sign in sign up
<<
Home , Archaeohyracidae, Astrapotheria, Dinocerata, Macrauchenia, Palaeolama, Parallel evolution

&HQR]RLF0DPPDOLDQ+HUELYRUHVIURPWKH$PHULFDV5HFRQVWUXFWLQJ$QFLHQW'LHWVDQG 7HUUHVWULDO&RPPXQLWLHV $XWKRU V %UXFH-0DF)DGGHQ 6RXUFH$QQXDO5HYLHZRI(FRORJ\DQG6\VWHPDWLFV9RO  SS 3XEOLVKHGE\Annual Reviews 6WDEOH85/http://www.jstor.org/stable/221724 . $FFHVVHG

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at . http://www.jstor.org/action/showPublisher?publisherCode=annrevs. .

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Annual Reviews is collaborating with JSTOR to digitize, preserve and extend access to Annual Review of and .

http://www.jstor.org Annu. Rev. Ecol. Syst. 2000. 31:33-59 Copyright? 2000 by Annual Reviews. All rights reserved

CENOZOICMAMMALIAN HERBIVORES FROM THE AMERICAS:Reconstructing Ancient Diets and TerrestrialCommunities

BruceJ. MacFadden Museumof Natural History, Universityof Florida, Gainesville, Florida 32611; e-mail: [email protected]

Key Words browsers,grazers, isotopes, teeth, morphology, paleoclimate * Abstract Herbivoryfirst evolved in terrestrialmammals during the late Cre- taceous, -100 million ago (Mya). Of the -35 ordinal-levelclades of extinct or extanteutherian from the , -24 have been adaptedto her- bivoryin one formor another.Dental for specializedterrestrial browsing are firstrecognized during the early Cenozoic (-).Mammalian her- bivoresadapted for grazingdid not become widespreadin the New Worlduntil the middleCenozoic; it seems thatthis adaptationand the spreadof grasslandsoccurred duringthe late (30 Mya) in SouthAmerica 10 million yearsearlier than in NorthAmerica (20 Mya).Carbon isotopic evidence from fossil herbivoreteeth indi- catesthat C3 plantspredominated until the late (8 Mya).Thereafter, C3 and C4 terrestrialcommunities diversified. Late 10,000 yearsago decimatedthe diversity of mammalianherbivores, particularly those of largerbody size.

INTRODUCTION

As primaryconsumers of plantbiomass, herbivores represent the majorityof diver- sity in ancientmammalian radiations. The fossil recordof mammalianherbivores in North and South Americais relativelywell representedover the past 65 million years (My). During this time therehave been considerablechanges in climate and plantdiversity that affected the structureand distributionof mammalianherbivore communities.In the past severaldecades, some importantfactors have influenced our understandingof, and our ability to reconstruct,ancient mammalian herbivore communities.Paleontological discoveries continuously improve our knowledge of the fossil recordand oftentimesfill in critical gaps. New techniques,such as anal- yses of stable isotopes and enamel microwear,have advancedour ability to make paleodietaryinterpretations. Continuous refinements in dating techniquesallow a better understandingof the time sequence of mammalianherbivore community .

0066-4162/00/1120-0033$14.00 33 34 MACFADDEN

The fossil recordreveals a time dimension not availableto modem ecologists. Paleontologistscan track discrete communities throughmillions of years during which basic communitystructure is preservedbut new taxa originateand fill eco- logical niches vacatedby taxa thatbecame extinct. This is called a chronofauna,a term originallyproposed by Olson (64,65). This paperreviews the fossil recordof terrestrialmammalian herbivore commu- nities overthe past 65 my, the Cenozoic Erain Northand SouthAmerica. Whenever possible, emphasisis placed on recentdiscoveries and new techniquesthat enhance understandingof this subject. The Cenozoic was a time of great global change, and it is somewhatartificial to devote this review to the New World,when parallel faunal changes occurredin the Old World.However, this focus is determinedby the availablespace; the interestedreader can also consult previousreviews on this general subject (31, 36, 94, 95).

BACKGROUNDAND METHODS

In 1873, the Russian paleontologist Kowalevsky published a classic paper (38) describing fossil from Europe. He asserted that the evolution of horses with short-crownedteeth to those with high-crownedteeth during the Miocene indicated a correspondingchange in diet from browsing to grazing. Since that time, the crown height of fossil herbivoreshas been used to interpretthe diets of ancient mammals (Figure 1). Within the past few decades, other tech- niques have addedindependent evidence to an understandingof ancientherbivore diets. Complementinggross tooth morphology,studies of skull morphology,and enamelmicrowear, stable carbon isotopes now can be used to reconstructCenozoic herbivorecommunities.

Morphology Modern herbivorousmammals with short-crowned(brachyodont) teeth, e.g., the (Tapirus) or deer (), are generallyadapted to feeding on soft, leafy vegetation and hence are browsers. In contrast, modem herbivorousmammals with high-crownedteeth, e.g., the zebra (Equus) and bison (Bison), are generally grazers.High-crowned is defined as unwornpremolar or teeth in which the height exceeds the occlusal length of the tooth (Figure2A). High-crownedteeth are either of finite growth,like those of horses, or ever-growing(hypselodont) during the 'slifetime, like those of some grazingrodents. The adaptivesignificance of high-crownedteeth is, in most cases, related to the abundanceof phytoliths in grasses. Phytoliths are microscopicbodies of silica (SiO2, the same compoundas glass) within grasses that tend to wear teeth down and are an adaptationof the plant against herbivory(59). There are, however, some exceptions to this general short-crownedbrowser/high-crowned grazer pattern. For example, short-crowned (Lama)are principally grazers (29, 52), whereassome extincthigh-crowned CENOZOICMAMMALIAN HERBIVORES 35

* Pulp cavity B [ H Enamel

Cementum

,1,.

Figure 1 Comparison of short-crowned(brachyodont) human tooth (a) versus high- crowned tooth (b) showing expansion of the crown relative to the root area. From Ref. 33 and reproducedwith permissionof CambridgeUniversity Press. horses were principallybrowsers (53). Nevertheless, tooth morphology serves as a general model to interpretextinct herbivorediets. There also is a correlationbetween high-crownedteeth and open-countryhabi- tats. Thus, some high-crownedherbivores may incorporateconsiderable amounts of grit into their diets from dust on the plant foods that they eat close to the land surface (29). Terrestrialgrazers are generally considered to include that feed predominantly(>90%) on grasses (29), althougha species is also a grazerif it crops plants other than grasses (e.g., forbs) that form the low ground cover in some biomes. Certaincranial characters are also highly correlatedto diet in extantherbivorous mammals.For example, the muzzle shape and incisor width of modem distinguish browsers from grazers (Figure 2). Grazerstend to have more trans- versely straight muzzles and incisors of generally similar size, which together form a functionalcropping mechanism to procuregrass and other plants near the ground surface. In contrast,browsers have more rounded muzzles and differen- tiated incisors, which togetherform a cropping mechanism for selective feeding from trees and shrubs(32, 84). These differencesare also apparentin fossil herbi- vores, e.g., Oligocene notoungulatesin SouthAmerica (78) andMiocene sympatric horse species in North America (47). Othercharacters, like the depth of the jaw, developmentof the bony masseterprominence on the cheek, and position of the orbit, also indicate browsing versus grazing diets (85). 36 MACFADDEN

A B il LEN M~3EN

ilLEN i3LEN UMAPL E N

cc: cc: MUZWDTH

CJAWA~~~~prsnt)~~(present) ~~" LEWAGL (absent)

Figure 2 Dental and skull charactersused to assess browsingversus grazing. (A) crown height, or hypsodontyindex (HI) is the ratioof unworncrown height (UCRNHT)to molar anteroposterior length (UMAPL); in this case the HI is -2.6, indicating a grazer. (B) Relative incisor lengths (ilLEN and i3LEN) and muzzle width (MUZWDTH)indicate a browser(narrow, left) or a grazer (broad, right). (C) The position of the orbit (ORBITPOS),presence or absence of the masseter prominence(MASSPROM), and depthand posterior angle of thejaw (JAWDEPTH,JAWANGLE) indicatea browser(left) or a grazer(right). Modified from Ref. 52 andreproduced with permission of the PaleontologicalSociety.

