sign in sign up
<<
Home , Amphitherium, Dorsetisaurus, Dryolestes, Dryolestida, Dryolestidae, Eozostrodon

NEW MAMMIALS FROM , WYOMING, AND THE INTERRELATIONSHIPS OF NON-TRIBOSPHENIC

DONALD R. PROTHERO

BULLETIN OF THE AMERICAN MUSEUM OF NATURAL HISTORY VOLUME 167: ARTICLE 5 NEW YORK: 1981 7~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

34 v,~~~~~~~~~~~~~~~~~

"AT~~~ ~ ~ ~

4K~~~~~~~~~~~~~~~4

4 ' j4'~~"T'4"444 4 ~~~~Z '4'''44444444~ 44~ *4~~4.4.4"~ " '- -,44~'4~':J"'-.'4 '~ " '" 4 4'', 4 ;q~W'7'44f~~~44~~~~~~~~444.4''"4~ ~ ~ '4'4"'~~~~~~~~~~~~~~' '4 '~~~~~~~~~~4''4,'7 ~ 4 ~ '.4 '4 4444.~ .444?,k4~4.4~X'4 '4~~~~/4:~~~~~.~~~'44444444.,~~~~~~~~~~ ~~ ~ ~ ~ ~~ 4"'' "'4 4" .~~~~~~~~~~~~~.. ' 4,. ~ ~ ~ ~ *44~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~N~4';i.44444'~

* g4 " ' 'A.~ ~~~" 74:'4'14~4 ' 7 -44-7p""'''4.'''444.44 *44~~~~~~~4.4...4'44 '~~~~~~~~~~~~~~'4 y,~'~ f ' ~ n #44~ 4 4~.44444'4 " 4 44',~~~~~~~~~~~4'~~~~~~~ 4. 4~~~~~~~~~~~~~~~~"4 7~~~~~~~~~'4 '4"''''4'''

