correlated response, competition, and female canine size in primates

Correlated Response, Competition, and FemaleCanine Size in Primates

J. MICHAEL PLAVCAN*Department of Anatomy, New York College of Osteopathic Medicine, NewYork Institute of Technology, Old Westbury, New York 11568

KEY WORDS canine teeth; sexual dimorphism; primates;correlated response

ABSTRACT Recently, comparative analyses of female canine tooth size inprimates have yielded two hypotheses to explain interspecific variation infemale relative canine size. Greenfield ([1992] Int. J. Primatol. 13:631–657;[1992] Yrbk. Phys. Anthropol. 35:153–184; [1996] J. Hum. Evol. 31:1–19)suggested that covariation in male and female canine size across speciesindicates that female canine size reflects correlated response (in which theexpression of a trait in one sex causes the expression of the same trait in theother sex). Plavcan et al. ([1995] J. Hum. Evol. 28:245–276) noted that femalecanine size in primates is associated with variation in categorical estimates ofthe intensity of female-female agonistic competition, suggesting that selectionfavors large female canine size in many species. While it may seem that thetwo models are in conflict, they are not. To simultaneously evaluate these twomodels, this analysis examines the joint relations between male canine size,female canine size, and estimates of female-female competition in a sample of108 primate species. Overall, female canine size is correlated with variation inmale canine size. Controlling for variation in male canine size, female caninesize is also correlated with estimates of the intensity of female-female agonisticcompetition. The relation between these variables differs strongly between anthro-poid and strepsirhine primates. In anthropoids, the data suggest that selection forthe development of large canines in females is not constrained by any affect ofcorrelated response. In strepsirhines, the evidence suggests that sexual selec-tion may affect male canine size but that correlated response affects femalecanine size, resulting in monomorphism for most species. These observationshelp reconcile the observations of Greenfield ([1992] Int. J. Primatol. 13:631–657; [1996] J. Hum. Evol. 31:1–19) and Plavcan et al. ([1995] J. Hum. Evol.28:245–276) and provide a more precise model for understanding interspecificvariation in female canine size and hence canine dimorphism. Am J PhysAnthropol 107:401–416, 1998. r 1998 Wiley-Liss, Inc.

Commonly, primate canine tooth sizesexual dimorphism is viewed as a product ofsexual selection acting on males. Thus, selec-tion favors big male canines because bigcanines help males to win fights for access tobreeding females. A few studies recognizethat female canine size is highly variableamong species. Harvey et al. (1978) presentevidence that female-female competition for

access to resources in large groups selectsfor large canine teeth, though data fromSmith (1981) contradicts this. Lucas (1981)and Lucas et al. (1986) note that there issubstantial variation in female canine size

*Correspondence to: J. Michael Plavcan, Department ofAnatomy, New York College of Osteopathic Medicine, Old Westbury,NY 11568.

Received 25 June 1996; accepted 3 September 1998.

AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 107:401–416 (1998)

r 1998 WILEY-LISS, INC.

and suggest that this may be related tofemale competition but offer no comparativedata to back this up.

More recently, two comparative analysespresent apparently contradictory explana-tions for interspecific variation in primatefemale canine size. On the one hand, Plav-can et al. (1995) offer evidence that selectionfavors the development of large canine teethin many female primates in association withfemale-female agonistic competition. On theother hand, Greenfield (1992b, 1996)presents evidence that interspecific varia-tion in female canine size reflects correlatedresponse to variation in male canine size.

Plavcan et al. (1995) suggest that, wherefemales regularly engage in dyadic, agonis-tic contest competition, selection favors thedevelopment of relatively large canines forfighting, as long as there is a fitness conse-quence of winning and losing fights andlarge canines help females win fights. Theyalso suggest that, where the outcome ofescalated contests among females is regu-larly determined by alliances (or coalitions)of individuals rather than individuals fight-ing alone, selection for the development ofweaponry is reduced. In support of this, theypresent evidence that female canines aresmall in species where female agonistic com-petition is either of low intensity or regu-larly occurs between coalitions of femalesand that female canines are larger in specieswhere females compete intensely.

The correlated response model whichGreenfield (1992b, 1996) favors may be unfa-miliar to many readers. Lande (1980)presents a quantitative genetic model whichpredicts that, where traits are polygenic andautosomal, selection for the development ofa trait in one sex should result in theexpression of the trait in the other sex. Thisphenomenon is called correlated response.For example, if canine tooth size is deter-mined autosomally and selection favors thedevelopment of large canine teeth in males,then females too should develop large ca-nines, even in the absence of selection forlarge female canines. The model predictsthat the primitive state for autosomallydetermined dimorphic traits should be mono-morphism. Dimorphism should developwithin a lineage only after a period of time,

pending that there is selective pressure toreduce the trait in the sex lacking selectionto develop or maintain the trait (Lande,1980). In the canine dimorphism example,the model predicts that after enlarged ca-nine teeth were expressed in both males andfemales in the ancestral anthropoid lineage,canine dimorphism would subsequentlyevolve through a reduction in female caninesize. This reduction in female canine sizecould be associated with the development ofeither sex-linked modifier genes that re-strict the expression of the trait in females(Rice, 1984) or through the development ofgenes that alter the response of the develop-ment of the trait to sex hormones (Wright,1993). Regardless of the exact genetic mech-anism underlying this process, the modelpredicts that this phase in the evolution ofdimorphism should be relatively slow, possi-bly requiring millions of generations (Lande,1980). Furthermore, the Lande model pre-dicts that if the trait is correlated betweenmales and females, then selection againstdevelopment of the trait in one sex couldactually limit, or constrain, the expressionof the trait in the other sex. In the caninedimorphism example, this latter phenom-enon would suggest that the development ofhypertrophied canines in some male pri-mates could actually be limited by selectionagainst the development of large canines infemales.

Greenfield (1992b, 1996) presents evi-dence that interspecific variation in femalecanine size covaries with that of male caninesize. He interprets this as evidence thatcorrelated response explains most variationin female canine size and that selection forthe development of canines as weapons oper-ates in male primates but not females.1Plavcan et al. (1995) also find a correlationbetween male and female relative caninecrown height but suggest that any effect ofcorrelated response is secondary to that ofselection for the development of female ca-nines as weapons. They note that the wide-

1In support of the correlated response model, Greenfield (1996)claims that Plavcan and van Schaik (1997a) (incorrectly cited asPlavcan and van Schaik, 1996) demonstrate no relation betweenrelative female canine size and the female competition intensityclassification. This citation is incorrect (see Plavcan et al., 1995);no such evidence was found, and no such claim has been made inany publication by Plavcan.