The mastication of various foodstuffs imparts distinctive microscopic wear patterns to the tooth enamel of modem herbivorous mamnmalswith known diets (83,86). Using modem analogs, extinct herbivore diets can be interpreted using dental microwear. These enamel-microwear patterns can be quantified by deter- mining the proportion of scratches versus pits on the enamel occlusal surface. Thus the enamel of a browser like the black-fronted duiker (Cephalophusniger) will have a large number of pits, which are caused by percussing leaves, in contrast to a grazer like the bison (Bison bison), which has many scratches caused by chewing abrasive grasses (Figure 3). CENOZOICMAMMALIAN HERBIVORES 37

30- Litocranius ~C)Browsers Cephalophus weller 0 Grzessipce niger []Fossil species

Cephalophus

Dinohippus Qdorsals mexicanus LJ

CTagelfaphus } 20 -strepsieros Kobus *4w. ellipsiprymnus o Cervus duvauceli O)&0 TrageJaphus/Trglpu dUUCi Q Acimbarls 0 *Alceaphus : ? / @ ~~~~~~~~~~~~~buselaphus Axis axis 0 ~~~~~~~~0 N

Boocercus Equlus geqvis eurycerus burchesl greyvil Pseudhlpparion

Giraffa Q camelopardis Connochaetes Bison taurinus Wbison

0 10 20 30 40 Mean numbhAr of scratches

Figure 3 Enamel microwearstudies the proportionof pits versus scratchesper unit area (in this case 0.5 mm2) for modem mammalianherbivores with known diets. Browsers have a high proportionof pits, whereas grazershave a high proportionof scratches.These modem analogs are used as a model to interpretthe diets of fossil mammalianherbivores. Of the late Miocene fossil horses depicted here, Dinohippus mexicanus is a browser and Pseudhipparionsimpsoni is a grazer.Modified from Ref. 53 andreproduced with permission of the AAAS.

Carbon Isotopes Stable carbon isotopes have recently been used to interpret the diets and ecology of fossil mammals. For extinct herbivores, analysis of tooth enamel carbonate in- dicates whether the animal was feeding on C3 plants, C4 grasses, or a mixture of these foodstuffs (e.g., 37, 41, 52, 74). This technique can be used to discrimi- nate diets because plants photosynthesize carbon by two principal pathways. The dominant photosynthetic pathway for terrestrial plants is the Calvin cycle, in which 38 MACFADDEN

carbon is first incorporatedinto 3-carboncompounds, hence the term C3. About 85% of terrestrialplant biomass, including trees, most shrubs, forbs, and high- latitude or high-elevation grasses, use the Calvin cycle. In contrast, about 10% of terrestrialplant biomass photosynthesizescarbon using the Hatch-Slackcycle, in which carbon is first incorporatedinto 4-carbon compounds, hence C4. (The third photosyntheticpathway, CAM, or Crassulaceanacid metabolism, is used by a small fraction of terrestrialplant biomass such as succulents. It is of minor relevancein the currentstudy.) Present-day tropical and temperategrasses are pre- dominantlyC4, and they are adaptedto more highly seasonal and arid climates thanare C3 plants (17, 18). Not only do C3 and C4 photosyntheticpathways incor- poratecarbon into differentcompounds, they also fractionatethe stableisotopes of carbon(12C and 13C)in differentproportions. The ratioof 13C/12Cis conventionally expressed as

313C(in parts per mil, %o) = (Rsample/(Rstandard- 1) X 1000,

whereR = 13C/12C.All measurementsof an unknown(such as fossil-toothenamel) are comparedto the standardPDB (a Cretaceousmarine mollusk PeeDee belem- nite, which has a S13C of 0 %0). C3 plants characteristicallyhave S13C values ranging from -34 %oto -23 %o, with a mean of -27 %o; C4 plants characteris- tically have S13C values ranging from -17 %oto -9 %o,with a mean of -13 %o (3, 16, 19). When mammalseat plants,the S13Cis enrichedin theirskeletal tissues, so the S13Cof tooth enamel is -12-14 %omore positive than the corresponding plant foods (8, 37; Figure4). The use of carbonisotopes for dietaryand communityreconstructions of mam- malian herbivoresis best suited to the late Cenozoic, after the diversificationof C3 and C4 plant communities -8 Mya. Prior to this time, most of the dietary reconstructionspresented below rely principally on morphological interpreta- tions.

TERRESTRIALHERBIVORE PHYLOGENY, ORIGINS,AND DISTRIBUTION

Herbivory is a widespread adaptationin the history of mammals. Of the -35 ordinal-level of extinct and extant mammals from the New World, two thirds contain herbivorous species and half are exclusively herbivores (33; Figure 5). These clades include such diverse specializations as folivores, frugi- vores, granivores,succulent feeders, mixed-plantfeeders, and grazers. During the late throughearly (225 to 120 Mya), all mam- mals were small-bodied(<5 kg) and lacked the distinctivedental adaptationsthat indicate plant-eating.Although some of these primitivemammals probably were omnivores, with some percentage of plants in their diets, it is difficult to make meaningfuldietary interpretations of most Mesozoic mammals (12). CENOZOICMAMMALIAN HERBIVORES 39

5

_ _ ~ ~~~~~Toothenamel E ~10i' :'::: _ /

-15 ::::i:......

0-20 (erG , F'

-30 Fraction C4(%)

Figure 4 Carbonisotopic values, expressed in 813 CPDB of end-memberC3 plants (left side, bottom = 0% C4) and C4 grasses (right side, bottom = 100% C4) as comparedto 813Cvalues of tooth enamel of C3 feeders (left side, top) versusC4 grazers(right side, top). Relative to plant foods, tooth enamel carbonateis enrichedby about 12-14 0/o0 (8, 37).

Eutheriandental evolution is markedby a key innovationduring the early Cre- taceous, -100 Mya, the tribosphenictooth (4,69). Priorto this time, mammalian teeth were relatively simple and probablywere principallysectorial (adaptedfor cutting), with a lesser crushing or grinding function. The tribosphenictooth is characterizedby an internalprincipal cusp, the protocone, in the uppers, which occludes with a talonid basin in the lowers. The evolution of the protocone and talonid basin, the function of which has been likened to that of a mortarand pes- tle, resulted in a morphologicallycomplex tooth with increased surface area for mastication. Some of the eutherianmammal clades that later developed highly specialized adaptationsfor herbivoryoriginated during the late Cretaceous,100 to 65 Mya. In North America duringthe late Cretaceous,some eutherian"" and condy- larth clades had dentitions that were apparentlyadapted for masticating fibrous plant foods of low nutritivevalue and/orseeds and nuts of higher energy (1). The fossil record of Mesozoic mammals in is relatively fragmentary (11, 13), although some notable recent discoveries have been made (68). Given the overall paucity of well-preservedmaterial, little can be said about the dental 40 MACFADDEN

Paleocene Eocene Oligocene Miocene Plio PI Multituberculata

Condylarthra

...... Taeniodonta ...... XXXXXXXXXXXXXXTillodonta , ...... ......

Pyrotheria f

Notoungulata Xenarthra Rodentia_ - Lagomorpha Perissodactyla _ Artiodactyla

Proboscidea 60 40 30 20 10 C MillionYears Ago

...... N America ...... S America - N & S America

Figure 5 Geological distributionof the major ordinal-level groups of terrestrialmam- malian herbivoresfrom the Cenozoic of the Americas. Othergroups with a low percentage of herbivores(Marsupialia, Chiroptera) or of low diversitythat are not discussedin text (Xe- nungulata)are not shown here. Compiledfrom numeroussources, e.g. Refs. 11, 34, 67, 68.

function and diet of South American Mesozoic mammals. Nevertheless, in the New World (i.e., mostly North America), we have evidence of the primitive clades that later developed specialized herbivorous adaptations.