4.''--. 4

~~~~~~~ "'<''~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~.4"

44444 "'.4 44V..

~~'.'~~~4'.'.~~~~44"."4.. 4 4~~- '4' .4 4'' 4 4444 4' 4''~~~~~~~4~ ~ NEW JURASSIC FROM COMO BLUFF, WYOMING, AND THE INTERRELATIONSHIPS OF NON-TRIBOSPHENIC THERIA

DONALD R. PROTHERO Graduate Student, Department of Geological Sciences Columbia University

BULLETIN OF THE AMERICAN MUSEUM OF NATURAL HISTORY VOLUME 167: ARTICLE 5 NEW YORK: 1981 BULLETIN OF THE AMERICAN MUSEUM OF NATURAL HISTORY Volume 167, article 5, pages 277-326, figures 1-12, tables 1-9 Issued February 26, 1981 Price: $3.20 a copy

ISSN 0003-0090

Copyright ©) American Museum of Natural History 1981 CONTENTS

Abstract ...... 281 Introduction ...... 281 Abbreviations ...... 282 Acknowledgments ...... 282 Stratigraphy and Localities ...... 283 Systematic Descriptions ...... 286 Taxa Known from Upper Jaws ...... 286 Comotherium, New ...... 286 Melanodon and Herpetairus ...... 289 Melanodon Simpson, 1927 ...... 290 Herpetairus Simpson, 1927 ...... 292 Taxa Known from Lower Jaws ...... 293 Amblotherium Owen, 1871 ...... 293 Marsh, 1878 ...... 293 Laolestes Simpson, 1927 ...... 295 Kepolestes Simpson, 1927 ...... 297 Discussion ...... 297 General Characters ...... 299 Lower Jaw Characters ...... 301 Lower Tooth Characters ...... 305 Upper Tooth Characters ...... 309 Cladogram Discussion ...... 312 Symmetrodont Monophyly ...... 313 Position of Amphidontids ...... 317 Position of ...... 319 Classification ...... 320 Literature Cited ...... 323

ABSTRACT A joint American Museum of Natural History- trary to the condition reported in Kimmeridgian Yale University expedition reopened Quarry Nine Portuguese dryolestids. at Como Bluff, Wyoming, in 1968-1970. This lo- The systematics of primitive Theria are re- cality had produced all but six of about 250 Ju- viewed, and an hypothesis of relationships for the rassic specimens known from North non-tribosphenic Theria is proposed. This hy- America prior to 1968. During the renewed col- pothesis corroborates the hypothesis of McKenna lecting, four additional prospects were found in (1975). It is concluded that the symmetrodonts are the vicinity, and 25 fossil mammal jaws and nu- a monophyletic group, and that Amphitherium is merous teeth were collected. most closely related to the dryolestoids. A cladis- A new genus and of dryolestid, Com- tic classification of the non-tribosphenic Theria is otherium richi, is described. The dryolestid gen- proposed. Six new taxa are named: sublegions era Melanodon and Herpetairus are reviewed and Amphidontoidea and Spalacotherioidea, and in- rediagnosed. The anterior dentition of Melanodon fraclasses Tinodontida, Spalacotheriida, Am- goodrichi and better material of Herpetairus hu- phitheriida, and . It is strongly urged milis are described. The crown and labial views that the term "pantothere" be abandoned, be- of Kepolestes, previously obscured in matrix, are cause it has misleading phylogenetic connotations described. No dryolestid jaw in this collection and because it obscures non-tribosphenic therian shows any trace of a coronoid or splenial, con- diversity.

INTRODUCTION The history of the Mammalia during the sponsored an expedition that reopened Quar- Jurassic Period has always been one of the ry Nine in the summers of 1968 to 1970. This most interesting, yet poorly documented expedition was led by Thomas H. Rich (then phases of mammalian evolution. Mammals of the American Museum) and Charles R. first appeared in the Rhaeto-Lias (late Trias- Schaff (then of Yale University). The result sic-, about 190 Ma), but are of this expedition was a large collection of known from only three medial Jurassic lo- fragmentary and invertebrates, calities and four localities. With including 25 mammal jaws and numerous the exception of one skeleton (Henkel and teeth. This collection has been expertly pre- Krebs, 1977), no Jurassic therian mammal is pared and given preliminary identification by known from more than fragmentary teeth Mr. Schaff, but has remained undescribed and jaws. until I took up the task at the suggestion of In North America, Jurassic mammals are Dr. Malcolm C. McKenna. Dr. Eugene S. recorded only from the , Gaffney has described the turtles in the col- which is (latest Jurassic), about lection (Gaffney, 1979). Dr. A. W. Crompton 145 Ma. All but six of more than 250 mammal discusses aspects of dental occlusion else- specimens collected prior to 1968 have come where. In the present paper I discuss the from Quarry Nine at Como Bluff, Wyoming. therian mammals in the new collection, in- In the late 1870s and 1880s, the crews of Oth- cluding a new genus and species of dryoles- niel C. Marsh discovered these specimens in tid. Quarry Nine. They were redescribed by While my study of the new form described Simpson (1927, 1929). Except for the pau- herein was in progress, it became necessary rodont Araeodon (Simpson, 1937), no North to review the systematics of the rest of the American Jurassic mammals have been re- dryolestoids. New developments in phylo- ported or described since 1929. genetic analysis, known as "cladistics" or For this reason, at the suggestion of Mal- "phylogenetic systematics" (Hennig, 1965, colm C. McKenna,- the American Museum 1966) are employed for reasons discussed of Natural History and Yale University below. The interrelationships of the numer- 281 282 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167 ous taxa of "pantotheres" have been ne- terial will serve as an immediate test of this glected since Simpson's monographs (1928, hypothesis. I hope that this preliminary hy- 1929). Simpson himself never attempted to pothesis of relationships stimulates other analyze the relationships of the individual workers to attempt to falsify it and to pro- taxa. A cladistic framework not only makes duce a more parsimonious explanation of the this possible but demands it. distribution of characters. In the analysis of primitive versus derived character states and morphocline polarities ABBREVIATIONS in dryolestoids, the analysis naturally ex- panded into all the primitive Theria and the- AMNH, Department of Paleon- rian sister groups. The cladogram (a branch- tology, American Museum of Natural His- ing diagram of relationships using only tory. shared derived character states) was arbi- USNM, Division of Vertebrate Paleontolo- trarily restricted to the non-tribosphenic gy, National Museum of Natural History, Theria. Eventually it will be possible to in- Smithsonian Institution clude the rest of the Mammalia in a much YPM, Peabody Museum of Natural History, larger cladogram currently in progress Yale University (McKenna, 1975). Kimmeridgian (late Jurassic) dryolestoids ACKNOWLEDGMENTS from Guimarota, Portugal, have recently Special thanks are due to three people who been described (Kiihne, 1961, 1968; Krebs, made the project possible. Dr. Thomas H. 1969, 1971; Henkel and Krebs, 1977). How- Rich initiated the project as the potential ever, none of the taxa except Crusafontia, subject for a Ph.D. thesis. He spent many from the Lower of Unia, Spain months in the field and laboratory working (Henkel and Krebs, 1969), has been diag- on it, but did not see it to its completion. Mr. nosed or figured adequately in the literature, Charles R. Schaff put in many months in the and so could not be incorporated into this field and with his customary expertise pre- study. A number of tantalizing figures of pared the specimens. Dr. Malcolm C. named, but still undescribed forms appeared McKenna first suggested the project to both in a paper by Kuhne (1968), but without de- Dr. Rich and to me, and has guided the pres- scriptions or diagnoses.' I have examined all ent manuscript through several revisions. the North American material, and have stud- The field crew at Como Bluff was large ied excellent sterophotographs of the British and changed from year to year, but included material, courtesy of Dr. A. W. Crompton. (in alphabetical order): B. Abel, M. Albert, Most workers who have studied W. Barone, T. Buckley, J. Furst, P. Gooris, mammals have avoided explicit phylogenetic W. Graustein, J. Kadish, S. Koehle Bell, P. hypotheses, or have made them too vague to LaFollette, J. McCarthy, D. McKenna, S. be tested. Admittedly, the fossil record of Mendlinger, R. Ratkevich, A. Sigler, J. Mesozoic mammals is poor, and there is Thomson, P. Whybrow, and G. Winters; much we need to know. There is, however, Drs. M. Chalifoux Coombs, C. Gawne, and a good deal of morphology in these P. V. Rich. I also thank Drs. A. W. Cromp- that we do know, and some sort of phylo- ton and F. A. Jenkins, Jr. (Harvard Univer- genetic hypothesis is warranted. The clado- sity), R. J. Emry (Smithsonian Institution), gram I present represents a preliminary hy- and J. H. Ostrom (Yale University) for al- pothesis, based on all data available as of lowing me to examine specimens in their 1978. The publication of the Guimarota ma- care. Drs. E. Gaffney (American Museum of Natural History), C. Patterson (British Mu- 1 Since these names and figures are published without seum of Natural History), and P. Galton descriptions, diagnoses, or designation of type speci- (University of Bridgeport) have aided in the mens, they are nomina nuda under Article 13 of the In- identification of the turtles, fish, and dino- ternational Code of Zoological Nomenclature. saurs, respectively. Mr. C. Schaff (Harvard 1981 PROTHERO: JURASSIC MAMMALS 283

University) did most of the preliminary iden- manuscript and made helpful suggestions. tification of the mammals, and Mr. E. Man- Mr. Clifford R. Prothero contributed his tal- ning (American Museum of Natural History) ents to the illustrations. Mr. Chester Tarka made preliminary identifications of the lower has given me useful advice on the illustra- vertebrates and curated the collection. Drs. tions. I have gratefully received support M. C. McKenna, P. M. Butler, E. S. Gaff- from a Columbia University Faculty Fellow- ney, T. H. Rich, and Messrs. E. Manning ship and an NSF Graduate Fellowship during and C. Schaff have read all or parts of the this project.

STRATIGRAPHY AND LOCALITIES

In July 1879 William H. Reed, one of the This confirmed statements by Marsh's work- collectors for 0. C. Marsh, discovered Quar- ers that the richest part of the quarry was on ry Nine at Como Bluff (Ostrom and Mc- the west side, and explained the lack of suc- Intosh, 1966, p. 24). Marsh's collectors cess of the 1897 expedition. From photo- worked almost continuously in Quarry Nine graphs in the American Museum archives, for the next 10 years, and sent numerous one can conclude that the efforts of 1897 cans and jars of small bones to New Haven. were directed at the barren east side of the Finally, in June 1889 Marsh's operations at quarry. Como Bluff ceased when Quarry Nine ap- In addition to trenches and test pits, the peared to have been exhausted. Quarry Nine area immediately behind the west face of the yielded almost 250 mammal specimens dur- quarry was bulldozed to expose about 150 ing those 10 years. In 1897 the American square feet of the producing layer. The re- Museum of Natural History worked Quarry sults of the work in the producing layer are Nine, but collected only one mammal jaw. listed in table 1. This specimen remained hidden in matrix Another deposit was found in the east wall until it was discovered and described by of the quarry face, to the northeast of the Simpson in 1937. In the summers of 1968, 1897 American Museum excavation. These 1969, and 1970 a joint American Museum- fossils came from a gray mudstone that im- Yale University expedition reopened Quarry mediately overlies the resistant sandstone Nine intending to find the site of the original bed. The greatest concentration was found work and extending the excavations. During in Pit A (see table 1), which yielded the most this time, four additional mammal localities numerous and best preserved mammals in were also discovered in the vicinity2 (fig. 1). the collection. For this reason, the work in The first phase of renewed operations con- 1969 and 1970 was concentrated more on the centrated on locating the layer from which eastern part of the quarry, and on Pit A. The all but three of the original Quarry Nine specimens recovered from the layer above mammals came. It was known to underlie a the resistant sandstone are listed in table 1. bed of resistant sandstone according to cor- Both localities in Quarry Nine occur approx- respondence between Marsh and his men. imately 80 feet below the Morrison-Cloverly Test pits and trenches were dug, and the lim- contact. The local generalized section for its of both Marsh's excavation and the 1897 Quarry Nine is shown in table 2. American Museum excavation were found. Four additional prospects that produced mammals were also found. In 1968 Dr. Mal- colm C. McKenna found a locality about a 2 The following discussion is abstracted from the field near the crest of notebooks of Thomas H. Rich. These are stored in the mile west of Quarry Nine, Archives of the Department of , Como Bluff. Since it seems to lie near the American Museum of Natural History, and may be con- Morrison-Cloverly contact, it was designat- sulted for further details. ed "Delta T" to indicate its unknown but 284 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

TABLE 1 Faunal List for the AMNH-Yale University Expedition to Como Bluff (x denotes presence of taxon; six-digit numbers are AMNH catalogue numbers of specimens.) Quarry Nine Producing Chuck's Dead layer Pit A Prospect Rabbit Hill Delta T Class Gastropoda x x Class Bivalvia x x Class Osteichthyes Order Amiiformes Ophiopsis sp.a x x x Order Dipnoi Ceratodus sp. x x x Class Amphibia Order Anura x Class Reptilia Order Cheloniab- Family Baenidae Glyptops plicatulus x x x x Dinochelys whitei- x x Order Rhynchocephalia Family Sphenodontidae Opisthias rarus x x Order Infraorder Eolacertilia Cteniogenys antiquus x x x Infraorder Lacertilia sp. x x Dorsetisaurus sp. x x Order Crocodilia Family Goniopholidae Goniopholis sp. x x x - x Order Saurischia carnosaur fragments x x Order Ornithischia( Family Stegosauridae x Family Fabrosauridae x Class Mammalia Order Triconodonta Family 101145 Order victor 104798 104796 104799 104801 104797 104800 104828 Order Family Plagiaulacidae 104795 104794 104793 101146 104802 104803 1981 PROTHERO: JURASSIC MAMMALS 285

TABLE 1--(Continued)

Quarry Nine Producing Chuck's Dead layer Pit A Prospect Rabbit Hill Delta T Legion Cladotheria Family Laolestes eminens 104827 101121 101122 101123 101125 - 101126 - 101127 - 101138 Dryolestes priscus 101136 101140 101137 101139 - 101141 Amblotherium gracilis 101134 Amblotherium debilis 101133 Amblotherium sp. 101143 101135 Herpetairus humilis 101128 - 101129 Melanodon goodrichi 101131 Comotherium richi 104826 101132 Dryolestoid indeterminate 101142 104927 101144 Therian petrosal 104830 Mammalia indeterminate 104926 104928 a Identification of Ophiopsis courtesy of Dr. Colin Patterson. This is the first notice of the occurrence of the genus in North America. bIdentification of the turtles courtesy of Dr. Eugene Gaffney. c Identification of ornithischian dinosaurs courtesy of Dr. Peter Galton. clearly younger age. "Delta T" produced a by Charles R. Schaff. Called "Chuck's Pros- multituberculate , another mam- pect" it seems to be equivalent to "Stego- malian premolar, a badly fractured docodont saurus Quarry 1899" shown in Loomis tooth, and much fragmentary lower verte- (1901, plate XXVI, fig. 2). If so, it occurs in brate material. Loomis's "No. 28 gray sandstone," approx- Another locality was discovered by Dr. imately 40 feet from the top of the Morrison Patricia V. Rich about a mile west of Quarry formation. "Chuck's Prospect" yielded a Nine. Designated "Dead Rabbit Hill" be- number of mammalian jaws before the de- cause a dead rabbit crowned it, the site was posit was apparently exhausted. A list of the about 150 feet below the Morrison-Cloverly specimens from this locality is shown in ta- contact. The deposit was screenwashed and ble 1. produced some Ceratodus teeth, turtle ma- A fourth locality was found by M. Albert terial, and a lower of Docodon. in July 1969. It was designated "M area" and In 1969 while prospecting in Bone Cabin was 390 feet due east of Quarry Nine at the Draw (see fig. 1), another locality was found base of Como Bluff. The exact stratigraphic 286 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

WYOMIedNGnMONTIN Bon Cabin

lff

1 0SMi.0 0 10 Km

30~ ~ ~ ~ ~ 7 7~~~~~~~~8

W~~~~~~~~~ ~Rc Riverl

> 9 \ ~Carbon Co. |Albany Co. \

MEDICINE BOW \ ,> MOUNTAINS LARAMIE BASIN

1060

FIG. 1. A. Index map of Wyoming; stippled area is enlarged in B. B. Map of localities described in text. Stippled area indicates outcrops of Morrison Formation. Q9 = Quarry Nine, Dead Rabbit Hill and "M" area; T = Delta T; CP = Chuck's Prospect

position of this locality has not been deter- about 15 percent larger than any so far mined. "M area" produced a fragmentary known from Como Bluff. triconodont jaw, a gastropod, and some lac- The work of 1970 produced no new mam- ertilian material. The triconodont teeth are mals or localities.

SYSTEMATIC DESCRIPTIONS CLASS MAMMALIA LINNAEUS, 1758 KNOWN DISTRIBUTION: Latest Jurassic INFRACLASS THERIA PARKER AND HAS- (Tithonian), Quarry Nine, Como Bluff, Al- WELL, 1897 bany County, Wyoming. LEGION CLADOTHERIA McKENNA, 1975 ETYMOLOGY: Como, from the type local- SUBLEGION DRYOLESTOIDEA BUTLER, 1939 ity, Como Bluff, Wyoming; therium, Latin- FAMILY DRYOLESTIDAE MARSH, 1879 ized form of the Greek Orhptov, "beast." DIAGNOSIS: Dryolestid upper jaws with TAXA KNOWN FROM UPPER JAWS stylocone placed at anterolabial end of par- COMOTHERIUM, NEW GENUS acrista. Heart-shaped labial border with Figures 2, 3A-C large, raised, two-cusped metastyle contact- TYPE SPECIES: Comotherium richi, new ing shelflike, enlarged parastyle anterolin- species. gually. No ridges or cusps in basin of trigon. 1981 PROTHERO: JURASSIC MAMMALS 287

