Journal of Vertebrate Paleontology 12:162-182, June 1987 SKELETAL INDICATORS OF LOCOMOTOR BEHAVIOR IN LIVING AND EXTINCT CARNIVORES BLAIRE VAN VALKENBURGH Department of Earth and Planetary Sciences, The Johns Hopkins University, Baltimore, Maryland 21218 (Present address: Department of Biology, University of California, Los Angeles, California 90024) ABSTRACT ~In this paper, [assess the correspondence between locomotor function (climbing, digging, and running) and skeletal morphology in living carnivores. Published behavioral observations are used to assign fifty-three extant carnivore species to four broad locomotor categories: arboreal, scansorial, terrestrial, and semi-fossorial. Body weight and measurements of claw shape, relative size of fore and hind claws, olecranon size and orientation, astragalar shape, and manus and hindlimb proportions are presented for each species. The correspondence between locomotor type and morphologic indices is ‘examined with bivariate and multivariate (discriminant function) analyses. Results demonstrate that ‘osteological indices are good predictors of locomotor behavior among extant carnivores. Arboreal and scansorial carnivores exhibit relatively more curved claws, longer proximal phalanges, and shorter ‘metatarsals than do terrestrial taxa; semi-fossorial carnivores differ from other locomotor types in having relatively long olecranon processes, and longer fore claws than hind claws. Bivariate plots suggest that some characters, such as astragalar shape, reflect ancestry more than function. Applicability Of the indices to other species is shown with a sample of eight non-carnivores and six Oligocene predators, including three sabertooth cats, a bear-dog, and two creodonts. INTRODUCTION There is much to be gained from the study of fossil faunas that is not apparent from studies of single species. For example, by analyzing suites of fossil species we can explore community-wide patterns of species di- versity, morphological convergence, and evolution over time (cf. Olson, 1952, 1971). Moreover, by analogy with modern animal and plant associations, fossil fau- nas suggest aspects of the paleoflora and can be used in conjunction with paleobotanical evidence to recon- struct paleocommunities (cf. Andrews et al., 1979). Studies of entire or partial paleofaunas are rare, how- ever, because of the time and effort required to study the paleobiology of many, as opposed to one, species. Asa result, some workers have interpreted the behav- ior of extinct species as being similar to that of either their nearest living relative or their nearest well-studied extinct relative, with minimal analysis of morphology (cf. Andrews et al., 1979; Van Couvering, 1980; Col- linson and Hooker, 1985). While this may be an ac- ceptable procedure for relatively young paleocom- munities, it becomes increasingly problematic as we progress backwards in time. Clearly, it would be better to infer behavior from morphology whenever possible, using a set of skeletal characters that are predictive of critical aspects of the ecology of an organism, such as diet, foraging, and predator-avoidance behavior. This paper presents such, a set of skeletal indices designed for a study of past and present guilds of mammalian carnivores (Van Val- 162 kenburgh, 1984, 1985). The skeletal characters reflect, locomotor function within four categories of loco- motor behavior (arboreal, scansorial, terrestrial, and semi-fossorial). Although the skeletal indices were de- veloped for camivores, they are probably useful for other mammals, especially those with relatively un- specialized forms of locomotion. As a demonstration of their generality, they are here applied to a small number of rodents, marsupials, insectivores, and creo- donts, as well as to fifty-seven carnivores (53 extant, 4 extinct) MATERIALS AND METHODS A total of sixty-one species of Recent mammals were ‘measured: 53 carnivores, 3 rodents, 3 marsupials, and 2insectivores (Table 1). All measurements were taken ‘on specimens of adult, wild-caught animals if avail- able; z00-raised adults were substituted when neces- sary. The number of individuals measured per species varied from one to twelve with an average of two (Ap- pendix 1). Larger samples include members of both sexes; significant sexual dimorphism was not apparent in the morphometric indices. All skeletons are housed at the USS. National Museum of Natural History (USNM). The source and identity of the fossil speci- mens are listed in