Texas Proboscidea

The distribution of fossil proboscidea from the Texas Coastal Plain is summarized in Table 1.

Hemingfordian Proboscidea

A survey of the collections at TxVP resulted in a single proboscidean specimen from TMM 40081 that J.A. Wilson apparently considered to be from the Oakville Formation and part of the “Garvin Gully Fauna.” TMM 40081-14 (renumbered from TMM 40081-3) was originally assigned the specimen number TMM 31158-1. Archival records at TxVP suggest that Wilson revisited this locality and collected additional material after which time all specimens were reassigned to TMM 40081. Based on the Geologic Map of Texas (Barnes et al., 1992), the upper boundary of the Oakville Formation is located just north of Brenham, Texas, near TMM 40081. Unfortunately, the small collection from TMM 40081 otherwise only includes undiagnostic, fragmentary specimens of equids and artiodactyls. TMM 40081-14 is a fragment of a thoracic vertebra of a large proboscidean (Figure 8). It may represent the earliest record of a proboscidean from the Miocene of the Texas Coastal Plain. Unfortunately, uncertainty about stratigraphic location and taxonomic identification limit the significance of this specimen. Based on size alone, it appears to be larger than would be expected for the small gomphotheres, that are described below, from Ba1 faunas in this region.

Early Barstovian Proboscidea

Noble Farm Local Fauna. A gomphothere from the Noble Farm L.F. was described by Hay (1925) based on a collection of “teeth and tusks” that was sent to him by Dr. Mark Francis at Texas A&M. They had been collected in 1915 at the farm of Edward Noble which was located 5 miles southeast of Navasota, Texas. No specimen numbers were included with Hay’s description; however, the figures can be used to identify specimens now curated at the Texas Vertebrate Paleontology Collections (Table 2).

As described by Hay (1925), the proboscidean from Noble Farm includes down-curved upper tusks with an enamel band and no spiral. The tusks are slightly oval in cross-section. He hypothesized that at least four individuals were represented by the tusk fragments. At least one was a young individual represented by two lower tusks (TMM 40775-48.1 and 48.2) that are interpreted as deciduous and exhibit longitudinal striations with no spiral (Figure 9). Although broken, there is no evidence of a growth cone as seen in the ever-growing permanent tusks (lower and upper). Neither is there evidence of an enamel band, although both exhibit a wear facet along the distal end. The larger of the two juvenile lower tusks is ~90 mm long and both are slightly oval in cross-section with dimensions of 19.5 mm x 15.3 and 20.2 mm x 15.5 mm.

TMM 40775-59 is a partial left upper tusk 488 mm long with a proximal cross-section of 61 mm x 50 mm and a distal cross-section of 48 mm x 41 mm (Figure 10A, B). The tusk is down-curved with an enamel band on the labial side. The enamel band is 31.5 mm wide at the distal end of the specimen and 25 mm wide at the proximal end, where it interfingers with cementum. Although the most anterior portion of the tusk is missing, a large wear facet is present on the ventro-lingual side of the tusk. That wear facet is sub-oval in shape and extends to the ventral edge of the enamel band. A relatively sharp cutting edge is preserved at the boundary between the wear facet and the enamel band. The preserved wear facet is 196 mm long and 36 mm wide (Figure 10A).

TMM 40775-2121 includes a nearly complete right upper tusk and part of the right premaxillary bone (Figure 10C, D). The distal part of this tusk (figured by Hay, 1925) was previously removed and incorporated into a 1939 TAM museum mount of a partial gomphothere skull from the Ba2 Cold Spring L.F. It was recently removed from that mount and returned to the material from Noble Farm. When reconstructed, the length of the tusk is ~517 mm. It is down-curved with a labial enamel band that is ~28 mm wide and exhibits inter-fingering with cementum at the proximal end. The distal tip of the tusk is rounded from wear and the enamel band shows clear signs of having been broken into a saw-blade pattern during the lifetime of the gomphothere (Figure 10B). The broken edge of the enamel band experienced post-damage wear pre-mortem. Whether this damage was the result of fighting, feeding, or some other behavior is unknown.

A linear feature in the premaxilla was interpreted by Hay (1925) as the suture between the premaxilla and maxilla (his fig. 4, page 11), but he apparently did not recognize that TAM 2213 was part of this specimen. Not only is it part of the original specimen, TAM 2213 preserves the termination of the feature suggesting it may be a taphonomic crack associated with desiccation prior to burial (Figure 10D). Additional desiccation cracks, subparallel to this feature, are present. The cross-sectional dimensions of the tusk at the broken anterior end are 60 mm x 45 mm.