EARLY TERTIARY MAMMALIAN HERBIVORES AND CLIMATE

The first half of the Cenozoic, represented by the Paleocene and Eocene epochs (65 to 34 Mya), represents a time of major global change. Stable isotopic evidence from both marine and terrestrial sequences at the Paleocene/Eocene boundary (-55 Mya), indicates there was a relatively abrupt global warming event result- ing in much warmer "hothouse" conditions during the Eocene (37,71). The mean annual temperature in low and middle latitudes during the early Eocene is esti- mated to have been between 25?C and 30?C (Figure 6), -15?C to 20?C warmer CENOZOICMAMMALIAN HERBIVORES 41

30

25 -

cm 20 -

< 15 _

E 0ID

0 Pale Oligo- Cretaceous cene Eocene cene Miocene

100 90 80 70 60 50 40 30 20 10 0 MillionYears Ago

Figure 6 Change in mean annual temperatureover the past 100 My as reconstructed from deep-sea oxygen isotope and paleobotanicaldata. From Ref. 63 and reproducedwith permissionof the Missouri Botanical Garden. than today. This was the most significantglobal warmingevent duringthe Ceno- zoic, and it had a profound effect on the terrestrialbiota. For example, in the well-sampled Paleocene/Eocenesequence in northernWyoming, plant macrofos- sil diversity doubled (25 to 50 recognizable species) between 55 and 53 Mya (100; Figure 7). Thus, the fossil recordindicates a dramaticshift from Paleocene mesic woodlands to Eocene subtropicalvegetation throughoutmuch of North America. The presence of cold-intolerantturtles, tortoises, crocodylians,and pri- mates throughoutNorth America, including at extreme northernlatitudes (e.g., in the CanadianArctic at 800 N; 24), indicates global expansion of subtropical and some temperate belts during this time (71). Although grasses originated during the early Cenozoic (the oldest reported fossil grass is from the Pale- oocene/Eocene of Tennessee, 14), therewere no extensive grasslandsthen as there are today. By the Paleocene, faunal interchangebetween the Americas, as represented by the pan-Americangroups Condylarthraand Pantodonta,had ended. Herbivore diversificationthereafter until the late Cenozoic was representedby separateadap- tive radiationsof endemic groups in South and North America. The isolation of South America during most of the Cenozoic gives us a unique opportunityto compareradiations of mammalianherbivores that originatedfrom differentclades and to observe parallelevolution of adaptations.Similarly, the reconnectionof the 42 MACFADDEN

50 -

40 - a)

al)01 CL cu 30 -

0M a) E

20-

1-Cretaceous j Paleocene Eocene

70 65 60 55 50 MillionYears Aqo Figure7 Fossilplant diversity across the Paleocene-Eocene boundary. Notice the doubling in theearly Paleocene and subsequent drop from 40 to 25 specieswithin about 2 My (from 57 to 55 Mya).From Ref. 100and reproduced with permission of Elsevier.

Americas at the end of the Cenozoic had a profoundimpact on the native biotas. These "naturalevolutionary experiments" are discussed below. Morphological evidence from both North and South American mammalian faunas duringthe Paleocene and Eocene reveals considerablebody size diversifi- cation (particularlyincrease; 11) and dentalspecializations. Dental morphologyof numerousclades indicates a complex tooth apparentlyspecialized for herbivory, including increased tooth area for masticationof plant foodstuffs. In the smaller range of body sizes (<5 kg), the abundanceof forested biomes during this time yielded a correspondinglydiverse array of arboreallyadapted herbivores, including folivores, frugivores,and granivores(71). In the New World,many of the modem groups of terrestrialherbivores origi- nated and/orradiated during this time. The ordersinclude Xenarthra(edentates), Lagomorph,Rodentia, Perissodactyla,and Artiodactyla(Figure 5). Now-extinct ordersof New Worldterrestrial herbivorous mammals alive at that time included the Tillodontia,Taeniodontia, Condylarthra, Dinocerata, Astrapotheria, Pyrothe- ria, Pantodonta,and non-therianMultituberculata. With a few exceptions, all of these Paleocene and Eocene herbivoreshad relativelybrachyodont dentitions, in- dicating a principally browsing diet. They exploited the niches created in open, CENOZOICMAMMALIAN HERBIVORES 43

nonarboreal,habitats. Taeniodonts and tillodonts from North America were high- crowned exceptions. Their diets have been reconstructedas having contained a considerableproportion of undergroundroots andtubers. They probablycontended with grit on their foods, which explains the hypsodontteeth (42,43). The earliesthorse Hyracotherium("eohippus"; family )is known from the early Eocene of Holarctica. Because of its widespread abundance,and its importance near the base of both the equid and perissodactyl diversifications, Hyracotheriumis an instructiveexample of an early Cenozoic mammalianher- bivore. Traditionally,because of its short-crowneddentition, Hyracotherium was considereda forest-dwellingbrowser (e.g., 62, 80, 81). However,an extraordinary populationof 24 individualsfrom a single quarryin Coloradois interpretedto rep- resent an open-countryform, which fed in early savanna-like,woodland mosaics (22). But instead of grass, this horse may have fed on soft groundvegetation, in- cluding herbaceousdicots and dry-adaptedferns, as well as some low shrubs(e.g. hackberry-Celtis). Fossilized brain endocasts of Hyracotheriumindicate a rela- tively advancedneocortex, suggesting increased tactile sensitivity of the lips for selective feeding (75). Dental microwearstudies of Hyracotheriumshow mostly pits, indicatinga mixed-browsingdiet, the animalshaving fed primarilyon fruits, bushes, seeds, ferns, and other leaves (5). The mean (13C of Hyracotheriumteeth is -12.5 %0, indicating a diet of C3 plants (91). The early diversificationof the ordersPerissodactyla and Artiodactylaoccurred duringthe Eocene. By the Oligocene, -34 to 30 Mya, more advancedartiodactyls diversified, particularlywithin the ruminants.It has been suggested that rumi- nantdigestion provideda competitiveadvantage relative to hind-gutperissodactyl fermentation.Perissodactyls remained of modest diversity while the artiodactyls increased dramaticallyin diversity later during the Cenozoic, and some workers (81) have suggested a causal interdependencebetween these two orders.It seems, however, that this is an oversimplificationand actually climatic and vegetative changes duringthis time are also causativefactors in the relativediversities of the Perissodactylaand Artiodactyla(10, 30). In South America, the fossil recordof herbivorousmammals is predominantly representedby brachyodontforms, indicatingprincipally browsing (70). An ex- ception, the Archaeohyracidae,a family of small-bodiednotoungulates, had ever- growing teeth. This may have been an early (Eocene) shift to grazing (82), an adaptationthat greatly expanded during the later Cenozoic. Another group, the polydolopid ,was -like,an adaptationalso seen in the multitu- berculatesin North America duringthe Paleocene and Eocene (31, 67, 68).

EOCENE GREENHOUSE TO OLIGOCENE ICEHOUSE

For almost a centurypaleontologists have recognized a major change in Eocene to Oligocene mammalianfaunas from Europe, termed the GrandCoupure ("big cut") (71, 73). A series of climatic events (includingthe GrandCoupure) occurred 44 MACFADDEN

duringa period of -7 My from -39 to 32 Mya (71, 73). Climatic reconstructions indicate significantglobal change from the peak Eocene "greenhouse"conditions to early Oligocene colder "icehouse"conditions. Temperaturesare estimated to have dropped -20?C from 50 to 35 Mya (Figure 6). Profoundfaunal and floral changes, including extinctions, originations,and geographicrange shifts, during the late Eocene and into the early Oligocene occurredin such diverse groups as plantmacrofossils, marine microfossils, marine invertebrates, amphibians, reptiles, and land mammals.Recent analysis of the ecology across the Eocene/Oligocene boundaryindicates a change from principallywarm, humid forest types duringthe Eocene to more zonal climates during the Oligocene that included arid, colder, and more open savanna-likehabitats (71, 73). The shift from greenhouse to icehouse conditions coincided with sweeping changes in New Worldherbivore communities. Archaic groups such as the condy- larths, primitive primates, tillodonts, taeniodonts, and uintatheres (Dinocerata) became extinct by the end of the Eocene. As global climate shifted towardmore glacial conditions, sea levels dropped,providing opportunities for intercontinental dispersalroutes, such as the Bering land bridge connecting Holarctica.Immigrant herbivores from the Old World that first appearedduring the Eocene in North Americainclude lagomorphs, , and severalclades of artiodactyls(e.g., ). With the possible exception of the lagomorphs,the othergroups of mam- malian herbivoreshad short-crownedteeth and were presumablybrowsers (31). Some of the best early Tertiarymammal faunas of South American endemic groups are known from classic early and middle Eocene localities in , -53 to 45 Mya (21). These faunas include xenarthrans,notoungulates, and litop- tems, which togetherrepresent the majorityof the herbivorediversity on thatcon- tinent at that time (several other ordinal-level groups that also existed in South America during this time are either poorly representedby fossils or were of low diversity and enigmatic phylogenetic relationship).Before the late Eocene, 80% of the herbivorousmammals were short-crownedand thus are interpretedto have been browsers,and 10% were either partiallyor fully high-crownedand thus are interpretedeither to have been early grazersor to have fed on grittyfoods (67, 70). Fossil phytolithsfirst become common in the late Eocene of South America, sug- gesting the spread of early grasslandbiomes (27, 87). As we will see below, the of high-crowned grazers and presumed spread of grasslands occurredat differenttimes in North and South America.

