American Journal of Primatology 67:177–198 (2005) RESEARCH ARTICLE Personality Traits in Captive Lion-Tailed Macaques (Macaca silenus) JACQUELINE H. ROUFF1n, ROBERT W. SUSSMAN1, and MICHAEL J. STRUBE2 1 Department of Anthropology, Washington University in St. Louis, St. Louis, Missouri 2 Department of Psychology, Washington University in St. Louis, St. Louis, Missouri Personality influences an individual’s perception of a situation and orchestrates behavioral responses. It is an important factor in elucidating variation in behavior both within and between species. The major focus of this research was to test a method that differs from those used in most previous personality studies, while investigating the personality traits of 52 captive lion-tailed macaques from four zoos. In this study, data from behavioral observations, a P-type principal components analysis (PCA), and bootstrapped confidence intervals as criteria for judging the significance of factor loadings were used rather than subjective ratings, R-type factor analyses, and arbitrary rules of thumb to determine significance. We investigated the relationships among individual component scores and sex, hormonal status, and dominance rank (controlling for age and social group) using a multiple regression analysis with bootstrapped confidence intervals. Three personality dimensions emerged from this analysis: Component 1 contained Extraversion-like behaviors related to sociability and affiliativeness. The higher mean Component score for females suggests that they are more ‘‘extraverted’’ than males. Only agonistic behaviors were significantly related to component 2. High-ranking individuals exhibited higher mean Component 2 scores than mid- or low-ranked individuals. Bold and cautious behaviors both loaded positively on Component 3, suggesting a dimension related to curiosity. The mean Component 3 score for females was higher than the mean score for males. The method used in this study should facilitate intraspecific and general interspecific comparisons. Developing a standardized trait term list that is applicable to many species, and collecting trait term data in the same manner and concurrent with behavioral observations (and physiologic measures when feasible) could prove useful in primate research and should be explored. Am. J. Primatol. 67:177–198, 2005. r 2005 Wiley-Liss, Inc. Contract grant sponsor: National Science Foundation; Contract grant number: 9816165; Contract grant sponsor: Sigma Xi; Contract grant sponsor: Scientific Research Society; Contract grant sponsor: Washington University in St. Louis. n Correspondence to: Jacqueline Rouff, 1633 Shepard Road, Glencoe, MO 63038. E-mail: firstname.lastname@example.org Received 22 April 2004; revised 15 February 2005; revision accepted 16 February 2005 DOI 10.1002/ajp.20176 Published online in Wiley InterScience (www.interscience.wiley.com). r 2005 Wiley-Liss, Inc. 178 / Rouff et al. Key words: personality; personality traits; lion-tailed macaques; Macaca silenus INTRODUCTION Personality–identifiable styles of behavior that differentiate one individual from another–reflects the dynamic organization within the individual of the psychobiological systems that modulate adaptation to a changing environment [Allport, 1937; Svrakic et al., 1996]. Researchers who use the ‘‘trait approach’’ view human personality as consisting of a limited number of bipolar dimensions. These are the stable, superordinate factors that influence an individual’s perception of a situation and orchestrate the behavioral responses [Capitanio, 1999; Capitanio et al., 1999]. The trait approach to personality is particularly appropriate for nonhumans because one can assess personality traits by measuring observable behaviors, such as sociability and aggression. In addition, many phenotypic personality traits appear to have a genetic component and/or are associated with individual differences in emotional experience in humans, which has led to a growing recognition that major personality traits represent basic psychobiological dimensions of temperament [Clark & Watson, 1999; Gillespie et al., 2003; Watson, 2000]. Understanding individual differences in personality traits would increase our knowledge of variation both within and between primate species. Primate researchers studying diverse topics have suggested that personality traits are important factors in explaining the results of their studies. For example, personality has been used to explain why certain individuals gain dominance [Burton, 1992; Huff, 2002; McGuire et al., 1994; Pavelka, 1997; Rhine & Maryanski, 1996; Saltzman et al., 1996; Sapolsky, 1990; Virgin & Sapolsky, 1997; Zumpe & Michael, 1996], differences in maternal behavior [Bahr, 1995; Bard, 1995; Brock, 1998; Fairbanks, 1996], grooming preferences and activity [Borries et al., 1994], and an animal’s approach to learning in studies of cognition [Boysen, 1994]. The personality of available males may influence female mate choice [Goodall, 1986; Keddy-Hector, 1992; Price, 1990; Small, 1989; Smuts, 1985, 1987]. Certain personality traits predict cortisol levels in male baboons and differences in response to stress and immune response in baboons and macaques [Capitanio & Mendoza, 1995; Capitanio et al., 1994; Kemeny & Laudenslager, 1999; Lilly, 1995; Sapolsky, 1990, 1991, 1999; Sapolsky & Ray, 1989; Suomi, 1991]. In addition, understanding individual differences in personality can enhance our ability to study other determinants of primate behavior. The omission of individual differences in personality relegates it to the error term, which can make it harder to detect the effect of other key independent variables [Chamove et al., 1972]. For example, failure to account for personality differences may cloud the results of stress studies, because the effects of a particular stressor may vary with the characteristics of the individuals involved. The results of numerous studies indicate a link between several traits and the monoaminergic neurotransmitter systems. Serotonin levels are known to correlate positively with affiliation, and negatively with overly aggressive and impulsive behaviors (see Higley and Bennett  and Higley et al.  for summaries of this research). The noradrenergic system appears to be related to differences in reactivity to novel or challenging circumstances [Suomi, 1991]. Species differences have been noted in behavioral and/or physiological responses Lion-Tailed Macaque Personality Traits / 179 to novel or challenging stimuli [Bernstein et al., 1963; Clarke et al., 1994, 1995; Singh & Manocha, 1966; Vitale et al., 1991], as well as in levels of CSF 5-HIAA and affiliative and aggressive behaviors [Champoux et al., 1997; Higley & Bennett, 1999]. A better understanding of the biological basis and heritability of personality traits and formative environmental factors could help us understand how individuals are constrained by evolved predispositions, and how these are related to social systems [Boinski, 1999; Clarke & Boinski, 1995; Steklis, 1993]. For example, do personality traits differ between female-bonded and non-femalebonded societies? Are differences between female and male dispersal patterns related to personality traits? Developing reliable methods to facilitate within- and cross-species comparisons is essential for nonhuman primate personality research, because the potential contribution to primatology of such research remains largely untapped. Methodological Differences From Past Studies The purpose of this research was to test a method that differs from those used in most previous personality studies, while investigating the personality traits of the lion-tailed macaque (Macaca silenus) and the relationships of sex, hormone status, and dominance rank to these traits. In this study, observed behaviors were used as data, which were analyzed with a P-type principal components analysis (PCA). Bootstrapped confidence intervals were then constructed to evaluate which behaviors were significantly associated with a component. This method differs from those used in most previous studies, which used rules of thumb to determine the significance of factor loadings, as well as R-type factor analytic methods, and which relied primarily on subjective assessments. With subjective assessments, each monkey or ape is rated on adjective trait terms (such as sociable, excitable, and popular), most often only once or one time per year, which makes these ratings vulnerable to memory biases. Both subjective rating and behavioral coding methods have their advantages and disadvantages, and behavior-based studies are not bias-free. Coding methods can introduce bias through the behaviors chosen to represent traits and the methods of observation and coding selected. However, relative to rating data, behavioral data have a more absolute metric and are less vulnerable to framing effects, and can provide meaningful comparisons among individuals, groups, and species–provided that equivalent species-specific behaviors are compared. The behavioral data in this study were analyzed with a P-type PCA rather than the R-type used in past studies. The P-type approach provides a perspective on personality that is a valuable alternative to that produced by the R-type analysis. In an R-type analysis, the data for each trait or behavioral category are averaged across the study for each subject. A single correlation matrix is calculated that represents covariation of behaviors across the sample of animals, and this matrix is then subjected to a PCA. In a P-type analysis, the data for each individual are not averaged across sample periods. Rather, a correlation matrix for each individual is calculated using the sample periods for each monkey in place of the aggregated data of each subject (as in the R-type analysis). In this study, the correlation matrix for each individual indexed the covariation of behaviors across the sample periods for that individual. These individual matrices were then averaged to form the correlation matrix for the PCA. In this sense, the P-type approach used in this study is best thought of as a combination of idiographic and nomothetic approaches, in that the goal of averaging the 180 / Rouff et al. individual animal matrices is to allow inferences to be made about the average animal. The key difference between the two approaches results from the nature of the correlations that enter the PCA, and consequently the nature of the components that emerge. In the P-type analysis, the correlations in the PCA reflect the typical