AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 120:171–181 (2003) Dietary Constraints on Encephalization in Primates Jennifer L. Fish1* and Charles A. Lockwood1,2 1 2 Department of Anthropology, Arizona State University, Tempe, Arizona 85287 Institute of Human Origins, Arizona State University, Tempe, Arizona 85287 KEY WORDS hypothesis brain size; comparative methods; independent contrasts; expensive tissue ABSTRACT Encephalization, and its relationship to potential selective forces, have been a focus of many studies of primate adaptation. It has been argued that gut size may constrain brain mass because these two types of “expensive tissue” (among others) compete in their metabolic requirements (Aiello and Wheeler  Curr. Anthropol. 36:199 –221). Following from the inverse correlation of gut size with diet quality, the expensive tissue hypothesis predicts that differences in diet quality are positively correlated with differences in brain mass, once the correlation of each variable with body mass is taken into account. We tested this prediction using both nonphylogenetic and phylogenetic methods. The results of both methods are consistent with predictions made by the expensive tissue hypothesis. We also discuss several examples of independent contrasts that are consistent with the hypothesis (e.g., Colobinae vs. Cercopithecinae), as well as some that are not (e.g., Tarsius vs. anthropoidea). Overall, the results indicate that improved diet quality, by allowing reduction in relative gut mass, is one mechanism involved in increased encephalization. Am J Phys Anthropol 120:171–181, 2003. © 2003 Wiley-Liss, Inc. Relatively large brains, along with enhanced visual and grasping capabilities, are adaptive features characteristic of all primates. Together, these three features are critical to primate foraging behavior and have provided a productive area of inquiry within the fields of primatology and paleoanthropology. Variation in relative brain mass among primates foraging strategies within the order Primates has provided the opportunity to test not only why primates as a whole have relatively large brains, but also why primates vary in encephalization. Explanations for the evolution of relatively large brains in primates have predominantly cited selection for behaviors in response to either social (e.g., Byrne and Whiten, 1988; Cheney and Seyfarth, 1985; Barton, 1996; Barton and Dunbar, 1997) or ecological (e.g., Clutton-Brock and Harvey, 1980; Harvey and Clutton-Brock, 1985; Foley and Lee, 1992; Milton, 1981, 1988, 1993) challenges. To a lesser extent, the role of biological constraints in limiting brain size has been debated (e.g., McNab and Eisenberg, 1989; Deacon, 1990; Allman et al., 1993; Aiello and Wheeler, 1995; Martin, 1996). In this paper, we focus on the latter subject and test predictions that follow from negative correlations between gut size and brain mass, specifically, the predictions of the expensive tissue hypothesis (Aiello and Wheeler, 1995). We investigate whether diet explains significant variance in brain size independent of that explained by body size. forefront of the encephalization debate with the introduction of the “expensive tissue hypothesis.” This hypothesis attempts to explain how the metabolic cost of relatively large brains can be met without increasing the basal metabolic rate (BMR). Of particular interest is the fact that humans do not show an increase in BMR corresponding to their observed increase in brain mass, which is approximately 5 times that expected for their body mass. In short, the expensive tissue hypothesis proposes that brain mass can increase without a corresponding increase in BMR if another metabolically expensive tissue, such as the gut, decreases in mass. Aiello and Wheeler (1995) supported their hypothesis with two essential points. First, they found a negative correlation between relative gut size and relative brain size in a small sample of primates. Second, when the observed and expected mass of organs in humans was analyzed using a regression model based on all primates, the total organ mass fit expectations, but brain mass was considerably larger than expected, and gut mass Grant sponsor: Institute of Human Origins; Grant sponsor: Arizona State University. *Correspondence to: Jennifer Fish, California Institute of Technology, Biology Division, MC 156-29, Pasadena, CA 91125. E-mail: email@example.com Received 8 January 2001; accepted 13 May 2002. THE EXPENSIVE TISSUE HYPOTHESIS Aiello and Wheeler (1995) brought the issue of biological constraints on relative brain mass to the © 2003 WILEY-LISS, INC. DOI 10.1002/ajpa.10136 Published online in Wiley InterScience (www.interscience.wiley. com). 