PRECOCIOUSHYPSODONTY AND GRAZERS IN SOUTH AMERICA

Patterson& Pascual (70, also 67) noticed a fundamentalchange in dental mor- phology and presumeddiets of middle Cenozoic mammalianherbivores in South America.The early and middle Eocene (45 Mya) herbivoresare principallyshort- crowned, presumed to be browsers. In younger faunas of late Eocene/early Oligocene "Tinguirican"(-35 Mya; 101) and late Oligocene/early Miocene CENOZOICMAMMALIAN HERBIVORES 45

4- O Notohippids * Equids 3 0

0 -a 0 2 0~~~~. E ~~~~~~8 0 Hypsodont - ---40 U 0 W 0 Brachyodont ?[>*8@ *| ,, J 0 60 50 40 30 20 10 0 MillionYears Ago

Figure 8 Adventof "precociouslyhypsodont" South American notohippid notoungulates at -25 to 30 Mya comparedto hypsodontyin North Americanhorses at -15 to 20 Mya. Modified from Ref. 78 and reproducedwith permissionof the Society of VertebratePaleontology.

Deseadan (-30 to 20 Mya) ages, the dominant small- to medium-sized herbi- vores are representedby the ,which include a strikingdiversity of high-crownedforms-with hypsodontyindices between 1.5 and 2 (Figure8). The teeth are characteristicallyevergrowing (hypselodont). The abundanceof South Americanmammals with high-crownedteeth (70) in the middle Tertiaryhas been suggested to indicate the advent of grasslandcom- munities,but this advancedmorphology occurs as much as 15 My earlierin South Americathan in NorthAmerica. This asynchronyhas led to the concept of "preco- cious hypsodonty"(70) for SouthAmerican notoungulates relative to, for example, horses (family Equidae) in North America (Figure 8). Carbonisotopic evidence from both short- and high-crownedherbivores from localities in Ar- gentina and uniformly yields 513Cvalues more negative than -10%o (50,54), indicatingthat these presumedgrazers were feeding on C3 grasses. This indicates a major difference from modem terrestrialgrasslands, which are pre- dominantlyC4. There are both climatic and physiological explanationsfor this difference,as is discussed below. The deseadanSalla Beds of Bolivia, which containprecociously hypsodont her- bivores, have been precisely dated between 28 and 25 Mya (35). The Salla fauna includesa diversityof short-crownedherbivores (, astrapotheres,litop- tems), small- to medium-sized high-crowned notoungulates, early caviomorph , and the Branisella. Modem primates occur in rain forests, so by analogy it might be suggested that Salla was tropical. However, given the 46 MACFADDEN

5cm Figure 9 Ventral views of upper dentitions of notoungulatesfrom the late Oligocene of Salla, Bolivia, showing differences in incisor series. In Pascualihippus(a), the incisor series is transverselylinear, indicating a grazer.In Eurygenium(b) and (c), the incisor series is rounded,indicating a browser,or more selective feeder. From Ref. 78 and reproducedwith permissionof the Society of VertebratePaleontology.

preponderanceof high-crownedherbivores, and based on associatedevidence from the sediments, Salla was reconstructedas a relatively arid grasslandhabitat (46). The Salla fauna contains three notoungulates,Pascualihippus, Eurygenium, and Rhynchippus,all of which have high-crowned teeth (HIs of 1.5, 1.8, and 1.4, respectively; also see Figure 2a), but they have different incisor morphologies. Pascualihippushas a broad, transverselylinear incisor series, indicating a grazer (Figure9A). In contrast,Eurygenium and Rhynchippus have roundedmuzzles (Fig- ure 9B,C), which otherwisemight suggest a browser(84). The incisor morphology of Eurygeniumand Rhynchippussuggest browsing, whereas their high-crowned teeth suggest grazing. How then does high-crowneddental morphology fit with the roundedincisor series? Eurygeniumand Rhynchippusare both interpretedas selective feeders, and thus with niches more specialized than that of Pascuali- hippus (78). Eurygeniumand Rhynchippus,although of similar incisor and cheek tooth morphologies,differed in size, which apparentlyfacilitated their niche sep- arationswithin the same community,perhaps correspondingto "Hutchinsonian ratios"(26, 77). In roughly equivalent 30 Mya Oligocene sediments from North America, as representedby the classic Badlandsof South Dakota and Nebraska,the terrestrial mammalianherbivore community was relativelyprimitive in its morphologicaland presumedfeeding adaptations.Although there was a significantdiversification of the principalungulate groups Artiodactyla and Perissodactylain the North Amer- ican Oligocene, they had uniformly short-crownedteeth with rounded incisors, indicating a browsing diet. Enamel-microwearstudies of the common Oligocene CENOZOICMAMMALIAN HERBIVORES 47

horse Mesohippusindicate that it was a browserand fed on a varied diet of fruits, shoots, and leaves (5). Thus, in contrastto their North America counterparts,the South Americanmammalian herbivores were indeed precociously hypsodont.

MIOCENESAVANNAS OF NORTHAMERICA

The extensiveNorth American fossil recordreveals major floral and faunal changes during the late Oligocene and early Miocene, -20 to 25 Mya. Although fos- sil grasses occur in earlier Cenozoic sediments in North America (14), the first evidence of abundantfossil grasslandsis from the early Miocene (55, 87). The adaptiveradiation of hypsodontherbivorous mammals occurred during the middle Miocene, 15 to 20 Mya. Membersof the Rodentia,, Perissodactyla, and Artiodactylaall developed hypsodontclades. This resultedin a strikingarray of herbivores,including browsers, mixed feeders, and the grazingguild (48), with a diversityand communitystructure similar to that of the modem African savanna (94, 96). Simpson (80) called this the GreatTransformation, a time when mammals invadeda new adaptivezone (81) thatallowed them to exploit grasslandresources. The evolutionary"cost" of becoming grazerswas rapidly acceleratedtooth wear caused by feeding on abrasivegrasses containingphytoliths. The responsein graz- ers was the evolution of hypsodonty. This time of fundamentalmorphological change (76), as groups of herbivorousmammals in North America evolved from primarilybrowsers to primarilygrazers, resulted in the terrestrialgrazing guild. Horses (family Equidae) diversifiedrapidly during the Great Transformation with pronouncedmorphological reorganization of the skull anddentition, resulting in a total of 13 genera in late Miocene faunas (Figure 10). Evidence from

15 13 Genera

i~~~~~~~~~10 ~ ~ ~ ~ ~ ~ ~~~~ Z%10 / (E; ~ > Hemp)

. Maximum Diversity During .C ~~~~~ClarendonianChronofauna

<>5- 5- \ \3 e 1

20 15 10 5 0 Million Years Ago

Figure 10 Generic diversity of Miocene and horses (Equidae) from North Amer- ica showing decline after about 15 Mya. Hemp =_ Hemphillian land-mammal age. From Ref. 47 and reproduced with permission of Cambridge University Press. 48 MACFADDEN

Figure 11 SEM showing enamel microwearin modem and fossil horses. Top: Equus burchelli,the modem Burchell's zebra, showing a predominanceand density of elongated scratches charactenisticof a grazer. Bottom: Late Miocene horse Cormohipparionfrom North America showing a lower density of scratchesand the presence of pits; this pattern is characteristicof a miixedfeeder. From Ref. 23 and reproducedwith permission of the Finnish Zoological PublishingBoard.