covariation of behaviors across samples. In the R-type analysis, the correlations represent the typical covariation of behaviors across animals. A positive correlation in the R-type approach means that when a particular animal exhibits a high amount of X (averaged over situations), that same animal also tends to exhibit a high amount of Y (averaged over situations). However, that correlation does not provide any information about the covariation of these traits or behaviors within each individual across situations. By contrast, a positive correlation in the P-type approach means that when a particular behavior occurs frequently across situations, another behavior tends to occur frequently across situations as well, for the average animal. The difference in the two approaches explains why traits from nomothetic analyses yield information on trends or tendencies to behave in certain ways, but are poor at predicting or describing momentary behaviors [Fleeson, 2004; Mischel, 1999]. The relationship between different traits within an individual yields important information about personality that cannot be obtained through nomothetic comparisons alone [Allport, 1937]. An R-type analysis is certainly an appropriate choice for investigating the structure of personality. However, important information about personality can be lost that might otherwise be captured if idiographic and nomothetic techniques are integrated, as they are in the P-type analysis used in this study [Allport, 1937; Fleeson, 2001, 2004; Mischel, 1999; Pelham, 1993]. The P-type analysis is a good choice when the goal is to describe personality with the patterns of behaviors over time for individuals. One problem with factor-analysis techniques in primate research is that it is often difficult to obtain the large sample sizes required to obtain stable results. Researchers in many earlier studies used too few subjects for the number of variables, which can produce factors that fail to replicate, and large factor loadings by chance alone. To decrease the possibility that we would interpret chance factor loadings as meaningful in this study, we constructed confidence intervals for the factor loadings using bootstrapping techniques rather than relying on rules of thumb. MATERIALS AND METHODS The traits that were included in this study were drawn from previous nonhuman primate personality research, as well as from aspects of two popular human personality measures: the ‘‘Big Five Factor Model’’ and Cloninger’s Temperament and Character Inventory (TCI) [Cloninger et al., 1994]. We selected two temperament factors (Persistence and Novelty Seeking) from the TCI, and three factors (Extraversion/Introversion, Neuroticism/Emotional Stability, and Agreeableness) from the Big Five Factor Model. The traits associated with these factors appeared similar to traits that were of interest to researchers in previous nonhuman primate studies, and could also be represented by behaviors commonly observed in captive macaques. Many studies of nonhuman primates and other species (e.g., rats [GarciaSevilla, 1984], guppies [Budeav, 1997], cats [Gosling, 1999; Saxton et al., 1987], octopuses [Mather & Anderson, 1993], hyenas and dogs [Gosling, 1999], and nonhuman primates [Bard & Gardner, 1996; Bolig et al., 1992; Capitanio, 1999; Lion-Tailed Macaque Personality Traits / 181 Chamove et al., 1972; Dutton et al., 1997; Gold & Maple, 1994; King & Figueredo, 1997; Laudenslager et al., 1999; McGuire et al., 1994; Mondragon-Ceballos et al., 1991; Mondragon-Ceballos & Santillan-Doherty, 1994; Murray, 1996; StevensonHinde et al., 1980a; van Hooff, 1970]) have found some evidence of factors that appear similar to some or all of the three Big Five Factors selected for this study. In humans, these three factors have shown cross-cultural stability and heritability [McCrae & Costa, 1997; Rowe, 1994; Steen, 1996]. There is also evidence of a biological basis and heritability in humans for Novelty Seeking and Persistence, the two factors selected from the TCI [Gusnard et al., 2003; Youn et al., 2002]. Trait Adjectives and Ethogram The human personality models and nonhuman primate personality factors found in previous studies aided us in selecting traits for this study that were important and of interest to personality researchers. Five Extraversion facets (gregariousness, activity, warmth, seeks excitement, and assertiveness), two Neuroticism facets (anxiety and vulnerability), and four adjective pairs for the domain of Agreeableness (aggressive/protective, agonistic/gentle, suspicious/ trusting, and insensitive/sensitive) were selected. Facets such as positive emotions (cheerful vs. serious) from Extraversion, and depression (hopeless vs. hopeful) from the Neuroticism factor were omitted because these required us to infer an emotional state that seemed difficult to assess from observed behaviors. Facets such as impulsiveness were not included, because a behavioral measure could not be found that was applicable to all captive groups and was feasible in a zoo setting (where researchers must accommodate zoo policies and keeper routines). The adjectives ‘‘welcome’’ (others allow focal animal to approach) and ‘‘avoided’’ (others move away when the focal animal approaches) were added as a measure of whether other group members found the focal animal agreeable or disagreeable. We included aspects of Cloninger’s  TCI factors by adding ‘‘persistence’’ and defining bold and cautious behaviors as ‘‘response to novelty.’’ We adapted definitions of selected traits for use with nonhuman primates using Stevenson-Hinde et al.’s [1980b] adjectives and definitions whenever possible. Specific behaviors to represent trait adjectives were selected from an ethogram of the lion-tailed macaque in captivity [Johnson, 1985; Skinner & Lockard, 1979], from Stevenson-Hinde et al.’s [1980b] behavioral definitions of traits for rhesus macaques (when available and appropriate), or based on behavioral observations of two captive groups of lion-tailed macaques at the St. Louis Zoo. The names of several trait adjectives were changed to terms that are frequently used in other primate studies, or to link trait terms more closely to their representative ethogram behaviors. The following substitutions were made: ‘‘proximity’’ for ‘‘gregariousness,’’ ‘‘affiliativeness’’ for ‘‘warmth,’’ ‘‘response to novelty’’ for ‘‘seeks excitement,’’ ‘‘dominance’’ for ‘‘assertiveness,’’ ‘‘reactivity’’ for ‘‘vulnerability,’’ ‘‘intervenes’’ for ‘‘protective,’’ ‘‘moves away’’ for ‘‘suspicious,’’ ‘‘approachable’’ (i.e., permits approach) for ‘‘trusting,’’ and ‘‘willing to interact’’ for ‘‘gentle.’’ To check the accuracy and reliability of the ethogram, three naive observers independently coded videotaped lion-tailed macaque behaviors. The ethogram definitions were revised until the observers could agree on the coding and agree that the behaviors represented the trait adjective. The revised ethogram was further refined and tested for reliability while two observers coded behaviors of a group of lion-tailed macaques at the St. Louis Zoo. The reliability of the ethogram was tested again in a group of lion-tailed macaques at the Mesker Park Zoo, by one naive observer and one individual who had participated in the prior test. 182 / Rouff et al. Cohen’s Kappa for behavioral categories ranged from 0.88 to 1.00. Values of Kappa greater than 0.75 indicate excellent agreement beyond chance [SPSS, 1997]. Table I contains more details on the traits included in this study, how they were defined, and the types of behaviors recorded. TABLE I. Personality Facets and Traits Included in This Study High Low Extraversion-like facets and behaviors I. Proximity Sociable – within 1 body length of Solitary – more than 1 body length another from others II. Activity level Active – moves faster than 1 body Slow – moves at or slower than 1 body length/second length/second III. Affiliativeness Affiliative – allogrooming, play, Aloof – autogrooms, refuses to play, refuses to non-aggressive physical contact allogroom, stops interacting IV. Response to novelty Bold – interested in and closer than Cautious – interested in but farther than 3 3 body lengths to a novel stimuli body lengths from a novel stimuli V. Dominance Assertive – displaces another without Submissive – yields place, food, or affiliative threat or aggression relationship to another Neuroticism-like facets and behaviors I. Anxiety Anxious – stiff, restrained posture Relaxed – relaxed posture and movement and movement; surveys environment; when 2 others display anxious behaviors relocates in response to interactions of others when not approached; stereotypical behavior; scratching; yawning (in non-sleepy context) II. Reactivity Reactive – reacts strongly to Unreactive – does not react to nonthreatenenvironmental change ing stimuli when 2 others are reactive Agreeableness-like facets and behaviors I. Intervenes (protective) – aids victim Aggressive –attacks or chases another of aggression or juvenile when play becomes too rough II. Interacts (gentle) – no threat or Agonistic – threatens another aggression when others initiate an interaction III. Approachable – does not move away Moves away (suspicious) – moves away when approached by others when approached IV. Sensitive (to distress of others) – Insensitive (to distress of others) – does notices (looks at) an agonistic or not notice (does not look at) an aggressive aggressive interaction; touches or agonistic interaction; steals or tries to embraces, or grooms a fearful steal an infant; pulls out hair of another individual V. Welcome – others allow the focal Avoided – others move away when focal animal to approach animal approaches Persistent Persistent – attempts to gain an item or outcome after failing to obtain it (working for a food item; re-approaching an individual following an agonistic or aggressive rebuff; or subsequent attempts to initiate play, grooming, or copulation after previous rejection) Lion-Tailed Macaque Personality Traits / 183 Subjects The subjects were 52 lion-tailed macaques in eight different groups at four different zoos (three groups at the Baltimore Zoo, two at the Center for Reproduction of Endangered Species (CRES) of the San Diego Zoological Society, one at the Mesker Park Zoo (Evansville, Indiana), and two at the Saint Louis Zoo). The groups included in this study were selected to provide a sample that reflected the different demographic and environmental conditions of lion-tailed macaques in captivity. Procedure Focal animal data. Each group was observed