ALLOMETRIC SCALING OF BODY LENGTH: ELASTIC OR GEOMETRIC SIMILARITY IN MAMMALIAN DESIGN MARINA SILVA Department of Biology, University of Prince Edward Island, 550 University Avenue, Charlottetown, Prince Edward Island CiA 4P3, Canada Allometric scaling of body length was examined for 1.733 estimates of body length for a broad taxonomic and size range of mammals. Mammal species were classified by locomotion, habitat, and taxonomic relatedness. Scaling exponents found in different groups of mammals were compared to those predicted by geometric (0.333) and elastic (0.250) similitude models. The scaling exponent for the length: mass relationship 0.359 agreed better with the geometric-similitude model. However, the relationship between body length and body mass was not linear as previously postulated, and differences in the scaling exponent were found between marine and terrestrial mammals and between volant and non-volant forms. Allometric scaling of body length also varied among orders and families of mammals. Key words: length, mass, mammals, scaling, morphology Several studies have examined the relationship between body length and body mass and suggested that this may be an important tool for the investigation of ecology, biology and physiology in organisms (Economos, 1983; Eisenberg, 1981; Niklas. 1994; Wolff and Guthrie, 1985). The allometric exponent of the relationship between body length and body mass has been used to explain morphological design of organisms by relating that design to the elastic or geometric model (Alexander, 1988; Economos, 1983; Pollock and Shadwick, 1994: )Vainwright et aI., 1982). According to the elastic model, trunk diameter increases faster than body length, so that body length scales to the 0.250 power of body mass. Alternatively, the geometric model assumes that trunk diameter increases isometrically with body length, such that the value of the exponent of the relationship between body length and body mass is ca. 0.333. Although some studies have dealt with the relationship between body length and body mass in mammals (Calder, 1984; Jerison, 1970: Radinsky, 1978), few of them (Cristoffer, 1991; Economos, 1983; Niklas, Journal of Mamrnalogy, 79(1):20-32. 1998 1994) investigated the importance of the scaling exponent as a tool for the examination of morphology and physiology in mammals. Furthermore, the few studies that did address the relationship between body length and body mass generally were based on small data sets or were restricted to one or few taxonomic orders (Economos, 1983; Jerison, 1970; Radinsky, 1978: Wolff and Guthrie, 1985). Most studies show that mammals, except the Bovidae, conform better with the geometric-similitude model than with the elastic one (Alexander, 1977, 1988; McMahon, 1973). However, several studies suggested that terrestrial mammals should conform better to the elastic model due to gravity (Alexander, 1977, 1988; Economos, 1983; McMahon, 1973). If this is true, aquatic mammals should deviate less from the geometric model than terrestrial forms because they live in a weightfree environment. Yet, for marine mammals, Economos (1983) showed that the scaling exponent of the length: mass relationship was closer to the exponent predicted by the geometric similarity model. Economos' study (1983) was, however, 20 February 1998 21 SILVA-SCALING OF BODY LENGTH based on a small data set of marine mammals from the order Cetacea. Marine mammals include two other mammalian taxonomic orders, Pinnipedia and Sirenia. Unlike cetaceans, pinnipeds inhabit both terrestrial and aquatic environments. It is unknown, however, if pinnipeds show a scaling exponent closer to the one of cetaceans or to the one of terrestrial mammals. Economos (1983) also suggested that the relationship between mammalian body mass and body length may be non-linear. Econom~s noted that phyletic increase in body mass of terrestrial mammals was accompanied initially by an increase in linear dimensions according to the principle of geometric similarity because gravitational loading does not present a structural problem. Therefore, at a certain body mass, we should expect to find a reduction of the scaling exponent of the length: mass relationship. Although Economos (1983) did not provide any statistics, he proposed a threshold for the change in the exponent at ca. 20 kg. If this is valid, I should expect to find a significant decrease in the slope of the length: mass relationship around this body mass. Furthermore, study of taxonomic deviations in the allometric exponent of the length: mass relationship from the global trend may be a powerful tool in the investigation of mammalian adaptations. However, little is known about the relationship between body length and body mass across and within taxonomic groups in mammals. For example, Eisenberg (1981) noted that volant mammals are smaller than non-volant forms. This adaptation may allow bats to minimize wing loading, which is basically body mass (length cubed) divided by area (length squared). Animals with low wing loading may be better able to carry more mass, and thus, a scaling exponent significantly smaller than the one expected from geometric-similitude model might be selectively advantageous for volant mammals considering aerodynamic consequences of flight (Cristoffer, 1991; Hayssen and Kunz, 1996). In addition, if selection for a reduction in body mass in the Chiroptera increases progressively with increasing body length (Eisenberg, 1981), then the scaling exponent of the length: mass relationship in bats would not only be steeper than the one of non-volant forms, but it also should differ significantly between small and large bats. No analyses across species were conducted to examine these predictions, although Cristoffer (1991) examined rodents and bats from both Nigeria and Thailand and found that large bats are less massive than rodents of comparable body length. Other differences with respect to the scaling exponent across mammalian orders and within them also can be predicted when one considers ecology or behavior. For example, Wainwright et al. (1982) suggested that mammals like cats (Carnivora: Felidae) and gazelles (Artiodactyla: Bovidae) reduce their mass to outrun other mammals of similar body length. This predicts scaling exponents significantly steeper than the one expected from the geometric-similitude model in these families. My purpose was to examine scaling of body length in mammals and provide equations that can be used for predicting body length. In the context of these scaling exponents, I compared mammals that use different means of locomotion and belong to different taxonomic groups. METHODS Data on body mass and body length were obtained from the literature (Anderson et al., 1993; Ceballos and Miranda, 1986; Eisenberg, 1989; Emmons. 1990; Hufnagl, 1972; Husson. 1978; Hutterer, 1986; Mares et al., 1989; Marshall, 1967; Nowak, 1991; Straham, 1983). Trunk diameter was not used because those values were rarely reported in the literature. Body mass (g) and body length (rom) were mean values for adults; when the mean values were unavailable, I used the midpoint of ranges. Body length was length of the head and body (Le.. excluding length of tail); for cetaceans, total length of the body was used. Only a small number of sources 22 Vol. 79. No.1 JOURNAL OF MAMMALOGY reported the method used to measure body length. Therefore, bias associated with the use of different methods to measure body length was difficult to evaluate accurately, although I suspect it was negligible due to strong correlations found in this study. Mammals were grouped by taxon, habitat, and means of locomotion. Comparisons were performed by classifying mammals into the following categories: marine (Cetacea, Pinnipedia, and Sirenia); wholly-marine (live strictly in the sea: Cetacea and Sirenia); and terrestrial. Terrestrial mammals were sorted into volant (Chiroptera) and nonvolant fonns. Taxonomic analyses were performed for both order and family following the classification of Corbet and Hill (1991). All variables were transformed logarithmically to stabilize variance, linearize responses, and normalize residuals. When detransformation from the logarithmic to the original scale were necessary, I corrected values following Zar (1996). Model II regressions were thought to be most appropriate for determining scaling exponents (La Barbera, 1989; Martin and Barbour, 1989); thus, I employed major axis regression (Jolicoeur, 1990) in this analysis. Although there are other model II regression methods, in this case major-axis regression was the most appropriate because both body length and body mass were log-transformed (Jolicoeur, 1990; Mazer and Wheelwright, 1993; Mesple et a1., 1996). Regression analyses were performed for taxonomic groups (orders or families) represented by at least six species. Thus, for several orders the sum of their respective familial samples for which regression was perfonned was not equal to the sample size when all species of the order were pooled. Non-linearity of the relationship between body length and body mass was determined using locally weighted sequential smoothing (LOWESS; Cleveland, 1979), a model-free method for detennining the unbiased fonn of the relationship between two variables, while guarding against the influence of deviant points (Cleveland and McGill, 1985). LOWESS was developed by Tokey (Cleveland and McGill, 1985) and is designed to be robust to variations in distribution and provide an unbiased indication of trends in large refractory data clouds. I used polynomial regression analysis to test for statistical significance of nonlinearities indicated by LOWESS (sensu Silva and Downing, 1995). 4.5 ~ E E ~ - , " 4.0 3.5 .c C> c 3.0 ~ 2.5 ~ "0 0 <D ~ 2.0 C> 2 1.5 1.0 -4 -3 -2 -1 0 1 2 3 4 5 log10 Body Mass (kg) - 4.5 ~ 3.5 .c b C> 4.0 C ~ "0 0 <D 3.0 "0 0 •>0 2.' 0 2.0 • .D o ::r 1.5 r2 '" 0.98 P < 0.0001 1.0 ':--:-'::--:":-----::':---:':--:::--:":-.....-:-' 1.0 1.5 2.0 2.5 3.0 3.5 •. 0 •. 5 log10 Estimated Body length (mm) FIG. 1.-(a) Relationship between logarithm of body length (nun) and logarithm of body mass (kg) of all mammals and (b) between predicted body length and observed body length using the all-mammals scaling fonnula (Table I) based on 1,733 estimates of body length and body mass. RESULTS I collected 1,733 published estimates of body length and body mass. Body length varied from 20 mm in the southern pygmy mouse (Baiomys musculus) to 15,200 mm in the spenn whale (Physeter catodon). Body mass varied from 0.002 kg in Kitti's hog-nosed bat (Craseonycteris thonglogyai) to 42,500 kg in the spenn whale (Fig. la). February 1998 SILVA-SCALING OF BODY LENGTH 23 TABLE l.