muzzle morphologyand hypsodonty indicates that these coexisting horses spanned a broadrange of diets (47). Enamel microwearof North American Cormohippar- ion (23; Figure 11) and Old WorldHipparion (5) indicates that these horses fed on, respectively,a mixed diet and principallygrass. (23). Evidence of grazingalso comes from the transverselylinear incisor morphologyshape of the late Miocene equid Calippus(25). Duringthe height of the Clarendonianchronofauna - 10 Mya, horses reachedtheir peak diversityof 13 genera, declining in diversitythereafter. The Clarendonianchronofauna is a NorthAmerican mammalian assemblage from -12 to 7 Mya (92,94,96). The first horses to become extinct during this time were short-crowned,browsing anchitheres,presumably in response to the decline of forested communities. Thereafter,during the late Miocene, diversity dropped to eight genera by the late Hemphillian,5 Mya (Figure 8). Evidence of diverse herbivorousadaptations in the Clarendonianchronofauna comes from a variety of Miocene mammals. At the height of the Clarendonian CENOZOICMAMMALIAN HERBIVORES 49

chronofaunain Nebraska, mammalianherbivore taxa were representedby 87% grazers, 10%mixed feeders, and 3%browsers (96). Aepycamelinecamels evolved stilt-legs, elongatedcervical vertebrae,and body heights approaching6 m, thereby occupying a niche similar to modem-day giraffes (96). Gomphothereprobosci- deans, with elongated,flattened, and spatulatelower incisor tusks, were previously thoughtto have scooped up aquaticvegetation, but morerecently have been reinter- pretedas mixed feeders, with theirincisors performing a varietyof food-procuring roles (39). Definitive paleontological evidence of grazing is representedby the presence of fossil grass in the dentalcavities of the late Miocene rhinocerosTeleo- ceras from the Ashfall Fossil Beds in Nebraska(90). Is thereevidence of carbonisotopic discriminationlike thatseen duringthe tran- sition from browserswith short-crownedteeth to grazerswith high-crownedteeth duringthe middle Miocene? Although horses became hypsodontbetween 15 and 20 Mya, thereis no correspondingshift of 813Cin the Equidae(9, 91). Mammalian herbivores had tooth enamel carbonate 813Cvalues characteristicallyless than -10% (7, 49, 50, 54, 91). Thus, mammalherbivores in the New Worldprior to the late Miocene lived in C3 grass biomes thatwere widespreadthroughout temperate and tropicalclimes, a situationvery differentfrom modem-day ecosystems.

LATEMIOCENE GLOBAL CARBON SHIFT

Priorto the late Miocene, carbonisotopic evidence from Northand South America indicatethat mammalianherbivores existed in a C3 world, with tooth enamel car- bonate 813Cvalues characteristically<100%o (7,49,50,54,91). The dominance of C3 plants in terrestrialecosystems throughoutthe early and middle Cenozoic relatesto atmosphericCO2 concentrations. C3 plantsare favoredin regimes of ele- vatedlevels of atmosphericC02, as has been modeled for most of the Cenozoic (2). Duringthe late Miocene, however,major global change reducedatmospheric CO2 to below a criticalthreshold and increased aridity and seasonality,all of which favor C4 photosynthesis(17, 18). Carbonisotopic values from fossil mammals,ancient soils, and deep-sea cores indicate a majorglobal change -7-8 Mya (6,7,40,74). 813Cvalues of post-late-Miocene fossil grazingmammals and grasslandpaleosol carbonateshave S13Cvalues around00%o indicating C4 terrestrialcommunities. (Figure 12). This late Miocene global carbon shift had a profound effect on the history of Cenozoic ecosystems: It resulted in a latitudinalgradient of C3/C4 grasses, with C3 grasses predominatingin colder, more polar regions, and C4 grasses predominatingin temperateand tropicalregions (51, 89). C4 grasses are generally found in ecosystems of lower productivity,which in turn supporta lower overall biomass and diversity. After 7 Mya, horses continuedto decline in diversity.It is plausible thathorses were part of the overall drop in diversity supportedby lower productivitysteppe- like C4 grasslands. Which equid taxa survived this ? Isotopic and 50 MACFADDEN

0

0 5-

0 Late Miocene global carbon shift c's cu 10 O

15 * N. Americanhorse teeth O%Qr r Pakistan mammal teeth

-a-- Pakistan paleosols Ocean carbonate shift 20 -15 -10 -5 0 5 13 CPDB(parts per mill %o)

Figure 12 Isotopic evidence for the late Miocene global carbon shift between -7 and 8 Mya. Modified from Ref. 6 and reproducedwith permission of Macmillan.

microwearstudies of the six coexisting high-crownedequid generafrom 5-My-old sediments in Florida indicate that these horses included grazers, mixed feeders, and C3 browsers (53). Two grazing and one browsing species became extinct during the latest Miocene at 5 Mya. Other taxa of North American mammalian herbivoresalso experiencedextinctions at 5 Mya, includingthe family Rhinoceroti- dae, and genera and species of Rodentia,Artiodactyla, and Proboscidea(96). The correspondingSouth Americanrecord at 5 Mya is not well known. Although C4 grasses are known to have been part of South American grazing diets (50,54), it is difficultto determineif extinctionswere above backgroundlevels at this time.

GREATAMERICAN INTERCHANGE AND PLEISTOCENE MEGAHERBIVORES

For most of the Cenozoic, North and South America were geographicallyisolated from one another,and uniquemammalian herbivore communities evolved on each CENOZOICMAMMALIAN HERBIVORES 51 of these continents. During the Pliocene (5 to 2 Mya), however, this faunal iso- lation changed with the formationof a dry-landconnection by the closing of the Isthmusof Panama -3 Mya. This closure resultedin a land-bridgedispersal corri- dor for the variousimmigrants. Paleontologists have studiedthis "GreatAmerican Interchange"(GAI; 88), the height of which occurred during the middle Pleis- tocene, 1 Mya. The communitystructure and faunal equilibriumwere disrupted by the additionof immigrantmammals into the residentbiotas. Immigrantherbi- vores from NorthAmerican that dispersed into South Americaincluded phyllotine rodents, , , horses, peccaries, llamas, and deer. Herbivoresfrom South Americathat immigratedinto North Americaincluded caviomorphrodents and the edentates,including armadillos,, and giant ground . It has been suggested that duringtimes of faunal interchange,the immigrants, or "invaders,"generally have a competitive advantage over the native, or resi- dent biota, as Simpson (79,82) asserted was the case during the GAI. Although the concept of competition is difficult to quantify and test in the fossil record, nevertheless, some general patterns of the GAI are instructive here. First, in South America (Figure 13), the native fauna of ungulate mammals consisted of 12-13 genera before the formationof the land bridge, and decreased to five and then three genera duringthe GAI. Correspondingly,the North Americaninvaders

Ungulate Genera in South America

American Ungulates

< American 3 M E ~ ' Ungulates E cu

E CLInterchange

(0 CL)

Diversity

Figure 13 Turnoverof ungulate genera in South America during the Plio-Pleistocene GreatAmerican Interchange. Modified from Ref. 92 and reproducedwith permissionof the PaleontologicalSociety. Land-mammalabbreviations and boundariesare as follows (Ref. 21): Huay, ,Miocene, 7 to 9 Mya; (hiatus);Chap, Chapadmalalan,Pliocene, 3.4 to 4 Mya; Uquian, Pliocene-early Pleistocene, 1.5 to 3 Mya; Ense, Ensenadan,middle Pleistocene, 0.5 to 1.5 Mya; Luja, Lujanian,, 10,000 years ago to 0.5 Mya. 52 MACFADDEN

