Behavioral Ecology Vol. 14 No. 6: 823–840 DOI: 10.1093/beheco/arg072 Ecological and behavioral correlates of coloration in artiodactyls: systematic analyses of conventional hypotheses C. J. Stoner, T. M. Caro, and C. M. Graham Department of Wildlife, Fish, and Conservation Biology, University of California, Davis, CA 95616, USA To test the generality of adaptive explanations for coat coloration in even-toed ungulates, we examined the literature for hypotheses that have been proposed for color patterns exhibited by this taxon, and we derived a series of predictions from each hypothesis. Next, we collected information on the color, behavioral, and ecological characteristics of 200 species of even-toed ungulates and coded this in binary format. We then applied chi-square or Fisher’s Exact probability tests that pitted presence of a color trait against presence of an ecological or behavioral variable for cervids, bovids, and all artiodactyls. Finally, we reanalyzed the data by using concentrated-changes tests and a composite molecular and taxonomic phylogeny. Hinging our findings on whether associations persisted after controlling for shared ancestry, we found strong support for hypotheses suggesting even-toed ungulates turn lighter in winter to aid in concealment or perhaps thermoregulation, striped coats in adults and spotted pelage in young act as camouflage, side bands and dark faces assist in communication, and dark pelage coloration is most common in species living in the tropics (Gloger’s rule). Whereas white faces, dark legs, white legs, dark tails, and white tails did not appear to assist in communication alone, legs and tails that were either dark or white (i.e., conspicuous) did seem to be linked with communication. There was moderate support for hypotheses that countershading aids concealment, that white faces are a thermoregulatory device, and that white rumps are used in intraspecific communication. There was weak support for spots in adults and stripes in young providing camouflage and for dark leg markings being a form of disruptive coloration. We found little or no evidence that overall coat color serves as background matching, that side bands are disruptive coloration devices, or that white rumps help in thermoregulation. Concealment appears the principal force driving the evolution of coloration in ungulates with communication, and then thermoregulation, playing less of a role. Key words: bovids, cervids, communication, concealment, countershading, disruptive coloration, pattern blending, phylogeny, thermoregulation, ungulates. [Behav Ecol 14:823–840 (2003)] he adaptive significance of coloration in ungulates has never been addressed systematically. Indeed, for mammals, many of the current hypotheses about coloration are still the same ones formulated more than 100 years ago (Beddard, 1892; Buxton 1923; Mottram, 1915, 1916; Poulton, 1890; Roosevelt, 1911; Thayer, 1909; Wallace, 1889) and have received remarkably little attention ever since (but see Ortolani, 1999). Hypotheses to explain the function of coloration in mammals can be grouped into three broad categories: concealment, communication, and regulation of physiological processes (Cott, 1940). Animals can remain concealed when their overall color resembles, or matches, the natural background in which they live (Endler, 1978). This phenomenon of background matching, also known as general color resemblance, includes crypsis in which overall body color resembles the general color of the habitat, or pattern blending, in which color patterns on the body match color patterns of light and dark in the environment (Cott, 1940; Poulton, 1890). Background matching may be subtle in that an individual may change its pelage or skin coloration annually if backgrounds change between seasons (also termed variable background matching), or may vary with age if the background environments of immatures and adults differ in predictable ways. Alternatively, T Address correspondence to C.J. Stoner. E-mail: cjstoner@ucdavis. edu. Received 17 August 2001; revised 16 January 2002; accepted 21 January 2003. 2003 International Society for Behavioral Ecology animals may achieve concealment through disruptive coloration by sporting contrasting colors or irregular marks that break up the body’s outline (Cott, 1940). Lastly, animals may attain concealment by having a pelage that lightens toward the ventral surface. This counteracts the sun’s effect of lightening the dorsum and shading the ventrum when it shines from above (Kiltie, 1988, 1989; Thayer, 1896, 1909). All of these types of coloration are found in even-toed ungulates. When coloration is conspicuous, as in the black and white faces of oryx species, intraspecific communication (Wickler, 1968) or communication between prey and predators is thought to be involved (Rowland, 1979). For example, patches of color may be used to signal danger to conspecifics (Alvarez et al., 1976) or to signal young to follow their mothers (Leyhausen, 1979). Alternatively, they may be used to signal to predators, the classic example being aposematic coloration in species that have noxious defenses, such as skunks (Spilogale putorius) (Guilford, 1990; Johnson, 1921). Many ungulates exhibit conspicuously colored heads, legs, and anal regions (Guthrie, 1971; Meyer and Kranzle, 1999). Finally, coloration may be related to thermoregulation or protection against ultraviolet rays (Hamilton, 1973). Thus, dark-colored animals may be expected to absorb solar radiation better than light-colored animals do and might therefore be found in colder climates (Burtt, 1979). Even-toed ungulates exhibit a range of hues from black through gray and red to pale. In contrast to birds, none of the major hypotheses for coloration in mammals pertains to mate choice (see Badyaev and Hill, 2000). Behavioral Ecology Vol. 14 No. 6 824 Table 1 Summary of suggested hypotheses for the adaptive significance of coloration in even-toed ungulates and examples Hypothesis, Example Source Overall body coloration Uniform coloration and background matching Reddish coat matches rocky habitats in markhor Uniform coloration and thermoregulation Pale coat minimizes heat exposure in hartebeest in desert Lighter coats in winter and background matching Seasonal color changes reduces conspicuousness in red deer in snowy arctic Spots on adults and pattern blending Dappled coat provides camouflage for giraffe in dappled light of woodland Spots on adults and reduced aggression Dappled coat assists in lowering aggression in axis deer living in groups Stripes on adults and pattern blending Striped coat in greater kudu blends in with riverine habitat Stripes on adults and reduced aggression Striped markings in group-living eland lower aggressive interactions Spots on young and pattern blending Dappled coat in young mule deer provides camouflage in dappled light of woodland Stripes on young and pattern blending Striped coat in young red river hog aids concealment in grassy vegetation Side bands and disruptive coloration Dark stripe breaks up springbok’s form in open environment Side bands and maintaining gregariousness Side bands assist Thomson’s gazelle in maintaining group cohesion Countershading and concealment Light ventrum in the diurnal dibitag counteracts shading from the sun Body parts Dark face markings and communication Black face assists in signaling to group members in oryx White face markings and communication White face assists in signaling to group members in vicuna White face markings and thermoregulation White face helps pronghorn reflect heat in desert Conspicuous face and communication Black and white face in impala assists in signaling to conspecifics Dark leg markings and disruptive coloration Dark leg markings in lechwe break up form in open environments Dark leg markings and communication Black knees assist in signaling to group members in addax White leg markings and communication White legs in bontebok act as visual signals in daylight Conspicuous leg markings and communication White legs in okapi signal to stalking predators Dark tail and communication Black tail in Grant’s gazelle signals to coursing predators White tail and communication White tail in blue duiker is prominent at night Conspicuous tail and communication White tail in white-tailed deer signals to other group members White rump and communication White rump in Soemmering’s gazelle signals to other group members White rump and thermoregulation White rump in bighorn sheep acts to deflect heat in desert 1, 2 2 3 3 4 5, 6 7 5, 8 1 6 4 3, 9 2, 10 11 12 13 14 2, 6, 12 12, 15, 16 4 1 1, 6, 17 18 1, 6, 19 2, 6 Sources: 1 Kingdon, 1982; 2 Spinage, 1986; 3 Cott, 1940; 4 J. Wolff, personal communication; 5 Thayer, 1909; 6 Estes, 1991; 7 Lobao Tello and van Gelder, 1975; 8 Linnell and Andersen, 1998; 9 Fogden and Fogden, 1974; 10 Hasson, 1986; 11 Poulton, 1890; 12 Geist, 1987; 13 Guthrie and Petocz, 1970; 14 Roosevelt and Heller, 1914; 15 Kingdon, 1997; 16 Danilkin, 1996; 17 Bildstein, 1983; 18 Geist, 1978; 19 Ward, 1979. Within these three broad categories, a number of more specific hypotheses have been advanced for overall body coloration and specific parts of the body in artiodactyls (Table 1). Most of these functional hypotheses for coloration were formulated by using one or a handful of ungulate species, and are based on qualitative arguments. Although they may well apply to the species in question, the extent to which they are generally applicable to other species in the order Artiodactyla is completely unknown. The purpose of this study was to test these conventional hypotheses systematically in order to determine their generality within the order. Our main focus was on overall body coloration and color of specific parts of the body rather than a constellation of body parts such as ‘‘contrasting markings’’ (see Geist, 1978) because hypotheses Stoner et al. • Color of artiodactyls have, in general, been formulated for specific areas of the body in this order and our data set allowed us to test such hypotheses at this relatively fine-grained level. The value of the comparative approach used here is that it allows us to investigate the adaptive significance of morphological traits by using a very large number of species, and make conclusions that have wide generality. Similar comparative analyses on coloration in birds have been successful in opening up debate on coloration signaling unprofitability to predators (Baker and Parker, 1979), on linking coloration to lekking (Hoglund, 1989), and in explaining patches of white plumage in waders (Brooke, 1998). METHODS Species Recent discoveries of new species in the order Artiodactyla (see Groves and Schaller, 2000), a growing number of phylogenetic studies focusing on this order, and controversies regarding the distinctness of numerous species and subspecies have led to continual changes in artiodactyl taxonomies. We started analyses based on the species listed in Nowak (1999) and refined this list to exclude (1) purely domesticated species; (2) species that are, or were at one time, extinct in the wild; and (3) species for which very little morphological or behavioral data could be found. Purely domestic species, such as domestic cattle (Bos taurus), domestic goats (Capra hircus), and domestic sheep (Ovis aries), were excluded because their pelage coloration may reflect artificial selection rather than adaptations to natural environments. The llama (Lama glama) and alpaca (Lama pacos) were similarly excluded as Nowak (1999) and Franklin (1982) note that these are generally thought to be domestic descendants of the wild guanaco (Lama guanicoe). Dromedary camels (Camelus dromedarius), were excluded because some taxonomists (see Corbet and Hill, 1991, 1992) believe they never existed in the wild (see Nowak, 1999). Although we included some species that are found both in the wild and in domestic or feral populations (e.g., wild boars [Sus scrofa], Bactrian camels [Camelus bactrianus], banteng and domestic Bali cattle [Bos javanicus], guar [Bos guarus], and yaks [Bos grunniens]), we focused on the coloration or ecological characteristics of wild populations for our analyses. One such species, the water buffalo (Bubalus bubalus), was excluded because some taxonomists feel this scientific name refers only to domestic populations of the species (see Corbet and Hill, 1991, 1992 and discussion in Nowak, 1999). To be conservative, we excluded species that are currently, or were at one time, extinct in the wild: the cave goat, Myotragus balearicus; three gazelle species, Gazella arabica, Gazella rufina, and Gazella saudiya; the blue buck, Hippotragus leucophaeus; and Schomburgki’s deer, Cervus schomburgki. We similarly excluded Pere David’s deer, Elaphurus davidianus, European bison, Bison bonasus, and the Arabian oryx, Oryx leucoryx, because at one time they existed only in captive environments (see Nowak, 1999). Thus, the species retained in our analyses are currently, and have presumably always been, extant in the wild. Finally, we excluded species for which little information on pelage coloration or behavior was available. For example, information regarding Pseudonovibos spiralis is based primarily on horns and accounts from local hunters. Gazella bilkis and Naemorhedus baileyi were also excluded, as their descriptions are based on rare specimens collected in the past (see Grzimeck, 1990; Nowak, 1999). As a result of this selection process, our analyses were based on a total of 200 artiodactyl species (see Supplementary Material online). 825 Data collection Information on overall body coloration, seasonal coloration change, and coloration of underparts was obtained from descriptions in the literature; face, leg, rump, and tail coloration was determined from photographs (Table 2 and appendix). Information on ecological characteristics was taken from descriptions in the literature and was placed into categories; latitude characteristics were determined from range maps (Table 3). With a few exceptions, we principally looked for associations between coloration and ecological rather than behavioral variables because behavioral hypotheses for a color marking, such as ‘‘follow-me’’ signals, are difficult to test. Such signals might apply between mother and young (Schaller, 1967), between members of small groups on the move (Hickman, 1979), or between members of large social groups; they are more amenable to tests through direct observation. For all variables, a value of one was assigned to species demonstrating a given trait, zero to species that did not display the trait, and a ‘‘?’’ for species for which little or no information on the variable was available (supplementary appendix online). Variables were scored present or absent as variation in pelage characteristics and ecological and behavioral characters was unknown for most species. Categories were not mutually exclusive. For example, a species demonstrating light red coloration would be assigned a value of one in both the ‘‘pale’’ and ‘‘red’’ color categories. Analyses Nonparametric tests To identify general trends between coloration traits and ecological variables, we first applied chi-squared tests separately to the two largest families, cervids (N ¼ 39 species) and bovids (N ¼ 125 species), to identify associations that were present within particular families. When sample sizes were inadequate to perform chi-squared statistics, we instead used Fisher’s Exact probability tests. We then performed these analyses for all the artiodactyls (N ¼ 200 species) in order to include important families such as the Suidae and Giraffidae and to determine whether associations held across the whole order. For each test, we included only those species for which information was available for the coloration and ecological variables of interest. Only significant results are reported in the text (p , .1 because we were searching for associations using very coarse behavioral and ecological measures). We felt it important to present cross-species comparisons in light of the current controversy over the applicability of phylogenetically controlled methods (Irschick et al., 1997; Price, 1997), although in our discussion we attach far more weight to results that controlled for shared ancestry (see below). Phylogenetic comparisons Cross-species comparisons fail to account for the fact that species values are nonindependent: shared character states may reflect common ancestry rather than independent adaptations (Harvey and Pagel, 1991). To control for the potential effects of shared ancestry, we analyzed the same hypotheses as in the nonparametric tests by using a composite phylogenetic tree and Maddison’s concentrated-changes test (Maddison, 1990) as implemented in MacClade (Maddison and Maddison, 1992). Concentrated-changes tests have been reported to detect such correlated changes statistically based on fewer character state changes compared with that of other methods (Cooper, 2002; Cooper et al., 2002; Ridley, 1983; Sillen-Tullberg, 1993). Although concentrated-changes tests rely on the topology of a given phylogeny rather than on Behavioral Ecology Vol. 14 No. 6 826 Table 2 Descriptions of coloration variables (see Appendix for sources) Variables Overall body colorationa Pale Red Gray Dark Seasonal coloration change Lighter in winter Underparts Countershading a Descriptions White, sandy, light brown, pale gray, light gray, buffy gray, light buff, light red, fawn, tawny, or yellowish-brown coloration Red, rusty, chestnut, tawny, rufous, ochraceous, cinnamon, or auburn tones Gray or grayish coloration Black, blackish, or dark brown coloration Pelage becomes white or lighter seasonally Ventral coloration described as paler than upper parts, white, dingy-white, whitish, or tinged with white Color patterns Presence of spots in young Presence of spots on adults Presence of stripes (vertical or horizontal) on young Presence of stripes (vertical or horizontal) on adult Presence of side band Black or white spots on dorsal surface, rear or legs Black or white spots on dorsal surface, rear or legs Black or white stripes on dorsal surface, rear or legs Black or white stripes on dorsal surface, rear or legs Prominent line along side of body Tail Light Dark Markings on tail white or lighter than coloration of rump Markings on black or darker than coloration of rump Rump White markings Presence of rump patch or white markings on rump Face Conspicuous light markings Conspicuous dark markings White or light markings on face that contrast with the overall body coloration Dark or black markings on face that contrast with the overall body coloration Legs Conspicuous light markings Conspicuous dark markings White or light markings on legs that contrast with the overall body coloration Dark or black markings on legs that contrast with the overall body coloration Sexually dichromatic species were scored as both colors. incorporating branch lengths estimates like more rigorous methods (see Nee et al., 1996; Pagel, 1994; Read and Nee, 1995), we deemed these tests appropriate for our analyses owing to the scarcity of branch length and divergence time estimates for ungulate taxa (Pérez-Barberı́a and Gordon 1999, 2000, 2001; Pérez- Barberı́a et al. 2002). Figure 1 demonstrates the hypothesized phylogenetic tree on which analyses were based. Owing to the lack of consensus over the placements of Artiodactyla taxa in the literature, we based this composite tree primarily on the findings of recent molecular studies and supplemented these with taxonomies and studies based on morphological traits. The general positions of the Cervidae, Bovidae, Giraffidae, Antilocapridae, Tragulidae, Hippopotamidae, Suidae, and Camelidae were based on cladistic analyses by Gatesy et al. (1999). Additional sources were used to resolve the genus Moschus (Su et al., 1999), the Tragulidae (based on taxonomy in Grubb, 1993), Suidae (Groves, 1981; Randi et al., 1996; Thenius, 1970), Peccaridae (Theimer and Keim, 1998), and Camelidae (based on Gatesy et al., 1999: Figures 11 and 12). Cervidae branch. The general positions of the CervinaeMuntiacus, Hydropotes-Alces-Capreolus, and Rangifer-New World odocoileinae clades were inferred from Cronin et al. (1996), Douzery and Randi (1997), Gatesy et al. (1999), and Randi et al. (1998). Resolution within Cervinae-Axis-Dama was based on Emerson and Tate (1993) and Randi et al. (2001), the Elaphodus-Muntiacus clade was based on Amato et al. (2000), and the placements of Hydropotes, Alces and Capreolus were based on Randi et al. (1998). The placements of Hippocamelus, Pudu, Blastoceros, and Ozotoceros (in relation to Rangifer, Odocoileus, and Mazama) were based on Webb (2000). Bovidae branch. All information gained from the simultaneous analysis of data sets in Gatesy and Arctander (2000) was retained in the Bovidae tree. Gatesy and Arctander (2000) was used to infer the basic branches within Bovidae and supplemented with information from the following sources. Resolution within the bovini and tragelaphini tribes was based on Schreiber et al. (1999) (Bulbus, Bos), Gatesy and Arctander (2000) (Syncerus, Psuedorx, Boselaphus), Hassanin and Douzery (1999a,b) (Tetracerus quadricornis in clade with Boselaphus tragocamelus), and Mathee and Robinson (1999) (Tragelaphus, Taurotragus). Resolution within the Antilopini clade was based on Gatesy et al. (1999), Mathee and Robinson (1999), Rebholz and Harley (1999) (general positions of Rachicerus and Madoqua clades), and Brashares et al. (2000) (resolution within the Rachicerus and Madoqua clades). Resolution within the Gazella-Antilope clade was based on Rebholz and Harley’s (1999) parsimony cladogram. Placements of Litocranius walleri, Ammodorcas, Antidorcus, Dorcatragus were based on Brashares et al. (2000). Within the Alcelaphini, the position of the Neotragus/Aepyceros clade (Hassanin and Douzery, 1999b; Mathee and Robinson, 1999) was