-Major axis regression analyses of the relationships between the logarithm of body length (L, mm) and the logarithm of body mass (M, kg) in mammals. Group Slope (el) 0.359 0.340 0.316 0.357 0.317 0.334 All mammals Marine mammals Wholly-marine mammals Terrestrial mammals Volant mammals Non-volant mammals (0.357-0.362) (0.311--0.369) (0.275--0.358) (0.354--0.360) (0.304-0.330) (0.331-0.338) Intercept nO 2.501 2.587 2.717 2.498 2.369 2518 1,733 85 34 1,648 479 1,169 • Number of observations included in the analysis. Pearson's coefficient of correlation; all correlations are significant at b The regression slope of the log-log relationship between body length and body mass for all mammals pooled was 0.359 (Table 1). The confidence interval of this value did not include expected values of 0.250 and 0.333, corresponding to the elastic- and the geometric-similitude models, respectively. Correlation between body length and body mass was extremely strong suggesting that body mass was a good pre- dictor of variation in body length (Table I). The slope of the relationship between predicted body length and observed body length was not significantly different from I (slope = 0.998; Fig. Ib) showing that the all-mammals equation (Table 1) may be used for predicting body length. The LOWESS analysis revealed that the '.5 ~ E E - 4.0 3.5 ""'" 3.0 "- 2.5 C ~ "0 0 <D 2.0 ~ '" ~ 1.5 - -, 1.0 -3 -2 -1 0 Lowess Malor Qxls 2 3 4 5 10910 Body Mass (kg) FIG. 2.-The log-log linear fit (solid line) to the 1,733 estimates bf body length and body mass of mammalian species shown in Fig, 1; the dotted line represents LOWESS. P = '" 0.99 0.93 0.94 0.99 0.92 0.99 Average log L Average log M 2.365 3.400 3.567 2.312 1.803 2.520 -0.379 2.427 2.723 -0.523 -1.818 0.007 0.001. log-log relationship between body length and body mass was distinctly nonlinear (Fig. 2). Polynomial regression analysis showed that the nonlinearity indicated by LOWESS was significant (P < 0.0001). The allometric exponent behaved differently over ranges of body mass from 0.002 to 0.1 kg, 0.1-100 kg, and 100-42,500 kg. Separate scaling exponents were 0.410, 0.346 and 0.344, respectively. Confidence intervals of the regression slopes suggested that small manunals differ significantly from intermediate and large mammals (P < 0.05), but the difference between intermediate and large mammals is not significant (P > 0.05). Confidence intervals also indicated that scaling values for small and intermediate mammals differ significantly from both the geometric- and the elasticsimilitude models, while the values for large manunals were consistent only with the geometric-similitude model. ANOVA showed that marine mammals were larger than terrestrial mammals (P < 0.05) and mean body mass for the respective groups differ significantly (Table I). However, this may not indicate that the marine environment requires that mammals be larger, because marine mammals do not occur randomly across all taxa but are restricted to three orders. Thus, difference in size could reflect taxon and means of locomotion, instead of habitat. Differences in the scaling exponent also existed between terrestrial and marine mammals (Table 1). When marine mammals (Cetacea, Pinnipe- 24 Vol. 79, No.1 JOURNAL OF MAMMALOGY TABLE 2.~Major axis regression analyses of the relationships between the logarithm of body length (L, mm) and the logarithm of body mass (M, kg) in several mammalian taxonomic orders and families. log L Average logM 2.34 -0.58 Average Taxon Marsupialia Dasyuridae DideJphidae Macropodidae Xenarthra Dasypodidae Mynnecophagidae Insectivora Erinaceidae Soricidac Talpidae Tenrecidae Scandentia Chiroptera Emballonuridae Hipposideridae Mo1ossidae Mormoopidae Noctilionidae Phyllostomidae Pteropodidae VespertiIionidae Primates Callitrichidae Cebidae Ccrcopithecidac Lorisidae Carnivora Canidae Felidae Herpestidae Mustelidae Procyonidae Ursidae Viverridae Pinnipedia Otariidae Phocidae Cetacea Perissodactyla Artiodactyla Bovidae Camelidae Cervidae Suidae Tayassuidae Rodentia Caviidae Ctenomyidae Dasyproctidae Echimyidae Erethizontidae Slope (el) 0.311 0.355 0.301 0.313 0.350 0.306 0.375 0.307 0.242 0.375 0.272 0.299 0.297 0.317 0.332 0.309 0.355 0.288 0.400 0.303 0.293 (0.292-0.330) (O.31l-DAOJ) (0.251-0.352) (0.240-0.390) (O.30fKl.394) (0.278-0.334) (0.351-0.399) (0.283-0.332) (0.134-0.356) (0.297-0.458) (0.195-0.351) (0.041-0.599) (0.193-0.408) (0.304-0.330) (0.294-0.371) (0.190-0.437) (0.286-D.426) (0.202-0.379) (0.292-0.517) (0.285-0.322) (0.210-0.380) 0.300 (0.271-0.3311 0.298 0.254 0.250 0.263 0.373 0.306 0.393 0.304 0.346 0.286 0.235 0.317 0.285 0.291 0.271 0.331 0.315 0.249 0.290 0.279 0.274 0.295 0.200 0.168 0.332 0.323 0.258 0.351 0.212 0.283 (0.283-0.313) (0.155-0.358) (0.208-0.292) (0.193-0.336) (0.294-0.457) (0.294-0.318) (0.345-0.442) (0.275-0.333) (0.235-0.466) (0.259-0.313) (0.167-0.305) (0.170-0.476) (0.239-0.333) (0.264-0.318) (0.247-0.295) (0.276-0.388) (0.273-0.359) (0.192-0.308) (0.270-0.310) (0.252-0.307) (0.160-0.395) (0.260-0.332) (-0.040-0.465) (0.023-0.320) (0.324-0.340) (0.205-0.451) (0.007-0.543) (0.264-0.442) (0.141-0.286) (0.164-0.409) Intercept n' 2.520 2.532 2.543 2.506 86 12 37 14 37 23 9 45 8 19 7 6 9 479 2.460 2.429 2.515 2.477 2.447 2.595 2.425 2.475 2.494 2.369 2.425 2.299 2.442 2.331 2.472 2.333 2.374 2.329 2.491 2.457 2.522 2.522 2.511 2.571 2.516 2.577 2.560 2.572 2.568 2.554 2.608 2.645 2.688 2.554 2.721 2.740 2.623 2.633 2.704 2.642 2.718 2.749 2.504 2.512 2.433 2.547 2.480 2.538 31 7 65 15 IO 214 23 99 123 20 49 24 II 227 40 47 16 71 19 8 21 51 26 23 31 16 138 59 6 48 7 II 455 10 6 12 16 II " P 0.96 0.99 0.90 0.94 0.94 0.98 0.99 0.97 0.91 0.92 0.97 0.85 0.93 0.92 0.96 0.95 0.79 0.89 0.95 0.91 0.84 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0016 0.0001 0.0003 0.0334 0.0003 0.0001 0.0001 0.0013 0.0001 0.0001 0.0001 