Later, in the southwestern part of the range of the MSA stock (southern Texas and northern México), the SA and muticus stocks were separated. The muticus stock occurred to the northeastward, and presumably no farther south than the area included within the present drainage basin of the Colorado River. Southward, the SA stock was isolated into several populations that are today represented by ater and T. s. emoryi, the most variable subspecies; the distribution of the most distinctive population of emoryi indicates a former isolated inland drainage. The multiple fragmentation of the SA stock presumably terminated by the end of the Pliocene. The progenitors of T. ater probably closely resembled ferox. Trionyx ater and T. ferox resemble each other morphologically and in habitat. Therefore, the progenitors of ater are considered to have undergone comparatively little differentiation.
The spinifer stock, occurring principally in the area included within the present drainage basin of the Río Grande, extended its geographic range eastward and became sympatric with muticus and ferox. An expansion of range necessarily demands more mesic conditions; these were perhaps afforded by the pluvials (wet, rainy ages) that were coincident with the glacial periods in the Pleistocene (Antevs, 1948:168). The pluvials permitted the isolated populations of the spinifer stock to unite, and permitted that stock to extend its range eastward. The concurrent continental glaciation permitted the spinifer stock to extend its range eastward only in a belt approximately 300 miles wide along the Gulf Coast, and also displaced the ranges of ferox and muticus to southern latitudes. Perhaps ferox was less tolerant of decreased temperatures or changes in habitat than was the spinifer stock but, for some unknown reason, ferox did not extend its range westward. Because T. ater closely resembles T. s. emoryi, continued isolation of ater since the beginning of the Pleistocene seems unlikely and ater may have been reunited in subsequent pluvial periods with the spinifer (emoryi) stock. A climatic fluctuation between relatively wet and dry periods is corroborated by studies of soil profiles in Trans-Pecos Texas (Bryan and Albritton, 1943).
The separation of the range of spinifer in the general region of western Louisiana, resulting in the differentiation of the spinifer group of subspecies to the east and the emoryi group of subspecies to the west, and the differentiation of T. s. asper and T. m. calvatus, both having corresponding western limits of distribution (Mississippi River drainage), are associated with the activities of the Mississippi River and its flood-plain. The combined effects of the pluvials and interpluvials seem responsible for changes in the lower Mississippi Valley. Great volumes of summer melt-water in the glacial stages greatly increased the breadth of the channel of the lower Mississippi River (corresponding to the northern extent of the Mississippi embayment; Hobbs, 1950), and this, coupled with the encroachment of Pleistocene seas (especially in the Mississippi embayment) in the interglacial periods, perhaps separated populations eastward represented today by T. m. calvatus and T. s. asper. The spinifer-hartwegi stock probably developed in southern Louisiana in association with the meandering of the Mississippi River and its tributaries, and its broad alluvial plain. The biota of that plain differed from that adjacent to the east or west (see discussion in Viosca, 1944) and constituted a barrier, of a sort, to free communication between the east and west. Westward the emoryi group of subspecies differentiated, its eastern limit probably being the Red River, which followed its own course to the Gulf along the lowlands on the west side of the Mississippi Valley and did not empty directly into the Mississippi until Recent times (Holland, 1944:20). There was not an equally-marked, corresponding separation of the range of muticus. However, the juvenal pattern of the subspecies muticus that inhabits the Gulf Coast streams is slightly different (having less short lines) from that of muticus elsewhere.
The Río Grande (inhabited by emoryi) presumably had its own exit to the Gulf whereas rivers westward to (and including) the Red River (inhabited by pallidus-guadalupensis cline) probably were joined near their mouths forming a large drainage system. Hubbs (1957:93) pointed out that the Río Grande-Nueces divide also limits a large number of species of fish. The differentiation of pallidus and guadalupensis is possibly due to a difference in the salt content of waters that drain the Edward's Plateau (see page 547), or to isolation of those subspecies in separate drainage systems that had their own exits to the Gulf.
In the lower Mississippi drainage, the spinifer-hartwegi stock extended its range northward following the retreat of the last glacial stage, and differentiated into those two subspecies in the upper Mississippi drainage and Great Lakes-St. Lawrence drainage system.
I have seen one specimen (UMMZ 59198) from the eastern part of the Tennessee drainage (inhabited by T. s. spinifer) that resembles T. s. asper (occupying the Gulf Coast drainages of the southeast). This resemblance tends to support the thesis of a former confluence of the Coosa (Alabama River system) and Tennessee drainages as believed by some malacologists to explain resemblances in molluscan fauna and as corroborated by physiographical evidence (see discussion in van der Schalie, 1945).
The Importance of the Study of Turtle Populations in Relation to the History of River Systems
In the Río Grande drainage the geographic distribution of the population of emoryi having orange color in males is approximately the same as that of Pseudemys scripta gaigeae; the corresponding distributions suggest that a part of the Río Grande drainage consisting of the Río Conchos in Chihuahua and the Big Bend region of Texas was isolated in former times. Accordingly, the known aquatic chelonian fauna in the basin of Cuatro Ciénegas in central Coahuila, México, is endemic (except T. s. emoryi). And the coincidence of the geographic ranges of T. muticus calvatus and Graptemys pulchra in the southeast suggest a former association of the included (Pearl to Escambia) river systems. The occurrence of T. s. pallidus in the Red River drainage indicates that the Red River was formerly associated with the Gulf Coast streams of eastern Texas and western Louisiana (inhabited by pallidus) and not with the Mississippi River drainage. The lower Mississippi River valley forms a prominent barrier to the eastern and western dispersal of many kinds of species and subspecies of turtles. T. m. calvatus and T. s. asper, which occur in rivers of the Gulf Coast drainage east of the Mississippi, are well-differentiated subspecies showing little or no evidence of intergradation with their relatives in the Mississippi River. The large faunal break provided by the Mississippi River would seem to indicate greater age for that river than for other rivers of the Gulf Coast drainage.
A comparison of the distributions of Trionyx and Graptemys in Texas suggests a faunal break between the drainage systems of the Brazos and Colorado rivers. Graptemys versa occurs in the Colorado and Guadalupe-San Antonio drainages. To my knowledge versa hitherto has not been recorded from the latter drainage system. I have seen one specimen of Graptemys (custody of Gerald Raun, University of Texas) from the Guadalupe River drainage, which I judge to be representative of versa, and Olson (1959:48) has reported Graptemys (probably versa) in the San Antonio River. The distribution of G. versa parallels in a general way, the distribution of T. s. guadalupensis. G. kohni and T. s. pallidus occur in the Brazos River and eastward. Also, it is notable that the population of T. m. muticus occurring in the Colorado River drainage differs slightly (more black pigmentation) from the same subspecies in the adjacent Brazos River system.
There is much difference in the patterns of distribution and degree of differentiation of different genera of aquatic turtles in the eastern United States. Tinkle (1958:41-43, Figs. 49-55) concluded that a general resemblance in the patterns of distribution of the different genera of turtles was evidence that the rates of evolution were essentially the same, assuming that each genus had had a similar time interval for differentiation (op. cit.:42). If this is true, corresponding patterns of distribution might indicate the same relative age of the population of turtles concerned. Generally, the genera of turtles that on morphological grounds are considered the oldest and most primitive (Macroclemys, Chelydra) show less differentiation into species and subspecies than those considered younger and more recently evolved (Graptemys, Pseudemys). In the genus Graptemys, much differentiation occurs in the geologically, recently formed, Gulf Coast drainage systems of the southeastern United States. It would seem then, that faster rates of differentiation denote more recent genera, whereas older genera are endowed with a "genetic senility" and are less subject to change.