went from one to 10 to 14 to 20 genera before the late Pleistocene extinctions (93, 95). Second, the patternof extinctionsduring the GAI is asymmetric;that is, the immigrantsapparently affected the native herbivoresin South America, but the reverse was not true, i.e. in North America the immigranttaxa from South America did not seem to have a competitive advantageover the native herbivores (82,98,99). It is tempting to suggest simple explanations(e.g. competition) for complex biological phenomenasuch as the GAI. Otherfactors, however, includ- ing climate change and loss of specialized habitats,almost certainlywere related to the demise of the native herbivorefaunas throughoutthe Americas during the Pleistocene. The Tarijafauna of southernBolivia is a classic middlePleistocene GAI locality. It serves as a good examplefrom which mammalianherbivore community structure can be examined. The Tarijamedium- and large-bodiedherbivores comprise two dozen species within the endemic ordersEdentata, Notoungulata, Litopterna, and caviomorph Rodentia and to these were added the North American immigrant ordersProboscidea, Perissodactyla, and Artiodactyla.A multivariatestudy of the carbon isotopes, cranial morphology,and hypsodonty of these Tarijaherbivores (except for the edentates,which lack enamel for carbonisotopic analysis)revealed some interestingpatterns within the herbivorecommunity (52). Of the 13 herbivore species analyzed, three were browsers, five were mixed feeders, and five were grazers. The grazers span the largest range of body size, from Vicugna to the Cuvieronius.The larger body sizes (to the right on the first principal component axis, Figure 14) supportsthe idea that grazers need longer retention times for processinglow-nutritive-value forage in theirgut (58, 66). Withinfamilies consisting of more than one herbivorespecies at Tarija,there is evidence of niche differentiation.For example,of the threesympatric llamas, two are grazers(Lama, Vicugna)and one is a mixed feeder (Palaeolama), and of the horses, two species are mixed feeders (Hippidion,Onohippidium) and one is a pure C4 grazer(Equus; Figure 14). Large-bodiedherbivores (megaherbivores) have a significantimpact on terres- trialcommunities (66). In NorthAmerica, the ecological impactof megaherbivore immigrationswas profoundduring the GAI. Sloths were consummatebrowsers, as has been documentedby the presence of low shrubtwigs and roots in the dung of the late Pleistocene Nothriotheriopsfrom Arizona (44). Glyptodontshave heavily infolded, high-crowned teeth, and were probably mixed feeders with a significantamount of grass in their diets. An added componentof biogeographic complexity resulted from immigrationsacross Holarcticaduring the Pleistocene. Two Old World immigrant megaherbivores,in particular,played a role in the hypergrazerniche, which during the Pliocene was occupied by Equus (49). The ,Mammuthus, and bison, Bison, dispersed into North America at re- spectively 1.5 and 0.5 Mya (45). Studies of the carbonisotopic values of the teeth of mammothand bison indicate thatthey were principallygrazers, and Equus was also a grazerat this time, althoughthe horse had more of a grazing/mixedfeeding componentto its diet duringthis time (15, 20, 36). CENOZOICMAMMALIAN HERBIVORES 53

1 Hippidion Neochoeris principale tarijensispltni Onohippidium patensis 0.5 Palaeolama Pa/aeo/ama devil/el

, Sp. Macrauchenia O 0 Tapirus patachonica QL fHippo- tarijensis E camelus CuvieroniLl 0p

Q- 0.5 - E uus/ co C P / ~~~~~~~~~~insulatus

-1.5 cf. Vicugna provicugna

angustimaxilla -2 -2 1 0 1 2 3 4 1st Principal Component

Figure 14 Principalcomponents analysis of 13 herbivoretaxa from the middle Pleistocene of Tarija,Bolivia, using carbon isotopes, molar length, muzzle width, and hypsodonty index. The grazers(G, black circles), mixed feeders (MF, shadedcircles), and browsers(B, white circles) fall into discrete groups in multivariatespace. From Ref. 52 and reproducedwith permission of the PaleontologicalSociety.

Only a fractionof the vast herbivorediversity that existed throughoutthe Amer- icas during the Pleistocene persists today. In a very short period of time, from -15,000 to 10,000 years ago, megaherbivoressuch as , mastodons, horses, sloths, glyptodonts, and giant armadillosbecame extinct throughoutthe Americas, and other taxa including llamas, tapirs,and bison have greatly reduced ranges today relative to those before 15,000 years ago (97,99). Two principal hypotheses have explainedthese late Pleistocene megaherbivoreextinctions. The first is climate and habitatchange; the other is that humansrapidly populated the Americas, resulting in "Pleistocene overkill" of large prey species (60,61). As 54 MACFADDEN

with all complex naturalsystems, these two explanationsundoubtedly both played roles as causal factors of late Pleistocene extinctions.

CONCLUDINGCOMMENTS

The analysisof Cenozoic herbivorecommunities in the Americasillustrates numer- ous evolutionaryand ecological principles and patterns.We see the of grass and grazers, the evolution of similar adaptations(e.g., hypsodonty) in differentgroups on differentcontinents, the biotic reaction to major climate per- turbations,the change in diversityof browsing and grazingherbivore groups with correspondingchanges in vegetation, and the extinctions resulting from climate and/orbiotic factors.These examples are more clearly interpretedusing a combi- nation of morphologicaland isotopic techniquesnow availableto paleontologists. The fossil record of Cenozoic mammalianherbivores reveals both change and stability of overall community structurethrough time, the idea inherent in the concept of the chronofauna.The fundamentalquestion about long-term changes in community evolution is whether biotic equilibriumand change are affected principally by external, climatic factors, or by internal factors such as species interactionsand competition (28,31,72). Naturalsystems are rarely simple and, in reality,long-term changes are almost certainly structuredby both climatic and biotic factors.The fossil recordprovides definitive and unique evidence fromwhich we can understandthe history of evolving communitiesof Cenozoic herbivoresof the Americas.

ACKNOWLEDGMENTS I thankRM Hunt and SD Webbfor makinghelpful comments on this manuscript. This paper was written during support from NSF grants EAR-9506550, IBN- 9528020, and EAR-9909186. This is Universityof FloridaContribution to Paleo- no. 515.

Visit the Annual Reviews home page at www.AnnualReviews.org

LITERATURE CITED 1. Archibald JD. 1998. Archaic ungulates zoic time. Am. J. Sci. 294:56-94 ("Condylarthra").In Evolution of Tertiary 3. Boutton TW. 1991. Stable carbon isotope Mammalsof NorthAmerica. Vol. 1: Terres- ratios of naturalminerals: II. Atmospheric, trial Carnivores,Ungulates, and Ungulate- terrestrial,marine, and freshwaterenviron- like Mammals,ed. CM Janis, KM Scott, LL ments. In Carbon Isotope Techniques,ed. Jacobs,pp. 292-331. New York:Cambridge DC Coleman,B Fry,pp. 173-95. San Diego, Univ. Press CA: Academic Press 2. Bemer R. 1994. GEOCARB II: a revised 4. Bown TM, Kraus MJ. 1979. Origin of model of atmospheric CO2 over Phanero- the tribosphenic molar and metatherian CENOZOICMAMMALIAN HERBIVORES 55

and eutherian dental formulae. In Meso- mals: TheFirst Two-Thirdsof Mammalian zoic Mammals: The First Two-Thirdsof History, ed. JA Lillegraven, Z Kielan- MammalianHistory, ed. JA Lillegraven,Z Jaworoska,WA Clemens,pp. 7-58. Berke- Kielan-Jaworoska,WA Clemens, pp. 172- ley: Univ. Calif. Press 81. Berkeley: Univ. Calif. Press 14. CrepetWL, FeldmanGD. 1991. The earli- 5. CapriniA. 1998. The food habits of some est remains of grasses in the fossil record. Eocene to present-day Equidae deduced Am. J. Bot. 78:1010-14 from observation of the teeth under the 15. Cronin SL, BetancourtJ, Quade J. 1998. S.E.M. Palaeontolograph.Italica 85:153- Late Pleistocene C4 plant dominance and 76 summerrainfall in the southwesternUnited 6. Cerling TE, Wang Y, Quade J. 1993. States from isotopic study of herbivore Global ecological change in the late teeth. Q. Res. 50:179-93 Miocene:expansion of C4 ecosystems.Na- 16. Deines P. 1980. The isotopic composition ture 361:344-45 of reducedorganic carbon. In Handbookof 7. Cerling TE, et al. 1997. Global vegetation EnvironmentalIsotope Chemistry.Vol. 1: through the Miocene/Pliocene boundary. The TerrestrialEnvironment, ed. P Fritz, Nature 389:153-58 JC Fontes, pp. 329-406. Amsterdam:El- 8. Cerling TE, HarrisJM. 1999. Carboniso- sevier tope fractionationbetween diet and bioa- 17. Ehleringer JR, Sage RF, Flanagan LB, patite in ungulate mammals and implica- Pearcy RW. 1991. Climate change and tions for ecological and paleoecological the evolution of C4 photosynthesis.Trends studies. Oecologia 120:347-63 Ecol. Evol. 6:95-99 9. Cerling TE, Harris JM, MacFadden BJ. 18. EhleringerJR, Cerling TE, Helliker BR. 1998. Carbon isotopes, diets of North 1997. C4 photosynthesis, atmospheric American equids, and the evolution of C4 C02, and climate. Oecologia 112:285-99 grasslands.In Stable Isotopes: Integration 19. FarquaharGD, EhleringerJR, Hubick KT. of Biological, Ecological, and Geochemi- 1989. Carbonisotopic discriminationand cal Processes, ed. H Griffiths,pp. 363-79. photosynthesis.Annu. Rev. Plant Physiol. Oxford, UK: Bios Sci. Pub. Plant Mol. Biol. 40:503-37 10. Cifelli RL. 1982. Patterns of evolution 20. Feranec RS, MacFadden BJ. 2000. Evo- among Artiodactyla and Perissodactyla. lution of the grazing niche in Pleistocene Evolution 35:433-40 mammals: evidence from stable isotopes. 11. Cifelli RL. 1985. South American un- PalaeogeogrxPalaeoclimatol. Palaeoecol. gulate evolution and extinction. In The In press GreatAmericanBiotic Interchange,ed. FG ?1. Flynn JJ, Swisher CC. 1995. Ceno- Stehli, SD Webb, pp. 249-66. New York: zoic South American land mammal Plenum ages: correlation to global chronolo- 12. Clemens WA, Keilan-JaworoskaZ. 1979. gies. In Geochronology,Time Scales and Multituberculata.In Mesozoic Mammals: Global StratigraphicCorrelation, ed. WA The First Two-Thirds of Mammalian Berggren,DV Kent, MP Aubry,J Harden- History, ed. JA Lillegraven, Z Kielan- bol, pp. 317-33. Tulsa,OK: Soc. Sed. Geol. Jaworoska, WA Clemens, pp. 99-149. Publ. 54. Berkeley:Univ. Calif. Press A2. Gingerich PD. 1981. Variation,sexual di- 13. Clemens WA, Lillegraven JA, Lindsay morphism,and social structurein the early EH, Simpson GG. 1979. Where, when, Eocene horse Hyracotherium(Mammalia, and what-a survey of known Mesozoic Perissodactlya).Paleobiology 7:443-55 mammal distribution.In Mesozoic Mam- D3. Hayek LA, Bernor RL, Solounias N, 56 MACFADDEN

