Cliffs, NJ. 573 pp.

Daoust, P.-Y., and H.W. Ferguson. 1984. The Pathology of Chronic Ammonia Toxicity

in Rainbow Trout,

Salmo gairdneri

Richardson. J. Fish Dis. 7:199-205.

DeGraeve, G.M., W.D. Palmer, E.L. Moore, J.J. Coyle, and P.L. Markham. 1987. The

Effect of Temperature on the Acute and Chronic Toxicity of Un-ionized Ammonia to

Fathead Minnows and Channel Catfish. Final Report to U.S. EPA by Battelle,

Columbus, OH. 39 pp.

Delos, C. 1999. Design Flow for a 30-Day Average CCC. Memorandum, September

1, 1999. U.S. EPA.

Diamond, J.M., D.G. Mackler, W.J. Rasnake, and D. Gruber. 1993. Derivation of Site-

Specific Ammonia Criteria for an Effluent-Dominated Headwater Stream. Environ.

Toxicol. Chem. 12:649-658.

Draper, N.R., and H. Smith. 1981. Applied Regression Analysis. Second Edition.

Wiley, New York. 709 pp.

Emerson, K., R.C. Russo, R.E. Lund, and R.V. Thurston. 1975. Aqueous Ammonia

Equilibrium Calculations: Effect of pH and Temperature. J. Fish. Res. Bd. Can.

32:2379-2383.

Erickson, R.J. 1985. An Evaluation of Mathematical Models for the Effects of pH and

Temperature on Ammonia Toxicity to Aquatic Organisms. Water Res. 19:1047-1058.

Fromm, P.O. 1970. Toxic Action of Water Soluble Pollutants on Freshwater Fish.

EPA Water Pollution Control Research Series. 18050 DST 12/70. PB 201650,

National Technical Information Service, Springfield, VA. 56 pp.

Gersich, F.M., and D.L. Hopkins. 1986. Site-Specific Acute and Chronic Toxicity of

Ammonia to

Daphnia magna

Straus. Environ. Toxicol. Chem. 5:443-447.

Gersich, F.M., D.L. Hopkins, S.L. Applegath, C.G. Mendoza, and D.P. Milazzo. 1985.

The Sensitivity of Chronic Endpoints Used in

Daphnia magna

Straus Life-Cycle Tests.

In: Aquatic Toxicology and Hazard Assessment: Eighth Symposium. R.C. Bahner and

D.J. Hansen, Eds. ASTM STP 891. American Society for Testing and. Materials,

Philadelphia, PA. pp. 245-252.

92

---

Goudreau, S.E., R.J. Neves, and R.J. Sheehan. 1993. Effects of Wastewater

Treatment Plant Effluents on Freshwater Molluscs in the Upper Clinch River, Virginia,

USA. Hydrobiologia 252:211-230.

Gulyas, P., and E. Fleit. 1990. Evaluation of Ammonia Toxicity on

Daphnia magna

and

Some Fish Species. Aquacultura Hungarica (Szarvas) 6:171-183.

Hasan, M.R., and D.J. Macintosh. 1986. Acute Toxicity of Ammonia to Common Carp

Fry. Aquaculture 54:97-107.

Hazel, C.R., W. Thomsen, and S.J. Meith. 1971. Sensitivity of Striped Bass and

Stickleback to Ammonia in Relation to Temperature and Salinity. Calif. Fish Game

57:138-153.

Heber, M., and K. Ballentine. 1992. Memorandum to Water Quality Standards

Coordinators. July 30. U.S. EPA.

Henderson, C., Q.H. Pickering, and A.E. Lemke. 1961. The Effect of Some Organic

Cyanides (Nitriles) on Fish. Industrial Waste Conf. 15:120-130.

Hermanutz, R.O., S.F. Hedtke, J.W. Arthur, R.W. Andrew, and K.N. Allen. 1987.

Ammonia Effects on Microinvertebrates and Fish in Outdoor Experimental Streams.

Environ. Pollut. 47:249-283.

Iwama, G.K., J.C. McGeer, P.A. Wright, M.P. Wilkie, and C.M. Wood. 1997. Divalent

Cations Enhance Ammonia Excretion in Lahontan Cutthroat Trout in Highly Alkaline

Water. J. Fish Biol. 50:1061-1073.

Johnson, C.G. 1995. Effects of pH and Hardness on Acute and Chronic Toxicity of

Un-ionized Ammonia to

Ceriodaphnia dubia.

M.S. Thesis. University of Wisconsin,

Stevens Point, WI. 69 pp.