The only specimens of adult lower tusks are TMM 40775-51 and TMM 40775-56 that are here interpreted as being from the same individual. They are associated with a partial symphysis (Figure 11). The anterior cross-section of TMM 40775-51 is sub-oval in shape with dimensions of 47 mm x 32 mm and shows tightly packed laminations with dense crosshatching (Figure 11C). The mandibular tusks are placed closely together with only 5 mm separating them at the distal end of the preserved portion of the symphysis. The cross-sectional shape of the lower tusks differs from the pyriform lower tusks seen in forms like Gomphotherium angustidens and Gomphotherium subtapiroideum. Only the proximal ends of these lower tusks are preserved, so the geometry of the distal ends is unknown. The distal ends of the juvenile tusks (TMM 40775-48.1, TMM 40775-48.2) are sub-oval in cross-section.

There are no enamel bands on the lower tusks although a 1-2 mm thick layer of cementum covers the underlying dentin. Shallow longitudinal ridges are present in the dentin. The lower tusks are ever-growing, as evidenced by the growth cones apparent in the pulp cavity at the proximal end. The partial symphysis is approximately 50 mm wide and slightly concave on the dorsal surface (Figure 11A). The fragment of symphysis is not sufficient to directly determine the length or morphology of the mandible, but the size and geometry of the tusks suggests that the gomphothere from Noble Farm was likely a longirostrine gomphothere.

TMM 40775-49 is a worn RM2, 105.1 mm long by 65.7 mm in maximum width, with three lophs, a well-developed lingual cingulum and simple pretrite trefoils (Figure 12). Measured widths at each loph include first: 65.5 mm, second: 65.7 mm, and third: 65.3 mm. The crown height, at the slightly worn third loph, is 43.8 mm. An antero-labial and a posterior cingulum are weakly developed. Pretrite valleys are relatively wide, while postrite valleys are narrow. The median sulcus is weakly developed and minor amounts of cementum are present in transverse valleys.

TMM 40775-46 is a Lm3 with three main lophids, a smaller fourth lophid and a small talonid (Figure 13). The tooth is slightly worn (young adult) but would develop simple pretrite trefoils on the three anterior lophids with additional wear. It has a maximum length x width of 136.6 mm x 64.0 mm (at second lophid) with a crown height of 44.7 mm at the second lophid and 42.7 mm at the third. The m3 lacks a well-defined lateral cingulum although a weak antero-labial cingulum is present. The median sulcus is well developed. Transverse valleys are relatively tight and lack secondary lophids. Minor amounts of cementum are present in the talonid and between the third and fourth lophids.

TMM 40775-47 is a RM3 that is slightly worn and exhibits strong lingual and antero-labial cingula (Figure 14). There are three main lophs and a talon consisting of five cones. The main lophs would develop simple pretrite trefoils with additional wear. This tooth is 136.6 mm x 70.4 mm, with a crown height at the second loph of 48.7 mm and 43.9 mm at the third. The median sulcus is relatively shallow, and the transverse valleys are relatively narrow. An accessory cone is present on the postero-labial side of the third loph.

Figure 15 illustrates the enamel pattern from an arbitrary slice (white line in A and D) through the CT volumes for TMM 40775-46 and TMM 40775-47. These images illustrate the simple trefoil pattern characteristic of gomphotheriids that would be observed in these specimens with additional wear (labial in m3 and lingual in M3).

The third molars of the Noble Farm gomphothere are relatively small and plot near the lower end of all Gomphotherium taxa sampled by Wang et al. (2017) and are much smaller than specimens identified as Gomphotherium productum. They are also significantly shorter than measurements compiled by Madden and Storer (1985) for M3 of Gomphotherium obscurum, as well as being smaller and lower crowned than the M3 of Gomphotherium osborni (190 mm x 87 mm) (Madden and Storer, 1985). Similarly sized taxa, based on the data in Wang et al. (2017), include Gomphotherium inopinatum, Gomphotherium angustidens, Gomphotherium mongoliense, Gomphotherium subtapiroideum, and Gomphotherium annectens. M2 (TMM 40775-49) is similar morphologically to Gomphotherium subtapiroideum including the loph and valley geometry emphasized by Wang et al. (2017).