TABLE 2 St ps Local section at Quarry Nine Thickness Lithology in meters Variegated gray claystones and sandstones, with occasional chert lenses. Locally fossiliferous (see table 1, Pit A). 2-4 White sandstone with minor feldspar, orange staining and iron concretionary lenses near top. Unfossiliferous. 0.3-2 Green siltstone and sandstone. Locally fossiliferous (see table 1, producing 2mm layer). 0-0.35 FIG. 2. Comotherium richi, AMNH 101132, genoholotype. Cross-hatched area is damaged. Symbols are as follows: ide = interdental embra- Interdental embrasure between trigons nar- sure, me = metacone, mec = metacrista, mt = rower than in any other dryolestid. metastyle, pa = paracone, pac = paracrista, ps = parastyle, st = stylocone, tb = trigon basin. Comotherium richi, new species TYPE SPECIMEN: AMNH 101132, a left anterior and posterior margins of a central maxillary fragment with three molars. trigon basin. No cusps or ridges are found in REFERRED SPECIMEN: AMNH 104826, an the basin, unlike those of most other dry- isolated right upper molar. olestoids. A small, conical stylocone occurs HORIZON AND LOCALITY: Same as for ge- on the labial end of the paracrista. This ex- nus. tremely anterior position of the stylocone is ETYMOLOGY: In honor of Dr. Thomas H. unusual for dryolestids. The stylocone is set Rich, who led the 1968-1970 American Mu- anterolingual to the medial external border, seum-Yale expedition to Como Bluff, and unlike the situation in most other dryoles- who discovered the first specimen of Com- tids. The metacone is a small swelling mid- otherium. way along the metacrista. It is clearly seen SPECIFIC DIAGNOSIS: Same as for genus. only on the middle molar. DESCRIPTION: AMNH 101132 consists of The stylar region of this specimen is much a left maxillary fragment with three molars. larger than in other dryolestids, forming a The anterior molar has been rotated postero- heart-shaped outer margin of the tooth. The lingually, so that the interdental embrasure metastyle of one tooth is closely appressed is extremely narrow. This is partly due to anterolingually to the parastyle of the pos- crushing and partly natural. The embrasure terior tooth, so that they appear to be fused between the second and third molars is also in the case of the first and second teeth. The much narrower than in any other dryolestid, parastyles curve anteriorly from the stylo- but the roots are firmly in place, and the axes cone, and in the first tooth form a right angle of the second and third molars are still in with the line of the paracrista. The parastyles line. also seem to show a small ridge running Except for the broken metacrista of the along their anterolabial margins. The en- first tooth, the teeth are unworn and undam- larged metastyle is separated from the rest aged. The molars have high prominent para- of the tooth by a distinct valley running di- cones with two ridges: a paracrista and a agonally across the tooth. Where the pres- metacrista that descend labially to form the ervation is good, the metastyles show two 288 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

'.N.

B A

1mm /- i,

.f r --.8E v

;v w. wI .....

.1 @-ii .*-.ffRE.r 8

X I.,~~~~

/ .. .. ~ ...:` V **X,/,..\ '4

lak W !e 'Ali -I., .Zle:

t -- I .-- D

1mm

FIG. 3. Comotherium richi, AMNH 101132, genoholotype. A. Crown view. B. Labial view. C. Lingual view. D. Herpetairus humilis, AMNH 101130. Crown view.

cuspules, one at the labial end of the meta- than the anterior metastyle- . Tne para- crista, the other on the anterolingual margin style is quite low, flat, and shelflike in labial of the metastyle. The posterior metastyle view. cusp is approximately equal in height to the Alveoli for the tooth posterior to those de- stylocone and metacone. All three cusps are scribed indicate that the molars are three much lower than the paracone, but higher rooted. The largest root is for the paracone; 1981 PROTHERO: JURASSIC MAMMALS 289 a second large root, anterolingual to it, sup- and separated from the trigon by a valley; ports the stylocone and parastyle, and a this does ndt occur in Euthlastus. broad, anteroposteriorly compressed root 5. Euthlastus has a single discrete para- supports the metastyle. It is posterolingual style cusp equal in size to the stylocone; to the parastyle-stylocone root, and postero- Comotherium has no parastyle cusp. labial to the paracone root. Thus, AMNH 101132 and 104826 could not The extreme anterobasal part of the zy- be referred to any genus of dryolestid upper gomatic arch is preserved on this specimen; teeth yet known, and are given a new genus. it begins at the level of the last preserved Interpretation of primitive and derived char- molars. If the position of the zygomatic arch acter states, and of the systematic position with respect to the molars is comparable in of Comotherium is given below. Melanodon and Comotherium, then the mo- lars preserved would be M35. AMNH 104826, an isolated right upper MELANODON AND HERPETAIRUS molar from Quarry Nine (see table 1), is also In 1927 and 1929 Simpson erected the gen- referred to Comotherium richi. It shows all era Melanodon and Herpetairus, each with the characters of the type specimen, includ- two species. Both forms are very similar in ing the strong paracrista and metacrista, the basic morphology, and are the most common high paracone, and the same shape and upper-teeth taxa at Como Bluff. Butler placement of the stylocone. It differs only in (1939, p. 338) suggested that the distinction that the stylocone is not so strong as in the between the genera was insignificant. He fig- type, and the parastyle is not so strongly ured a specimen (Melanodon goodrichi, hooked. These differences, however, fre- YPM 13750) that has a median transverse quently occur on more anterior molars of ridge on one molar (a Melanodon character) other dryolestids. Since the type seems to and lacks this ridge on the molar anterior to consist of M35, it is probable that AMNH it (a Herpetairus character). According to 104826 is an M1 or M2. Butler, this was the primary character used MEASUREMENTS (IN MM.): M3 length by Simpson (1929, p. 75, not p. 85, as listed 1.01, width 1.75; M4 length 0.96, width 1.69; in Butler, 1939, p. 338) to distinguish the two M5 length 0.84, width 1.63. genera, and therefore the distinction was DIscUSSION: Comotherium was tentative- doubtful. ly referred to Euthlastus cordiformis by the I have examined the entire Marsh collec- original workers, since both forms have tion carefully, and disagree with this sugges- strongly hooked parastyles, resulting in a tion. The specimen in question does show a heart-shaped labial margin. This character is median transverse ridge on the anterior mo- much more strongly pronounced in Como- lar. There is a small break or gap in the cen- therium, whose parastyle and metastyle are ter of the ridge, but its trace can be clearly much more closely appressed than in Euth- seen on the labial face of the paracone. I lastus. Direct comparison of the two reveals could find no evidence of a stylocone-meta- some striking differences: cone ridge (diagnostic of Herpetairus arcu- 1. The teeth of Comotherium are about atus) shown in Butler's text-figure 5. In all twice the size of Euthlastus. other characters, particularly stylocone size 2. Comotherium has a higher paracone, and position, this specimen is typical of Me- stronger paracrista and metacrista, and a lanodon goodrichi. more acute trigon angle than does Euthlas- In examining the Marsh collection, I found tus. that all four of Simpson's Melanodon-Her- 3. The stylocone in Comotherium is dis- petairus species could be distinguished eas- tinct and conical, and lingual to the para- ily and unambiguously. All the new speci- style. In Euthlastus, the stylocone is a swell- mens from Quarry Nine could be referred to ing on the posterolabial part of the parastyle. one or another of these four species without 4. In Comotherium, the metastyle is raised difficulty. Because there are four diagnos- 290 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

TABLE 3 Comparison of Herpetairus and Melanodon Character H. arcuatus H. humilis M. goodrichi M. oweni Stylocone size moderate moderate large very large Stylocone position on marginal marginal marginal central middle molars Stylocone position on posterior slightly posterior posterior paracrista posterior Median transverse ridge absent faint-moderate moderate strong Metacone-stylocone ridge present absent absent absent Paracrista cusp small small small large Labial margin of straight straight straight concave middle molars Stylocone shape transversely transversely pyramidal conical compressed compressed plast internal cingulum present ? ? absent plast external cingulum one, posterior ? ? three cusps able clusters of morphology present, I con- transverse ridge than Herpetairus or Melan- tinue to recognize Simpson's four species. odon goodrichi. Revised diagnoses are given below and in ta- DISCUSSION: Several writers (Butler, 1939, ble 3 to make the distinctions more clear. p. 337; Kiihne, 1968, among others) have For interpretation of primitive and derived suggested that Malthacolestes osborni is a character states, see Discussion. dP4 and M1 of Melanodon. Unfortunately, M1 of the type of Melanodon oweni is poorly MELANODON SIMPSON, 1927 preserved. One specimen (YPM 13752) in the Malthacolestes Simpson, 1927 Marsh collection appears to be M1-2 of Me- lanodon oweni, based on size and position of TYPE SPECIES: Melanodon oweni. the stylocone, and by comparison with M2 in REVISED DIAGNOSIS: Dryolestid upper the type. M1 of YPM 13752 is practically jaws with large stylocone, placed posterior identical with the second tooth of Maltha- to paracrista. Strong parastyle, and two- colestes osborni. Therefore, Malthacolestes cusped metastyle. Distinct transverse medi- is here synonymized with Melanodon oweni. an crest on trigon. Last premolar without in- ternal cingulum; external cingulum with Melanodon goodrichi Simpson, 1927 three cusps: anterior, central, and posterior. TYPE SPECIMEN: YPM 13738, right maxilla oweni 1927 with M>5. Melanodon Simpson, REFERRED SPECIMENS: YPM 13750, 13749; TYPE SPECIMEN: YPM 10663, a right max- AMNH 101131. illa with P4 and M1-5. REVISED SPECIFIC DIAGNOSIS: Stylocone REFERRED SPECIMENS: YPM 13752 (re- large, pyramidal in shape, placed on straight ferred to Melanodon goodrichi by Simpson, labial margin. Moderate median transverse 1929); YPM 13751, (type of Malthacolestes ridge compared to M. oweni. osborni Simpson, 1927). DESCRIPTION: AMNH 101131 (fig. 4A) REVISED SPECIFIC DIAGNOSIS: Stylocone consists of a left maxilla with Ml-4 and the very large, completely conical, labial to line base of M5. The preservation of Simpson's of parastyle and metastyle. Labial margin of hypodigm is very poor in these teeth, so this middle molars concave. Stronger median specimen deserves detailed description. 1981 PROTHERO: JURASSIC MAMMALS 291

-1 4 1mm i '.., A

.IF"kr 1mm

t'r-, 1,17, .,4 11 '*No- W-1*1- B "A. :.

w "4W., 0i 41 ll.- .1&

c ., 'I

FIG. 4. A. Melanodon goodrichi, AMNH 101131, crown view. B. Amblotherium debilis, AMNH 101133. Posterior ramal fragment in lingual view. C. Oblique crown view of same specimen, showing M8 erupting in crypt.

M1 is narrower labiolingually than the oth- apex. The stylocone is larger than on the oth- er molars and has a simple, slightly basined er molars, and is roughly conical with a dis- crown, shaped like an equilateral triangle. A tinct lingual and posterior facet. It is not so partly broken paracone rises from the lingual pyramidal as on the other molars. The sty- 292 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167 locone arises from the central labial margin transverse ridge weak or absent. Last pre- of the trigon. The parastyle is broken, but molar with both external and internal cin- the base indicates that it extends out antero- gula, with a single small posterolabial cusp. labially from the trigon. The metastyle is flat, forming the posterior corner of the equilat- Herpetairus arcuatus (Marsh, 1879) eral triangle of the trigon; two small incipient TYPE SPECIMEN: YPM 11822, right maxilla cusps can be seen. No metacone appears on with W2-7. the metacrista. A very weak ridge connects REFERRED SPECIMENS: YPM 13741, 13742, the stylocone and paracone. 13743, 13744, 13746, 13747, 13740; USNM M2-4 become progressively more wedge- 2724, 2762, 2800, 2818, 2845. shaped and anteroposteriorly compressed. REVISED SPECIFIC DIAGNOSIS: Distinct The paracrista and metacrista are distinct No trans- and form valleys that are divided by the sty- metacone-stylocone ridge. median locone-paracone ridge. The parastyle of M2 verse ridge. interlocks posterolabially with the metastyle of M1, but succeeding parastyle-metastyle Herpetairus humilis Simpson, 1927 contacts are less labial and more posterior. TYPE SPECIMEN: YPM 13745, left maxilla Stylocones are high and in line with the sty- with M-6. lar cusps (as opposed to M. oweni) but are REFERRED SPECIMENS: AMNH 101128, not so pyramidal in shape as on the type of 101129, 101130. M. goodrichi. However, this could be a REVISED SPECIFIC DIAGNOSIS: Faint to function of wear, and therefore the alignment moderate median transverse ridge. No meta- of the cusps is seen as a better indicator of cone-stylocone ridge. affinity. A small ridge runs down the anterior DESCRIPTION: AMNH 101130 (fig. 3D) is face of the stylocone and connects it with the a left maxillary fragment bearing M-5 and paracrista. alveoli for M2 and M6. The paracope of M5 Parastyles all project anteriorly from the and parastyle of M3 are broken off, but oth- trigon with a valley between them. Each erwise the teeth are well-preserved. This parastyle has a low, rounded cusp with a lin- specimen compares closely with YPM 13745, gually inclined wear facet. The metacrista the type of H. humilis, but is much less has two low cusps, the larger of which ap- worn. It has a weak transverse ridge and no pears to be the metacone. Both cusps show stylocone-metacone ridge, diagnostic of H. well-developed wear facets inclined postero- humilis. The stylocone is slightly more an- lingually. Two small cusps are present on the terior than in the type specimen, but other- metastyle. One, at the posterior end of the wise its size, shape and position are typical labial margin of the trigon, is connected to of Herpetairus. Because of the lack of wear the stylocone by a small ridge. The other on AMNH 101130, the metacone is much metastyle cusp occurs at the most labial end more distinct than on the type. The meta- of the metacrista. cone is roughly hemispherical in shape and The maxillary fragment includes the an- cut in half by the vertical truncation of the teroventral part ofthe zygomatic arch, which metacrista. The metastyle also shows less begins at the level of M4. wear than in the type of H. humilis. It is narrower and more extended pQsteriorly. There is a distinct trough between the meta- HERPETAIRUS SIMPSON, 1927 crista edge and the stylocone-metastyle TYPE SPECIES: Herpetairus arcuatus. ridge. The latter ridge terminates with a dis- REVISED DIAGNOSIS: Dryolestid upper tinct metastyle cusp. jaws with moderate, transversely com- AMNH 101130 came from Quarry Nine, pressed stylocone, placed posterior to para- but its exact stratigraphic position is un- crista. Parastyle and metastyle approximate- known, because the field number on the ly equal to stylocone in size. Median specimen proved to be in error. 1981 PROTHERO: JURASSIC MAMMALS 293

AMNH 101129 is a right maxillary frag- from Guimarota, Portugal (Krebs, 1971). ment with (?) M-5 and an alveolus for (?) Nor does the specimen show any trace of a M3. The teeth are very heavily worn, so that coronoid sutural surface that is found in Por- the metacone is gone and the trigon has a tuguese forms. The pterygoid fossa is very saddle-shaped appearance. The size and po- deep on the internal surface of the jaw. The sition of the stylocone, and a faint median coronoid process is fragmentary but quite transverse ridge, are in agreement with H. high; the condylar process is broken. The humilis. jaw apparently had a weak angular process. AMNH 101128 is a right maxillary frag- ment with one molar (?M4), followed by two Amblotherium gracilis Simpson, 1927 three-rooted alveoli and part of a third al- veolus. The tooth is moderately worn, with TYPE SPECIMEN: YPM 11883. a broken parastyle, but the metacone has not REFERRED SPECIMENS: AMNH 101134; yet been worn off. The stylocone propor- YPM 13732, 13731, 13733; USNM 2142, tions and faint median transverse ridge agree 2693. with H. humilis. DESCRIPTION: AMNH 101134 is a left ra- mal fragment with one molar and three al- veoli anterior to it. The molar is badly worn TAXA KNOWN FROM LOWER JAWS with a broken metaconid and talonid. How- AMBLOTHERIUM OWEN, 1871 ever, the size (25 percent larger than A. de- bilis) and presence of a strong labial cingu- Amblotherium debilis Simpson, 1927 lum compares closely with A. gracilis. TYPE SPECIMEN: YPM 11821. REFERRED SPECIMENS: AMNH 101133; Amblotherium sp. YPM 13728, 13730, 13734; USNM 2726. REFERRED SPECIMENS: AMNH 101135, DESCRIPrION: AMNH 101133 (fig. 4B-C) 101143. consists of two right ramal fragments: one DESCRIPTION: AMNH 101135 is a mandi- segment has three molars (which three is ble fragment with four worn . It is hard to determine), and the other segment doubtfully referred to Amblotherium on the consists of M6, alveoli for M7.8, and the rear basis of size. portion of the jaw. The last "alveolus" is of AMNH 101143, an isolated left lower mo- particular interest in that it has a molar still lar, is most similar in morphology to A. de- inside it and just about to erupt. The small bilis, but is the size of A. gracilis. Unlike all size of this tooth (relative to the fully erupted other specimens of Amblotherium except MK) and the partly formed trigonid suggest YPM 13732 (referred to A. gracilis), the that this "alveolus" is a crypt for an M8 that paraconid is completely erect, and the tal- had not yet erupted at death. M7 alveolus onid is cusplike, rather than forming a ledge. gives every indication that M7 has been fully If this unique morphology occurred in more erupted. Apparently the eruption of M8 took than just two isolated teeth, a new species place much later than the rest of the teeth. might be justified. For the moment, this AMNH 101133 compares closely with A. specimen is referred to Amblotherium sp. debilis in that it is 20 to 25 percent smaller than A. gracilis, has a faint labial cingulum DRYOLESTES MARSH, 1878 on the molars, and in that the paraconid is lower than the metaconid on MK. One char- Dryolestes priscus Marsh, 1878 acter not noted by Simpson is the fact that TYPE SPECIMEN: YPM 11820. the paraconid overlaps the preceding talonid. REFERRED SPECIMENS: AMNH 101140, The specimen has a marked inner groove 101139, 101136?, 101137?, 101141; YPM ending at the dental foramen on the inside of 11884, 10646; USNM 2722 the jaw, but there is no sign of splenial DESCRIPTION: AMNH 101140 is a right grooves found in Kimmeridgian dryolestids jaw fragment with M4-6 preserved and 294 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

The teeth are more worn than in other specimens of Dryolestes and consequently the trigonids are very narrow anteroposte- riorly. The bifid metaconid of Laolestes is Z-O.; absent in this specimen; this can often be due .-z-fIW4" to wear, however. This specimen is distinct from Amblotherium in its lack of labial cin- gulum on the molars and presence of strong- ly procumbent paraconids. MK in this speci- men has an apparent lingual "cingulum" that A seems to be caused by a break through the crown of the tooth. The talonids are rela- 1mm tively broad and shelflike, with a small ridge on the lingual side that runs anteroposterior- ly. AMNH 101139 (fig. 5) is a right fragment with five molars, possibly M3-7. The last two are broken and only the lingual por- tions of the crowns remain. The crowns of

I U~ the other three molars are excellently pre-