based on Gatesy and Arctander (2000). Resolution within Neotragus was taken from Brashares et al. (2000). Positions of Connochaetes, Damaliscus, Acelaphus, and Sigmoceros were taken from Gatesy and Arctander (2000), whereas Brashares et al. (2000) provided Stoner et al. • Color of artiodactyls 827 Table 3 Descriptions of ecological and behavioral variables (see Appendix for sources) Variables Descriptions Latitude categories Tropics Subtropics Transitional Subarctic Arctic Part Part Part Part Part Environmental categories Open Closed Grasslands, deserts, or tundra habitats Swamp/riverine, or dense forest habitats Habitat categories Grassland Bushland Woodland Light forest/woodland Dense forest Desert Rocky Tundra Swamp/riverine Occupies prairie, savannah, meadows, or steppe grasses habitats Occupies scrub, bushland, riparian, thicket, or shrub vegetation habitats Occupies woods, woody areas, or woodlands All species included in the woodland category, plus those occupying open, sparse, or light forests Occupies alpine, tropical, boreal, deciduous, mixed, timberland, or dense forests Found in deserts or semideserts Occupies rocky areas such as talus, boulders, rocky outcrops, crevices, or cliffs Found in tundra Occupies swamp, marsh, bogland, moorland, reedbeds or riverine habitats Activity pattern categories Diurnal Crepuscular/nocturnal Primarily active during the day Primarily active during the night, or in the morning or late/afternoon evening Group size categories Solitary Intermediate groups Large groups Primarily found alone or in pairs Primarily found in groups of 250 individuals Forms aggregations of more than 50 individuals Hider/follower behavior Hider Follower Young lie concealed for more than 1 week after birth Young follow mothers within 1 week of birth Hunting style of principal predators Stalker Courser Predator usually observes/follows/stalks prey before attacking Predator usually runs down/exhausts prey of of of of of range range range range range falls falls falls falls falls in in in in in tropic tropic tropic tropic tropic the topography within Damaliscus. Within the Caprini-Rupicaprini-Ovibovini clade, the general placement of Ovis, Oreamnos, Capra, Hemitragus (species assumed to be monophyletic), Pseudois (assumed to be monophyletic), Pantholops, Ovibos, Rupicapra, Nemorhaedus (assumed to be monophyletic), and Capricornis was inferred from Gatesy and Arctander (2000). Resolution within Ovis was based on Hassanin et al. (1998), Ludwig and Knoll (1998), Ludwig and Fisher (1998), and Hiendleder et al. (1998). The placement of Budorcas taxicolor is based on Hassanin and Douzery (1999a,b). The topography of Capra species is based on Hassanin et al. (1998) and Manceau et al. (1999). The placement of Ammotragus with Rupicapra was based on Hassanin and Douzery (1999b), whereas the topography of Capricornis species was based on Hassanin et al. (1998). In the tribe Hippotragini, the placement of Oryx, Addax, and Hippotragus was based on Gatesy and Arctander (2000). The general placements of Redunca, Kobus, Pelea, and Cephalophus branches in the Redunci-Cephalophini tribes were based on Gatesy and Arctander (2000). Relationships within these genera were derived from additional sources: Brasheres et al. (2000) (Redunca, Cephalophus), Gatesy and Arctander (2000) (Kobus), Hassanin and Douzery (1999b), and Mathee and Robinson (1999) (placement of Oreotragus in clade with Cephalophus). Some species were excluded from the composite phylogeny because relatively little information could be found regarding their phylogenetic position (e.g., Bos sauveli, Tragelaphus latitudes latitudes latitudes latitudes latitudes (010 degrees from equator) (1130 degrees) (3150 degrees) (5160 degrees) (more than 60 degrees) buxtoni, Cephalophus rufilatus, Cephalophus natalensis, Cephalophus nigrifrons, Modoqua piacentii), or great disagreements exist regarding their placements (e.g., Saiga saiga, Oribi oribi, and Procapra species). To resolve polytomies in the composite tree (as MacClade requires a fully dichotomized tree for concentrated-changes tests), we used an online library database (BIOSIS) to count the number of citations retrieved for each species in the polytomy (Nunn C, personal communication). Species with the least number of citations (indicating those for which less information is known) were excluded from the polytomies until a fully dichotomized tree remained. These species included Ovis vignei, Ovis nivicola, Hemitragus jayakari, Capra wallie, Capra falconeri, Capra pyrenaica, and Oryx dammah. To prepare for phylogenetic analyses, we first used MacClade to map the coloration (dependent) and ecological (independent) variables onto the phylogenetic tree using the matrix depicting each species’ character states (one, zero, or ‘‘?’’; supplementary appendix online). From the distribution of character states across all species, MacClade reconstructs the evolutionary history of a given trait throughout the tree by using parsimony. This makes it possible to count the number of evolutionary gains (change in a character state from a zero to a one) and losses (change from a one to a zero) in either the coloration or ecological variables (Maddison and Maddison, 1992). In those instances in which character reconstruction was ambiguous (i.e., the tree contained areas in which both zero and one were equally parsimonious), within 828 Behavioral Ecology Vol. 14 No. 6 Figure 1 The hypothesized phylogeny artiodactyl for species included in concentrated-changes tests was based on previously published phylogenetic studies. Upper star denotes origin of the Cervidae; lower star, origin of the Bovidae. Stoner et al. • Color of artiodactyls MacClade we used ‘‘most parsimonious reconstruction mode,’’ which includes all reconstructions produced by the deltran and acctran options, to generate all possible reconstructions of the character. Because multiple ambiguities for the same character can lead to many possible reconstructions, we selected the first and the last reconstructions for the first analysis (following the method of Ortolani 1999; Ortolani and Caro 1996). We examined the first and last reconstructions as these are the reconstructions with the maximum number of gains and fewest losses, and minimum number of gains and most losses, respectively. Next, we used concentrated-changes tests to test the probability that gains in the coloration variable were associated with the ecological variable more than could be expected by chance, and losses in the coloration variable occurred in the presence of the ecological trait less than expected by chance. Thus, we examined whether the presence of the ecological trait facilitated the maintenance of the color trait over evolutionary time. The null hypothesis, tested against a distribution derived through simulation and 10,000 replicates, was that gains and losses in a given coloration variable were randomly distributed on the tree with respect to the ecological variable. When ambiguity in character reconstruction existed, the multiple reconstructions resulted in two or four probability values for each test, depending on whether one or both of the dependent and independent variables displayed ambiguity. Because, in concentrated-changes tests (but not the nonparametric tests), a dependent variable could be matched to an independent variable more than once because of multiple phylogenetic reconstructions, we applied a standard Bonferroni correction. This is a more conservative way to apply this test than by using a sequential Bonferroni (Rice, 1989). Thus, if four tests had been conducted between a coloration and ecological variable, we multiplied each p value by four and reported it only if the product was less than 0.1. Following the method of Rice (1989), we limited application of the Bonferroni correction to multiple tests of the same pair of dependent and independent variables rather than across one dependent and several independent variables because each independent variable exhibited a different phylogenetic reconstruction and was used to test a different hypothesis. If several concentrated-changes tests were conducted for a given prediction, we report only the most significant p value corrected using the standard Bonferroni technique. For the sake of convenience, even-toed ungulates is a general term we use loosely to refer to all species that we analyzed, cervids refers to Cervidae, bovids to the Bovidae, and artiodactyls to all species in the supplementary appendix online. As a short hand, ‘‘cross-species comparisons’’ refer to species counts analyzed by using chi-square tests or Fisher’s exact probability tests that do not control for phylogeny, whereas ‘‘phylogenetically controlled tests’’ refer to the results of concentrated changes tests using the reduced species list, termed CCT. For Fisher’s and chi-square tests, true p values are given, but for CCT, p values are given after Bonferroni corrections. We consider an association worth discussing if any of the phylogenetic reconstructions was significant after a Bonferroni correction. RESULTS Coloration for concealment Uniform coloration To discover whether ungulate coloration aids in concealment, we first tested whether uniform pelage color typically matches habitat background. Cross-species comparisons showed that 829 pale bovids and artiodactyls were likely to be found in open environments (N ¼ 125 species, v2 ¼ 9.28, p , .01; N ¼ 194, v2 ¼ 13.26, p , .01, respectively), but results failed to reach significance when confounding effects of shared ancestry were taken into account. Pale bovids and artiodactyls were also associated with one habitat category, deserts, but only in cross-species comparisons (N ¼ 125, v2 ¼ 6.91, p ¼ .03; N ¼ 194, v2 ¼ 10.19, p , .01, respectively). Next, we tested gray and red pelage coloration against rocky habitats and found gray bovids (N ¼ 125, v2 ¼ 8.20, p ¼ .02) and artiodactyls (N ¼ 194, v2 ¼ 8.80, p ¼ .01) were significantly more likely to live in this habitat, but again, these results disappeared using phylogenetic tests. Surprisingly, red pelage was associated with living in habitats that were not rocky (bovids: N ¼ 125, v2 ¼ 9.80, p , .01). Dark bovids and artiodactyls were found significantly more often in closed environments (N ¼ 125, v2 ¼ 4.37, p , .0001; N ¼ 195, v2 ¼ 33.2, p , .0001, respectively), and dark cervids, bovids, and artiodactyls were found in dense forest habitats (Fisher’s test: N ¼ 38, p ¼ .02; N ¼ 125, v2 ¼ 16.16, p , .01; N ¼ 195, v2 ¼ 28.23, p , .0001, respectively), although not using CCTs. An association between dark bovids and swamp habitats approached significance (N ¼ 125, v2 ¼ 5.30, p ¼ .08). Lighter coats in winter If seasonal pelage coloration contributes to concealment, species displaying a lighter winter coat should occupy habitats which become lighter seasonally. Cervids, bovids, and artiodactyls were likely to turn lighter in winter if they lived in arctic latitudes (Fisher’s test: N ¼ 39, p ¼ .07; N ¼ 125, v2 ¼ 2.72, p , .01; N ¼ 197, v2 ¼ 24.79, p , .0001, respectively; CCTs for both bovids and artiodactyls: p , .001). Bovids and artiodactyls with lighter winter coats were also associated with tundra habitats (N ¼ 125, v2 ¼ 12.72, p , .01; N ¼ 197, v2 ¼ 24.37, p , .0001, respectively), and this association held true for all three taxonomic groups using CCTs (p ¼ .07, p , .01, p , .001, respectively). In short, there was strong backing for ungulates taking on lighter coats in arctic and tundra regions. Spotted coats If spotted coats facilitate crypsis through pattern blending, having the appearance of dappled light would be favorable for diurnal, solitary, forest-dwelling species and those that hide their young. Cervids with spotted adult coats were diurnal (CCT: p ¼ .06) but were less likely to be solitary (Fisher’s test: N ¼ 37, p ¼ .03). Conversely, bovids with spotted young and spotted adult coats were likely to be solitary (N ¼ 125, v2 ¼ .79, p ¼ .03, N ¼ 125, v2 ¼ 5.17, p ¼ .08, respectively), as were species with spotted young across all artiodactyls (N ¼ 193, v2 ¼ 13.90, p , .01; CCT: p ¼ .02). Spotted adult bovids were associated with light forests (N ¼ 125, v2 ¼ 9.97, p , .01; CCT: p ¼ .08). Cervids with spotted young were strongly tied to grassland/bushland habitats (CCT: p , .0001), bovid spotted young to light forests habitats (N ¼ 125, v2 ¼ 11.66, p , .01; CCT: p ¼ .06), and spotted young across all artiodactyls to dense forests (N ¼ 196, v2 ¼ 13.62, p , .01). Species with spotted young were significantly more likely to hide after birth in all taxonomic groups (cervids, Fisher’s test: N ¼ 38, p ¼ .09; bovids: N ¼ 123, v2 ¼ 5.60, p ¼ .06; artiodactyls: N ¼ 194, v2 ¼ 23.40, p , .0001, CCT: p ¼ .05) (Figure 2). Indeed, spotted cervid, bovid, and artiodactyl young were significantly less likely to follow their mothers (Fisher’s: N ¼ 38, p ¼ .09; N ¼ 123, v2 ¼ 5.90, p ¼ .05; N ¼ 194, v2 ¼ 24.61, p , .0001, respectively). In short, there was patchy support for adult spotted coats helping with pattern blending, whereas in young, spotted coats were found in forest bovids, grassland cervids, and hiding and solitary artiodactyls, strongly supporting a concealment hypothesis. 830 Behavioral Ecology Vol. 14 No. 6 Figure 2 Association between spotted coats in young artiodactyls and hider species. The first reconstruction was used for both phylogenies, and the associated p value is .05 after Bonferroni correction. Black lines indicate species with spotted coats; white lines indicate nonspotted species. For the boxes on the right, black boxes indicate species that are hiders; white boxes indicate species that are not hiders. Stoner et al. • Color of artiodactyls Striped coats Similar analyses of striped patterns revealed that adults displaying stripes were significantly associated only with light forests in bovid and artiodactyl species (N ¼ 125, v2 ¼ 15.15, p , .001, CCT: p , .1; N ¼ 197, v2 ¼ 11.32, p , .01, CCT: p ¼ .09, respectively) (Figure 3). We found marginal associations between bovids with striped young and light forests (N ¼ 125, v2 ¼ 15.15, p , .001, CCT: p , .1) and artiodacyls with striped young and dense forests (N ¼ 197, v2 ¼ 4.88, p ¼ .09). There was a tendency for artiodactyls with striped young to be solitary (N ¼ 194, v2 ¼ 5.33, p ¼ .07), but this did not hold after controlling for phylogeny. Bovids and artiodactyls with striped young were chiefly those that exhibited a hiding strategy (N ¼ 123, v2 ¼ 5.87, p ¼ .05; N ¼ 195, v2 ¼ 11.60, p , .01, respectively), and they did not exhibit following behavior (N ¼ 123, v2 ¼ 6.34, p ¼ .04; N ¼ 195, v2 ¼ 12.21, p , .01, respectively); however, these results did not hold up after controlling for phylogeny. In short, striped coats in adults were found in light forests, and striped young were associated with hiding species. Dark side bands and dark leg markings If side bands act as disruptive coloration by acting to break up the body shape, they may aid in concealment for species that are diurnal or are found in open and sunny environments in which other forms of crypsis are limited. Artiodactyls with prominent side bands were principally diurnal (N ¼ 187, v2 ¼ 6.41, p ¼ .04) and found in open environments (N ¼ 194, v2 ¼ 14.73, p , .001), as were bovids (N ¼ 124, v2 ¼ 11.84, p , .01). There were strong associations between side bands and desert-dwelling for both bovids (N ¼ 124, v2 ¼ 10.00, p , .01) and artiodactyls (N ¼ 194, v2 ¼ 14.71, p ¼ .0006). Reindeer (Rangifer tarandus), the only cervid with side bands, is found in tundra (Fisher’s: N ¼ 38, p ¼ .03). None of the analyses of side bands remained significant after controlling for phylogeny. Dark leg markings, which may serve as another form of disruptive coloration, should similarly be associated with open environments. We found that artiodactyls with dark legs were likely to live in deserts (N ¼ 196, v2 ¼ 8.36, p ¼ .0.02; CCT: p ¼ .07), although there was no association with other open habitats such as tundra. Countershading As predicted if countershading acts to conceal shadows cast on the lower body, countershaded bovids were diurnal (CCT: p ¼ .06) and strongly associated with desert habitats (CCT: p ¼ .01). Comparisons across all countershaded artiodactyls similarly revealed an association with desert-dwelling (CCTs: p ¼ .03). Conversely, there were fewer countershaded species living in tundra than expected by chance (bovids: N ¼ 125, v2 ¼ 12.11, p , .01; artiodactyls: N ¼ 195, v2 ¼ 10.03, p , .01). Coloration for communication Markings may be most effective as visual signals if they are associated with conditions in which they are most visible (e.g., diurnal activity patterns and open habitats), associated with gregarious species (for intraspecific communication), or associated with particular predator types (for interspecific communication). Spotted and striped coats in adults Spotted adult pelages might provide a mechanism to reduce intraspecific aggression because spotted coats resemble those of young in many species and might therefore act as a signal of subordination. Spotted cervids were more likely to be 831 found in intermediate-sized groups (two to 50 individuals; Fisher’s test: N ¼ 37, p , .01), but there were no associations between spotted coats in adults and living in intermediatesized or large (more than 50) groups after controlling for phylogeny. There were no significant associations between striped coats in adults and intermediate or large group sizes. Side bands If side bands are used in communication between conspecifics, perhaps to maintain gregariousness, they should be found in group-living species. Artiodactyls with side bands lived in intermediate-sized (N ¼ 192, v2 ¼ 8.56, p , .01) and large groups (N ¼ 192, v2 ¼ 15.91, p , .0001; CCT: p ¼ .01), providing support for this hypothesis. Face markings Dark visual signals for intra or interspecific communication might be most noticeable in diurnal species occupying open habitats. Cross-species comparisons revealed that cervids with dark faces tended to be diurnal (N ¼ 36, v2 ¼ 6.77, p ¼ .03), but there was no evidence that dark-faced species in any of the three clades lived in open environments. Dark-faced bovids and artiodactyls lived in intermediate (CCTs: p , .01; p , .01, respectively) and large-sized social groups (CCTs: p ¼ .02; p ¼ .02, respectively) (Figure 4), as is expected if dark faces are used to communicate with conspecifics. Although dark-faced artiodactyls were apparently subject to predation by stalking predators (N ¼ 188, v2 ¼ 8.79, p ¼ .01), there were also indications that bovids and artiodactyls with dark faces were pursued by coursers after controlling for shared ancestry (CCTs: p , .01; p , .01, respectively). If white face markings are used in communication, they might be prominent signals for either diurnal or nocturnal species. Artiodactyls with white faces were likely to be diurnal (CCT: p ¼ .01) but not nocturnal, and to live in intermediatesized social groups (CCT: p ¼ .06). Bovids with white faces were likely to be pursued by coursers (N ¼ 119, v2 ¼ 5.92, p ¼ .06), although this result was not upheld using phylogenetic tests. Because any conspicuous markings might provide an effective signal, rather than dark or white facial markings per se, we repeated the analyses for species with either dark or white markings, or a combination of the two. Cervids and artiodactyls with conspicuous faces inhabited grassland/ bushland habitats (CCTs: p ¼ .03; p ¼ .03, respectively), but no other associations were significant. Bovids with conspicuous faces were less likely to be found in large groups than expected by chance (N ¼ 125, v2 ¼ 6.05, p ¼ .01). In short, there was rather little support for conspicuous faces having an obvious communicatory role. Leg markings Artiodactyls with dark leg markings were associated with deserts (N ¼ 196, v2 ¼ 8.36, p ¼ .02; CCT: p ¼ .07), where this color pattern is presumably prominent. Artiodacyls with dark legs are found in large social groups (CCT: p , .1), as expected if such markings facilitate intraspecific communication; there was no association with stalking or coursing predators for any taxonomic group. White leg markings were associated with diurnality in artiodactyls (CCT: p ¼ .04), but we found no associations between these markings, open habitats, gregariousness, or predator types. We repeated leg marking analyses combining species with either dark or white leg coloration and found that both bovids and artiodactyls with conspicuous legs were diurnal (CCTs: p ¼ .02; p ¼ .08, respectively) and principally lived in deserts 832 Behavioral Ecology Vol. 14 No. 6 Figure 3 Association between striped coats in adult artiodactyls ungulates and living in light forests. The first reconstruction was used for both phylogenies, and the associated p value is .09 after Bonferroni correction. Black lines indicate species with striped coats; white lines indicate nonstriped species. For the boxes on the right, black boxes indicate species living in light forests; white boxes indicate species not living in light forests. Stoner et al. • Color of artiodactyls 833 Figure 4 Association between artiodactyls with dark faces and living in large groups. The first reconstruction was used for both phylogenies, and the associated p value is .02 after Bonferroni correction. Black lines indicate species with dark faces; white lines indicate species lacking dark faces. For the boxes on the right, black boxes indicate species living in large groups; white boxes indicate species that do not live in large groups. Behavioral Ecology Vol. 14 No. 6 834 (CCTs: p ¼ .04; p ¼ .08, respectively). Artiodactyls with conspicuous legs were associated with also grassland/bushland habitats (CCT: p ¼ .05). These findings suggest a role in communication, but because no significant associations were uncovered with group sizes or predator hunting style, it is unclear whether intra or