LOCOMOTOR ADAPTATIONS IN THE LIMB SKELETONS OF NORTH AMERICAN MUSTELIDS
by
Thor Holmes
A Thesis Presented to The Faculty of Humboldt State University
In Partial Fulfillment of the Requirements for the Degree Master of Arts
June, 1980 LOCOMOTOR ADAPTATIONS IN THE LIMB SKELETONS OF NORTH AMERICAN MUSTELIDS
Approved by the Master's Thesis Committee Chairman
Approved by the Graduate Dean
TABLE OF CONTENTS Page
Acknowledgments iv
Abstract vi
Introduction 1
Materials and Methods 11
Results and Discussion 30
Intraspecific Variation 30
Species Comparisons 46
Forelimb 46
Hindlimb 68
Conclusions 91
Literature Cited 97
Appendices 105 ACKNOWLEDGMENTS
This study was partially funded by a thesis grant from the Biology Graduate Student Association at HSU. I thank the following museums and their curators for the use of materials in their collections: American Museum of Natural History; California Academy of Science; Carnegie Museum of Natural History; Field Museum of Natural History; Humboldt State University, Museum of Vertebrate Zoology; University of Kansas Natural History Museum; Los Angeles County Museum of Natural History; Museum of Zoology, University of California, Berkeley; Michigan State University, The Museum; National Museum of Natural History; San Diego Museum of Natural History; San Diego Natural History Museum; University of California at Los Angeles; University of Michigan Zoology Museum; University of Montana Department of Zoology; University of Puget Sound, Museum of Natural History; I would like to thank particularly Murray Johnson, Helen Kafka, and Shiela Kortlucke for help during the times I used their collections.
Suzanne Edwards, Judy Fessenden, Cynthia Hofmann, Rebecca Leuck, Bob Sullivan, my brother, Thomas, and my wife, Elaine helped me make measurements. Mike Gwilliam, Tim Lawlor, Steve Smith, and Bob Sullivan helped me with computer programs. Cynthia Hofmann, Elaine Holmes, and Kathy McCutcheon helped type the manuscript. Kathy typed the final draft. V
I would like to thank the members of my committee, Jake Houck, Frank Kilmer, Tim Lawlor, John Sawyer, and Jim Waters for time they invested in my thesis and my education. I would like to thank particularly my major professor, Tim Lawlor, for help, friendship, encouragement, and sundry other contributions to my life and my education far too numerous to mention here. Finally I wish to thank Elaine, my wife, for a decade of support in every sense of the word. And I wish to acknowledge that she, more than any other person, is the reason that I can look back from this point in my life and smile. ABSTRACT
The morphology and proportions of the limb skeletons of thirteen species of North American mustelids are examined.
A series of forty-nine ratios is generated for each species.
Ratios are analyzed using standard descriptive statistics; mean, standard deviation, variance, standard error, coefficient of variation, and range. Ratios are also analyzed with a closest connection (Prim) network. Qualitative comparisons of appendicular skeletons are made with drawings of each limb element. Progressive specialization from an hypothesized primitive condition to fossorial, arboreal-cursorial, aquatic, and ambulatory modes of locomotion is revealed in limb skeletons of the Mustelidae. Relationships between morphology and proportions of mustelid limb skeletons, and modes of locomotion are discussed. INTRODUCTION
The family Mustelidae, comprising twenty-five recent and some seventy fossil genera (Anderson and Jones, 1967), is one of the most diverse families of carnivores. Only its ecological counterpart in the Old World tropics, the family Viverridae, contains more extant genera. Mustelids range in size from small (35 g) to medium (37 kg). They are distributed throughout the world except for Australasia and Antarctica. The Mustelidae apparently arose from the most ancient carnivorans, family Miacidae, about 35 mybp. The origin of the Mustelidae within the miacids seems to be separate from that of the other canoid carnivores. They quickly adopted the typical mustelid specializations of a strongly carnivor- ous dentition, short powerful jaws, prominent pre- and post- glenoid processes, and a long cylindrical body (Ewer, 1973; Romer, 1966). Mustelids as a group also are characterized by a weak zygoma. Ewer (1973) and Gambaryan (1974) suggested that the weak zygoma and a long cylindrical body are adaptations to hunting small mammals within their burrows, a method of predation that still characterizes the largest and one of the most primitive living genera, Mustela (Table 1). Despite specializations of cranial and axial osteology the mustelids retain a very generalized limb structure. They do, however, show fusion of two carpals, the scaphoid 2
Table 1. -- Earliest record for ten genera of the family Mustelidae (after Romer, 1966).
Genus Time of Probable Origin Mustela Upper Miocene Martes Middle Miocene Eira Pleistocene Gulo Pleistocene Spilogale Lower Pleistocene Mephitis Pleistocene Conepatus Upper Pliocene Taxidea Upper Pliocene Lutra Lower Pliocene Enhydra Pleistocene
PLEISTOCENE Mephitis Eira Gulo Enhydra Spilogale
Conepatus Taxidea PLIOCENE Lutra
Mustela MIOCENE Martes 3 and lunar. They also have lost a third carpal, the centrale. These are interpreted as cursorial specializations and are the common heritage of all carnivores (Ewer, 1973; Vaughn, 1978). Mustelids are plantigrade to digitigrade. The extent to which they are digitigrade is never as pronounced as in the Felidae or Canidae. The mustelids are pentadactyl and all of the digits touch the substrate. Members of the family retain relatively short, stocky limbs and do not have retractile claws. Gambaryan (1974) suggested that the basic cursorial gait in the Mustelidae, the bound or half bound, is a result of their long narrow body. Variations on the basic gait include a slow ambulatory walk (Mephitis, Conepatus), a speedy trot (Taxidea, Mustela), and a bear-like shuffle (Gulo). The Mustelidae, at least as regards appendicular anatomy, more closely resemble basal fissipeds than any extant group of carnivores. While a generalized limb skeleton characterizes most mustelids, some species have evolved ambulatory, arboreal, semi-aquatic, aquatic, and fossorial habits. Mustelids, therefore, present the student an opportunity to observe a spectrum of locomotor adapta- tions in a single family. There are numerous studies on locomotor specialists. Horses, gazelles, cheetahs, whales, and moles have all been examined. Studies of this kind are valuable in that they reveal the types of adaptations to specific modes of 4 locomotion that typify specialists. I have chosen, however, to examine the array of locomotor adaptations that charac- terize the Mustelidae. No mustelid is highly specialized for a particular mode of locomotion. For that reason I think that an investigation of the Mustelidae is of particular interest because it promises to elucidate locomotor trends in types of animals often ignored in the locomotor literature. Some recent members of the Mustelidae have been the object of considerable osteological or myological study (Fisher, 1942; Howard, 1973, 1975; Leach, 1977a, 1977b; Leach and Dagg, 1976; Ondrias, 1960, 1961; Tarasoff, 1972). As yet no one has studied the entire spectrum of locomotor adaptations in the Mustelidae either myologically or osteo- logically. Some studies on the limb skeletons of other families of carnivores have been made: Hildebrand (1952, 1954), on canids; Goynea and Ashworth (1975) and Hopwood (1947), on fends; and Taylor (1970, 1974, 1976), on viverrids. Techniques used in those studies could produc- tively be applied to the Mustelidae. The burgeoning literature on primate locomotion (Ashton et al., 1971; Ashton and Oxnard, 1964; Jenkins, 1976; Lewis, 1972; Reisenfeld, 1974) contains techniques with possible applica- tions to mustelids. A similar, if somewhat older, literature on locomotion in mammals in general (Brown and Yalden, 1973, and citations therein) provides a valuable backdrop against which to compare mustelids. Finally there are studies of 5 the effect of modes of life on the skeleton, or on particular skeletal elements (Chapman, 1919; Jones, 1953; Lehmann, 1963; Taylor, 1914; Yalden, 1972). These papers are especially useful in that they elucidated the type of morphological indicators that suggest specializations to a particular mode of life. The taxonomic position of some mustelids is not well understood (Simpson, 1945), but the family lends itself to separation into four locomotor categories corresponding roughly to the four main recognized subfamilies (Fig.1 ). The Mustelinae are weasel-like forms which typically have sub-cursorial, scampering habits. This is the oldest and most primitive subfamily. Martens, fishers, and tayras, the larger members of this group, have arboreal proclivities. The mink, another member of the Mustelinae, is amphibious. The Mephitinae, the skunks, are ambulatory. The Melinae include most of the badgers; they are fossorial. Finally, the Lutrinae comprises the most aquatic fissipeds, the otters. The American badger, Taxidea, and the wolverine, Gulo, are the subject of some taxonomic debate. Nevertheless, they may be conveniently grouped with the badger and weasel subfamilies, respectively, in regard to locomotion. North American mustelids encompass all of the locomotor types that characterize the family as a whole. Hence they should constitute a representative subset of the range of locomotor specializations that characterize mustelids as a whole. It appears that the different locomotor patterns in 6
FIGURE 1
1) Hypothesized trends in the appendicular skeletons of twelve North American species of the family Mustelidae. Species are arranged according to proposed continua of progressive specializations for arboreal, cursorial (sub-cursorial), aquatic, fossorial, and ambulatory modes of locomotion. Species (continua) radiate from a hypo- thetical primitive condition simulated among extant forms by Mustela frenata. 7 8 mustelids evolved from a more or less generalized, cursorial creature that was, nevertheless, somewhat specialized to hunt small, hole-frequenting prey in their burrows. That primitive mustelid is probably best represented among living forms by the weasels. If an artificial line is drawn from the hypothetical primitive mustelid condition, simulated among extant forms by Mustela frenata, through an amphibious form (Mustela vison) and a semi-aquatic form (Lutra canadensis) to a strongly aquatic form (Enhydra lutris), a sequence toward an aquatic tendency can be observed. It follows that inspection of the limb skeletons of the animals along this line should reveal morphological and proportional changes that are related to increasing specializations for life in water. Similarly, the limb skeletons of Mustela frenata, Martes pennanti, Martes americana, and Eira barbara should show progressively greater specialization for life in the trees, because each of those animals is progressively more arboreal. The limb skeleton of the American badger should be strongly modified for fossorial life. Mustela nigripes, a subterranean, if not strictly fossorial, form obligately tied to life in prairie dog (Cynomys) towns, may have specializations of the limb skeleton intermediate between weasels and badgers. Although not a cursor in the strictest sense, the wolverine, with a home range up to 500,000 acres (Ewer, 1973), is the widest-ranging mustelid. The limb skeleton of wolverines should show specializations for 9 the life of a cursor, albeit an ungainly one. Finally a line connecting Mustela frenata, Spilogale putorius,
Mephitis mephitis, and Conepatus mesoleucus extends from small, predaceous carnivores to medium-sized, ambulatory omnivores. This line may describe an actual relaxation of locomotor specialization since there probably is no particular premium on speed or agility in striped or hognose skunks. There is, therefore, the possibility that the limb skeletons of Mephitis mephitis and Conepatus mesoleucus may be more variable anatomically than those of
Mustela frenata or any other non-mephitine mustelid. In other words, skunks (particularly Conepatus and Mephitis) may have the most generalized, although not the most primitive, skeletons in the Mustelidae. Selection in the
Mephitinae seems to have favored extreme development of circumanal glands and aposematic coloration rather than locomotor ability.
