URSID SYSTEMATICS a Brief History of Taxonomy

HANGINGBEARS FROMPHYLOGENETIC TREES: INVESTIGATINGPATTERNS OF MACROEVOLUTION

JOHNL. GITTLEMAN,' Department of Ecologyand EvolutionaryBiology, University of Tennessee, Knoxville,TN 37996, USA

Abstract: Phylogeneticinformation of the family Ursidaeis well resolvedand readily available for investigatingmacroevolutionary questions. Using completephylogenies of the ursidsand related terrestrial carnivores, I investigatewhether patterns of body size andlife historyevolution in bearsdiffer fromother carnivores with respectto cladogenesis,species richness,and overall phyletic trends. Largebody size in bearsis not relatedto theirphyloge- netic history,in contrastto most othercarnivore taxa; this may relateto bears' relativelyrecent evolutionary history or theirlarge body size, which is flexible for utilizinglow-quality foods, thusbuffering environmental change. Also, ratesof body size evolutionin bearsare average or perhapsslightly slower thanother carnivores. Certain life historytraits (birth weight, age eyes open, inter-birthinterval, longevity) arevery differentin bearsrelative to other carnivores,even after accountingfor body size and phylogeny. In general,large body size, flexibility in phyletic change in size, and slow life historiesof ursidsmay be an effective evolutionarystrategy for dealingwith recentenvironmental stresses.

Ursus 11:29-40

Key words: bears,body size, life history,macroevolution, Ursidae

Bears are evolutionarilyrelated to terrestrialcarnivores lytical techniquescan be used to test macroevolutionary including canids, mustelids, hyaenas, and felids. Sys- hypotheses using phylogenies. tematically, ursids are separatedfrom these other taxa Two caveats are needed. Many of the patternsdetected by: (1) unspecialized incisors and molars with broad, here are preliminary;validity will rest with more and flat, tuberculatedcrowns; (2) the shearing function of betterdata. Also, emphasisis placed on patternsof mac- carassials in ancestraltaxa largely replacedby a crush- roevolutionarydifferences between bears and other car- ing function; (3) large and powerful legs, with planti- nivores rather than explanations for these differences. gradeposture; (4) an omnivorousdiet, with most species Futurework, especially at the populationlevel, will sort leaning towardfrugivory and folivory; and, (5) most dis- out why bears are unique. tinctive of all, a gigantic body. Clearly, these features are not only useful for systematics,they also elicit inter- esting ecological and evolutionaryquestions. Does large URSIDSYSTEMATICS body size in bearscontribute to higherextinction (or lower speciation) rates relative to other carnivores? Are bears A BriefHistory of Taxonomy hampered reproductively with respect to life histories Reviews of ursid systematics are found elsewhere becauseof the lengthof time it takesto grow large? These (Simpson 1945; Ewer 1973; Stains 1984; Flynn et al. questions are comparativein nature,specifically requir- 1988; Wayne et al. 1989; Wozencraft 1989a,b; Flynn ing detailed informationabout relative differences with 1996). Excluding the giant (Ailuropodamelanoleuca) other carnivore species and knowledge of evolutionary andred (Ailurusfulgens)pandas, the family Ursidaecom- history. prises the following species: Fortunately,we now have considerable information Helarctos malayanus, Malayan sun bear frombasic naturalhistory field studiesand, more recently, Melursus ursinus, sloth bear fromphylogenetic systematics. Armedwith phylogenetic Tremarctosornatus, Andean (spectacled) bear trees, we can examine the timing and patternof macro- Ursus americanus, Americanblack bear evolutionary changes among bears and closely related Ursus arctos, brown bear lineages. In this paper, I present findings from recent Ursus maritimus,polar bear studies comparingtrends in bears relative to other carni- Ursus thibetanus,Asiatic black bear vores with respect to: (1) phylogenetic trees, (2) macro- Relative to other carnivorefamilies, the taxonomy of evolutionarytrends of speciation and body size changes bears is uncontroversial. Ursidae is one of the most re- using these trees, and (3) variationin life historypaterns. cent taxonomicunits for familial designationin the class My aim is to show how bears differ from other carni- Mammalia (Simpson 1945), the last family of arctoids vores as well as to illustrateby example how mod ana- to appearin the fossil record (Stains 1984), and one of