on three separate occasions at least 90 days apart. On each of these three occasions, each animal was observed for a total of 4 hr (12 hr per animal over the entire study, and 616 hr total for all animals). One researcher collected the majority of the data included in the analysis (596 hr), and a second individual collected the remaining 20 hr of data. The monkeys were observed for a 30-min period using instantaneous sampling for a focal animal, with ethogram behaviors coded as present or absent at 1-min intervals, yielding 30 sample points [Lehner, 1996]. The onset of all ethogram event behaviors was also recorded. Only one 30-min sample was collected on any individual per day (24 samples total per individual). All macaque groups could see the observers and were habituated to their presence. The Mesker Park, Baltimore Zoo Veterinary Center, and CRES groups could easily hear the observers as well. Before the data were collected, two observers checked the reliability of the data collection by simultaneously coding the behavior of each individual in each group for 30 min. A Cohen’s Kappa for interrater reliability was calculated for each behavioral facet (e.g., sociable behaviors/solitary behaviors or reactive behaviors/unreactive behaviors) for each group. Table II shows the mean Cohen’s Kappa for all groups on each behavioral facet. Response to novelty data. A novel-objects study was conducted at the conclusion of the focal animal study. Twelve novel objects (a plastic owl, dinosaur head puppet, small magnifying mirror, large toy stuffed bird, plastic pumpkin, laptop computer with the screen-saver program ‘‘Catz’’ running, large mirror, rubber snake, poster of the head of a large felid, plastic frog that croaked when approached, pinwheel blown by a battery-operated fan, and plastic human skull) were placed outside the enclosure for 15 min each. Data were collected through scan sampling at 1-min intervals, and it was noted whether each individual was bold (within three body lengths and looking at the object), cautious (farther than three body lengths away, and looking at the object), or uninterested in it. Before the study began, two observers simultaneously recorded the bold, cautious, and uninterested behaviors of each group in response to three different novel objects to check the reliability of the data collection. Cohen’s Kappa ranged from 0.82 to 0.92. Persistence data. The collection of persistence data was complicated by the different circumstances of each group and the different zoo policies, which created different opportunities for ethogram persistent behaviors. Therefore, PVC tube feeders, which required manipulation for an occasional food reward, were 184 / Rouff et al. TABLE II. Mean Value of Cohen’s Kappa Calculated For Reliability of Facets Averaged Across Groups Extraversion Neuroticism Agreeableness Proximity .96 Anxiety .88b Activity level Affiliativeness .91 .95 Reactivity .98c Response to Novelty Dominance a .92 Aggressive/ intervenes Agonistic/interacts Moves away/ approachable Insensitive/sensitive Welcome/avoided Persistence 1.00d Persistent .99e .95 .94 .92 .94 a Behaviors observed in only one group during sample periods (K=0.90). 100% agreement that behaviors did not occur in other groups. b Behaviors observed in only five groups during sample periods. 100% agreement that behaviors did not occur in other groups. c Behaviors observed in only six groups during sample periods. 100% agreement that behaviors did not occur in other groups. d Behaviors observed in only three groups during sample periods. 100% agreement that behaviors did not occur in other groups. e Behaviors observed in only four groups during sample periods. 100% agreement that behaviors did not occur in other groups. periodically placed in the enclosures. However, this was not feasible for the CRES groups. In those groups, persistence was defined as digging through the grass/ weeds for seeds, bugs, etc. prior to the morning produce feeding, or prior to any isolation of females for a hormonal study by another research group. Data Analysis PCA. The focal animal data were analyzed with a P-type PCA, and bootstrapped confidence intervals were calculated for the resulting principal components loadings. Twenty-six variables were included in the analysis: welcomed, avoided, sociable, solitary, aloof (breaks off interaction + refuses to play/groom), affiliative, self-grooming, active, slow, assertive, submissive, bold, cautious, anxiety (anxious + submissive gestures + salute + scratch + stereotypical behaviors + gape yawn), relaxed, reactive, unreactive, aggressive, intervenes, moves away, approachable, interacts, agonistic, insensitive, sensitive, and persistent. In the P-type PCA, a correlation matrix was first calculated for each monkey. Then the correlation matrices for each of the 52 subjects in the study were aggregated by simple averaging. With simple averaging, any bias will be slightly negative. This makes it a more conservative approach than Fisher z0 transformation, which can produce a positive bias [Strube, 1988]. By aggregating the correlations from all individuals, we also dealt with the problem of increased standard errors for correlations in the individual matrices that were based on a small number of observations, as