Evolution, 35(3), 1981, pp 433-440 PATTERNS OF EVOLUTION AMONG THE ARTIODACTYLA AND PERISSODACTYLA (MAMMALIA) R. L. CIFELLI Department of VertebratePaleontology, American Museum of Natural History, New York, New York 10024 Received March 28, 1980. Revised September 16, 1980 Diversity changes through time of the have been compiled from Romer (1966). mammalian ungulate orders Artiodactyla Current revisions might alter the data and Perissodactyla have been touted long somewhat, but it is doubtful that the dif- and widely as exemplifying ordinal level ferences would be systematic or great. taxonomic (and presumably ecologic) dis- Question-marked taxa or those not desig- placement (Simpson, 1953; Stanley, 1974) nated chronostratigraphically to subepoch since, as Van Valen (1971) has pointed were ignored. The time scale used is from out, they have a nearly identical "way of Harland et al. (1964). life" (=adaptive zone, Simpson, 1953; plus competitive interactions, Van Valen, DOMINANCE AND DISPLACEMENT 1971). The controversial assumption that Competitive interactions leading to the herbivores are food limited (cf. Hairston relative dominance of the artiodactyls are et al., 1960; Murdoch, 1966; Ehrlich and expected to be reflected in changes of di- Birch, 1967; Slobodkin et al., 1967) is im- versity and turnover rates through time. plicit to this generalization. Both orders Generic and familial diversity of the two presumably appeared in the late Paleo- orders are shown in Figures 1 and 2. Here, cene, as they are well differentiated by the as elsewhere, I have systematically over- early Eocene (Figs. 1 and 2). While peris- estimated chronologic ranges for genera- sodactyls seem to have undergone taxo- and therefore calculated standing diver- nomic diversification at both generic and sity at any given point in time-by con- familial levels somewhat earlier than ar- sidering originations to have occurred at tiodactyls, the difference is much less than the beginning of the subepoch of first ap- most authors suggest. Perissodactyls de- pearance and extinctions at the end of the clined in importance after the middle subepoch of last known occurrence. Al- Eocene, while the artiodactyls maintained though it would be desirable to conduct a stable "dominance" from the early Oli- the analysis at the species level, the fossil gocene until their apparently enormous record is inadequate to the task, and gen- Plio-Pleistocene radiation. Various deter- era are used. While sampling of families ministic hypotheses have been forwarded is doubtless better than of genera, the lat- in explanation of this pattern. Such hy- ter are more appropriate in the present potheses usually invoke the acquisition of context, since they reflect more accurately one or more characters (double trochleated events which ultimately must occur at the astragalus, selenodont molars, ruminant species level. Whereas biases in the fossil digestion) by the artiodactyls as giving record (preservation, relative exposures of that group a relative adaptive advantage. fossiliferous strata, "monographic"effects; Here I review the evolutionary histories Raup, 1972) hamper comparison through of these two orders in search of evidence time, I assume that relative diversities of supporting these hypotheses. Further, I the two orders at any given point in time analyze overall patterns to detect possible are unaffected by such biases. The possi- correspondence to or support for the var- bility of differences in geographic distri- ious "empirical" (Valentine, 1973) and bution is discussed below. "equilibrium" (Mark and Flessa, 1977) The adaptive advantage conferred by models of Phanerozoic diversity. All data the acquisition of a double trochleated as- 433 434 R. L. CIFELLI 0 es - 65 CD o A 25 - EEO MEO LEO EOL MOL LOLEMI MMI LMI PLI PLS R 0 0 ~~~~~~~~~~0 0 B 0 ~~~~~~~~~~~~00 N- ~ ~ ~ ~ ~ ~ ~ ~ ~ W 0 0 ~~~~~~~~~~~~02 0EEO MEO~~~~~~~~~~<0 LEO EOL MOL LOLEMI MMI LMI PLI PLS 0 0D O ~~~~~~~~~~~~o N ~~~~~~~~~~~~0 FIG. 2. Familial diversity (circles), originations (triangles), and extinctions (diamonds) of perissodac- tyls (A) and artiodactyls (B). Originations and ex- 0 0 tinctions are cumulative. R; Recent; other abbrevi- ations as in Fig. 1. into (tinls an xicin(daod)o e ochs (Raup, 1972) must be confronted: while vertical corrections to absolute time have been made in the figures, relative diversity is biased. This is because, all rissodactyls (A) and artiodactyls (B). Abbreviations: other conditions being equal, a long inter- EEO; early Eocene, MEO; middle Eocene, LEO; late Eocene, EOL; early Oligocene, MOL; middle val is expected to show greater diversity Oligocene, LOL; late Oligocene, EMI; early mio- than a short one. In the present case, high cene, MMI; middle Miocene, LMI; late Miocene, diversity in the early Miocene (duration = PLI; Pliocene, PLS; Pleistocene. 