172 J.L. FISH AND C.A. LOCKWOOD TABLE 1. Raw data and species included Species Strepsirhini Avahi laniger Cheirogaleus major Cheirogaleus medius Daubentonia madagascariensis Galago demidoff Galago senegalensis Indri indri Loris tardigradus Microcebus murinus Nycticebus coucang Perodicticus potto Propithecus verreauxi Haplorhini Alouatta seniculus Aotus trivirgatus Ateles belzebuth Ateles geoffroyi Callicebus moloch Cebuella pygmaea Cebus apella Cercocebus albigena Cercocebus galeritus Cercopithecus aethiops Cercopithecus ascanius Cercopithecus lhoesti Cercopithecus mitis Colobus badius Gorilla gorilla Hylobates agilis Hylobates lar Hylobates moloch Macaca fascicularis Macaca nemestrina Macaca sinica Miopithecus talapoin Nasalis larvatus Pan troglodytes Papio anubis Papio hamadryas Papio ursinus Pongo pygmaeus Presbytis entellus Presbytis obscura Saguinus midas Tarsius bancanus 1 Index of diet quality1 Log brain mass (g)2 Log body mass (g)2 0.42 0.70 0.69 0.84 0.87 0.63 0.55 0.95 0.76 0.71 0.62 0.56 1.00 0.83 0.50 1.65 0.53 0.68 1.58 0.82 0.25 1.10 1.15 1.43 3.02 2.65 2.25 3.45 1.91 2.27 3.80 2.51 1.73 2.90 3.06 3.54 0.51 0.63 0.65 0.61 0.59 0.67 0.87 0.69 0.65 0.69 0.69 0.67 0.68 0.40 0.38 0.54 0.71 0.53 0.74 0.62 0.68 0.78 0.53 0.61 0.66 0.66 0.63 0.59 0.51 0.51 0.77 1.00 1.72 1.23 2.07 2.03 1.28 0.65 1.84 2.02 2.28 1.82 1.83 1.97 1.88 1.89 2.70 1.94 2.01 1.99 1.85 2.09 1.92 1.60 1.99 2.61 2.30 2.21 2.26 2.55 2.30 1.81 1.02 0.56 3.81 2.92 3.95 3.90 2.95 2.15 3.37 3.90 4.03 3.64 3.53 3.93 3.80 3.85 5.02 3.87 3.76 3.79 3.70 3.94 3.92 3.08 4.15 4.66 4.40 4.15 4.39 4.80 4.33 3.85 2.54 2.10 Compiled from Sailer et al. (1985) and Fleagle and Reed (1996). was significantly smaller than expected. While their interpretation was limited by the small sample available, the hypothesis of Aiello and Wheeler (1995) derived from these observations and the postulated correlated evolution of brain and gut mass. Thus, “if the changes in the proportions of the two organs were contemporary evolutionary events, there is no reason that the BMRs of hominids would ever have been elevated above those typical of other primates as a consequence of the energetic costs of encephalization” (Aiello and Wheeler, 1995, p. 205). A direct test of the evolutionary relationship of gut size and brain size in primates is limited by the lack of quality data on primate gut mass, a difficulty noted by Aiello and Wheeler (1995). However, they also emphasized the relationship between gut mass and diet quality. Drawing primarily on the data of others, they suggested “that the relationship between relative brain size and diet is primarily a 2 Compiled from Stephan et al. (1981) and Marino (1998). relationship between relative brain size and relative gut size, the latter being determined by dietary quality” (Aiello and Wheeler, 1995, p. 207). It was shown elsewhere that gut morphology and size are related to diet quality in mammals (Chivers and Hladik, 1980). In this context, diet quality refers to the nutrient content and ease of digestion of a food source. Mammals consuming low-quality foods, such as leaves, have large, complex guts. Faunivores have simple stomachs, and frugivores display an intermediate morphology (Chivers and Hladik, 1980). In primates, this pattern is exemplified by colobine monkeys, who have complex guts and a diet consisting predominantly of leaves (Chivers, 1994). Given that diet quality is a strong correlate of gut mass, the expensive tissue hypothesis predicts that differences in diet quality are positively correlated with differences in brain mass, once the effects of potentially confounding factors, such as body mass, are taken into account. DIET AND ENCEPHALIZATION 173 Fig. 1. All primates: brain mass residuals vs. diet quality residuals. Residuals were obtained through linear least-squares regression of each variable on body mass. The regression line has a significantly positive slope (y ⫽ 0.62x). Squares, haplorhines; circles, strepsirhines. Although Aiello and Wheeler (1995) concluded that reducing biological constraints by balancing the cost of metabolically expensive tissues was essential to encephalization, the generalization that changes in brain mass and diet quality were correlated evolutionary events across primates was not tested. Here, we use multiple comparative methods to examine the evolutionary relationship of diet quality and brain mass in primates, and test for the correlated evolution of these two traits. MATERIALS AND METHODS Variables The three variables used in this study are whole brain mass, body mass, and diet quality. Data on whole brain mass and body mass were