E-H. _Pliosaccomys dubius_, left upper tooth-row, including P4-M2 (M3 unknown). Patterns based on Nos. 1798 and 1799 (LAM) from Smiths Valley (Middle Pliocene), Lyon Co., Nevada.

I-L. _Pliosaccomys dubius_, right lower tooth-row, including p4-m3.

Patterns based on Nos. 1796 (holotype), 1804, and 1806 (LAM) from Smiths Valley (Middle Pliocene), Lyon Co., Nevada.

The lineage of the Th.o.m.omyini is essentially rectilinear and without the major branching seen in the tribe Geomyini. The one genus, _Th.o.m.omys_, appears first in the Upper Pliocene (early Blancan time), and the specializations characterizing the lineage had already developed by that time. Evidently, the early stages of divergence from the ancestral stock resulted in the development of rootless, ever-growing, more hypsodont cheek teeth, simplification of M3, and enlargement of the ma.s.seteric ridge on the mandible. The enamel investment on the sides of the molariform teeth is interrupted owing to intrusion of tracts of dentine on the sides of each column. Even so, complete anterior and posterior plates are retained on all of the cheek teeth (Fig. 5, K and L) and there is no trend toward additional loss of enamel as in the Geomyini. The enamel on the sides of the column has little functional value, and its elimination probably reduces friction during the anteroposterior movements of the lower jaw, thereby increasing the efficiency of the cutting blades on the anterior and posterior wall of the tooth. The simplification of M3 was achieved by union of the two columns of the primitive pattern into a single column and obliteration of both the l.a.b.i.al and lingual re-entrant folds in the first stages of wear. The adult tooth (see Fig. 5L) is without trace of the bilophate pattern and is not elongated; therefore, its structure is essentially the same as that of the first and second upper molars.

In the Th.o.m.omyini, the two lophs of the unworn molars unite entirely across the width of their surfaces with the first traces of wear (see Fig. 5, I and J), owing to the shallow and uniform depth of the transverse valley. In the molars, the final pattern is acquired, therefore, before the deciduous premolar has been replaced by the permanent tooth. A relatively shallow re-entrant inflection between the ends of the parameres sometimes is retained, although it also will disappear with slight additional wear. Therefore, both lophs tend to unite completely with the first stages of wear in the Th.o.m.omyini, thus omitting both U and H patterns from the sequence of wear. This is the highest degree of specialization attained in the Geomyidae in regard to the patterns of wear, since a sequence of bilophodont patterns appear in both the Dikkomyini and Geomyini before the monoprismatic pattern is developed.

[Ill.u.s.tration: FIG. 5. Drawings of molariform dent.i.tions representative of the tribes Geomyini and Th.o.m.omyini depicting patterns of wear on the occlusal surface. A-D represent, in ontogenetic sequence from left to right, upper tooth-rows of the tribe Geomyini. E-H represent, in the same sequence of stages, lower tooth-rows of the tribe Geomyini. I-L represents both upper and lower tooth-rows of both pre-final and final stages of wear in the tribe Th.o.m.omyini. All 5.

A and E. _Geomys bursarius majusculus_, No. 2948 (KU), Douglas Co., Kansas. Right upper (A) including DP4-M3; lower left (E) including dp4-m3.

B and F. _Pappogeomys bulleri burti_, No. 100444 (KU), 10 mi. NNW Barra de Navidad, Jalisco. Right upper (B) including P4-M3; right lower (F) including p4-m3 (both P4 and p4 with unworn enamel caps).

C and G. _Pappogeomys bulleri albinasus_, No. 31044 (KU), 10 mi. S and 8 mi. W Guadalajara, Jalisco. Right upper (C) including P4-M3; right lower (G) including p4-m3.

D and H. _Pappogeomys bulleri albinasus_, No. 31002 (KU), W side La Venta, 13 mi. W and 4 mi. N Guadalajara, Jalisco. Right upper (D) including P4-M3; right lower (H) including p4-m3.