From the literature (Brown and Yalden, 1973; Gambaryan,
1974; Hildebrand, 1974; Howell, 1944; Smith and Savage,
1956) I would expect to see the following trends in the
Mustelidae. Species with cursorial tendencies should have long legs. Distal limb elements should be disproportionately long. Limb bones should be slender with long out-levers and short in-levers. Arboreal species should share many of the specializations of sub-cursorial species. This is because arboreal mustelids simply run in trees. These species should have recurved claws and mobile ankles. 10
Fossorial mustelids should have short legs, with a disproportionate shortening or distal limb elements. Limb bones of diggers should be robust and highly sculptured with short out-levers and long in-levers. The epicondyles of the humerus should be wide, and the claws should be long and powerful. Aquatic mustelids should be typified by short legs with a disproportionate shortening of proximal limb elements. Distal limb elements may be modified to be paddle-like. The pubic symphysis may be weak. The skulls of aquatic species may be foreshortened. The potential value of the comparison of the limb skeletons of all these species is enhanced by the fact that they are all phylo- genetically close.
The objectives of my study are to:
(1)describe the morphology and proportions of elements of the limb skeletons of several species of Mustelidae, and
(2)explore the relationships between modes of locomotion and the limb skeletons of those species. MATERIALS AND METHODS
A total of 558 individuals from twenty-two species of the family Mustelidae were examined. This sample encompassed sixteen genera and all of the mustelid subfamilies (Appendix
1). Nomenclature in this study follows Hall and Kelson
(1959) for North American material. The taxonomy of Old
World material is from Simpson (1945). Both males and females were included. Individuals of unknown sex were also examined, but were treated statistically only when males and females were lumped together. Large samples
(n = 30) were obtained for eight species. Eight other species were only represented by one specimen each and were, therefore, unsuited to any statistical characterization.
However, single specimens of some species were compared as individuals against the patterns of variation developed by
the better represented species. Subspecific names and locality data were noted for use in the examination of geographic variation in the locomotor
skeleton of the species studied. Because specimens for this
study were obtained from fifteen separate museum collections
(Appendix 2), and because specimens from any particular
collection often were taken at different times and from
different places, none of my samples actually represent
natural populations. So, for reasons of pragmatism, sub-
species are treated as being broadly equivalent to
geographical races. 12
There are three North American species of weasels, one of which I chose to simulate the hypothetical primitive mustelid. I chose Mustela frenata as the representative weasel for this study because it is the largest of the weasels. All three species of North American weasels lead essentially similar lives (Ewer, 1973; Powell, 1979; Rosenzweig, 1966) and are all three likely to have similar locomotor adaptations. All three species are strict carni- vores which hunt hole-frequenting prey in its burrows. Except for size differences all three species are strikingly similar. Using the largest species serves to minimize potential allometric problems arising from comparisons of animals that differ considerably in size such as weasels, otters, and wolverines. Mustela frenata is also the widest- ranging North American weasel, and is better represented in most collections than M. rixosa and M. erminea. Similarly, the three species of Spilogale examined in this study (S. gracilis, S. putorius, and S. angustifrons) probably have similar modes of locomotion. In contrast to the situation with weasels, however, the similarity of adaptation could be tested in Spilogale. The testing procedures, delineated elsewhere, justified in my mind lumping all of the species of Spilogale for my examination of limb skeletons. The decision was not taxonomic, but functional, and had the added practical effect of improving the sample size for a critical position in. my hypothetical arrangement of locomotor trends. 13
To assess limb dimensions and proportions, I measured thirty-seven structures. Dimensions for left and right sides were measured in 296 specimens, and for only one side in 262 specimens. Measurements were chosen from Hildebrand (1952, 1954), Leach (1977a), Ondrias (1961) and Taylor (1974, 1976), and are illustrated in Figs. 2-5. Measurements were made with a vernier caliper accurate to .05mm. Those measure- ments not illustrated are noted by an asterisk (*) and are briefly explained in the list.
*a) Total length of body Taken from 1 *a ) Length of tail standard museum *b) Length of hindfoot specimen tags. c) Total length of skull d) Zygomatic breadth e) Length of scapula f) Height of scapula
*g) Maximum height of scapular spine--measured perpendicular to the blade of the scapula *h) Length of acromion process--measured from the deepest point in the notch between the acromion process and the glenoid fossa i) Length of humerus
j) Posterior displacement of humeral head, consisting of two separate measures: 14
ja--the distance between the lateral aspect of the humeral head and the most medial aspect of the lesser tuberosity. jp --the distance between the medial aspect
of the humeral head and the most lateral aspect of greater tuberosity. k) Transverse width of medial epicondyles 1) Maximum diameter of humerus at midshaft m) Minimum diameter of humerus at midshaft n) Total length of ulna o) Length of olecranon process
*p) Angle of olecranon process with shaft-- measured with a protractor; the angle formed by the olecranon process and the shaft of the ulna in anterior view
q) Length of radius r) Maximum distal width of radius s) Length of third metacarpal t) Total length of pelvis u) Length of pubic symphysis v) Preacetabular length of pelvis w) Dorsoventral breadth of ilium x) Total length of femur
y) Mediolateral width of femoral condyles z) Anteroposterior width of femoral condyles A) Maximum diameter of femur at midshaft B) Minimum diameter of femur at midshaft C) Total length of tibia 15
D) Total length of tibia minus length of
medial malleolus
E) Anteroposterior width of tibial head
F) Mediolateral width of tibial head
G) Maximum diameter of tibia at midshaft
H) Minimum diameter of tibia at midshaft
I) Length of fibula
J) Postastragalar length of calcaneus
K) Length of third metatarsal
1 *a-a ) Body length--Total length minus tail length n-o) Total length of ulna minus length of
olecranon process
*i+n-o+s) Sum of length of humerus plus length of
ulna minus length of olecranon process
plus length of third metacarpal
*x+C+K) Sum of length of femur plus length of tibia
plus length of third metatarsal
I initially intended to use total length of body as a standard against which to compare limb measurements, but of the species measured only about half (232) were accompanied by external measurements from standard museum specimen tags.
For that reason I searched for a more readily available measure that might still provide a body size standard. A number of similar studies have used a measure of the combined lengths of the thoracolumbar vertebrae (Hildebrand, 1952; 16
FIGURES 2-5
2) Bones, aspects of bones, and dimensions examined in this
study. Letters are identified in text. Figures are of
Martes americana.
3) Same
4) Same
5) Same 17
Fig. 2 18 ULNA lateral aspect anterior aspect medial aspect
RADIUS lateral aspect anterior aspect
WRIST AND HAND palmar aspect dorsal aspect
Fig. 3 19 PELVIC GIRDLE dorsal aspect ventral aspect
lateral aspect
FEMUR lateral aspect posterior aspect medial aspect
Fig. 4 20 TIBIA FIBULA lateral aspect anterior aspect lateral aspect
CALCANEUM CLAW
anterior aspect fore
medial aspect hind
ANKLE AND FOOT dorsal aspect plantar aspect
Fig. 5 21
Ondrias, 1961; Thorington, 1972) as an indicator of body
size. Because of the disarticulated state of the material
I was using, such a measure in this study would have been
prohibitively tedious. I sought to find some other useful
and convenient measure of body size. The correlation coefficient was computed between length of the body and length of the skull (n = 132). A highly significant correlation was found between body length and length of skull (r = .92; p<.001) for all mustelids. I recalculated the correlation coefficient without otters (r = .97; p<.001). Otters have relatively shorter skulls than other mustelids and therefore including otters diminishes the correlation. Because of the divergent nature of the length of the skull relative to body length in otters care must be taken in the interpretation of any ratio with length of skull (c) in it. While length of skull is a good indicator of length of body in most mustelids it over-estimates the length of the body in otters. Still the correlation between length of skull and length of body is highly significant in the Mustelidae. Therefore I felt justified in using length of skull as an indicator of length of the body.
In mustelids, and particularly the Mustelinae, males are substantially larger than females (Ewer, 1973; Haley, 1975;
King, 1975; Powell, 1979). For this reason and because wide-ranging mustelids (e.g., weasels) are geographically variable (King, 1975) .I felt that it was statistically suspect to combine raw measurements from different sexes or 22 different subspecies. Yet these individuals should have proportionately similar limb dimensions despite absolute size differences. The influence of sex and subspecific variation was therefore assessed in species containing sufficiently large samples. The use of ratios is a time-honored technique in morpho- metric studies (Desmond, 1976; Howell, 1944; Tarasoff, 1972). It is important to note that ratios are derived numbers. Some of the attendant problems in the use of ratios are treated by Atchley et al. (1976) and Simpson et al. (1960). Despite some drawbacks there are several advantages to the use of ratios (Corruccini, 1977). Of particular interest to me is the value of ratios as a means of comparing the limb skeletons of animals that differ widely in size. Stahl (1962) notes, "When measured by the length of his own fore- arm every man is the same size as every other". That is precisely the quality of ratios that I wished to exploit. A total of forty-nine ratios was generated for each side of all 558 specimens. The ratios are listed here.
