Most of these characters vary but do not overlap. Subgeneric rank is here accorded to the two groups of Proechimys characterized immediately above.
The primary cause of the subgeneric differentiation is thought to have been geologic changes in the continental area. As already pointed out (see [Paleontology]), decreasing humidity in the Central Plateau of Brazil may have caused a migration southwestward of one or more of the species along with the forests. Once isolated geographically, the species probably differentiated at an accelerated rate.
The fact that a much larger number of subspecies occupies the larger geographic range of the subgenus Proechimys would not be sufficient to prove that this subgenus, Proechimys, is nearer to the primitive group than Trinomys, the subgenus occupying the smaller range with fewer subspecies. The paleontological evolution of the rodents, however, consistently points to teeth with a larger number of counterfolds (as seen in Proechimys) as the primitive condition. The extension of the main fold, tending to set apart one lamina in each upper molariform tooth, seems to be a specialization; reduction in the size of the head and body, increase in length of tail and decreasing size of molars posteriorly also may be specializations. The main point, however, is to establish if Trinomys is a relic group rather than a "differentiated" one. If an intermediate form were known which connected Trinomys with one species of Proechimys more than with another or even if Trinomys itself more closely resembled one of the groups of species of the subgenus Proechimys than it did another, we would assume that divergence and selection accounted for the subgeneric variation. The lack of any such connecting link favors the first idea, namely that Trinomys differentiated rapidly with the aid of geographic variation.
If Trinomys is, as I am inclined to consider it, the result of "differentiation," its subgeneric features are to be admitted as "new" and therefore the most primitive species in the genus should be found in the subgenus Proechimys.
It is a matter of common sense to admit the two groups considered above as subgenera rather than genera. Since the two structural plans were established they would, and do, act as different sources of variation. On the other hand, the morphological differences do not give the two groups an amount of morphological differences that would justify full generic rank for each.
SPECIFIC VARIATION IN THE SUBGENUS
PROECHIMYS
Most of the described forms in the subgenus were initially named as distinct full species. More recently, however, in accordance with the ideas now prevalent in systematic work, many of the named kinds were reduced to the rank of subspecies. [Tate] first made a geographic arrangement (1935:399-400) and later (1939:177-178) provisionally synonymised several named kinds of Proechimys with Proechimys "cayennensis cayennensis." A similar tendency was clearly displayed by [Ellerman] (1940:115-122) who allocated 29 names, out of 33 (in the subgenus, as here understood), to the species Proechimys guyannensis and gave full specific rank to four other named kinds. [Osgood] (1944) also had the same viewpoint; that is to say, he appeared to have the idea that there were only two full species in the subgenus in Brazil—admitting this orally—and consequently he synonymised some full species where two or more occurred in the same place, thinking that he was dealing with individual, rather than specific, differences. Evidently the number of species in the subgenus cannot be great because the known kinds show few patterns worthy of specific designation and therefore the majority of the existing names should be suspected of having no more than subspecific value. Nevertheless none of the above writers presented real evidence in support of his arrangement.
Criteria for the recognition of full species are most easily recognized where two or more different species live together. In the literature, P. goeldii and P. "oris" were mentioned by [Thomas] (1912:89) as having been collected in the same place; P. mincae and P. canicollis, by H. H. Smith (in [Allen], 1904:440); P. "leucomystax," from Utiarití, by Miranda [Ribeiro] (1914:42) and P. "longicaudatus," from the same place, by [Allen] (1916:569) were other examples. In these, and other alleged instances of two or more kinds occurring together, detailed study of the specimens concerned was necessary to learn the true facts. Also with the opportunity to compare collections from several different places, new facts emerged. P. longicaudatus, as it was conceived of by [Allen], was a composite species, but in one locality, Utiarití, [Ribeiro] and [Allen] actually were dealing with two distinct species.
The species, or subspecies belonging to different species, living together are: goeldii and hyleae, at Fazenda Paraiso; goeldii and riparum in Manaus; boimensis and hyleae in Tauarí; leucomystax and villicauda in Utiarití; mincae and canicollis in Bonda; gularis and hendeei on the banks of Rio Napo ("same trap lines," according to P. Hershkovitz, In Litt.). Study of samples of the above named pairs of kinds of Proechimys showed the following specific differences: goeldii is large with narrow aristiforms, has a large and strongly built skull, with four counterfolds in one or more upper molars: hyleae is smaller, has wide aristiforms, smaller skull with less pronounced ridges, and never has more than three counterfolds in the upper molariform teeth; riparum closely resembles hyleae; boimensis has thin aristiforms, small skull and no more than three counterfolds in the upper molariform teeth in contrast to hyleae, already discussed; leucomystax closely resembles boimensis; villicauda closely resembles both hyleae and riparum; mincae is similar to hyleae-riparum-villicauda; canicollis has the number of counterfolds in all molars reduced to two; gularis is large, has a strongly built and ridged skull, some upper molariform teeth with four counterfolds and wide aristiforms; hendeei closely resembles leucomystax and boimensis.
The evidence obtained from study of specimens where two or more species occurred together was applied to the remaining samples and the geographic distribution was worked out. As a result the arrangement below was made, including all valid kinds already named and those here newly named from Brazil. The names of kinds I do not consider as belonging to the subgenus (and genus) are excluded. These are Echimys macrourus Jentink, not seen, and Proechimys cayennensis hoplomyoides [Tate] (= genus Hoplomys). The application of names is tentative, however, because the types deposited in Europe have not been seen. An asterisk denotes the forms not seen by me.