1 Presentaddress: Department of Biology, GilmerHall, Universityof Virginia,Charlottesville, VA 22903-2477, USA, e-mail: JLGittleman@ Virginia.edu 30 Ursus 11:1999 the most homogeneous Carnivoranfamilies, as reflected Table 1. Indices relating to the distribution of taxonomic minimal subfamilial levels coverage for and the resolution of a composite tree of the by (i.e., Tremarctinae) (see Carnivora Bininda-Emonds et al. The in (from 1999). Wozencraft1989b, 1993). Populationvariance many parenthetical value of percent resolution for the herpestids ursids is significant and continues to be assessed, par- refers to when safe taxonomic reduction(Wilkinson 1995) was ticularly in the brown bear and polar bear (see Wayne used to improve the resolution of this family and Kurten and Anderson Koepfli 1996). Indeed, Number (1980:184) exclaimed in referenceto brown and grizzly of Numberof Percent Elements/ sourcetree/ source elements resolution taxon taxon bear systematics: "Some 232 Recent and 39 fossil "spe- Taxon trees cies" and "subspecies"(list in Erdbrink,1953) have been Higher groups 62 202 100 16.8 0.27 for this taxon-a waste of effortwhich, proposed systematic Mustelidae 30 155 72.7 3.4 0.11 as far as we is with the know, unparalled.".Fortunately, Otariidae 15 46 69.2 3.3 0.22 of some names above list- exception generic (see species Phocidae 21 118 94.4 6.2 0.30 and identification ing; also, Zhang Ryder 1993), species Ursidae 28 50 85.7 6.2 0.22 is the is not stable;therefore, present analysis hampered Canidae 36 180 69.7 5.3 0.15 Of course, there is debate about by species uncertainty. Felidae 40 282 97.1 7.8 0.20 taxonomic placementof the pandas,especially the giant Hyaenidae 6 8 66.7 2.0 0.33 panda,which sharesmany features(herbivorous, planti- Herpestidae 9 53 27.8 1.4 0.16 grade and gigantic) with ursids (O'Brien et al. 1985, (58.3) 1991; Schaller et al. 1985; Mayr 1986.) Both familial Viverridae 9 90 97 2.6 0.29 and ordinalanalyses (e.g., Wyss and Flynn 1993, Zhang and Ryder 1993, Vranaet al. 1994, Bininda-Emondset gruence among phylogenies places the giant panda and al. 1999) stronglyindicate that the giant pandais closely Andean (spectacled)bear as initial lineages in ursidevo- relatedto the bears. The red pandais more of an enigma, lution, the sloth bear as an ancient monotypic lineage, as it arguably can be placed in its own family, in the and the black and polar bears as recent taxa. Also, Procyonidae or in the Ursidae (Roberts and Gittleman throughoutthe family,there is completeresolution among 1984, Bininda-Emondset al. 1999). The comparative nodes, a major improvementfrom one of the first mo- analyses presentedhere will consider the red pandaas a lecular phylogenies (see Wayne et al. 1989). Of course, procyonidbecause of the manyfunctional characters sug- with the many phylogenies available for the ursids and gesting this placementand to not obscureobserved trends no clear methods for how to weight some characters(or between the ursids and other carnivores. phylogenies) against others, it is important to assess agreementamong all the availablephylogenetic hypoth- Phylogenies:Molecular and eses. Morphological Due to the relatively small numberof taxa, accessibil- A CompletePhylogeny Bininda-Emondset al. assembled a ity of sample material,the extendedtime thatbears have (1999) complete for all 271 of derivedfrom evolved, and a fairly complete fossil record,ursids have phylogeny species Carnivora, 166 matrix with received a greatdeal of phylogeneticstudy (Martin1989, phylogenies using representation parsi- Wayne et al. 1989). At least 28 phylogenetic trees are mony analysis (Baum 1992, Ragan 1992). Essentially, combination of available for ursids (Table 1), which is a relatively high matrix representationpermits phylogenetic treesderived from various data sources, even numbercompared to other carnivorefamilies, consider- though the data not be different and ing the numberof species. Further,many differenttypes may congruent; phylogenies are of characters,including morphological, molecular, and elements given equal weighting. source trees are available for carnivore behavioral and ecological elements, have been used in Most "higher" ursids have the fourth numberat the fam- phylogenetic studies (Wayne et al. 1989, Wozencraft taxa; greatest and mustelids 1989a, Bininda-Emondset al. 1999). As for all organ- ily level behind felids, canids, (Table 1). The that have been studiedmost also isms (Page and Holmes 1998), molecular phylogenies groups intensively tend to have the most elements used for recently have become plentiful for ursids; for example, study, although the ursids behind in this A expla- new phylogenies for the entire family are availablefrom lag respect. possible nation for this trend is that material is less cytochromeb gene (Zhang and Ryder 1993), combined postcranial accessible for such animals as bears, due to cytochrome b/tRNAPro and tRNAThrgenes (Talbot and large-bodied difficulties in and them in mu- Shields 1996), and partial sequence informationfrom 6 the transporting housing seum collections. Ursids however, favor- regions of mtDNA (Waitset al. 1999). In general, con- do, compare INVITEDPAPER * MACROEVOLUTION* Gittleman 31

Ursus arctos, Brown Ursus maritimus, Polar Ursus thibetanus, Asiatic black Helarctos malayanus, Sun Melursus ursinus, Sloth Ursus americanus, American black Tremarctos ornatus, Andean Ailuropoda melanoleuca, Giant panda