Steigerwald P. 1991. Preliminarystudies tional durability of their dentitions, with of hipparionine horse diet as measured special reference to limiting factors. Bio. by tooth microwear. Ann. Zool. Fennici Rev. 63:197-230 28:187-200. 34. Janis CM, Scott KM, Jacobs LL, eds. 24. Hickey LJ, West RM, Dawson MR, Choi 1998. Evolution of TertiaryMammals of DK. 1983. Arctic terrestrial biota: pa- North America. Vol. 1: TerrestrialCarni- leomagnetic evidence of age disparity vores, Ungulates, and UngulatelikeMam- with mid-northern latitudes during the mals. New York: CambridgeUniv. Press. late Cretaceousand early Tertiary.Science 691 pp. 221:1153-56 35. Kay RF, MacFadden BJ, Madden RH, 25. HulbertRC. 1988. Calippusand Protohip- SandemanH, Anaya F. 1997. Revised age pus (Mammalia,Perissodactyla, Equidae) of the Salla Beds, Bolivia, and its bearing from the Miocene (Barstovian-early on the age of the Deseadan South Ameri- Hemphillian) of the Gulf Coastal Plain. can land mammal"age." J. Vert.Paleontol. Bull. Florida Mus. Nat. Hist., Biol. Sci. 18:189-99 33:229-338 36. Koch PL. 1998. Isotopic reconstructionof 26. Hutchinson GE 1959. Homage to Santa past continentalenvironments. Annu. Rev. Rosalia, or why are there so many kinds Earth Planet. Sci. 26:573-613 of animals?Am. Nat. 93:145-59 37. Koch PL, Zachos JC, GingerichPD. 1992. 27. Jacobs BF, KingstonJD, Jacobs LL. 1999. Correlationbetween isotope recordsin ma- The origin of grass-dominatedecosystems. rine and continentalcarbon reservoirs near Ann. Missouri Bot. Gard. 86:590-643 the Palaeocene/Eocene boundary.Nature 28. Janis CM. 1984. The use of fossil ungulate 358:319-22 communities as indicators of climate and 38. Kowalevsky V. 1873. Sur L'Anchitherium environment.In Fossils and Climate,ed. P aurelianense Cuv. et sur l'histoire Brenchleypp. 85-104. New York:Wiley pal6ntologique des chevaux. Mem. 29. Janis CM. 1988. An estimation of tooth L'Acad. Impe'riale Sci. St. Petersbourg, volume and hypsodonty indices in ungu- 7th Ser. 20(5):1-73, 3 plates late mammals,and the correlationof these 39. Lambert WD. 1992. The feeding habits factors with dietary preference. In Teeth of the shovel-tusked gomphotheres: evi- Revisited: Proc. VIIth Int. Symp. Dental dence from tusk wear patterns.Paleobiol- Morphology, ed. DE Russell, JP Santoro, ogy 18:132-47 D Sigogneau-Russell, pp. 367-87. Me'm. 40. LatorreC, Quade J, McIntosh WC. 1997. Mus. Nat. Hist. Nat., Paris, Se'r.C, 53 The expansion of C4 grasses and global 30. JanisCM. 1989. A climaticexplanation for change in the late Miocene: evidence from patternsof evolutionarydiversity in ungu- the Americas. Earth Planet. Sci. Lett. late mammals.Palaeontology 32:463-81 146:83-96 31. Janis CM. 1993. Tertiarymammal evolu- 41. Lee-ThorpJA, Van der Merwe NJ. 1987. tion in the context of changing climates, Carbonisotope analysis of fossil bone ap- vegetation,and tectonic events.Annu. Rev. atite. S. Afr J. Sci. 83:712-75 Ecol. Syst. 24:467-500 42. Lucas SG, Schoch RG, Williamson TE. 32. Janis CM, EhrhardtD. 1988. Correlation 1998. Taeniodonta.In Evolutionof Tertiary of relativemuzzle width andrelatve incisor Mammalsof NorthAmerica, ed. CM Janis, width with dietarypreference in ungulates. KM Scott, LL Jacobs, pp. 260-67. New Zool. J LinneanSoc. 92:267-84 York:Cambridge Univ. Press 33. Janis CM, Fortelius M. 1988. On the 43. Lucas SG, Schoch RG. 1998. Tillodonta.In means whereby mammals achieve func- Evolution of TertiaryMammals of North CENOZOICMAMMALIAN HERBIVORES 57

