Proechimys goeldii is the species which has the largest number of characters that are judged to be primitive, and it may be the oldest stock. P. semispinosus, P. longicaudatus and P. guyannensis may have been derived from an early splitting of the genus or they may have branched off the main stem at different times. P. canicollis, however, seems clearly to be an offshoot of P. guyannensis; canicollis shows greater resemblance to guyannensis than to any other species. P. g. vacillator is another close relative of P. guyannensis with the number of counterfolds almost as much reduced as in P. canicollis. Conceivably, vacillator is a full species, but the reduction in number of counterfolds in the teeth more probably expresses only one extreme of a gradient, as will be discussed below.
SUBSPECIFIC VARIATION IN THE SUBGENUS
PROECHIMYS
In spite of the lack of specimens from areas in which Proechimys certainly occurs, it is evident that the genus has great plasticity and that the number of subspecies will be greatly increased as additional material is studied. Only perfunctory examination of samples from outside the area of Brazil shows me that there are several unnamed subspecies there. My impression is that [Allen's] trinitatis, of Trinidad, the genotype of Proechimys, will eventually be split.
There are two main lines of subspeciation in Proechimys guyannensis. The one south of the Amazon River includes P. g. bolivianus, in Bolivia, P. g. villicauda, and P. g. ribeiroi occurring on the divide of the headwaters of the Amazon and Parana rivers, in Brazil, and P. g. hyleae in the lower Tapajoz and P. g. nesiotes in the lower Tocantins. All six subspecies have a large number of counterfolds in the molariform teeth. In these six subspecies, p4 has four counterfolds and the lower molars have three each. Toward the northeastern coast the number of counterfolds decreases to three in p4 and to two in the lower molars, as in P. g. arescens, P. g. leioprimna and P. g. oris.
In northern South America, north of the Amazon River, the subspecies with the greatest number of counterfolds is P. guyannensis warreni (known from only the Demerara River area); p4 has four counterfolds and the lower molars have three each. The number decreases in all the adjacent populations: P. g. guyannensis, in the Guianas, P. g. trinitatis, and P. g. urichi (going westward from the Guianas to Venezuela) have the counterfolds reduced to three in p4, but the lower molars still have the same number of counterfolds, namely, three, although there is a tendency for them to coalesce; farther west, on the coast, the number decreases to three counterfolds in p4 and to only two in the lower molars as in P. g. guairae and P. g. mincae. Subspecies south of the coast show the same reduction of counterfolds, P. g. cherriei and P. g. o'connelli being examples; P. g. ochraceus and P. g. poliopus have the reduction carried to the upper molars, M3 having usually only two counterfolds; P. g. chrysaeolus in the valley between the Madalena and the Cauca rivers seems to be somewhat isolated and shows reversion to three counterfolds in the lower molariform teeth; directly southward of the range of P. g. warreni the number of counterfolds decreases to three in all lower cheekteeth (population at Ayan-Tepuy, southern Venezuela), and then to three in p4 and to two in the lower molars, as in P. g. arabupu on the Brazilian side of Mount Roraima, and the reduction is extended to the upper molars in P. g. vacillator.
On the north bank of the Amazon, the only population of P. g. hyleae known to me (from Obidos) has four counterfolds in p4 and three in the lower molars; P. g. riparum, from Manaus, also on the north bank of the Amazon, has three counterfolds in p4 and two counterfolds in the lower molars. P. g. hyleae occurs also on the south bank of the Amazon. P. g. riparum, therefore, may be the northern part of the southern cline, instead of the southern end of the northern cline.
The whole picture, as outlined above, may be explained by assuming that the species P. guyannensis differentiated somewhere on the Central Plateau of South America, with three counterfolds in each upper molariform tooth, four counterfolds in the lower premolar and three counterfolds in the lower molars. The species might have extended its range to the Guianas and then all the biotypes with reduced number of counterfolds might have slowly developed by natural selection. The gradient is, broadly, from subspecies with greater number of counterfolds in more humid areas, to a gradually lessening number of counterfolds in less humid areas.
Proechimys longicaudatus is limited in the south to the headwaters of the Parana River drainage, where the subspecies P. l. roberti and P. l. longicaudatus are found. The species ranges northward through the Tapajoz drainage, with P. l. leucomystax in the headwaters and P. l. boimensis in the lower course. To the northwest and west the species is represented in Bolivia by P. l. securus; P. l. elassopus, P. l. simonsi, P. l. pachita, and P. l. hendeei occur in Peru and P. l. brevicauda in Peru and Brazil; and P. l. nigrofulvus occurs in Ecuador. Again in P. longicaudatus it seems that the number of counterfolds follows a gradient from more humid areas with four counterfolds in p4, as seen in nigrofulvus, pachita, simonsi, elassopus and brevicauda, decreasing to three or four in securus, to three in longicaudatus, but with m3 having only two counterfolds in leucomystax and roberti. P. l. boimensis, widely separated in the lower Tapajoz (no samples being known from the intervening range) may be the end of a cline started by leucomystax with only 2 counterfolds in m3 and ending to the northward with four counterfolds in m3. Over the same area the counterfolds in p4 increase from 3 to 4.
Of Proechimys goeldii I have had inadequate material but there seems to be a similar gradient in it which may be traced from P. g. steerei to P. g. goeldii. P. g. steerei has four counterfolds in more upper molars than occurs in the other subspecies.