Two small cheek teeth were interpreted by Hay (1925) as L and R Dp3. These teeth are only slightly worn with three lophids and a small posterior heel (Figure 16). Based on the analysis of dental ages within a large sample of Gomphotherium angustidens by Tassy (2013), TMM 40775-50.1 and 50.2 are now interpreted as L and R Dp4. TMM 40775-50.1 is an L Dp4 with maximum length of 59.4 mm, width at the anterior lophid of 27.4 mm, and width at posterior lophid of 35.8 mm. TMM 40775-50.2 is essentially the same size, although the anterior portion is broken. When compared to TMM40418-10, from the Ba1 Brenham L.F., these teeth are consistent with identification as Dp4. Dp3 and Dp4 are replaced by p3 and p4 while m1 is in early occlusion in Gomphotherium angustidens (Tassy, 2013).

Additional cheek teeth, apparently associated with the Noble Farm gomphothere, include R and L p3 as interpreted by Hay (1925, his plate 3 and figs. 2, 3). These specimens are unworn and include caps with no roots (Figure 17). Their assignment as permanent premolars is uncertain. Length x width measurements for these two teeth are 28.5 mm x 20.1 mm and 26.2 x 18.9 mm. An additional juvenile cheek tooth (TMM 40775-61) contains no TAM number and was not included by Hay in his 1925 paper. The provenance of this specimen is uncertain, but it may represent a premolar (P/p2?).

The proboscidean from the Noble Farm L.F. belongs to Gomphotheriinae based on the presence of accessory conules on the pretrite half-lophids forming trefoil patterns, a relatively narrow median sulcus, trilophodont M2/m2, incipient tetralophodonty on M3/m3, down-turned upper tusks with enamel bands, and lower tusks without enamel bands. The specimens were identified by Hay (1925) as “Cope’s form” Tetrabelodon serridens cimarronis which he renamed Gomphotherium cimarronis (Cope). A single tooth was assigned by Cope (1893) as Tetrabelodon serridens?, although he stated significant doubt as to that identification because of the small size. In fact, he said that the small size “renders the identification doubtful, unless it indicates a small race of the same species.” “It may pertain to a different species, but for the present I refer it to a small race of the T. serridens, under the name of T. serridens cimarronis.” This cautious attempt to recognize a subspecies (Cope’s “race”) was elevated to species status by Hay (1925). The “type” specimen of Tetrabelodon serridens cimarronis (no specimen number and no explicit holotype designation) was described by Cope (1893) as a left m3 from the “Loup Fork beds north of the South Fork of the Red River.” Hay (1925) disagreed and correctly interpreted the tooth as an M3. The specimen is curated in the Texas Vertebrate Paleontology Collection at the University of Texas, Austin (TMM 40282-12). It is included with several specimens from the Clarendonian Stanton Ranch locality (TMM 40282) near Clarendon, Texas. Gomphotherium cimarronis was interpreted by Tobien (1973) as a junior synonym of Gomphotherium productum.

In their review of North American occurrences of Gomphotherium, Lambert and Shoshani (1998) only recognized Gomphotherium angustidens and Gomphotherium productum, but also listed many previously named species under Gomphotherium sp. They include Gomphotherium cimarronis in the later group suggesting uncertain validity for the taxon. Regardless of the validity of the name Gomphotherium cimarronis, the “type” specimen of the taxon from the Stanton Ranch L.F. is clearly different from the gomphothere from Noble Farm (Figure 18). TMM 40282-12 from Stanton Ranch is 76.6 mm wide at the first loph, 70.0 mm wide at the second, and at least 58 mm wide at the third (now partially broken, even though it was shown as complete by Cope (1893)). TMM 40775-47 (RM3) is significantly narrower at the first loph (69.9 mm). The crown height of the “holotype” of Gomphotherium cimarronis is much taller: 65.7 mm at the first loph and 62.7 mm at the second, compared to 43.9 mm and 48.7 mm for the same lophs on the gomphothere from Noble Farm. The talon is also different being composed of only three conules, instead of the five to six seen in the specimen from Noble Farm. The lingual (pretrite) cingulum is also weaker in Gomphotherium cimarronis, although it is partially missing now along the posterior part of the tooth. Although specimens from both the Stanton Ranch and Noble Farm Local Faunas represent a relatively small gomphothere with simple pretrite trefoils, they are not the same taxon. Furthermore, Gomphotherium cimarronis, from the Clarendonian of the Texas panhandle, is considered nomen dubium because the “holotype” is a partial M2 lacking diagnostic characters.