taken from Stephan et al. (1981) and Marino (1998) and logtransformed (base 10) prior to use in analysis. These data represent species averages. Both brain and body mass data were derived from the same source for all species. Some recent analyses on primate encephalization preferred to analyze neocortex mass, rather than whole brain mass. This is sometimes justified because the neocortex, a uniquely mammalian struc- ture in the forebrain associated with higher cognitive functions, is largely responsible for the increased mass of the whole brain in primates (Allman, 1999). However, metabolic constraints operate on the whole brain; therefore, whole brain mass is the appropriate variable for our analysis. Diet quality has often been quantified using the percentage of a specific food type in the diet (e.g., percent frugivore; Dunbar, 1992). For our purposes, it is necessary to create a value for diet quality that is able to reflect the contribution of a range of food types to a primate’s diet. We determined diet quality from the percent of time spent foraging for different food items, as reported in Sailer et al. (1985) and Fleagle and Reed (1996). A continuous, quantitative index for absolute diet quality was established using the following equation: DQ ⫽ 0.33L ⫹ 0.67F ⫹ M, where DQ is the index of diet quality, and L, F, and M are the percentages of time spent foraging for leaves, fruit, and meat, respectively. The possible range of DQ values is 0.33–1. The coefficients of 0.33, 0.67, and 1 serve as a weighting for each food 174 J.L. FISH AND C.A. LOCKWOOD TABLE 2. Results of regression analyses of diet quality vs. body mass Nonphylogenetic (species values) Phylogenetic (independent contrasts) Group n Slope r F P* n Slope r F P* All primates Strepsirhines Haplorhines Anthropoids 44 12 32 31 ⫺0.09 ⫺0.11 ⫺0.11 ⫺0.08 ⫺0.55 ⫺0.46 ⫺0.59 ⫺0.46 17.98 2.71 15.97 7.88 ⬍0.001 0.13 ⬍0.001 0.009 43 11 31 30 ⫺0.10 ⫺0.03 ⫺0.13 ⫺0.09 ⫺0.34 ⫺0.08 ⫺0.53 ⫺0.35 5.61 0.07 11.98 4.01 0.023 0.80 0.0016 0.055 * Two-tailed. TABLE 3. Results of regression analyses of brain mass (size-adjusted) vs. diet quality (size-adjusted) Nonphylogenetic (species values) Phylogenetic (independent contrasts) Group n Slope r F P* n Slope r F P* All primates Strepsirhines Haplorhines Anthropoids 44 12 32 31 0.62 0.58 0.44 0.65 0.50 0.72 0.42 0.59 14.32 11.05 6.32 15.44 ⬍0.001 0.008 0.018 ⬍0.001 43 11 31 30 0.56 0.62 0.51 0.68 0.60 0.82 0.44 0.57 23.30 20.10 7.36 13.85 ⬍0.001 0.001 0.011 ⬍0.001 * Two-tailed. type. The values of these coefficients were chosen to represent simply the rank of relative nutrient quality and ease of digestion of each type of food (e.g., Fleagle, 1999; Janson and Chapman, 1999), and not any quantitative measure of these features. This equation is similar to that determined by Sailer et al. (1985) for estimating dietary quality, differing only in that we chose a more conservative weighting for animal matter. Data for all three variables used in the analysis were available for 44 primate species (Table 1). Because brain mass is highly correlated with body mass (Jerison, 1973; Martin, 1990), we performed a size-adjustment based on the use of residuals, as described below. Diet, like many ecological variables, also correlates with body mass (Leonard and Robertson, 1994). For our data, diet quality and body mass are significantly negatively correlated in all groups except within strepsirhines. We report details of this correlation below, and consequently, we use size-adjusted values of diet quality to evaluate the relationship of brain mass and diet quality. Comparative methods Both nonphylogenetic and phylogenetic comparative methods were used. All analyses used linear least squares regression analysis, the null hypothesis being that the independent variable does not explain a significant amount of variance in the dependent variable. The significance of each regression was established with the F-test. Separate analyses were conducted for four clades: all primates, strepsirhines, haplorhines, and anthropoids. Nonphylogenetic methods. The nonphylogenetic methodology first adjusted for body mass through linear least-squares regression analyses of species values for brain mass and diet quality (dependent variables) on body mass (independent variable). Significant results for regressions of diet quality on body mass justify the size-adjustment. Residual values in the form observed minus predicted are left in logarithmic form and represent size-adjusted data (see review by Jungers et al., 1995). To determine the relationship between brain mass residuals and diet quality residuals, further regression analyses were run, with diet quality