402 J.M. PLAVCAN

spread presence of canine dimorphismamong anthropoid primates implies that thegenetic control of male and female caninesize is decoupled. Even so, unless male andfemale canine tooth size are under com-pletely separate genetic control, some variationin relative female canine size may reflect corre-lated response. Finally, strepsirhines are char-acterized by relatively slight canine dimor-phism, regardless of mating system (Kappeler,1996). Correlated response, therefore, mightexplain the lack of canine dimorphism in strep-sirhines.

The results of the studies of Plavcan et al.(1995) and Greenfield (1992a,b, 1996) ap-pear contradictory. However, neither hasevaluated the joint relation between maleand female canine size and female agonisticcompetition in primates. Thus, the relativeroles of selection for the development ofweaponry and correlated response in theevolution of primate female canine size areunclear. Fortunately, each model generatestestable predictions about the relation be-tween relative male and female canine toothsize. This analysis therefore sets out toevaluate whether the models are in conflictand whether comparative evidence supportseither or both models.

This analysis focuses only on these twomodels. Evaluation of other factors whichmay or may not be related to male or femalecanine size is beyond the scope of this analysis.

ASSUMPTIONS AND PREDICTIONSOF THE MODELS

The correlated response and intrasexualcompetition models make a variety of predic-tions about the relation between male andfemale canine size and estimates of female-female agonistic competition in primates. Inorder to define these predictions, we mustfirst establish the appropriate measure ofcanine size or dimorphism.

The correlated response model specificallyaddresses the evolution of dimorphism overtime within a lineage. It might seem besttherefore to simply use a ratio estimate ofdimorphism to evaluate correlated response.However, ratios cannot reveal whether achange in the magnitude of dimorphismresults from a relative change in the value ofthe male or female trait. This is particularly

important for evaluating the female compe-tition hypothesis, which specifically ad-dresses the evolution of relative female ca-nine size without regard to male canine size(Plavcan et al., 1995). Consequently, thisstudy uses measures of relative male andfemale canine size derived as residuals froman allometric analysis.

A second problem is that the correlatedresponse model refers only to the evolutionof dimorphism within a single lineage. Thedissociation of male and female canine sizecould theoretically proceed at different ratesin different lineages. Ideally, the best way totest the correlated response model is to tracechanges in male and female traits through adetailed fossil record (for an example of thistype of analysis using peccaries see Wright,1993). Consequently, a strong phylogeneticbias could be present in a cross-sectionalanalysis of extant species. For simplicity, thefollowing analysis begins with the assump-tion that the initial dissociation of male andfemale canine size is a primitive trait forprimates and that the rate that geneticcontrol of male and female canine size wasdecoupled was the same in each lineage.This implies that any effect of correlatedresponse is uniform among all species. Theanalysis of Greenfield (1992b, 1996) alsoimplicitly makes this assumption. Furtheranalyses were repeated within lower taxo-nomic groups to check for biases.

In the absence of any correlated response,the competition model predicts simply thatspecies showing either low intensity or coali-tionary agonistic female-female competitionshould be characterized by relatively smallerfemale canines than those showing high-intensity female-female competition (thedefinitions of competition are briefly pre-sented below, and the reader is referred toPlavcan et al. (1995) for an extensive discus-sion of the definitions and classifications offemale agonistic competition).

Figure 1 illustrates a series of predicted,idealized distributions of relative male andfemale canine size considering both the cor-related response and competition modelsjointly. The panels show six hypotheticaldistributions of relative male and femalecanine tooth size (derived from a comparisonto some measure of overall size), plotted on

403PRIMATE FEMALE CANINE SIZE

Figure 1. Six plots showing the predicted relation between male and female relative canine size andfemale behavioral classifications. Data and distributions are hypothetical. Crosses indicate speciesclassified as high-intensity female-female competition, and zeros indicate species classified as low-intensity female competition. See text for a discussion of each plot.

the x- and y-axes, respectively. Points repre-sented by a cross indicate high-intensityfemale-female agonistic competition species,and those represented by a circle indicatelow-intensity female-female agonistic compe-tition species.

Figure 1A shows the predicted distribu-tion of species if correlated response is theonly mechanism influencing female caninesize. There should be a strong correlationbetween female and male canine size, withno clear relation between female canine sizeand the behavioral classifications. Note thatthe tight correlation depicted in Figure 1Aimplies either relatively uniform degrees ofdimorphism among all species if the slope ofthe relation is isometric or a strong allome-tric effect on dimorphism if the slope devi-ates from isometry.

Figure 1B shows the predicted distribu-tion of species if selection for the develop-ment of weaponry is the only mechanisminfluencing female canine size, with no corre-lated response. In this case, there is nocorrelation between male and female rela-tive canine size, and the high-intensity fe-male competition species have relativelylarger canines than the low-intensity femalecompetition species.

Figure 1C shows the predicted distribu-tion of species if both mechanisms equallyaffect female canine size. Note that withineach cluster corresponding to the competi-tion classifications there is a strong correla-tion between male and female relative ca-nine size. However, for any given male caninesize, females classified as showing high-intensity competition have relatively largercanines than those classified as showinglow-intensity competition.

Figure 1A–C assumes that there are nocorrelations between male and female intra-sexual competition classifications and thatthere are no sexual differences in thestrength of selection for the development ofweaponry. Unfortunately, correlations be-tween male and female behavior, as well assexual differences in the relation betweenintrasexual competition and selection forthe development of large canines, consider-ably complicate the predictions of the model.Figures 1D–F shows predicted distributionsof relative male and female canine size

assuming strong covariation between maleand female competition classifications (i.e.,in each species males and females are bothclassified with either high-intensity or low-intensity competition).

Figure 1D shows a predicted distributionof species if there is a perfect correlationbetween classifications of male and femalecompetition and correlated response affectsfemale canine size. Note that there is astrong correlation between male and femalecanine size both across the competition clas-sifications and within them. This suggeststhat correlated response plays an importantrole in the development of female caninesize. Note that if the analysis were limited tofemale canines only (a univariate analysis offemale canine size on the y-axis), we wouldconclude that the high-intensity femaleshave relatively larger canines than low-intensity females, possibly erroneously cor-roborating the competition hypothesis. Onthe other hand, this pattern could spuri-ously corroborate the correlated responsemodel if selection associated with the maleand female competition classifications wasidentical in nature and was perfectly corre-lated across taxa (see Leroi et al., 1994).This latter situation seems unlikely butcannot be ruled out completely.