0.0001 0.0001 0.90 0.0001 0.96 0.78 0.87 0.85 0.96 0.96 0.94 0.95 0.87 0.93 0.87 0.90 0.95 0.95 0.98 0.94 0.94 0.93 0.93 0.94 0.96 0.93 0.69 0.66 0.97 0.91 0.82 0.94 0.86 0.87 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0021 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0028 0.0001 0.0857 0.0278 0.0001 0.0003 0.0464 0.0001 0.0001 0.0005 2.11 -1.l9 2.28 2.64 2.67 2.58 2.82 2.07 2.31 1.88 2.09 2.32 2.22 1.80 1.74 1.67 1.85 1.78 1.94 1.82 2.01 1.72 2.59 2.35 -0.90 2.64 2.76 2.34 2.78 2.84 2.94 2.60 2.64 2.67 3.21 2.78 3.29 3.24 3.32 3.57 3.40 3.14 3.15 3.28 3.14 3.12 2.98 2.22 2.39 2.30 2.69 2.38 2.63 0.43 0.60 0,49 0.80 -1.31 -0.59 -1.95 -1.24 -0.52 -0.95 -1.82 -2.08 -2.04 -1.77 -1.91 -1.35 -1.73 -1.27 -2.07 0.32 -0.43 0.47 0.92 -0.47 0.68 0.84 1.19 0.1I 0.25 0.44 2.10 0.59 2.23 2.05 2.36 2.74 2.66 1.81 1.87 2.12 1.72 2.08 1.41 -0.86 -0.39 -0.53 0.42 -0.49 0.34 TABLE Taxon Heteromyidae Muridae Sciuridae Lagomorpha 25 SILVA-SCALING OF BODY LENGTH February 1998 Slope (Cl) 0.383 0.333 0.341 0.337 (0.312--0.458) (0.318--0.347) (0.315-0.368) (0.286--0.390) 2.-Continued. Intercept 2.569 2.501 2.514 2.542 n" 6 289 60 18 ,. P Average log L Average logM 0.99 0.94 0.96 0.96 0.0001 0.0001 0.0001 0.0001 2.02 2.12 2.31 2.59 -1.43 -1.15 -0.59 0.16 , Number of observations included in the analysis. to Pearson's coefficient of correlation. dia, and Sirenia) were analyzed alone, the regression slope of the log-log relationship between body length and body mass was 0.340, and when wholly-marine marrunals (Cetacea and Sirenia) were analyzed, the value was 0.316. Cetacea, however, contributed unequally to this analysis (Cetacea, n = 31; Sirenia, n = 3). A similar scaling analysis of terrestrial marrunals resulted in an allometric exponent of 0.357. Confidence intervals around regression slopes for these groups indicated that there were no significant differences between them (P > 0.05). The scaling exponent of marine mammals was not significantly different from that predicted by geometric similarity, while that of terrestrial mammals was significantly different. When terrestrial mammals were sorted into volant and non-volant forms, analyses indicated that non-volant mammals complied better with geometric similarity than elastic similarity; however, elastic and geometric models must be rejected for volant marrunals (Table 1). Scaling of body length differed among taxonomic orders (Table 2). Values of confidence intervals of regression slopes indicated that of the 13 taxonomic orders analyzed for which adequate (n :z:: 6) data sets were available, nine slopes differed significantly from the one found when all mammals were pooled. For example, confidence intervals indicated that the scaling exponent of Rodentia was significantly different from that of Artiodactyla, Carnivora, Perissodactyla, Pinnipedia, and Primates. On the other hand, analyses indicated that Lagomorpha and Rodentia, and Cetacea and Pinnipedia, did not differ significantly. These results must be interpreted with caution, because differences in the allometric exponents might reflect different ranges of body mass covered by these orders. Non-linearity of the log-log relationship between body length and body mass was less obvious than when mammals were pooled, but small departures from linearity occurred for the smallest or the largest organisms in several orders (Figs. 3 and 4). Range of body mass for the order Insectivora is similar to that for the Chiroptera, but body length of Chiroptera increased 5% faster with body mass. Although the scaling value found for Pinnipedia did not differ significantly from that of Cetacea, only Cetacea showed a scaling exponent not significantly different from that predicted by geometric similarity. With the exception of Perissodactyla, all taxonomic orders showed only small deviations from the theoretical value of 0.333 expected on the basis of the geometric similitude. For Perissodactyla, the exponent of the length: mass relationship was not significantly different from the theoreti"cal value of 0.250, suggesting that predicted and observed exponents were not significantly different. Within families, body mass was correlated significantly with body length (Table 2). Confidence intervals of regression slopes of the log-log relationship between body length and body mass suggested that most of the differences were not significant. Within the Carnivora, scaling exponents of Canidae and Felidae differ significantly; scaling value for Canidae differed signifi- 26 JOURNAL OF MAMMALOGY '.0 ~---------, 3.5 Vol. 79, No.1 ~----------, ,< Carnivora Artiodactyla 3.5 2.5 3.0 - - ll1lton .. _-.- '.' "'" E E 2.5 0 '-'" ••• ..c +m '.0 C 0 2 • 3 2.' ., , ," , 2.0 ...." ." /~~o 3.5 , .5 - ...... en ~ -, 2.0 Chlrcptera 0 0 m 0 • 3 Cetacea --.J >- 2 Taxon •.. _- All mammals 3.0 Q) "0 - ...... Law ••• " Low ••• ._. AU mammal. 3.0 , - Taxon LowI.I _. All mammals • 3 2 5 Taxon --_ ... Low ••• "'-' All mammals 1.. -3 -, -2 0 3 •• 3.' Xenarthra Insectivore 2.5 3.0 2.0 2.' 1.5 - , -Taxon Taxon ...... ..... - Low,,, Low ••• .,,_. All mommlll, ... _. All mammal. 2.0 -2 -, 0 2 10910 3 Body 1.0 -3 -2 Mass -, 0 (kg) FIG. 3.-Relationships between the logarithm of body length (rum) and the logarithm of body mass (kg) of Artiodactyla, Carnivora, Cetacea, Chiroptera, Xenarthra, and Insectivora. The solid lines are simple log-log linear fits to data of each taxonomic order (Table 2), and the dotted lines represent LOWESS fits to the data. The solid curves represent predictions made by the all-mammals scaling fonnula (Table 1). SILVA-SCALING OF BODY LENGTH February 1998 3.0 r------------, 3.5 27 r------------, Lagomorpha tdarsuplalla 3.0 2.5 2.5 Taxon ...... Low ••• _.. _. AU mammol. 