Jude, D.J. 1973. Sublethal Effects of Ammonia and Cadmium on Growth of Green

Sunfish. Ph.D. Thesis. Michigan State University, East Lansing, MI. 193 pp.

Knoph, M.B. 1992. Acute Toxicity of Ammonia To Atlantic Salmon

(Salmo salar)

Parr.

101 C:275-282.

Lee, D.R. 1976. Development of an Invertebrate Bioassay to Screen Petroleum

Refinery Effluents Discharged into Freshwater. Thesis. Virginia Polytechnic Institute

and State University. 105 pp.

Lemly, A.D. 1996. Wastewater Discharges may be most Hazardous to Fish during

Winter. Environ. Pollut. 93:169-174.

93

---

Lloyd, R. 1980. Toxicity Testing with Aquatic Organisms: A Framework for Hazard

Assessment and Pollution Control. In: Biological Effects of Marine Pollution and the

Problems of Monitoring. A.D. McIntyre and J.B. Pearce, Eds. Rapp. P.-v. Reun. Cons.

int. Explor. Mer. 179:339-341.

Lloyd, R., and D.W.M. Herbert. 1960. The Influence of Carbon Dioxide on the Toxicity

of Un-ionized Ammonia to Rainbow Trout

(Salmo gairdnerii

Richardson). Ann. Appl.

Biol. 48:399-404.

Macek, K.J., and B.H. Sleight, III. 1977. Utility of Toxicity Tests with Embryos and Fry

of Fish in Evaluating Hazards Associated with the Chronic Toxicity of Chemicals to

Fishes. In: Aquatic Toxicology and Hazard Evaluation. F.L. Mayer and J.L. Hamelink,

Eds. ASTM STP 634. American Society for Testing and Materials, Philadelphia, PA.

pp. 137-146.

Malins, D.C. 1982. Alterations in the Cellular and Subcellular Structure of Marine

Teleosts and Invertebrates Exposed to Petroleum in the Laboratory and Field: A Critical

Review. Can. J. Fish. Aquat. Sci. 39:877-889.

Marchetti, R. 1960. New Studies on the Toxicology of Fish with Respect to Control of

Waste Waters. (English translation used.) Ann. Sta. Centr. Hydrobiol. Appl. 8:107-

124 .

Mayes, M.A., H.C. Alexander, D.L. Hopkins, and P.B. Latvaitis. 1986. Acute and

Chronic Toxicity of Ammonia to Freshwater Fish: A Site-Specific Study. Environ.

Toxicol. Chem. 5:437-442.

McCormick, J.H., S.J. Broderius, and J.T. Fiandt. 1984. Toxicity of Ammonia to Early

Life Stages of the Green Sunfish

Lepomis cyanellus.

(with erratum) Environ. Pollut.

(Series A) 36:147-163.

McKim, J.M. 1977. Evaluation of Tests with Early Life Stages of Fish for Predicting

Long-Term Toxicity. J. Fish. Res. Board Can. 34:1148-1154.

Ministry of Technology. 1967. Effects of Pollution on Fish. In: Water Pollution

Research 1967. H.M. Stationery Office, London. pp. 56-65.

Mitchell, S.J., and J.J. Cech, Jr. 1983. Ammonia-Caused Gill Damage in Channel

Catfish

(lctalurus punctatus):

Confounding Effects of Residual Chlorine. Can. J. Fish.

Aquat. Sci. 40:242-247.

Monda, D.P., D.L. Galat, S.E. Finger, and M.S. Kaiser. 1995. Acute Toxicity of

Ammonia (NH 3-N)

in Sewage Effluent to

Chironomus riparius:

II. Using a Generalized

Linear Model. Arch. Environ. Contam. Toxicol. 28:385-390.

94

---

Moore, D.R.J., and P.-Y. Caux. 1997. Estimating Low Toxic Effects. Environ. Toxicol.

Chem. 16:794-801.

Mount, D.I. 1982. Memorandum to R.C. Russo. August 6.

Mount, D.R., D.D. Gulley, J.R. Hockett, T.D. Garrison, and J.M. Evans. 1997.

Statistical Models to Predict the Toxicity of Major Ions to

Ceriodaphnia dubia, Daphnia

magna

and

Pimephales promelas

(Fathead Minnows). Environ. Toxicol. Chem.

16:2009-2019.

Nimmo, D.W.R., D. Link, L.P. Parrish, G.J. Rodriguez, W. Wuerthele, and P.H. Davies.