residuals as the independent variable. Intercepts for the latter analyses are always zero. Phylogenetic methods. Nondirectional and directional phylogenetic comparative methods were also used, following the extensive recent literature on these methods (e.g., Harvey and Pagel, 1991; Garland et al., 1992; Sillen-Tullberg and Moller, 1993; Garland and Janis, 1993; DiFore and Rendall, 1994; Mitani et al., 1996; Lindenfors and Tullberg, 1998; Pagel, 1999). All analyses were based on the primate phylogeny determined by Purvis (1995). The first set of analyses used independent contrasts, with each contrast representing a difference between two sister groups. This required estimation of nodal values throughout the phylogeny, and these were determined using the model of Felsenstein (1985) and equal branch lengths throughout (i.e., a “punctuated” model of character evolution). Branch lengths were not used because divergence dates are missing for most of the nodes in the phylogeny of Purvis (1995). Also, Martins and Garland (1992) showed that independent contrast analyses which assume equal branch lengths are robust to deviations from that assumption. Together, the independent contrasts represent all possible comparisons between sister groups in the primate phylogeny. Contrasts were determined, using the computer program “Comparative Analyses by Independent Contrasts” (Purvis and Rambaut, 1995). Size-adjustment and tests of association between contrasts followed methods and formulae described and justified by Garland et al. (1992). To control for the effects of body mass, independent contrasts of the dependent variables (diet quality or brain mass) were regressed on positivized con- 175 DIET AND ENCEPHALIZATION Fig. 2. All primates: size-adjusted contrasts in brain mass vs. size-adjusted contrasts in diet quality. The regression line has a significantly positive slope (y ⫽ 0.56x). Squares, haplorhines; circles, strepsirhines; triangle, root contrast. trasts in body mass, using linear least-squares regression. Slope values and correlation coefficients (r) were determined for a regression line forced through the origin (see Harvey and Pagel, 1991; Garland et al., 1992). Residual values from these analyses represent size-adjusted contrasts in the same manner that residuals from the regression analysis of species values represent sizeadjusted species values. To test whether diet quality explains a significant amount of variance in brain mass, size-adjusted contrasts for each variable were then subjected to a linear least-squares regression, with diet quality contrasts as the independent variable. The second phylogenetic comparative method was a test for directional relationship that helps illustrate the phylogenetic relationship between brain mass and diet quality. In this case, size-adjusted values for diet quality and brain mass were derived for each species, using methods described above in Nonphylogenetic Methods. Residual species values of diet quality from linear regression on body mass ranged from ⫺0.3 to 0.3. This range was arbitrarily divided into six tenths, converting diet quality into an ordered discrete character with six character states. With each species thus assigned a character state, the evolution of diet quality was reconstucted in MacClade, version 3.07, using accelerated transformation parsimony (Maddison and Maddison, 1992). Brain mass species residuals were evaluated as a continuous trait, using a model of squared-change parsimony, also in MacClade. At branches where changes in diet quality residuals occurred (positive or negative), descendant and ancestral values of brain mass residuals were compared to identify whether changes in brain mass were in the same direction as changes in diet quality. The significance of these results was evaluated using Fisher’s exact test. RESULTS Diet quality and body mass Our index of diet quality is negatively correlated with body mass in primates as a whole, within haplorhines, and within anthropoids (Table 2). This is true using phylogenetic or nonphylogenetic methods, although the anthropoid analysis of independent contrasts is only significant at P ⫽ 0.055. These variables are not significantly correlated in our small sample of strepsirhines. All subsequent analyses use residuals from these regressions as size-adjusted values for diet quality. This does not affect the results for strepsirhines, 176 J.L. FISH AND C.A. LOCKWOOD Fig. 3. Strepsirhines: size-adjusted contrasts in brain mass vs. size-adjusted contrasts in diet quality. The regression line has a significantly positive slope (y ⫽ 0.62x). Note that without the contrast between Daubentonia and Indriidae, the