Proc. Nati. Acad. Sci. USA Vol. 91, pp. 10403-10406, October 1994 Evolution , body size, and paleotemperature THOMAS M. BOWN*, PATRICIA A. HOLROYD*, AND KENNETH D. ROSEt *U.S. Geological Survey, MS 919, Box 25046 Federal Center, Denver, CO 80225; and tDepartment of Cell Biology and Anatomy, Johns Hopkins University School of Medicine, Baltimore, MD 21205 Communicated by Elwyn L. Simons, July 15, 1994

ABSTRACT There is a general inverse relationship be. coinciding with the two biohorizons. Increased floodplain tween the natural logarithm of tooth area (a body size indica- instability and greater paleosol hydromorphy (wetness) also tor) of some and paleotemperature during resulted from increased sediment accumulation rate. These approximately 2.9 million years of the early In the correlations support Schankler's postulated causes for faunal Bighorn Basin of northwest Wyoming. When mean tempera- turnover and they are further advocated by correlations of hures became warmer, tooth areas tended to become smaller. tooth size and paleotemperature. During colder times, larger species predominated; these gen- erally became larger or remain the same size. Paleotemper- Paeotemperature and Tooth Size ature trends also markedly affece patterns of local (and, perhaps, regional) e con d immigration. New species Studies of stratigraphically ordered collections of Willwood appeared as immigrants during or near the hottest(smaller mammals (e.g., refs. 13-16) show that the means and ob- forms) and coldest (larger forms) intervals. Paleotemperature served ranges of the natural logarithm of the area of the first trend reversals commonly resulted In the ultimate e n of lower molar [In(length x width) of ml] generally change both small forms (during cooling intervals) and larger forms through time. Tooth size, especially that of the first lower (during warming intervals). These immgtions and extinc- molar, has been demonstrated by many authors to correlate tions mark faunal turnovers that were also modulated by sharp well with body size (13, 17-21) and sustained shifts in tooth increases in sediment accumulation rate. size through time, therefore, strongly suggest evolutionary change in body size. Although it is possible that tooth size The lower Eocene Willwood Formation ofthe Bighorn Basin, shifts reflect other parameters (e.g., dietary change), such Wyoming, contains one of the densest known records of parameters are also related to body size and, thus ultimately, Tertiary mammalian evolution (1) and has the most detailed to paleotemperature as well. biostratigraphic resolution, with more than 1100 fossil mam- A paleotemperature curve derived from leaf-margin anal- mal localities (and more than 100,000 mammal ) tied-to yses of stratigraphically controlled Paleocene-Eocene floras measured sections (2). Fluvial sedimentologic, paleosol, and from the Bighorn Basin corresponds sufficiently with curves paleontologic analyses of Willwood rocks and faunas in produced from 81 0 measurements ofbenthic and planktonic recent years have focused on Willwood sediment accumula- foraminifera (9) to indicate that the leaf-margin data docu- tion rates, time stratigraphy [with percent of elapsed Will- ment mean annual temperature fluctuations in the intermon- wood time depicted in temporal values or TV (3)], and the tane Bighorn Basin between 90C and 180C from about 55.7 correlation of biotic and sedimentary patterns (3-9). million years ago to about 52.8 million years ago. At TV intervals 0-33 and TV 82-97, mean temperatures gradually Willwood Mammalian Bostratigraphy rose from 13 to 150C and from 16.5 to 180(, respectively (Fig. 1). In TV 33-72, there was a gradual decrease in temperature Schankler (10) recognized four biostratigraphic zones, from 150( to about 90(, and in TV 72-82 mean temperature bounded by three major events of faunal turnover that he rose very rapidly from 9 to 16.50C. termed biohorizons A, B, and C. These biohorizons record To test the hypothesis that tooth size (and thereby body times ofboth "" (here termed disappearances) and size) fluctuations through time are related to temperature "immigration" (appearances). Reconstructed Willwood time changes, the leaf-margin paleotemperature curve was plotted stratigraphy (3) demonstrates that biohorizons B and C against temporal records of tooth areas for each of four actually record the same episode of faunal turnover (here groups of Willwood mammals (Figs, 1-4). called B). Biohorizon A records faunal events during the time Hyopsodus (Condylarthra: Hyopsodontde). A founding occupied by the 190- to 210-m interval (TV = about 50; Fig. small species of the small Hyopsodus disappears 1), as measured above the base of the Willwood Formation, at biohorizon A, in the middle of a cooling phase (Fig. 1, and biohorizon B encompasses the 380- to 425-m interval (TV Hyopsodus sp. 1). This event was succeeded by the abrupt = 78.9-82). appearance ofa slightly larger species (Hyopsodus sp. 2) and its slight decrease in size during a warming trend begun just Sediment Accumulation and Extinction before biohorizon B. This species continued to decrease in size and survived until about TV 82. After the biohorizon B Schankler (10) believed that faunal transitions at biohorizons interval, four species of Hyopsodus appeared as immigants A and B resulted from changing climatic or ecological con- (Hyopsodus spp. 3-6). One species was very large bodied ditions, whereas Badgley and Gingerich (12) suggested that (Fig. 1), another was relatively small, and the other two were biohorizon A phenomena might instead reflect nothing more intermediate in size between the large species and the small- than small sample sizes. Bown and Kraus (3) showed that est species. All four postbiohorizon B species decreased in Willwood sediment accumulation rate was influenced by size during continued warming from TV 82 to TV 97. tectonic pulses and that it increased sharply at two intervals Phenacodus and Copecion (Condylarthra: Phenacodon- tidae). The much larger Phenacodus (two spe- The publication costs ofthis article were defrayed in part by page charge cies) and Copecion (one species) show a similar pattern. payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact. Abbreviation: TV, temporal value(s). 10403 Downloaded by guest on September 25, 2021 10404 Evolufion: Bown et al. Proc. Natl. Acad. Sci. USA 9f (1994)