Fig. 1. A composite tree of the Ursidae (from Bininda-Emondset al. 1999). Numbers refer to primarynodes supported when merging trees. ably with respect to the numberof elements studied per pling, and accessible to questions about speciation and taxon and the number of elements per source tree per rates of evolution. The following results are based on taxon. the complete carnivorephylogeny of Bininda-Emondset What does a composite bearphylogeny look like, how al. (1999; see Fig. 2) unless otherwise noted. resolved is it, and how robust? Generally, the tree is fairly well resolved (Fig. 1), showing 86.7% resolution, Speciation Rates and is reasonablyrobust based on Bremersupport values Regardingthe entire phylogeny of the carnivores,do for the nodes (Bininda-Emondset al. 1999). Details of some lineages contain significantly more species than the tree indicate: others? The null expectationcomes from a model where 1. The giant panda and the Andean (spectacled) bear all extanttaxa have the same possibility to diversify (Nee are the oldest lineages, supportingrecent molecu- et al. 1995). The comparativetest identifies ancestral lar results (Zhang and Ryder 1993, Talbot and lineages that have given rise to a disproportionatenum- Shields 1996). ber of extant species in comparisonto their contempo- 2. The position of the American black bear remains raries (Purvis et al. 1995). The results show that 8 unclear and merits further work, especially with lineages containsignificantly more species thanexpected the monophyly of Ursus uncertain,as pointed out (Fig. 2). (It should be cautioned that the radiationsare by others (Goldmanet al. 1989, Zhang and Ryder not independent;a radiationis more likely to give rise to 1994). This is an especially vexing result given more taxa.) The bears are not one of these, despite the that non-monophylyis rare across the entire order fact that all of the significant radiations lie in the (e.g., Phoca, Vulpes, Lutra, Mustela, Leopardus, caniforms. It is interestingto speculatewhether some of Onciphelis). the generalcharacteristics uniting this group(e.g., primi- 3. Thereare sistertaxon links betweenthe brownbear- tive bullar construction,basicranial arterial circulation, polar bear and sun bear-sloth bear, also suggested postcranial skeleton) has encouraged species richness in other studies (Cronin et al. 1991, Talbot and relative to the feliforms. Shields 1996). Clearly,the majorweakness in the composite phylog- Body Size eny is the ancestral clade leading to the ursines; also, The most common explanationfor the unequaldistri- elevation of the polar bear to its own genus does not ap- bution of species among lineages rests with body size pear appropriate. (Gittlemanand Purvis 1998): small body size is associ- ated with high diversity because of niche partitioning, metabolic rate, reproductiverate, or brain size. Carni- MACROEVOLUTIONARYTRENDS vores are a good group to test this hypothesis in because Completespecies-level phylogenies permitmacroevo- the range of body sizes is greaterthan any other mam- lutionarystudy hithertonot possible (Purvis 1996). Phy- malian order, spanning over 4 orders of magnitude logenies arenecessary for any test involving evolutionary (Gittleman 1985). Morover, as shown, we know that history,but this informationis often unavailableor mis- there are significant differences in species-richness leading from taxonomies. A complete phylogeny makes among lineages of the same age. tests of macroevolutionarypattern more robust, less likely Using the complete species-level phylogenies of the to producemistakes simply due to biased taxonomicsam- carnivores and the ursids, we can test whether larger- .. .. .

32 Ursus 11:1999

"' " ' " -,,, A, ,,' 2 2I I I I I 60 40 20 0 Millions of yearsbefore present et al. 1999), including estimated times of Fig. 2. A composite tree for all 271 species of Carnivora(from Bininda-Emonds divergence as well as which lineages have given rise to significantly more extant descendents than expected. INVITEDPAPER * MACROEVOLUTION* Gittleman 33

4A '

a 2- cr3 31 agOoD a CA X Mu 0 B 0 3 u 0-2 I PI,C ;ba Carnivore contrasts 0 ?b UrlN ?i5a CA1 1 Andean vs other 'truebears' -2 - r-- u 2 Asiatic black bear vs brown and 03O polar -4 aE 3 Giantpanda vs rest of bears -6 0 1 2 3 4 5 4 Bears vs sister taxa (mustelids, procyoids,pinnipeds, and red panda) Body size contrasts