1 1) (a-a )/a Length of body/Total length 1 2) b/(a-a ) Length of hindfoot/Length of body 1) 3) c/(a-a Length of skull/Length of body 4) e/c Length of scapula/Length of skull 5) e/(i+n-o+s) Length of scapula/Total length of forelimb 6) f/e Height of scapula/Length of scapula 23
7) f/c Height of scapula/Length of skull 8)i/c Length of humerus/Length of skull 9)i/e Length of humerus/Length of scapula
10) i/(i+n-o+s) Length of humerus/Total length of forelimb 11) ja/jp Posterior displacement of humeral head 12) k/i Transverse width of epicondyles/ Length of humerus 13) m/1 Minimum diameter of humerus at midshaft/Maximum diameter of humerus at midshaft 14) n/c Length of ulna/Length of skull 15) n/i Length of ulna/Length of humerus
16) n/(i+n-o+s) Length of ulna/Total length of forelimb 17) o/n Length of olecranon process/Length of ulna 18) (n-o)/c Length of ulna minus length of olecranon process/Length of skull 19) (n-o)/i Length of ulna minus length of olecranon process/Length of humerus 20)(n-o)/(i+n-o+s) Length of ulna minus length of olecranon process/Total length of forelimb 24
21)q/c Length of radius/Length of skull 22) q/i Length of radius/Length of humerus 23) r/q Distal width of radius/Length of radius 24) s/c Length of third metacarpal/Length of skull 25) s/(i+n-o+s) Length of third metacarpal/Total length of forelimb 26) t/c Length of pelvis/Length of skull 27) t/(x+C+K) Length of pelvis/Total length of hindlimb 28) u/t Length of pubic symphysis/Length of pelvis 29) v/t Preacetabular length of pelvis/ Length of pelvis 30) x/b Length of femur/Length of hindfoot 31) x/c Length of femur/Length of skull 32) x/(x+C+K) Length of femur/Total length of hindlimb 33) z/y Anteroposterior width of femoral condyles/Mediolateral width of femoral condyles 34) y/x Mediolateral width of femoral condyles/Length of femur 35) z/x Anteroposterior width of femoral condyles/Length of femur 25
36) B/A Minimum diameter of femur/Maximum diameter of femur 37) D/C Length of tibia minus length of medial malleolus/Length of tibia 38) C/b Length of tibia/Length of hindfoot 39) C/c Length of tibia/Length of skull 40) C/x Length of tibia/Length of femur 41) C/(x+C+K) Length of tibia/Length of hindlimb 42) E/F Mediolateral width of tibial head/ Anteroposterior width of tibial head 43) H/G Minimum diameter of tibia/Maximum diameter of tibia 44) I/C Length of fibula/Length of tibia 45) J/K Postastragalar length of calcaneus/ Length of third metatarsal 46) J/b Postastragalar length of calcaneus/ Length of hindfoot 47) J/(x+C+K) Postastragalar length of calcaneus/ Total length of hindlimb 48) K/(x+C+K) Length of third metatarsal/Total length of hindlimb 49) b/(x+C+K) Length of hindfoot/Total length of hindlimb 26
The following standard statistics were computed for the ratios in samples of each species: range, mean, standard deviation, variance, standard error, coefficient of variation, and sample size.
To avoid complications due to age variation I used only adult animals. Adult status was determined by closure of cranial sutures and fusion of epiphyseal caps to long bones
(Hildebrand, 1952; Leach, 1977a, 1977b; Taylor, 1970, 1974,
1976). Variation between left and right sides was examined by comparing the mean + two standard errors of ratios for left and right sides of males of the same subspecies. A total of fifteen ratios (6, 8, 9, 10, 12, 13, 14, 16, 17,
19, 22, 23, 26, 39, 40) were examined for thirteen species.
Variation between males and females was examined by comparing the mean ± two standard errors of ratios for the right side of males and females of the same subspecies. A total of twenty ratios (1, 3, 6, 8, 9, 10, 12, 13, 14, 15, 17, 18,
19, 22, 23, 26, 28, 31, 36, 39, 40, 43) was examined for thirteen species. Geographic variation was examined by comparing the mean + two standard errors of ratios for the right side of males in different subspecies. The same ratios and species were examined for geographic variation as for sex variation. The ratios examined embraced fore and hindlimbs and girdles, ratios of length and robusticity within limb, and limb/body ratios. 27
Histograms of frequency distributions for over 300
ratios were visually inspected to determine whether ratios
were distributed normally about their means.
Because sample sizes were small in a number of cases, and because there were some deviations from normalcy in the frequency distributions of some of the samples a series of
Student's t-tests were performed to check the accurace of
the mean ± two standard errors as a predictor of significant difference.
Patterns among species were examined by use of Dice
grams of the mean ± two standard errors (Fig. 6), shortest
connection (Prim) networks, and qualitative appearances from drawings. Prim networks are generated by summing the
differences between the means of any given sample of ratios
for the species to be compared. The summed differences are
used to produce a closest connection network which arrays
species based on least difference (closeness). Salient to
the interpretation of a Prim network are the relative
distance between two species along the network, and the
identity of nearest neighbors in the network. The branching
pattern (e.g. angle of branching) is not important to the
interpretation of a Prim network. Prim networks were
generated using eleven species (Mustela frenata, Mustela
vison, Martes americana, Martes pennanti, Eira barbara,
Gulo luscus, Spilogale, Mephitis mephitis, Conepatus
mesoleucus, Lutra canadensis, Enhydra lutris). Nine
combinations of ratios were used to generate Prim networks 28 of the limb skeletons of mustelids (e.g. forelimb lengths, hindlimb lengths, robusticity). Networks were generated for each combination using males only, females only, and males and females combined. Networks were then compared for similarity of pattern. Similarity of networks for males only and females only was taken as evidence that the two sexes of different species are very similar in regard to the ratios used in that particular combination. By virtue of the similar patterns for males only and females only I felt justified in conducting my species-level examinations on males and females combined. Networks based upon larger numbers of species (thirteen or fifteen) were subsequently generated to examine the degree of closeness of species of
Spilogale, or to incorporate species that were only represented by males (badgers, black-footed ferrets). Of course, utilization of species consisting only of males and species consisting of males and females in the same sample reduces resolution. Yet valuable information was provided by including badgers and black-footed ferrets in networks with other species of mustelids. Limb skeletons of the twelve best-represented species were compared by standardizing the linear dimension of each limb element according to techniques developed by Hildebrand
(1952, 1954), Hopwood (1947), Jenkins and Camazine (1977), and Taylor (1974, 1976). Drawings permitted qualitative comparisons of shape, proportion, and sculpturing of various bones that might have been missed in the numerical treatment. 29
The following caveats must be given. Some of the species in this study are represented by relatively small sample sizes. Statistical characterizations of such small samples must be viewed askance and interpreted carefully.
The natural history of some of the species in this study
(Eira barbara, Conepatus mesoleucus) is poorly known, making statements about their locomotion provisional.
Finally there is a duality to the use of ratios. There are two ways to achieve a high value for a ratio: the numerator can be large, or the denominator can be small.
It is possible to get a high value for length of femur/length of hindlimb by having a relatively long femur, or by having a relatively short leg. More to the point, it is possible to get high values for length of humerus/length of skull, length of ulna/length of skull, length of femur/length of skull, etc. in Enhydra lutris not because the individual limb elements are relatively long, but because the skulls of sea otters are relatively short. This does not abrogate the value of ratios, but rather dictates caution.
Fortunately most of the limb elements I examined were treated in a series of ratios (e.g., i/c, i/e, i/(i+n-o+s)) and are therefore less likely to be misunderstood. RESULTS AND DISCUSSION
Intraspecific Variation
Histograms of frequency distributions (Appendix 4) that were plotted for samples of ratios permit the following statements regarding the normalcy of the distribution of the ratios about their mean. The effect of lumping the ratios for males and females of a species generally leads to a better approximation of a normal distribution than either sex alone. This is no doubt partly the result of increased sample size. In small mustelines the effect of lumping ratios of males and females is usually a somewhat platykurtotic curve if the denominator of the ratio is length of skull (c). This apparently results from dispro- portionate sexual dimorphism in the size of skulls of males and females. In other words, females of small mustelines have disproportionately larger skulls than males of the same species. Nevertheless, when ratios for males and females are lumped together, they are approximately normally distributed about their mean. The effect of lumping the ratios of different subspecies of a species is much the same as lumping sexes. It is interesting to note that lumping subspecies of Mustela frenata and Martes americana resulted in somewhat more platykurtotic curves than those for other species. I suspect that my larger samples for Mustela frenata and Martes americana produce this result. My sampling of subspecies 31 is better for Mustela frenata and Martes americana than the other species and leads, therefore, to a better picture of geographic variation. At the species level the distribution of ratios about their mean provides the best approximation of a normal curve of any sample of ratios examined. There is some kurtosis and some skewness, particularly in the poorly represented species (Conepatus mesoleucus, Eira barbara, Mustela nigripes). Yet lumping both sexes and all subspecies of a species has the general effect of improving the normalcy. In any case, it is at the species level that I wish to make my comparisons. Results of Student's t-tests supported by use of the mean ± two standard errors as an indicator of differences between samples. Of 122 Student's t-tests, 113 (93%) corroborated my estimates based upon the mean ± two standard errors. Of the eleven t-tests that were unsupportive, nine nonetheless showed a significant difference at the 95% level when the mean ± two standard errors suggested that the two samples were not quite significantly different at the 95% level. If there is a preferred direction in which to be wrong, it is in predicting that two samples are not signifi- cantly different when they are. This means that if I err in my interpretation of the degree of difference between two samples I will tend to underestimate it based upon the mean ± two standard errors. 32
To test for differences between the ratios for left and right sides I used my best-represented subspecies. As Fig. 7 indicates there is no significant difference between left and right sides for Martes americana. A series of fifteen t-tests showed no significant difference between any ratio for left or right sides of any species. Dice grams of the mean + two standard errors and t-tests for differences between males and females of a subspecies produced some interesting results. No sexual dimorphism was found in any ratio for any species except when (c), length of skull, was the denominator of the ratio (Figs. 7-9). In ratios which have length of skull as the denominator a significant difference is observed for small mustelines and Spilogale (Table 2). As already noted females of the small mustelines (and perhaps Spilogale) evidently have skulls that are disproportionately large relative to the size of their bodies when contrasted to males of the same species. There are some exciting resource-partitioning implications in this observation. It certainly merits further study. In species that are sexually dimorphic (e.g. Mustela frenata, Martes americana), males of different subspecies show the same pattern relative to one another that females do (Figs. 7-9). This observation improves my confidence about combining males and females in order to estimate overall patterns of variation in ratios for a subspecies or a species. 33
Dice grams and t-tests for significant difference between subspecies of a species document the existence of geographic variation. Fig. 8 illustrates perhaps the most striking example. Martes americana abietinoides and Martes americana actuosa are decidedly different animals from other martens, at least as regards relative length of olecranon process. It is compelling to postulate that since M. a. abietinoides and M. a. actuosa are residents of stunted, boreal forest, and therefore are obliged to be somewhat more cursorial than other martens, they should show a well-known cursorial adaptation (reduction of the olecranon process) relative to their conspecifics from more heavily forested situations. A number of other ratios show similar disparities between these subspecies of martens. Geographic variation also exists in some of the ratios for Mustela frenata, Mustela vison, Martes americana, Martes pennanti, Gulo luscus, Spilogale, Mephitis mephitis, Taxidea taxus, Lutra canadensis and Enhydra lutris (Table 4). It is particularly pronounced in Mustela frenata and Martes americana, the two best-represented species in this study. I suspect that analysis of large samples of other mustelids would also reveal extensive geographic variation. So while a high score in Table 4 is a good indicator of geographic variation in the ratios for a species, a low score more likely reflects insufficient data than the absence of geographic variation.. Of course, it is also likely that some of the variation in Mustela frenata and Martes 34
FIGURES 6-9
6)Dice grams of mean, ± two standard errors, and range for ratio #17 (o/n--length of olecranon process/length of ulna) for thirteen mustelids. Sample sizes are shown in parentheses. Mf, Mustela frenata; Mn, Mustela nigripes; Mv, Mustela vison; Ma, Martes americana; Mp, Martes pennanti; Eb, Eira barbara; Gl, Gulo luscus, S, Spilogale; Mm, Mephitis mephitis; Cm, Conepatus mesoleucus; Lc, Lutra canadensis, El, Enhydra lutris.