,f,f... served and show the least amount of wear of any specimen in the collection. As a result, the conids are much more rounded, the notches between them deeper, and distinct B wear facets are poorly developed. The meta- conid is not bifid nor out of the paraconid- talonid line, distinguishing this specimen from Laolestes. Unlike Amblotherium, the paraconid is strongly procumbent and there is no labial cingulum. The talonid on these teeth is relatively un- worn and consequently has a strong antero- posterior ridge in its center. This gives the talonid a peaked, convex shape that disap- pears as the ridge wears down and is reduced to a remnant on the lingual corner, as in AMNH 101140. The paraconid of the most FIG. 5. Dryolestes priscus, AMNH 101140. A. anterior tooth seems to have a cusp on it, Lingual view. B. Labial view. C. Crown view. whereas the paraconids of the two teeth be- hind it are flat, triangular shelves. Finally, the enamel on these teeth is markedly lighter alveoli for M7,8. Most of the back of the jaw in color than any other specimen in the col- is preserved, except for the angle, coronoid lection. process, and condyle. The deep pterygoid AMNH 101137 is a left mandible fragment fossa has a fairly large dental foramen in the with P4-M3. The teeth are well worn, and the anterior part. Around the internal groove molars have trigonids shaped more like a U there are no traces of insertion grooves for than a V. The talonids are shelflike but have a splenial bone. Because the coronoid pro- a distinct cusp on the posterolingual corner cess is damaged on the lingual side, it is dif- with a ridge running anterior to it. The pro- ficult to determine if there was any trace of cumbent paraconids, lack of labial cingulum, a coronoid bone. and alignment of the lingual cusps seem to 1981 PROTHERO: JURASSIC MAMMALS 295 assign this specimen to Dryolestes. How- umn developed on the posteroexternal sur- ever, P4 has no anterior accessory cusp and face of the jaw. The angle is also preserved, the lingual cingulum is continuous, which making the length of the mandible from M1 8 contradicts Simpson's (1929, p. 57) diagnosis nearly equal to the jaw segment posterior to for Dryolestes. the tooth row. Simpson's figure 25 shows the AMNH 101136 is a right ramal fragment molar part about one-third longer than the with three molars of uncertain designation. posterior part of the jaw. A single mental fo- The procumbent paraconids, lack of labial ramen appears below the alveoli for P,4. cingulum, and non-bifid metaconids suggest The teeth, unfortunately, are extremely Dryolestes. However, the metaconids are worn and blunt. The only diagnostic char- more labial to the paraconid-talonid line than acters that are preserved are the presence of other Dryolestes, and like most Laolestes. a labial cingulum on the molars and the pro- The paraconid shelf overlaps the preceding cumbent paraconids, which together suggest trigonid to a greater extent than in any other Laolestes. However, AMNH 101121 is specimen. Consequently, the relatively un- about 25 percent smaller than the holotype worn talonids have a strong ridge on them, of L. eminens, and smaller than any other oriented anteroposteriorly. Laolestes in the collection. The talonids are AMNH 101141 is a right ramal fragment broken on M2_3, and M1 is so worn as to be with M78 and the back of the jaw poorly reduced to a knob. preserved. M8 is about one-third smaller than AMNH 101138 is a left ramal fragment the broken M7. Both teeth have talonids that with M2,4 and alveoli for P,4 and M1. The are much smaller than in other specimens bifid metaconids labial to the paraconid-tal- and are strongly convex. The metaconid is onid line, as well as the labial cingulum, about equal to the protoconid in height, but clearly indicate Laolestes. The talonids are otherwise the characters of the teeth suggest narrower than in other specimens, but do Dryolestes. show a distinct posterolingual cusp with a ridge running anteriorly to the metaconid. LAOLESTES SIMPSON, 1927 The trigonids are much broader than the ho- lotype of L. eminens. There is no mental fo- Laolestes eminens Simpson, 1927 ramen beneath the P,4 alveoli. TYPE SPECIMEN: YPM 13719. AMNH 101124 is a left jaw fragment with REFERRED SPECIMENS: AMNH 101121, slightly worn M_6 and alveoli for M78. It 101138, 101124, 101126, 101125; YPM 10662, clearly shows the bifid, relatively labial 10658, 10660, 10661, 13720, 13721, 13723, metaconids and labial cingula characteristic 13724, 13728, 13726; USNM 2727, 2729, of Laolestes. The paraconids tend to have 2731, 2732. anteriorly directed wear facets that become DESCRIPTION: AMNH 101121 (fig. 6A-B) progressively more worn anteriorly. The tal- is the most complete jaw in the collection, a onids become progressively narrower pos- left ramus with M_7 and the alveoli for M1 teriorly, but all have a distinct posterolingual and M8. Except for the coronoid process, the cusp and an anteriorly-directed ridge. back of the jaw is complete. The inside of AMNH 101125 is a right ramus with P4-M5 the jaw, however, is too badly fractured to and alveoli for I4, C, and P1.3. The bifid discern any internal groove or splenial molar metaconids become progressively grooves, and even the pterygoid fossa is bad- more labial relative to the paraconid-talonid ly broken. No trace of a coronoid bone or a line in more posterior molars. The labial cin- sutural surface for one is present. The con- gulum is also present, corroborating the dyle is well preserved, and is much higher identification of this specimen as a member and more vertical than in Simpson's (1929) of the genus Laolestes. The talonids pre- figure 25. The chief articular surface is most- served on M12 are narrow but have the ly dorsal (rather than posterodorsal, as it has characteristic posterolingual cusp and ridge. been illustrated) and it is supported by a col- P4, which is missing on the holotype, is much 2g6 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

3' v 'f -'S'

B _

C.R' .

''Y',=-w h % ;:-._ W.ll

D

FIG. 6. Laolestes eminens, AMNH 101121. A. Lingual view. B. Labial view. C. Kepolestes color- adensis, USNM 2723. Crown view. D. Labial view. E. Lingual view.

larger than P13 or M1, and has a faint lingual strong labial cingulum and apparently bifid cingulum and a strong labial cingulum. There metaconids, suggesting Laolestes. The tal- is no anterior accessory cusp. onids are quite narrow and heavily worn, so AMNH 101126 is a left mandible with M35 that the posterolingual talonid cusp is also and alveoli for M1,2 and M-68. The back of quite worn. This wear is greatest on M3 and the jaw is not preserved. The molars have a least on M5. 1981 PROTHERO: JURASSIC MAMMALS 297

KEPOLESTES SIMPSON, 1927 M3-7 show the Figure typical dryolestid molar 6C-E pattern. The amount of wear has reduced the Kepolestes coloradensis Simpson, 1927 protoconid and metaconid to a low, saddle- shaped ridge, and consequently there is a TYPE AND ONLY SPECIMEN: USNM 2723. of REVISED DESCRIPTION: narrowing the trigonid. The wear has also Kepolestes color- reduced the paraconids to blades or wings, adensis is the only known Morrison therian with that deeply worn, U-shaped grooves sepa- did not come from Como Bluff. It was rating them from the metaconids. Two dis- found in Marsh's dinosaur quarries at Gar- tinctive derived characters clearly indicate den Park, Colorado. Simpson named (1927) the phylogenetic affinities of Kepolestes: and described it (1929, p. 67) on the basis of A labial the 1. cingulum, previously obscured lingual view of the jaw. The other side of in is visible on the jaw matrix, M-6. It is not so remained unprepared and unde- strongly developed as in Laolestes, but is scribed. When this specimen (USNM 2723) nevertheless distinct. This is shared with was graciously lent to me for study by Dr. Laolestes, Crusafontia, and Amblotherium R. J. Emry, it was decided that the prepa- (see ration should be completed. fig. 8). Mr. Otto Si- 2. The metaconids are not so heavily worn monis masterfully executed this delicate on and task. M3 M7, and further preparation re- veals that they are slightly bifid, contrary to Kepolestes coloradensis consists of a right Simpson's (1929, p. 70) statement. This char- lower jaw with P4 and M1 8. Alveoli for a acter is found elsewhere canine and are only in Laolestes P1,3 present. There may be and thus unites Kepolestes and Laolestes as indications of at least one incisor alveolus on sister taxa (see the broken anterior end of the fig. 8). jaw. The jaw Kepolestes presents an additional bizarre is very slender and tapers upward at its an- feature revealed further terior end. An internal by preparation. M8 groove runs parallel is reduced to a cylindrical peg, approximate- to the ventral border and immediately above the same it. The back of the ly height as M7. This odd tooth jaw lacks an angular pro- morphology does not seem to be due to cess and condyle, but is otherwise complete. The coronoid breakage or excessive wear, as the surfaces process is heavy, slanting up- are smooth and the alveolus is the same di- ward at a 30-degree angle to the tooth row. Its ameter as the crown. The cross section anterior border has a broad surface that is roughly ovoid and only one-third of the is directed anterolabially. The pterygoid fos- diameter of A sa is very shallow M7. small, triangular, antero- and poorly developed. The labially-facing wear facet is present, but oth- masseteric fossa, however, is quite deep and erwise extends the tooth is featureless. This tooth is forward as a trough that terminates difficult to interpret, and its is below M4. A mental foramen appears below morphology the alveolus apparently unique among primitive Theria. for P2 It may be due to abnormal development in The teeth are highly worn, so that M1l2 this are reduced particular individual, or to other possible to featureless nubbins. P4 is high causes which must remain speculative. If and recurved, with no trace of anterior cusps another or any specimen of Kepolestes were found cingula. However, its preservation is that showed this unique morphology, the very poor, so these features cannot be too form of heavily relied M8 would be considered an autapo- upon. morphy for the genus. DISCUSSION The study of the phyogeny of the non-tri- mammalian paleontologists have not had to bosphenic Theria has been hampered by contend: some special problems with which most (1) Nearly all the known taxa occur in the 298 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167 late Jurassic of North America and Europe. primitive, ancestral morphotypes to the ac- Thus, stratigraphic superposition cannot be quisition of shared derived characters. Cla- used to reconstruct phylogeny. distic systematics, in the strictest sense, (2) The taxa are known almost exclusively does not recognize ancestors (Schaeffer et from fragmentary jaws and teeth, whose as- al., 1972; Englemann and Wiley, 1977), since sociations are unknown or doubtful in most ancestors are unessential to the analysis. But cases. The skeleton of one Kimmeridgian even non-cladistic authors (e.g., Clemens, dryolestid has been reported (Henkel and 1970, p. 369; Clemens and Mills, 1971, pp. Krebs, 1977), but this does not resolve the 110-111; Fox, 1977) have been forced to ad- of the rest of the forms, which are mit that the generalized forms have special- primarily "upper jaw animals" and/or "low- izations (or "Iautapomorphies" in cladistic er jaw animals." terminology) which exclude them from an- (3) The samples are exceedingly small, and cestry of more derived taxa. If one is inter- come primarily from two localities: Quarry ested in the ancestral morphology, it can be Nine, Como Bluff, Wyoming, and the Pur- extrapolated from the nodes of a cladogram, beck Beds of England (with new forms from without making untestable assertions about Guimarota, Portugal, yet to be published ad- a particular specimen "giving rise" to later equately). Fourteen of the 29 taxa in figures animals. 8 and 9 are known from single specimens. The lack of association between most up- Also, most of the forms are quite distinctive per and lower dentitions made it necessary and too specialized to represent ancestral for Simpson (1928, 1929) to assign different morphotypes. generic names for each, except in the few As a result, most workers have avoided cases where the association was proved. discussing the relationships of the various This proliferation of names gives a false taxa, limiting their work to morphological impression of diversity, but no evidence is transitions between Amphitherium, Pera- presently available to resolve the dilemma. mus, and some "typical" dryolestoid or Unfortunately, the diagnostic characters of "typical" symmetrodont. The unfortunate the "lower jaw taxa" such as Laolestes and consequence of this practice is that the di- Dryolestes occur primarily on the lingual versity of forms available has been widely side of the teeth. Thus, these features would ignored, simply because they were "too spe- not appear on the upper teeth as wear facets, cialized to be the ancestor of later forms." since the lingual side of the lower teeth is not In view of these conditions, a cladistic ap- in direct occlusion. Similarly, most of the proach to therian phylogeny is especially ap- distinguishing characters of taxa based on propriate. Readers unfamiliar with the prac- upper teeth, such as those of Herpetairus or tice and terminology of cladistic systematics Melanodon, occur on the labial margin or in are referred to papers by Hennig (1965, the center of the trigon. Again, identification 1966), Brundin (1966, 1968) Cracraft (1972), of diagnostic wear facets to determine the Griffiths (1972), Schaeffer et al. (1972), El- synonymy of these names would be very dif- dredge and Tattersall (1975), and Wiley ficult, and is not attempted here. (1976, pp. 7-13). Cladistic systematics has Clemens and Lees (1971) have described the following additional advantages over an isolated upper tooth and lower tooth they competing systems of phylogeny reconstruc- call Melanodon hodsoni, from the early Cre- tion for the primitive Theria: taceous (Wealden) of England. The upper (1) Cladistic systematics does not rely on tooth appears to be indistinguishable from stratigraphic data to reconstruct phylogeny Melanodon goodrichi, except by size. The (Schaeffer et al., 1972). lower tooth, as far as it is preserved, shows (2) Cladistics presents a technique of ana- a bifurcated metaconid, which is shared with lyzing diverse morphologic data and con- Laolestes. However, it lacks the labial de- structing a parsimonious phylogeny from it. flection of the metaconid that is usually (3) Cladistics changes the emphasis from found in Laolestes. The most consistent di- 1981 PROTHERO: JURASSIC MAMMALS 299 agnostic feature of Laolestes, a labial cin- unique to the Theria. In all Theria, the mo- gulum, is not preserved, since the protoconid lars are slightly "(skewed"; that is, the pro- is broken off. Clemens and Lees were reluc- tocristid-paracrista shear is more vertical tant to synonymize Melanodon and La- and transverse, whereas the paracristid-me- olestes on this evidence, and I concur with tacrista shear is more oblique and diagonal this decision. (Crompton, 1971, p. 80). These characters In addition, there are four distinct species are partly correlated with the reversed tri- of "upper jaw animals" (Melanodon and angle arrangement, but did not develop si- Herpetairus) that are the right size to be as- multaneously. In , Amphi- sociated with three species of "lower jaw don, and (Crompton, 1971, p. 82), animals" (Laolestes and Dryolestes). Which a single wear facet is developed on the an- belongs to which? We have gotten no further terior face of the trigon and posterior face of on this problem than Simpson did. Lower the trigonid (paracrista-protocristid), but jaws of two species of Amblotherium are there are still multiple wear facets (A and B, known from Como Bluff, but no uppers have Crompton, 1971, fig. 8A) on the posterior been found so far. There are now four Como trigon-anterior trigonid (paracristid-meta- Bluff genera of "upper jaw animals" (Peli- crista). The elimination of these metacrista copsis, Miccylotyrans, Euthlastus, and wear facets and the development of a single Comotherium) that seem to have no likely metacrista-paracristid wear facet is appar- candidate for lower jaws. Pelicopsis has ently the derived condition. It occurs in spa- been suggested as the upper teeth of a pau- lacotherioid symmetrodonts, dryolestoids, rodont (Simpson, 1929, p. 83), but there are , and the . Amphi- five species of paurodont lower jaws known. therium, however, still retains facet B and In view of this lack of correspondence be- part of A on the paraconids of its anterior tween "upper jaw" and "lower jaw" taxa, molars; the upper teeth are unknown it was necessary to analyze them separately (Crompton, 1971, fig. 8B). and generate a cladogram for each. These are Butler (1972, fig. 3) has shown that the in- then combined into a cladogram of both up- clination of shear movement in per and lower jaws, using only the taxa that (?=), and Kuehneotherium have both, or whose associations are reason- ranges from 30 to 40 degrees. This is also ably well demonstrated. true of Peramus and spalacotherioid sym- metrodonts and occurs in the primitive Tri- GENERAL TOOTH CHARACTERS bosphenida. Amphitherium and the dryoles- toids develop a shear inclination of 40 to 50 OCCLUSAL RELATIONS: The "reversed tri- degrees, which indicates highly transverse angle" dental pattern of the posterior teeth, mandibular movement (Butler, 1972, p. 478). resulting from the labial shift of the proto- Shear movement inclination of less than 40 conid, has been widely recognized as a degrees appears to be primitive for the Mam- shared derived character (or "synapomor- malia; an inclination of 40 to 50 degrees is phy" in cladistic terminology) uniting the therefore a derived character for the Dry- Theria (Osborn, 1907; Patterson, 1956; Mills, olestoidea, including Amphitherium. 1964; Kermack et al., 1968; Crompton and DENTAL FORMULA: Recently, there has Jenkins, 1968; Crompton, 1971, 1974; Mc- been renewed debate about the number and Kenna, 1975). This arrangement of cusps is homologies of postcanine teeth in the Theria derived with respect to the in-line cusps of (McKenna, 1975). If one accepts Kuehneo- morganucodonts and other triconodonts, and therium as the sister group of all remaining is not found in any other group of primitive Theria (), the primitive therian mammals. Crompton (1971) also pointed out dental formula is still open to question. Ker- that the development of broad wear facets mack and others (1968) stated that Kueh- joining the cusps, rather than the primitive neotherium could have had nine to 11 post- intercusp wear facets of triconodonts, is canine teeth, with five or six lower premolars 300 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

MANDIBLE LOWER MOLARS UPPER AND LOWER MOLARS lingual view lablial view crown view

A , A- /4

C

E Jaw fragmentary

F Upper molars unknown