Fig. 3. Clade size and body size contrasts in carnivores (Gittlemanand Purvis 1998). The comparison on the extreme right is between Ailurusand the mustelid/procyonid clade. bodied clades (i.e., bears) contain more species than either within the ursids or relative to other carnivores. smaller-bodiedclades. Once body masses are calculated Again, using a phylogeny,we can investigatethis lack of for species withinclades, differences between sister clades pattern. For such a test, though, it is necessary to use are used to search for changes in size (Gittleman and phylogenetic informationthat is totally independentof Purvis 1998). Five sets of (clade) contrasts are useful the characterevolution we are tracing (i.e., body size), here: (1) contrastsacross all carnivores;(2) ursids ver- otherwise the test may not be independent(Gittleman et sus sister taxa (mustelids, procyonids, pinnipeds, red al. 1996). Taxonomic ranks (Wozencraft 1989b) are panda); (3) contrastsbetween Tremarctosand the other therefore employed along with a statistic (Moran's I; 'true' bears; (4) contrastsbetween Ursus thibetanusand Gittlemanand Kot 1989) to assess whetherphylogenetic U. arctos-U. maritimus; and (5) contrasts between change of body size in bears is unusualrelative to other Ailuropodaand all the otherbears. Overall,the caniform carnivores (data on carnivore body weight are from carnivores (canids, procyonids, pinnipeds, ursids and Gittleman and Purvis 1998 and available from the au- mustelids) do show a tendency for smaller bodied lin- thor). The simple answeris yes. A 'correlogram'shows eages to have more species (Gittlemanand Purvis 1998). that all carnivore families other than the ursids, However,on closer inspectionthe contrastsinvolving the herpestids, and hyaenids show significant correlation bears do not reveal a consistent pattern(Fig. 3): clade between taxonomicrank and body size (Fig. 4). In other contrasts of 3 and 4, for example, both have low clade words, close phylogenetic relationshipin the 3 families size contrastsbut very differentsize contrasts. Although does not necessarily predict variationin size. body size is often implicated as a correlate of species- If a taxonomy reflects phylogenetic relationshipsand richness in mammals,much of the variationin diversity closely related taxa are phenotypicallymore alike, then cannot be attributedto size differences. body size shouldcorrelate with phylogeny (or in this case, taxonomic ranks). Ursids are clearly very different in PhylogeneticPattern this respect: along with hyenids and herpestids, ursids The above phylogenetic tests fail to show that clado- show no phylogenetic patternin body size. An obvious genesis in bears is unusual relative to other carnivore explanationfor this is small sample size. This may ex- taxa or that large body size is, as predicted, correlated plain some of the lack of correlation,but not all, as the with species richness. An obvious explanation is that procyonidsare a smallfamily andshow significancewhile body size in bears does not reflect phylogenetic history, the herpestids are a relatively large group and do not show significant correlation. 34 Ursus 11:1999

Rates of Evolution and time (Fig. 5), as expected, given that measures of One mechanismfor the decouplingof body size evolu- ratechange decline over longerperiods (Gingerich 1993). tion from phylogenetic distance lies in differentialrates Against this overall trend we can see that the ursids ap- of evolution. If body size in ursidseither changes slowly pear to change slightly slower than the other carnivore or increases and decreases at unequal rates, as Kurten taxa. This suggests that perhaps the bears are indeed (1968) once proposed,then statisticaltests of phylogeny adoptingan evolutionarystrategy of "lifein the slow lane" and body size will reveal differentpatterns than that pro- (Macdonald1992), a classic K-selectionstrategy for deal- ducedby a typical gradualmodel of evolutionarychange. ing with stable but stressful environments. Although it One method for exploring this idea is to calculate rates has clearly been shown that the r-K selection theory is of evolutionary change in body size through phyloge- not useful at population level (Stears 1992), phylogenetic time. We (Gittlemanet al. 1996) calculated rates netic analyses are now showing that the theory may ap- of evolutionin Darwin's- quantitativechange in a char- ply to significant divergences in size, life histories and acter from its initial dimension to its final dimension abundanceacross higher taxonomic levels (McKinney divided by the amountof time elapsed (Gingerich 1993) and Gittleman 1995, McKinney 1997). In the size of bears is an obvious -for body size and othertraits among variousmammal summary, gigantic to other terres- taxa and found that, in general, morphological traits distinguishing characteristiccompared tests show evolved at slower rates than ecological or life history trialcarnivores. The above phylogenetic that, rates across carnivore traits. Might therebe differencesin rates of evolution of despite unequal speciation taxa, the size of bears is unrelatedto rich- body size change within carnivores? The general pat- large body species a test of correlationbetween tern is a negative relationshipbetween rates of evolution ness. Further, simple phy-

z 0

-10 g sbf f

Fig. 4. Correlogram showing observed patterns of z scores (from Moran's /) for body weight across carnivore families. Phylogenetic patterns reflectcorrelation of body weight (fromleft to right)among species with genera, genera withinsubfamilies, and subfamilies withinfamilies. INVITEDPAPER * MACROEVOLUTION* Gittleman 35

01 , + Other-Other O Bear-Other - Bear-Bear 100

+ +

+ c102 :++ ++++ -_ -?10- + -

+ c +1 - +

E10 oc +'- + - ,~~~+t 10

' ' ' ' ' 10'o 1 1 10 100 Time (MY)