interspecific communication is involved. Tail markings Chi-square analyses gave some backing for dark tail markings acting as visual signals. Dark tails were found among diurnal bovids (N ¼ 121, v2 ¼ 9.00, p ¼ .01) and artiodactyls (N ¼ 188, v2 ¼ 7.17, p ¼ .03). Bovids with dark tails tended to form large groups (N ¼ 125, v2 ¼ 8.76, p ¼ .01), whereas artiodactyls with dark tails were associated with intermediate (N ¼ 192, v2 ¼ 8.33, p ¼ .02) and large groups (N ¼ 192, v2 ¼ 8.69, p ¼ .01). Dark-tailed bovids and artiodactyls were associated with stalking predators (N ¼ 119, v2 ¼ 9.47, p , .01; N ¼ 187, v2 ¼ 6.16, p ¼ .05, respectively), and there were no significant associations with coursing predators. None of these results were upheld in phylogenetic tests, giving little support to conventional hypotheses for dark tail coloration. White tails were associated with diurnality in bovids (N ¼ 121, v2 ¼ 7.01, p ¼ .03) but not with being diurnal or living in open environments in other clades. There were few associations between white tails and gregariousness, except whitetailed artiodactyls were weakly associated with large-sized groups (CCT: p ¼.1). There was an association between whitetailed bovids and stalking predators, but only in cross species comparisons (N ¼ 119, v2 ¼ 10.82, p , .01). Thus, there was little support for white tails having signal value. In examining species with either black or white tails, we found that bovids and artiodactyls with any conspicuous markings were diurnal (N ¼ 121, v2 ¼ 9.34, p , .01, CCT: p , .01; N ¼ 188, v2 ¼ 7.40, p , .01, CCT: p ¼ .06, respectively). Artiodactyls with tail markings were associated with intermediate-sized (N ¼ 193, v2 ¼ 3.09, p ¼ .08, CCT: p ¼ .07) and large (N ¼ 193, v2 ¼ 4.47, p ¼ .03, CCT: p ¼ .09) groups, as were bovids (N ¼ 125, v2 ¼ 5.88, p ¼ .02). Conspicuous tails could be involved in interspecific communication, as bovids with conspicuous tails were attacked by stalking predators (N ¼ 119, v2 ¼ 12.48, p , .001), although cervids were actually less likely to be attacked by stalkers (Fisher’s test: N ¼ 36, p , .01); significance for these analyses were lost after controlling for phylogeny. In short, there was reasonably strong support for conspicuous tails being associated with intraspecific communication. White rumps In examining whether white rumps serve a communicatory role, we found white rumps were associated with diurnality in artiodactyls (N ¼ 188, v2 ¼ 5.15, p ¼ .07; CCT: p ¼ .09) and open habitats in bovids (N ¼ 125, v2 ¼ 21.14, p , .0001; CCT: p ¼ .05) and artiodactyls (N ¼ 195, v2 ¼ 33.29, p , .0001; CCT: p , .0001). There were strong associations using crossspecies comparisons between white rumps and living in intermediate-sized groups for cervids (Fisher’s test: N ¼ 36, p , .0001; CCT: p , .01), bovids (N ¼ 125, v2 ¼ 16.48, p , .001), and artiodactyls (N ¼ 193, v2 ¼ 32.32, p , .0001) and living in large groups (cervids: N ¼ 36, v2 ¼ 7.29, p ¼ .04; bovids: N ¼ 125, v2 ¼ 23.64, p , .0001; artiodactyls: N ¼ 193, v2 ¼ 33.37, p , .0001). Bovids and artiodactyls with white rumps were likely to be pursued by coursers (N ¼ 119, v2 ¼ 13.88, p , .1; N ¼ 187, v2 ¼ 8.48, p ¼ .01, respectively), but absence of significance in phylogenetically controlled tests casts doubt on their role in signaling to predators. No associations were found between species sporting white rumps and those preyed on by stalking predators. Coloration for thermoregulation Uniform coloration Bovids and artiodactyls with dark pelages were found principally in the tropics (bovid CCT: p , .0001; artiodactyl CCT: p ¼ .01), as expected from Gloger’s rule. Pale bovids and artiodactyls were associated with deserts in chi-square tests (N ¼ 125, v2 ¼ 6.91, p ¼ .03; N ¼ 194, v2 ¼ 10.19, p , .01, respectively) but not after controlling for phylogeny. Face markings To test whether white face markings serve to reduce heat stress, we checked whether they are found in species living in open and warm environments. White-faced cervids were more likely to live in open environments (Fisher’s test: N ¼ 39, p ¼ .06; CCT: p ¼ .06). White faces were associated with living in grassland/bushland in bovids (N ¼ 125, v2 ¼ 6.19, p ¼ .05) and in artiodactyls (N ¼ 196, v2 ¼ 4.82, p ¼ .09; CCT: p , .01). White faces were also found in desert bovids (N ¼ 125, v2¼ 6.19, p ¼ .05). In sum, white faces were associated with diurnality, open environments and grassland/bushland habitats. These findings provide evidence for a thermoregulatory role. White rumps Turning to the idea that rumps may also deflect heat in eventoed ungulates, we found significant relationships between species having white rumps and those occupying deserts (cervids, Fisher’s test: N ¼ 38, p ¼ .08; bovids, N ¼ 125, v2 ¼ 24.37, p , .0001; artiodactyls, N ¼ 195, v2 ¼ 33.29, p , .0001) and grassland/bushland (bovids, N ¼ 119, v2 ¼ 7.92, p ¼ .05; artiodactyls, N ¼ 187, v2 ¼ 8.39, p ¼ .02). These associations were not upheld when shared ancestry was taken into account, however. DISCUSSION After taking phylogeny into account, we found that many aspects of coloration in ungulates are related to concealment and are less involved in communication. Thermoregulation apparently played only a minor role in the evolution of coloration in this taxon. Although our results are preliminary, they point to predation pressure being a key selective force acting on coloration in this order of mammals. The importance of mate choice was not assessed but is likely to be of minor import, judging from the small number of ungulate species that are sexually dichromatic. If correct, this contrasts sharply with birds, although a phylogenetically controlled analysis of sexual dichromatism and mating system in ungulates would still be useful. Before discussing our findings in more detail, we must allude to five points that could have influenced our results. First, we know that our behavioral and especially our ecological categories are broad. Thus, dense forest constituted alpine, tropical, boreal, deciduous, mixed dense forests, and timberland, where lighting conditions are likely to be extremely variable (Endler, 1993) and would influence the extent to which prey might remain cryptic. Indeed, variation in lighting within subcategories of dense forest might even be greater than between dense forest and light forest/woodland, but it is impossible to measure this across the range of environments in which species live. More refined categories might increase our ability to detect associations. Second, the large scope of the study meant that we lacked detailed data on most species and were forced to use a coarse categories that could be ascribed to most species, such as habitat or group size. If analyses were restricted to a smaller sample of well-known species, one might be able to use more Stoner et al. • Color of artiodactyls sensitive measures, such as ‘‘is’’ or ‘‘is not’’ preyed up by cheetahs (Acinonyx jubatus) and thereby be able to make far more specific predictions about prey-predator signaling, for example (see Caro, 1994). Third, our attempts to tease out predictions that would lend support or cast doubt on conventional hypotheses may have been misplaced. Thus, it may be inappropriate to suggest that if a color patch is used in communication, we would expect it to be found in diurnal or open country species; perhaps they would be found in crepuscular species or forest-dwelling species too. Fourth, our phylogenetic reconstruction was of low resolution, as a complete molecular phylogeny is not available. In the absence of such data, we constructed a composite tree based on the most recent molecular studies and supplemented this with taxonomies based on morphological traits. Certain species were additionally excluded because of lack of information or strong disagreements over their phylogenetic position, and we resolved polytomies by using the arbitrary procedure of excluding species with fewest citations. Nevertheless, this is the most comprehensive tree based primarily on molecular studies that is currently available for Artiodactyla. Although a more resolved tree might in the future change our results because concentrated changes tests are notoriously sensitive to inclusion or exclusion of species (Sillen-Tullberg, 1993), we think that this would not produce more significant associations. Nonetheless, future reanalysis of data in the supplementary appendix online by using a thorough molecular phylogeny will definitely be worthwhile. Fifth, we used stringent criteria to accept that an association between a coloration pattern and an ecological or behavioral variable was indeed significant by focusing attention on phylogenetic reconstructions that were significant only after a standard and therefore conservative Bonferroni correction. With these points in mind, we now discuss specific associations summarized in Table 4. Overall body coloration If analyses had been restricted to simple cross-species comparisons, there would have been strong support for overall coat coloration being an adaptation for background matching or thermoregulation (Kingdon, 1982; Spinage, 1986). Pale coats were found in open environments and desert habitats, and dark coats were found in closed environments and dense forests. Gray coats were found in rocky habitats, which would support only a background-matching role. Nonetheless, none of these significant associations were maintained once shared ancestry was taken into account. Moreover, on some of these tests, species counts revealed that there were actually more species not associated with the variable of interest (pale bovids not found in deserts; gray artiodactyls not found on rocks), despite there being more pale species in deserts than non-pale species in deserts. In contrast, parallel comparative analyses of overall body coloration in carnivores (Ortolani and Caro, 1996) and lagomorphs (Stoner et al., 2003) revealed that pale coloration was associated with desert habitats, making the even-toed ungulate results even more surprising. In conclusion, despite several extraordinarily experienced naturalists having linked overall coat color to habitat on the basis of observations made on single species or guilds of ungulates, there is no comparative systematic support for background matching across the even-toed ungulates. Although a more resolved phylogeny might, in the future, be more supportive of background matching, we speculate that overall coat coloration, such as being pale, serves different functions in different species, and that ascribing the function of all pale coats either 835 to background matching or to reflecting heat in hot climates is too simplistic. Cross taxonomic support for Gloger’s rule is equivocal. This study showed that dark even-toed ungulates live in the tropics; similarly, in canids, ursids, and herpestids, dark species are found in tropical forests (Ortolani and Caro, 1996). Assuming many tropical areas are humid, this result