2.0 ' - - - - - " - - - - - - ' o -1 4.0 r----------, 2.0 , 1.5 ' : ·· r 0"1 C 3.5 "-:'-:'~"" >- 2 r----------, ,o 3.5 o a o o Taxon ....•• Law ••• . _. All mammal. OJ 3.0'-_ _ _ _" -_ _ _ _-' 2 0"1 o 0 o .' -0 o -1 / Q) -l -2 Plnnlpedla PerIssodactyla ..c ..... '---_"-_"-_-L_-'--_-' -3 4.0 Low ••• All mammal. 4 3 3.5 - - Toxan ...... Low ... • .. _. All mammal. 3.0 '---'-_-'--_ _-.1.._ _--' 1 2 3 4 3.5 Primates . r------------, Rodentia , , 3.0 , • 0 3.0 2.5 2.5 2.0 o - T"xa" Law ••• All mammal. 2.0 '--_"-_"-_-L_-'--_-' -1 2 3 -2 o - - Taxon .•.•.. Low ••• _.. _. All mammal. 1.5 ~-~-L--~-L--~__: -3 -2 -1 0 2 3 log10 Body Mass (kg) FIG. 4.-Relationships between the logarithm of body length (mm) and the logarithm of body mass (kg) of Lagomorpha, Marsupialia, Perissodactyla, Pinnipedia, Primates, and Rodentia. The solid lines are simple log-log linear fits to the data of each taxonomic order (Table 2), and the dotted lines represent LOWESS fits to the data. The solid curves represent predictions made by the all-mammals scaling fonnula (Table 1). 28 JOURNAL OF MAMMALOGY cantly from both elastic- and geometric-similitude models. For the Felidae, the geometric model could not be rejected. Within Xenarthra, the regression slope of Mynnecophagidae was steeper than that for Dasypodidae but also was based on a smaller sample size (Table 2). Based on confidence intervals of the regression slopes, the geometric-similitude model could be accepted for Dasypodidae, whereas elastic and geometric models were rejected for Myrmecophagidae (Table 2). For Primates, the scaling exponent differed between families of simians and prosimians. When 123 primates species (Table 2) were sorted as simians (Callithricidae, Cebidae, and Cercopithecidae) or prosimians (Lorisidae), the regression slope for simians (slope = 0.289; r = 0.96; n = 97) was significantly different (P < 0.05) from that of prosimians (slope ~ 0.348; r ~ 0.97; n ~ 26). Based on the confidence intervals, neither scaling model fit simians; whereas the geometric model could not be rejected for prosimians. Within rodents, most families complied with the geometric-similitude model. The regression slope for Hetereromyidae was steeper than that of other families but also was based on a small sample size. For Echimyidae, however, the allometric exponent was not significantly different from the one expected from the elastic-similitude model. Log-log relationships between body length and body mass were· similar across most families of bats. There were no significant differences between families of Megachiroptera (Pteropodidae, n = 23) and Microchiroptera (all other bats, n = 456, Table 2). When all microchiropteran species were pooled, the regression slope (slope = 0.309; CI ~ 0.296--D.323) was not significantly different from that of rnegachiropteran species (Pteropodidae, Table 2). DISCUSSION This study is the most comprehensive analysis of the scaling of body length in marrunals. It covers a broad range of taxa and the entire range of size in marrunals. Vol. 79, No.1 Body mass is highly correlated with body length whether marrunals are pooled or grouped by type of locomotion or taxon. Results presented provide several equations that can be used for predicting body length in mammals. Note, however, that the regression for all mammals ignores non-linearity of the length: mass relationship and scaling differences among and within tax0nomic orders. The global-allometric exponent of the relationship between body length and body mass is 0.359. While this value suggests that neither the geometric- nor the elasticsimilitude models accurately describe the morphological design of mammals, it is in better agreement with the geometric similarity model than with the elastic one. Analyses presented here also upholds expectations of Economos (1983) concerning the non-linearity of the log-log relationship between body length and body mass. Results show that the linear relationship does not accurately fit data of the smallest «0.0 I kg) and largest (> 100 kg) mammals. Differences in scaling exponents over the range of body mass indicate that no single similarity model can represent morphological design of mammals. Small manunals tend to increase their body length with body mass faster than the common linear relationship would predict, while large mammals increase their body length less rapidly, approaching the theoretical values expected from geometric or elastic similitude. Analyses also indicate that taxonomic groups composed of predominantly small or large species should have different scaling exponents. One explanation for these findings might be a physiological restriction associated with the surface law that causes small mammals to be heavier at a given body length than a common linear relationship predicts (McMahon, 1973; Peters, 1983). This reduces the ratio of surface area to body mass and may help reduce heat loss in small manunals (Peters, 1983; Scholander, 1955). On the other hand, differences in the scaling exponent between species of February 1998 SILVA-SCALING OF BODY LENGTH small and large mammals lend general support to Economos' (1984) prediction that elastic forces in a gravitational field may surpass the safety factor in the mechanical s'trength of mammalian bones in large mammals; thus, they show an exponent