The third body of physiologic data bearing on the nature of species is to be drawn from the host relations of our present insects. Every species of the 93 in the genus occurs on white oaks, including, however, oaks of both the groups Leucobalanus and Protobalanus of the Trelease classification (1924). In common with many other Cynipidae, these insects thus throw doubt on the validity of separating Protobalanus from the other white oaks.
The restriction of our genus, as here classified, to the white oaks is of especial interest because previous monographs of the group (Dalla Torre 1893, Dalla Torre and Kieffer 1910, and Beutenmüller 1911) have included many black oak species. Here is an illustration of the importance of basing biologic conclusions on sound taxonomic classifications.
The most significant of the host relations of long phylogenetic standing within the genus may be summarized as follows:
1. The Pacific Coast subgenus Beabicus represents three stocks, multipunctata, maculosa, and mirabilis, which are restricted to the groups of oaks centering about Q. lobata, Q. dumosa, and Q. garryana respectively. The subgeneric stock must have reached the Sierras, where it split into the three stocks, before the Great Basin became arid during the Miocene. Today, every individual of all the species of this subgenus shows sensory reactions of essentially the same nature as those shown by the ancestral stocks many millions of years ago.
2. Similarly, the ancestral stocks of Cynips echinus and C. guadaloupensis, of the Pacific Coast, have perpetuated their host preferences, Q. lobata-dumosa and Q. chrysolepis respectively, thruout the 9 species which these groups now represent.
3. One of the most special host restrictions in this genus is that of the four species of the Cynips centricola group on Q. stellata. This evidences the persistence in heredity of a specialized physiologic character.
4. In the Cynips mellea group, 8 of the 11 species similarly occur on Q. stellata. In this case, however, one may find stray individuals on Q. alba and other hosts, especially in regions where Q. stellata is rare or lacking.
5. The species grouped under Cynips pezomachoides and C. gemmula are obviously very close relatives. Nevertheless, the seven species of pezomachoides east of the Great Plains are confined to Q. alba and its close relatives while the three species of gemmula occur on chestnut oaks of the Q. Prinus group. The ancestors of each group must have been separated on the basis of host preferences, and their descendents still maintain the ancestral choice.
It would appear, then, that species in nature may be differentiated wholly or largely either on physiologic or morphologic