I and J. _Th.o.m.omys talpoides bridgeri_, No. 6865 (KU), 2 mi. up Mink Creek, Pocatella, Bannock Co., Idaho. Left upper (I), DP4-M3; left lower (J), dp4-m3.

K and L. _Th.o.m.omys talpoides fossor_, No. 13205 (KU), Wa.s.son Ranch, 3 mi. E Creede, Mineral Co., Colorado. Right lower (K), p4-m3; left upper (L), P4-M3.

Relationship of the Geomyini with the ancestral Dikkomyini is most clearly demonstrated in the sequence of wear on the occlusal surfaces of the molars. As in all geomyids, the upper part of the crown is biprismatic in the newly erupted tooth, and the two columns are separated by an intervening valley. With slight attrition on the unworn enamel cap, the weakly developed cusps merge and form a transverse enamel loop on each of the two columns (see third molar in Fig. 5, A and E), each loop enclosing a core of dentine that had become exposed. The valley between the two columns is shallow, and upon further wear of the tooth, the two loops unite. The two columns become joined at different points in the upper and lower molars depending on the varying depth of the valley in different teeth.

Therefore, upper and lower molars develop distinctly different occlusal configurations.

In the lower molars, the pattern characteristic of _Dikkomys_ (Fig.

4C) is preserved without significant modification, as ill.u.s.trated in an immature specimen of _Geomys_ (see Fig. 5E). The H-pattern and modified H-pattern are developed in the same sequence of wear in the Geomyini. A juvenal female (not ill.u.s.trated), KU 2931, provides an example of the intermediate H-pattern. In this specimen, the protolophid and hypolophid of the left m2 are united only at their mid-points, indicating that the pattern of wear occurs in the same sequence in the Geomyini as it did in the Miocene genus _Dikkomys_.

After the two columns have become united at their mid-points, a secondary union is formed at the edge of their protomeres, thus enclosing the enamel fossette as ill.u.s.trated in Figure 5E (this is the modified H-pattern mentioned above). However, the fossette itself is shallow and soon disappears with slight wear. At this stage, the occlusal configuration would be in a U-pattern (m1 in Fig. 5E). The lingual re-entrant fold is also shallow in vertical depth; therefore, it is obliterated by wear following the eradication of the l.a.b.i.al fossette. Consequently, the two columns are united into one. In m3 (see Figs. 5E, F, and G), the two columns merge by progressive lateral expansion of the medial isthmus.

In the first and second upper molars, the two columns unite across the entire surface of their protomeres from near the lingual edge of the crown to near its center. A minute inner inflection may be temporarily retained in some teeth. At this stage (see Fig. 5B), the parameres are still separated by the l.a.b.i.al fissure, and the occlusal pattern is in the shape of a U, resembling, but not exactly duplicating, the pre-final pattern of Ml and M2 in the genus _Pliosaccomys_ (see Fig.

4H). The l.a.b.i.al fissure is shallow, and, with further wear, the inflection is worn away and the parameres also unite, thereby forming a monoprimatic crown in the final stage. In M3, the two lophs first become united near the edge of their protomeres (see Fig. 5B), therefore forming a U-pattern similar to that developed in Ml and M2 of _Pliosaccomys_. The connection of the two lophs is not directly at the end of the protomere; consequently a shallow lingual inflection remains. The lingual edge of the valley is also shallow, and, with continued wear a second union of the two lophs takes place near the ends of their parameres, and the deeper, interior part of the valley remains as an isolated enamel fossette (see Fig. 5C). The two primary lophs of the tooth are now joined near both sides, having shallow lingual and l.a.b.i.al re-entrant angles on the sides and the enamel island in the center. With continued effacement of the occlusal surface, the fossette will be eradicated, and the pattern of the occlusal surface will become the partially biprismatic pattern of the final stages (adult) of wear (see Fig. 5D). M3's of _Dikkomys_ and _Pliosaccomys_ are not known; however, it seems reasonable to a.s.sume that the pattern of wear in the M3 of Dikkomyini was not essentially different from that of the Geomyini, except that it is likely that the U-pattern of the second stage of wear in the Geomyini was probably the final stage in the genus _Dikkomys_.