I' 1981 PROTHERO: JURASSIC MAMMALS 301 and three to six lower molars. They did not four (Clemens and Mills, 1971). Spalaco- figure any specimen, however, that shows theriids and dryolestoids (including Amphi- molar alveoli, so it is not clear how they de- therium) have increased the number of mo- termined the number of molars. Their figure lars to at least seven, with eighth and ninth 8 indicates alveoli for six premolars, al- molars appearing variably in some dryoles- though the last two premolars have double tids. The late eruption of the eight and ninth roots and could also be molars. If six pre- molars (see p. 293) may explain this variation. molars is the primitive number for the Ther- North American paurodonts have only four ia, then reduction to five would be derived molars, but Kuhne (1968) reported pauro- for the Trechnotheria (McKenna, 1975). Re- donts with five or six molars. Perhaps this duction to three premolars is found in spa- discrepancy is also due to late eruption, and lacotherioid symmetrodonts and is therefore to the poor condition of the few known considered a synapomorphy for the group. North American paurodonts. In any case, Dryolestids have reduced the number of pre- the distinctive feature of the molar count is molars to four, and in North American pau- the increase from the primitive three to five, rodonts, it is further reduced to two. Kuhne to at least six or seven that occurs in spala- (1968) reported paurodonts with three, four, cotheriids, Amphitherium, dryolestids, and and five premolars, however. possibly some paurodonts. Amphitherium was originally described with four premolars, but Clemens (personal commun.) has reexamined the specimens LOWER JAW CHARACTERS and now agrees with McKenna (1975) that it MANDIBLE CHARACTERS: The primitive had five premolars. therian jaw morphotype can be seen in McKenna thinks Peramus had five pre- Kuehneotherium and is closely comparable molars, and reductions of this formula oc- to Morganucodon (?=Eozostrodon) and cuffed variously in the Tribosphenida. Cer- some (Kermack et al., 1973). tainly the teeth called MI by Clemens and Kuehneotherium has no angular process, but Mills (1971) and P5 by McKenna (1975) look possesses a large dentary trough for the ac- more like premolars than molars. If, how- cessory jaw bones (which are not preserved), ever, this hypothesis is falsified, then a re- a facet for the rudimentary coronoid bone, duction to four premolars would be a syn- a medial flange passing from the tooth row apomorphy for the cladotheres (excluding to the condyle, and a coronoid process at a Amphitherium; see fig. 11). The position of very low angle to the tooth row. The condyle Peramus as the sister group of the Tribo- is on, or slightly below, the level of the tooth sphenida is supported by several other char- row. All of these characters appear in Mor- acters (see fig. 12) and does not stand or fall ganucodon (?=Eozostrodon) and thus, by on the premolar-molar count. outgroup comparison, are plesiomorphic in The number of molars in Kuehneotherium Kuehneotherium. Kuehneotherium is more was given by Kermack and others (1968) as derived than Morganucodon (?=Eozostro- three to five, depending upon whether there don) in that the medial flange is continuous, are nine, 10 or 11 postcanine teeth. Amphi- without an interruption of the abductor fos- don and Tinodon have four molars, and Per- sa. amus has either three (McKenna, 1975) or Using Kuehneotherium as the sister group

FIG. 7. Comparison of representative taxa mentioned in text. A. Kuehneotherium. B. Tinodon (upper teeth of Eurylambda). C. Amphidon. D. (upper teeth of Peralestes). E. Spa- lacotheroides. F. Amphitherium. G. Paurodon (upper teeth of Pelicopsis). H. Laolestes (upper teeth of Melanodon). I. Peramus. I- lop!uoqdsoqu,L + e + I+ 0 snwlWacd + u t II 0 eK. C-. t- t cK- cK. ct- c- P qlool oJiaqUId OpJOd "it UOpOUOXOd v7vJ7 cK. cb cK-**et.c<-C cK-* e. C- 0 wnlJladtoqwV + + + + + + N ~~+ + - +± t bN + + o m + + cl. vnjuofvsnijD + en. + t oo + + + ;0 salsalov77 t-It 00 I- + + + + + t 00 + + -+ 0 0 SdisdOJSVtfJ, +I + + + + + +

salsaloday cK.cl. +0I + + + b +

sa,sajofia +- N o + + + 4) + t- coos +j + K - I+ "it + +It+ + + 0 + + xvlvdsvJiaJ L-. + C) lNl~+ + ++ t +-t-c UOpOdJ2V 0

t- + UOpOlflVJ el+ cI- +

U0o140wotaivq C'- e-+-c I++I +

uopoJnvd L--

0 cK.e,. I wnudatt2dw 0 +- + t + + W* + ±4)4 + + eo*et oo g cK.fK.- Q0 wnlnJdu,?1yoavjvd5 Sdp!O)UOpO4ljdWWS + I + e0n + II sap!oJdzfjo0tlvdS c~~.I c~~~.c e'i c~~. + c~~~. IclI~~1-0 + 8l- UOpOUIJ: ~~*c~~*~~ c~~* e~~. ~ C- C- WI) +~~~~~~ uopo.inp,uv,pV 4-00. o A uopnidwuv 'C +en~--+~ ~ c } r )4 wfnlhddflOduuldfn 0 o- .-0o'

4)4-44-4~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ 5- 4 0 '4) 4) -a 0 -4 E0

0 l0o0w 04 S Z _4 Ci e) u. oc o

302 iop!uoqdsoquj + -I + +- snwvJiag + + + + OJJOd qtool OJ!94U[qd + + + UOPOUOXOdtVItcI + + + uonlou,lo.qulvd wnuJatgojqwy + + +

vlyuofvsnij + + + +

+ + + + salsaIQo'2U + + sawsaIoday + + + + + +

xvlvdsvJia + + + + uopoavJv + + + UOpOfIW}pL + + +

UO,tujOdtyallv + +

uopornpid + +

+ + + wniiay,loivlvdg + + +

S9P!O1UOPO4dWWSSX + +

sdp!OJldLfjovIdVd + +

UOpOU.tL +

uopo.inpouvI +

uop!ydw,V + cl

wnuaylioapouzany +

tQ~~~~~~~~~~~~~~~~~~~~C°so 8 3sQoJ

* u £ b 0oo o o - _ _ _ _ _ N N N

303 304 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167 of the remaining therians, all Trechnotheria ther anterior, and in shape dissimilar to the share the loss of the medial flange extending angular process of any cladothere. Indeed, from the tooth row to the condyle, and show the long, slender, somewhat dorsally deflect- an increase in the angle of the coronoid pro- ed angular process seen in the Dryolestidae cess. This angle ranges from about 45 de- seems to be unique to that group; the angular grees in Peramus to 90 degrees in Tinodon, process is unknown in paurodonts. The an- but the distribution of this characters shows gular process of Amphitherium, Peramus, no consistent pattern within the Trechnoth- and the Tribosphenida is stout, triangular in eria. Trechnotheres share a condyle that is profile, and ventrally deflected, unlike that above the level of the tooth row. of dryolestids, but much farther posterior The condyles of Morganucodon (?= than the process in Morganucodon, Doco- Eozostrodon) (Kermack et al., 1973) and don, or any other non-therian. Thus, the Kuehneotherium (Kermack et al., 1968) are presence of a true angular process is consid- directed posteriorly and are dorsoventrally ered a synapomorphy for the Cladotheria. flattened. The condyle of Kuehneotherium The morphology of the dryolestid angle is partly broken (Kermack et al., 1968, fig. would then be a derived form of the primitive 10), but it has been reconstructed so that it cladothere angle. closely resembles Morganucodon (?= Symmetrodonts, lacking an angular pro- Eozostrodon) (Kermack et al., 1968, fig. 11). cess of any kind, have a modification in that Where the condyle is preserved in higher area that seems synapomorphous for the therian jaws, it has a distinct trochlea that is group. Tinodon, Amphidon, and Spalaco- usually dorsally or posterodorsally directed. therium (the only symmetrodonts with the This can be seen in Tinodon (Simpson, 1929, back of the jaw known) show a distinct pter- figs. 15 and 16), Spalacotherium (Osborn, ygoid crest that passes from the ventral bor- 1907, fig. 11), Laolestes (Simpson, 1929, fig. der of the jaw continuously into the condylar 25), Amblotherium (Simpson, 1929, fig. 26), process. This feature is unique within the Phascolestes, Peramus, and Amphitherium Theria, although , an am- (Simpson, 1928, fig. 35), and Dryolestes (Os- philestid triconodont, shows something like born, 1907, fig. 32). it. However, because the Theria are a mono- A trochlear condyle can also be seen in phyletic group based on other character dis- non-therians such as Docodon (Osborn, tributions, this feature would have to be de- 1907, figs. 20 and 21), but not in rived in parallel from the primtiive or Phascolotherium (Osborn, 1907, figs. 1 la Kuehneotherium-Morganucodon morpholo- and 6). The distribution of this character sug- gy. An adaptive-functional explanation might gests that the lack of a distinct trochlea on be suggested for this parallelism in jaws lack- the condyle is primitive for the Mammalia. ing angles, but will not be considered here. The condition in Docodon and in the Trech- Krebs (1969, 1971) has shown the exis- notheria would then be derived in parallel, tence of a rudimentary coronoid and splenial if the other character distributions are cor- in dryolestids from Guimarota mine. These rect in indicating that these taxa are not are primitively present in Kuehneotherium, closely related. For purposes of this discus- but absent in all Trechnotheria not from Por- sion, a distinct, posterodorsally, directed tugal. I have examined all the North Amer- trochlear condyle is considered a shared de- ican material with this problem in mind. No rived character for the Trechnotheria (see specimen that is adequately preserved shows fig. 12). any trace of a coronoid or splenial. Because Morganucodon (?=Eozostrodon) has a other characters establish that dryolestids feature labeled the angular process (Ker- are trechnotheres, and derived ones at that, mack et al., 1973), but Patterson (1956, p. two explanations are possible: (1) These 76) has given reasons for doubting its ho- primitive bones were lost or fused indepen- mology with the angular process of the Cla- dently in symmetrodonts, all dryolestoids dotheria. This process is certainly much far- except the Guimarota forms, and the Za- 1981 PROTHERO: JURASSIC MAMMALS 305 theria; (2) the coronoid and splenial are neo- However, the extreme inequality of dryoles- tenic retentions of the primitive condition in tid roots is an unambiguous synapomorphy the Guimarota specimens. Unfortunately, for the dryolestids. none of the Guimarota dryolestoids has been Another character found in all the Dry- described adequately or named, so it was im- olestidae is the unequal height of the alveolar possible to include them in the cladogram. border on each side of the molar. It is mark- Therefore, the distribution of this character edly deeper on the labial side (Simpson, has been left out of the cladogram pending 1929, fig. 24C), and at one time gave the the publication of full descriptions of the impression that the lingual and labial views Guimarota taxa. of Amblotherium molars were of different In most primitive Theria the mandible is animals. This character is not found in pau- long and slender, with a long symphysis. rodonts or Amphitherium, so it seems to be Three genera of paurodonts (Araeodon, a shared derived character for the Dryoles- Paurodon, and Archaeotrigon) develop a tidae. jaw that is much shorter and stouter, with a TALONID DEVELOPMENT: Kuehneother- short symphysis. This character is unique to ium has a distinctly expanded posterior cin- the group, and is interpreted here as a syn- gulum which is larger than the anterior cin- apomorphy for them. The fourth genus of gulum and seems homologous with the paurodont, Tathiodon, appears to retain the talonid. It has a single cusp which Kermack primitive slender condition of the mandible. and others called the hypoconulid, but which In other characters (see fig. 8 and table 4), Freeman (1976) considered the hypoconid. Tathiodon is a good paurodont. It is there- The talonid is not basined, and slopes away fore the primitive sister group of the remain- from the hypoconid crest. A posterior cin- ing three genera. gulum of some sort is primitive for the mam- malian lower molar, and the enlargement LOWER TOOTH seen in Kuehneotherium seems to be derived CHARACTERS for the Theria. RoOTS: The Dryolestidae have long been With this polarity established, the sym- recognized (Simpson, 1929, fig. 24D) by their metrodonts all share a reduction of the tal- unequal roots in the lower molars, the small- onid until it is equal to the anterior cingulum er of which is posterointernal and supports in size, or lacking. Indeed, talonid reduction the reduced talonid. This feature is unique to in this manner gives their molars the sym- the Dryolestidae, excluding the paurodonts. metrical appearance that is responsible for According to Butler (1939, p. 334), Amphi- the name. The talonid of Tinodon is only therium has slightly unequal roots. Yet slightly smaller than that of Kuehneotherium Simpson (1928, p. 117) stated that Amphith- (Crompton and Jenkins, 1967, 1968), but the erium has subequal roots, "although in pos- talonid is highly reduced in the Spalacother- terior molars the posterior root does not ex- iidae, and absent in the Amphidontidae. tend quite so far externally as does the In paurodonts, the talonid is larger than in anterior one." Butler also reported that two Kuehneotherium, but it becomes broad and paurodonts (YPM 13775, 13778) have shelflike. The cusp on the talonid is present, "slightly unequal roots." Unfortunately, but becomes progressively reduced in size, YPM 13775 has apparently been lost and and is lost in Paurodon. Paurodon and Ar- YPM 13778 is broken so that the roots are chaeotrigon are united by having a talonid no longer present. Most of the specimens of that is semicircular in crown view. A trian- paurodonts examined by the writer have gular outline is the primitive condition be- equal roots, but YPM 13776 has a slightly cause it is found in Amphitherium, dryoles- larger anterior root. If Butler's observations tids, and the paurodonts Tathiodon and are correct, then slightly unequal roots Araeodon. would unite Amphitherium and some pau- In the dryolestids, the talonid is slightly rodonts as sister groups of the Dryolestidae. larger than in Kuehneotherium, but it be- 306 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167 comes anteroposteriorly compressed along 9). The anterior cingulum persists in all sym- with the rest of the tooth. In most dryolestids metrodonts, in Peramus (as a cusp), and in it is a low shelf on the posterior side of the many Tribosphenida. However, Palaeoxon- trigonid, triangular in crown view, with the odon and the Porto Pinheiro mammal have hypoconid at the posterolabial apex. There apparently lost it, insofar as can be deter- is usually a ridge (possibly homologous to mined from single, isolated teeth. It is also the cristid obliqua) from the base of the tri- lost in all dryolestoids except Crusafontia gonid terminating at the hypoconid. Devel- (Henkel and Krebs, 1969). The lower molar opment of this characteristic talonid is hy- lingual cingulum is lost in amphidontids (see pothesized to be derived for the Dryolestidae. Position of Amphidontids, below) and in all Amphitherium presents a problem. Its tal- the Cladotheria. The loss of lower molar lin- onids are more expanded than those of the gual cingula seems to be a good synapomor- dryolestids, and in this regard it seems to be phy for the Cladotheria. more closely related to the . The de- Primitive non-therians such as Eozostro- velopment of a cristid obliqua is also much don (?=Morganucodon) and stronger than in the dryolestids. This may be (Crompton and Jenkins, 1968) have several in conflict with other characters that unite cuspules on the lower molar lingual cingu- Amphitherium with the Dryolestida (new lum. The largest of these, cusp "g" of rank; see Position of Amphitherium). If a Crompton and Jenkins (called the "Kiihne- broad talonid is hypothesized to be primitive cone" by Parrington, 1967) is lost in British for the Cladotheria, then the shelflike pau- (but not French-see Sigogneau-Russell, rodont condition and the secondary reduc- 1978) specimens of Kuehneotherium and all tion in dryolestids would be modifications of higher Theria. The loss of the "Kuhnecone" the Amphitherium condition. Amphitherium therefore unites British Kuehneotherium and also shows a unique character in the exten- the trechnotheres. sion of its talonid over the base of the tri- Primitively, there is no labial cingulum gonid of the succeeding molar (Mills, 1964; on the lower molars. It develops in the sym- Clemens and Mills, 1971, p. 106). This char- metrodonts Spalacotherium and Symmetro- acter is not seen in any other therian, and is dontoides, and in the dryolestids Amblo- therefore an autapomorphy which excludes therium, Crusafontia, Laolestes, and Amphitherium from ancestry of any other Kepolestes. form (Clemens and Mills, 1971, p. 108). MOLAR INTERLOCK: The lower molars of Palaeoxonodon (Freeman, 1976) shows Kuehneotherium show an indentation on the further development of the talonid. It is shal- posterior rim of the cingulum, which re- lowly basined and has an incipient hypocon- ceives the main cusp of the talonid. This ulid and entoconid. Peramus has a fully ba- "tongue-in-groove" molar interlock is seen sined talonid with all three talonid cusps. in the type specimen of Tinodon bellus These characters unite Peramus with higher (YPM 11843) and, to a lesser extent, in the Theria and are synapomorphous for the lower teeth of Eozostrodon (?=Morganu- Zatheria. Krusat (1969) has described, but codon) and Thrinaxodon (Crompton and not named, a tooth from Porto Pinheiro, Por- Jenkins, 1968, fig. 9), , and tugal, that has three distinct talonid cusps (Crompton, 1974). The distri- but is not basined. Thus, it would seem to be bution of this character suggests that it is intermediate on the morphocline between primitive for the Mammalia, although it is Palaeoxonodon and Peramus. lost in the rest of the non-Theria. In the Ther- MOLAR CINGULA: Kuehneotherium has a ia, it is lost in all symmetrodonts except one strong anterior and lingual cingulum on its specimen of Tinodon and in all cladotheres lower molars, but no labial cingulum. This is except Amphitherium and Crusafontia primitive for the Mammalia, as seen in (Krebs, 1971). However, this character is Eozostrodon (?=Morganucodon) and Thri- variably developed in the taxa named above, naxodon (Crompton and Jenkins, 1968, fig. and succeeding teeth in the same jaw may 1981 PROTHERO: JURASSIC MAMMALS 307 differ with respect to interlock. In view of developed in Amblotherium, Crusafontia, this fact, multiple loss of this character Laolestes, and Phascolestes. seems a more parsimonious explanation of A character unique to the spalacotheriid its distribution within the Theria (see fig. 8). symmetrodonts (but still primitive in Am- TRIGONID CHARACTERS: Kuehneotherium phidon and Tinodon) is an extremely high has a moderately high trigonid, with a meta- crown, with long pointed cusps. This seems conid-protoconid-paraconid angle of about to be a synapomorphy for symmetrodonts 110 degrees. The metaconid is much lower minus Amphidon and Tinodon. than the paraconid, whereas it is as high or A characteristic feature of the dryolestoids higher than the paraconid in all other Theria, is the strongly procumbent paraconid. The and in outgroups such as Morganucodon primitive condition is seen in Kuehneother- (?