minute adjustments in pH were made as appropriate. Mixing of the test water 

 and additives was virtually instantaneous, ensuring uniform water chemistry 

 conditions throughout any one tank. This was confirmed by repeated sampling 

 studies. 



The average size of rainbow trout was 9-11 g, all fish from the same 

 stock, and that for the fathead minnows was 1.8-2.0 g, again from a single 

 stock. Tests were conducted on successive weeks, three at a time: one 

 acid, one base, and one at the normal pH of the test water. The normal pH 

 test was repeated each time the acid and base tests were conducted; com- 

 parable results from the normal pH tests verified that test conditions from 

 week to week were comparable, and that the test fish stock had not changed 

 appreciably over time. 



The results of the tests on rainbow trout are illustrated in Figure 6. 

 Ninety-six hour LC50 values and their confidence limits in terms of both 

 total ammonia-nitrogen and un-ionized ammonia-nitrogen are plotted for each 

 pH test. A log scale for the LC50 values has been used so that visual com- 

 parison of total ammonia and NH3 values can easily be made. The excellent 

 reproducibility of the tests run at normal test water pH is apparent. If 

 the un-ionized form of ammonia (NH3) were solely responsible for the toxic 

 action on the test fish, then one would expect that the LC50 values, in 

 terms of NH3, would be reasonably constant for all tests regardless of the 

 solution pH and total ammonia present. This did not turn out to be the 

 case. Figure 7 illustrates the results of the tests on fathead minnows. 

 The LC50 values, in terms of both total ammonia-nitrogen and un-ionized 

 ammonia-nitrogen, are higher than those for rainbow trout because the fat- 

 head minnow is a more ammonia- tolerant fish, but the LC50 vs. pH trend is 

 the same. 



Our findings provide support for the conclusions of Tabata (1962) and 

 Armstrong et ^. (1978), and are in conflict with the more widely accepted 

 notion that the toxicity of NH3 is independent of pH. The LC50 values in 

 terms of NH3 for our 96-hour acute toxicity tests on rainbow trout are 

 strikingly similar to those reported by Robinson-Wilson and Seim (1975) for 

 coho salmon within the pH range 7.0 to 8.5. These authors explain the cor- 

 relation of solution pH with NH3 LC50 values to be related to changes in the 

 CO2 concentration, hence pH, at the surface of the fish gill tissue. Our 

 conclusion at this time is that the NH4''' ion exerts a heretofore not fully 

 recognized toxic effect on fishes, and/or that the toxicity of NH3 increases 

 as the H+ ion concentration increases. 



Regardless of the explanation for it, the correlation between LC50 in 

 terms of NH3 and pH has been demonstrated, and the rationale for water 

 quality criteria for ammonia needs to address this. 



CONCLUSION 



I have discussed briefly just four factors affecting the toxicity of am- 

 monia. I have used these as examples of how the many chemical and physical 

 parameters involved in aqueous systems are interrelated in affecting the 



131 