Judging from the pre-final stages of wear, the dent.i.tion of the Geomyini provides a curious combination of patterns that resemble in part the Miocene genus _Dikkomys_ and in part the early and middle Pliocene genus _Pliosaccomys_. There is no significant variation in the premolars or third molars (at least in the lower teeth) of the Geomyinae from the early Miocene to late Pliocene; therefore, deviations of major significance are in the character of the first and second molars. In the Geomyini, the patterns of wear of m1 and m2 are the same as those of _Dikkomys_, and are distinctly different from those of _Pliosaccomys_ where the two columns first unite at the edge of their protomeres to form a U-pattern, rather than at their mid-points to form an H-pattern. Even though the intermediate stages of ontogeny in m1 and m2 of _Pliosaccomys_ and the Geomyini are entirely different, the bicolumnar crowns of both eventually unite, upon wear, into a single column. On the other hand, the patterns of M1 and M2 in the Geomyini most closely resemble those of _Pliosaccomys_, rather than _Dikkomys_. In this regard it should be pointed out that the upper molars of _Dikkomys_ are presently represented by only one tooth, an M1 in an early stage of wear. As described already, the patterns of M1-2 evidently would be mirror images of m1-2 in corresponding stages of wear. However, the initial union of the two columns, in the M1 that is known, is somewhat to the lingual side of center and the relatively small lingual valley does not reach the base of the crown, indicating, that eventually with wear, the two columns of _Dikkomys_ might have become united across the entire surface of their protomeres as in _Pliosaccomys_. Even so, the two columns of M1 do initially join closer to their mid-points than they do in _Pliosaccomys_, and, if they did actually unite across their protomeres, the union would have occurred with subsequent wear. That is, the first occlusal pattern would be H-shaped (but with the connection closer to the lingual than the l.a.b.i.al side), as in m1 and m2, and it would become U-shaped only after additional wear. This sequence of patterns of M1 and M2, as already pointed out, does not pertain in _Pliosaccomys_ or the Geomyini, since the U-pattern is formed with the first union of the two columns at the edge of their protomeres, and the primitive H-pattern is never developed, unless one counts the slight lingual inflection, that occasionally is formed just after the two columns unite, as being indicative of the primitive pattern. As in the lower teeth, the bicolumnar crowns of early ontogeny in both _Pliosaccomys_ and the Geomyini become eventually united, with wear, into a single column.

Based upon the foregoing evidence, it would seem likely that the Geomyini evolved from an early Pliocene (perhaps late Miocene) Dikkomyini ancestor that had evolved the specializations of M1 and M2 that characterize its relative, _Pliosaccomys_, but had not also evolved the specializations of m1 and m2 that distinguish _Pliosaccomys_. Therefore, the ancestor of the Geomyini differed from the _Pliosaccomys_-Th.o.m.omyini lineage in its retention, unmodified, of the primitive patterns in m1 and m2 that characterized the earliest known Geomyines (_Dikkomys_). The same patterns are preserved in m1 and m2 of its modern descendents, the living Geomyini. In the _Pliosaccomys_-Th.o.m.omyini lineage the pattern of m1 and m2 are entirely different, as described above.

The earliest record of the Geomyini is the extinct genus _Pliogeomys_ (see Fig. 6) in the latest Hemphillian (middle Pliocene) and earliest Blancan (late Pliocene). _Pliogeomys_ is more primitive than any modern genus of the Geomyini, seems to have been a late survivor of the primitive stock, but was itself probably a collateral lineage and not on the direct line of descent. The cheek teeth in _Pliogeomys_ are rooted and less hypsodont than in the late Pliocene examples of the modern genera, and the anterior enamel plate of the lower molars shows no indication of reduction, as would be expected if _Pliogeomys_ were in the direct line of evolution. Separation of _Pliogeomys_ from the main stem of the Geomyini probably occurred after several specializations had already been achieved by the Geomyini. Two inheritances might have been grooving on the upper incisors and some reduction in amount of enamel on the sides of the cheek teeth. The dentine tracts on the sides of the cheek teeth of _Pliogeomys_ are narrow (see Fig. 7A) and barely separate the enamel blades and there is no discernible reduction in the anterior enamel blades on its lower molars. Those blades evidently were lost in the main lineage before the Pleistocene radiation of the living genera took place.