Based on L x W measurements for M2, M3, and m3, the gomphothere from Noble Farm plots near the center of the distribution of sizes compiled by Sanders and Miller (2002) for Gomphotherium angustidens. However, significant sexual dimorphism within M3 length for Gomphotherium angustidens was documented by Tassy (2014), and the gomphothere from Noble Farm plots near the low end of the males. Unlike Gomphotherium angustidens, the upper tusks show no evidence of the spiral architecture described by Tassy (2014), and the lower tusks appear to be straighter and sub-oval in cross-section, rather than pyriform. Differences in tusk morphology, with respect to both Gomphotherium productum and Gomphotherium angustidens were also discussed by Hay (1925). The M3 of Gomphotherium agustidens has four lophs while the gomphothere from Noble Farm has only three main lophs and an incipient fourth loph with a posterior heel. Central conules are also less well developed in the gomphothere from Noble Farm.

The gomphothere from Noble Farm differs from Gomphotherium angustidens, Gomphotherium productum, and from the “holotype” of Gomphotherium cimarronis. The cheek tooth morphology is similar to Gomphotherium subtapiroideum, while the mandibular tusks are similar to Gomphotherium productum. Similarity to Gomphotherium subtapiroideum (Mastodon angustidens subtapiroidea, Schlesinger, 1917) was noted by Hay (1925), but Wang et al. (2020) have described the lower tusks of Gomphotherium subtapiroideum as pyriform, rather than sub oval as seen in the gomphothere from Noble Farm (Figure 11). They interpret pyriform lower tusks to be a plesiomorphic character for elephantimorphs with rounded or oval tusks being a derived morphology. The gomphothere from Noble Farm is a relatively primitive morphotype that is smaller and lower crowned than Gomphotherium productum and differs in dental morphology from Gomphotherium angustidens. It is possible that the gomphothere from Noble Farm is a primitive morphotype of Gomphotherium productum, but it is here referred to as Gomphotherium. sp. Although representing a similarly conservative morphotype, it differs in detail from the gomphothere in the Ba1 Belts Creek L.F. described below.

Brenham Local Fauna. TMM 40418-10 is a partial left dentary of a juvenile gomphothere with a deciduous premolar and a matrix-filled alveolus for a deciduous lower tusk (Figure 19B). The tooth is a slightly worn Dp3 with a maximum length of 42.6 mm. There are two primary lophids, an incipient anterior lophid, and a small posterior heel. The tooth is 29.0 mm wide at the second lophid and 23.8 mm wide at the first lophid. The median sulcus is moderately well developed and a small accessory conulid is present in the labial side of the transverse valley between the lophids. A relatively large mental foramen is located on the labial side of the dentary, just anterior to Dp3. A sub-circular alveolus for a deciduous lower tusk is approximately 17 mm in diameter. This specimen compares favorably with Gomphotherium sp. from the Noble Farm L.F., extends the geographic range of the taxon and supports the Ba1 faunal interpretation (Figure 20). TMM 40418-10 was previously misidentified as Dinohyus.

Belts Creek Local Fauna. A gomphotheriid proboscidean from the Belts Creek L.F. is represented by a single cheek tooth (Figure 21). TMM 40622-25 is a moderately worn Rm3 with an occlusal pattern that is characterized by simple pretrite trefoils, with three main lophids, a smaller fourth lophid and a small posterior conulid. The maximum length of this tooth is 150 mm and the width varies from 69.1 mm at the first lophid, to 76.0 mm at the second, and 76.2 mm at the third lophid. The fourth lophid is 55.6 mm wide, and the posterior heel is a single conulid. A minimum crown height of ~44.3 mm was measured on the lingual side of the slightly worn third lophid. The anterior cingulum is worn nearly flat and accessory conulids are present only on the lingual (postrite) margin between lophids. A weak median sulcus is present and accessory conulids are absent from the transverse valleys. Postrite valleys are relatively broad at their base while pretrite valleys are narrow. There is no lingual cingulum and no cementum between the lophids.