=Eozostrodon). Therefore, this high para- ium, symmetrodonts, and Peramus, with the conid is autapomorphic for Kuehneotherium paraconid pointing dorsally or slightly an- and excludes it from the ancestry of the rest terodorsally. In paurodonts and dryolestids, of the Theria. In all other characters, how- it points very strongly anteriorly, at about a ever, Kuehneotherium seems to be primitive 70 degree angle to the vertical, metaconid for the Theria, and is used as the outgroup axis. The condition in Amphitherium is dif- for comparison. ficult to interpret. It appears to be more pro- The distribution of trigonid angles is cumbent than the primitive condition (Simp- shown in table 5. In symmetrodonts that son, 1928, fig. 38; Mills, 1964, fig. 3; have complete tooth rows, the trigonid angle Crompton and Jenkins, 1968, fig. 11), but this becomes progressively more acute posterior- is difficult to determine from sketches and ly. Tinodon and Symmetrodontoides have stereophotographs that differ greatly. For an M1 with cusps nearly in line. M3 of Tin- purposes of this discussion, it is considered odon is shown to have an angle of about 99 "moderately procumbent" and any sort of degrees (Crompton and Jenkins, 1967), and procumbent paraconid is a synapomorphy in Symmetrodontoides, M6 has an angle of for the Dryolestoidea, including Amphither- about 30 degrees (Fox, 1976). Patterson ium. (1956, fig. 13A) shows angles ranging from Paurodonts have further modified the 55 degrees in M1 to 45 degrees in M6 of Spa- paraconid so that it becomes shelflike, with lacotherium, so this trend is present, but not a reduced cusp. In the most derived form so pronounced. Unfortunately, Kuehneo- (Paurodon) the paraconid is completely flat therium has no jaws with teeth in them, so without a cusp. Paurodonts are also charac- the polarity of this character is difficult to terized by a reduction of the metaconid establish. Parrington (1971, 1978) showed a height relative to the protoconid. Tathiodon number of teeth of Kuehneotherium which shows the beginnings of this condition. It is have angles ranging from 180 degrees to al- most pronounced in Araeodon, Archaeotri- most 100 degrees. Presumably, the in-line gon, and Paurodon. cusps of the anterior molars are the retained Amblotherium and Crusafontia show a primitive condition, and the acute angle in character that is apparently unique to them. the posterior molars is derived. In this case, The paraconid and metaconid in these ani- all therians share a posterior trigonid angle mals are much more slender and rodlike, of 100 degrees or less (except the amphidon- rather than shelflike or conical as in the rest tids; see Position of Amphidontids). of the Theria. I have never seen the actual Dryolestids show an extreme anteropos- specimens of Crusafontia, but figure 1 in terior compression of the trigonids and nar- Henkel and Krebs (1969) shows this char- rowing of the angle, corresponding to the acter. general shortening of the tooth. This seems LOWER PREMOLAR CHARACTERS: Ker- to be a synapomorphy for the Dryolestidae, mack et al. (1968) showed some isolated pre- since it does not occur in paurodonts, Per- molariform teeth that are referred to Kueh- amus, or Amphitherium. It is most strongly neotherium. None of these teeth show either 00 t3p!uoqdsoqup ten + ,v _ + I+ + + + + _- _

snwvjac me + ; + e- - + + + +

!ZP1JPOOX UOPOUV1I4I t*ct-t-t-* + + ++ +++ ++ I

!UdMO UOpOUVDaJ4 + + + + +++ +++

snfvJnl.W snJqn,adlaH Cv- ++ ° o + ++ +++ +++

C- m ) Wwny snJrn,adJi H C-. et- + r^ 0 + + ++ cK. +++ +++

+ + +++ ++ I + SUVJQ1X(33.lI4I C-- - C co o + + + +++ +++ uopoujny -tC + ++ +++ I +++ t-C-cKI + wnuru ,swn!uayLjojqwV,, 0 + I + + + ++ I ++

wnualoylqwV tCK-b+YStI + + +++ +++

snfsvljln3 C- + + +++ +++ r 4 + sisdo3yact CKCK < oI + + +++ ++ *5- + s9ppoJayjo3PDvd,V C-o - t o + + II.

Cd + + + + salsailwaj C-\c -+ I It-I I+ + + + + q + + + t-C-t-t-N K-+ vpqwvuning 0 0

wnafloJyOauyand t-CKCt No +I + + + * S)

>- Cd~~ ~ ~ ~ ~~~'

'a2 '0a ia

- - 0 z_O0Wz to)_ _ ~ _ ~_ ~ 1981 PROTHERO: JURASSIC MAMMALS 309 a labial or a lingual cingulum, except for a sidered to be a left upper molar. Simpson tooth shown in figure 6. It is not clear wheth- (1971, p. 188) considered it a probable syn- er the cingulum on the tooth in figure 6 is onym of Kuehneotherium. Considering the labial or lingual. All of the figured premolar- preservation and the isolated nature of all the iform teeth have an anterior cingulum or Rhaetic fissure-fill teeth, it is placed with cusp of some sort. Kuehneotherium in the cladogram. Its distin- There is no lingual cingulum on the last guishing characters, if they are valid, appear lower premolar of Amphidon or Tinodon. A to be autapomorphous. lingual cingulum occurs (where known) on Eurylambda is based on an upper tooth the last lower premolars of spalacotheriids, referred to Tinodon by Simpson (1929). This Amphitherium, and dryolestids except Cru- is corroborated by the work of Crompton safontia. It is absent from Peramus and and Jenkins (1967). It lacks a trigon basin primitive Tribosphenida. An interpretation and has a trigon angle of about 160 degrees of the distribution of this character is shown (Crompton and Jenkins, 1967, fig. lE). The in figures 8 and 10. metastyle cusp, although damaged, appears The anterior cingulum or cusp on the last to be similar in size to that of Kuehneoth- lower premolar is lost in all Cladotheria ex- erium. Eurylambda retains the primitive cept Peraspalax and Dryolestes. This loss is multiple wear facets on the metacrista interpreted as a derived character for the cla- (Crompton, 1971). It has one character, how- dotheres, and the condition in Peraspalax ever, that is derived and shared with the rest and Dryolestes would therefore be second- of the Theria. In Eurylambda, there is no arily regained. sign of a parastyle-metastyle interlock, al- though the lower teeth (Tinodon) still retain an interlock. UPPER TOOTH CHARACTERS Most of the remaining symmetrodonts and Kuehneotherium (Kermack, Kermack and cladotheres are known from more than iso- Mussett, 1968) is also known from isolated lated upper teeth. Peralestes, considered by upper teeth; in all respects these upper teeth Simpson (1929) to be the upper teeth of Spa- show the primitive morphology for the Ther- lacotherium, shows progressively more ia. Kuehneotherium still retains a complete acute trigon angles on the more posterior cingulum around the tooth, a high central molars (see Patterson, 1956, fig. 13A). Upper paracone, a large metastyle cusp and a para- tooth rows are unknown from the other sym- style tab that interlocks with a groove be- metrodont taxa. Judged from the lower tween the metastyle and metacone. These teeth, however, the retention of the obtuse are primitive for the Mammalia, as demon- angle in the anterior molars is primitive, and strated by outgroup comparison with Eozos- the acute angle of the posterior molars is de- trodon (?=Morganucodon) and other forms rived for all therians except Eurylambda and (Crompton and Jenkins, 1968, fig. 2). Other Kuehneotherium. According to Crompton primitive characters include presence of only and Jenkins (1967), Eurylambda is probably two molar roots, retention of wear facets A based on an M1, so that its posterior molars and B (Crompton, 1971) on the metacrista, could have been more acute (as suggested by paracrista and metacrista poorly developed the lower molars in Tinodon). This condition (i.e., cusps still high and pointed), and lack may even occur in Kuehneotherium. Ker- of any trigon basin. Kuehneotherium shows mack et al. (1968, fig. 2) show an upper mo- the reversed triangle dentition and conse- lariform tooth with a more acute angle than quent labial deflection of the stylocone and the type upper molar. Acute posterior trigon metacone, which is a shared derived char- angles, then, may be a shared derived char- acter for the Theria. acter for the Theria as a whole, but in the Kuehneon ("Duchy 33" of Kiihne, 1950; absence of better evidence, this character named by Kretzoi, 1960) is a poorly pre- can only be used to unite Peralestes with the served tooth that Kermack et al. (1968) con- Cladotheria. 310 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

The presence of a trigon basin, and the dition is only true of the penultimate molar corresponding development of a paracrista (?M3 or M2). The molar anterior to it (M2 or and metacrista, appears in Peralestes, Spa- ?M1) has a metacone shearing the groove be- lacotheroides, and the cladotheres, but not tween the paraconid and the hypoconid, pro- in Eurylambda or Kuehneotherium. If Eu- ducing facets on each (Crompton, 1971, fig. rylambda is based on an M1, this character 8C). This is distinct from the condition in would not be very pronounced, and might Kuehneotherium, in which cusp c sheared have existed in more posterior molars. For the notch between the paraconid and proto- the purposes of the cladogram, this character conid and the talonid had not yet expanded can only be used to unite the symmetrodonts posteriorly to meet it (Crompton, 1971, fig. (excluding Eurylambda) and cladotheres. 8A). METACRISTA CHARACTERS: "CUSP C" VS. The source of controversy lies in the re- METACONE: Crompton (1971) has analyzed construction of the unknown upper teeth of the wear facets of the primitive Theria and Amphitherium. The lower teeth have facets concluded that Kuehneotherium, Eurylamb- that show the Kuehneotherium condition on da, and Amphitherium retain facets A and B anterior teeth and the Peramus condition on on the metacrista. The upper molars of Am- posterior teeth (Crompton, 1971, fig. 6C). phitherium are unknown, however, so this is Crompton (personal commun.) considers it inferred from the corresponding wear on unlikely that a metacone could shift occlusal lower molars. Presumably, the loss of these relations in the same jaw, i.e., shear the facets is derived for all Theria except these paraconid in the anterior part of the jaw and three taxa. the hypoconid in the posterior part. There- Crompton (1971) has recently cast doubt fore he postulated (1971, fig. 6) a metacone on the homologies of the main metacrista arising de novo in front of cusp c on the up- cusp. According to his argument, in primi- per teeth of Amphitherium. In his scenario, tive Theria this cusp is not the metacone of the metacone would enlarge and cusp c all higher Theria, but what he has labeled would reduce until it is barely visible on a "cusp c" in Kuehneotherium. Cusp c shears single tooth of Peramus, and absent else- down a notch between the paraconid and where. protoconid of Kuehneotherium, giving rise An alternative hypothesis might also be to facets "2" and "A" (Crompton, 1971, fig. suggested. The posterior expansion of the 7C). The paraconid shears up the notch be- talonid and posterolingual retreat of the para- tween cusp c and the paracone, producing conid in posterior molars of Amphitherium facets "A" and "B" in figure 7C. would naturally cause a change in occlusal In Peramus, the condition is quite differ- relations. The posterior metacrista cusp that ent. The paraconid has retreated posterolin- shears down between the protoconid and gually, so that only part of the crista between paraconid would be "captured" as the para- the metacone and the metastyle shears conid retreats and the hypoconid rises in its against it. On the penultimate molar (?MN, or place. This scenario is more parsimonious M! according to McKenna, 1975) of the sole than one involving de novo addition of the specimen of Peramus upper teeth (BM[NH] metacone, and has been suggested by Butler M21887), there is a minute cusp between the (1972). metacone and metastyle. Crompton called Crompton (personal commun.) objects to this cusp c and homologized it with the large the latter hypothesis for the following rea- metacrista cusp of Kuehneotherium, since sons: its occlusal relations are similar. Due to the (1) It involves a shift in occlusal relation expansion of the talonid, the true metacone for the main metacrista cusp. in Peramus does not shear the paraconid, (2) Cusp c still remains on Peramus. but its anterolingual side shears against the (1) The degree of change in the molars of posterolabial edge of the talonid (Facet 4, Amphitherium is quite striking, so that a Crompton, 1971, fig. 5). However, this con- shift in occlusal relations seems quite plau- 1981 PROTHERO: JURASSIC MAMMALS 311 sible. Such shifts do occur; Crompton him- STYLAR REGION: All primitive Theria ex- self documented an example (Peramus) in cept Kuehneotherium and Eurylambda the same paper. Indeed, if one postulates show a noticeable enlargement of the stylar both a metacone and cusp c on the metacris- region as the trigon angle becomes narrower. ta of Amphitherium, then the metacone must In most of these forms, the stylar regions of change its occlusal relations (i.e., gradually adjacent teeth contact each other. The para- come in contact with the hypoconid on more style projects anterolabially and usually con- posterior teeth). Thus, the problem has not tacts the posterolabial part of the metastyle. been eliminated, but instead becomes more In a few forms the parastyle becomes ex- complex. tremely hooked and closely interlocked with (2) "Cusp c" in Peramus is so insignificant the preceding metastyle. Related to the con- as to have been ignored by Clemens and dition is the slightly indented stylar margin, Mills (1971, fig. 2 and text) in their analysis. which seems to be primitive for all Theria Equating this small metacrista swelling on a except Kuehneotherium and Eurylambda. single tooth of a single specimen of Peramus In Herpetairus and Melanodon goodrichi with the well-defined cusp c on all teeth of the outer margin is straight. In Euthlastus Kuehneotherium does not seem warranted. and Comotherium, on the other hand, the Crompton (personal commun.) claims that labial margin is so indented as to give the , Prokennalestes and Cimo- crown a heartlike shape. Both of these con- lestes also show cusp c. Ifthe figures ofthese ditions are interpreted here to be indepen- taxa are accurate, the metacrista cusps are dently derived, since they conflict with all barely worth mentioning, let alone being dig- the remaining character distributions of nified with the "cusp c" homology. these taxa (see fig. 9). It is impossible to settle this dispute until Clemens and Mills (1971, p. 111) have the upper teeth of Amphitherium are found. pointed out that the stylar region, stylocone, For the purposes of this paper, I shall utilize and posterior cingulum of the upper molars Occam's razor and consider "cusp c" ho- of Peramus are far more reduced than in mologous with the metacone. If Crompton's higher Theria, such as Pappotherium and hypothesis is correct, then the metacone oc- Holoclemensia. These characters are appar- curs only in Amphitherium, Peramus, and ently autapomorphous and exclude Peramus the Tribosphenida. It would then be a syn- from the ancestry of the Tribosphenida. apomorphy for these animals (as in fig. 11C, Primitively, the metastyle is large with a this paper). Other characters (see fig. 1IB) large single cusp, as in Kuehneotherium and make this hypothesis less parsimonious than Eurylambda. The advanced forms show a the one adopted here. considerable reduction of the metastyle and If cusp c is equivalent to the metacone, especially its apex. In Miccylotyrans, Como- Peramus and the dryolestoids all retain the therium, Herpetairus, and Melanodon the primitive metacrista shear facet between the metastyle is divided by a valley running an- metacone and metastyle, seen in Kuehneo- terolingually-posterolabially. Distinct cusps therium. The derived condition is seen in are present on each side of the metastyle val- spalacotherioid symmetrodonts, which ex- ley in Comotherium, Herpetairus, and Me- tend this facet to the paracone (Crompton, landon. These conditions are interpreted 1971, p. 85). here as derived conditions uniting these taxa. The metacone undergoes reduction and is TRIGON CHARACTERS: As Crompton and sometimes lost due to wear in all the dry- Jenkins (1968, p. 455) have pointed out, sym- olestoids except Pelicopsis. Because Peli- metrodonts (excluding Eurylambda) are de- copsis is considered to be based on the upper rived in that their trigons are expanded ver- teeth of a paurodont (Simpson, 1929), the re- tically and increase the paracone height. duction of the metacone would then be a Dryolestoids, on the other hand, expand shared derived character for the Dryolesti- their molars transversely (or compress them dae. anteroposteriorly, in other words). Both of 312 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167 these conditions appear to be derived inde- cusp, about equal in size to the metacone pendently from the primitive condition, and and located on the paracrista. This is modi- neither is exhibited by Peramus. The Tri- fied in several ways in the dryolestids: bosphenida make their upper molars trans- (1) Primitive dryolestids have this stylo- versely wider by addition of the protocone, cone morphology, with some reduction in which is different from the mechanism em- height, along with the reduction of the meta- ployed by the dryolestoids. cone. In Comotherium, Herpetairus, and The trigon of all the dryolestoids except Melandon, however, the stylocone is dis- Euthlastus and Pelicopsis has a high, dis- tinct and conical, projecting high above the tinct paracrista and metacrista that gives it labial margin of the tooth, and much higher a basined appearance. This seems to be a than the metacone. shared derived character for the remaining (2) In Herpetairus and Melanodon, the dryolestoid taxa. stylocone shifts posteriorly, so it is placed in In a number of dryolestoids, there is a the middle of the tooth and no longer on the transverse ridge on the trigon, which is not paracrista. found in any other group. It appears faintly (3) In Comotherium and Melanodon ow- on MP of Amblotherium nanum (Simpson, eni, the stylocone shifts lingually to the sty- 1928, p. 137) and is distinct on most molars lar margin. This character is fairly equivocal, of Kurtodon, Miccylotyrans, and Herpetai- though, since it is a function of the size of rus humilis. It is very pronounced on all the parastyle and can be less pronounced in species of Melanodon, and this degree of more anterior molars. development is diagnostic for the genus. (4) All species of Melanodon have a very However, it is absent in Pelicopsis, Euthlas- large stylocone which dominates the trigon. tus, Amblotherium pusillum, Comotherium, This character is diagnostic for the genus. and Herpetairus arcuatus. In view of this The most parsimonious distribution of sty- and other character distributions (see fig. 9 locone characters is shown in figure 9. This and table 5), it seems most parsimonious to requires parallelism in the lingual stylocone interpret this ridge as a synapomorphy for shift (character 3 above), but, as noted, this Amblotherium nanum, Kurtodon, Miccylo- character is often ambiguous. tyrans, Herpetairus, and Melandon, with CINGULUM CHARACTERS: A labial cingu- secondary loss in Comotherium and Her- lum is present on the upper molars of Kueh- petairus arcuatus. In keeping with this char- neotherium and primitive non-therians but acter distribution, Amblotherium nanum, lost in all more derived therians. A lingual Kurtodon, Miccylotyrans, Comotherium, cingulum is primitive for therians, since it Herpetairus, and Melanodon show a much occurs in Eozostrodon, Kuehneotherium, narrower trigon not seen in Pelicopsis, Euth- Peramus, and the Tribosphenida (where it lastus, or Amblotherium pusillum. This im- eventually develops a protocone). However, plies that Amblotherium nanum is more both dryolestoids and symmetrodonts (ex- closely related to the more derived dryoles- cept Eurylambda) have lost the lingual cin- tids than it is to A. pusillum. Therefore, Am- gulum. The loss of the upper molar lingual, blotherium (as presently constructed) is a cingulum might be used to unite these paraphyletic genus. Owen (1971) originally groups, but this character is in conflict with named it Achyrodon nanus and this was syn- a number of others (see Symmetrodont onymized with Amblotherium by Simpson Monophyly and below). (1928, p. 120). Since A. pusillum is the type species of the genus, "Amblotherium" nan- CLADOGRAM DISCUSSION um requires a different generic name if tax- The cladograms in figures 8 and 9 are not onomists revising the British material cor- strictly congruent. This is because fewer roborate this cladogram. taxa are known from upper teeth than from STYLOCONE CHARACTERS: In primitive lower teeth, and only a few taxa are known therians, the stylocone is a wedge-shaped from both (Kuehneotherium, Amblotherium, 1981 PROTHERO: JURASSIC MAMMALS 313