provides indirect support for Gloger’s rule, although the causal mechanisms maintaining the association remain opaque. In contrast, dark lagomorphs were not associated with tropical latitudes after controlling for phylogeny (Stoner et al., 2003), so the generality of this rule in mammals is still open to question. Ungulates that assume lighter coats in winter are associated with arctic latitudes and tundra habitats. This might be an adaptation to match their snowy background (Cott, 1940) or a thermoregulatory device, as white coats are translucent and let heat and sunlight penetrate, which are then caught by darker inner hairs. Dark outer hairs would only absorb heat remaining on the surface that is quickly carried away by wind (Wolff J, personal communication). Similar matches between white coats and living in the arctic have been found in phylogenetic comparisons in carnivores (Ortolani and Caro, 1996) but not in lagomorphs (Stoner et al., 2003). It seems obvious that deer are spotted to blend in with dappled light (Beddard, 1892; Schaller, 1967; Thayer, 1909), so we were surprised to find few significant associations between species in which adults are spotted and living in forests or grassland. This was undoubtedly owing to shared ancestry because spotted species are closely related, particularly in the cervids. Nonetheless, spotted cervids were diurnal, and spotted bovids were found in light forests (although there are only five species that are spotted). In sum, spotted coats may be an adaptation for concealment based on these two tests, but other predictions concerning habitats and being solitary were not supported. Among carnivores, dark spots were associated with living in closed habitats and having ungulates as the main prey item, suggesting that, in that order, their function is aggressive camouflage (Ortolani, 1999). There was definite support for striped coats aiding in concealment (Cott, 1940; Estes, 1991) given that both bovids and artiodactyls with such coats were more likely to inhabit light forests after controlling for ancestry. Nonetheless, there was no association with being diurnal, living in grassland, or being solitary and possessing a striped coat. In carnivores, however, vertically striped coats were associated with living in grassland and with being terrestrial, bolstering the background-matching hypothesis (Ortolani, 1999). In even-toed ungulates, there was no evidence for spotted or striped coats being involved in intraspecific communication because they were not associated with gregariousness. Many of the predictions regarding spotted coats in young as helping in concealment were well supported (Cott, 1940; Thayer, 1909). Each of the three taxonomic groups was found in forests or grasslands, and solitary artiodactyls were particularly likely to have spotted young (see Danilkin, 1996). Extraordinarily, every one of the 30 species with spotted young exhibited a hiding strategy, demonstrating a tight link between this morphological character and this type of antipredator behavior. Mothers of species that have spotted young all leave their neonates alone when they forage. Support for species with spotted young using the follower strategy (Macdonald, 1984) was lacking. Striped coats in young also aid in concealment. Bovids with striped young were found in lightly forested habitats. Species whose young had striped coats were significantly more likely to be hiders not followers, but based only on crossspecies comparisons. Thus, the relationship between pattern blending in spotted young and the hiding strategy was Behavioral Ecology Vol. 14 No. 6 836 Table 4 Degree of support for associations tested in the present study Concealment Coloration Overall body coloration Uniform color Lighter in winter Spotted adults Striped adults Spotted young Striped young Side bands Countershading Body parts Dark face White face Conspicuous face Dark legs White legs Conspicuous legs Dark tail White tail Conspicuous tail White rump a BM Communication PB DC CS Intra Physiological Inter T Yesa None Strong Weak Strong Strong Weak None None None Strong Moderate Weak Strong Weak None Weak Strong None None None Moderate Weak Strong None Weak None Moderate None None None None None BM indicates background matching; PB, pattern blending; DC, disruptive coloration; CS countershading; Intra, intraspecific communication; Inter, interspecific communication; and T, themoregulation. Refers to Gloger’s rule. replicated in striped young although without the same degree of certainty. It is difficult to interpret results regarding predictions about side bands acting as a form of disruptive coloration (Estes, 1991). Although artiodactyls with side bands were diurnal and lived in open environments and in both desert and tundra habitats, and although some of these findings were also found separately in cervids and bovids, none of these results held up after controlling for phylogeny. The clearest result from crossspecies analyses was that artiodactyls with side bands lived in open environments (24 out of 24 species), but this is not a clear-cut expectation from the disruptive coloration hypothesis (e.g., disruptively colored woodcock, Scolopax minor, chicks are found in leaf litter); thus, we cannot be certain that side bands serve a disruptive function. In contrast, artiodactyls with side bands were found in large groups, suggesting that they might be used in intraspecific communication, perhaps allowing individuals to keep in sight of each other. There was a strong association between bovids and artiodactlys that were countershaded and inhabiting deserts. Because countershading is thought to aid in concealment by reducing shadow in well-lit environments (Cott, 1940; Fogden and Fogden, 1974), the hypothesis is partially supported despite no significant associations with other open environments. In contrast to even-toed ungulates, lagomorphs that were countershaded were not more likely to be found in open habitats that included deserts (Stoner et al., 2003), so the widespread importance of countershading in helping to conceal mammals is still questionable. Body parts Focusing on phylogenetically controlled comparisons, darkfaced bovids and artiodactyls were found in intermediate- and large-sized social groups, which implies an intraspecific communicatory function (see Kingdon, 1982; Spinage, 1986). The prediction that dark faces might be used to signal to stalking predators that they have been noticed (see Caro, 1995) was supported only weakly (in a chi-square test on artiodactyls), but in contrast, there were highly significant associations between dark faces and being pursued by coursers, the biological significance of which is mysterious. Because even-toed ungulates are likely to flee from coursers, predators would be unlikely to see their faces. Perhaps the extent of blackness on the face signals ability to outdistance a predator, but this is pure speculation. In carnivores, dark patches around the eyes were associated with crepuscular activity, perhaps serving as a way to reduce glare (Ortolani and Caro, 1996). White faces were associated with being diurnal (all artiodactyls) and living in open environments (cervids and artiodactyls). Species with white faces were found in grassland/bushland habitats in all three clades. If, as we imagine, grassland habitats are sunny environments with relatively little shade, this would lend (albeit weak) support to a thermoregulatory role in which white faces are used to lower temperature, as suggested by Geist (1987). There was weak support for white faces being used in intraspecific communication based on patterns of grouping but no support for signaling to predators (see Poulton, 1890). Similarly, there was very little widespread support for conspicuous faces being used in communication. We found weak support for dark legs acting as disruptive coloration (see Roosevelt and Heller, 1914) because artiodactyls with dark legs were found in deserts. Phylogenetically controlled comparisons gave some support to a role for dark legs being used in intraspecific signaling, but only for artiodactyls and living in large groups. Hypotheses suggesting white legs might be an aid in communication (Danilkin, 1996; Geist, 1987; Kingdon 1982) received weak support. Conspicuous leg markings were more obviously associated with Stoner et al. • Color of artiodactyls diurnal communication in open habitats, but the type of communication, to predators or conspecifics, was unknown. Conventional hypotheses that dark tails are involved in intraspecific communication (Kingdon, 1982) or signal to predators were supported only in nonparametric tests. Backing for white tails acting in communication (Bildstein, 1983; Kingdon, 1982) was similarly weak, and cross taxonomic analyses did do not support the suggestion, based on single species studies, that white tails are used to signal to predators (for a review, see Caro et al., 1995). However, our examination of species with conspicuous tail markings (be they dark or white) revealed that these species are more likely to be diurnal and gregarious than those without distinct tail markings. Thus, conspicuous tails seem to be involved with intraspecific communication in even-toed ungulates. In lagomorphs, contrasting tail tips were similarly associated with sociality (Stoner et al., 2003) although not in carnivores (Ortolani and Caro, 1996), making it difficult to form general conclusions about mammals. White rumps might either act as signals to conspecifics (Guthrie, 1971; Ward, 1979) or to predators (Caro, 1986; Fitzgibbon and Fanshawe, 1988), or be used in thermoregulation (Spinage, 1986; Estes, 1991). White-rumped species in all three ungulate clades were found in species inhabiting open environments and, after controlling for phylogeny, in intermediate-sized social groups, lending weight to intraspecific communication. There was some evidence of signaling to predators because white-rumped species were pursued by coursers, although not in concentrated changes tests. That white rumps were found in the open, in deserts, and in grassland/bushland habitats lends credence to the thermoregulation hypothesis, but we did not replicate all these results by using phylogenetic controls. The function of white rumps remains an enigma, although intraspecific communication and thermoregulation both remain promising candidates. 