relationship is no longer significant (n ⫽ 10, r ⫽ 0.55, F ⫽ 3.96, P ⫽ 0.078). while it results in slightly higher correlations for analyses of other groups (as expected, given the negative correlation of diet quality with body mass in primates).1 Evaluation of predictions In all analyses of primates as a whole or of subgroups of primates, diet quality and brain mass are significantly positively correlated (Table 3). Figure 1 shows the nonphylogenetic regression analysis of size-adjusted values of diet quality and brain mass for all primates. Analyses of size-adjusted contrasts in brain mass vs. size-adjusted contrasts in diet quality are presented in Figure 2 (for all primates), Figure 3 (for strepsirhines), Figure 4 (for haplorhines), and Figure 5 (for anthropoids). The directional relationship between changes in brain mass and diet quality is also highly significant (Table 4). 1 Barton (1999) cautions that correcting two variables for their correlations with a third (such as body mass) can result in spurious correlations. The differences between our correlation values obtained using raw diet quality contrasts, and size-adjusted contrasts, range from 0 – 0.07. No interpretations about the significance of relationships change based on this size-adjustment, but the magnitudes of correlation and the probability levels are more appropriate. The results are also robust to varying the estimates of body mass used. For illustrative purposes, Figure 6 depicts the primate phylogeny used here, with character evolution mapped onto it. Changes in diet quality and brain mass are graphically represented by arrows. Evolutionary changes in diet quality are generally associated with corresponding changes in brain mass (Fisher’s exact test, P ⫽ 0.0032). Thus, in branches where diet quality increases, ancestral values for brain size tend to be less than descendant values, and vice versa. Therefore, the results of this study are consistent with predictions made by the expensive tissue hypothesis about the relationship between diet quality and brain mass. DISCUSSION Evaluation of methods It has been noted that phylogenetic comparative methods are often applied without first asking whether variation in the traits of interest is empirically related to phylogeny, and that this may encourage unnecessary error in the analysis, perhaps by choice of incorrect phylogeny or branch lengths (Gittleman et al., 1996; Abouheif, 1999). If variation in the trait is independent of phylogeny, it may be argued that traditional nonphylogenetic methods should be applied. However, we also think phylogenetic methods are useful because of DIET AND ENCEPHALIZATION 177 Fig. 4. Haplorhines: size-adjusted contrasts in brain mass vs. size-adjusted contrasts in diet quality. The regression line has a significantly positive slope (y ⫽ 0.51x). the perspective they offer, regardless of statistical benefits. In this study, therefore, we do not directly test whether diet quality or brain mass is strongly influenced by phylogeny prior to evaluating the relationship between them. Instead, by comparing results from different methods, we indirectly examine the impact of inherited similarity at different levels of the analysis. The magnitudes of the correlations (and associated F-tests) between diet quality and body mass, or diet quality and brain mass, are similar for nonphylogenetic and phylogenetic analyses, with the phylogenetic results slightly more significant. (Tables 2 and 3). This general similarity is noteworthy, given statements in the literature that nonphylogenetic comparative methods are essentially invalid (Harvey and Pagel, 1991; Harvey and Purvis, 1991; Martins and Garland, 1992; Purvis and Webster, 1999). Two inferences can be made from the similarity among our results: 1) the apparent significance of the relationship between brain mass and diet quality is not inflated by inherited similarity among close relatives, and 2) the phylogeny and assumptions used in the phylogenetic comparisons do not misrepresent the evolution of these traits. These inferences cannot be generalized beyond this data set. For example, our sampling of taxa is broad (across the order Primates) but not particularly dense within major subgroups. If we were to sample a large number of species within a speciose genus, a larger phylogenetic effect might become evident. Moreover, at the broad level of our study, phylogenetic relationships are relatively wellknown, and phylogenetic comparisons are not likely to be led astray by problems such as sampling errors in close relatives (Purvis and Webster, 1999). A more important result is that for primates as a whole, the relationship of