7)Dice grams of mean, ± two standard errors, and range for ratio #14 (n/c--length of ulna/length of skull) comparing sides, sexes, and subspecies of martens. Sample sizes are shown in parentheses.
8)Dice grams of mean, ± two standard errors, and range for ratio #17 (o/n--length of olecranon process/length of ulna) comparing sexes and subspecies of martens. Sample sizes are shown in parentheses.
9)Dice grams of mean, ± two standard errors, and range for ratio #12 (i/i--transverse width across epicondyles/length of humerus) comparing sexes and subspecies of martens. Sample sizes are shown in parentheses. abi, Martes americana abietinoides; act, M. a. actuosa; cau, M. a. caurina; sie, M. a. sierrae; -abi, all subspecies except M. a. abietinoides and M. a. actuosa; All, all subspecies of Martes americana. 35
Fig. 6 36 Fig. 7 37 Fig. 8 38 Fig. 9 39 americana is a reflection of the wide geographic range of these two species.
Prim networks were used to examine possible differences in the relationship between males and between females of eleven species of mustelids. Networks for males only and females only were compared for similarity of pattern using all nine combinations of ratios (Appendix 5). Figure 10 illustrates the poorest correspondence between Prim networks for males and females in this study. There are some differences between the network for males and the network for females. The relationships between species and between subfamilies are not significantly different for males and for females however. Figure 11 illustrates a better and more typical correspondence between Prim networks for males and females. This examination indicates that males and females of different species show the same relationships to each other when ratios that reflect proportions of the limb skeleton are used to generate a Prim network. It follows from this similarity then that males and females are similar to one another in appendicular anatomy, and that species comparisons by Prim network are reasonable using samples of males and females combined.
There are significant differences between males and females in some ratios (Tables 2 and 3). There are also some differences between ratios for different subspecies of a species (Table 4). Although combining sexes and different subspecies of a species tends to mask sexual and geographic 40
FIGURES 10 & 11
10)Prim networks for females (A) and males (B) using
eleven species of mustelids and eight forelimb length
ratios.
11)Prim networks for females (A) and males (B) using eleven species of mustelids and twenty-six fore and
hindlimb ratios. 41
FIGURE 10A
FIGURE 10B 42 FIGURE 11A
FIGURE 11B Table 2. -- (+) scores for significant differences (P<.05) between ratios for males and females. Data taken from best-represented subspecies in five species of Mustelidae; Mustela frenata, Martes americana, Spilogale putorius, Mephitis mephitis, Lutra canadensis.
f/c f/e i/c i/e n/c n/i (n-o)/c (n-o)/i q/c q/i
M. f. nevadensis + - + - + - + - + - 4 3 M. a. abietinoides + - + - - - + - + -
S. 2. interrupta - - + - - - - - - -
M. m. avia - - - - - - - - - -
L. c. canadensis - - - - - - - - - - 44
Table 3. -- Scores for occurrence of sexual dimorphism in
eleven species of Mustelidae. Each ratio was scored 1 point if there was a significant difference (P<.05)
between the sexes. Scores were added for each species.
For example four forelimb and four hindlimb ratios were
significantly different for males and females of
Mustela frenata.
Species Forelimb Hindlimb Combined
Mustela frenata 4 4 8
Mustela vison 4 3 7
Martes americana 4 3 7
Martes pennanti 4 0 4
Eira barbara 1 0 1
Gulo luscus 0 0 0
Spilogale spp. 2 1 3
Mephitis mephitis 0 0 0
Conepatus mesoleucus 0 0 0
Lutra canadensis 0 0 0
Enhydra lutris 1 0 1 45
Table 4. -- Scores for occurrence of geographic variation in
twelve species of Mustelidae. Each ratio was scored 1
point if there was a significant difference (P<.05)
between any two subspecies of a species. Scores for
each species were added. For example eight forelimb
and four hindlimb ratios in Mustela frenata revealed at
least one occurrence of significant differences between
subspecies.
Species Forelimb Hindlimb Combined
Mustela frenata 8 4 12
Mustela vison 1 3 4
Martes americana 8 4 12
Martes pennanti 1 0 1
Eira barbara 0 0 0
Gulo luscus 1 0 1
Spilogale spp. 4 2 6
Mephitis mephitis 1 2 3
Conepatus mesoleucus 0 0 0
Taxidea taxus 1 2 3
Lutra canadensis 3 4 7
Enhydra lutris 0 1 1 46 variation, I made species level comparisons using combined samples of males and females and all available subspecies of a species. I did this for the following reasons: (1) normalcy of the distribution of ratios about their mean results from combining males and females, or subspecies;
(2) subspecies of males and of females exhibit the same pattern of relationship to each other in Dice grams of mean
± two standard errors; males and females combined follow the same pattern (Figs. 7-9); (3) Prim networks for males, females, and males and females were similar. Appendix 6 summarizes the descriptive statistics for all ratios of males plus females and all subspecies of thirteen species of the family Mustelidae.
Species Comparisons
FORELIMB
Clavicle
The clavicle is vestigial in all of the mustelids.
Hall (1927) reported that it is absent in Taxidea taxus.
Fisher (1942) and Howard (1973) recorded the absence of the clavicle in the Lutrinae. Savage (1957) did not record a clavicle in Potamotherium, an early lutrine. Hall (1926) did not mention the occurrence of a clavicle in Spilogale or Mephitis. Klingener (1972) noted that the clavicle of mink is "a tiny vestigial bone buried between the clavo- trapezius and clavodeltoid muscle". Martes has the most 47 highly developed clavicle of all mustelids observed. This perhaps is owing to the arboreal habits of martens and fishers. Even in Martes, however, the clavicle is quite small (<1cm.), and does not articulate with the scapula
(Leach and Dagg, 1976). Ondrias (1961) did not mention the clavicle in his study of the forelimbs of European mustelids.
Scapula
The scapula in all twelve species figured (Fig. 12) is fairly uniform anatomically. The spine is well developed in all mustelids. The acromion process is located above the glenoid fossa anteriorly in the mustelines (excepting
Mustela vison), Spilogale, and Taxidea taxus. The meta- cromion process is poorly developed in the mephitines. The metacromion process is well developed in the other species examined and is particularly prominent in Mustela frenata,
Mustela vison, the lutrinae, and Taxidea taxus.
The profile of the scapula is essentially triangular in Mustela frenata but tends to become more rectangular in the Mephitinae and Taxidea taxus. The scapula of Enhydra lutris is rectangular, but it is triangular in Mustela vison, and very triangular in Lutra canadensis. The scapulae of
Martes, Eira barbara, and Gulo luscus are essentially like that in Mustela frenata but are rounder, particularly in
Martes pennanti and Eira barbara.
The triangular or sub-round profile of the scapula is a characteristic of generalized mammals (Hildebrand, 1974; 48
Jenkins, 1974; Taylor, 1974). A narrower more rectangular
profile suggests some limitation of movement, and specializa-
tion to strengthen those limited movements. A narrow
scapula can indicate fossorial or cursorial habits (Gray,
1968; Hildebrand, 1974).
Taxidea taxus, Lutra canadensis, and the larger mustelines have a prominent post-scapular fossa which extends the site of origin of the teres major muscle. The teres major contributes to posterior movement of the humerus as in digging or swimming. The presence of the post-scapular fossa in the larger mustelines is owing perhaps to their greater weight in comparison to Mustela frenata. Ewer
(1973) equated the possession of a post-scapular fossa in the ursids to the combined effects of arboreality and great weight. The absence of a post-scapular fossa in Enhydra lutris is not surprising in light of the fact that sea otters do not use the forelimb much in swimming (Tarasoff et al., 1972). Taylor (1914) noted that the forelimb in
Enhydra lutris is an organ of prehension, not propulsion.
The scapula is long relative to the body in Conepatus mesoleucus, Mephitis mephitis, and Gulo luscus. It is shortest in Mustela and Martes. The scapula is long relative to the forelimb in Enhydra lutris, but sea otters have relatively short legs (Kenyon, 1969; Taylor, 1914). The scapula is relatively short compared to the long legs of
Martes, Eira, and Gulo. The height of the scapula relative to its length is greatest in Taxidea taxus, Lutra canadensis, 49
Enhydra lutris, and Gulo luscus, which have scapulae most
unlike that of Mustela frenata of all the species examined.