FIG. 8. Cladogram of taxa known from lower teeth and jaws. See table 6 for explanation.

Peramus, and the Tribosphenida3). Eury- grams are compared, one major difference lambda and Peralestes are believed to be- can be seen: there are no synapomorphies in long to Tinodon and Spalacotherium respec- the upper teeth that unite the symmetrodonts tively, and this is assumed in the composite (Tinodon-Eurylambda, Spalacotheroides and cladogram (fig. 12). Patterson (1956, p. 10) Spalacotherium-Peralestes). However, this referred an upper molar to his lower molar may be due to the fact that Eurylambda is genus Spalacotheroides (Patterson, 1955). based on an isolated M1 and no posterior The suggestion that Melanodon and La- molars are known. olestes might be synonymous (Clemens and Lees, 1971) is not followed here for reasons given above. In particular, the critical genus SYMMETRODONT MONOPHYLY Amphitherium has no known upper teeth. Because symmetrodonts are not just iso- This makes the composite cladogram more lated upper teeth, but animals with both up- difficult to construct (see Position of Am- per and lower dentitions, characters of both phitherium, below). When the two clado- must be combined in evaluating their possi- ble monophyly. Four possible hypotheses of 3Freeman (1976) referred an isolated upper tooth to relationships are suggested: Palaeoxonodon, but published figures and descriptions (1) Tinodon plus Amphidon are the sister are inadequate to make any meaningful comparisons or group of the spalacotherioids plus clado- to list it in table 5. theres (fig. IOA). 314 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

TABLE 6 Explanation of characters in Figures 8 and 9 (* indicates parallelism) FIGURE 4: LOWER TOOTH CHARACTERS FIGURE 5: UPPER TOOTH CHARACTERS 1. Reversed triangle molar pattern, single wear facet 1. Reversed triangle cusp pattern, single wear facet on on anterior trigon-posterior trigonid, broad wear anterior trigon, broad wear surfaces joining cusps surfaces join cusps, lose Kuhnecone 2. Lose parastyle tab interlock 2. Coronoid angle increases, lose medial condylar 3. Lose metacrista facets A and B, posterior trigon flange, reduce to five premolars, angle of posterior angle acute, trigon basin formed, reduce metastyle molar trigonids less than 100°, condyle above level 4. Metacrista shear facet extends to paracone, molars of tooth row, trochlear condyle develops increase crown height, lose lingual cingulum on mo- 3. Greatly reduced talonid, strong internal pterygoid lars* ridge, reduce to four premolars* 5. Stylar regions enlarge and interlock, molars expand 4. Reduce metaconid and paraconid to crenulations on transversely, angle of anterior molar trigons less cristids, angle of trigonids about 140° than 100°, metacone with facet from hypoconid 5. Reduce to three premolars shear, three roots on molars 6. High slender trigonids, lose paraconid facets A and 6. Reduce to three molars, reduce stylocone B (single paraconid wear facet)* 7. Reduce to four premolars, lose lingual cingulum on 7. Increase to seven molars*, molars acquire labial molars*, paracrista shear becomes transverse cingulum*, continuous lingual cingulum on last pre- 8. Reduce metacone molar* 9. Distinct paracrista and metacrista, higher paracone 8. Lose lingual cingulum on molars, lose anterior cusp 10. Narrower trigon, weak transverse ridge on M5 on last premolar, anterior molar trigonid angle re- 11. Distinct transverse ridge on most molars duces to less than 100°, develop true angular pro- 12. Bifurcated metastyle with incipient cusps cess, expand talonid, hypoconid wear facet for 13. Conical stylocone higher than metacone, metastyle metacone with two cusps 9. Slightly unequal molar roots, continuous lingual cin- 14. Straight labial margin, stylocone shifts posteriorly gulum on last premolar*, more than six molars*, lose anterior cingulum on molars*, slightly procum- bent paraconid (2) Amphidon, Tinodon, and the spala- 10. Transverse metacristid shear, strongly procumbent cotherioids are a monophyletic sister group paraconid, reduce to four premolars* of the cladotheres 11. Reduce to two premolars, broad shelflike paraconid (fig. lOB). and talonid, reduced paraconid and talonid cusps, (3) Spalacotherium, Spalacotheroides, metaconid slightly shorter than protoconid Symmetrodontoides, and Amphidon plus 12. Metaconid much shorter than protoconid, jaw short Tinodon are progressively more primitive and stout with short symphysis, last lower premolar sister groups of the Cladotheria (fig. lOC). loses anterior cingulum (4) Spalacotherium plus dryolestids are a 13. Talonid semicircular in crown view monophyletic group, and the sister group to 14. Anterior molar root much larger than posterior root, the Zatheria (fig. lOD). trigonid and talonid anteroposteriorly compressed, Figure 10 and table 7 summarize the situ- lingual alveolar border lower than labial border, an- ation with respect to character conflicts in gular process slender and dorsally deflected the symmetrodonts. Characters 5 and 15. Anterior cusp on last premolar* 6 must 16. Narrower trigonids be parallelisms under Hypotheses 1, 2, and 17. Labial cingulum on molars* 3, because uniting dryolestoids and Spala- 18. Bifid metaconid cotherium (Hypothesis 4) generates the 19. Slender metaconids and paraconids, three mental greatest number of conflicts. Characters 2 foramina (Henkel and Krebs, 1969, p. 460) and 3 argue for symmetrodont monophly 20. Incipient hypoconulid and entoconid (Hypothesis 2), and 7, 8, and 9 against it 21. Strong hypoconulid and entoconid (Hypotheses 1 and 3). But Hypothesis 3 has 22. Basined talonid, reduce to three molars almost as many conflicts as Hypothesis 4. Hypotheses 1 and 2 each have five instances 1981 PROTHERO: JURASSIC MAMMALS 315

FIG. 9. Cladogram of taxa known from upper teeth. See table 6 for explanation.

of parallelism, and are therefore equally par- name. It is not found in any other primitive simonious. The distribution of character 15 Theria. (loss of upper molar lingual cingulum) con- (3) Character 7 (loss of metacrista-para- flicts with all the hypotheses above. This ap- conid facets A and B occurs in spalacother- parent parallelism was left off figure 10 to ioid symmetrodonts as a result of the meta- reduce clutter. crista shear extending to the paracone Since Hypotheses 1 and 2 are equally par- (Crompton, 1971, p. 85). In cladotheres, simonious, they should be expressed as an these facets are lost, but the metacrista shear unresolved trichotomy, to be tested by fur- never goes beyond the metacone. Amphith- ther characters or by new specimens. For the erium, a good cladothere by other charac- present, I consider Hypothesis 2 (symmetro- ters, has lost these facets on it its posterior dont monophyly) to be the most parsimoni- teeth, but retains them on anterior teeth. ous for the following reasons: Therefore, this character seems to develop (1) Character 2 (strong internal pterygoid in different ways and may not be homolo- ridge) is very distinctive and is unique within gous. the Theria. A case for splitting up the (2) Character 3 (highly reduced talonid) is is not convincing. I therefore take the con- the diagnostic derived character that gives servative course and recognize the Symme- the group the symmetry responsible for the trodonta as a monophyletic group (including A A T S. Sm D B A T S. S. D Z 1 _

2 m

3

4

5

6 I?

7

8

9 ?

10 -14

C D-

HYPOTHESIS 3 HYPOTHESIS 4

FIG. 10. Four hypotheses of symmetrodont relationships. See p. 313. Taxa are symbolized as fol- lows: A = Amphidon; T = Tinodon; Ss = Spalacotheroides; Sm = Spalacotherium; D = dryolestoids; Z = Zatheria. Numbers correspond to character states in table 7. Primitive character states are shown by open bar; derived states by solid bar. Question mark indicates condition is unknown. 1981 PROTHERO: JURASSIC MAMMALS 317

TABLE 7 Character Distribution in Symmetrodonts (see figure 10) Spala- Spala- Amphi- Tino- cothe- cothe- Clado- Character don don roides rium theria Notes 1. Reduce to three premolars - + ? + _ (two in some paurodonts) 2. Strong internal pterygoid + + ? + ridge on mandible 3. Highly reduced talonid + + + + 4. Trigons very high, pointed - - + + 5. Increase to seven molars - - ? + +1- (in some dryolestoids) 6. Plast with continuous - - - + +/- (in some dryolestoids lingual cingulum 7. Lose metacrista-paraconid - - ? + + (except Amphitherium) facets A and B 8. Anterior trigon-trigonids - - + + + (except Symmetro- more acute, with basin dontoides) formed in trigons 9. Reduced metastyle cusp ? - + + + 10. True angular process - - - + 11. Upper molars expand - - - - + transversely 12. Stylar region enlarges - - - - + and interlocks 13. Hypoconid shears metacone - - - - + 14. Lose anterior cusp on last - - - + lower premolar 15. Lose upper molar lingual ? + - +1- (lost in dryolestids) cingulum the amphidontids; see below). This decision talonids and a strong pterygoid ridge on the is by no means final, but simply a challenge mandible. for other workers to test it further. The dental formula (Pl_4M1-4) may be primitive, although five premolars seems more likely as the primitive trechnothere POSITION OF AMPHIDONTIDS condition (see p. 299). If so, then reduction to Amphidon superstes is known from a sin- four or less premolars is a synapomorphy for gle specimen (YPM 13638), a right ramus symmetrodonts, and reduction to three pre- with alveoli for three premolars, and crowns molars unites the non-amphidontid symme- of the fourth premolar and four molars. Its trodonts (see fig. 8). systematic position has been controversial, The most puzzling character of Amphidon because it has been used in the dispute be- is that the molar "cusps" are no more than tween the "cusp rotation hypotheses" and swellings or crenulations on the cristids. This the "in situ hypothesis" of the origin of the makes the teeth functionally unicuspid. tritubercular molar pattern (Crompton and Small anterior and posterior cingula are pres- Jenkins, 1968, p. 447). Simpson (1925a, 1929) ent, but there is no lingual or labial cingulum. assigned Amphidon to the Symmetrodonta Simpson (1925a) interpreted this condition as on the basis of its general morphology. It evidence of incipient cusp formation on the does share two unique derived characters cristids. Crompton and Jenkins (1968, p. 448) with other symmetrodonts: greatly reduced have pointed out that the specimen is heavily 318 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167 worn, and the cusps may have been reduced on the morphocline between the Theria and by wear. Outgroup comparison with Kueh- outgroups with in-line cusps on posterior neotherium and triconodonts indicates that molars. However, in-line cusps on anterior this condition is due to secondary reduction. molars are primitive for the Theria; in this Thus, its condition would be derived with respect Amphidon is more derived than respect to the remaining Theria, rather than Kuehneotherium, Tinodon, and Symmetro- a retention of a primitive unicuspid reptilian dontoides. In the face of these contradictions tooth. and other character distributions (see fig. 8 In addition to its cusp reduction, Amphi- and table 4), both of these conditions are in- don is highly modified in other features. It terpreted here as autapomorphic for Amphi- has lost the lingual cingulum on the lower don. molars and the molar interlock. The fourth Manchurodon (Yabe and Shikama, 1938) molar is highly reduced in both size and mor- is a poorly preserved jaw from the Husin phology. In view of the amount of wear and coal field of Manchuria. Yabe and Shikama preservation of the single specimen of Am- considered it to be late Jurassic in age, but phidon, these characters would appear to be Patterson (1956, p. 29) believed it to be early autapomorphies. Cretaceous. The description and figures are The angle of the trigonid has led to some so poor that it is impossible to say much difficulties in determining the systematic po- about the specimen, except that it shares the sition of Amphidon. Amphidon has a trigonid molar morphology of Amphidon. Yabe and angle on M1 of about 140 degrees. M23 seem Shikama gave a dental formula of P1_3M,-5, to become only slightly more acute-angled, but Patterson (1956, p. 29, note 2) gave rea- and M4 is so reduced that there is no angle. sons for interpreting the formula as P1_4M14. Thus, it does not show the primitive therian The latter formula would agree with that character of a posterior trigonid angle of 100 of Amphidon. For purposes of this discus- degrees or less. This would place Amphidon sion, Manchurodon is considered the sister

TABLE 8 Character Distribution in Cladotheres (see figure 11) Amphi- Dryo- Character therium lestoids Zatheria Notes 1. Lose lower molar interlock - + + (also in symmetrodonts, some non-therians) 2. Lose metacrista-paracristid +/- + + (also in symmetrodonts) facets A and B 3. Broad talonid + - + 4. Posteroventrally deflected + - + angular process 5. Last lower premolar with + + - (also in Spalacotherium) continuous lingual cingulum 6. Six or more molars + + - (also in Spalacotherium; not in paurodonts) 7. Molar roots unequal + + - 8. Lose anterior cingulum on + + - (also in Palaeoxonodon, molars Porto Pinheiro tooth) 9. Procumbent paraconids + + - 10. Shear inclination greater + + - (secondarily in later than 40° therians) 1981 PROTHERO: JURASSIC MAMMALS 319

A B C

2 2 2

3 3 3

4 4 4

5 5 5

6 6 6

7 7 7

8 8 8

9 9 9

10 10 10

HYPOTHESIS 1 HYPOTHESIS 2 HYPOTHESIS 3 FIG. 11. Three hypotheses of cladothere relationships. Numbers correspond to character states in table 8. Primitive character states shown by open bar; derived states by solid bar.

group of Amphidon, until reexamination of 1 iB). This hypothesis has been suggested by the specimen reveals more diagnostic infor- McKenna (1975). mation. (3) Amphitherium is more closely related to zatherians than it is to dryolestoids (fig. 1 IC). This point of view has been expressed POSITION OF AMPHITHERIUM by Hopson and Crompton (1969, fig. 12), Amphitherium is known from lower jaws Crompton and Jenkins (1968, fig. 11), and only. There has been a considerable diver- Crompton (1971). sity of opinion about its systematic position, The character distributions are shown in although it is most commonly considered table 8 and the results are apparent from fig- "on the main line of therian evolution" ure 11. Amphitherium is the sister group of (Mills, 1964). This and other similarly vague the dryolestoids (Hypothesis 2), since it assertions can be stated more precisely as shares at least six good characters with follows: them, and this requires less parallelism than (1) Amphitherium is the sister group of either Hypothesis 1 or Hypothesis 3. Char- dryolestoids (Dryolestidae plus Paurodonti- acter conflicts 1 and 2 are discussed above. dae) plus Zatheria (Fig. 1lA). This hypoth- As in the case of symmetrodonts, they seem esis corresponds to a cladistic interpretation to be most parsimoniously explained as par- of Parrington's (1971, fig. 15) phylogeny. allelism. Character 3 (expanded talonid) is (2) Amphitherium is more closely related the only good character that contradicts Hy- to dryolestoids than it is to zatherians (fig. pothesis 2. However, it seems most parsi- 320 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

FIG. 12. Hypothesis of relationships of the non-tribosphenic Theria, using shared derived characters of taxa known from both upper and lower teeth. For explanation of characters, see table 9. monious to call this a parallelism in the face known from both, gives the cladogram of the other character distributions. It is also shown in figure 12. An exception is made for possible that the expanded talonid is a syn- Amphidon and Amphitherium (see discus- apomorphy for the Cladotheria. Then the sion above) because they are sufficiently well anteroposteriorly compressed talonids (but known from lower jaws to be considered. broader than the primitive condition) of dry- The cladogram can then be converted into a olestoids would be a modification of the dichotomous classification as follows: broad cladothere talonid. In view of the evi- dence, this interpretation is adopted here. CLASS MAMMALIA LINNAEUS, 1758 Likewise, character 4 (posteroventral de- Subclass Theria Parker and Haswell, 1897 flection of the angular process) would seem Superlegion Kuehneotheria McKenna, 1975 to be the primitive condition for the clado- Kuehneotherium there angular process. The dryolestoid an- Superlegion Trechnotheria McKenna, 1975 gular process is therefore a derived variant Legion Symmetrodonta Simpson, 1925 of this condition. Sublegion Amphidontoidea, new rank Amphidon CLASSIFICATION Manchurodon Sublegion Spalacotherioidea, new rank Combing the cladograms of upper- and Infraclass Tinodontida, new rank lower-tooth taxa, and neglecting the taxa not Tinodon =Eurylambda 1981 PROTHERO: JURASSIC MAMMALS 321