Paleotemperature 7141 r f76 15 4 17 12 10l 5 C

,J 0 52.79 r-7L Arr:Ar datr: QW.::ua 30 58 F- 85 80 ~~~BIOHORIZON B -75C K

a) 65 nOK E TV av 55 - coI. 50 K BIOHORiZON A '>,-.~; f;; s ,,s; 5 .,, aL.. 4W :;: ., < 45 - E X 40 Hyopsodus 35 N = 2055 FIG. 1. Stratigraphic distribution of the natural n s s.- - logarithm offirst lower molar(M1) areas ofthe small 30 *Hyvpsod mammal Hyopsodus (six species) Hyopsodus c sP > condylarthran 25 from the lower Eocene Wiliwood Formation. Ma, * HyopsOads 1r age in million years before present based on radio- Hyopsodus C metric date of 52.79 Ma at TV 97.93 and of Hycpsodus c (9) age I 5 55.7 Ma for base of Willwood Formation (5, 11). * Hyoosodus c The TV in the Willwood Formation was based on 1 0 paleosol maturity indices (3). Heavy curve and 50 upper abscissa are leaf-margin paleotemperatures

55.7 L derived from Wing et al (9). Note that paleotem- 0 perature scale is reversed. The intervals encom- 1 .5 2.0 2.5 3 - passing biohorizons A and B are shaded. Symbols In M1 area designate different taxa as shown. Below biohorizon B, Phenacodus was relatively rare and to TV 75. This large species barely persisted beyond the remained approximately the same size until just before bio- pronounced warming at biohorizon B and disappeared horizon A (TV 50) when it began to increase in size. After shortly thereafter (TV 92). At TV 72 (nearthe beginning ofthe biohorizon A, this species continued to increase gradually in most rapid rise in Willwood paleotemperature), the much size, especially during a period of sharp cooling from TV 45 smaller Phenacodus vortmani appeared, followed closely in Paleotemperature 79 18 .7 16 75 14 13 12 10 R Q,'- ..... 100 52.79 tAr Ar dater 95 90 1 53 .I..; 4 85

> 80 ..-~~ ~ ~~~~~~~A .rI 75 70 65 K am 60 K 0) 55 a)X - BIOHORIZON A Eco ITV 50 Q 54 0 45 K a-aL 40 K- 35 Phenacodontidae 30 N = 331 25 K Pirenacodus trriobtah),m

20 K * Pt1enacCous vct, .1a !. Flu 2. StraltigraphiLc distrihu- 15 Copecion, trca leia lion of first lower molar cM, ) areas .it' the large condylarthran mam- nLo I'cnucoldu.s (lvmo species) i fron' 5 .- nd (CPccitu (one spexksi ,he lower Eocene Wiliwood For- ....2 . . ... lation. S\ mhols des".naite diftfe- talxii as shown. Othet srvmhbol in M area .-e as explained in Fri- Downloaded by guest on September 25, 2021 Evolution: Bown et al. Proc. Natl. Acad. Sci. USA 91 (1994) 10405

Paleotemperature 19 18 17 16 15 14 13 12 10 90C

52.79 (Ar/Ar date) 95 0 0 m 5 90 w 0 0 E 53 1 a : : 85 - 1! 1!. q a mm-%% m 5 80 BIOHORIZON B - 75 70 65 0) 60 55 CD TV . 54 x 0 0. 0L Apheliscus N = 141 B A. nitidus

F" W. A. wapitriensis W. :R * A. new so. 1 MI 8 VA M A. new sp. 2 FIG. 3. Stratigraphic distribu- r.:: EJ S tion of fourth lower premolar (P4)