Morphologically, the Belts Creek proboscidean is assignable to the genus Gomphotherium and the bunodont, pretrite trefoil morphology distinguishes TMM 40622-25 from Zygolophodon. However, species-level identification becomes difficult partly because it is a single tooth and partly because of the current state of the taxonomy of the genus. Eurasian species of Gomphotherium were reviewed by Wang et al. (2017), and, based on their compilations, the Gomphotherium from Belts Creek compares well with the small- to moderate-sized taxa with m3 dimensions similar to Gomphotherium inopinatum, Gomphotherium subtapiroideum, Gomphotherium angustidens, Gomphotherium cooperi, and Gomphotherium sylvaticum. In terms of size, crown height and occlusal morphology, TMM 40622-25 appears most similar to Gomphotherium inopinatum figured by Wang (2014; IVPP V 18700). The fourth lophid on TMM 40622-25 is slightly better developed and the posterior conulid appears absent on Gomphotherium inopinatum. TMM 40622-25 also has postrite conulids between the lophids, a character that appears absent in Gomphotherium inopinatum. Gomphotherium inopinatum is known from the Linxia Basin in central China, from the lower part of the Dongxiang Formation. According to Wang (2014), this formation is late Early Miocene in age or approximately equivalent to the Hemingfordian NALMA. Gomphotherium inopinatum was interpreted as a basal species of Gomphotherium and possibly ancestral to Gomphotherium angustidens Wang (2014). The type locality of Gomphotherium inopinatum is from the Turgai region in Kazakhstan and was considered to be Lower Miocene by (Tassy, 1985). In their phylogenetic analysis of Gomphotherium, Wang et al. (2017) recovered Gomphotherium inopinatum as a basal member of the “G. angustidens group” with an age range of approximately 15-17 Ma. The morphological differences between the gomphothere from Belts Creek and Gomphotherium inopinatum suggest a slightly more derived morphotype in the Texas form.

Although known from a single molar, the gomphothere from the Belts Creek L.F. represents a conservative species of Gomphotherium and is herein identified as Gomphotherium cf. Gomphotherium inopinatum. This suggests that a species of Gomphotherium, like Gomphotherium inopinatum, immigrated from Eurasia to North America, probably during the Hemingfordian, arriving in Texas by the Early Barstovian. Gomphotherium cf. Gomphotherium inopinatum from the Belts Creek L.F. is morphologically distinct from the Gomphotherium sp. from Noble Farm described above. The third lower molar is longer (150 mm vs. 136.6 mm), has a weaker median sulcus and a less complex trefoil pattern. The posterior heel is simpler (single conulid), and the postrite valleys are wider at the base with better developed accessory conulids. At least two different species of Gomphotherium are present in Ba1 faunas of the Texas Coastal Plain.

Discussion. There is significant confusion in the literature regarding interpretation of Early Barstovian proboscideans from the Miocene Fleming Group localities of the Texas Coastal Plain. Specimens from the Burkeville L.F. were cited by Tedford et al. (2004) as a possible Ba1 proboscidean occurrence, but Prothero et al. (2006) stated “Only the High Plains and the Texas Gulf Coastal Plain show no evidence of early Barstovian proboscideans,...” Previous authors misinterpreted the ages of the Noble Farm, Belts Creek, and Brenham Local Faunas, but all of these are here shown to be Ba1 based on stratigraphic position and faunal evidence. It is now clear that at least two different species of Gomphotherium are known from Ba1 local faunas in the lower to middle Lagarto Formation. Gomphotherium cf. Gomphotherium inopinatum is known from a single m3 from the Belts Creek L.F. while Gomphotherium sp. is known from multiple specimens from the Noble Farm and Brenham Local Faunas. Gomphotherium sp. from Noble Farm should not be referred to Gomphotherium cimarronis as suggested by Hay (1925). The Ba1 gomphotheres from the Lagarto Formation are all relatively small with primitive dental morphology and probably represent early immigrants from Asia.

Late Barstovian Proboscidea

Fossil proboscidea are known from the Cold Spring (including Sam Houston) and Chappell Hill local faunas from the upper part of the Lagarto Formation.

Cold Spring Fauna. Trilophodon sp. was included in faunal lists by Hesse (1942) for both the “Sam Houston” and “Cold Spring fauna”. As discussed above, the “Sam Houston fauna” is now considered a locality within the Cold Spring L.F. Material cited by Hesse (1942) in the faunal list for the ‘Cold Spring Fauna’ included “ Trilophodon, fair skull in A&M Museum”. This description undoubtedly refers to TMM 38-3046 that was mounted for display at TAM in 1939. TMM 38-3046 includes fragments of both a partial skull and a partial left dentary, however, the amount of plaster reconstruction limits the ability to provide a useful description (Figure 22, Figure 23). The upper tusk, incorporated into the mount, is TMM 40775-2121 from the Ba1 Noble Farm L.F. and does not belong to the mounted skull. Other than the fragments of tusk associated with TMM 38-3046, there are no known upper tusks from gomphotheres found in any of the Ba2 local faunas of the upper Lagarto Formation. The tooth associated with the partial skull identified as an M3 by Tobien (1973), is largely reconstructed from plaster and has a coating of plaster covering the occlusal surface.