The height of the scapula is least in the mephitines, particularly Conepatus mesoleucus and Mephitis mephitis.
This character suggests some restriction of forelimb mobility, and is an unusual finding in light of the locomotor habits of skunks.
There are trends in the morphology of the scapula in the Mustelidae which serve to support my notion of progressive specialization to different modes of locomotion. The scapula becomes progressively longer relative to the body in the following species sequences: Mustela frenata--Mustela nigripes--Taxidea taxus; Mustela frenata--Mustela vison--
Lutra canadensis; Mustela frenata--Spilogale--Mephitis mephitis--Conepatus mesoleucus; Mustela frenata--Martes americana--Martes pennanti--Eira barbara--Gulo luscus. The length of scapula relative to the forelimb follows essentially the same trends, except that because of their relatively long legs, the scapulae of Martes, Eira, and Gulo are relatively shorter than that of Mustela frenata.
Both indices of scapular height (relative to body length, and relative to length of scapula) show essentially the same progressive increases in height as are described above for length of scapula. The skunks show a progressive narrowing rather than widening of the scapula relative to length of scapula. 50
Humerus
The humerus of msutelids (Figs. 13, 14) is most gracile in Martes americana, Mustela frenata, Martes pennanti, Eira and Mustela vison, and is most robust in Conepatus, Taxidea,
Gulo, Lutra, and Enhydra. Trends of increasing robusticity closely follow proposed trends of locomotor specialization:
Conepatus, Taxidea, Lutra, Enhydra, and Gulo have the most robust and elaborately sculptured humeri. I am obliged to
note that there is a size increase in all of my proposed
sequences; of course, an increase in mass will lead to an increase in the robusticity of a limb element (Hildebrand,
1974; Pedley, 1977). In this case, however, increases in
robusticity and sculpturing are probably the result of both
increased size and different modes of locomotion between
the species. For instance the humerus of Spilogale is more
robust than that of Mephitis. Since Mephitis is heavier than
Spilogale, the greater robusticity of the humerus in Spilogale
cannot be attributed to greater weight. On the other hand
increasing robusticity in Martes americana, Martes pennanti,
Eira and Gulo is probably related to increased weight along
that sequence, since locomotion in those species is very
similar.
The humerus is longest relative to the body in Gulo,
Eira, Martes pennanti, Conepatus, Mephitis, and shortest
relative to the body in Mustela frenata. Ondrias (1961)
suggested that a shortening of the humerus is typical of
fossorial mammals that must turn around in narrow burrows. 51
Both Mustela nigripes and Taxidea have longer humeri than
Mustela frenata, making me suspect the accuracy of Ondrias' generalization. The humerus is long relative to the body in the long-legged Martes americana, Martes pennanti, Eira, and Gulo; short relative to the body in otters and skunks; and short in the badger, which, of course, has powerful forelimbs. No mustelid has a humerus that is longer relative to the forelimb than Mustela frenata, although the relation- ship is about the same for Enhydra. Martes and Gulo have humeri that are short relative to length of forelimb, but this is a reflection of lengthening of distal limb elements
(see below). Long legs and disproportionate lengthening of distal limb elements is a cursorial specialization (Brown and Yalden, 1973; Gray, 1968; Hildebrand, 1974).
The shaft of the humerus is curved in Lutra, Enhydra and Taxidea. Savage (1957) noted that it is curved in
Potamotherium. Ondrias (1961) reported that it is relatively straight in Meles.
The transverse width across the epicondyles of the humerus is greatest relative to length of the humerus in
Conepatus, Taxidea, Lutra, Enhydra and Mephitis (Fig. 13).
Ondrias (1961) noted that it is also great in Meles
(European badger). The transverse width across the epicondyles is narrowest in Martes americana and Mustela frenata. There is a widening trend in the epicondyles along all of the sequences I propose. A prominent medial epicondyle is indicative of powerful flexors of the manus. 52
A prominent lateral epicondyle is indicative of powerful extensors and supinators of the manus. The relatively deeper groove between trochlea and capitulum in Taxidea, Lutra,
Enhydra, Mustela frenata, and Martes americana suggests rigidity of the elbow in these species. Ondrias (1961) suggested that widening of the epicondyles was correlated with the powerful flexion of the elbow and pronation and extension of the carpus and manus, and that it is charac- teristic of fossorial and semi-aquatic specialists. The trends illustrated here corroborate that suggestion. No mustelid has a trochlea as narrow as do the cursorial
Canidae. Among mustelids, the trochlea is narrowest in Martes americana.
The posterior placement of the humeral head is somewhat more enigmatic. The head is most anterior on the humerus in
Taxidea, Conepatus, Mephitis, and, oddly enough, in Martes americana. The head is most posterior in the lutrines.
Taylor (1974) noted that the humeral head was most posteriorly placed in cursors and less so in more generalized forms. Unfortunately there are no aquatic viverrids with which to contrast otters. Ondrias (1961) said the head of
Enhydra is placed almost as far posteriorly as in the
Phocidae (seals).
The head of the humerus is divergent, suggesting mobility of the humerus, in Martes, Eira, Gulo, Lutra, and
Enhydra. The head is more firmly buttressed and less divergent in Taxidea taxus and the skunks, suggesting that greater stresses are passed through the head and along the 53 shaft of the humerus. Taylor (1974) noted from cinero- radiographs that the head of the humerus can rock in the glenoid in viverrids, making sphericity of the head a poor predictor of mobility at the shoulder. The shaft of the humerus is roundest in Martes americana, Mustela nigripes,
Eira, Martes pennanti, Mustela frenata, and Gulo, and is most elliptical in Enhydra, Lutra, Mustela vison, Taxidea, and skunks. The pectoral and deltoid ridges in Enhydra,
Lutra, Taxidea, and the skunks, are very strong and dominate the shaft of the humerus, in contrast to Martes, Eira; and
Gulo. Hildebrand (1974) noted that round shafts of the long bones can be correlated with cursorial or arboreal habits. The greater tuberosity of the humerus is prominent in Lutra, Taxidea, and the skunks, which correlates well with power in lateral rotation of the humerus. The lesser tuberosity is prominent in Enhydra. The greater and lesser tuberosities are subequal in other mustelids.
Trends in length of humerus relative to body length and other forelimb elements generally correspond to the proposed sequences, as do trends in ellipticity of the shaft.
Ulna
The ulna is gracile in Martes americana, Mustela frenata, Martes pennanti, Eira, Spilogale and robust in
Lutra, Enhydra, Mustela vison, and Gulo (Fig. 15).
Trends of increasing robusticity of the ulna generally follow proposed continua from Mustela frenata. The ulna 54 of Enhydra is slightly less robust than that of Lutra. The ulna of Taxidea is not so robust as might be expected for a fossorial mammal, but the radius of badgers is quite robust.
The length of ulna relative to body length is greatest in Gulo which suggests some cursorial specialization. The ulna is also long in Conepatus, Taxidea, and Mephitis, but that is due partly to the disproportionately long olecranon
process in these species. The ulna is relatively short in
Mustela frenata, Mustela nigripes, Mustela vison, Lutra, and
Enhydra. The length of ulna relative to the humerus is greatest in Taxidea, Conepatus, Mephitis (all three of which have long olecranon processes), Lutra, Enhydra, and Gulo.
It is important to recall the relatively short humerus of the otters when considering this relationship, because it exaggerates the length of the ulna. The humerus and ulna are about the same length in Martes and Eira. The ulna is relatively short in Mustela frenata and Mustela vison. The length of ulna relative to the forelimb shows essentially the same relationship between species as length of the ulna relative to body length. The olecranon process of Taxidea, Conepatus, and Lutra is long, providing a considerably greater mechanical advantage, in extension of the forearm compared to other mustelids. The olecranon process is short in Martes, Gulo, and Eira. A long olecranon process is often correlated
with fossorial life or swimming movements of the forearm.
The long olecranon process of Conepatus and Mephitis may be 55 an adaptation to turning over logs and rocks in search of grubs. A short olecranon process has generally been attributed to cursorial or arboreal habits (Flower, 1885;
Gambaryan, 1974). Trends in the length of the olecranon process in the Mustelidae also closely follow my proposed continua, except that the relative length of the olecranon process is about the same in Mustela frenata and Enhydra.
Ondrias (1961) stated that the length of the olecranon process in Lutra could not be correlated with semi-aquatic life. Noting the progression in length of the olecranon process in Mustela frenata, Mustela vison, and Lutra canadensis, I would respectfully disagree.
I generated ratios using as the numerator the length of the ulna minus the length of the olecranon process reasoning that such a measure was more likely an indicator of the relative contribution of the ulna to the total length of the forelimb than length of ulna. Relative to body length, length of ulna minus length of olecranon process is longest for Gulo, Eira, Martes pennanti, and Martes americana and reflects cursorial specialization in those species. Relative to body length this measure is also long for Taxidea and the skunks, albeit not as long, relatively, as suggested by measures including the length of the olecranon process. The length of the ulna minus length of olecranon process is short in Mustela frenata, Mustela vison, and Lutra. Relative to length of humerus, length of ulna minus length of olecranon process is longest in Gulo and shortest in Mustela 56 frenata. Length of ulna minus length of olecranon process relative to both body length and length of humerus follows the sequences I suggest. Relative to total length of the forelimb length of ulna minus length of olecranon process is longest in Enhydra, Taxidea, and the skunks. This relationship must be considered in light of the short limbs of the otters. Lutra and Mustela vison have the shortest length of ulna minus length of olecranon process, relative to total limb length. This ratio generally follows my proposed sequences. There is a trend toward distal broad- ening of the ulna in Taxidea, Gulo, Mustela vison, and
Lutra.
The angle between the olecranon process and the shaft of the ulna is greatest in Mephitis, Mustela nigripes,
Lutra, and Mustela frenata; the angle is smallest in Mustela vison, Martes pennanti, Conepatus, and Taxidea. The smaller angle causing the olecranon process to be more in line with the shaft of the ulna in Taxidea, Conepatus, and Mustela vison seems as though it might improve leverage relative to a greater angle. The function of the greater angle in
Mephitis, Lutra, Mustela nigripes, and Mustela frenata is a mystery to me. Taylor (1974) suggested that a smaller, more acute, angle between the olecranon process and the shaft of the ulna would lengthen the triceps, slowing its action and permitting more controlled movements of the forelimb. He also proposed that a greater angle between the olecranon process and the shaft of the ulna would shorten 57 the triceps muscle, thus leading to faster movements of the forelimb. My data neither strongly support or negate
Taylor's suggestion.