9 million years) follows low diversity in the late Oligocene (duration= 3 million years). Estimates of standing diversity tragalus (Schaeffer, 1947) cannot be eval- may be made by subtracting cumulative uated in the present context, since it ap- extinctions from cumulative originations pears as a basic ordinal character of the at any given point in time. Figure 3 shows earliest (basal Eocene) artiodactyls. There that the late Oligocene constriction and is no discernible change in pattern during the early Miocene expansion are real, al- the late Eocene, when the highly efficient though the magnitude of difference is not selenodont dentitions arose, although this great. This is further supported by might have been expected to have stimu- changes in familial diversity (Fig. 2b). lated an artiodactyl radiation. The in- The great Plio-Pleistocene artiodactyl crease in artiodactyl diversity in the early radiation (comprised mainly of bovids and Miocene corresponds roughly to the evo- cervids; Table 1) apparently had no no- lution of the fully ruminating stomach (as ticeable effect on perissodactyl diversity, projected by Janis, 1976), which is hy- which seems to have increased over this pothesized to have conferred great adap- time period (although this may be a sam- tive advantage ("eat and run") upon its pling bias, as the Plio-Pleistocene mam- bearers (see, for instance, Young, 1962). mal record is particularly good). However, the Miocene artiodactyl radia- If, as Newell (1952) suggests for Meso- tion is comprised in large part by the non- zoic nautiloid-ammonoid diversities, com- ruminating Merycoidodontidae (Janis, petitive advantage of one group is indi- 1976; see Table 1). Here, the problem of cated by its diversification occurring differences in relative duration of subep- concomitantly with the decline of its com- PATTERNS OF MAMMALIAN EVOLUTION 435 TABLE 1. Generic diversity of perissodactyl and artiodactylfamilies by subepoch, early Eocene to Pleisto- cene. Abbreviations: E; early, M; middle, L; late. Eocene Oligocene Miocene E M L E M L E M L Plio. Pleis. A. Perissodactylfamilies Equidae 3 5 5 3 2 1 4 5 6 8 6 Palaeotheriidae 1 2 2 Brontotheriidae 3 7 22 13 3 1 Eomoropidae 1 1 4 Chalicotheriidae 1 2 1 3 4 3 2 5 4 Isectolophidae 1 1 1 1 Helaletidae 2 6 4 1 1 1 Lophialetidae 1 2 Deperetellidae 1 2 1 Lophiodontidae 3 2 2 Tapiridae 1 1 2 4 2 2 2 1 Hyracodontidae 2 4 3 1 1 1 Amynodontidae 7 3 2 1 1 Rhinocerotidae 1 10 13 10 14 7 9 10 13 7 B. Artiodactylfamilies Diacodectidae 3 Leptochoeridae 1 1 1 Homacodontidae 1 3 4 Dichobunidae 1 9 4 3 1 Achaenodontidae 1 2 Choeropotamidae 3 2 1 Cebochoeridae 2 2 2 Entelodontidae 2 3 3 1 1 Suidae 1 1 1 2 6 6 7 12 12 Tayassuidae 1 2 1 3 2 4 5 6 Anthracotheriidae 4 9 6 7 8 9 5 5 7 1 Hippopotamidae 1 2 Cainotheriidae 2 2 3 2 1 1 Anoplotheriidae 3 6 3 2 Xiphodontidae 2 3 2 Amphimerycidae 1 2 2 Agriochoeridae 1 1 1 1 1 Merycoidodontidae 7 9 8 15 7 3 2 Oromerycidae 2 1 Camelidae 1 2 3 5 3 3 12 12 Hypertragulidae 5 4 5 3 5 Protoceratidae 2 1 1 2 1 1 2 Gelocidae 3 4 3 1 Tragulidae 1 1 3 3 Palaeomerycidae 1 1 3 10 10 7 5 1 Cervidae 3 5 5 24 33 Giraffidae 3 5 15 5 Antilocapridae 1 2 9 6 Bovidae 1 3 6 8 67 105 petitor(s), the record of odd- and even- ment (Fig. 5). Although increase in artio- toed ungulates gives no such indications; dactyl diversity corresponds roughly to nor is there any apparent negative corre- perissodactyl decline in North America, lation between generic diversities of the "clades" on all other continents resemble two orders as would be predicted (Fig. 4). those of the combined data; in fact, pe- An analysis by zoogeographic province riods of increased diversity of artiodactyls might be more meaningful, but here, again, and perissodactyls seem to coincide, sug- the evidence is not supportive of displace- gesting that both are either dependent on 436 R. L. CIFELLI 0 _ A e- A 14- >-0 0 V) Lo A 0 0 A A A tsAAAiA C\L A 0CC) 0 25 50 75 100 125 150 175 0 ART IODACTYLS 0 FIG. 4. Correlation of perissodactyl and artio- dactyl generic diversities. EEO MEOLEO EOL MOLLOL EMI MMI LMI PLI PLS during the middle and early late Tertiary, 0 o B but this is probably due to an inadequate fossil record. 0 Changes in turnover rates, which may 0 not be indicated by generalized diversity 0 0 trends, must also be considered. If selec- tive advantage of one group causes an in- 0 0 crease in turnover rates of a competitor, 0 an increase in perissodactyl extinction rate 0 should accompany an artiodactyl radia- tion. Taxonomic survivorship curves, which indicate extinction rates, have been plotted for six perissodactyl cohorts (Fig. 6; vertical scale is log percent of surviving FIG.