_Pliogeomys_ is in an intermediate stage in evolution, and was not so advanced as was the main lineage at the time _Pliogeomys_ died out.

Its structure does provide clues as to phyletic development that took place in the main lineage.

Specialized trends in the early phylogeny of the Geomyini included: development of rootless, ever-growing cheek teeth and an increase in hypsodonty; loss of the bicolumnar structure of the first and second molars, and, consequently, the formation of a single elliptical column in the final stage of wear; interruption of the enamel investment of the molariform teeth and formation of anterior and posterior enamel plates; and enlargement of the ma.s.seteric ridge and fossa. Each of these trends occurred independently in the Th.o.m.omyini, and each is an example of parallelism in the phyletic evolution of the two lineages.

Three additional specializations lacking in the Th.o.m.omyini are the grooving on upper incisors, loss of anterior enamel plate in lower molars, and development of a basitemporal fossa on the mandible.

Evidently, two grooves evolved in the ancestral incisors in the same bisculcate pattern preserved in _Pliogeomys_, _Zygogeomys_ and _Geomys_. The innermost groove is weakly developed in _Pliogeomys_, suggesting that this character was in an intermediate stage of evolution in the ancestral lineage at the time that _Pliogeomys_ split off. Numerous other specializations in the Geomyini appeared later, but evolved in the different genera that diverged from the ancestral lineage and are discussed separately in the next account. Only two of the major features characterizing the Dikkomyini are retained in the Geomyini: the H-pattern on the occlusal surface of the m1 and m2 developed during the initial stages of wear, and the bicolumnar pattern of M3. Adaptive radiation produced the living genera of the Geomyini in the late Pliocene and early Pleistocene (see Fig. 6) and subsequent specialization of the ancestral morphology followed.

Parallelism in the molars of later geomyines and the Entoptychinae is ill.u.s.trated by the lateral interruption of the enamel investment and loss of enamel plates and by the omission of the H-pattern stage in the first and second molars (in _Pliosaccomys_). Resemblance of dent.i.tions in certain stages of wear in _Pliosaccomys_ and in entoptychines led some investigators, for instance, Hibbard (1953:357), to suggest that _Pliosaccomys_ descended from one of the less specialized entoptychines, possibly _Grangerimus_ but probably _Gregorymys_. Actually, the highly specialized upper and lower premolars and third molars of the entoptychines rule them out as ancestors of the later geomyines. The evolution of entoptychine-like features in _Pliosaccomys_ is regarded as an example of iteration, a pattern of parallelism (see Simpson, 1953:248-253) where an allochronic and independent lineage undergoes the same evolutionary trend that phyletically characterized an earlier lineage, usually after the latter has become extinct. In this case, the lineage giving rise to _Pliosaccomys_ pa.s.sed through the same phyletic stages in its evolution in the early Pliocene (and possibly the late Miocene) as did the entoptychines in the late Oligocene and early Miocene.

Another parallelism by iteration, occurring in the middle and late Pliocene in both the Th.o.m.omyini and Geomyini, is the loss of enamel from the lateral surfaces of the cheek teeth, and, in the Geomyini only, the eventual loss of the anterior plate in the lower teeth and the posterior plate in the upper teeth. Both features were evolved more than an epoch earlier in the specialized entoptychid genus _Entoptychus_ of the lower Miocene. In _Entoptychus_, only the posterior plate of the lower molars and the anterior plate of the upper molars remained in the final stages of attrition, although a central enamel fossette, a remnant of the re-entrant fold, remained throughout life. Iteration is also expressed in the subfamily Geomyinae by the development of grooving on the upper incisor and the formation of the basitemporal fossa. A shallow but distinct basitemporal fossa occurs between the coronoid process and the third lower molar in the genus _Entoptychus_ and a sulcated upper incisor, a single shallow groove usually near the median border of the tooth, is found in the genus _Gregorymys_ of the subfamily Entoptychinae. Both features are regarded as advanced specializations in the tribe Geomyini, even though each was evolved in the entoptychines of the Lower Miocene.