The styloid process of the ulna is most divergent in
Mustela frenata, Mustela vison, and Martes americana. The styloid process of the ulna is least divergent in Gulo,
Taxidea and the skunks.
Radius
The radius is most gracile in Martes americana, Martes pennanti, Gulo, Mustela frenata, Spilogale, and Conepatus.
A much more robust radius typifies Lutra, Enhydra, Taxidea, and Mephitis (Fig. 16). Trends from gracile to robust generally follow my proposed sequences, although the radius in Conepatus is unusual among skunks. The shaft of the radius is curved dorsoventrally in Mustela vison, Spilogale,
Conepatus, Taxidea, and Lutra. The curve is particularly noticeable in Mustela vison, Lutra, and Taxidea. The radii of other mustelids examined were more or less straight.
Hildebrand (1974) correlates bowing of the radius with power in supination of the manus.
The length of the radius relative to the length of the body is greatest in Gulo and closely parallels the relation- ships already noted for the ulna. Trends in this ratio closely follow proposed continua with even the typically unusual Enhydra conforming. The length of the radius relative to the humerus is greatest for Gulo, which is 58 followed by Taxidea, Conepatus, Mephitis, and Spilogale; the latter also have relatively short humeri. My proposed sequences are only partially supported by trends in this ratio.
The distal width of the radius is greatest in Lutra and
Taxidea, and narrowest in Mustela frenata and Martes. The trends I postulate in my sequences are supported. The only deviation is that the distal end of the radius is wider in
Mustela frenata than in Martes; probably this condition is due to the fossorial habits of Mustela frenata and selec- tive pressure to strengthen the carpus.
Carpus and Metacarpus
The wrist and hand (Fig. 17) are particularly gracile in Mustela frenata, and Martes and are robust in Taxidea,
Lutra, Mustela vison, and the skunks. The hand is particu- larly broad in Taxidea, Lutra, and the skunks. Ondrias
(1961) noted that the hand of Meles was very broad compared to that of other European mustelids. The length of the third metacarpal relative to the body is greatest in Gulo,
Martes, and Eira. The third metacarpal is shortest in
Enhydra, Taxidea, Spilogale, Mephitis, and is somewhat longer in Mustela frenata. Trends in this ratio are somewhat equivocal. Curiously, both Mustela vison and Mustela
nigripes have longer hands than Mustela frenata, even
though the hands of Enhydra and Taxidea are much shorter.
Relative to the length of the forelimb the length of the 59 third metacarpal is greatest in Mustela vison, and Lutra, and shortest in Taxidea and Enhydra. The short hand of
Taxidea is a logical correlate of the power necessary for fossorial life. The long hands of Mustela vison and Lutra are probably due to the paddle-like function of the hand.
The relatively short hand in Enhydra is probably related to the use of the hand for functions other than swimming
(Kenyon, 1969; Tarasoff, et al., 1972). The third meta- carpal is actually relatively long in Martes, Eira, and
Gulo, but seems relatively short in this ratio due to the relatively long legs of these species. The long hands of these species probably represent adaptations to running either in trees or on the ground; and correlate nicely with the tendency toward a digitigrade stance in the arboreal, sub-cursorial mustelines.
Foreclaws
The foreclaws (Fig. 18) are longest in Taxidea and are undoubtedly correlated with fossorial life. The skunks have very long claws as well. I suspect their length is related to the opportunistic feeding habits of skunks, which turn things over in search of food. The claws in Martes, Eira, and Gulo are all strongly curved and correlate well with the known arboreal tendencies of these species (Banfield,
1974; Coues, 1877, Walker et al., 1975). The foreclaws of Mustela frenata are morphologically intermediate. In
Mustela vison, the foreclaws are rather strongly curved and 60
FIGURES 12-18
12) Lateral aspect of scapulae of twelve species of Mustelidae, arrangement like that in Figure 1.
13) Anterior aspect of humeri of twelve species of Mustelidae, arrangement like that in Figure 1.
14) Lateral aspect of humeri of twelve species of Mustelidae,
arrangement like that in Figure 1.
15) Lateral aspect of ulnae of twelve species of Mustelidae,
arrangement like that in Figure 1.
16) Anterior aspect of radii of twelve species of Mustelidae, arrangement like that in Figure 1.
17) Dorsal aspect of carpus and metacarpus of ten species of
Mustelidae, arrangement like that in Figure 1.
18) Foreclaws (right) and hindclaws (left) of twelve species
of Mustelidae, arrangement like that in Figure 1. 61
Fig. 12 62
Fig. 13 63
Fig. 14 64
Fig. 15 65
Fig. 16 66
Fig. 17 67
Fig. 18 68 may aid in scrambling on stream sides or in prey capture.
The foreclaws are small in otters, particularly in Enhydra.
Bony support embracing the base of the keratinized sheath of the claws is most noticeable in Enhydra, Taxidea,
Conepatus, and Gulo.
HINDLIMB
Pelvis
The pelvis is particularly narrow and gracile in Martes americana but is fairly narrow in Martes pennanti, Eira,
Mustela frenata, Mustela vison, Spilogale, and Lutra (Fig.
19). The pelvis is wider and more robust in Mephitis, Gulo, Taxidea, and particularly in Enhydra. Taylor (1914) correlated flaring ilia with swimming movements of the hind- limb, noting that the anterior ends of the ilia of seals are almost perpendicular to the vertebral column. The angle that the ilium forms with the pubic symphysis is closest to perpendicular in Martes, Mustela, and Eira (Fig. 20).
The angle is much more acute in the skunks, Taxidea, and
Enhydra. The angle in Lutra and Gulo is intermediate in being more acute than Martes and less acute than Taxidea.
A less acute angle between the ilium and the rest of the inominate is a specialization for speed (Smith and Savage,
1956). A rodlike ilium more or less parallel to the vertebral column is a fossorial specialization (Shimer,
1903). A prominent constriction in the ilium just anterior 69 to the acetabulum is found in Taxidea, Enhydra, Lutra, Gulo,
Martes pennanti, and Eira. The constriction is less distinct in Mustela frenata and in skunks.
The length of pelvis relative to body length is great- est in otters, Conepatus, and Mephitis. Mustela, Martes, and Spilogale have a much shorter pelvis. This trend parallels my proposed sequences closely. The length of pelvis relative to length of hindlimb exhibit the same trends as length of pelvis relative to length of body.
The pubic symphysis is shortest in the skunks and
Taxidea. The pubic symphysis is longest in the mustelines.
Symphysis length in the otters is intermediate between weasels and skunks. Hall (1926) noted that the adductor mass in Spilogale and Mephitis mephitis was not so expansive as in Martes. He did not mention the pubic symphysis at all, however. Hildebrand (1974) considered a poorly fused pubic symphysis to be associated with fossorial habits.
Taylor (1914) noted that the poorly fused pubic symphysis of sea otters was structurally antecedent to the condition of the pubic symphysis in seals. The relative length of the pubic symphysis corresponds to trends I propose for aquatic and fossorial specializations. My sequences for arboreal and ambulatory specialization are not followed closely.
The preacetabular length of the pelvis is one of the least variable ratios in my study. Preacetabular length is slightly longer in Mephitis than other mustelids and slightly shorter in Taxidea and in otters. Taylor (1976) 70 noted that preacetabular length in viverrids is relatively constant.
Femur
The femur of mustelids is most gracile in Martes, Mustela frenata, Gulo, and Eira. A robust femur is found only in Enhydra and Lutra; Taxidea and skunks are intermediate
(Figs. 21, 22). Trends of increasing robusticity follow the sequences I propose. A constricted neck between the head and shaft of the femur characterizes members of the Mustelinae and Enhydra. Skunks, Lutra, and Taxidea, have a thicker neck; the head of the femur is therefore more firmly buttressed.
The length of femur relative to hindfoot length is greatest in the skunks and Taxidea. Otters and Mustela vison have the shortest femora. The hindfeet of otters
(and to a lesser degree mink) are somewhat longer than those of other mustelids. The trends in this ratio follow my proposed sequences except for skunks, which do not differ appreciably among themselves. The length of femur relative to the body is greatest in Conepatus, Eira, Gulo,
Martes pennanti, and Mephitis. The femur is shortest in
Lutra, Mustela vison, and Mustela frenata. The long femur in arboreal species and runners (Eira Barbara, Gulo luscus, and Martes pennanti) follows well-established trends for cursorial specialization (Hildebrand, 1974; Howell, 1944;
Smith and Savage, 1956). The long femur of Conepatus and Mephitis is mystifying. Relative to body length all 71
other mustelids have longer femora than Mustela frenata.
The femur is longest relative to hindlimb length in Taxidea, suggesting that the distal limb elements have become shortened in this species. The femur is shortest relative to total length of the limb in the otters. This ratio follows my proposed sequences closely.
Ratios examining the relative width across and depth of the femoral condyles (z/y, y/x, z/x) show that otters,
Taxidea, Mustela vison, and skunks have femoral condyles
widened mediolaterally, whereas the condyles are narrow in
Mustela frenata, Martes, and Eira. The femoral condyles
are deep anteroposteriorly in otters and Mustela vison and
shallow in Martes americana, Martes pennanti, Eira, and
Spilogale. These ratios follow my proposed continua fairly
closely. The shaft of the femur is roundest in Martes,
Eira, Gulo, and Mustela frenata. The shaft of the femur is most elliptical in the otters, Mustela vison, Taxidea,
Mustela nigripes, and Spilogale. The shaft of the femur is expanded mediolaterally in Lutra and particularly in Enhydra.
Round shafts on long bones are correlated with running and arboreal habits, while more elliptical shafts are typical of aquatic and fossorial specialization (Hildebrand, 1974).
This ratio follows my proposed continua for running,
arboreal, and aquatic specializations. Skunks, Taxidea,
and Mustela nigripes do not conform to my continua.
Mustela nigripes and Spilogale both have more elliptical 72 femora than Taxidea taxus; Conepatus and Mephitis do not differ appreciably from Taxidea.