Figure 1E shows the predicted distribu-tion of species if classifications of male andfemale behavior are the same but correlatedresponse does not operate. The key observa-tion here is that there is no correlationbetween male and female relative caninesize within the female competition classifica-tions. Therefore, the position of the distribu-tions probably reflects only an interactionbetween male and female behavior, andcorrelated response probably plays little rolein determining female canine size (though itcannot be ruled out entirely).

Finally, Figure 1F shows the predicteddistribution of species similar to Figure 1E,but in this case the effect of selection for thedevelopment of weaponry is unequal be-tween the sexes. In this example, high-intensity male-male competition is associ-ated with large increases in relative malecanine size, while high-intensity female-female competition is associated with onlymodest increases in female canine size. Con-


sequently, the high-intensity female clusteris shifted to the right in the plot.

The above hypotheses can be tested in arelatively straightforward fashion usingsimple correlation and regression analyses.

MATERIALS AND METHODSData

Data on average male and female maxil-lary canine crown height for 108 specieswere taken from Plavcan (1990) for anthro-poid primates and Kappeler (1996) for strep-sirhine primates. Among canine dimen-sions, maxillary canine crown height is moststrongly correlated with male and femalebehavioral variables (Plavcan, 1993; Plav-can and van Schaik, 1992; Plavcan et al.,1995; Greenfield, 1992a,b,c, 1996). Furtherdetails of the exact measurements andsamples of species measured are presentedin Plavcan (1990), Plavcan and van Schaik(1992), and Plavcan et al. (1995).

Adult body mass data were abstractedfrom Plavcan and van Schaik (1997b) andSmith and Jungers (1997). These two sourcespresent body mass data on wild-caught speci-mens abstracted from the primary litera-ture. Data on M/1 tooth lengths were avail-able only for the anthropoid sample (Plavcan,1990).

Measurement of relative canine size

The measurement of relative male andfemale canine size for the purposes of thisanalysis presents a difficult problem. Astraightforward regression of female againstmale canine size might seem logical, butsuch a procedure is confounded by interac-tions between male canine size, male bodysize, male intrasexual agonistic competi-tion, and phylogeny (see Greenfield andWashburn, 1991; Mitani et al., 1996; Plav-can, 1993; Plavcan and van Schaik, 1992,1997b). A majority of the species showingthe highest degrees of male-male competi-tion and the largest canines are also thelargest species. Monomorphic species, onthe other hand, occur across a wide range ofbody sizes. This results in a skewing of theprimate distribution at the large end of thesize range when female canine size is plottedagainst male canine size (Fig. 2). Conse-

quently, variation in a residual measure offemale canine size derived from a directregression against male canine size willreflect primarily variation in male caninesize in dimorphic species. Therefore, malecanine size alone is a poor baseline fromwhich to establish an allometric criterion ofsubtraction for measuring relative femalecanine size.

Instead, male and female relative caninesizes were estimated as the least-squaresresidual deviation from an isometric linepassed through a comparison of canine sizeto either body mass or the mesiodistal lengthof M/1. In both cases, the isometric line waspassed through pooled male and female datasets, treating males and females as indepen-dent points (i.e., both males and femaleswere compared to the same isometric line).The isometric line is preferred over an em-pirical line following the recommendationsof Smith (1980), Plavcan (1993), and Plav-can et al. (1995) and corresponds to a simpleratio of canine size and the independentvariable.

Behavioral classifications

Classifications of male and female intra-sexual competition were taken from Plavcanet al. (1995), to which the reader is referredfor detailed definitions and discussion. Inshort, agonistic, escalated competition (fight-ing) in either males or females is classifiedon the basis of intensity, potential frequency,and context (coalitionary competition).

The intensity and potential frequency cat-egories are dichotomized into high and lowclasses. High-intensity and high-frequencyspecies are predicted to have larger caninesthan low-intensity and low-frequency spe-cies, respectively.

High-intensity contest competition is rec-ognized when either adults are intolerant ofone another or stable dominance hierarchiesare maintained through agonistic interac-tions. Low-intensity competition is recog-nized when individuals are reported to betolerant of one another, dominance hierar-chies are difficult to detect, and escalatedintrasexual competition is reported as rareor absent. To render the analysis conserva-tive, a sex of a species was classified as highintensity whenever escalated agonistic com-

406 J.M. PLAVCAN

petition was reported as common amongadults of the same sex. High-frequency com-petition is recognized when more than oneadult of a sex typically occurs in a group,while low-frequency competition is recog-nized when only one adult of a sex typicallyoccurs in a group.

Plavcan et al. (1995) hypothesize thatwhen agonistic competition regularly occursbetween alliances, or coalitions, of individu-als, selection for the development of weap-onry should be reduced in comparison to the

situation where the outcome of contests isdetermined by individuals acting alone. Thisis based on the hypothesis that, where com-petition occurs between groups of individu-als, the number of individuals enlisted in acoalition will be as important, if not more so,than individual weaponry in determiningthe outcome of contests. Such coalitionarycompetition is rare among males but com-mon among female anthropoids.

The coalitionary competition category isdivided into coalitions present (coalitionary

Figure 2. Bivariate plots of ln-transformed male and female maxillary canine crown height for 108species of primates. An isometric line is fitted through monomorphic species.


competition) or absent (noncoalitionary com-petition). Males or females of species show-ing coalitionary competition are specificallypredicted to have relatively smaller caninesthan high-intensity, high-frequency competi-tion species. Notably, the males of severalspecies are classified as showing coalition-ary competition. These species show coali-tionary competition only in between-groupconflicts and otherwise would be classifiedas showing low-intensity competition (Plav-can et al., 1995). The classification of thesespecies as coalitionary renders any test ofthe coalitions hypothesis conservative.

Behavioral classifications are not avail-able for the entire data set. Therefore,whereas the correlation between male andfemale canine size can be evaluated for all108 samples, there are only 66 species forwhich competition classifications are avail-able for either or both males and females.For a number of species, behavioral classifi-cations could be made for only one sex. Thus,sample sizes vary among the analyses.

Plavcan et al. (1995) demonstrate that,while relative male canine size is associatedwith variation in all three competition types,relative female canine size is significantlyassociated only with the intensity and con-text of competition but not the frequency.For this reason, the analysis presented be-low considers only the effects of female com-petition intensity and context on relativefemale canine size.