Fig. 5. Log of evolutionary rate (darwins) versus log time interval (million years ago) observed across designated carnivores for body weight. logenetic distance and body size is not significant in the COMPARISONSWITH OTHER bears, which is unusual for carnivores and mammals in CARNIVORES general. Thereare many potentialreasons for these negative results, including the possibility that the phyloge- Tinkeringwith Size, Life Histories, and netic information used here is wrong. One biological explanationis that rates of evolutionary change in body Phylogeny Life history patterns in carnivores are extremely di- size are differentin the bears;here, some supportis given verse 1986, For com- to this view. Also, the dramatic size increase in bear (Gittleman 1993). example, simple parison of the largest and smallest species is profound: body size may have surpassedexpected change given their By the time a female polar bear reaches sexual maturity relativelyrecent evolutionary history. Anothermore eco- at around5-6 years of age, a least weasel (Mustelanivalis) logical explanation (Stirling and Derocher 1990) is that would already have 56 descendants! Such a fundamen- the large size in bears confers an adaptive strategy,lend- tal difference has significant effects on population dy- ing flexibility to environmentalchange, which separates namics, ecological adaptations,and evolutionarychange. bears from expected phylogenetic pattern. Clearly, the Comparisonsof mammalianlife history patternsinvolve massive within-species variability in body size, for ex- traits such as length, birth weight, inter- ample the 2-3-fold increase in total body mass of polar many gestation birth interval, and longevity. By comparing single traits bears during the spring foraging season (Cattet et al. across the carnivores,it is to isolate which vari- 1997), is indication that size evolution is unusual possible body ables are differentin the thus reveal- in bears. significantly bears, ing relative macroevolutionaryshifts in the ursids and 36 Ursus 11:1999

Table 2. Comparative relations of life history patterns among carnivore families after statistically removing effects of size bear and *P < 0.05; **P < 0.001. All other tests are not . y 0 Brown taxonomy. (a) ./ aPolar bear significant. 6 Blackbear At ^ a Lifehistory trait Familycomparisons (sample size) - .Gianta_0 5 panda Gestationlength 2.02 (6,90) A - 4 Birth weight 3.12 (6,61)** 0: Canidae 'S 0: Ursidae 1.11 (6,64) 3 Weaningage Ailurus *o^~ 21^ ~ ~ A: 2.50 E 0: Ailuropoda Longevity (6,48)* S 2. 4 Procyonidas .*^'C~~/^y'^r~~~~2: Mustelldae Age at sexual maturity 0.80 (6,58) C5 / : Viverridae 1t /X /l: Hyaenidae Inter-birthinterval 2.64 (6,50)* 1 A.: Felidae Age eyes open 7.01 (6,63)** -1 Growthrate 0.81 (3,36)

4 -3 -2 -1 0 1 2 3 4 5 6 Logarithm of female body weight (kg)

/7"K-selected" y .17x + 2.143, r2 .461 3.75, ~375.~ ~ = / ? - .., (b) // -o . 3.5- ?'*cladea A Brown size bear size ?;..-::./,,' , .-'cladea . (It Black bear 0 ,- dcadeb .. - 'stress-selected" - 3.25-5 X O .. . _ AA ^ Polarbear E 3- age asaGiant * panda a 2.75. temporal traits ._ / ' 2.5- lc2X m AXO

) 2.25- O: Canidae litter -.":_~~~ /*?V~ ~ ~ t0: Ursidae size, 2 O A: Ailurus E / : Ailuropoda fecundity Y ~1.75* ~- */~+.:*J~ Procyonidas etc. 1.75- 0 ODA A A X: Mustelidae t D ?: Viverridae 1J.5: Hyaenidae Fig. 7. A model of clade diffusion with respect to size and 1 5 size-related traits (e.g., life histories). Species in more -4 3e 0 b 2 3 4 5 6 closely related clades such as the bears tend to be more of Logarithm female body weight (kg) similar and reflect how increased size and life histories may decouple when "stress-selected". After McKinney and Gittleman (1995). y .132x + 2.945, r = .122 4- GiantpandaG life features that are critical for man- lack bear C Pollar bear potential history X i B 3.5, X 3SW2 (C)(c) and conservation decisions. l x [ Bro\wn bear agement 21 To perform systematic comparisons of life histories

0? across carnivores, it is necessary to account for 2 vari- C 2.5, O*A A 0 0D 0 AA I ables. First, size is involved: a least weasel O body clearly 0 A A A cannot give birth to young the size of a polar bear, nor a' Asiatic black bear 0 Canidae 05 0 Ursidae can a bear as often as a weasel. 0. .5 A Ailurus polar reproduce Thus, O Ailuropoda of life histories must include . A + Procyonidae comparisons initially plot- E X Mustelldae * Viverrtdae ting the allometry of life history traits against body size * cm Hyaenidae o .5 A Felidae (typically female body weight, as we are interested in reproductive traits), then using residuals for compara- 0 0 tive tests (see Gittleman 1986, 1993). Second, as with ------histories are -4 -3 -2 1 0 1 2 3 4 5 6 most traits (Harvey and Pagel 1991), life Logarithm of female body weight (kg) correlated with phylogeny, thus comparative tests may be weakened not data In- 6. Plots of residual values - allometric and by using independent points. Fig. following in a of 13 life traitsacross phylogenetic autoregressive conversions - for carnivore life deed, study history carnivores, history traits and female body weight. Included are (a) birth each trait was significantly related to phylogeny weight, (b)inter-birth interval, and (c) age eyes open. INVITEDPAPER * MACROEVOLUTION* Gittleman 37