ESB 35EB533W to in Q EU areas of the minute condylarthran BW9In amI (?) mammal Apheliscus (four spe- cies) from the lower Eocene Will- 55.7 L wood Formation. Symbols desig- nate different taxa as shown. 0.9 1.4 1.9 Other symbols are as explained in In P4 area Fig. 1.

time by appearance of the exceedingly small phenacodontid exhibits the same trends as the first lower molar, but more Copecion brachypternus (Fig. 2). dramatically. The lower first molar in the earliest Willwood Apheliscus (Condylarthra?). In Fig. 3, the size distributions species of Apheliscus, Apheliscus nitidus, decreased in size ofthe natural logarithm ofthe fourth premolar in four species as paleotemperature increased during the interval from TV 0 of the minute mammal Apheliscus are illustrated. This tooth to TV 33. A. nitidus disappeared shortly after paleotemper-

Paleotemperature 19 18 17 16 15 14 13 12 10 90C 100 52.79 Ar./Ar date) 95 -

53 90K- 85 80 _ IZONB 75 70 65 - (U 60 - cn 55 -

)- TV 50 _ 0) 54 45 - x Omomyidae Co N = 430 0- 40 - 0. Teirrnardlna amercana- Q. 35 - crassidens Ihneage Teiiha rdina tenuicuva 30 - Tetonius matthewi- 25 - PseudotetonUs aMtgUUs ireagE! 20 - AnemorhysIs SpD 15 - Chiororrysiss& Ta. - a.' .s Steinius vespertnus FIG. 4. Stratigraphic distribu- 10 Absarokius abpotti don offirst lower molar (M1) areas 5 - Absarok.us re toecuc ofomomyid primates (eight genera and 12 species) from the lower 55.7 - 0 _____- EoceneWillwoodFormation. Sym- 0.75 1.25 1.75 bols designate different taxa as shown. Other symbols are as ex- In Ml area plained in Fig. 1. Downloaded by guest on September 25, 2021 10406 Evolution: Bown et al. Proc. Natl. Acad. Sci. USA 91 (1994) atures began to decrease at TV 33. Subsequent to biohorizon states that body size typically increases with decreasing mean A, a new and slightly larger species ofApheliscus (Apheliscus temperatures. Although studies of temperature/body size wapitiensis) appeared. This species increased in size (in- relationships are largely based on latitudinal distributions in versely following decreasing paleotemperature) until it dis- extant mammals (26), this principle should apply equally to appeared during warming at TV 78. At TV 76, after the mammals of an evolving fauna from a specific region that coldest period in Willwood mean temperature, a new and experienced significant mean temperature changes over tens very large undescribed species ofApheliscus was introduced of thousands to millions of years. (Fig. 3, large open circle), and this event was followed closely In a related study using only 18 mammal tooth data points, by the disappearance of both it and its smaller congener. A Rea et al. (27) maintained that lower first molar areas of the new fourth species, introduced after biohorizon B (TV 83), large pantodont mammal Coryphodon demonstrate apprecia- succumbed to at least local extinction at about TV 93, during ble molar size decrease during earliest Eocene warming. Our continued mean paleotemperature rise from 16.5 to 18TC. work, from a data base of nearly 3000 specimens of taxo- Omomyidae (Primates). Omomyid lineages are well repre- nomically diverse fossil mammals ranging over about 3 sented in the Willwood Formation (14). The Teilhardina million years of the early Eocene, strongly supports these lineage (Teilhardina americana, Teilhardina crassidens, and previous positions and amplifies upon them. Moreover, this Teilhardina tenuicula; Fig. 4, shaded triangles) showed de- analysis of the correspondence between paleotemperature crease in ml area as temperature increased from TV 7-33 and and both mammalian tooth size and faunal turnover is doc- beyond, until TV 42, after the beginning of a cooling phase. umented from well-studied lineages of Eocene mammals. It The slightly larger T. crassidens evolved from T. americana provides a strong causal explanation not only for patterns of during cooling from TV 46 to TV 48 but disappeared at faunal turnover but also for changes in the tempo and biohorizon A. It was succeeded by the newly introduced and direction of mammalian evolution. short-lived T. tenuicula. The Tetonius-Pseudotetonius lin- We are grateful to J. G. Fleagle and S. L. Wing for discussion and eage (Fig. 4, open squares) first appeared at about TV 20 comments on the manuscript and acknowledge support of National (Tetonius matthewi). These primates apparently developed Geographic Society Grant 3985-89 to T.M.B. and National Science smaller molar areas during warming up to TV 28 and then Foundation Grant BSR-8918755 to K.D.R. increased in size during cooling until biohorizon A. Between biohorizons A and B, molar area remained about the same 1. Gingerich, P. D. (1980) Univ. Mich. Pap. Paleontol. 