Data analyses

Data analysis is presented in two stages.First is a straightforward test of the hypoth-esis that male and female canine size arecorrelated. The analysis is carried out for allspecies values and controlling for phyloge-netic effects. Phylogeny is controlled in twoways: first by reanalyzing data in progres-sively lower taxonomic groups and, second,using the phylogenetic contrasts method(Felsenstein, 1985; Pagel and Harvey, 1988;Pagel, 1992), following the algorithms andrecommendations set forth in Pagel (1992)and setting branch lengths equal to 1 in theabsence of reliable data on branch lengths.Phylogenetic contrasts were calculated byhand. Phylogenies used for the analysiswere drawn from Purvis (1995). Unresolved

nodes in the Purvis phylogeny were resolvedusing phylogenies or taxonomies from Na-pier (1985) for Presbytis and Rumpler andDutriaeux (1986) for lorises and galagoes.Phylogenetic contrasts were also calculatedusing phylogenies for papionines from Stras-ser and Delson (1987), for cercopithecinesfrom Ruvolo (1988), for platyrrhines fromFord (1986), Kay (1990), Rosenberger (1981),and Schneider et al. (1993), and for strepsi-rhines from Rumpler and Dutriaeux (1986)and Stanger and Macedonia (1992). Resultsdid not differ substantially using the alterna-tive phylogenies.

Next the pattern of male and female rela-tive canine tooth size is evaluated againstthe predictions of the correlated responseand selection models outlined in Figure 1(see above). Phylogenetic control was diffi-cult in these analyses. Meaningful statisticscould not be generated for phylogenetic con-trasts because variation in male and femalebehavioral classifications did not allowenough meaningful contrasts between nodeswithin the phylogeny. Reanalysis of the datawithin lower taxonomic levels was, for themost part, limited by small sample sizes.The lack of strong phylogenetic control inthe analysis of the behavioral correlationsmeans that phylogeny cannot be ruled out asa potentially confounding factor.

RESULTSCorrelation between male

and female canine size

Across the whole primate sample, ln-transformed estimates of male and femalecanine size are strongly correlated with oneanother (N 5 108, r 5 0.909, P , 0.001)(Fig. 2). This is not surprising because of thelarge range of overall sizes encompassed bythe sample.

Both measures of residual canine crownheight show strong correlations betweenmale and female values (Table 1; Fig. 3). Theonly large samples showing no correlationbetween male and female residual caninesize are cercopithecines (N 5 26), but onlyfor residuals derived from a comparison tobody mass, and colobines (N 5 18), but onlyfor residuals derived from a comparison toM/1 size. Two small samples also fail toachieve statistical significance (hylobatids

408 J.M. PLAVCAN

(N 5 5) and callitrichids (N 5 4), but thelatter only for residuals from the compari-son to M/1 length).

The phylogenetic contrast method yields acorrelation of r 5 0.668 (N 5 92) for thecomparison of male and female residualcanine size. This correlation is stronger thanthat derived for the species values and dem-onstrates that the correlation between maleand female canine size is not a taxonomicartifact. The sample size is less than 108because of unresolved nodes in the phylog-eny.

Joint effect of female competitionand male canine size

There is a strong interaction betweenmale and female competition classifications(Table 2). Specifically, because the males ofmost anthropoid species are classified asshowing high-intensity competition, mostanthropoid species classified as showinghigh-intensity female competition also showhigh-intensity male competition. Consistentwith this observation, species classified asshowing either high-intensity or coalition-ary female competition show significantlygreater relative male canine size than thosecharacterized by low-intensity female compe-tition (Table 3). This interaction between

male and female competition classificationsmeans that the relations between male andfemale canine size and female competitionclassifications should resemble the distribu-tions in Figure 1D–F.

Male and female relative canine size arecorrelated in the low-intensity female compe-tition classification (Table 4) using residualsderived from either a comparison to bodymass or M/1 size but are correlated withincoalitionary species using only residuals de-rived from a comparison to M/1 size. Withinthe high-intensity competition classifica-tion, there is no significant correlation be-tween male and female relative canine sizeusing either set of residuals.

Considered separately, anthropoids andstrepsirhines display very different distribu-tions (Fig. 4). Figure 4A shows the anthro-poid distribution of male and female re-sidual canine size (derived from a comparisonto body mass), with a reduced major axis(RMA) regression line passed through thelow-intensity female competition species.Note first that there is a strong correlationbetween male and female residual caninesize within the low-intensity female competi-tion species but none in the high-intensity orcoalitionary female competition species (thisis the reason the RMA line is passed only

TABLE 1. Correlations between female and male residual canine size

Body mass residuals1 M1 residuals2

N r P N r P

All primates 108 0.573 ,0.001 — —Strepsirrhines 31 0.927 ,0.001 — —

Lorisids 13 0.856 ,0.001 — —Lorisines 6 0.986 ,0.001 — —Galagines 8 0.819 0.013 — —

Lemurids 18 0.938 ,0.001 — —Lemurines 11 0.815 0.002 — —Indriids 3 0.999 0.027 — —Cheirogalids 3 0.998 0.041 — —

Anthropoids 77 0.649 ,0.001 79 0.554 ,0.001Catarrhines 54 0.647 ,0.001 57 0.404 0.002

Cercopithecoids 44 0.607 ,0.001 47 0.524 ,0.001Cercopithecines 26 0.278 0.169 29 0.449 0.015Colobines 18 0.712 0.001 18 0.393 0.107

Hominoids 10 0.969 ,0.001 10 0.950 ,0.001Hylobatids 5 0.155 0.803 5 0.176 0.777Great Apes 5 0.987 0.002 5 0.968 0.007

Platyrrhines 23 0.761 ,0.001 22 0.814 ,0.001Cebines 19 0.809 ,0.001 18 0.817 ,0.001Callitrichids 4 0.989 0.011 4 0.927 0.073

1 Least-squares residuals from an isometric line passed through the pooled male and female data on maxillary canine crown heightand body mass.2 Least squares residuals from an isometric line passed through the pooled male and female data on maxillary canine crown heightand the mesiodistal length of the mandibular first molar. Molar tooth data was not available for the strepsirhines.


through the low-intensity species). Next notethat the high-intensity female competitionspecies fall above the RMA line, while themajority of coalitionary female competitionspecies fall below the line. ANOVA of re-sidual values derived from the RMA regres-sion line corroborates this visual evaluation(Table 5). These results suggest that withinthe low-intensity species, relative femalecanine size strongly covaries with male ca-nine size. Within the high-intensity species,

female canine size does not covary with malecanine size, and females have relativelylarger canines than low-intensity femalesfor any given male canine size. The coalition-ary species have relatively small caninescompared to the high-intensity species.