The postcranial skeleton of living genera of pocket gophers, as befits animals that spend most of their life within underground burrows, are highly specialized for a fossorial life. Elements of the postcranial skeleton recovered from Lower Miocene deposits indicate that the entoptychines were only semi-fossorial (see Cope, 1884:857; Wood, 1936:4-5; Wilson, 1949:117-118). One of the basic trends of the entoptychines was towards greater fossorial adaptation; the skeleton of _Entoptychus_ shows a greater degree of fossorial adaptation than earlier genera of the subfamily. There is no reason to suppose that the geomyine genus _Dikkomys_, which lived at the same times as the entoptychines, had acquired any more advanced fossorial adaptations than had the entoptychines.

The most p.r.o.nounced fossorial adaptations seem to have evolved only in the ancestral lineage of the modern geomyines, probably in the latter part of the Miocene and in the early Pliocene, before the modern Th.o.m.omyini and Geomyini diverged. Extreme fossorial adaptations in herbivorous rodents, such as those characteristic of the modern pocket gophers and their immediate ancestors, are thought to have evolved only in response to p.r.o.nounced arid conditions. The Entoptychinae and evidently the early geomyines lived in environments that were either tropical or temperate, and under conditions more mesic than I would consider necessary to bring about selection pressure resulting in fossorial specializations. In late Oligocene and early Miocene, according to Axelroad (1958:433-509), arid conditions did not exist in the United States, and the only xerophytic environments in North America occurred on the Central Plateau of Mexico. Moreover (Axelroad, _loc. cit._), arid conditions did not develop in the western United States until the early Pliocene. Geomyids evidently became extinct in this region at the close of the Middle Miocene, and none appear in fossil deposits in the western United States until the latest Lower Pliocene (Clarendonian). The reappearance of geomyids, _Pliosaccomys_, in the western United States coincides with a trend toward aridity and the northward movement of the Madro-tertiary geoflora into the Great Basin and Great Plains from its place of origin on the Central Plateau of Mexico (Axelroad, _loc. cit._). Later, in the middle and later Pliocene, the Madro-tertiary geoflora gave rise to the modern xerophytic plants that now characterize the desert vegetation of North America.

The Madro-tertiary climax does not appear as a major flora until the Miocene, but probably originated earlier. According to Axelroad (_loc.

cit._), this xerophytic flora evolved from elements of the Neotropical-tertiary geoflora that became adapted to arid conditions that developed in the rain shadow of the high mountains flanking the Central Plateau of Mexico. Originally, the Madro-tertiary flora consisted of small trees, shrubs, and gra.s.ses. Although some elements of this flora moved northward in the late Miocene, the major part of it remained in Mexico until the early Pliocene. In the western United States, mountain formation increased in intensity in the Pliocene and continued on into the early Pleistocene. As the mountains became more elevated, especially the Sierra Nevada and Cascade ranges, they blocked the prevailing winds from the Pacific Ocean and extensive aridity developed on their leeward side. As xeric conditions became widespread, the Madro-tertiary flora successfully occupied the drier regions of southern California, the Great Basin, and the western parts of the Great Plains.

While the Entoptychinae probably evolved in response to the Arcto-tertiary flora, the late Tertiary geomyines probably evolved in response to the Madro-tertiary geoflora on the Central Plateau of Mexico. Some of these early geomyines, especially ancestors of the modern lineages, probably were pushed southward by compet.i.tion with the more specialized entoptychines. Most geomyines were pushed out of the northern area of distribution, except for _Dikkomys_ that survived in a.s.sociation with the entoptychids throughout the early and middle Miocene. During this time, and probably continuing on into the late Miocene, the geomyines occurring to the south in Mexico became adapted to the arid environments of the Madro-tertiary geoflora.