(Gittleman 1993). There are many ways to solve this This is particularlyevident in large predatoryspecies, statistical problem (Purvis et al. 1994, Martins and such as saber-toothedcats (Smilodon), which have had Hansen 1996). Here, I use an autoregressive method repeated extinction events (Van Valkenburgh 1991, because it is appropriatewhen using taxonomic rank in- 1999). In opting for a large body size, the ursids seem- formationand sample sizes commensuratewith the car- ingly have employed a safer strategy: herbivory,flex- nivores. In sum, the following analyses are based on 2 ibility in diet and size, and protracted life histories. sets of residuals(allometric and autoregressive), thus pro- Relative to other carnivores,this strategymay be work- viding conservative tests for which variables the bears ing, at least over macroevolutionarytime scales. The differ (or do not) from other carnivores. The carnivore frequency of extinction of ursids duringthe Pleistocene life history data, definitions of variables, and details of appearsrelatively low comparedto other carnivoretaxa methodology are in Gittleman(1986, 1993). (Kurten and Anderson 1980) and, at present, the only I examined 8 life history traitsfor differencesbetween bear species to receive endangeredclassification is the bears and the remainingcarnivore families (due to avail- giant panda. This is not a message for complacency. ability of data, most analyses only include brown, black, Rather,macroevolutionary trends suggest that the ursids and polar bears, and giant panda). Four traits showed have adopteda size and life history strategyfor the long- significant differences: birth weight, longevity, inter- haul. Bears, like all other animals, obviously have their birthinterval, and age eyes first open (Table2). In each limits; many large, charasmatic mammals like mast- of these the ursids lie at the extreme of bivariatedistri- odonts (Mammut),Irish elk (Megaloceros), and saber- butions: bears have comparativelysmall neonates (Fig. tooths are dramatic examples. We now need critical 6a), long inter-birthintervals (Fig. 6b), and late devel- informationto find out what precise environmentalfac- opmental ages for opening eyes (Fig. 6c). These find- tors influence changes in size and life histories, the ex- ings generally agree with previous work (e.g., Eisenberg act traitswhich makebears distinct from othercarnivores 1981; Gittleman 1986, 1993), although greaterempha- as well as vulnerableto extinction. sis now is placed on just how differentthe bears are with respect to some life histories, as a stringenttest here ac- counts for size and phylogeny. ACKNOWLEDGMENTS At least 6 general factors are considered to influence I thankM. Pelton and G. Burghardtfor inviting me to such life historydifferences (for reviews see Boyce 1988, present this work at the meeting, T. Fuller and M. Harvey et al. 1989, Gittleman 1994): body size, brain Vaughanfor comments on the manuscript,and the De- size, metabolic rate, phylogeny, ecology, and mortality. partmentof Ecology and EvolutionaryBiology, Univer- All of these may,to some extent,drive life historiesamong sity of Tennessee, for support. ursids. Along with the flexibility of body size evolution in bears, as shown above, some life histories appearto be relatively decoupled from allometricexpectation. If we LITERATURECITED consider both large body size and slow life histories a BAUM,B.R. 1992. Combiningtrees as a way of combining response to environmentalchange or stress (Fig. 7), then data sets for phylogenetic inference, and the desirabilityof a model for how these factorsare simultaneously respond- combininggene trees. Taxon42:637-640. ing may be applicable (McKinney and Gittleman 1995, BININDA-EMONDS,O.R.P., J.L. GITTLEMAN,AND A. PURVIS.1999. Buildinglarge trees by information: McKinney 1997). Some life history traits (birthweight, combiningphylogenetic A completephylogeny of theextant Carnivora age eyes open, inter-birth interval) are much (Mammalia). clearly BiologicalReviews 74:143-175. slower in bears than in other carnivores. Environmental BOYCE,M.S., EDITOR. 1988. Evolution of life histories of stressessuch as food or fluctuating supply declininghabi- mammals.Yale University Press, New Haven, Connecticut, tat availabilty may be sufficiently handled by reduced USA. reproductiverates during these periods. CATTET,M.R.L., S.N. ATKINSON,S.C. POLISCHUK,AND M.A. RAMSAY.1997. Predicting body mass in polar bears: is morphometricsuseful? Journal of Wildlife Management IMPLICATIONSFOR LONG-TERM 61:1083-1090. CONSERVATIONAND EVOLUTION CRONIN,M.A., S.C. AMSTRUP,G.W. GARNER,AND E.R. VYSE. 1991. Interspecificand mitochrondrialDNA the end of a intraspecific Extinction, phyletic line without replace- variationin NorthAmerican black bears Canadian has (Ursus). ment, occurredthroughout the history of life and is Journalof Zoology 69:2985-2992. the ultimate of destiny every species. Many mammals EISENBERG,J.F. 1981. The mammalianradiations. University have gone throughrapid and dramaticextinction events. of Chicago Press, Chicago, Illinois, USA. 38 Ursus 11:1999