The anterior portion of the skull (premaxilla, maxilla, frontals) includes some original bone, although much of the skull is either sculpted from plaster or coated with plaster and paint. The specific dimensions are of limited value, but the length of the “skull”, from the broken edge of the premaxilla to the posterior edge of the cranium, is 55.5 cm. The height, from the ventral side of the zygomatic arch to the dorsal side of the cranium, is 38.5 cm. The height, from the top of the tooth row to the top of the premaxilla, is 14 cm. The length of the zygomatic arch is approximately 32 cm. If the reconstruction honored at least some original bone morphology, this is a small gomphothere and similar in size to TMM 31219-237 and to the skull from the Ba2 locality near Chappell Hill, TX (TMM 31272-42).

The partial left dentary mounted with TMM 31219-3046 has no visible specimen number. Handwritten notes state “left side of skull and left mandible”, collected “1.3 miles north of the court house, Cold Spring, San Jacinto Co., Texas” in 1934 by Claude Riley. The worn lower m3 incorporated into the dentary only includes a small amount of original tooth (lingual margin) and is mostly modeled in plaster. The occlusal pattern is of no value and the tooth appears to be attached to the dentary with plaster. The lower tusk, associated with the dentary, also appears to be set in plaster at the alveolus with no actual fit and therefore is of uncertain provenance. This tusk is somewhat oval in cross-section (38 mm x 32 mm) at the broken proximal end (Figure 24). The total length of the tusk fragment is ~280 mm and is relatively straight with no enamel band. It is significantly smaller than the lower tusk of Gomphotherium sp. from the Sam Houston locality (TMM 31191-2233) (53 mm x 47 mm cross-section diameters and 416 mm length). The latter appears less compressed in cross-section, but the significance of these differences is uncertain and could easily reflect ontogeny or sexual dimorphism. Neither specimen exhibits growth cones, suggesting the proximal ends are missing. The sub-oval cross-section of the mandibular tusks differs from the dorso-ventrally flattened lower tusks seen in the holotype of Gomphotherium productum from New Mexico (Morgan 2015).

The presence of cement between lophs on the M3 of TMM 38-3046 was noted as a distinctive and important character by Tobien (1973). However, the occlusal surface of this heavily worn and partially reconstructed tooth was coated in plaster for mounting in 1939. It is not possible to observe cement as described by Tobien (1973). The description of TMM 38-3046 in Tobien (1973) is misleading and the species of Gomphotherium represented by this specimen is unclear. A significant re-preparation effort might reveal characters that could be used for more confident identification. Based on the actual bone observable at present, all that can be said is that the specimen represents a small gomphothere, possibly with peg-like lower tusks.

Tobien (1973) incorrectly stated that “Hay’s specimens” (referring to “Gomphotherium cimarronis” from Noble Farm) “belong to the Coldspring fauna.” As described elsewhere in this paper, the Noble Farm L.F. is clearly from the lower part of the Lagarto Formation and is a Ba1 fauna while the Cold Spring L.F. is from the upper part of the Lagarto Formation and is a Ba2 fauna. Perhaps influenced by the statement in Tobien (1973), MacFadden et al. (2015) also stated that Gomphotherium cimarronis (Hay 1925) was “from the Cold Spring Fauna of the Texas Gulf Coastal Plain.”

Additional specimens of proboscideans are known from the Cold Spring L.F. TMM 31219-44 is an unworn Lm1 with a maximum length of 83.0 mm and a maximum width of 49.8 mm (Figure 25A, B). There are three lophids, a posterior heel, and both anterior and pretrite cingula. This tooth would develop pretrite trefoils with wear. There is no cement on this tooth. TMM 31219-177 is an unworn premolar with a maximum length of 35.4 mm and width of 29.8 mm (Figure 25C). The tooth displays two primary lophids with a small anterior, posterior, and pretrite cingula. The morphology of this tooth compares favorably with other permanent premolars recognized in some gomphotheres.

TMM 31219-144 is a partial left maxilla of a juvenile Gomphotherium sp. from the Cold Spring L.F. (Figure 26). The specimen includes partial roots of DP2, a nearly complete DP3, and a partial DP4. The DP3 is 37.6 mm in length and 31.6 mm in maximum width with a subtle pretrite trefoil pattern.

TMM 31219-237 (incorrectly cited by Tobien [1973, p. 218] as 21219-237) was described by Tobien as a partial skull with “deeply worn right M3, min length 116 mm.” This partial skull from the Cold Spring L.F. represents a mature individual (Tassy [2013] age class XXIII). The worn RM3 has enamel only around the posterior loph and heel. Alveoli are present for RM2 and for LM2-M3. The ventral portion of the skull is relatively complete but was locally restored with plaster. The palate and basicranium are well preserved, as are the occipital condyles (Figure 27).