Tibia
The tibia is most gracile in Mustela frenata, Martes,
Eira, Spilogale, and Gulo. The tibia is much more robust in Taxidea, Lutra, Enhydra, and Mustela vison (Fig. 23).
The tibia is fused proximally to the fibula in Gulo and
Enhydra. This fusion likely restricts motion at the ankle to flexion and extension of the pes (Barnett and Napier,
1953ab; Walmsley, 1918). The length of the medial malleolus is greatest in Enhydra, Lutra, Mustela vison, Taxidea, and
Mustela nigripes, and is probably also an indication of reduced mobility at the ankle of those species. A shorter medial malleolus is characteristic of Martes, Eira, and
Gulo, which are decidedly more agile beasts than badgers or otters. Trapp (1972) mentioned the ability to rotate the hindfoot in Martes and other arboreal carnivores as an adaptation permitting head-first descent from trees. Taylor
(1976) recorded more potential movement in the ankle of
Nandinia, an arboreal viverrid, than in any other African species of the family.
The length of tibia relative to hindfoot length is greatest in skunks, and least in the badger and otters. The extraordinary claws of badgers explain partially the relative shortness of the badger tibia in this ratio, because the length of hindfoot measure that I used included 73 claws. Shortening of distal limb elements is a fossorial adaptation (Smith and Savage, 1956). The long, paddle-shaped feet of otters may account in part for the condition of this ratio in otters. The tibia is longest relative to the body in skunks but is also long in Martes, Eira, and Gulo.
A shorter tibia occurs in Taxidea, Mustela nigripes, and
Mustela frenata. The length of tibia relative to length of femur is one of the classical relationships of locomotor studies (Desmond, 1976) and has been widely used to illus- trate cursorial adaptation. Gregory (in Desmond, 1976) suggested that the higher the ratio of length of tibia/length of femur (T/F) the faster an animal is structurally capable of running. Desmond (1976) reported a T/F ratio of 0.92 in race horses and 1.25 in gazelles. All of the species I examined have a T/F ratio greater than 0.92 except Taxidea, and Enhydra has a T/F ratio of more than 1.25. Mustelids, and certainly otters, are not specialized cursors. Taylor
(1976) gave T/F ratios of 1.02-1.26 for viverrids. These high T/F ratios point up one of the pitfalls of making hasty comparisons. The femora of all carnivores are shorter relative to their body lengths than the femora of ungulates and render the comparison of T/F ratios between carnivores and ungulates suspect.
Trends in the ratios of tibia length follow proposed sequences except in skunks, which have surprisingly long distal limb elements. The length of the tibia relative to the length of the hindlimb is greatest in Lutra, Mephitis and Conepatus. The tibia is shortest in Taxidea. 74
The mediolateral width of the proximal end of the
tibia is greatest in otters, and correlates well with the
wide femur of these species. The anteroposterior depth of
the proximal end of the tibia is greatest in Gulo. Trends
in widening or deepening of the proximal end of the tibia
do not fit my proposed sequences well.
The tibia of mustelids is most round in Mustela vison
and Martes, and most elliptical in the otters, Taxidea, and
Mustela frenata. Ellipticity of the tibia in Mustela
frenata is greater than that in any mustelids except
Taxidea and otters. Trends in ellipticity of the tibia do not follow my sequences.
The lengthening of distal elements in the limb of
skunks is not reported in the literature to the best of my
knowledge . It seems reasonable that specializations for increasing the length of stride in cursorial species would be selectively advantageous in ambulatory species as well.
An increase in the length of stride should reduce the energy investment to move a unit of distance regardless of speed.
Fibula
The fibula of mustelids is gracile in Martes, Mustela,
Eira, Gulo, and the skunks. The fibula is more robust in
Taxidea, Lutra, and Enhydra (Fig. 24). Trends in robusticity of the fibula follow my proposed sequences in the mustelines, and in Lutra and Enhydra. The fibula of the three 75 mephitises I examined are almost identical. The fibula is longest relative to the tibia in Martes, Eira, Gulo and
Spilogale. Mustela nigripes, Mustela vison, Taxidea, Lutra, and Enhydra have short fibulas.
Calcaneus
The calcaneus is gracile in Martes, Gulo, Mustela vison and Conepatus. A more robust calcaneus typifies Mustela frenata, Taxidea, and Enhydra (Fig. 25). Postastragalar length of the calcaneus is greatest in Conepatus, Taxidea, and Mephitis. The postastragular length is much less in Mustela frenata, Martes americana, and Enhydra. My proposed sequences conform very well with trends in the postastragular length of the calcaneus except, as usual, Enhydra is very unlike other aquatic mustelids. The postastragalar length relative to length of the third metatarsal and relative to length of hind foot are essentially similar ratios. The postastragalar length relative to length of hindlimb is greatest in Taxidea, Lutra, Gulo, and Conepatus. The postastragalar length is least in Mustela frenata and
Martes americana. This ratio parallels my proposed sequences except in the case of Enhydra in which the postastragalar length relative to hindlimb length is essentially the same as in Mustela vison. Stains (1976) noted that the calcanea of mustelids were at least subfamilially distinct, but drew no functional conclusions from his work. Probably an increase in postastragalar length increases the power of 76 extension of the pes in otters, badgers, and relatively heavy-bodied sub-cursorial forms such as Gulo.
Pes
The hindfoot is most gracile in Mustela, Martes, and
Spilogale. The hindfoot of Taxidea, Lutra, Conepatus, and
Enhydra is much more robust (Fig. 26). Trends in robust- ness of the hindfoot follow my proposed sequences closely.
The length of the third metatarsal relative to the total length of the hindlimb is greater by far in Enhydra, which has large paddle-like feet and short legs. The third meta- tarsal is much shorter in Eira, Spilogale, Mephitis,
Conepatus, and Taxidea. Trends in skunks, arboreal mustelines, Gulo, and fossorial mustelids correspond to my proposed sequences.
Hindfoot length relative to length of hindlimb is greatest in Enhydra and least in Spilogale and Mephitis.
Sample sizes for this ratio were the smallest in this study.
The length of the hindfoot relative to length of body is greatest in Enhydra and Gulo. The hindfoot is shortest relative to length of body in Mustela vison and Spilogale.
The long hindfoot of Gulo could be an adaptation to sub-cursorial locomotion in this wide-ranging mustelid.
However, the size of hindfoot may be a reflection of the large size and massiveness of Gulo compared to other smaller mustelids. The long hindfoot of Enhydra may be a function of the size of sea otters as well. Sea otters, 77
however, are not particularly adept at overland travel
(Kenyon, 1969; Tarasoff et al., 1972) and the long hindfoot
is most likely an adaptation to aquatic locomotion. The
relatively small hindfoot of Lutra correlates well with the
fact that more of the propulsive forces in river otters are
provided by the forelimb and tail than in sea otters
(Tarasoff et al., 1972). Indeed, terrestrial locomotion
is much more common in river otters than in sea otters. The
ratio of hindfoot length to body length in Lutra is more
similar to terrestrial mustelids than to Enhydra. The
smallest hindfeet in the Mustelidae belong to Spilogale,
Mustela frenata, Mustela vison, and Mustela nigripes, which implies along the same line suggested for Gulo luscus that
little animals have relatively small feet.
Hindclaws
The hindclaws in the Mustelidae follow essentially the
same patterns described for the foreclaws (Fig. 18). In
skunks and badgers the hindclaws are much shorter than the foreclaws. The structure of foreclaws and hindclaws is
very similar.
Tail
Ratio number 1, body length/total length documents, a
short tail in Taxidea, Gulo, and Enhydra. Long tails
characterize Lutra, Conepatus, and Mephitis. Short tails
are typical of fossorial mammals (Hildebrand, 1974; Shimer,
1903; Vaughan, 1978). Tarasoff et al. (1972) noted that 78
FIGURES 19-26
19) Dorsal aspect of pelvic girdles of twelve species of
Mustelidae, arrangement like that in Figure 1.
20) Lateral aspect of pelvic girdles of twelve species of
Mustelidae, arrangement like that in Figure 1.
21) Medial aspect of femora of twelve species of Mustelidae,
arrangement like that in Figure 1.
22) Posterior aspect of femora of twelve species of
Mustelidae, arrangement like that in Figure 1.
23) Anterior aspect of tibiae of twelve species of
Mustelidae, arrangement like that in Figure 1.
24) Lateral aspect of fibulae of twelve species of
Mustelidae, arrangement like that in Figure 1.
25) Anterior (left) and medial (right) aspects of calcanea
of twelve species of Mustelidae, arrangement like that in Figure 1.
26) Dorsal aspect of tarsus and pes of eleven species of
Mustelidae, arrangement like that in Figure 1. 79
Fig. 19 80
Fig. 20 81
Fig. 21 82
Fig. 22 83
Fig. 23 84
Fig. 24 85
Fig. 25 86
Fig. 26 87 the tail was important in aquatic locomotion to both Lutra and Enhydra. That information makes the short tail of
Enhydra somewhat enigmatic. The long bushy tail of skunks is well known and serves more as a warning device than a locomotor aid. The long tail of Martes pennanti is not surprising in an arboreal runner and leaper (Hildebrand,
1974; Taylor, 1970). What is peculiar is the decidedly shorter tail of the equally, if not more, arboreal Martes americana (Powell, pers. comm.).
Figure 27 summarizes the results of Prim networks for various aspects of the limb skeletons of thirteen species of Mustelidae.
Prim networks show a strong similarity of form in the limb skeletons of otters, badgers, and skunks. These three subfamilies of the Mustelidae are particularly close with regard to the forelimb skeleton and robusticity ratios. The hindlimbs of otters are more reminiscent of members of the genus Mustela than of skunks. Otters are always relatively distant from whatever species they join in the networks.
Considering both limbs and robusticity, it is clear that otters are closest to skunks and badgers. This similarity in form and function between swimmers and diggers is not a new observation (Gray, 1944, 1968; Hildebrand, 1974; Smith and Savage, 1956). One interesting observation about otters from the Prim networks is that in regard to forelimbs, hindlimbs, and robusticity, Lutra canadensis, is more distantly connected to other mustelids than is Enhydra 88 lutris. The implication is that Enhydra lutris, at least regarding appendicular osteology, is more generalized than Lutra canadensis. 89
FIGURE 27
Prim networks for males and females combined using thirteen species and (A) twenty-six fore and hindlimb ratios;
(B) fifteen forelimb ratios; (C) nineteen hindlimb ratios;
(D) eleven robusticity ratios. 90 CONCLUSIONS
Taken as a whole my proposed sequences are borne out by the data. However, there are some interesting divergences between my predictions and my observations. Enhydra lutris is not on the same continuum of locomotor specialization as
Mustela vison and Lutra canadensis. Enhydra is more aquatic, and presumably more specialized for aquatic life, than
Lutra. Enhydra emphasizes different bone-muscle systems in aquatic locomotion than Lutra or Mustela vison.