ERDBRINK,D.P. 1953. A review of fossil and recent bears of Pages 536-568 in J.L. Gittleman, editor. Carnivore the Old World with remarkson their phylogeny based on behavior,ecology, and evolution. CornellUniversity Press, their dentition. Jan de Lange, Deventer, Netherlands. Ithaca,New York,USA. EWER,R.F. 1973. The carnivores. Cornell University Press, MARTINS,E.P., AND T.F. HANSEN. 1996. The statisticalanalysis Ithaca,New York,USA. of interspecificdata: a review andevaluation of phylogenetic FLYNN,J.J. 1996. Camivoranphylogeny andrates of evolution: comparativemethods. Pages 22-75 in E.P.Martins, editors. morphological,taxic, and molecular. Pages 542-581 in J.L. Phylogeniesand the comparativemethod in animalbehavior. Gittleman, editor. Carnivore behavior, ecology, and Oxford University Press, New York,New York,USA. evolution. Volume2. CornellUniversity Press, Ithaca,New MAYR,E. 1986. Uncertaintyin science: is the giant panda a York,USA. bear or a raccoon? Nature 323:769-771. _ , N. NEFF, ANDR. TEDFORD.1988. Phylogeny of the McKINNEY,M.L. 1997. Extinctionvulnerability and selectivity: Carnivora. Pages 73-116 in M.J. Benton, editor. The combining ecological and paleontological views. Annual phylogeny and classification of tetrapods. Volume 2. Review of Ecology and Systematics 28:495-516. ClarendonPress, Oxford, United Kingdom. _ , ANDJ.L. GITTLEMAN.1995. Ontogeny and phylogeny: GINGERICH,P.D. 1993. Quantification and comparison of tinkeringwith covariationin life history, morphology and evolutionaryrates. AmericanJournal of Science 293A:453- behaviour. Pages 21-47 in K.J. McNamara, editor. 478. Evolutionarychange and heterochrony.John Wiley & Sons, GITTLEMAN,J.L. 1985. Carnivorebody size: ecological and Chichester,United Kingdom. taxonomic correlates. Oecologia 67:540-554. NEE, S., E.C. HOLMES, A. RAMBAUT, AND P.H. HARVEY. 1995. _ . 1986. Carnivore life history patterns: allometric, Inferringpopulation history from molecular phylogenies. phylogenetic, and ecological associations. American PhilosophicalTransactions of the Royal Society of London Naturalist 127:744-771. B 344:25-31. . 1993. Carnivore life histories: a reanalysis in the O'BRIEN,S.J., R.E. BENVENISTE,W.G. NASH, J.S. MARTENSON, light of new models. Pages 65-86 in N. Dunstone and M M.A. EICHELBERGER,D.E. WILDT,M. BUSH,R.K. WAYNE,AND Gorman,editors. Mammals as predators.Oxford University D. GOLDMAN.1991. Molecular biology and evolutionary Press, Oxford, United Kingdom. theory: the giant panda'sclosest relatives. Pages 225-250 . 1994. Are the pandas successful specialists or in L. Warrenand H. Koprowski,editors. New perspectives evolutionaryfailures? BioScience 44:456-464. on evolution. Wiley-Liss, New York,New York,USA. , C.G. ANDERSON,M. KOT, AND H.K. LUH. 1996. , W.G. NASH,D.E. WILDT,AND R.E. BENVENISTE.1985. Comparativetests of evolutionarylability and rates using A molecular solution to the riddle of the giant panda's molecular phylogenies. Pages 289-307 in P.H. Harvey, phylogeny. Nature 317:140-144. A.J.L. Brown, S. Nee, and J. MaynardSmith, editors. New PAGE,R.D.M., ANDE.C. HOLMES.1998. Molecular evolution. uses for new phylogenies. OxfordUniversity Press, Oxford, Blackwell, Oxford, United Kingdom. United Kingdom. PURVIS,A. 1996. Using interspecies phylogenies to test _ , ANDM. KOT. 1989. Adaptation: statistics and a null macroevolutionary hypotheses. Pages 153-168 in P.H. model for estimating phylogenetic effects. Systematic Harvey, A.J.L. Brown, S. Nee, and J. Maynard Smith, Zoology39:227-241. editors. New uses for new phylogenies. OxfordUniversity _ , ANDA. PURVIS.1998. Body size and species-richness Press, Oxford, United Kingdom. in carnivoresand primates. Proceedings of the Royal Society ?____, J.L. GITTLEMAN,AND H.-K. LUH. 1994. Truth or of London B 265:113-119. consequences: effects of phylogenetic accuracy on two GOLDMAN,D., P.R. GIRI,AND S.J. O'BRIEN.1989. Molecular comparative methods. Journal of Theoretical genetic-distanceestimates among Ursidae as indicated by Biology167:293-300. one- andtwo-dimensional protein electrophoresis. Evolution _ , S. NEE, ANDP.H. HARVEY. 1995. Macroevolutionary 43:282-295. inferencesfrom primate phylogeny. Proceedings of the Royal HARVEY,P.H., ANDM.D. PAGEL.1991. The comparative method Society of LondonB 260:329-333. in evolutionarybiology. Oxford University Press, Oxford, RAGAN,M.A. 1992. Phylogenetic inference based on matrix United Kingdom. representation of trees. Molecular Phylogenetics and , A.F. READ,AND D.E.L. PROMISLOW.1989. Life history Evolution 1:53-58. variation in placental mammals: unifying the data with ROBERTS,M.S., ANDJ.L. GITTLEMAN.1984. Ailurus fulgens. theory. Oxford Surveys in EvolutionaryBiology 6:13-31. MammalianSpecies Account 222. KURTEN, B. 1968. Pleistocene mammals of Europe. SCHALLER,G.B., H. JINCHU,P. WENSHI,AND Z. JING. 1985. The Weidenfieldand Nicolsen, London, United Kingdom. giant pandas of Wolong. University of Chicago Press, _ ANDE. ANDERSON.1980. Pleistocene mammals of North Chicago, Illinois, USA. America. Columbia Univ. Press, New York, New York, SIMPSON,G.G. 1945. The principles of classification and a USA. classification of mammals. Bulletin of the American MACDONALD,D. 1992. The velvet claw: a naturalhistory of Museum of NaturalHistory 85:1-350. the carnivores. BBC Books, London, United Kingdom. STAINS,H.J. 1984. Carnivores. Pages 491-521 in S. Anderson MARTIN,L.D. 1989. Fossil history of the terrestrialCarnivora. and J.K. Jones, Jr., editors. Orders and families of recent INVITEDPAPER * MACROEVOLUTION* Gittleman 39