Appendix 3. Locomotor Behavior Categories— As shown in Table 1, each of the Recent species is classified into one of four locomotor categories on the basis of behavioral information, without reference to the morphologic data. © 1987 ye Sokty of Verne PaontlonsThe four categories are: (1) arboreal; (2) scansorial; (3) terrestrial; and (4) semi-fossorial. As defined in Table 2, the categories are broad and emphasize foraging and escape behavior. The locomotor mode employed in escaping danger may differ from, and be much less frequently employed than the foraging mode, but both are equally critical to survival (see Van Valkenburgh, 1985). Most species were easily classified because their escape and foraging modes were similar; when the two modes differed (e.g. black bear, lion, tiger), species were placed according to escape, or potential escape mode, The locomotor categories are broadly defined so as to be generally applicable to mammals of diverse func- tion and ancestry. More narrow locomotor categories, such as long-distance runner, are usually useful for only a few species that are often closely related. Hence, mor- phologic similarities within the group might reflect common ancestry rather than function and thus be poor predictors of locomotor function in distantly re- lated species. By defining broad categories that include representatives of several carnivore families, the pre- dictive power of the morphologic indices is greatly enhanced. Although each of my locomotor groups is dominated by a different carnivore family, all but one (semi-fossorial) includes members of several families (Table 3). Morphometric Indices Body weight and ten characters of the post-cranial skeleton were chosen as indices of locomotor function because of a previously described association with lo- comotor behaviors, or an association predicted from biomechanical principles. In addition to functional ‘considerations, characters were selected for ease of ‘quantification and probability of preservation in fossil material. The latter was determined from the tapho- nomic literature (Voorhies, 1969; Behrensmeyer, 1975; Shipman, 1981). Thus, despite the proven utility of scapula shape for functional analyses of carnivores (cf. Hopwood, 1945; Davis, 1949; Maynard Smith and Savage, 1956; Oxnard, 1968; Taylor, 1974), the scap- ula was excluded from this study because it is rarely preserved without damage (Hill, 1980; Brain, 1981: 18). The values of the morphometric indices are listed by species in Appendix 1 Body Size (LBW) —Body size was calculated as the log of body weight. Body weights for Recent taxa were taken from the literature as listed in Table 1. For fossil taxa, weights were estimated from head-body length and skull length (see Appendix 2). ‘Ungual Phalanx Shape (UD and ARCH)—As shown in Figure 1, ungual phalanx shape was estimated with two indices, depth (UD, st/rs) and curvature of the dorsal arc (ARCH, xz/wy), both on terminal phalanges of forefeet only. Whenever possible, the phalanges of the third and fourth digits were measured, and aver- aged for each species. If the ungual phalanges were not identifiable to digit, then 8-10 ungual phalanges from both forefeet were measured and averaged. Measure- ments were made from color transparencies. Hind/Fore Ungual Phalanx Length (HF)—Length of the ungual phalanx of the third digit on the hindfoot is divided by that of the homologous ungual phalanx of the forefoot. Maximum length was measured with dial calipers as shown in Figure 1B (wy). As above, if ungual phalanges were not identifiable to digit, then mean values were employed. Manus Proportions (MCP)— An estimate of manus proportions is given by the ratio of the length of the third metacarpal to that ofthe proximal phalanx of the third digit (Fig. 1C and D). Both lengths were measured with dial calipers. Olecranon Orientation (OLA)—The anterior-pos- terior orientation of the olecranon process with respect to the ulna shaft was measured on color transparencies of lateral views of the ulna as shown in Figure 2 (A and B). ‘Otecranon Length (OLL)—Otecranon length is de- fined as length of the olecranon process divided by length of the ulnar shaft. As shown in Figure 2, these lengths were measured from an estimated center of rotation of the humeral-ulnar joint to the distal and proximal ends of the ulna, respectively. Maynard Smith and Savage (1956) compared olecranon length to that of the entire forelimb below the elbow, whereas this paper follows Hildebrand (1974:529) and compares it to ulna shaft length (Fig. 4C and D). Although the former might be a better estimate of triceps in-force relative to resistance at the manus, both of the above Papers reached similar conclusions and Hildebrand’s method is easier with incomplete fossil material, OLL ‘was measured from color transparencies. Hindlimb Proportions (FMT)—Maximum length of femur is divided by maximum length of fourth meta- tarsal. Femur length was measured with a tape mea- sure; metatarsal length was measured with dial cali- pers. Depth of Astragalar Trochlea (ASD)—The trochlea, ‘or proximal articular surface of the astragalus, artic ulates with the tibia. The trochlea is somewhat pulley shaped and the depth of grooving of the pulley was estimated as groove depth (bd) divided by trochlea width (ac), as measured from color transparencies (Fig. 3) STATISTICAL ANALYSES ‘The utility of the morphometric indices for the pre- diction of locomotor behavior was tested with bivar- iate plots and the multivariate procedure of discrim- inant analysis (Neff and Marcus, 1980; Reyment et al., 1984). The bivariate analyses are particularly useful for the interpretation of behavior in extinct species because fossil material is often incomplete and there- fore unsuitable for discriminant analysis. Because a primary objective of this work was to produce a data base for future morphological com- parisons between extinct and extant species, I have presented a bivariate plot for most of the indices. In addition, I have calculated the variance of each of these 163TABLE |. List of species examined, their common names, weights, locomotor classification, references for weight (Ws. refer.) and for locomotor behavior (Loco. refer.). ID No. is the number used for each species in the bivariate plots. The locomotor category abbreviations are: arb (arboreal); scan (scansorial), ter (terrestrial); sf (semi-fossorial), and °? (uncertain). If two locomotor categories are listed for a species, the one in parentheses is that predicted by morphology in the discriminant analysis. ‘The other is that predicted by behavioral data, If only one category is listed, the two approaches gave the same result. The key to references follows the table. Body 1D weight Locomotor we Loco. No, Species Common name es) eategory sefer refer Marsupalia 1 Pseudocheirus sp. Cuscus 1 arb 1 18 2 Dasyurops maculatus Ringtail possum 25 sean 1 129 3 Sarcophilus harrsii Tasmanian devil 68 ter 1 ' Insectivora 4 Solenodon paradoxus Solenodon 0.64 2s) 28 29 Rodentia 5 Sciurus carolinensis Carolina gray squirrel 0.s4 arb 1 120 6 — -Marmota monax Ground hog 34 sf 21 un 7 Erethizion dorsatum Porcupine 53 arb) 1 H21 Scandentia 8 Tupaia gis ‘Tree shrew 0.03 scan 1 Ls Carnivora 9 Urocyon cinereoargenteus Gray fox 45 scan(tem) 1,6 121 10 Vulpes vulpes Red fox 9 ter 1 n21 HL Cuon alpinus Dhole ter 1 127 12 Dusicyon griseus Chico gray fox 65 ter 23,47 46 13 Speothas venaticus Bush dog. 6 ter t 144 14 Nyctereutes procvonoides Raccoon dog. 7 ter I nas 1S Lyeaon pictus Wild dog 25 ter 3 3.34 16 Canis lupus Gray wolf 430 ter 1,25 1,25 17 Canis latrans Coyote 1S ter ne n21 18 Canis aureus Golden jackal 7 tere 2 3.35 19 Canis mesomelas Blackbacked jackal 7 ter 2 3,35 20 Tremarctos ornatus Spectacled bear 140 scan 1 1,37 21 Melursus ursinus Sloth bear 102 scan(tert) us 22 Helarctos malayanus Sun bear 49 scan 131 15 23 Selenarctos thibetanus Asiatic black bear 155 scan us 5.39 24 Ursus americanus ‘American black bear 150 scan L6 121 25 Ursus arctos Grizaly bear 24 tere 6 121 26 Eira barbera Tayra 4.2 scan(ars) 14 1,10 27 Gulo gulo Wolverine 24 sean H612 0 13 28 Martes pennanti Fisher 34 scan New 17 29° Galictis vinarus Grison 32 tert) ua 1,10 30 Mephitis mephitis Striped skunk 3 ost ne 20,2 31 Tetonyx striatus Zorilla Bo sf 3 3 32 Taxidea taxus North American badger BS sf 16 121 33 Meles meles: European badger Bo sf 1 1,40 34 Arctonyx collaris Hog badger 0s sf 1 1 35 “Mellivora capensis Ratel 100s) 3 3,43, 36 Potos flavus Kinkajou 2.5 arb 14 1,20 37 Procyon lotor Raccoon un scan(arb) 1,612 1,21 38 Nasua narica Coatimundi 5 sean 6,20 nis 39 Bassariscus asturus Ringtailed cat 1 sean(arb) 1,6 15,2 40 Felis wiedii Margay 27 arb 6 27 41 Neofelis nebulosa Clouded leopard 200 arb 12310 19 42 Lynx canadensis ‘Canadian lynx nn sean (arb) 1.6 221 43 Puma concolor Puma 60 sean 12.6 2.21 44 Caracal caracal Caracal 65 scan 2 3,42 45° Felis pardalis Ocelot 13 sean 2,6 20 46 Felis temmincki ‘Temminck’s cat 3S sean 31 5.9 V8 70) Joe 1987