Of course, information is lacking about climates in several parts of the late Miocene and early Pliocene. When such information becomes available it conceivably could modify the hypothesis outlined immediately above.

The princ.i.p.al trend of evolution in these semi-fossorial rodents was toward more complete fossorial adaptation, and the p.r.o.nounced fossorial features characteristic of the modern pocket gophers were perfected. This trend continued in response to the intense selection pressures in this arid environment. The princ.i.p.al structural characters effected were in the postcranial anatomy, especially in the skeletal and muscular systems. Consequently, it is not surprising that in skull and dent.i.tion, _Pliosaccomys_ differs but little from _Dikkomys_. Therefore, most of the basic structural specializations so far developed for subterranean existence probably had evolved by the time geomyines moved back north in the early Pliocene. Both modern lineages, the tribes Th.o.m.omyini and Geomyini, have essentially the same fossorial features, and it seems unlikely that these features were acquired independently in the relatively short period of time available to them after their divergence; probably they were inherited from a common ancestor. These probabilities indicate that the evolution of the fossorial specialization was in the later phyletic development of the tribe Dikkomyini.

Plio-Pleistocene radiation of Geomyini

Unlike the lineage of the Th.o.m.omyini that remained essentially rectilinear through out its history, the Geomyini in the late Pliocene and the early Pleistocene underwent adaptive radiation in a degree comparable to the earlier radiation of the Entoptychinae, and all of the later history of the tribe is dominated by the radiation--the resulting structural diversity. At least four lineages were produced by the Plio-Pleistocene radiation (see Fig. 6); each originated at essentially the same time (late Pliocene) presumably from the same ancestral stock. Each of these lineages within the Geomyini has given rise to one of the four modern genera: _Zygogeomys_, _Geomys_, _Orthogeomys_, and _Pappogeomys_.

[Ill.u.s.tration: FIG. 6. Plio-Pleistocene radiation of the Tribe Geomyini.]

_Morphotype_

The immediate, unknown, ancestor probably lived on the Central Plateau of Mexico. After the radiation began the ancestors of _Geomys_ and _Zygogeomys_ extended their ranges northward.

Features of the hypothetical morphotype, that would permit derivation of the modern genera would include the following: (1) Skull generalized, neither excessively long and narrow or short and broad; (2) skull smoothly rounded, without p.r.o.nounced angularity, rugosity or cresting (sagittal crest probably lacking, even in old individuals); (3) zygomata slender, without lateral platelike expansions; (4) rostrum moderately broad; (5) upper incisors bisulcate, two grooves in pattern found in _Pliogeomys_, _Zygogeomys_ and _Geomys_; (6) lateral re-entrant angles of premolars obtuse; (7) p4 having four enamel plates (one on anterior wall, one on posterior wall, and two lateral plates) and lower molars having one enamel plate on the posterior wall of tooth (anterior plate is lacking); (8) P4 having four enamel plates, in same pattern as described for p4, M1 having two enamel plates (one anterior and one posterior), M2 same as M1, M3 having three plates (one anterior, two lateral on sides of posterior loph, none posterior); (9) M3 subtriangular in cross-section, distinctly bicolumnar, two columns marked by shallow re-entrant folds and connected by broad isthmus; (10) ma.s.seteric ridge large, forming high crest bordering ma.s.seteric fossa; (11) basitemporal fossa shallow; (12) angular process of mandible short, its lateral projection barely exceeding that of zygomatic arch.

_Specializations in Genera_

In relation to the primitive morphotype, increase in size, simplification of dent.i.tion, and changes in shape of skull are regarded as specializations. Considerable parallelism between the four lineages is seen. But each lineage is distinguished by a combination of specialized features, and three by a few unique specializations.