There are two problems with my proposed continua for arboreal specialization. First, the species are not in the appropriate order. Martes americana is, if anything, a bit more specialized for arboreal life than Martes pennanti or Eira. Second, arboreal and cursorial speciali- zations are very similar in the Mustelinae. This statement is demonstrated by the tendency of Martes, Eira, and Gulo to group closely together in the majority of my ratios.
Gulo does have more pronounced cursorial adaptations (e.g. increase in length, disproportionate increase in length of distal limb elements) than other mustelids. Martes americana, however, is nearly as specialized for running as Gulo, even though Martes americana is a more arboreal mustelid; Eira and Martes pennanti are intermediate between Mustela frenata and Martes americana.
My data for skunks present some surprises. Skunks have combined cursorial specializations (e.g. lengthening 92 of distal limb elements) and fossorial specializations (e.g. robust skeletons, long claws, powerful forelimbs). This combination of specializations suggests a lack of channeliza- tion to any particular mode of locomotion in skunks. In oter words, skunks are more generalized in locomotor tendencies with specializations characteristic of several modes of locomotion.
The data for fossorial and aquatic locomotion (ignoring for a moment Enhydra) provide strong support for my proposed sequences. Mustela vison is intermediate between
Mustela frenata and Lutra. Mustela nigripes is inter- mediate between Mustela frenata and Taxidea.
The apparent similarity of form and function between the skunks and badgers, and between the larger mustelines, is probably at least partly a reflection of phylogenetic closeness between the Mephitinae and the Melinae, and between Martes, Gulo, and Eira (Anderson, 1970; Simpson,
1945; Winge, 1941). Still, there is little doubt in my mind that the Prim networks, the drawings, and the statis- tical estimates of similarity reflect actual similarities in mode of locomotion. My confidence is strengthened by the following comparisons between New and Old World species.
Meles meles, the European badger, has generally similar ratios to Taxidea taxus. The single specimen of Meles has a longer, narrower scapula, a shorter humerus with wider epicondyles, and a shorter ulna than Taxidea taxus. Meles meles also have a longer pelvis, pubic symphysis, and ilium 93 than Taxidea taxus. Meles also has a slightly longer femur, tibia, and hindfoot. Meles appears to be somewhat more fossorial than Taxidea taxus.
Ratios for Mellivora sagulata (the ratel or honey badger) are more similar to Taxidea taxus than Meles moles.
Ratios for Helectis moschata, the ferret badger, are inter- mediate between those for Taxidea taxus and Mustela nigripes.
Finally, the ratios for Arctonyx collaris, the hog badger, are very close to those for both Taxidea taxus and Mellivora sagulata.
The ratios for Amblonyx cinerea (the Oriental small clawed otter) are more like Lutra canadensis than any other
North American mustelid. Amblonyx differs from Lutra in having a shorter, wider scapula, a shorter humerus with wider epicohdyles, a shorter ulna with a longer olecranon process, and a radius that is wider distally. The pelvis of Amblonyx is shorter than that of Lutra as is the tibia, whereas the femur is a bit longer. The ratios of Pteronura brasiliensis, (the giant otter of South America) are more like Amblonyx than Lutra.
The ratios for Mustela putorius, the European polecat, are similar to those for Mustela frenata, Mustela vison, and
Mustela nigripes which makes both functional and taxonomic sense.
The implication to me, from my North American data and the relatively scanty data on other mustelids, is that a 94 larger sample of mustelids worldwide would strengthen the generality of my proposed sequences.
There clearly are different locomotor patterns among species of Mustelidae. These locomotor patterns are followed in type and degree by patterns of specialization in the limb skeletons of the Mustelidae. Patterns of specialization in the family are revealed by sampling North American forms. The more cursorial forms show an increase in the length of the limb bones, particularly the distal limb bones, relative to more generalized species. Species with cursorial (and arboreal) tendencies have longer bones with rounder shafts compared to fossorial or aquatic species.
In mink and otters, and in black-footed ferrets and badgers, there are progressive decreases in the length of long bones.
Mink and otters on one line, and black-footed ferrets and badgers on another, show progressive increases in the length of in-levers (olecranon process, calcaneus) and progressive decreases in the length of out-levers (forearm, hindfoot) compared to weasels. Certainly the specializations for running in mustelids are not so pronounced as in ungulates or even canids. Neither are fossorial adaptations as pronounced as in moles, nor aquatic specializations as spectacular as in pinnipeds. Trends are evident, however, and to a considerable extent they appear to progress from the hypothetical generalized condition observed in weasels.
Skunks appear to be quite generalized. Hall (1926) noted that the muscular anatomy of Mephitis mephitis was 95
primitive (generalized?) compared to that of Martes which
he felt showed specializations for speed and agility.
Hall's view contradicts subsequent positions taken by Leach
(1976, 1977a, 1977b) and Ondrias (1961), who felt that the
anatomy of Martes was very generalized. When contrasted
to ungulates or dogs, the appendicular anatomy of martens and fishers is generalized. In comparison to Mephitis, and
I presume other skunks as well, the opposite seems to be
true. The more or less central position of skunks in the
Prim networks suggests a generality in their limb skeletons,
although martens and fishers are often centrally located
in the networks as well.
Coefficients of variation were examined for all species
having samples of n>9 for all ratios (Appendix 6). The
most variable species for each of the 49 ratios was noted.
The number of instances in which each species exhibited the
highest coefficient of variation is as follows: Mustela frenata, 4; Mustela vison, 5; Mustela nigripes, 0; Martes americana, 1; Martes pennanti, 0; Eira Barbara, 1; Gulo luscus, 0; Spilogale, 6; Mephitis mephitis, 16; Conepatus mesoleucus, 0; Taxidea taxus, 6; Lutra canadensis, 2;
Enhydra lutris, 8. A relative indication of variability at
the subfamilial level can be computed by dividing the number of occurrences of the highest coefficient of variation in a
subfamily by the number of species examined in that subfamily.
Values for that computation are: Mustelinae, 1.57: Mephitinae,
7.33; Melinae, 6.0; Lutrinae, 5.0. This index suggests that 96 the Mephitinae are the most variable subfamily of mustelids while the Mustelinae are the least. The variability in the
Mephitinae is actually underestimated in this calculation because the sample of Conepatus mesoleucus is small
(n = 6); hence Conepatus was not scored although it has the highest coefficient of variation in nine of the ratios. The high coefficient of variation characteristic of skunks suggests that they are generalized. I would argue that the generalized limb skeletons of skunks and their variability suggest that there are few selective pressures to be swift or agile in skunks (particularly M. mephitis and C. mesoleucus) compared to other mustelids. LITERATURE CITED
Anderson, E. 1970. Quaternary evolution of the genus Martes (Carnivora, Mustlidae). Acta Zoologica Fennica, Helsinki-Nelssingfors, 132pp. Anderson, S. and J. K. Jones. 1967. Recent mammals of the world; a synopsis of families. Ronald Press, New York, pp335-338. Ashton, E. H. and C. E. Oxnard. 1964. Functional adapta- tions of the primate shoulder girdle. Proc. Zool. Soc. London, 142:49-66. Ashton, E. H., R. M. Flinn, C. E. Oxnard, and T. F. Spence. 1971. The functional and classificatory significance of combined metrical features of the primate shoulder girdle. J. Zool., London, 163:319-350. Atchley, W. R., C. T. Gaskins, and D. Anderson. 1976. Statistical properties of ratios. I Empirical results. Syst. Zool., 25:137-148. Banfield, A. W. F. 1974. The mammals of Canada. University of Toronto Press, Toronto, pp315-346. Barnet, C. H. and J. R. Napier. 1953a. The rotary mobility of the fibula in eutherian mammals. J. Anat., 87:11-21. . 1953b. The form and mobility of the fibula in metatherian mammals. J. Anat., 87:207-213. Brown, C. J. and D. W. Yalden. 1973. The description of mammals. II Limbs and locomotion of terrestrial mammals. Mamm. Rev. , 3(4):107-134. 98
Brown, J. H. and R. C. Lasiewski. 1972. Metabolism of weasels: the cost of being long and thin. Ecology, 53(5):939-943. Chapman, R. N. 1919. A study of the correlation of the pelvic structure and the habits of certain burrowing mammals. Amer. J. Anat., 25:185-219. Corruccini, R. S. 1977. Correlation properties of morpho- metric ratios. Syst. Zool., 26:211-217. Coues, E. 1877. Fur-bearing animals: a monograph of North American Mustelidae. Government Printing Office, Washington, 348pp. Desmond, A. J. 1976. The hot-blooded dinosaurs; a revolution in paleontology. Dial Press, New York, 238pp. Ewer, R. F. 1973. The Carnivores. Cornell University Press, New York, 494pp. Fisher, E. M. 1942. The osteology and myology of the California river otter. Stanford University Press, California. Flower, W. H. 1885. An introduction to the osteology of the Mammalia. MacMillan and Co., London, 373pp. Gambaryan, P. P. 1974. How mammals run. John Wiley and Sons, New York, 367pp. Goynea, W. and R. Ashworth. 1975. The form and function of retractile claws in the Felidae and other representa- tive carnivorans. J. Morph., 145:229-238. Gray, J. 1944. Studies in the mechanics of the tetrapod skeleton. J. Exper. Biol., 20:88-116. 99
1968. Animal locomotion. Weidenfeld and Nicholson, London, 479pp. Haley, D. 1975. Sleek and savage: North America's weasel family. Pacific Search Books, Seattle, 128pp. Hall, E. R. 1926. The muscular anatomy of three mustelid mammals, Mephitis, Spilogale, Martes. Univ. California Publ. Zool., 30:7-38. . 1927. The muscular anatomy of the American badger (Taxidea taxus). Univ. California Publ. Zool., 30:205-219.