mammalsof the world. Wiley, New York,New York,USA. Ithaca,New York, USA. STEARNS,S.C. 1992. The evolution of life histories. Oxford , R.E. BENVENISTE,D.N. JANCZEWSKI,AND S.J. O'BRIEN. University Press, Oxford, United Kingdom. 1989. Molecular and biochemical evolution of the STIRLING,I., AND A.E. DEROCHER.1990. Factors affecting the Carnivora. Pages 465-494 in J.L. Gittleman, editor. evolution and behavioral ecology of the modern bears. Carnivore behavior, ecology, and evolution. Cornell InternationalConference of BearResearch and Management University Press, Ithaca,New York, USA. 8:189-204. WILKINSON,M. 1995. Coping with abundantmissing entries TALBOT,S.L., ANDG.F. SHIELDS. 1996. A phylogeny of the in phylogenetic inference using parsimony. Systematic bears (Ursidae) inferredfrom complete sequences of three Biology 44:501-514. mitochrondrial genes. Molecular Phylogenetics and WOZENCRAFT, W.C. 1989a. The phylogeny of the recent Evolution 5:567-575. Carnivora. Pages 495-535 in J.L. Gittleman, editor. VAN VALKENBURGH,B. 1991. Iterative evolution of Carnivore behavior, ecology, and evolution. Cornell hypercarnivory in canids (Mammalia: Carnivora): University Press, Ithaca,New York, USA. evolutionary interactions among sympatric predators. ___ . 1989b. Classification of the recent Carnivora. Pages Paleobiology 17:340-362. 569-593 in J.L. Gittleman, editor. Carnivore behavior, . 1999. Major patterns in the history of carnivorous ecology, and evolution. Cornell University Press, Ithaca, mammals.Annual Review of Earthand PlanetarySciences New York, USA. 27:463-493. . 1993. Order Carnivora. Pages 279-348 in D.E. Wilson VRANA,P.B., M.C. MILINKOVITCH,J.R. POWELL,AND W.C. and D.M. Reeder, editors. Mammals species of the world. WHEELER.1994. Higher level relationshipsof the arctoid SmithsonianInstitution Press, Washington,D.C., USA. Carnivora based on sequence data and "total evidence". WYSS,A.R., ANDJ.J. FLYNN. 1993. A phylogenetic analysis Molecular Phylogenetics and Evolution 3:47-58. and definitionof the Carnivora.Pages 32-52 in F.S. Szalay, WAITS,L.P., J. SULLIVAN,S.J. O'BRIEN,AND R.H. WARD. 1999. M.J. Novacek and M.C. McKenna, editors. Mammal Rapid radiationevents in the family Ursidae indicated by Phylogeny. Springer-Verlag,New York,New York,USA. likelihood phylogeneticestimation from multiplefragments ZHANG, Y.P., AND O.A. RYDER. 1993. Mitochondrial DNA of mtDNA. MolecularPhylogenetics and Evolution. 13:82- sequence evolution in the Arctoidea. Proceedings of the 92. National Academy of Sciences 90:9557-9561. WAYNE,R.K., AND K.-P. KOEPFLI. 1996. Demographic and , AND . 1994. Phylogenetic relationships of bears historical effects on genetic variationof carnivores. Pages (Ursidae) inferred from mitochondrial DNA sequences. 453-484 in J.L. Gittleman, editor. Carnivore behavior, MolecularPhylogenetics and Evolution 3:351-359. ecology, and evolution.Volume 2. Corell UniversityPress,