1989. Comparison of On-Site and Laboratory Toxicity Tests: Derivation of Site-Specific

Criteria for Un-ionized Ammonia in a Colorado Transitional Stream. Environ. Toxicol.

Chem. 8:1177-1189.

Rankin, D.P. 1979. The Influence of Un-ionized Ammonia on the Long-term Survival

of Sockeye Salmon Eggs. Technical Report No. 912, Fisheries and Marine Service,

Nanaimo, British Columbia, Canada. 17 pp.

Reichenbach-Klinke, H.-H. 1967. Investigations on the Influence of the Ammonia

Content on the Fish Organism. (English translation used.) Archiv fur

Fischereiwissenschaft 17:122-132.

Reinbold, K.A., and S.M. Pescitelli. 1982a. Effects of Exposure to Ammonia on

Sensitive Life Stages of Aquatic Organisms. Report to U.S. EPA by Illinois Natural

History Survey, Champaign, IL.

Reinbold, K.A., and S.M. Pescitelli. 1982b. Acute Toxicity of Ammonia to the White

Sucker. Report to U.S. EPA by the Illinois Natural History Survey, Champaign, IL.

Reinbold, K.A., and S.M. Pescitelli. 1982c. Acute Toxicity of Ammonia to Channel

Catfish. Report to U.S. EPA by the Illinois Natural History Survey, Champaign, IL.

Rice, S.D., and J.E. Bailey. 1980. Survival, Size, and Emergence of Pink Salmon,

Oncorhynchus gorbuscha,

Alevins after Short- and Long-term Exposures to Ammonia.

Fish. Bull. 78:641-648.

Robinette, H.R. 1976. Effect of Selected Sublethal Levels of Ammonia on the Growth

of Channel Catfish

(lctalurus punctatus).

Prog. Fish-Cult. 38:26-29.

Robinson-Wilson, E.F., and W.K. Seim. 1975. The Lethal and Sublethal Effects of a

Zirconium Process Effluent on

.

Juvenile Salmonids. Water Resour. Bull. 11:975-986.

Roseboom, D.P., and D.L. Richey. 1977. Acute Toxicity of Residual Chlorine and

Ammonia to Some Native Illinois Fishes. Report of Investigation 85, Illinois State

Water Survey, Urbana, IL. 42 pp.

95

---

Russo, R.C., D.J. Randall, and R.V. Thurston. 1988. Ammonia Toxicity and

Metabolism in Fishes. In: Protection of River Basins, Lakes, and Estuaries. R.C.

Ryans, Ed. American Fisheries Society, Bethesda, MD. pp. 159-173.

Samylin, A.F. 1969. Effect of Ammonium Carbonate on the Early Stages of

Development of Salmon. (English translation used.) Uch. Zap. Leningr. Gos.

Pedagog. Inst. Im. A.I. Gertsena 422:47-62.

Schubauer-Berigan, M.K., P.D. Monson, C.W. West, and G.T. Ankley. 1995. Influence

of pH on the Toxicity of Ammonia to

Chironomus tentans

and

Lumbriculus variegatus.

Environ. Toxicol. Chem. 14:713-717.

Schulze-Wiehenbrauck, H. 1976. Effects of Sublethal Ammonia Concentrations on

Metabolism in Juvenile Rainbow Trout

(Salmo gairdneri

Richardson). Ber. dt. wiss.

Kommn. Meeresforsch. 24:234-250.

Sheehan, R.J., and W.M. Lewis. 1986. Influence of pH and Ammonia Salts on

Ammonia Toxicity and Water Balance in Young Channel Catfish. Trans. Amer. Fish.

Soc. 115:891-899.

Smart, G. 1976. The Effect of Ammonia Exposure on Gill Structure of the Rainbow

Trout

(Salmo gairdneri). J.

Fish Biol. 8:471-475.

Smith, C.E. 1972. Effects of Metabolic Products on the Quality of Rainbow Trout.

Amer. Fish. Trout News. 17:7,8,21,22.

Smith, C.E. 1984. Hyperplastic Lesions of the Primitive Meninx of Fathead Minnows,

Pimephales promelas,

Induced by Ammonia: Species Potential for Carcinogen Testing.

Natl. Cancer Inst. Monogr. 65:119-125.

Smith, C.E., and R.G. Piper. 1975. Lesions Associated with Chronic Exposure to

Ammonia. In: The Pathology of Fishes. W.E. Ribelin and G. Migaki, Eds. U. Wisc.

Press, Madison, WI. pp. 497-514.