Dimensions, following Tassy (2013), are shown in Table 3. Both the pattern of variation and the absolute values of TMM 31219-237 are similar to SEP185 that Tassy (2013) measured for age class XXI of Gomphotherium angustidens. The palatine foramina are large and elongated antero-posteriorly. The occipital condyles are relatively rounded with a maximum width of 16.2 cm with an oval-shaped foramen magnum. The dorsal portion of the skull is missing, as are the tusks. The width of the skull, as well as the length and width of the basicranium, are significantly larger in Gomphotherium sp. from the Cold Spring L.F. (TMM 31219-237). The single worn M3 is of little use for identification. This specimen supports the observation that the gomphotheres from the Ba2 faunas in the upper Lagarto Formation were generally similar in size to Gomphotherium angustidens, but with a broader skull. The gomphothere from the Cold Spring L.F. was significantly smaller than the Early Clarendonian cf. Gomphotherium. sp. from the Lapara Creek Fauna (May, 2019).

Additional specimens of Gomphotherium sp. are known from the Sam Houston locality within the Cold Spring L.F. (TMM 31191). TMM31191-1 is an Rm3 with four main lophids exhibiting pretrite trefoils. A well-developed posterior heel includes three main conulids and a small posterior conulid (Figure 28). There is no lingual cingulum, and the median sulcus is poorly developed. A lingual conulid is present between both the first and second lophids. The m3 is 152 mm in length. The first and second lophids are partially broken, but the minimum width at the first lophid is 65 mm. The widths at the third and fourth lophids are 65.7 mm and 60.4 mm, respectively. The crown height, on the lingual margin of the third and fourth lophids, is 42.7 mm and 38.6 mm, respectively. TMM 31191-1 is significantly larger than Gomphotherium sp. from Noble Farm (TMM 40775-46) and the fourth lophid, the posterior heel and accessory conulids are all better developed (Figure 28). These characters are all similar to Gomphotherium angustidens and TMM 31191-1 clearly represents a larger and more derived species of Gomphotherium in the Ba2 Cold Spring L.F. compared to the Ba1 Noble Farm L.F. Although similar in size and crown height, it differs from the holotype of Gomphotherium productum from Late Barstovian faunas of New Mexico, in having four lophids with a well-developed posterior heel.

Chappell Hill Local Fauna. TMM 31272-42 is another partial skull of a relatively small gomphothere with both left and right M3 (Figure 29). This skull was collected by J. Quinn in 1953 from a roadcut on Hwy 290 near Chappell Hill, Texas, and is from the upper Lagarto Formation.

The left zygomatic arch on TMM 31272-42 is partially preserved, with some plaster restoration. The top of the skull is crushed, and the left anterior portion is missing. The left side of the skull is imbedded in a plaster cradle. The length of the preserved skull, from the occipital condyles to the anterior part of the premaxilla, is 68.5 cm and the length of the zygomatic arch is approximately 35 cm. The height of the preserved skull, from the frontals to the dorsal edge of the tooth row, is 34.5 cm. The width of the occipital condyles is approximately 160 mm, although they are not well preserved. The posterior portion of the skull is crushed and not well preserved. The worn LM3 exhibits simple pretrite trefoils, a lingual cingulum, is 147 mm in length, with widths of 75 mm at the first loph, 71 mm at the second, 64 mm at the third and 46 mm at the fourth. The fourth loph is relatively well developed with a small posterior heel. There is a small amount of cement in the lingual transverse valley between the second and third lophs. No cement is obvious on the RM3. The LM2 is extremely worn with only partial second and third lophs preserved. The LM3 is significantly longer than Gomphotherium sp. from the Noble Farm L.F. with a better developed fourth loph. It is not possible to measure crown height due to the worn nature of the tooth.

Trilophodon sp. was included by Hesse (1942) in his faunal list for the “Goodrich fauna” (TMM 31183) located south of Livingston in Polk County, Texas, and considered by Tedford et al. (2004) to be a late Ba2 fauna. The specimen was apparently collected approximately 0.8 km northeast of the main locality from near the top of the exposed Fleming Group (Hesse, 1942). Both localities are referred to as Smith Ranch in the TxVP collections, but the locality that included a proboscidean, according to Hesse (1942), is now labeled TMM 31200. Although 13 specimens are known from TMM 31200 (Hesse, 1942 noted 12), none are proboscidean.