24, 1-6. until about TV 70 when it began to decrease again during 2. Bown,T. M.,Rose,K. D.,Simons,E. L.&Wing,S. L.(1994) rapid warming. This lineage (now Pseudotetonius ambiguus) U.S. Geol. Surv. Prof. Pap. 1540. disappeared at biohorizon B. The very small omomyids 3. Bown, T. M. & Kraus, M. J. (1993) Palaios 8, 68-80. Steinius, Anemorhysis, and Arapahovius (Fig. 4) were intro- 4. Bown, T. M. & Kraus, M. J. (1987) J. Sediment. Petrol. 57, duced near the acme of biohorizon B warming (TV 80) but 587-601. 5. Bown, T. M. & Kraus, M. J. (1993) Evol. Anthropol. 2, 11-21. disappeared prior to TV 82 at the top of the biohorizon 6. Kraus, M. J. (1987) J. Sediment. Petrol. 57, 602-612. interval. Continued warming above biohorizon B (TV 83-97) 7. Kraus, M. J. & Bown, T. M. (1993) Geology 21, 743-746. was accompanied by the appearance ofthe small Anemorhy- 8. Bown, T. M. & Beard, K. C. (1990) Geol. Soc. Am. Spec. Pap. sis wortmani (upper two shaded diamonds) and brief appear- 243, 135-151. ances of the minute omomyids Tatmanius szalayi and Chlo- 9. Wing, S. L., Bown, T. M. & Obradovich, J. D. (1991) Geology rorhysis incomptus [black hexagon (14, 22)]. Also appearing 19, 1189-1192. were two species of the larger omomyid Absarokius (Absa- 10. Schankler, D. M. (1980) Univ. Mich. Pap. Paleontol. 24, 99- rokius abbotti and Absarokius metoecus; Fig. 4, shaded 114. squares and black squares, respectively). A. metoecus de- 11. Obradovich, J. D. (1988) Paleoceanography 3, 757-770. 12. Badgley, C. & Gingerich, P. D. (1988) Palaeogeogr. Palaeo- creased in size through time and A. abbotti remained about climacol. Palaeoecol. 63, 141-157. the same size. 13. Gingerich, P. D. (1974) Nature (London) 248, 107-109. 14. Bown, T. M. & Rose, K. D. (1987) Paleontol. Soc. Mem. 23, Condusions 1-162. 15. Gunnell, G. F. (1989) Univ. Mich. Pap. Paleontol. 27, 1-157. All of the tooth area plots presented here exhibit variations 16. Thewissen, J. G. M. (1990) Univ. Mich. Pap. Paleontol. 29, on repeated patterns that reflect evolutionary change in 1-107. mammalian body size. These size changes correspond well 17. Bookstein, F. L., Gingerich, P. D. & Kluge, A. G. (1978) with gradual increase and decrease in early Eocene paleo- Paleobiology 4, 120-134. 18. Gingerich, P. D. & Schoeninger, M. J. (1979) Am. J. Phys. temperatures through time. Body size is closely linked to Anthrop. 51, 457-466. character evolution (14, 23), and environmentally modulated 19. Conroy, G. C. (1987) Int. J. Primatol. 8, 115-137. changes in body size are known to be one of the most 20. Legendre, S. & Roth, C. (1988) Hist. Biol. 1, 85-98. fundamental adaptations in mammals (24, 25). The data used 21. Dagosto, M. & Terrenova, C. (1992) Int. J. Primatol. 13, in this study show three responses ofearly Eocene mammals 307-344. to shifts in paleotemperature, (i) their immediate disappear- 22. Bown, T. M. & Rose, K. D. (1991) J. Hum. Evol. 20, 465-480. ance or disappearance after a short interval oftime (generally 23. Bown, T. M. & Fleagle, J. G. (1993) Paleontol. Soc. Mem. 29, at a biohorizon boundary), (ii) tooth area in these mammals 1-76. inversely tracks paleotemperature for some time but this 24. Fleagle, J. G. (1988) Primate Adaptation and Evolution (Aca- demic, San Diego). phenomenon is followed shortly thereafter by disappearance 25. Jungers, W. L. (1985) Size and Scaling in Primate Biology ofthe species, and (iii) new forms appear that have relatively (Plenum, New York). smaller molar areas in warmer times and relatively larger 26. Peters, R. H. (1983) The Ecological Implications ofBody Size molar areas during cooler periods. (Cambridge Univ. Press, Cambridge, U.K.). Body size in mammals can be correlated in a general way 27. Rea, D. K., Zachos, J. C., Owen, R. M. & Gingerich, P. D. with mean temperature. For example, Bergmann's "rule" (1990) Palaeogeogr. Palaeoclimatol. Palaeoecol. 79, 117-128. Downloaded by guest on September 25, 2021