28 VICTOR E. SHELFORD. 



habits are known is correlated icith the distribution of the conditions 

 necessary for breeding. 



While our tables show that there is considerable bare bottom 

 in the pond $c, there is good evidence that this is largely due to 

 building of the road and of the Lake Shore and Mich. Southern 

 R. R. which separated this pond from the others and from the 

 lake and probably excluded fish since 1851. The exposures of 

 bare sandy bottom which are due to natural causes are usually 

 not covered with more than six inches ol water. 



Turning to the perch which is abundant here we note that the 

 eggs are extruded in the open water or vegetation as well as over 

 terrigenous bottom. Terrigenous bottom is less necessary than 

 to the other food fishes. 



Turning to the spotted bullhead we note that the nests are 

 probably usually made in water shallower than any of the other 

 fishes. Only one specimen has been taken from pond I.; they 

 are numerous in pond $c and ja. There are some old logs and 

 stumps and a very narrow zone of bare sand in o in. and less of 

 water in these ponds. This is commonly shaded by vegetation. 



In connection with oxygen content we note that it is greatest 

 in 5c where the first four species of Table XXV. do not breed. 

 However, this pond must be regarded as in a measure non- 

 comparable because of contamination and small amount of plank- 

 ton. 



The low oxygen content on the muck bottoms of the older 

 ponds, at depths used by the fishes present in pond I., and absent 

 from these older ones, certainly is a sufficient reason for their 

 absence, though it is not to be expected that this is the sole 

 cause. It is apparent also that A. nebulosits, which is present 

 in the older ponds, not only breeds in shallower water but also 

 has superior means of aerating the eggs (Smith and Harron, '02). 



Succession of fish then becomes succession of breeding condi- 

 tions and breeding mores. While the major factors as indicated 

 here are related to deptli and bottom, there are doubtless others. 



IV. GENERAL DISCUSSION. 



There is great danger of error in dealing with such complex 

 problems when compilation is necessary and especially when the 