Discussion. Ba2 local faunas, from the upper Lagarto Formation, include at least one additional species of Gomphotherium that shares characteristics with Gomphotherium angustidens and Gomphotherium productum. The small gomphotheres from Texas with “peg-like lower tusks” were interpreted by Tobien (1973) to be similar to Gomphotherium angustidens but chose to synonymize all of the “Texas material” as Gomphotherium productum. Whereas Gomphotherium productum has down-curved upper tusks with a straight enamel band, those of Gomphotherium angustidens have a spiral architecture (Tassy, 2014). The lack of upper tusks from the Ba2 gomphothere from the upper Lagarto Formation precludes comparison of this distinguishing character. Acknowledging both the limited number of specimens, and the need for a complete taxonomic revision of the Gomphotheriidae, the group of small gomphotheres from Ba2 faunas of the Texas Coastal Plain are here identified as Gomphotherium sp. Whether or not this clade of gomphotheres was derived from the more primitive Ba1 taxa in this region, or represents a separate immigration event from Eurasia, is unclear given the limited amount of material presently available.

Two proboscideans were described from the Early Clarendonian Lapara Creek Fauna in the Goliad Formation that overlies the Fleming Group (May, 2019). The Early Clarendonian Ten Mile Waterhole Creek and Bridge Ranch Local Faunas both contain material that May (2019) assigned to cf. Gomphotherium sp. The specimens include partial upper tusks with enamel bands, isolated cheek teeth, and a segment of a pyriform lower tusk. The cheek teeth include three third molars that range in size from 150-170 mm x 66-75 mm with four main lophs and a posterior heel (May, 2019). These teeth are generally longer and more hypsodont than Gomphotherium sp. from the Ba2 faunas of the upper Lagarto Formation described above. A second molar (TMM 30936-331) exhibits a relatively zygodont morphology (see fig. 70 of May, 2019). Although material is limited, the lower tusk and cheek teeth of cf. Gomphotherium sp. from Ten Mile Waterhole Creek and Bridge Ranch, are more derived than the Ba2 gomphotheres from the Lagarto Formation. This proboscidean is the only Miocene gomphotheriid from the Texas Coastal Plain with pyriform lower tusks. Gomphotherium is absent in the Cl2 Farrish Ranch and Buckner Ranch Local Faunas of the Goliad Formation where it is replaced by Blancotherium buckneri (May, 2019). Blancotherium buckneri is a longirostrine gomphothere that is especially abundant at the Buckner Ranch site TMM 30896. A large amount of both cranial and post-cranial material from this locality makes Blancotherium buckneri the best known gomphotheriid from the Miocene of the Texas Coastal Plain. Numerous additional field jackets, clearly representing a large proboscidean, are curated at TxVP awaiting preparation and almost certainly represent additional specimens of Blancotherium buckneri.

Blancotherium buckneri exhibits clear gomphotheriid affinity with respect to cheek tooth morphology, hyoid morphology, mandibular morphology including a long, shoe-horn shaped symphysis, and lower tusks (May, 2019). Unlike most other gomphotheres, Blancotherium buckneri has large, upwardly curving upper tusks that lack enamel bands. Although several mandibles from the Buckner Ranch locality possess alveoli for lower tusks, no lower tusks have been found within dentaries. Two, relatively small, isolated lower tusks with enamel bands were described by May (2019), but it appears that lower tusks were commonly shed from the mandible prior to maturity. Blancotherium represents a stage of longirostrine gomphotheriid evolution associated with the loss of lower tusks. This corresponds to the development of much larger, upward-curving, upper tusks that lacked enamel bands and show no consistent wear facets as seen in proboscideans from the Barstovian faunas. Ontogenetic loss of lower tusks was also described by Mothé et al. (2016) for the gomphotheriid Cuvieronius.

The phylogenetic relationship between Blancotherium and Gomphotherium is uncertain, although Blancotherium is clearly a more derived gomphotheriid with large, up-turned, upper tusks and apparent ontogenetic loss of lower tusks. Blancotherium is significantly different from all older gomphotheres of the Texas Coastal Plain and appears to be an immigrant that replaced Gomphotherium sp. during the Early Clarendonian. Blancotherium is currently only known from the Lapara Creek Fauna.

All Miocene and Pliocene proboscideans from the Texas panhandle have been referred to Rhynchotherium, Amebelodon, or Stegomastodon (e.g., Hemphillian: Higgins, Box T Local Faunas, Amebelodon (Schultz, 1990), Coffee Ranch, Rhynchotherium (?) sp. (Schultz, 1990); Blancan: Cita Canyon Local Fauna, Stegomastodon mirificus (Savage, 1955), Mt. Blanco Rhynchotherium praecursor, Stegomastodon mirificus (Dalquest, 1975), Red Corral S. mirificus (Schultz, 2016).