Figure 4B shows the strepsirhine distribu-tion, again with an RMA regression linepassed through the low-intensity female com-petition species. Unlike the anthropoids,there is a significant correlation betweenmale and female canine size in both the low-and high-intensity competition species, and

TABLE 2. Matrices of male (rows) and female(columns) intrasexual competition intensity

classifications within taxonomic groups1

All primates Anthropoids Prosimians

LI HI LI HI LI HI

LI 26 3 3 3 23 0HI 20 29 18 25 2 41 HI, high intensity; LI, low intensity.

TABLE 3. ANOVAs for differences in relative malecanine size (residuals derived from a comparison tomale body mass) among species of different female

competition classifications1

Effect SS DF MS F P

AnthropoidsLI vs HI 0.860 1 0.860 13.528 0.001Error 2.161 34 0.064LI vs. Coal. 1.542 1 1.542 23.801 ,0.001Error 2.073 32 0.065HI vs. Coal. 0.108 1 0.108 2.272 0.144Error 1.240 26 0.048

StrepsirhinesLI vs HI 0.320 1 0.320 6.554 0.016Error 1.320 27 0.049

1 Coal., coalitionary female competition; HI, high-intensityfemale competition; LI, low-intensity female competition.

TABLE 4. Correlations between residual maxillarycanine crown height of females and males within

different classifications of femaleintrasexual competition1

Female classification N r P

Residuals from a com-parison to body mass

Coalitions 13 0.475 0.101LI 46 0.686 ,0.001HI 19 0.231 0.342

Residuals from a com-parison to M1 size

Coalitions 13 0.657 0.015LI 20 0.656 0.002HI 15 0.334 0.224

1 HI, high intensity; LI, low intensity.

Figure 3. Bivariate plots of relative male and femalecanine size in primates. A: Residuals derived from acomparison of ln-transformed male and female maxil-lary canine crown height to an isometric line using bodymass as the independent variable. B: Residuals derivedfrom a similar comparison as A, except mandibularmolar tooth mesiodistal length was substituted for bodymass as the independent variable.

410 J.M. PLAVCAN

the high-intensity female competition spe-cies fall close to the regression line. Thisdistribution closely resembles that pre-dicted in Figure 1D. ANOVA of residualvalues derived from the regression line showsno significant difference between high- andlow-intensity female competition species(Table 5). This suggests that variation inrelative female canine size in strepsirhinescovaries with that of the males.

DISCUSSION

The results of this analysis provide cor-roborative evidence that variation in femalecanine size among primates is a function ofboth correlated response and selection forthe development of weaponry in females.However, anthropoids and strepsirhines dif-fer strongly in the pattern of covariationbetween relative male and female caninesize and the behavioral classifications usedhere. This suggests that the relative influ-ence of correlated response and selection forthe development of female weaponry differsstrongly between these two groups.

Strepsirhines

Interspecific variation in strepsirhine rela-tive female canine size appears to reflectcorrelated response. This is not surprising,since all strepsirhines are characterized bynearly monomorphic canines (Kappeler,1996), even though relative canine size var-ies among species. For example, even thoughVarecia variegata have large projecting ca-nines and Propithecus diadema have shortcanines, both species are nearly monomor-phic.

A clear explanation for strepsirhine mono-morphism has so far evaded comparativeanalyses (Kappeler, 1996). Kappeler (1996)found no relation between variation in strep-sirhine canine dimorphism and a series ofbehavioral/ecological variables. In fact, Kap-peler found that only phylogeny correlateswith the relatively modest variation in strep-

Figure 4. Bivariate plots of residual female vs. re-sidual male maxillary canine crown height in anthro-poids (A) and strepsirhines (B). Small squares representspecies classified as showing low-intensity female compe-tition, open circles represent species classified as show-ing high-intensity female competition, and filled circlesrepresent species classified as showing coalitionary fe-male competition. The line in each panel is a reducedmajor axis line passed through the low-intensity femalecompetition species.

TABLE 5. ANOVAs for differences in relative femalecanine size (measured as a residual from a RMA linepassed through the low-intensity female competition

species in a comparison of relative female canine size vs.relative male canine size) between female

competition classifications1

Effect SS DF MS F P

AnthropoidsLI vs. HI 0.313 1 0.313 11.199 0.002Error 0.949 34 0.028LI vs. Coal. 0.108 1 0.108 5.379 0.027Error 0.642 32 0.026HI vs. Coal. 0.647 1 0.647 16.311 ,0.001Error 1.032 26 0.040

StrepsirhinesLI vs. HI 0.005 1 0.005 0.622 0.437Error 0.226 27 0.008

1 Coal., coalitionary female competition; HI, high-intensityfemale competition; LI, low-intensity female competition.


sirhine canine dimorphism. Kappeler wenton to suggest that sexual selection may notfavor the development of traits such as largebody size or canine size to help males winfights, because male reproductive successdepends on mechanisms other than physicalcombat.

Plavcan et al. (1995) found that femalestrepsirhines ranked as showing high-inten-sity competition tend to have relativelylarger canines than those ranked as low-intensity. However, the analysis presentedhere suggests that this result may be anartifact of the covariation in male and fe-male intensity classifications: where fe-males are classified as high intensity, so tooare males.

It is important to note that male strepsi-rhines classified as showing high-intensitycompetition have significantly larger rela-tive canine size than those classified asshowing low-intensity competition (ANOVA,F 5 4.839, P 5 0.037). This result holdsusing phylogenetic contrasts (in all five pair-wise contrasts between high- and low-intensity males, the high-intensity specieshas larger canines). The fact that malecanine size is correlated with the male-malecompetition classifications suggests, contraKappeler (1996), that sexual selection fromphysical combat might favor the develop-ment of big male canines. If true, thencorrelated response might act as a con-straint on the evolution of canine dimor-phism in strepsirhines. This is the firstanalysis to provide evidence that sexualselection may affect male canine size instrepsirhines and at the same time to sug-gest a mechanism to explain why strepsi-rhine canines are nearly monomorphic.

This interpretation for strepsirhine ca-nine monomorphism is still uncertain. Inthe first place, strepsirhine behavior is rela-tively poorly known, and any of the competi-tion classifications used here may changewith the advent of better data. Furthermore,because male and female competition classi-fications are the same in almost all strepsi-rhine species, the covariation in relativemale and female canine size may reflectequal selective pressures on canine size.While perhaps less likely than the corre-lated response model, this possibility cannot

yet be discounted (for a discussion of thepotential inability of comparative analysisto distinguish between such hypotheses seeLeroi et al., 1994). Finally, left unansweredare why the genetic control for male andfemale canine size in strepsirhines has notdecoupled, as in anthropoids and whetherthe lack of body mass dimorphism in strepsi-rhines is also a function of correlated re-sponse. Such analyses are beyond the scopeof this paper.