Smith, WE., T.H. Roush, and J.T. Fiandt. 1984. Toxicity of Ammonia to Early Life

Stages of Bluegill

(Lepomis macrochirus).

EPA-600/x-84-175. In-house report, U.S.

EPA, Duluth, MN.

Snell, T.W., and G. Persoone. 1989. Acute Toxicity Bioassays Using Rotifers. II. A

Freshwater Test with

Brachionus rubens.

Aquatic Toxicol. 14:81-91.

Soderberg, R.W., and J.W. Meade. 1991. The Effects of Ionic Strength on Un-ionized

Ammonia Concentration. Prog. Fish-Cult. 53:118-120.

96

---

Soderberg, R.W., and J.W. Meade. 1992. Effects of Sodium and Calcium on Acute

Toxicity of Un-ionized Ammonia to Atlantic Salmon and Lake Trout. J. Appl.

Aquaculture 1:83-92.

Soderberg, R.W., J.B. Flynn, and H.R. Schmittou. 1983. Effects of Ammonia on

Growth and Survival of Rainbow Trout in Intensive Static-Water Culture. Trans. Amer.

Fish. Soc. 112:448-451.

Soderberg, R.W., M.V. McGee, and C.E. Boyd. 1984. Histology of Cultured Channel

Catfish,

lctalurus punctatus

(Rafinesque). J. Fish Biol. 24: 683-690.

Solbe, J.F.L.G., and D.G. Shurben. 1989. Toxicity of Ammonia to Early Life Stages of

Rainbow Trout

(Salmo gairdneri).

Water Res. 23:127-129.

Sousa, R.J., T.L. Meade, and R.E. Wolke. 1974. Reduction of Ammonia Toxicity by

Salinity and pH Manipulation. IN: Proc. Fifth Annual Workshop, World Mariculture

Society 5:343-354.

Sparks, R.E., and M.J. Sandusky. 1981. Identification of Factors Responsible for

Decreased Production of Fish Food Organisms in the Illinois and Mississippi Rivers.

Final Report for Project No. 3-291-R, Illinois Natural History Survey, River Research

Laboratory, Havana, IL. 63 pp.

Stephan, C.E., and J.W. Rogers. 1985. Advantages of Using Regression Analysis to

Calculate Results of Chronic Toxicity Tests. In: Aquatic Toxicology and Hazard

Assessment: Eighth Symposium. R.C. Bahner and D.J. Hansen, Eds. ASTM STP 891.

American Society for Testing and Materials, Philadelphia, PA. pp. 328-338.

Stevenson, T.J. 1977. The Effects of Ammonia, pH and Salinity on the White Perch,

Morone americana. Ph.D. Thesis, University of Rhode Island, Kingston, RI. 154 pp.

Swigert, J.P. and A. Spacie. 1983. Survival and Growth of Warmwater Fishes

Exposed to Ammonia Under Low Flow Conditions. PB83-257535. National Technical

Information Service, Springfield, VA.

Szumski, D.S., D.A. Barton, H.D. Putnam, and R.C. Polta. 1982. Evaluation of EPA

Un-ionized Ammonia Toxicity Criteria. J. Water Pollut. Control Fed. 54:281-291.

Tabata, K. 1962. Toxicity of Ammonia to Aquatic Animals with Reference to the Effect

of pH and Carbonic Acid. (English translation used.) Tokai-ku Suisan Kenkyusho

Kenkyu Hokoku 34:67-74.

Thomas, P.C., C. Turner, and D. Pascoe. 1991. An Assessment of Field and

Laboratory Methods for Evaluating the Toxicity of Ammonia to

Gammarus pulex (L.) -

97

---

Effects of Water Velocity. In: Bioindicators and Environmental Management: Sixth

Symposium. D.W. Jeffrey and B. Madden, Eds. Academic Press, London. pp. 353-

363.

Thurston, R.V., and R.C. Russo. 1983. Acute Toxicity of Ammonia to Rainbow Trout.

Trans. Amer. Fish. Soc. 112:696-704.

Thurston, R.V., R.C. Russo, and C.E. Smith. 1978. Acute Toxicity of Ammonia and

Nitrite to Cutthroat Trout Fry. Trans. Amer. Fish. Soc. 107:361-368.

Thurston, R.V., C. Chakoumakos, and R.C. Russo. 1981a. Effect of Fluctuating

Exposures on the Acute Toxicity of Ammonia to Rainbow Trout

(Salmo gairdnen)

and

Cutthroat Trout

(S. clarki).

