finitely a correlation for the rainbow trout tests, but we cannot draw the 

 same conclusion for the fathead minnow tests. 



In an attempt to study the reduced dissolved oxygen effect on ammonia 

 toxicity in relation to time, we analyzed our data for the rainbow trout 

 tests, comparing the dissolved oxygen vs. LC50 correlations for the tests at 

 12, 24, 48, 72, and 96 hours. This showed a very clear and statistically 

 defensible trend (Figure 5); the shorter the time period, the more pro- 

 nounced the correlation. This trend suggests at least two possibilities: 

 either individual fish which require higher oxygen concentrations succumb 

 early in the tests, and/or those fish which do survive become increasingly 

 acclimated to the ammonia and oxygen test conditions as time progresses. 



The EPA Red Book (U.S. EPA 1977) has recommended a minimum concentration 

 of 5.0 mg/liter dissolved oxygen to maintain good freshwater fish popula- 

 tions. At that dissolved oxygen concentration the regression line for the 

 rainbow trout tests reported above indicates a 96-hour LC50 of 0.5 mg/liter 

 NH3 (Figure 4). At dissolved oxygen concentrations of 8.0 mg/liter and 

 above, more common to natural cold-water fish habitats, the test results re- 

 gression line indicates 96-hour LC50's in excess of 0.7 mg/liter NH3. For 

 this particular stock of test fish, tested under the given bioassay condi- 

 tions, there was a 30 percent decrease in the medium lethal concentration of 

 ammonia when the dissolved oxygen concentration dropped from 8 to 5 

 mg/liter. If this ammonia LC50/dissolved oxygen correlation bears up under 

 further testing using this and other species, the need for reconsideration 

 of both ammonia and dissolved oxygen criteria is clear. 



EFFECT OF pH 



A premise of both the EIFAC (1970) and the U.S. EPA (1977) criteria for 

 ammonia is that NH4'^ is not appreciably toxic to aquatic life. The empiri- 

 cal basis for this was mentioned earlier, and has been explained by the 

 ability of NH3 to diffuse across the gill membrane whereas NH4+ requires 

 active transport. The research by Tabata (1962), Robinson-Wilson and Seim 

 (1975) and Armstrong et al_. (1978), however, raises questions about the 

 criteria premise. 



We have conducted two series of bioassays to investigate the toxicity of 

 ammonia under different pH conditions. The fishes tested were rainbow 

 trout and fathead minnows, and the pH range was 6.5 to 9.0. We chose this 

 pH range because its limits are those recommended by the U.S. EPA (1977) as 

 being the limits acceptable to freshwater fishes. We treated the data from 

 each test by the trimmed Spearman-Karber method (Hamilton et aj_. 1977) to 

 determine both the total ammonia and the un-ionized ammonia 96-hour LC50 

 values. Again, for each bioassay there were five test tanks at different 

 ammonia concentrations and one control tank; eac*" tank contained 10 test 

 fish. The pH of the water in all tanks for any one test was uniform; this 

 was achieved by adjusting the normal pH (7.8) of the test water either up 

 by means of a metered sodium hydroxide solution, or down using a solution of 

 hydrochloric acid. During any given test, the ammonia concentration, pH, 

 and temperature in each test tank were monitored between 5 and 8 times, and 



129 