America, ed. CM Janis, KM Scott, LL 54. MacFadden BJ, Wang Y, Cerling TE, Jacobs,pp. 268-73. New York:Cambridge Anaya F. 1994. South American fos- Univ. Press sil mammals and carbon isotopes: a 25- 44. Lull RS. 1929. A remarkableground sloth. million- sequence from the Boli- Mem. PeabodyMus. Nat. Hist. 3:1-39 vian Andes. Palaeogeogr.Palaeoclimatol. 45. LundeliusEL, et al. 1987. The NorthAmer- Palaeoecol. 107:257-68 ican Quaternary sequence. In Cenozoic 55. MacGinitieHD. 1962. The Kilgore flora:a Mammalsof NorthAmerica: Geochronol- late Miocene florafrom northern Nebraska. ogy and ,ed. MO Wood- Univ. Calif. Pub. Geol. Sci. 35:67-158 burme,pp. 211-35. Berkeley: Univ. Calif. 56. McKenna MC. 1972. Was Europe con- Press nected directly to North America prior to 46. MacFadden BJ. 1990. Chronology of the middle Eocene? In EvolutionaryBiol- Cenozoicprimate localities in SouthAmer- ogy, ed. T Dobzhansky, MK Hecht, WC ica. J. Hum. Evol. 19:7-21 Steere, pp. 179-88. New York:Appleton- 47. MacFaddenBJ. 1992. Fossil Horses: Sys- Century-Crofts tematics, Paleobiology, and Evolution of 57. McKennaMC. 1975. Fossil mammalsand the FamilyEquidae. New York:Cambridge the earlyEocene land continuity.Ann. Mis- Univ. Press 369 pp. souri. Bot. Gard. 62:335-53 48. MacFaddenBJ 1997. Originand evolution 58. McNaughtonSJ. 1991. Evolutionaryecol- of the grazing guild in New World terres- ogy of large tropical herbivores.In Plant- trialmammals. Trends Ecol. Evol. 12:182- Animal Interactions: Evolutionary Ecol- 87 ogy in Tropical and TemperateRegions, 49. MacFaddenBJ, Cerling TE. 1996. Mam- ed. PW Price et al, pp. 509-22. New York: malian herbivore communities, ancient Wiley feeding ecology, andcarbon isotopes: a 10- 59. McNaughtonSJ, TarrantsJL, McNaughton million-yearsequence from the Neogene of MM, Davies RH. 1985. Silica as a defense Florida.J. Vert.Paleontol. 16:103-15 against herbivory and as a growth pro- 50. MacFaddenBJ, CerlingTE, PradoJ. 1996. moterin African grasses. Ecology 66:528- Cenozoic terrestrialecosystem evolutionin 35 Argentina:evidence from stable isotopes 60. MartinPS. 1973. The discovery of Amer- of carbonisotopes of fossil mammalteeth. ica. Science 179:969-74 Palaios 319-327 61. MartinPS. 1984. Prehistoricoverkill: the 51. MacFadden BJ, Cerling TE, Harris JM, global model. In QuaternaryExtinctions: Prado J. 1999. Ancient latitudinal gradi- A Prehistoric Revolution, ed. PS Martin, ents of C3/C4grasses interpretedfrom sta- RG Klein, pp. 354-403. Tucson: Univ. ble isotopes of New World Pleistocene Ariz. Press horse (Equus) teeth. Global Ecol. Bio- 62. MatthewWD. 1926. The evolution of the geogr 8:137-49 horse: a record and its interpretation.Q. 52. MacFaddenBJ, ShockeyBJ. 1997. Ancient Rev.Biol. 1:139-85 feeding ecology and niche differentiation 63. Novacek MJ. 1999. 100 million years of of Pleistocenemammalian herbivores from land vertebrateevolution: the Cretaceous- Tarija,Bolivia: morphologicaland isotopic early Tertiary transition. Ann. Missouri. evidence. Paleobiology 23:77-100 Bot. Gard. 86:230-58 53. MacFaddenBJ, Solounias N, Cerling TE. 64. Olson EC. 1952. The evolution of a Per- 1999. Ancient diets, ecology, and ex- mian vertebrate chronofauna. Evolution tinction of 5-million-year-oldhorses from 6:181-96 Florida.Science 283:824-27 65. Olson EC. 1966. Communityevolution and 58 MACFADDEN

the origin of mammals. Ecology 47:291- 78. Shockey BJ. 1997. Two new notoungulates 302 (family )from the Salla Beds 66. Owen-Smith N. 1988. Megaherbivores: of Bolivia (Deseadan:late Oligocene): sys- The Influence of Very Large Body Size tematics and functional morphology. J. on Ecology. Cambridge,UK: Cambridge Vert.Paleontol. 17:584-99 Univ. Press. 369 pp. 79. Simpson GG. 1950. The Nhistory of the 67. Pascual R, Ortiz JaureguizarEO. 1990. fauna of South America.Am. Sci. 38:261- Evolving climates and mammal faunas in 89 Cenozoic South America. J. Hum. Evol. 80. Simpson GG. 1951. Horses. Oxford, UK: 19:23-60 Oxford Univ. Press. 247 pp. 68. Pascual R, Ortiz JaureguizarEO. 1993. 81. Simpson GG 1953. The Major Features Evolutionarypattern of land mammalfau- of Evolution. New York: Columbia Univ. nas during the late Cretaceous and Pa- Press. 434 pp. leocene in South America: a comparison 82. Simpson GG. 1980. Splendid Isolation: with the North American pattern. Ann. The Curious History of South American Zool. Fennici 28:245-52 Mammals. New Haven: Yale Univ. Press. 69. PattersonB. 1956. Early Cretaceousmam- 266 pp. mals and the evolution of mammalianmo- 83. Solounias N, Hayek LA. 1993. New meth- lar teeth. Fieldiana Geol. 13:1-105 ods of tooth microwear analysis and ap- 70. Patterson B, Pascual, R. 1972. The fos- plication to dietary determinationof two sil mammal faunas of South America. In extinct antelopes.J. Zool. 229:142-55 Evolution,Mammals, and SouthernConti- 84. Solounias N, Moelleken SMC. 1993. Di- nents, ed. A Keast, FC Erk, B Glass, pp. etary adaptationof some ruminantsdeter- 247-309. Albany: State Univ. NY Press mined by premaxillary shape. J. Mamm. 71. ProtheroDR. 1994. TheEocene-Oligocene 74:1059-71 Transition: Paradise Lost. New York: 85. Solounias N, Moelleken SMC, Plavcan ColumbiaUniv. Press. 291 pp. JM. 1995. Predicting the diet of extinct 72. ProtheroDR. 1999. Does climatic change bovids using masseteric morphology. J. drive mammalian evolution? GSA Today Vert.Paleontol. 15:795-805 9:1-7 86. Solounias N, TeafordM, WalkerA. 1988. 73. Prothero DR, Berggren WA, eds. 1992. Interpretingthe diet of extinct ruminants: Eocene-Oligocene Climate and Biotic The case of a non-browsinggiraffid. Pale- Evolution. Princeton,NJ: PrincetonUniv. obiology 14:287-300 Press. 568 pp. 87. StebbinsGL. 1981. Coevolutionof grasses 74. Quade J, et al. 1992. A 16 million year and herbivores.Ann. Missouri Bot. Gard. record of paleodiet from Pakistan using 68:75-86 carbonisotopes in fossil teeth. Chem.Geol. 88. Stehli FG, Webb SD, eds. 1985. The Great (Isotope Geosci. Sec.) 94:183-92 American Biotic Interchange.New York: 75. Radinsky LR. 1976. Oldest horse brains: Plenum. 532 pp. more advanced than previously realized. 89. TerTiJA, Stowe LG. 1976. Climatic pat- Science 194:626-27 ternsand distribution of C4grasses in North 76. RadinskyLR. 1983. Allometryand reorga- America. Oecologia 23:1-12 nizationin horse skull proportions.Science 90. Voorhies MR, Thomasson JR. 1979. 221:1189-91 Fossil grass antothecia within Miocene 77. Roth VL. 1981. Constancy in the size skeletons: diet of an extinct rations of sympatric species. Am. Nat. species. Science 206:331-33 118:394-404 91. WangY, CerlingTE, MacFaddenBJ. 1994. CENOZOICMAMMALIAN HERBIVORES 59

Fossil horses and carbon isotopes: new 267-306. Chicago: Univ. Chicago Press evidence for Cenozoic dietary, habitat, 97. Webb SD. 1984. Ten million years of and ecosystem changes in North America. mammal extinctions in North America. Palaeogeogr Palaeoclimatol. Palaeoecol. In QuaternaryExtinctions: A Prehistoric 107:269-79 Revolution,ed. PS Martin,HE Wright,pp. 92. Webb SD. 1969. The Burge and Min- 189-210. Tucson:Univ. Ariz. Press nechaduzaClarendonian mammalian fau- 98. Webb SD. 1991. Ecogeography and the nas of north-centralNebraska. Univ. Calif. Great American Interchange.Paleobiol- Pub. Geol. Sci. 78:1-191 ogy 17:266-80 93. Webb SD. 1976. Mammalian faunal dy- 99. Webb SD, BarnoskyA. 1989. Faunaldy- namics of the greatAmerican interchange. namics of Pleistocene mammals. Annu. Paleobiology 2:220-34 Rev.Earth Planet. Sci. 17:413-38 94. Webb SD. 1977. A history of savannaver- 100. Wing SL, Alroy J, Hickey LJ. 1995. tebrates in the New World. Part I. North Plant and mammal diversity in the America.Annu. Rev.Ecol. Syst. 8:355-80 Paleocene to early Eocene of the 95. Webb SD. 1978. A history of savannaver- Bighom Basin. Palaeogeogr. Palaeocli- tebratesin the New World. Part II. South matol. Palaeoecol. 115:117-55 America and the GreatInterchange. Annu. 101. Wyss AR, et al. 1994. mam- Rev.Ecol. Syst. 9:393-426 mals from the Andes of central : a 96. WebbSD. 1983. The rise andfall of the late preliminarytaxonomic, biostratigraphic, Miocene ungulate fauna in North Amer- and geochronological assessment. Am. ica. In Coevolution, ed. MH Nitecki, pp. Mus. Novitates 3098: l-31