Water Res. 15:911-917.

Thurston, R.V., R.C. Russo, and G.A. Vinogradov. 1981 b. Ammonia Toxicity to

Fishes. Effect of pH on the Toxicity of the Un-ionized Ammonia Species. Environ. Sci.

Technol. 15:837-840.

Thurston, R.V., R.C. Russo, and G.R. Phillips. 1983. Acute Toxicity of Ammonia to

Fathead Minnows. Trans. Amer. Fish. Soc. 112:705-711.

Thurston, R.V., R.J. Luedtke, and R.C. Russo. 1984a. Toxicity of Ammonia to

Freshwater Insects of Three Families. Technical Report No. 84-2, Fisheries Bioassay

Laboratory, Montana State University, Bozeman, MT. 26 pp.

Thurston, R.V., R.C. Russo, R.J. Luedtke, C.E. Smith, E.L. Meyn, C. Chakoumakos,

K.C. Wang, and C.J.D. Brown. 1984b. Chronic Toxicity of Ammonia to Rainbow Trout.

Trans. Amer. Fish. Soc. 113:56-73.

Thurston, R.V., R.C. Russo, E.L. Meyn, R.K. Zajdel, and C.E. Smith. 1986. Chronic

Toxicity of Ammonia to Fathead Minnows. Trans. Amer. Fish. Soc. 115:196-207.

Tomasso, J.R., and G.J. Carmichael. 1986. Acute Toxicity of Ammonia, Nitrite, and

Nitrate to the Guadalupe Bass,

Micropterus treculi.

Bull. Environ. Contam. Toxicol.

36:866-870.

Tomasso, J.R., C.A. Goudie, B.A. Simco, and K.B. Davis. 1980. Effects of

Environmental pH and Calcium on Ammonia Toxicity in Channel Catfish. Trans. Amer.

Fish. Soc. 109:229-234.

U.S. EPA. 1985a. Ambient Water Quality Criteria for Ammonia - 1984. EPA-44015-85-

001. National Technical Information Service, Springfield, VA.

U.S. EPA. 1985b. Guidelines for Deriving Numerical National Water Quality Criteria

for the Protection of Aquatic Organisms and Their Uses. PB85-227049. National

Technical Information Service, Springfield, VA.

98

---

U.S. EPA. 1986. Ambient Water Quality Criteria for Dissolved Oxygen. EPA 440/5-

86-003. National Technical Information Service, Springfield, VA.

U.S. EPA. 1989. Ambient Water Quality Criteria for Ammonia (Saltwater) - 1989. EPA

440/5-88-004. National Technical Information Service, Springfield, VA.

U.S. EPA. 1991. Technical Support Document for Water Quality-based Toxics

Control. EPA/505/2-90-001. National Technical Information Service, Springfield, VA.

U.S. EPA. 1994. Interim Guidance on Determination and Use of Water-Effect Ratios

for Metals. EPA-823-B-94-001 or PB94-140951. National Technical Information

Service, Springfield, VA.

U.S. EPA. 1996. "Ammonia Criteria" in the Water Quality Criteria and Standards

Newsletter for January. EPA-823-N-96-001. Office of Water, Washington, DC. p. 5.

U.S. EPA. 1998. 1998 Update of Ambient Water Quality Criteria for Ammonia. EPA

822-R-98-008. Office of Water, Washington, DC.

U.S. EPA. 1999. Response to Comments on 1998 Update of Ambient Water Quality

Criterion for Ammonia. Office of Water, Washington, DC.

Wade, D.C. 1992. Definitive Evaluation of Wheeler Reservoir Sediments Toxicity

using Juvenile Freshwater Mussels (Anodonta imbecillis Say). TVA/WR--92/25.

Tennessee Valley Authority.

West, C.W. 1985. Acute Toxicity of Ammonia to 14 Freshwater Species. Internal

Report. U.S. EPA, Environmental Research Laboratory, Duluth, MN.

Williams, K.A., D.W.J. Green, and D. Pascoe. 1986. Studies on the Acute Toxicity of

Pollutants to Freshwater Macroinvertebrates. 3. Ammonia. Arch. Hydrobiol. 106:61-70.

Willingham, T. 1987. Acute and Short-term Chronic Ammonia Toxicity to Fathead

Minnows (Pimephales promelas) and Ceriodaphnia dubia Using Laboratory Dilution

Water and Lake Mead Dilution Water. U.S. EPA, Denver, CO. 40 pp.

Willingham, T. 1996. Letter to C. Stephan. November 8.