837 adaptive significance of coloration in animals and that they elevate the study of coloration from fascinating anecdotes to testable hypotheses about a wide array of species. APPENDIX Sources used to obtain morphological and ecological information for each species are as follows: Bauer, 1995, Chadwick, 1983; Chapman and Chapman, 1997; Clark, 1994; Cloudsley-Thompson, 1980; Dagg and Bristol-Foster, 1976; Danilkin, 1996; Estes, 1991, 1993, Grzimek, 1990; Heptner et al., 1989; Hoefs, 1985; Holmes, 1974; Huntingford and Turner, 1987; Johnson and Lockard, 1983; Jones et al., 1985; Kingdon, 1982; Lindsey et al., 1999; Macdonald, 1984; Mungall and Sheffield, 1994; Nabhan, 1993; Nowak, 1999; Miura et al., 1993; Oliver, 1993; Payne and Francis, 1985; Putman, 1988; Prior, 1987; Schaller, 1977, 1998; Shackleton, 1999; Shrestha, 1997; Soma, 1987; Sowls, 1984, 1997; Spinage, 1968, 1982; Spitz, 1991; Strahan, 1983; Stuart and Stuart, 1993, 1997; Valdez, 1982; Valdez and Krausman, 1999; Van Wormer, 1969; Walther, 1984; Walther et al., 1983; Wemmer, 1987; Whitehead, 1972, 1993; Wilson, 2000. We thank the University of California for supporting C.J.S. and the McNair Scholars Program for supporting C.M.G., Kate Trimlett and Janelle Vargas for help with coding variables, Marisa Flores for running supplementary analyses, John Eadie and Charlie Nunn for discussion, John Byers for drawing our attention to the possibility of white rumps being used in thermoregulation, Jerry Wolff for suggesting we examine conspicuous coloration patterns and more behavior, and Dave Westneat, Jerry Wolff, and an anonymous reviewer for very helpful comments. REFERENCES Conclusion We uncovered strong support for the following color patterns serving to conceal even-toed ungulates: lightening in winter, striped coats in adults, and spotted coats in young. We found some support for a communicatory role of color patterns, that is, side bands, dark faces, conspicuous tails, and white rumps. Finally, we found only moderate support for a color pattern acting in a thermoregulatory capacity, notably dark coats and white faces. We made no attempt to explore the role of mate choice in driving coloration patterns, as no specific hypothesis pertains to sexual selection in mammals, despite certain species, including some ungulates, being sexually dichromatic. Our clearest findings were concerned with hypotheses about concealment, particularly background matching and pattern blending, and our less conclusive findings tended to concern issues of communication and thermoregulation. In part, it is easier to formulate clear-cut predictions about habitat characteristics than about social or predator variables when testing these hypotheses about communication, and it is not always clear as to the way in which different habitats might contribute to heat stress (e.g., humid rain forests or parched deserts). Also, we suspect that several social variables may be intercorrelated (e.g., follower and social species, being social and being pursued by coursers), and this will inevitably confound findings that rely on behavioral variables to test predictions. That said, our findings suggest that some coloration patterns in ungulates probably have different functions in different species, and this is why the comparative data set sometimes failed to support hypotheses based on studies of one or two species. Despite these problems, we nonetheless feel that systematic comparative analyses are key to making strong and supportable generalizations about the Alvarez F, Braza F, Norzagaray A, 1976. The use of the rump patch in the fallow deer (D. dama). Behaviour 56:298–308. Amato G, Egan MG, Schaller GB, 2000. Mitochondrial DNA variation in Muntjac: evidence for discovery, rediscovery, and phylogenetic relationships. In: Antelopes, deer, and relatives: fossil record, behavioral ecology, systematics, and conservation (Vrba ES, Schaller GB, eds). New Haven, Connecticut: Yale University Press; 285–295. Badyaev AV, Hill GE, 2000. Evolution of sexual dichromatism: contribution of carotenoid- versus melanin-based coloration. Biol J Linn Soc 69:153–172. Baker RR, Parker GA, 1979. The evolution of bird coloration. Phil Trans R Soc Lond B 287:63–130. Bauer EA, 1995. Elk: Behavior, ecology and conservation. Stillwater, Minnesota: Voyageur Press. Beddard FE, 1892. Animal colouration. London: Hazell, Watson and Viney. Bildstein KL, 1983. Why white-tailed deer flag their tails. Am Nat 121:709–715. Brasheres JS, Garland T Jr, Arcese P, 2000. Phylogenetic analysis of coadaptation in behavior, diet, and body size in the African antelope. Behav Ecol 11:452–463. Brooke M de L, 1998. Ecological factors influencing the occurrence of ‘‘flash marks’’ in wading birds. Funct Ecol 12:339–346. Burtt EH Jr, 1979. Tips on wings and other things. In: Behavioural significance of color (Burtt EH Jr, ed). New York: Garland STPM Press; 75–110. Buxton PA, 1923. Animal life in deserts: a study of the fauna in relation to the environment. London: Edward Arnold. Caro TM, 1986. The functions of stotting in Thomson’s gazelles (Gazella thomsoni): some tests of the predictions. Anim Behav 34:663–684. Caro TM, 1994. Ungulate antipredator behavior: preliminary and comparative data from African bovids. Behaviour 128:189–228. Caro TM, 1995. Pursuit-deterrence revisted. Trends Ecol Evol 10: 500–503. 838 Caro TM, Lombardo SL, Goldizen AW, Kelly MJ, 1995. Tail-flagging and other antipredator signals in white-tailed deer: new data and synthesis. Behav Ecol 6:442–450. Chadwick DH, 1983. A beast the color of winter: the mountain goat observed. San Francisco: Sierra Club Books. Chapman D, Chapman N, 1997. Fallow deer: their history, distribution, and biology. Machynlleth: Coch-y-bonddu Books. Clark JL, 1994. The great ark of the wild sheep. Norman: University of Oklahoma Press. Cloudsley-Thompson JL, 1980. Tooth and claw: defensive strategies in the animal world. London: M. Dent. Cooper W Jr, 2002. Convergent evolution of plant chemical discrimination by omnivorous and herbivorous scleroglossan lizards. J Zool 257:53–66. Cooper WE Jr, Caldwell JP, Vitt LJ, Perez-Mellado V, Baird TA, 2002. Food-chemical discrimination and correlated evolution between plant diet and plant-chemical discrimination in lacertiform lizards. Can J Zool 80:655–663. Corbet GB, Hill JE, 1991. A world list of mammalian species, 3rd ed. London: British Museum (Natural History). Corbet GB, Hill JE, 1992. The mammals of the Indomalayan region: a systematic review. Oxford: Oxford University Press. Cott HB, 1940. Adaptive colouration in animals. London: Methuen. Cronin MA, Stuart R, Pierson BJ, Patton JC, 1996. K-Casein gene phylogeny of higher ruminants (Pecora, Artiodactyla). Mol Phylo Evol 6:295–311. Dagg AI, Bristol-Foster J, 1976. The giraffe: its biology, behavior and ecology. New York: Van Nostrand Reinhold. Danilkin A, 1996. Behavioural ecology of Siberian and European roe deer. London: Chapman Hall. Douzery E, Randi E, 1997. The mitochondrial control region of Cervidae: evolutionary patterns and phylogenetic content. Mol Biol Evol 14:1154–1166. Emerson BC, Tate ML, 1993. Genetic analysis of evolutionary relationships among deer (subfamily Cervinae). J Hered 84:266– 273. Endler JA, 1978. A predator’s view of animal colour patterns. Evol Biol 11:319–364. Endler JA, 1993. The color of light in forest and its implications. Ecol Monogr 63:1–27. Estes RD, 1991. The behaviour guide to African mammals. Los Angeles: University of California Press. Estes RD, 1993. The safari companion: a guide to watching African mammals. Post Mills, Vermont: Chelsea Green. Fitzgibbon CD, Fanshawe JH, 1988. Stotting in Thomson’s gazelles: an honest signal of condition. Behav Ecol Sociobiol 23:69–74. Fogden M, Fogden P, 1974. Animals and their colors. New York: Crown. Franklin WL, 1982. Biology, ecology and relationship to man of the South American camelids. Pymatuning Lab Ecol Spec Pub 6:457– 489. Gatesy J, Arctander P, 2000. Hidden morphological support for the phylogenetic placement of Pseudoryx nghinhensis with bovine bovids: a combined analysis of gross anatomical evidence and DNA sequences from five genes. Syst Biol 49:515–538. Gatesy J, O’Grady P, Baker RH, 1999. Corroboration among data sets in simultaneous analysis: hidden support for phylogenetic relationships among higher level artiodactyl taxa. Cladistics 15:271–313. Geist V, 1978. Life strategies, human evolution, environmental design: toward a biological theory of health. New York: Springer-Verlag. Geist V, 1987. On the evolution of optical signals in deer: a preliminary analysis. In: Biology and management of the Cervidae (Wemmer C, ed). Washington, DC: Smithsonian Institute Press; 235–255. Groves CP, 1981. Ancestors for the pigs: taxonomy and phylogeny of the genus Sus, technical bulletin no. 3. Canberra, Australia: Department of Prehistory, Research School of Pacific Studies, Australian National University. Groves CP, Schaller GB, 2000. The phylogeny and biogeography of the newly discovered Annamite artiodactyls. In: Antelopes, deer, and relatives: fossil record, behavioral ecology, systematics, and conservation (Vrba ES, Schaller GB, eds). New Haven, Connecticut: Yale University Press; 261–282. Behavioral Ecology Vol. 14 No. 6 Grubb P, 1993. Order Artiodactyla. In: Mammal species of the world (Wilson DE, Reeder DM, eds). Washington, DC: Smithsonian Institute Press; 377–414. Grzimek B, 1990. Grzimek’s encyclopedia of mammals. New York: McGraw-Hill. Guilford TC, 1990. The evolution of aposematism. In: Insect defences: adaptive mechanisms and strategies of prey and predators (Evans DL, Schmidt JO, eds). Albany: State University of New York Press; 23–61. Guthrie RD, 1971. A new theory of rump patch evolution. Behaviour 38:132–145. Guthrie RD, Petocz RG, 1970. Weapon automimicry among mammals. Am Nat 104:585–588. Hamilton WJ III, 1973. Life’s color code. New York: McGraw-Hill. Harvey PH, Pagel MD, 1991. The comparative method in evolutionary biology. Oxford: Oxford University Press. Hassanin A, Douzery EJP, 1999a. Evolutionary affinities of the enigmatic saola (Pseudoryx nghetinhensis) in the context of the molecular phylogeny of Bovidae. Proc R Soc B 266:893–900. Hassanin A, Douzery EJP, 1999b. The tribal radiation of the family Bovidae (Artiodactyla) and the evolution of mitochondrial cytochrome b gene. Mol Phylo Evol 13:227–243. Hassanin A, Pasquet E, Vigne JD, 1998. Molecular systematics of the subfamily Caprinae (Artiodactyla, Bovidae) as determined from cytochrome b sequences. J Mammal Evol 5:217. Hasson O, 1986. Predator decision-making and pursuit-deterrence signals. Anim Behav 35:45–54. Heptner VG, Nasimovich AA, Bannikov AG, 1989. Mammals of the Soviet Union (Hoffman RS, ed). New York: Brill. Hickman GC, 1979. The mammalian tail: a review of functions. Mammal Rev 9:143–157. Hiendleder S, Mainz K, Plante Y, Lewalski H, 1998. Analysis of mitochondrial DNA indicates that domestic sheep are derived from two different ancestral maternal sources: no evidence for contributions from urial and argali sheep. J Hered 89: 113–120. Hoefs M, 1985. Wild sheep: distribution, abundance, management and conservation of the sheep of the world and closely related mountain ungulates. Whitehorse: Northern Wild Sheep and Goat Council. Hoglund J, 1989. Size and plumage dimorphism in lek-breeding birds: a comparative analysis. Am Nat 134:72–87. Holmes F, 1974. Following the roe: a natural history of the roe deer. Edinburgh: J. Bartholomew. Huntingford F, Turner AK, 1987. Animal conflict. New York: Chapman and