Anthropoids

The results of this analysis are consistentwith the hypothesis that anthropoid relativefemale canine size is at least partially afunction of both selection for the develop-ment of weaponry and correlated response.Unlike the strepsirhine distribution, whichneatly fits into the predictions of Figure 1D,the anthropoid pattern of male and femalerelative canine size is complex and appearsto reflect a hybrid of the predictions ofFigure 1C–E.

In the first place, it is notable that thelow-intensity female competition speciesshow a very strong correlation between maleand female relative canine size, suggestingthat correlated response explains much ofthe interspecific variation in these species.In stark contrast, the high-intensity andcoalitionary female competition species showno correlation between relative male andfemale canine size. This seems to suggestthat within the high-intensity and coalition-ary female competition species, correlatedresponse does not explain variation in rela-tive female canine size.

The next question to address is whetherthe high-intensity and coalitionary competi-tion females have relatively larger or smallercanines for any given male canine size thanthe low-intensity females. The lack of corre-lation between male and female canine sizein two of three competition classificationsprecludes the use of analysis of covariance totest this hypothesis. Instead, we must makethe comparison using an RMA regressionline passed through the low-intensity distri-bution (use of a least-squares line does notalter the results). As shown in Figure 4A andTable 5, the high-intensity species fall abovethe line, and the coalitionary species fall

412 J.M. PLAVCAN

below it. This suggests that indeed the high-intensity female species have relativelylarger canines than the low-intensity femalespecies for any given male canine size. Fur-thermore, as predicted by Plavcan et al.(1995), the coalitionary species have rela-tively smaller canines than the high-inten-sity species.

On the whole, these results suggest thatcorrelated response does not operate in con-flict with selection for the development ofweaponry in female anthropoids. It appearsthat in the absence of selection for thedevelopment of weaponry, correlated re-sponse is an important factor influencingthe expression of anthropoid female caninesize. However, if there is selection for thedevelopment of weaponry, correlated re-sponse plays little part in explaining inter-specific variation in female canine toothsize. Thus, in anthropoids, unlike in strepsi-rhines, correlated response does not seem toconstrain the evolution of female caninesize. This result is, of course, consistent withthe presence of strong canine dimorphism inmany anthropoids. Interspecific variation indimorphism alone proves that male andfemale canine size can vary at least partlyindependently in anthropoids. Thus, corre-lated response should not be expected toconstrain the development of large caninesas weapons in some females. The analysisprovides evidence that at least some varia-tion in canine dimorphism reflects selectionacting on female canine size (Plavcan et al.,1995; Harvey et al., 1978; Lucas et al., 1986).

Problems with the models

The results of this analysis must be tem-pered with a careful consideration of theshortcomings of the behavioral classifica-tions, the weakness of comparative analysisfor differentiating between genetic and non-genetic mechanisms affecting character evo-lution, and the fact that the correlated re-sponse model is best tested using data withina single lineage rather than across lineages.

The classifications of female and malecompetition used in this analysis sacrificeprecision in order to allow a broad compara-tive analysis (Plavcan et al., 1995). Impor-tant variation in both male and female

behavior is subsumed within the behavioralclassifications. More refined evaluations ofthe relation between behavior and morphol-ogy may reveal other mechanisms that af-fect both male and female canine size or mayrefine our understanding of the selectivepressures underlying variation in caninedimorphism.

Leroi et al. (1994) argue that comparativeanalyses, such as those used here, cannotdistinguish between competing evolutionarygenetic models for the evolution of traits.For example, as noted above, the pattern ofvariation in strepsirhine relative canine sizecannot distinguish between the correlatedresponse model and the hypothesis thatthere is equal selective pressure on maleand female canine size. For reasons notedabove, the former seems a more likely hy-pothesis. Nevertheless, the results of thisanalysis only corroborate or refute the predic-tions of the two recently proposed modelsand do not in themselves offer proof thateither correlated response or selection affectfemale canine size. Furthermore, where co-variation in female and male canine size isposited to be a result of correlated response,the exact genetic mechanism determiningmale and female covariation in canine size isopen to question, because the analysis evalu-ates only whether male and female caninesizes covary. Thus, while the results of theseanalyses offer corroborative evidence for themodels, they certainly do not offer proof andshould not be taken as definitive evidencefor the operation of either correlated re-sponse or selection for the development ofweaponry in females.

As mentioned previously, the correlatedresponse model specifically addresses theevolution of sexual dimorphism through timewithin a lineage. Consequently, the best testfor the model would be to trace the evolutionof canines in males and females throughtime, demonstrating first the hypertrophy ofthe canines in both sexes and then thesubsequent reduction of canines in females.The fossil record for most extant species istoo poor to allow such a test.

Instead, Greenfield (1992b, 1996), Plav-can et al. (1995), and this analysis addressonly the hypothesis that interspecific varia-


tion in female canine size covaries withinterspecific variation in male canine size. Itis reasonable to suppose that covariation inmale and female canine size across speciesreflects correlated response, but, as justnoted above, neither the mechanism under-lying this covariation nor the exact degree towhich correlated response actually influ-ences canine size in any particular speciescan be demonstrated by this analysis.

Unless the genetic control of male andfemale canine size decoupled in the commonancestor of all anthropoids and the effect onfemale canines is uniform across all anthro-poids, we should expect different effects ofcorrelated response in different lineages.For example, there is an obvious differencein the covariation between male and femalecanine size between strepsirhines and an-thropoids. Likewise, even though this analy-sis suggests that there is an effect of corre-lated response within anthropoids, at thesame time female canines in most anthro-poids are obviously smaller than those ofmales. Thus, female canine size in mostspecies should reflect a balance betweenshared genetic control of male and femalecanine size, the mechanism that decouplesthis control and allows canine reduction infemales, and selection for canine reductionor increase in females. Obviously, the evolu-tion of these features could proceed differ-ently in different lineages.