brain mass to diet quality is stronger when using independent contrasts than when using species values. The basis for this is seen in comparing Figures 1 and 2. Strepsirhines generally have smaller brain masses for a given diet than do anthropoids. This difference obscures the relationship between brain mass and diet quality when traditional comparative methods are used. However, in the analysis of independent contrasts, the difference between clades is reduced to one point (Fig. 2), and the relationship between variables of interest is more evident. This result follows other examples where phylogenetic methods can reduce the confounding effect of differences between speciose clades, and lends support to the use of these methods (Nunn and Barton, 2001, p. 94). 178 J.L. FISH AND C.A. LOCKWOOD Fig. 5. Anthropoids: size-adjusted contrasts in brain mass vs. size-adjusted contrasts in diet quality. The regression line has a significantly positive slope (y ⫽ 0.68x). TABLE 4. Directional test of the relationship between diet quality and brain mass1 Brain mass Diet quality Increase Decrease 1 Increase Decrease 10 1 2 7 One-tailed Fisher’s exact test, P ⫽ 0.0032. Individual contrasts Phylogenetic methods serve not only to account for the statistical effects of inherited similarity; they are also useful in that they allow for further understanding of the comparative relationship through the examination of influential contrasts. This can be shown with several examples. First, the highest correlation between brain mass and diet quality occurs within Strepsirhini (Table 3). However, the significance of the correlation is limited due to small sample size. In fact, the high correlation depends largely on a single contrast, that of Daubentonia vs. Indriidae (Fig. 3). Relative to other strepsirhines, the insectivorous Daubentonia has a large brain and a high-quality diet. Daubentonia also displays high levels of sensorimotor intelligence, which may be linked to the extractive foraging behavior of these animals (Sterling and Povinelli, 1999). In contrast, the correlation between brain size and diet quality in anthropoids is consistent throughout the clade. Two examples in particular illustrate this relationship. These are the contrasts of Cercopithecinae vs. Colobinae and Pan vs. Gorilla, which are highlighted in Figure 5. Colobines and cercopithecines are morphologically distinct in their dietary adaptations. The sharply cusped teeth and large, complex stomachs of colobine monkeys reflect their adaptation to a high-cellulose diet (Fleagle, 1999). Differences in brain mass between the subfamilies correspond to differences in diet and gut morphology. A similar divergence in diet quality and brain mass characterizes the African hominoids. Gorillas have a primarily folivorous diet, while chimpanzees, with relatively larger brains, specialize in ripe fruits. Both of these examples reflect contrasts between sister groups that have very different lifestyles. In each case, there is a clear divergence in dietary adaptation that is also associated with a difference in relative brain size. By highlighting closely related species where there has been a significant evolutionary change in both brain mass and diet quality since their divergence from a common ancestor, we are DIET AND ENCEPHALIZATION 179 Fig. 6. Directional phylogenetic test for Primates (relationships derived from Purvis, 1995). Arrows indicate direction of change between nodes. Brain mass (size-adjusted) is represented by solid arrow-halves; diet quality (also size-adjusted) is represented by open arrow-halves. See text for further explanation. 180 J.L. FISH AND C.A. LOCKWOOD better able to evaluate the overall pattern of the relationship between these traits. Finally, the Tarsius vs. Anthropoidea contrast (noted in Fig. 4) demonstrates an interesting exception encountered in this study that deviates especially from the overall pattern within haplorhines. Tarsiers have small brains relative to anthropoids, even after adjusting for body mass. At the same time, they are completely insectivorous, which gives them the highest possible index value for diet quality. The difference represented by this contrast is sufficient to alter the correlation coefficient within haplorhines, as the addition of the tarsier as the sister group to anthropoids has a dramatic negative effect on the correlation (see Table 3). The effect can also be seen in Figure 4, where the contrast between tarsiers and anthropoids is exceptional, given the trend among other contrasts. It has been suggested that outlying contrasts in phylogenetic comparisons often indicate distinct evolutionary trajectories within