Hall. Irschick DJ, Vilt LJ, Zary PA, Losos JB, 1997. A comparison of evolutionary radiations in mainland and Caribbean anolis lizards. Ecology 78:2191–2203. Johnson CE, 1921. The hand-stand habit of the spotted skunk. J Mammal 2:87–89. Johnson RL, Lockard FR, 1983. Mountain goats and mountain sheep of Washington, biological bulletin no. 18. Olympia: Washington Department of Game. Jones JK, Armstrong DM, Choate JR, 1985. Guide to mammals of the plains states. Lincoln: University of Nebraska Press. Kingdon J, 1982. East African mammals: bovids, vol. 3. New York: Academic Press. Kingdon J, 1997. The Kingdon field guide to African mammals. New York: Academic Press. Kiltie RA, 1988. Countershading: universally deceptive or deceptively universal? Trends Ecol Evol 3:21–23. Kiltie RA, 1989. Testing Thayer’s countershading hypothesis: an image processing approach. Anim Behav 38:542–552. Leyhausen P, 1979. Cat behavior. New York: Garland STPM Press. Lindsey SL, Green ML, Bennett CL, 1999. The okapi: mysterious animal of Congo-Zaire. Austin: University of Texas Press. Linnell JDC, Andersen R, 1998. Timing and synchrony of birth in a hider species, the roe deer (Capreolus capreolus). J Zool 244:497– 504. Lobao Tello JLP, van Gelder RG, 1975. The natural history of nyala Tragelaphus angasi (Mammalia, Bovidae) in Mozambique. Am Mus Nat Hist 155:323–386. Ludwig A, Fisher S, 1998. New aspects of an old discussion: phylogenetic relationships of Ammotragus and Pseudois within Stoner et al. • Color of artiodactyls the subfamily Caprine based on comparison of the 12S rDNA sequences. J Zool Syst Evol Res 36:173–178. Ludwig A, Knoll J, 1998. Multivariate morphometrische analysen der gattund ovis linnaeus, 1758 (Mammalia, Caprinae). Z Saugetierkd 63:210–219. MacDonald D, 1984. The encyclopedia of mammals, vol. 2. London: George, Allen and Unwin. Maddison WP, 1990. A method for testing the correlated evolution of two binary characters: are gains and losses concentrated on certain branches of the phylogenetic tree? Evolution 44:539–557. Maddison WP, Maddison DR, 1992. MacClade: analysis of phylogeny and character evolution. Sunderland, Massachusetts: Sinauer Associates. Manceau V, Despres JB, Taberlet P, 1999. Systematics of the genus Capra inferred from mitochondrial DNA sequence data. Mol Phylo Evol 13:504–510. Mathee CA, Robinson TJ, 1999. Cytochrome b phylogeny of the family Bovidae: resolution within the Alcelaphini, Antilopini, Neotragini, and Tragelaphini. Mol Phylo Evol 12:31–46. Meyer W, Kranzle S, 1999. Zur Bedeutung von Farburg und Zeichnung der Korperpole (Kopf und Analregion) bei Saugetieren. Zool Garten NF 3:159–187. Miura S, Kaji K, Ohtaishi N, Koizumi T, Tokida K, Wu J, 1993. Social organization and mating behavior of the white-lipped deer in the Qinghai-Xizang Plateau, China. In: Deer of China: biology and management (Ohtaishi N, Sheng HI, eds). Amsterdam: Elsevier; 220–234. Mottram JC, 1915. Some observations on pattern-blending with reference to obliterative shading and concealment of outline. Proc Zool Soc Lond 1915:679–692. Mottram JC, 1916. An experimental determination of the factors which cause patterns to appear conspicuous in nature. Proc Zool Soc Lond 1915:383–419. Mungall EC, Sheffield WJ, 1994. Exotics on the range: the Texas example. College Station: Texas A&M University Press. Nabhan GP, 1993. Counting sheep: twenty ways of seeing desert bighorn (Nabhan GP, ed). Tucson: University of Arizona Press. Nee S, Read AF, Harvey PH, 1996. Why phylogenies are necessary for comparative analysis. In: Phylogenies and the comparative method in animal behavior (Martins EP, ed). Oxford: Oxford University Press; 399–411. Nowak RM, 1999. Walker’s mammals of the world, 6th ed. Baltimore, Maryland: Johns Hopkins University Press. Oliver WLR, 1993. Pigs, peccaries, and hippos: status survey and conservation action plan. Gland: IUCN. Ortolani A, 1999. Spots, stripes, tail tips and dark eyes: predicting the function of carnivore colour patterns using the comparative method. Biol J Linn Soc 67:433–476. Ortolani A, Caro TM, 1996. The adaptive significance of color patterns in carnivores: phylogenetic tests of classic hypotheses. In: Carnivore behavior, ecology, and evolution, vol. 2. (Gittleman J, ed). Ithaca, New York: Comstock; 132–188. Pagel M, 1994. Detecting correlated evolution on phylogenies: a general method for the comparative analysis of discrete characters. Proc R Soc Lond Ser B 255:37–45. Payne J, Francis CM, 1985. A field guide to the mammals of Borneo. Kuala Lumpur: World Wildlife Fund Malaysia. Pérez-Barberı́a FJ, Gordon IJ, 1999. The functional relationship between feeding type and jaw and cranial morphology in ungulates. Oecologia 118:157–165. Pérez-Barberı́a FJ, Gordon IJ, 2000. Differences in body mass and oral morphology between the sexes in the Artiodactyla: evolutionary relationships with sexual segregation. Evol Ecol Res 2:667–684. Pérez-Barberı́a FJ, Gordon IJ, 2001. Relationships between oral morphology and feeding style in the Ungulata: a phylogenetically controlled evaluation: Proc R Soc Lond B 268:1021–1030. Pérez-Barberı́a FJ, Gordon IJ, Pagel M, 2002. The origins of sexual dimorphism in body size in ungulates. Evolution 56:1276–1285. Poulton EB, 1890. The colours of animals. London: Kegan Paul, Trench, Trubner. Price T, 1997. Correlated evolution and independent contrasts. Phil Trans R Soc B 352:519–528. Prior R, 1987. Deer watch. London: David and Charles. Putman RJ, 1988. The natural history of deer. London: Christopher Helm. 839 Randi E, Lucchini V, Diong CH, 1996. Evolutionary genetics of the Suiformes as reconstructed using mtDNA sequencing. J Mammal Evol 3:163. Randi E, Mucchi N, Claro-Hergueta F, Amelie B, Douzery EJP, 2001. A mitochondrial DNA control region phylogeny of the Cervinae: speciation in Cervus and implications for conservation. Anim Conserv 4:1–11. Randi E, Mucchi N, Pierpaoli M, Douzery EJP, 1998. New phylogenetic perspectives on the Cervidae (Artiodactyla) are provided by the mitochondrial cytochrome b gene. Proc R Soc Lond B 265: 793–801. Read AF, Nee S. 1995. Inference from binary comparative data. J Theor Biol 173:99–108. Rebholz W, Harley E, 1999. Phylogenetic relationships in the bovid subfamily Antilopinae based on mitochondrial DNA sequences. Mol Phylo Evol 12:87–94. Rice WR, 1989. Analyzing tables of statistical tests. Evolution 43:223– 225. Ridley M, 1983. The explanation of organic diversity: the comparative method and adaptations for mating. Oxford: Oxford University Press. Roosevelt T, 1911. Revealing and concealing coloration in birds and mammals. Bull Am Mus Nat Hist 30:119–231. Roosevelt T, Heller E, 1914. Life histories of African game animals. New York: Scribner’s. Rowland WJ, 1979. The use of colour in intraspecific communication. In: The behavioral significance of color (Burtt EH Jr, ed). New York: Garland STPM Press; 379–421. Schaller GB, 1967. The deer and the tiger. Chicago, Illinois: University of Chicago Press. Schaller GB, 1977. Mountain monarchs: wild sheep and goats of the Himalaya. Chicago, Illinois: University of Chicago Press. Schaller GB, 1998. Wildlife of the Tibetan steppe. Chicago, Illinois: University of Chicago Press. Schreiber A, Seibold E, Notzold G, Wink M, 1999. Cytochrome b gene haplotypes characterize chromosomal lineages of Anoa, the Sulawesi dwarf buffalo (Bovidae: Bubalus sp.). Am Genet Assoc 90:165–176. Shackleton DM, 1999. Hoofed mammals of British Columbia. Vancouver: UBC Press. Shrestha TK, 1997. Mammals of Nepal: with reference to those of India, Bangladesh, Bhutan, and Pakistan. Kathmandu: Bimala Shrestha. Sillen-Tullberg B, 1993. The effect of biased inclusion of taxa on the correlation between discrete characters in phylogenetic trees. Evolution 47:1182–1191. Soma H, 1987. The biology and management of Capricornis and related mountain antelopes (Soma H, ed). New York: Croom Helm. Sowls LK, 1984. The peccaries. Tucson: University of Arizona Press. Sowls LK, 1997. Javelinas and other peccaries: their biology, management and use. College Station: Texas A&M University Press. Spinage CA, 1968. The book of the giraffe. London: Collins. Spinage CA, 1982. A territorial antelope: the Uganda waterbuck. New York: Academic Press. Spinage CA, 1986. The natural history of antelopes. Beckenham, Kent: Croom Helm Publishers. Spitz F, 1991. Proceedings of the International Symposium ‘‘Ongules/ Ungulates 91,’’ Toulouse, France: 2–6 September 1991 (Spitz F, ed). Toulouse: SFEPM-IRGM. Stoner CJ, Bininda-Emonds ORP, Caro TM, 2003. The adaptive significance of coloration in lagomorphs. Biol J Linn Soc (79:309– 328). Strahan R, 1983. Complete book of Australian mammals: the national photographic index of Australian wildlife. Sidney: Angus & Robertson. Stuart C, Stuart T, 1993. Field guide to the mammals of southern Africa, 2nd ed. Cape Town: Struik. Stuart C, Stuart T, 1997. Field guide to the larger mammals of Africa. Cape Town: Struik. Su B, Wang YX, Lan H, Wang W, Zhang Y, 1999. Phylogenetic study of complete cytochrome b genes in musk deer (genus Moschus) using museum samples. Mol Phylo Evol 12:241–249. Thayer AG, 1909. Concealing coloration in the animal kingdom. New York: Macmillan. 840 Thayer AH, 1896. The law which underlies protective coloration. Auk 13:124–129. Theimer TC, Keim P, 1998. Phylogenetic relationships of peccaries based on mitochondrial cytochrome b DNA sequences. J Mammal 79:566–572. Thenius E, 1970. Zur evolution und verbreitungsgeschichte der Suidae (Artiodactyla, Mammalia). Z Saugetierkd 35:321–342. Valdez R, 1982. The wild sheep of the world. Mesilla: Wild Sheep and Goat International. Valdez R, Krausman PR, 1999. Mountain sheep of North America. Tucson: University of Arizona Press. Van Wormer J, 1969. The world of the American Elk. Philadelphia, Pennsylvania: Lippincott. Wallace AR, 1889. Darwinism. London: Macmillan. Walther FR, 1984. Communication and expression in hoofed animals. Bloomington: Indiana University Press. Walther FR, Mungall EC, Grau GA, 1983. Gazelles and their relatives: a study in territorial behavior. Westwood, New Jersey: Noyes. Behavioral Ecology Vol. 14 No. 6 Ward P, 1979. Colour for survival. London: Orbis. Webb SD, 2000. Evolutionary history of new world Cervidae. In: Antelopes, deer, and relatives: fossil record, behavioral ecology, systematics, and conservation (Vrba ES, Schaller GB, eds). New Haven, Connecticut: Yale University Press; 38–64. Wemmer C, 1987. Biology and management of the Cervidae. In: Research symposia of the National Zoological Park, Front Royal, Virginia: 1–5 August 1982 (Wemmer C, ed). Washington, DC: Smithsonian Institute Press; 577. Whitehead GK, 1972. Deer of the world. London: Constable. Whitehead GK, 1993. The Whitehead encyclopedia of deer. Stillwater, Minnesota: Voyageur Press. Wickler W, 1968. Mimicry in plants and animals. London: Weidenfeld and Nicholson. Wilson EO, 2000. Sociobiology: the new synthesis. Cambridge, Massachusetts: Harvard University Press.
© Copyright 2025 Paperzz