TABLE 9 Explanation of Characters Used in Figure 12 (* indicates parallelism) 1. THERIA: reversed triangle occlusal pattern, broad pressed, talonid becomes narrow shelf, lose upper single wear surfaces joining cusps, lose "Kiihne- molar lingual cingulum* cone" 17. ZATHERIA: reduce to three molars, basined tal- 2. Kuehneotherium: metaconid lower than paraconid onid, add hypoconulid and entoconid, reduce sty- 3. TRECHNOTHERIA: coronoid angle increases, locone, lose anterior cingulum on lower molars* lose medial condylar flange, reduce to five premo- 18. Peramus: reduction of last molar, reduce stylar re- lars, angle of posterior molar trigons-trogonids less gion, reduce stylocone, reduce posterior cingulum than 1000, trochlear condyle above level of tooth on molars row* 19. TRIBOSPHENIDA: add protocone, paracone an- 4. SYMMETRODONTA: reduce to four premolars*, terior to metacone greatly reduced talonids, strong internal pterygoid ridge 5. AMPHIDONTOIDEA: trigonid angles about 1400, Infraclass Spalacotheriida, new rank reduce paraconid and metaconid to crenulations, lose lower molar lingual cingulum* Spalacotheroides 6. SPALACOTHERIOIDEA: reduce to three premo- Symmetrodontoides lars, lose upper molar lingual cingulum* Spalacotherium =Peralestes 7. Tinodon (=Eurylambda): coronoid process rises at Legion Cladotheria McKenna, 1975 right angle to tooth row Sublegion Dryolestoidea Butler, 1939 8. SPALACOTHERIIDA: high trigonids, wear facet Infraclass Amphitheriida, new rank on metacrista reaches lose facets A Amphitherium paracone, and Infraclass Dryolestida, new rank B on metacrista-paraconid*, trigon basined* Family 9. Spalacotheroides: paracristid parallel to protocris- Paurodontidae Marsh, 1887 tid Tathiodon Pelicopsis 10. Spalacotherium (=Peralestes): labial cingulum Araeodon continuous on lower molars* Archaeotrigon 11. CLADOTHERIA: lose lower molar lingual cingu- Paurodon lum*, expand talonid, lose anterior cusp on last low- Family Dryolestidae Marsh, 1979 er premolar, lose metacrista-paraconid facets A and Dryolestes Euthlastus B (but shear does not reach paracone), develop true Peraspalax Amblotherium angular process, expand upper molars Phascolestes transversely, "Amblotherium" nanum stylar region enlarges and interlocks, trigon ba- Laolestes Kurtodon sined*, hypoconid shears metacone, angle of ante- Kepolestes rior molar trigon-trigonids less than 1000, three up- Miccylotyrans per molar roots Crusafontia 12. DRYOLESTOIDEA: increase to six or more mo- Comotherium, new genus lars, procumbent paraconids, posterior lower molar Amblotherium Herpetairus root smaller than anterior root, continuous lingual Melanodon cingulum on last lower premolar*, lose anterior cin- (=Malthacolestes) gulum on lower molars* Sublegion Zatheria McKenna, 1975 13. Amphitherium: paraconid overlaps preceding tal- Infraclass Peramura McKenna, 1975 onid Peramus 14. DRYOLESTIDA: completely transverse paracris- Infraclass Tribosphenida McKenna, ta-metacristid shear, strongly procumbent paraco- 1975 nids, reduce to four premolars* All remaining therians 15. PAURODONTIDAE: broad shelflike paraconid and Cladotheria, incertae sedis talonid with reduced cusps, molars broaden antero- Brancatherulum posteriorly, lose anterior cusp on last lower pre- Palaeoxonodon molar* The genera within infraclass Dryolestida 16. DRYOLESTIDAE: reduce metacone, lower molar have been placed in sequential rank accord- roots markedly unequal, slender angular process ing to the cladogram, and separated into col- dorsally deflected, molars anteroposteriorly com- umns for genera known from upper teeth 322 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

(right column) and lower teeth (left column). erected infraclass Patriotheria to include the When the associations and synonymies are orders Pantotheria and Symmetrodonta. better known for these taxa, it will be pos- Kuehneotherium was placed in the latter or- sible to finish the dichotomous classification der, and the rest of the "Amphitheria" were to the specific level. Until that time, I believe relegated to the Pantotheria. that it is prudent to leave the classification The capricious and arbitrary nature of in the form given above. such taxonomic practice is readily apparent Once the classification has separated out from the vicissitudes of the names under var- individual genera, no further monotypic taxa ious workers. None of these classifications (e.g., "Legion Kuehneotheriida," "Suble- was supported by shared derived characters, gion Kuehneotherioidea," "Family Kueh- and "Pantotheria" from its inception has neotheriidae," etc.) are used, since these been a horizontal, "wastebasket" taxon, names are redundant and clutter the classi- roughly equivalent to "primitive therians." fication (Farris, 1976). It has never been a monophyletic group in the sense of Hennig (1966). McKenna (1975) THE OF THE pointed out that "Pantotheria" were not a USE TERM PANTOTHERE natural group, and so abandoned the term in The order Pantotheria was proposed by his classification. However, the most recent- Marsh in 1880 to include all non-multituber- ly published work on Mesozoic mammals culate Mesozoic mammals (which at that (Lillegraven, Kielan-Jaworowska, and Cle- time included triconodonts, docodonts, sym- mens, 1979) persists in using this term. metrodonts, and dryolestids). The Greek My paper has corroborated McKenna's roots also imply that the Pantotheria (7rcav- hypothesis of 1975 (with the original intent ros, "all"; Oijp, "beast") included every- of falsifying it), and supported his cladogram thing but the (acXXog, "other"; with many more characters. Therefore, un- fhAp, "beast"). The classification underwent less this cladogram is effectively falsified, numerous changes (documented by Simp- there should be no doubt that the "panto- son, 1929, pp. 45-47) until Simpson (1929, p. theres" are not a natural group, and should 47) restricted the term to Amphitherium, therefore have no formal status in taxonomy. Peramus, paurodonts, and dryolestids. He All the taxonomic names proposed here are excluded the symmetrodonts and tricono- based on clearly formulated cladistic princi- donts in this restricted usage. In his 1945 ples, and are supported by a parsimonious classification, Simpson created an additional arrangement of shared derived characters. infraclass Pantotheria and included the sym- As a cladistic classification, it carries the metrodonts within it. Kermack and Mussett maximum phylogenetic information content (1958) then renamed "Order Pantotheria" and is thus non-arbitary and maximally heu- the order Eupantotheria to distinguish it ristic. This hypothesis is not based on au- from infraclass Pantotheria. thoritarian fiat, or arbitary, capricious, The discovery and description of Kueh- "Gestalt"-type phenetic lumping, and can neotherium and further work on Amphither- (and should) be tested by anyone. ium and Peramus (Mills, 1964; Clemens and Some workers, however, will surely per- Mills, 1971) led to considerable debate about sist in using "pantothere" as a general term the classification. In 1968, Kermack et al. for "non-tribosphenic therian." This argu- listed three suborders within the Eupantoth- ment has been forwarded in favor of other eria: Dryolestoidea (dryolestids), Symmetro- paraphyletic and polyphyletic groups, on the donta, and Amphitheria (Kuehneotherium, grounds that they reflect "morphologic dis- Amphitherium, Peramus, and paurodonts). tance" or "genetic affinity" (Mayr, 1974). If Crompton and Jenkins (1968) returned the the non-tribosphenic Theria were a group symmetrodonts to ordinal status, separate with uniformly primitive morphology, whose from Pantotheria, and placed Kuehneother- affinities were highly ambiguous, this argu- ium in the Symmetrodonta. In 1971, Simpson ment might have some weight. However, 1981 PROTHERO: JURASSIC MAMMALS 323 this is definitely not the case with the non- sented here does far more to explain therian tribosphenic Theria. As a whole, the groups evolution than widespread, vague assertions are highly derived and quite distinctive to about "pantotheres" giving rise to higher anyone who has examined them more than mammals. Therefore, I strongly urge that the superficially. The use of a wastebasket term term "pantothere" be abandoned. If one like "pantothere" is pernicious, not only be- wishes to refer to "pantotheres," the term cause it obscures relationships, but also be- "non-tribosphenic Theria," although less cause it gives a false impression that the concise, does not carry misleading phyloge- whole group can be lumped together without netic implications. In fact, this term carries further thought or discussion. This is partic- more information content than "pantothere" ularly true of comparative studies between and focuses attention on an important "typical pantotheres" and some other form. change in mammalian evolution, namely the There is no such thing as a "typical panto- acquisition of the protocone to form the tri- there." bosphenic tooth. As a didactic tool, the cladogram pre-

LITERATURE CITED Brundin, Lars (Nat. Hist.), (Geol.), vol. 20, pp. 89- 1966. Transatlantic relationships and their sig- 113. nificance as evidenced by chironomid Cracraft, Joel midges. I. Principles of phylogenetic 1972. The relationships of the higher taxa of systematics and phylogenetic reason- birds; problems in phylogenetic rea- ing. Stockholm Kungl. Svenska Ve- soning. The Condor, vol. 74, pp. 379- tensk. Holl., vol. 11, pp. 1-472. 392. 1968. Application of phylogenetic principles 1974. Phylogenetic models and classification. in systematics and evolutionary theory. Syst. Zool., vol. 23, pp. 71-90. In 0rvig, Tor (ed.), Current problems Crompton, A. W. of lower vertebrate phylogeny. Nobel 1971. The origin of the tribosphenic molar. Symposium 4, New York, Interscience In Kermack, Doris M., and K. A. Ker- Publishers, John Wiley and Sons, pp. mack (eds.), Early mammals, Supp. 1, 473-495. Jour. Linn. Soc. (Zool.), vol. 50, pp. 1972. Evolution, causal biology, and classifi- 65-88. cation. Zool. Scripta, vol. 1, pp. 107- 1974. The dentitions and relationships of the 120. southern African mammals Butler, Percy M. parringtoni and Me- 1939. The teeth of Jurassic mammals. Proc. gazostrodon rudnerae. Bull. Brit. Mus. Zool. Soc. London (ser. B), vol. 109, (Nat. Hist.), vol. 24, pp. 397-437. pp. 329-356. Crompton, A. W., and Farish A. Jenkins, Jr. 1972. Some functional aspects of molar evo- 1967. American Jurassic symmetrodonts and lution. Evolution, vol. 26, pp. 474-483. Rhaetic 'pantotheres.' Science, vol. Clemens, William A., Jr. 155, pp. 1006-1009. 1970. Mesozoic mammalian evolution. Ann. 1968. Molar occlusion in mam- Rev. Ecol. Syst., vol. 1, pp. 357-390. mals. Biol. Reviews, vol. 43, pp. 427- Clemens, William A., Jr., and Patricia M. Lees 458. 1971. A review of English Eldredge, Niles, and Ian Tattersall mammals. In Kermack, Doris M., and 1975. Evolutionary models, phylogenetic re- K. A. Kermack (eds.), Early Mam- construction, and another look at hom- mals, Supp. 1, Jour. Linn. Soc. (Zool.), inid phylogeny. In Szalay, F. S., (ed.), vol. 50, pp. 117-130. Approaches to , Clemens, William A., Jr., and J. R. E. Mills Contributions to primatology, vol. 5. 1971. Review of Peramus tenuirostris (Mam- Basel, S. Karger, pp. 218-242. malia, Eupantotheria). Bull. Brit. Mus. Englemann, George F., and Edward 0. Wiley 324 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 167

1977. The place of ancestor-descendant re- Kermack, Kenneth A., and Frances Mussett lationships in phylogeny reconstruc- 1958. The jaw articulation of the Docodonta tion. Syst. Zool., vol. 26, pp. 1-11. and the classification of the Mesozoic Farris, James S. mammals. Proc. Royal Soc. (ser. B), 1976. Phylogenetic classification of fossils vol. 48, pp. 204-215. with recent species. Syst. Zool., vol. Kermack, Kenneth A., Frances Mussett, and H. 25, pp. 271-282. W. Rigney Fox, Richard C. 1973. The lower jaw of Morganucodon. 1976. Additions to the mammalian local fau- Journ. Linn. Soc. (Zool.), vol. 53, pp. na from the Upper Milk River Forma- 87-175. tion (Upper Cretaceous), Alberta. Can. Krebs, Bernard Jour. Earth Sci., vol. 13, pp. 1105- 1969. Nachweis eines rudimentaren Coro- 1118. noids im Unterkiefer der Pantotheria 1977. Notes on the dentition and relation- (Mammalia). Palaont. Zeitschr., vol. ships of the insectivore 43, pp. 57-63. Simpson. Can. Jour. 1971. Evolution of the mandible and lower Earth Sci., vol. 14, pp. 1823-1831. dentition in dryolestoids (Pantotheria, Freeman, Eric F. Mammalia). In Kermack, Doris M., 1976. Mammal teeth from the Forest Marble and K. A. Kermack (eds.), Early mam- () of Oxfordshire, En- mals, Supp. 1, Jour. Linn. Soc. (Zool.), gland. Science, vol. 194, pp. 1053- vol. 50, pp. 89-102. 1055. Kretzoi, M. Gaffney, Eugene S. 1960. Zur Benennung des altesten Symme- 1979. The Jurassic turtles of North America. trodonten. Vertebr. Hung., vol. 2, pp. Bull. Amer. Mus. Nat. Hist., vol. 162, 307-309. pp. 95-135. Krusat, G. Griffiths, Graham, C. D. 1969. Ein Pantotheria-Molar mit dreispitzi- 1972. The phylogenetic classification of the gem Talonid aus dem Kimmeridge von Diptera Cyclorrhapha, with special ref- Portugal. Palaont. Zeitschr., vol. 43, erence to the male postabdomen. 2. pp. 52-56. Principles and procedures of classifi- Kuhne, Walter G. cation. The Hague, W. Junk, pp. 5-28. 1950. A symmetrodont tooth from the Rhae- Henkel, Siegfried, and Bernard Krebs to-Lias. Nature, vol. 166, pp. 696-697. 1969. Zwei Saugetier-Unterkiefer aus der 1961. Une faune de mammiferes lusitaniens Unteren Kreide von Unia (Prov. Cuen- (raport provisoire). Comunicacoes dos ca, Spanien). Neues Jahrb. Geol. Pa- Servicos Geologicas de Portugal, vol. laont. Mh. 8, pp. 449-463. 45, pp. 211-221. 1977. Die erste fund eines Saugetier-Ske- 1968. Kimeridge [sic] mammals and their lettes aus der Jura-Zeit. Umschau 77, bearing on the phylogeny of the Mam- Heft 7, pp. 217-218. malia. In Drake, E. J. (ed.), Evolution Hennig, Willi and environment, New Haven, Yale 1965. Phylogenetic systematics. Ann. Rev. Univ. Press, pp. 109-123. Ent., vol. 10, pp. 97-1 16. Lillegraven, J. A., Z. Kielan-Jaworowska, and W. 1966. Phylogenetic systematics. Urbana, Chi- A. Clemens (eds.). cago, London, Univ. Illinois Press. 1979. Mesozoic mammals: The first two- Hopson, James A., and A. W. Crompton thirds of mammalian history. Berkeley, 1969. Origin of mammals. In Dobzhansky, Univ. Calif. Press, pp. 1-311. T., M. K. Hecht, and W. C. Steere McKenna, Malcolm C. (eds.), Evolutionary biology, vol. 3, 1975. Toward a phylogenetic classification of New York, Appleton-Century-Crofts, the Mammalia. In Luckett, W. P., and pp. 15-72. F. S. Szalay (eds.), Phylogeny of the Kermack, Doris M., Kenneth A. Kermack, and . New York, Plenum, pp. 21- Frances Mussett 46. 1968. The Welsh pantothere Kuehneother- Marsh, Othniel C. ium praecursoris. Jour. Linn. Soc. 1887. American Jurassic mammals. Amer. (Zool.), vol. 47, pp. 407-423. Jour. Sci., vol. 33, pp. 328-348. 1981 PROTHERO: JURASSIC MAMMALS 325

Mayr, Ernst Sigogneau-Russell, Denise 1974. Cladistic analysis or cladistic classifi- 1978. Decourverte de mammiferes rhetiens cation? Zeitschr. Zool. Syst. Evolu- (Trias superieur) dans 1'est de la tionforsch., vol. 12, pp. 98-128. France. C. R. Acad. Sc. Paris, vol. 287, Mills, J. R. E. pp. 991-993. 1964. The dentitions of Peramus and Am- Simpson, George Gaylord phitherium. Proc. Linn. Soc. London, 1925a. Mesozoic Mammalia. II. Tinodon and vol. 175, pp. 117-133. its allies. Amer. Jour. Sci., vol. 10, pp. Nelson, Gareth J. 452-470. 1973. Classification as an expression of phy- 1925b. Mesozoic Mammalia. III. Preliminary logenetic relationships. Syst. Zool., comparison of Jurassic mammals. vol. 22, pp. 344-359. Ibid., vol. 10, pp. 559-569. Osborn, Henry Fairfield 1927. Mesozoic Mammalia. VI. Genera of 1907. Evolution of mammalian molar teeth to Morrison pantotheres. Ibid., vol. 13, and from the triangular type. New pp. 407-416. York, Macmillan, pp. 1-250. 1928. A catalogue of the Mesozoic Mammalia Ostrom, John H., and John S. McIntosh in the Geological Department of the 1966. Marsh's Dinosaurs: The collections British Museum (Natural History). from Como Bluff. New Haven, Yale London, Brit. Mus. (Nat. Hist.), pp. 1- Univ. Press, pp. 1-388. 215. Parrington, Francis Rex 1929. American Mesozoic Mammalia. Mem. 1967. The origins of mammals. Advancement Peabody Mus., Yale, vol. 3, pp. 1-235. of science, London. vol. 24, pp. 165- 1937. A new Jurassic mammal. Amer. Mus. 173. Novitates, no. 943 pp. 1-6. 1971. On the Upper Triassic mammals. Phil. 1945. The principles of classification and a Trans. Royal Soc., Ser. B, vol. 261, pp. classification of the mammals. Bull. 231-272. Amer. Mus. Nat. Hist., vol. 85, pp. 1- 1978. A further account of the Triassic mam- 350. mals. Ibid., vol. 282, pp. 177-204. 1971. Concluding remarks: Mesozoic mam- Patterson, Bryan mals revisited. In Kermack, Doris M. 1955. A symmetrodont from the Early Cre- and K. A. Kermack (eds.), Early mam- taceous of northern Texas. Fieldiana, mals, Supp. 1, Jour. Linn. Soc. (Zool.), Zool., vol. 37, pp. 689-693. vol. 50, pp. 181-198. 1956. Early Cretaceous mammals and the Wiley, Edward 0. evolution of mammalian molar teeth. 1976. The phylogeny and biogeography of Fieldiana, Geol., vol. 13, pp. 1-105. fossil and Recent gars (Actinopterygii: Schaeffer, Bobb, Max K. Hecht, and Niles El- Lepisosteidae). Misc. Publ. Univ. dredge Kansas, no. 64, pp. 1-111. 1972. Phylogeny and paleontology. In Dob- Yabe, H., and T. Shikama zhansky, T. M. K. Hecht, and W. C. 1938. A new Jurassic mammal from South Steere (eds.), Evolutionary biology, Manchuria. Proc. Imp. Acad. Tokyo, vol. 6. New York, Appleton-Century- vol. 14, pp. 353-359. Crofts, pp. 31-46.

5, W..V Vig r.&;wx.7" -7 _4 t 41

x ft 7. CA V .4 VI X

lp N--- kv. IV p", V 4" n 41; v.il 77 Y., A Y Z j 4r V W 'k -7 X. "t X, .W; fk- g" .4 .0 o ;j- 'uT NA A ,ZNR 3

4. ;t%A N RV n 1.0; T. Iz '4 AE. k,:"T- 'id e., Z. 4. '47-4 'J" 03, 7- AC i7l Flw 'X A, .7 .4.4N 44t At 4 V, 4. .0 4-1"." ;4 .k iJE A T

VP4

4 F .J_ INI .f J, g 6 % le N, Ark,x .Qaw., A `Y. T A" --4 XV. PI' -y' 'w, ia, "It" M .11 T*i. n .11 Ploll;--, V k

-j N '41 4 -A k.. 4, 5- A V

T,lrW -M -a :741 Py 1,,A I tf WWI 4 V, 41"k, :i*x t .7, 'AK k-- 'Y; k oll W.-jt

O Q), .14 Ahr V, IT" g .K?" V 5Z .4- 4k p O" Al lk, A_. 4- 41P V. Z 4-X W. 4, V--- 0-T .A; NK 4. Tw'4 A uor A", 4 3 f."SJ ec,

P DL T 4 4"'A R. 4 JA 5 14.0 M11+1 V- -0, 4., k .l P -A .,i 4t v 4. WAM w. ls,- 7,

S, P .i., _7 ,4. p. F" 7.t ot ag"A, '2.1- Y- "A "S J ;'M :j gj llk Ai .k V .!44VN ..4 4 `.l A, w W. L 14 AN WV, X 4. "W" ; 'S 17-.,4 "A i V "S' .41 -, r v..,., p% 7.- 7" ,l 1A V, v, k _41 'J .N 1, W. Z. V T,., if%* 'R J_ W li, rl.-1, J.,. .k'-;"M '5- Y '4 V, bi. ._t ,U N, A q X 'A g w __,n _41 74, 44, 6" 'e M. .d IT4 X 14, 7k. WIA_ 1AIT 1- V. f4i: 14- p W. .;, A4 `4 .5j 7w A ..i xm )C",--f,V 47, F.4 ,WY NO -W 4',

4 '4"~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~. 41

/ I~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ .44~ ka,"PA"r~~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ 4"

If4

M.11~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~4

""4""'4~N ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~~~; f A ~ ~ ~ ~ ~~~~~~~~~~"k',~~~~~~~~~~~~~ ~ "" .4'4 "r-Y~~~~~~~~~~~~"""~ "'.' X4e*4"

~~~~~A'' '4 4

51VAA~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~'4' 4,

o zt~~~~~~~~~~~~~~~~~~~~~~~~~~~~~4 ' 4-4~~~~~~~~~~~4. '4~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

44,~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~4.

Y~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~eZ4 44 TM~~~~~~~~~~~~~~~~~~~~~~~~4 4

-.444.4

y'~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~4 '4