significantly smaller than the one found for smaller ones. Similar scaling values are found when all mammals are pooled, and when marine and terrestrial mammals are analyzed separately, although the average marine mammal is about twice as long and weighs over three times more than its terrestrial counterpart. However, when wholly-marine species are compared with terrestrial ones, body length increases 19% faster in terrestrial forms. These findings lend general support to studies suggesting that wholly-marine mammals conform better with the geometric similarity because they do not have problems with gravity (Economos, 1983, 1984), although other factors also may explain differences between terrestrial and marine species. One explanation might be that marine mammals need to be more massive than terrestrial forms to reduce heat loss. As water is abetter thermal conductor than air, aquatic mammals may have greater problems in temperature regulation than terrestrial forms, resulting in selection for larger body mass rather than body length (Wolff and Guthrie, 1985). It is also possible, however, that these results derive from the strong correlation between taxonomy and means of locomotion. For example, all volant mammals belong to the order Chiroptera, but no species of the order Cetacea can walk. The body shape and length: mass relationship may therefore reflect some constraints of swimming producing a streamline body in marine mammals (R. Channell, pers. comm.). Significant differences also were found when marine mammals that venture onto land (Pinnipedia) were compared to cetaceans; the geometric model seems to be appropriate only for cetaceans. Pinnipeds may be less modified fOIl aquatic life than are 29 wholly aquatic cetaceans. There is, however, considerable variability in Pinnipedia. Scaling of body length is lower for Otariidae than for Phocidae, and only for Phocidae, the geometric model cannot be rejected. Obviously, the reason for the difference between phocids and otariids is not due to their means of locomotion; apart from the fact that otariids use forelimbs and phocids hind limbs, locomotion is relatively similar. Results suggest that for a given increment in body mass, body length increases by a smaller amount in otariids than in phocids. This may be due to different ranges of body mass covered by these two families and nonlinearity of the log-log length: mass relationship. On average, phocids are heavier and larger than otariids (Table 2). These differences also may result from different ecological adaptations related to reproduction. Although both female phocids and otariids exhibit high levels of maternal investment and milk fat, fat, and water deposition is more important in phocids than otariids (Boness and Bowen, 1996; Iverson et al., 1993; Riedman and Ortiz, 1979). These different maternal strategies result in different patterns of mass change in females, which tend to lose considerable mass particularly when lactating (Boness and Bowen, 1996). Thus, relationships between body length and body mass may reflect different energetic strategies of females pinnipeds. Under this scenario, data for pinnipeds used in this analysis would need to be unequally represented by females. Unfortunately, sources used only sporadically reported sex of -individuals. Additional studies are necessary to test if sexual differences in pinnipeds affect the scaling exponent of body length in this group. Therefore, it is difficult to decide if differences between pinnipeds and cetaceans are due to different means of locomotion, different ecological adaptations, or both. Another question addressed in this study is whether volant mammals are less massive than non-volant forms of similar body length. Analyses indicate that insectivores 30 JOURNAL OF MAMMALOGY are heavier (ca. 25%) than bats of similar length and body length increases ca. 5% faster with increased body mass in Chiroptera than in Insectivora. Cristoffer (1991) reached similar conclusions in an examination of bats and rodents from Nigeria and Thailand. In addition, LOWESS analyses show that the increase of body length is different between smaller «30 g) and larger bats (>30 g). It is possible that the difference in the allometric exponent between small and large bats helps large bats minimize wing loading. Constraints imposed by flight on structural and physiological attributes of bats may become progressively greater as body mass increases, so that body length increases more rapidly than does body mass (Cristoffer, 1991). This also suggests that the difference between the regression slope of the log-log relationship between body length and body mass between volant and non-volant mammals depends on absolute body mass of the species in the study. The conclusion regarding similarity between microchiropterans and macrochiropterans with respect to the allometric exponent is limited by the fact that Megachiroptera only contains one family and Microchiroptera contains several. Alternatively, apparent similarity between microchiropterans and macrochiropterans may indicate that differences in Chiroptera result from physiological or morphologically adaptations associated with flight rather than taxonomy. Ordinal scaling suggests several similarities among mammals belonging to different orders but covering a comparable range of body mass. For most taxonomic orders, the allometric exponent complies better with the geometric-similitude model than with the elastic one. The only clear exception is Perissodactyla. Reasons