If an effect of correlated response did varywidely among different lineages, then wemight expect either a lack of correlationbetween male and female canine size acrosslineages regardless of competition classifica-tions or at least taxonomic variation in therelation between male and female caninesize. Obviously, the strong correlation be-tween male and female canine size in bothstrepsirhines and low-intensity female com-petition anthropoids proves that the firstpoint is not true. At the same time, thedifference between strepsirhines and anthro-poids suggests that there are indeed taxo-nomic differences in the expression of corre-lated response. However, within theanthropoid distribution, there is still littleindication that any of the results are anartifact of phylogeny. Within the low-inten-sity female competition species, there is a

mix of hominoids, platyrrhines, cercopithe-cines, and colobines, with no clear taxo-nomic differences in the relation betweenmale and female canine size. Within thehigh-intensity species, these taxa are alsorepresented. Cercopithecine males tend tohave hypertrophied canines, creating a taxo-nomic difference in relative male canine size(Plavcan, 1993), but this is not repeated forrelative female canine size. Within cercopi-thecines, there is still no correlation be-tween male and female canine size withinthe high-intensity female competition spe-cies. Thus, it does not appear that the re-sults of this analysis are an artifact ofvariation in the effect of correlated responseacross taxa. Even so, only an analysis ofchanges in canine size through time withindifferent lineages can directly address thisproblem.

Finally, it should be emphasized that thisanalysis tests broad hypotheses about theinfluence of two factors on the evolution ofrelative female canine size in primates. Thisdoes not by any means exclude other factorsfrom being important in the evolution ofmale or female canine size, either acrossspecies or within individual lineages. Forexample, other studies have suggested thatacross species relative male canine size isinfluenced by predation pressure (e.g., Leu-tenegger and Kelley, 1977; Plavcan and vanSchaik, 1992). It widely held that both maleand female pithecine canines are morphologi-cally specialized for opening hard fruit (e.g.,Greenfield, 1992b; Plavcan and Kelley, 1996).The goal of this analysis is not to explainvariation in all species by two isolated fac-tors, and the results should not be inter-preted this way. Rather, the goal is to evalu-ate whether interspecific variation in relativefemale canine size is consistent with eitheror both hypotheses. With the results pre-sented here, there is still a large amount ofvariation in relative male and female caninesize that is unexplained.

Phylogenetic implications

Recently, Kay et al. (1997) noted thatcanine tooth size sexual dimorphism is prob-ably a derived character for anthropoid pri-mates. This view is not new, and caninedimorphism has been used as evidence that

414 J.M. PLAVCAN

early cercomoniine adapiforms gave rise toanthropoid primates (e.g., Gingerich, 1995;Simons and Rasmussen, 1996). Conversely,Lucas et al. (1986) suggested that earlyanthropoids were characterized by smallcanines in both sexes. They suggested thatinitially sexual selection favored the evolu-tion of hypertrophied male canines, withphylogenetic inertia maintaining large malecanine size in descendent anthropoids.

The correlated response model makes sev-eral predictions about dimorphism in earlyanthropoids pertinent to these views. In thecase of the canine teeth, Lande’s (1980)model clearly predicts that sexual selectionfor large male canines initially should resultin the hypertrophy of both male and femalecanines. Thus, the primitive condition shouldbe analogous to that seen in, for example,Lemur catta, not a condition like Callicebusin which both males and females have smallcanines. Dimorphism should evolve onlylater by reduction of the female trait (thisdoes not mean that male canines cannot befurther enlarged or reduced). If this model istrue, then canine reduction in female anthro-poids might be the shared derived characteramong anthropoids, not necessarily dimor-phism arising through hypertrophy of themale canine.

It is well known that a number of adapidprimates show substantial canine dimor-phism (Krishtalka et al., 1990; Gingerich,1981, 1995; Simons and Rasmussen, 1996).Conversely, canine dimorphism has not beendocumented in Omomyid primates. If Omo-myids are the sister taxon to anthropoids(Kay et al., 1997), then this might constituteevidence that canine dimorphism has evolvedin parallel in at least two primate lineages.

This model also suggests that the en-larged canines of some female anthropoids,including hylobatids, callitrichids, and sev-eral subspecies of Colobus guereza, are sec-ondarily derived (see Lucas et al., 1986).Colobus guereza is particularly interesting.Variation in female canine size among thesubspecies of this species is tremendous,with C. g. occidentalis possessing nearlyhylobatid-like canines (Hayes et al., 1995;Lucas et al., 1986). Regardless of the under-lying mechanisms driving the evolution oflarge canines in female C. guereza (Hayes et

al., 1995), the fact that there is large varia-tion in female canine size without parallelvariation in male canine size suggests thatany effect of correlated response in thisspecies does not constrain the evolution offemale canine size.

CONCLUSIONS

Two hypotheses have recently been for-warded to explain interspecific variation infemale canine size in primates: correlatedresponse to male canine size (Greenfield,1992b, 1996) and selection for the develop-ment of weaponry in females (Plavcan et al.,1995). The operation of these two mecha-nisms is not mutually exclusive, and theresults of this analysis suggest that bothmechanisms operate simultaneously in an-thropoid primates. When male canine sizewas controlled for, females classified by Plav-can et al. (1995) as showing high-intensity,noncoalitionary competition show relativelylarge canine size, with no correlation be-tween male and female canine size. Con-versely, those females classified as showingeither low-intensity agonistic competition orcoalitionary competition possess on averagerelatively small canines as compared to high-intensity female competition species. Withinthe low-intensity female-female competitionclassification there is a strong correlationbetween female and male relative caninesize. This suggests that, overall, correlatedresponse is important in explaining femalecanine size only in those species where thereis little or no selection for the development ofweaponry.

Among strepsirhines (which show low de-grees of canine dimorphism), male and fe-male relative canine sizes are strongly corre-lated. At the same time, male canine size iscorrelated with intensity of male-male com-petition. This suggests that sexual selectionmay favor the development of large caninesin some male strepsirhines, while at thesame time the expression of canine dimor-phism is constrained by correlated response.

This analysis reconciles the seeminglyconflicting findings of Greenfield (1992a,b)and Plavcan et al. (1995), demonstratingthat both models are probably partially cor-rect.


ACKNOWLEDGMENTS

I thank Larry Witmer, Nikos Solounias,and Margeret Lewis for helpful discussions.Margeret Lewis, Carel van Schaik, and ScottSampson provided valuable comments on anearlier draft of this paper. Two anonymousreviewers also provided suggestions thatgreatly improved the manuscript.

LITERATURE CITED

Felsenstein, J. 1985. Phylogenies and the comparativemethod. Am Nat 125:1–15.

Ford SM. 1986. Systematics of the New World monkeys.In: Swindler DW, Erwin J, editors. Comparative pri-mate biology, vol. 1: systematics, evolution andanatomy. New York: Alan R. Liss. p 73–135.

Gingerich PD. 1981. Cranial morphology and adapta-tions in Eocene Adapidae. I. Sexual dimorphism inAdapis magnus and Adapis parisiensis. Am J PhysAnthropol 56:217–234.

