71 



lu other words, one-celled organisms and Kntomostraca are the natural, 

 and practically the only, food of an undillerentiatecl small fish; and to be 

 at liberty to grow, t he fish must either change its food (as is usually done) 

 or must develop a special ajjparatus (commonly a set of fine long gill-rakers) 

 for the separation of Entomostraca from the waters in which they swim. 



Of the tlshes which emerge from this earliest stage, through increase in 

 size with failure to develop alimentary structures especially fitted to the 

 appropriation of minute animal forms, some become mud-eaters, like Cam- 

 postoma and the gizzard shad: a few apparently l)ecome vegetarians at 

 once; but most i)ass into or tlirough an insectivorous stage. After this a 

 few become nearly onuiivorous. like the bullheads; others learn to depend 

 chiefly on moUuscan food — the sheepshead and the redhorse species— but 

 many become essentially carnivorous. In fact, unless the gars are an ex- 

 ception, as they now seem to be (attacking young fishes almost as soon as 

 they can swallow), all our specially carnivorous fishes make a progress of 

 three steps, marked respectively, by the predominance of Entamostraca, of 

 insects, and of tlshes, in their food : and the same is true of those strictly 

 fitted for a molluscan diet. 



While small tishes of all sorts are evidently competitors for food, lliis 

 competition is relieved to some extent by differences of breeding season, 

 the species (lro])ping in successively to the banquet, some commencing in 

 very early si)ring. or even, like the whitefish, depositing their eggs in fall, 

 that their young may be first at the board, while others delay luitil .lune 

 or July. The most active breeding period coincides, however, with that of 

 the greatest evolution of Entomostraca in the backwaters of our streams; 

 that is, the early spring. 



That large adult tishes, with tine and numerous rakers on the gills — 

 like the shovelfish and the river carp— may compete directly with the young 

 of all other species, and tend to keep their numbers down by diminishing 

 their food supply— especially in times of scarcity — is very probable, but is 

 not certainly true; for these larger fishes have other food resources also, 

 and may resort to Entomostraca only when these are superabundant, thus 

 appropriating the mere excess aljove what are required for the young of 

 other groups. 



On the Definitexess and Permanency of the Food Habits 



OF Fishes, 



It is always possible that the seemingly specific differences of food ex- 

 hibited by data derived from miscellaneous collections not strictly com- 

 parable as to dates and localities, are really due to differences of circum- 

 stance affecting the representatives of the species, and not to differences 

 in the food habits or the regimen of tlie species in general. Date, locality. 

 and other circumstantial conditions, may have more to do with the dis- 

 tinctions of food detected than structure and specific habit. It is true 

 that the probability of such errors of inference is reduced to a mininuim 

 where alimentary peculiarities can be clearly correlated with peculiarities 

 of structure, as has usually Ijeen done in my discussions: but to test still 

 further the distinctness of species and genera willi respect to food habits 

 and i)references. I have assorted my ol)servat ions according to dates and 

 localities of tlie collections on which they were made, and have compared 

 species with species as occurring under the same general conditions and 

 at the same time. If perch and cat tishes caught in the same haul of the 

 seine show more marked difterences in food between the two groups than 

 those exhibited l»y the individuals of each group among fliemselves, the 

 probability is considerable Ihat the differences are s])eciHc instead of acci- 

 dental : and such probability becomes greater the greater tlie number of 

 species found to present corresiioiiding differences under corresponding 

 circumstances. Altbougli it was rarely tlie case tlial examples enough of 