for this trend are not clear, but they may reflect disproportionate contribution of rhinos (n = 6) and tapirs (n = 6) in my analysis, both of which possess a morphological design clearly different from that exhibited by horses and zebras. The same pattern is ob- Vol. 79, No.1 served in Suidae (Artiodactyla, Table 2). Uris may reflect similar effects of environment on functional design or morphology of these three families. Species of Artiodactyla scale similarly to Pinnipedia and Carnivora. Lagomorpha and Rodentia also exhibit a comparable scaling exponent. At the family level, results indicate that scaling exponents also vary within the same order. Moreover, results indicate that scaling exponents derived for some families deviate significantly from those detennined when all the species in an order are pooled (Xenarthra, Primates and Carnivora, Table 2). In particular, variability among families within Carnivora suggests the great dissimilarity in their shape. Although there are some significant differences possibly due to small sample sizes, relative constancy of exponents in families in orders such as Rodentia may show that species in this order are more uniform in morphological design.,. Scaling differences between orders and families might reflect, therefore, greater heterogeneity across orders compared to within them. Thus, although all analyses agree on the importance of absolute body mass, taxa of species surveyed also may have a significant influence on the scaling exponent of the length: mass relationship. Scaling differences of body length across and within taxonomic orders might reflect some physiological, ecological, and evolutionary adaptations found in mammals. Several studies suggest the role of phylogenetic history as an alternate explanation for morphological design (Huey, 1987; Lauder, 1991; Wainwright. 1991). For example, evidence from paleontological, morphological, and genetic data shows that pinnipeds evolved from a common carnivore ancestor that was probably an ursid or a mustelid (Amasson and Widegren, 1986; Lento et aI., 1995; McLaren, 1960; Novacek. 1992; Tedford, 1976). It is possible, therefore, that similarities found between Carnivora and Pinnipedia lend general support to studies suggesting that these orders February 1998 SIL VA-SCALING OF BODY LENGTH are evolutionarily related. On the other hand, the difference in the scaling exponent between otariids and phocids may result from recent ecological adaptations and not from characteristics that they have inherited from a common ancestor. Findings also show that for a given increment in mass, body length increases 45% faster in Canidae than in Felidae. This difference upholds the idea that morphology influences ecology by limiting ability of an individual to perfonn important tasks in its daily life (Alexander, 1988; James, 1991; Wainwright 1991). For example, effects of morphological variation in Carnivora may explain why even though Felidae and Canidae are predators, each family uses different strategies to capture prey and exploit different types of prey (Earle, 1987; Eisenberg, 1981). Morphological design and large size of Felidae may allow them to capture prey much larger than themselves (Rosenzweig, 1966; Vezina, 1985; Wilson, 1975). ACKNOWLEDGMENTS I am grateful to J. A. Downing who encouraged me to do this study. I also thank R. Peters, F.-J. Lapointe, R. Channell, M. Prieto, and one anonymous reviewer for their comments and suggestions on the manuscript. LITERATURE CITED ALEXANDER, R. M. 1977. Allometry of the limbs of antelopes (Bovidae). Journal of the Zoological Society of London, 173:549-573. - - - . 1988. Elastic mechanisms in animal movement. Cambridge University Press, Cambridge, United Kingdom. 141 pp. ANDERSON, S., R. RIDDLE, T. L. YATES, AND J. A. COOK. 1993. Los mamfferos del Parque Nacional Amb6ro y la region de Santa Cruz de la Sierra, Bolivia. Special Publication of the Museum of Southeastern Biology, University of New Mexico Press, Albuquerque. 58 pp. ARNASSON, U., AND B. WIDEGREN. 1986. Pinniped phylogeny enlightened by molecular hybridizations using highly repetitive DNA. Molecular Biology and Evolution, 3:356-365. BONESS, D. J., AND W. D. BOWEN. 1996. The evolution of maternal care in pinnipeds. BioScience, 46:645654. CALDER, W. A., III. 1984. Size, function and life history. The Belnap Press of Harvard University Press, Cambridge, Massachusetts, 431 pp. CEBALLOS, G., AND A. MIRI"tNDA. 1986. Los mamfferos 31 de Chamela, Jalisco. Third ed. Universidad Nacional Autonoma de Mexico, Mexico, D.F., 436 pp. CLEVELAND, W. S. 1979. Robust locally weighted regression and smoothing scatterplots. Journal of the American Statistical Association. 74:829-836. CLEVELAND, W. S., AND R. McGILL. 1985. Graphical perception and graphical methods for analyzing scientific data. Science. 229:828-833. CORBET. G. B., AND J. E. HILL. 1991. A world list of mammalian species. Oxford University Press, Oxford, United Kingdom, 243 pp. CRISTOFFER, C. 1991. Scaling of body weight in bats and rodents. Journal of Manunalogy, 72:731-733. EARLE, M. 1987. A flexible body mass in social carnivores. The American Naturalist, 133:736-740. ECONOMOS, A. C. 1983. Elastic and/or geometric similarity in manunalian design? Journal of Theoretical Biology, 103:167-172. - - - . 1984. The surface illusion and the elastic! geometric similarity paradox, encore. Journal of Theoretical Biology, 109:463-470. EISENBERG, J. F. 1981. The manunalian radiations: an analysis of trends in evolution, adaptation, and behavior. The Chicago University Press, Chicago, Illinois, 610 pp. - - - . 