Among trends resulting in simplification of the dent.i.tion, reduction of enamel on the posterior wall of the upper cheek teeth has occurred in various degrees in all lineages of the Geomyini even to loss of all enamel on the posterior wall of the premolars and molars in two genera. Loss of some enamel is more common on P4 than on M1-2, and has occurred in all genera (see Figs. 7 and 9.)

In evolutionary sequence loss of enamel from M1 and M2 usually occurs after, but never preceding, the reduction of enamel on P4. Loss of enamel plates from the posterior face of M1 and M2 is a.s.sociated with the evolution of an efficient anterotransverse shearing action of the teeth.

On the anterior wall of those teeth no reduction of the cutting blade has been observed; a complete anterior plate is retained in all living Geomyini.

Presence of both the posterior and anterior plates decreases the efficiency of transverse shearing, by providing two upper plates (anterior plate of one tooth and posterior plate of the preceding tooth) over which the lower cutting blade _simultaneously_ must pa.s.s with each movement. The advantages of shearing over the more common mechanics of planing are largely lost unless the posterior plates are eliminated. Also, none of the living Geomyini have retained a definitive posterior enamel plate on M3, the last upper molar; but two well-developed lateral plates, that extend almost all of the way back to the posterior apex of M3, have been retained, and, together function as a posterior plate. Loss of either or both of the lateral plates of M3 is rare, and occurs only in old individuals. Their loss in the final stages of wear may represent the beginning of a new trend in those species where it occurs (the _castanops_-group of the subgenus _Cratogeomys_). In any case, reduction of enamel takes place by transverse shortening of the plate through the complete loss of enamel on one end, the diminution beginning first on the l.a.b.i.al end and proceeding by progressive atrophy to the lingual end of the plate.

Evidently, when enamel has been eliminated from the l.a.b.i.al end of a plate, the rate of loss decreases markedly, and the last stages of evolution, terminating in complete loss of an enamel plate, occurs more slowly. Evolution may be arrested before complete loss has occurred, and that part of the enamel that remains forms a short, vestigial plate restricted to the lingual one-fourth or one-third of the wall. The enamel pattern of the lower dent.i.tion is the same in all of the diverging lineages, with no evidence of additional loss of enamel from that which had already occurred in their common ancestor (see Figs. 7 and 9). Reduction and loss of enamel plates began and was terminated in the lower dent.i.tion before reduction began in the upper dent.i.tion.

Other dental specializations have occurred in the shape of the third upper molar and in the pattern of grooving in the upper incisor.

Unlike M3 of the Th.o.m.omyini, that of the Geomyini differs in shape from M2, and its enamel investment differs from that of M2.

Primitively, M3 was probably subtriangular in cross-section, and the posterior loph evidently projected posteriorly as a short, rudimentary heel that formed the apex of the triangle. Other shapes of M3 are considered to be specializations that have been derived from the primitive form. In addition to the primitive subtriangular pattern, the M3 of living Geomyini may be suborbicular, quadriform, elongate, or obcordate in shape. Usually each lineage is characterized by only one pattern, but in one genus (_Pappogeomys_) all patterns occur. Of the different forms, the elongate and obcordate seem to be the most highly specialized deviations from the triangular-shaped tooth. The bicolumnar pattern is accentuated in the elongate type (Fig. 7D, F, H) by deep lateral re-entrant folds, on both the lingual and l.a.b.i.al sides, and by the elongation of the posterior loph into a p.r.o.nounced heel. Teeth having this pattern have been ill.u.s.trated by Merriam (1895:76-82) in Figures 27 (6 and 7), 28 (c and d), 34 (7 through 15), and 35 (8).

[Ill.u.s.tration: FIG. 7. Molariform dent.i.tions of the Tribe Geomyini.

Drawings ill.u.s.trating enamel patterns characteristic of _Pliogeomys_, _Zygogeomys_, and the subgenera of _Orthogeomys_ (_Orthogeomys_, _Heterogeomys_ and _Macrogeomys_). 5.