The fossil record of the Geomyidae provides a sequence of morphotypes, each representing a stage in the phyletic development of the family. Most of the preserved specimens probably represent the stufenreihe rather than the ahnenreihe, as Simpson (1953:219-220) points out. Even so, the stufenreihe closely approximates the general trend of evolution, and the level of structural organization in the different stages of phyletic development may be ascertained. The actual ancestral series of most lineages probably will remain unknown, but hopefully some of the existing gaps will be filled by future discoveries. From the established record, several clearly defined lineages can be distinguished; in fact the sequence of origin, pattern of evolution, and specializations, of the principal lineages are reasonably well expressed.
In the earliest known geomyids from the Upper Oligocene and Lower Miocene, the premolars and molars are biprismatic and bilophodont. In rodents, this is itself a specialized pattern, and is thought to have evolved from a more primitive sextituberculate prototype by the union of individual cusps, and probably also cuspules, forming the two transverse enamel lophs. The primitive, common ancestor of the Geomyidae and Heteromyidae with sextituberculate teeth in the early Tertiary is unknown.
As soon as geomyids attained the early bilophodont stage of evolution, the basic morphological structure of the family was established. The family probably first became clearly distinguished from other Geomyoidea at this stage. In the early bilophodont stages of evolution, owing to the relatively deep valley between them, the two columns probably failed to unite in the normal cycle of wear, as they do in all later geomyids. Griphomys described by Wilson (1940:93) from the late Eocene of California, has a bilophate pattern in which the anterior and posterior lophs are separated by a persistent transverse valley. The occlusal pattern of Griphomys closely resembles a stage through which the ancestors of the early Miocene geomyids must have passed in their pre-Miocene evolution, as Wilson suggests (1949:115-116). Although he (1940:95; 1949:110-118) tentatively referred Griphomys to the superfamily Geomyoidea and Simpson (1945:80) went so far as to refer it to the family Geomyidae, with a notation of incertae sedis, its exact relationship to the pocket gophers is uncertain. However, the structure of the molariform dentition of Griphomys does not exclude it from the phyletic ancestry of the Geomyidae. In subsequent stages of evolution the anterior and posterior columns become united. Thereby part of the valley floor between the transverse prisms was progressively elevated, to the stage where attrition on the occlusal surface would unite the two columns. On the unworn enamel cap of living geomyids the two transverse enamel folds are separated by a shallow but well defined valley, briefly reflecting the ancient ancestral pattern.
Union of the lophs may have been either at the mid-points of the two columns or at the edge of their protomeres. [A protomere is the half of a tooth containing the protocone or protoconid—lingual side of upper tooth and labial side of lower tooth. The paramere is the opposite half of a given tooth—labial side of upper tooth and lingual side in lower tooth. See Miller and Gidley, 1918:434.] Union of the columns at the mid-points would have produced the figure-8 occlusal pattern (or H-pattern), which is characteristic of the early Miocene Geomyinae (Dikkomys). Union of the two columns at the protomeres would have produced the U-shaped pattern of the Entoptychinae, which also occurred in the early Miocene and were contemporary with the earliest Geomyinae. Since pre-Miocene geomyids are unknown, the actual phyletic development of the dentition is a matter of speculation. Probably the development of the two divergent lineages, one leading to the Entoptychinae and the other to the subfamily Geomyinae, occurred in the Oligocene (as depicted in [Fig. 3]). Of the two lineages, the subfamily Geomyinae, in my view, is the more primitive and less specialized. Support for this view is furnished by a reconstruction of the pattern of occlusal wear in Dikkomys and Pliosaccomys, especially on the first and second molars.
In Dikkomys, the anterior and posterior column first unite near their mid-points in the first stages of wear thus producing a figure-8 shaped (H-shaped) occlusal pattern in the premolar and all three molars. Evidently in the first two upper molars, the columns unite closer to their lingual margins than their mid-points, but at any rate both outer and inner re-entrant folds are evident at this stage of wear. With continued attrition on m1 and m2 of Dikkomys, the anterior and posterior columns secondarily unite at the edge of their labial margins thus enclosing a fossette of enamel in the labial half of the tooth. The lateral coalescence at the ends of the protomeres occurs because of the shallow vertical depth of the labial re-entrant fold, and the fossette itself does not reach the base of the crown and with continued wear it too would disappear, but not until the last stages of wear, at least in Dikkomys matthewi. The lingual re-entrant fold is deep, and therefore, persistent through all stages of wear. Although the amount of wear required for its effacement would be great, the occlusal configuration of the first and second lower molars in Dikkomys could be eventually ground down to a U-pattern as in the entoptychids. Only one upper molar of Dikkomys, the first, has been recovered (see Wood, 1936:23, fig. 32B). Although the tooth is in an early stage of wear, the lingual valley is minute. Less attrition than required in m1 and m2 would progressively reduce the lingual fold until it too would essentially form a U-pattern, perhaps retaining a slight lingual inflection. Hence, the first upper molar becomes a mirror image of the first lower molar, and the second upper molar probably had the same pattern as the first (at least it does so in Pliosaccomys). Both of the lateral re-entrant folds of the premolar are deep vertically, and consequently would not disappear with occlusal wear. Therefore, the H-pattern of the premolars is retained throughout life.
The m3 (M3 unknown for Dikkomys or Pliosaccomys) also has deep lateral folds; hence, it too retains the H-pattern in all stages of attrition, although the isthmus between the two prisms may become wider in the final phases of wear (as it does in Pliosaccomys).
In Pliosaccomys, the stages of wear are essentially the same as those described for Dikkomys, except that the anterior and posterior loph of the first and second molars tend to unite closer to one side of the tooth, lingual side in upper molars and labial in lower. Only a slight inflection of the re-entrant fold is evident on the side of union, and the inflection disappears in the first phases of wear as the columns unite. Concomitant with the lateral shift in the initial point of coalescence of the transverse lophs, the occlusal penetration of the re-entrant fold from the opposite side increases in horizontal depth, and the fold extends medially more than half way across the occlusal surface, thus forming a pattern essentially like that of the entoptychids. The U-pattern in Pliosaccomys appears in the initial stages of wear without going through an earlier H-pattern as is the case in its Miocene ancestors of the genus Dikkomys, unless the minute inflection is considered as indicative of that stage. The two columns of the premolar and m3 are joined near their mid-points as in Dikkomys; therefore, they retain their primitive H-pattern, a feature unique to the Geomyinae.