Gingerich PD. 1995. Sexual dimorphism in EarliestEocene Cantius torresi (Mammalia, Primates, Adapoi-dea). Contributions from the Museum of Paleontology:The University of Michigan 29:185–199.

Greenfield LO. 1992a. A nonadaptive dental trait. Int JPrimatol 13: 631–657.

Greenfield LO. 1992b. Origin of the human canine: anew solution to an old enigma. Yrbk Phys Anthropol35:153–184.

Greenfield LO. 1992c. Relative canine size, behavior,and diet in male ceboids. J Hum Evol 23:469–480.

Greenfield LO. 1996. Correlated response of homologouscharacteristics in the anthropoid anterior dentition. JHum Evol 31: 1–19.

Greenfield LO, Washburn A. 1991. Polymorphic aspectsof male anthropoid canines. Am J Phys Anthropol84:17–34.

Harvey PH, Kavanagh M, Clutton-Brock TH. 1978.Sexual dimorphism in primate teeth. J Zool Lond186:475–485.

Hayes VJ, Freedman L, Oxnard CE. 1995. The differen-tial expression of dental sexual dimorphism in subspe-cies of Colobus guereza. Int J Primatol 16:971–996.

Kappeler PM. 1996. Intrasexual selection and phyloge-netic constraints in the evolution of sexual caninedimorphism in strepsirhine primates. J Evol Biol9:43–65.

Kay RF. 1990. The phyletic relationships of extant andfossil Pitheciinae (Platyrrhini, Anthropoidea). J HumEvol 19:175–208.

Kay RF, Ross C, Williams BA. 1997. Anthropoid origins.Science 275:797–804.

Krishtalka L, Stucky RK, Beard KC. 1990. The earliestfossil evidence for sexual dimorphism in primates.Proc Natl Acad Sci U S A 87:5223–5226.

Lande R. 1980. Sexual dimorphism, sexual selection andadaptation in polygenic characteristics. Evolution 34:292–307.

Leroi AM, Rose MR, Lauder GV. 1994. What does thecomparative method reveal about adaptation? Am Nat143:381–402.

Leutenegger W, Kelly JT. 1977. Relationship of sexualdimorphism in canine size and body size to social,behavioral, and ecological correlates in anthropoidprimates. Primates 8:117–136.

Lucas PW. 1981. An analysis of canine size and jawshape in some Old and New World non-human pri-mates. J Zool Lond 195:437–448.

Lucas PW, Corlett RT, Luke DA. 1986. Sexual dimor-phism of tooth size in anthropoids. Hum Evol 1:23–39.

Mitani JC, Gros-Louis J, Richards AF. 1996. Sexualdimorphism, the operational sex ratio, and the inten-sity of male competition in polygynous primates. AmNat 147:966–980.

Napier PH. 1985. Catalogue of Primates of the BritishMuseum (Natural History) and Elsewhere in theBritish Isles Part III: Family Cercopithecidae, Subfam-ily Colobinae. London: British Museum (Natural His-tory).

Pagel MD. 1992. A method for the analysis of compara-tive data. J Theor Biol 156:431–442.

Pagel MD, Harvey PH. 1988. Recent developments in theanalysis of comparative data. Q Rev Biol 63:413–440.

Plavcan JM. 1990. Sexual dimorphism in the dentitionof extant anthropoid primates. PhD dissertation. AnnArbor, MI: University Microfilms.

Plavcan JM. 1993. Canine size and shape in male anthro-poid primates.Am J PhysAnthropol 92:201–216.

Plavcan JM, Kelley J. 1996. Evaluating the ‘‘dual selec-tion’’ hypothesis of canine reduction. Am J PhysAnthropol 99:379–387.

Plavcan JM, van Schaik CP. 1992. Intrasexual competi-tion and canine dimorphism in anthropoid primates.Am J Phys Anthropol 87:461–477.

Plavcan JM, van Schaik CP. 1997a. Interpreting homi-nid behavior on the basis of sexual dimorphism. JHum Evol 32:345–374.

Plavcan JM, van Schaik CP. 1997b. Intrasexual competi-tion and body weight dimorphism in anthropoid pri-mates. Am J Phys Anthropol 103:1–27.

Plavcan JM, van Schaik CP, Kappeler PM. 1995. Compe-tition, coalitions and canine size in primates. J HumEvol 28:245–276.

Purvis A. 1995. A composite estimate of primate phylog-eny. Philos Trans R Soc Lond [Biol] 348:405–421.

Rice WR. 1984. Sex chromosomes and the evolution ofsexual dimorphism. Evolution 38:735–742.

Rosenberger AL. 1981. Systematics: the higher taxa. In:Coimbra-Filho AF, Mittermeier RA, editors. Ecologyand behavior of Neotropical primates, vol. 1. Rio deJaneiro: Academia Brasileira de Ciencias. p 9–27.

Rumpler Y, Dutriaeux B. 1986. Evolution chromosomi-que de prosimiens. Mammal. 50:82–107.

Ruvolo M. 1988. Genetic evolution in the African gue-nons. In: Gautier-Hion A, Bourliere F, Gautier J, editors.A primate radiation: evolutionary biology of the Africanguenons. New York:Alan R. Liss. p 127–139.

Schneider H, Schneider MPC, Sampaio I, Harada ML,Stanhope M, Czelusniak J, Goodman M. 1993. Molecu-lar phylogeny of the New World monkeys (Platyrrhini,Primates). Mol Phyl Evol 2:225–242.

Simons EL, Rasmussen DT. 1996. Skull of Catopithecusbrowni, an early Tertiary catarrhine. Am J PhysAnthropol 100:261–292.

Smith RJ. 1980. Rethinking allometry. J Theor Biol87:97–111.

Smith RJ. 1981. Interspecific scaling of maxillary caninesize and shape in female primates: relationships tosocial structure and diet. J Hum Evol 10:165–173.

Smith RJ, Jungers WL. 1997. Body mass in comparativeprimatology. J Hum Evol 32:523–559.

Stanger KF, Macedonia JM. 1992. Phylogenetic affini-ties among the extant Malagasy lemurs (Lemurifor-mes). Paper presented at the XIVth Congress of theInternational Primatological Society, Strasbourg.

Strasser E, Delson E. 1987. Cladistic analysis of cercopi-thecid relationships. J Hum Evol 16:81–99.

Wright DB. 1993. Evolution of sexually dimorphic char-acters in peccaries (Mammalia, Tayassuidae). Paleobi-ology 19:52–70.

416 J.M. PLAVCAN

correlated response, competition, and female canine size in primates

Documents