No. 637] 



PARASITISM 



155 



ment, we might expect to find that egg-parasites affecting 

 the organism at an earlier and less highly differentiated 

 stage of ontogeny, are more catholic in their tastes. This 

 is, however, not borne out by observation to any extent, 

 and egg parasites are usually as closely restricted to 

 particular hosts as their relatives who confine their atten- 

 tion to larval insects. 



Small size is a prerequisite of all true, internal egg- 

 parasites except a few that occur in the oothecae of cock- 

 roaches, where the comparatively large species of Evania 

 undergo their development. Some parasites oviposit in 

 the host-egg, but live at the expense of the larva; they 

 are, except in polyembrj'onic forms, larger, and not 

 classed as egg-parasites. 



On the basis of size, then, practically all egg 'parasites 

 are either Chalcidoidea or Serphoidea and this habit 

 characterizes a number of families, and smaller taxo- 

 nomic groups {cf. Girault, '07, '11). Among them the 

 strange, tropicopolitan, genus Podagrion attacks only the 

 eggs of Mantidae. The large cosmopolitan genus Tele- 

 nomus occurs in the eggs of various insects, mainly Lepi- 

 doptera while the very similar genera Phanurus and 

 Trissolciis are restricted to eggs of Tabanidas and Penta- 

 tomidai. Again, Scelio and several related genera at- 

 tack only the eggs of the Orthoptera Saltatoria. Thus, 

 if used with due caution, egg-parasites are in the main 

 illustrative of close correlation between the taxonomy of 

 host and parasite in spite of the fact that we may nat- 

 urally regard insect eggs as more similar inter se, than 

 insect larvae. 



It is true that the ubiquitous little Trichogramma af- 

 fects eggs of several orders and many families of insects, 

 but like other less conspicuous examples, it stands quite 

 apart from its conmionplace associates. 



With their larger and more variable size, and great 

 diversity in habits and structure, larval insects present a 

 correspondiiiiily vni it .l sorics of opportunities for para- 

 sites. We liiul also that practically no genera are known 