1989. Mammals of the Neotropics: Panama, Colombia, Venezuela, Guyana. Surinam, French Guiana. The Chicago University Press, Chicago, Illinois, 449 pp. EMMONS, L. H. 1990. Neotropical rainforest mammals: a field guide. The Chicago University Press, Chicago, Illinois, 281 pp. HAYSSEN, v., AND T. H. KUNZ. 1996. Allometry of litter mass in bats: maternal size, wing morphology, and phylogeny. Journal of Mammalogy, 77:476490. HUEY, R. B. 1987. Phylogeny, history, and the comparative method. Pp. 76-97, in New directions in ecological physiology (M. E. Feder, A. F. Bennett, W. E. Burggren, and R. B. Huey, eds.). Cambridge University Press, Cambridge, United Kingdom, 364 pp. HUFNAGL, E. 1972. Libyan mammals. The Oleander Press, Stoughton, Wisconsin, 85 pp. HUSSON, A. M. 1978. The mammals of Suriname. ZoOlogische Monographieen van het Rijksmuseum van NatuurIijke Historie No.2. E. J. Brill, Leiden, The Netherlands, 569 pp. HUTfERER, R. 1986. The species of Crocidura (Soricidae) in Morocco. Mammalia, 50:521-534. IVERSON, S. J., W. D. BOWEN, D. J. BONESS, AND O. T. OFfEDAL. 1993. The effect of maternal size and milk energy output on pup growth in grey seals (Halichoerus grypus). Physiological Zoology, 66: 61-88. JAMES, F. C. 1991. Complementary description and experimental studies of dinal variation in birds. American Zoologist, 31:694-705. JERISON, R. R. 1970. Brain evolution: new light on old principles. Science, 170:1224-1225. JOLICOEUR, p. 1990. Bivariate allometry: interval estimation of the slopes of the ordinary and standardized normal major axes and structural relationship. Journal of Theoretical Biology, 144:275-285. LA BARBERA, M. 1989. Analyzing body size as a fac- 32 JOURNAL OF MAMMALOGY tor in ecology and evolution. Annual Review of Ecology and Systematics, 20:97-117. LAUDER, G. V. 1991. Biomechanics and evolution: integrating physical and historical biology in the study of complex systems. Pp. 63-78, in Biomechanics in evolution (D. B. Wake, and G. Roth, eds.). Cambridge University Press, Cambridge, United Kingdom, 273 pp. LENTO, G. M., R. E. HICKSON, G. K. CHAMBERS, AND D. PENNY, 1995. Use of spectral analysis to test hypotheses on the origin of pinnipeds. Molecular Biology and Evolution, 12:28-52. MARES, M. A., R. A. OJEDA, AND R. M. BARQUEZ. 1989. Guide to the mammals of Salta Province, Argentina. University of Oklahoma Press, Norman, 303 pp. MARSHALL, P: 1967. Wild mammals of Hong Kong. Oxford University Press, London, United Kingdom, 64 pp. MARTIN, R. D., AND A. D. BARBOUR. 1989. Aspects of line-fitting bivariate allometric analyses. Folia Primatologica,53:65-81. MAZER, S. J., AND N. 1: WHEELWRIGHT. 1993. Fruit size and shape: allometry at different taxonomic levels in bird-dispersed plants. Evolutionary Ecology, 7:556-575. McLAREN, I. A. 1960. Are the Pinnipedia biphyletic? Systematic Zoology, 9:18-28. McMAHON, 1: A. 1973. Size and shape in biology. Science, 179:1201-1204. MESPLE, E, M. TROUSSELLIER, AND P. LEGENDRE. 1996. Evaluation of simple statistical criteria to qualify a simulation. Ecological Modelling, 88:9-18. NIKLAS, K. J. 1994. The scaling of plant and animal body mass, length, and diameter. Evolution, 48:4454. NOVACEK, M. J. 1992. Mammalian phylogeny: shaking the tree. Science, 356: 121-125. NOWAK, R. M. 1991. Walker's mammals of the world. Fifth ed. The Johns Hopkins University Press, Baltimore, 1:1-642,2:643-1629. PETERS, R. H. 1983. The ecological implications of body size. Cambridge University Press, Cambridge, United Kingdom, 329 pp. POLLOCK, C. M., AND R. E. SHADWICK. 1994. Relationship between body mass and biomechanical Vol. 79, No.1 properties of limb tendons in adult mammals. The American Journal of the Physiological Society, 266: 1016-1021. RADINSKY, L. 1978. Evolution of brain size in carnivores and ungulates. The American Naturalist, 112: 815-831. RffiDMAN, M., AND C. L. ORTIZ. 1979. Changes in milk composition during lactation in the northern elephant seal. Physiological Zoology, 52:240-249. ROSENZWEIG, M. L. 1966. Community structure in sympatric Carnivora. Journal of Mammalogy, 47: 602-612. SCHOLANDER, P. E 1955. Evolution of climatic adaptation in homeotherms. Evolution, 9: 15-26. SILVA, M., AND J. A. DOWNING. 1995. The allometric scaling of density and body mass: a nonlinear relationship for terrestrial mammals. The American Naturalist, 145:704-727. STRAHAM, R. 1983. Complete book of Australian mammals. Angus & Robertson Publishers, London, United Kingdom, 530 pp. TEDFORD, R. H. 1976. Relationships of Pinnipedia to other carnivores (Mammalia). Systematic Zoologist, 25:363-374. VEZINA, A. F. 1985. Empirical relationships between predator and prey size among terrestrial vertebrate predators. Oecologia, 67:555-565. WAINWRIGHT, P. C. 1991. Ecomorphology: experimental functional anatomy for ecological problems. American Zoologist, 31:680-693. WAINWRIGHT, S. A., W. D. B!9GS, J. D. CURKEY, AND 1. M. GOSLINE. 1982. Mechanical design in organisms. Princeton University Press, Princeton, New Jersey, 423 pp. WILSON, D. S. 1975. The adequacy of body size as a niche difference. The American Naturalist, 109:769784. WOLFF, J. 0., AND R. D. GUTHRIE. 1985. Why are aquatic small mammals so large? Oikas, 45:365373. ZAR, J. H. 1996. Biostatistical analysis. Third ed. Prentice-Hall, Inc., Upper Saddle River, New Jersey, 662 pp. Submitted 23 July 1996. Accepted 1 April 1997. Associate Editor was Janet K. Braun.
© Copyright 2026 Paperzz