different clades (“grade shifts,” Martin and Harvey, 1985; Nunn and Barton, 2001). One possibility, then, is that the Tarsius-Anthropoidea contrast reflects a shift in brain mass relative to diet quality between prosimians and anthropoids. As can be seen in Figure 1, Tarsius is similar to many strepsirhines in its relatively small brain mass. Another possible explanation for its low brain mass is the fact that Tarsius has the largest orbits of all primates relative to skull size (Kay and Cartmill, 1977; Martin, 1990). Martin (1990) suggested that tarsier eyes are large to compensate for the absence of the tapetum lucidum in these nocturnal animals. It may be that orbit volume, in this extreme condition, has a role in constraining brain volume. Determining which of these alternative hypotheses best explains the pattern observed in Tarsius warrants further investigation. CONCLUSIONS Consistent with the expensive tissue hypothesis (Aiello and Wheeler, 1995), diet quality and brain mass are significantly positively correlated in primates as a whole, and evolutionary changes in diet quality are directionally related to evolutionary changes in relative brain mass. Moreover, comparisons between nonphylogenetic and phylogenetic comparative methods provide broadly similar results, increasing confidence in the overall trend. In primates as a whole, phylogenetic methods appear to be more appropriate, as they are less affected by the confounding differences in relative brain mass between strepsirhines and anthropoids. Relative brain mass may ultimately be influenced by multiple factors, perhaps including selection on social behavior and foraging strategy. However, we agree with suggestions that physiological constraints are a primary factor affecting brain mass. Our results support the hypothesis that improved diet quality is one mechanism involved in the evolution of relatively large brains, by allowing reduc- tion in relative gut mass. Deviations from the overall pattern (e.g., Tarsius) may illustrate the effect of other constraints and warrant further investigation. ACKNOWLEDGMENTS We thank John Fleagle, Bill Jungers, Bill Kimbel, Charlie Nunn, Kaye Reed, and three anonymous reviewers for their comments and suggestions on earlier drafts of the manuscript. LITERATURE CITED Abouheif E. 1999. A method for testing the assumption of phylogenetic independence in comparative data. Evol Ecol Res 1:895–909. Aiello LC, Wheeler P. 1995. The expensive-tissue hypothesis. Curr Anthropol 36:199 –221. Allman JM. 1999. Evolving brains. Washington, DC: Library of Congress. Allman JM, McLaughlin T, Hakeem A. 1993. Brain structures and life-span in primate species. Proc Natl Acad Sci USA 90: 3359 –3563. Barton R. 1996. Neocortex size and behavioral ecology. Proc R Soc Lond [Biol] 263:173–177. Barton R. 1999. The evolutionary ecology of the primate brain. In: Lee PC, editor. Comparative primate socioecology. Cambridge: Cambridge University Press. p 167–203. Barton R, Dunbar RIM. 1997. Evolution of the social brain. In: Whiten A, Byrne R, editors. Machiavellian intelligence. II: extensions and evaluations. Cambridge: Cambridge University Press. p 240 –263. Byrne RW, Whiten A, editors. 1988. Machiavellian intelligence: Social expertise and the evolution of intellect in monkeys, apes, and humans. Oxford: Oxford University Press. Cheney DL, Seyfarth RM. 1985. Social and non-social knowledge in vervet monkeys. Philos Trans R Soc Lond [Biol] 308:187– 201. Chivers DJ. 1994. Functional anatomy of the gastrointestinal tract. In: Davies A, Oates J, editors. Colobine monkeys: their ecology, behavior and evolution. Cambridge: Cambridge University Press. p 205–227. Chivers DJ, Hladik CM. 1980. Morphology of the gastrointestinal tract in Primates: comparisons with other mammals in relation to diet. J Morphol 166:337–386. Clutton-Brock TH, Harvey PH. 1980. Primates, brains and ecology. J Zool Lond 190:309 –323. Deacon TW. 1990. Problems of ontogeny and phylogeny in brainsize evolution. Int J Primatol 11:237–282. DiFore A, Rendall D. 1994. Evolution of social organization: a reappraisal for primates by using phylogenetic methods. Proc Natl Acad Sci USA 91:9941–9945. Dunbar RIM. 1992. Neocortex size as a constraint on group size in primates. J Hum Evol 20:469 – 493. Felsenstein J. 1985. Phylogenies and the comparative method. Am Nat 125:1–15. Fleagle JG. 1999. Primate adaptation and evolution. New York: Academic Press. Fleagle JG, Reed KE. 1996. Comparing primate communities: a multivariate approach. J Hum Evol 30:489 –510. Foley R, Lee PC. 1992. Ecology and energetics of encephalization in hominid evolution. In: Whitten A, Widdowson E, editors. Foraging strategies and natural diet of monkeys, apes, and humans. Oxford: Clarendon Press. p 63–72. Garland T, Janis C. 1993. Does metatarsal/femur ratio predict maximal running speed in cursorial mammals? J Zool Soc Lond 229:133–151. Garland T, Harvey PH, Ives AR. 1992. Procedures for the analysis of comparative data using phylogenetically independent contrasts. Syst Biol 41:18 –32. Gittleman J, Anderson CG, Kot M, Luh H-K. 1996. Phylogenetic lability and rates of evolution: a comparison of behavioral, morphological and life history traits. In: Martins E, editor. Phylogenies and the comparative method in animal behavior. New York: Oxford University Press. p 166 –205. DIET AND ENCEPHALIZATION Harvey PH, Clutton-Brock TH. 1985. Life history variation in primates. Evolution 39:559 –581. Harvey PH, Pagel MD. 1991. The comparative method in evolutionary biology. New York: Oxford University Press. Harvey PH, Purvis A. 1991. Comparative methods for explaining adaptations. Nature 351:619 – 624. Janson CH, Chapman CA. 1999. Resources and primate community structure. In: Fleagle J, Janson C, Reed K, editors. Primate communities. Cambridge: Cambridge University Press. p 237–267. Jerison HJ. 1973. Evolution of the brain and intelligence. New York: Academic Press. Jungers WL, Falsetti AB, Wall CE. 1995. Shape, relative size, and size-adjustments in morphometrics. Yrbk Phys Anthropol 38:137–162. Kay RF, Cartmill M. 1977. Cranial morphology and adaptations of Palaechthon nacimienti and other Paromomyidae (Plesiadapoidea, ?Primates), with description of a new genus and species. J Hum Evol 6:19 –53. Leonard WR, Robertson ML. 1994. Evolutionary perspectives on human nutrition: the influence of brain and body size on diet and metabolism. Am J Hum Biol 6:77– 88. Lindenfors P, Tullberg BS. 1998. Phylogenetic analyses of primate size evolution: the consequences of sexual selection. Biol J Linn Soc 64:413– 447. Maddison WP, Maddison DR. 1992. MacClade: analysis of phylogeny and character evolution. Sunderland, MA: Sinauer Associates. Marino L. 1998. A comparison of encephalization between odontocete cetaceans and anthropoid primates. Brain Behav Evol 51:230 –238. Martin RD. 1990. Primate origins and evolution: a phylogenetic reconstruction. Princeton: Princeton University Press. Martin RD. 1996. Scaling of the mammalian brain: the maternal energy hypothesis. News Physiol Sci 11:149 –156. Martin RD, Harvey PH. 1985. Brain size allometry: ontogeny and phylogeny. In: Jungers WL, editor. Size and scaling in primate biology. New York: Plenum. p 147–173. 181 Martins EP, Garland T. 1992. Phylogenetic analyses of the correlated evolution of continuous characters—a simulation study. Evolution 45:534 –557. McNab BK, Eisenberg JF. 1989. Brain size and its relation to the rate of metabolism in mammals. Am Nat 133:157–167. Milton K. 1981. Food choice and digestive strategies of two sympatric primate species. Am Nat 117:496 –505. Milton K. 1988. Foraging behaviour and the evolution of primate intelligence. In: Byrne R, Whiten A, editors. Machiavellian intelligence: social expertise and the evolution of intellect in monkeys, apes, and humans. Oxford: Oxford University Press. p 285–305. Milton K. 1993. Diet and primate evolution. Sci Am 269:86 –93. Mitani J, Gros-Louis J, Richards A. 1996. Sexual dimorphism, the operational sex ratio, and the intensity of male competition in polygynous primates. Am Nat 147:966 –980. Nunn CL, Barton RIM. 2001. Comparative methods for studying primate adaptation and allometry. Evol Anthropol 10:81–98. Pagel M. 1999. Inferring the historical patterns of biological evolution. Nature 401:877– 884. Purvis A. 1995. A composite estimate of primate phylogeny. Philos Trans R Soc Lond [Biol] 348:405– 421. Purvis A, Rambaut A. 1995. Comparative analysis by independent contrasts (CAIC): an Apple Macintosh application for analysing comparative data. Comput Appl Biosci 11:247–251. Purvis A, Webster AJ. 1999. Phylogenetically independent comparisons and primate phylogeny. In: Lee PC, editor. Comparative primate socioecology. Cambridge: Cambridge University Press. p 44 –70. Sailer LD, Gaulin SJC, Boster JS, Kurland JA. 1985. Measuring the relationship between dietary quality and body size in Primates. Primates 26:14 –27. Sillen-Tullberg B, Moller AP. 1993. The relationship between concealed ovulation and mating systems in anthropoid primates: a phylogenetic analysis. Am Nat 141:1–25. Stephan H, Frahm H, Baron G. 1981. New and revised data on volumes of brain structures in insectivores and primates. Folia Primatol (Basel) 35:1–29. Sterling EJ, Povinelli DJ. 1999. Tool use, aye-ayes, and sensorimotor intelligence. Folia Primatol (Basel) 70:8 –16.