well as the effects of outliers from a single correlation for any individual. A PCA with Varimax rotation was performed on the averaged correlation matrix for all monkeys in the study (n=52), with the use of Systat statistical software. Three components were selected for further analysis based on the Scree test and Horn’s  random Scree procedure. In addition, PCAs were conducted on the averaged matrices of subgroups of subjects to ensure that the Lion-Tailed Macaque Personality Traits / 185 components used to develop component scores for each individual (which were taken from the analysis of all animals) were representative of the entire sample. The PCAs of these subgroups (all males (n=20), all females (n=32), intact and vasectomized males (n=11), chemically or surgically castrated males or juvenile males (n=9), cycling females (n=19), and noncycling females (postmenopausal, spayed, contracepted, and juvenile females (n=13)) suggested that three components provided a reasonably stable solution. We then calculated regression-based component scores for each individual on each of the three components by multiplying the standardized scores of behavioral categories for each individual by a weight matrix derived from the PCA of the averaged matrices for all of the monkeys in the sample. Bootstrapping techniques [e.g., Hamilton, 1992; Hesterberg et al., 2003; Mooney & Duval, 1993] were used to empirically build sampling distributions for the component loadings. This made it unnecessary to rely on theoretically derived distributions, the values of which carry assumptions that may not be tenable for the data from this study. These sampling distributions were then used to construct confidence intervals for the loadings. Bootstrapped confidence intervals were calculated with a combination of commercial statistical software (Systat) and software designed for these data (Visual Basic) [Strube, 2002]. Resampling for the bootstrapping procedure was carried out in the following way: In each of the 2,000 bootstrapped samples, data from 52 animals were selected randomly and with replacement. Eleven observation periods were selected randomly and with replacement from the data for each of the 52 selected individuals. A correlation matrix was constructed from the 11 observation periods for each selected individual, and the 52 resulting matrices were then averaged, with each correlation weighted by the available sample size. Occasionally the resampling produced undefined correlates because there was no variability for a particular variable. These instances were considered missing data, and the averaged correlations were based on the remaining data. Occasionally some nondiagonal correlations were equal to 1.0, a situation that prevents inversion of the matrix in the PCA. These values were coded to .99 so the analysis could proceed. We calculated the end points of the confidence intervals using the percentile method, the bias-corrected percentile method, and the normal approximation method [Mooney & Duval, 1993]. The bias-corrected method should have been the method of choice for this analysis, but it did not appear to work well for these data. Therefore, only those variables that loaded significantly on a component in both the percentile and the normal approximation methods were considered significant loadings. These two methods differ when the median bootstrap estimate is quite different from the original parametric estimate, and requiring significance by both methods is a conservative approach. As a matter of interest, an R-type PCA with Varimax rotation was also performed on the mean behavior/hour data from all monkeys in the study to determine which components would emerge from this more traditional method of analysis. In the R-type analysis, four components were judged as meaningful according to the Scree test with sample data and a random data Scree test. Confidence intervals were then constructed for the four R-type components by means of the same method used in the P-type analysis. Novel-Objects Study Data To examine the relationships among behaviors in the novel-objects study and the individual scores on Component 3, we calculated correlations for novel-object 186 / Rouff et al. behaviors (bold, cautious, uninterested behaviors, and ‘‘curious’’ – bold + cautious behaviors) with individual component scores for Component 3, as well as with focal animal bold and cautious behaviors. Multiple Regression Analysis After the effects of age and zoo group were removed, the effects of sex, hormone status, and dominance rank on the individual component scores from the P-type PCA were investigated by means of a multiple regression analysis with bootstrapped confidence intervals. To determine dominance status, we calculated each individual’s rank within the social group using a dyadic interaction matrix [Lehner, 1996] of assertive and submissive behaviors. The resulting linear ranks were then divided into three equal groups (high-, mid-, and low-status). Bootstrapped confidence intervals were calculated using the percentile-t method [Mooney & Duval, 1993] with software designed for this project (Mathcad 2000 [Strube, 2001]). The bootstrapped sampling distribution was constructed in the following way: For each of the 2,000 bootstrapped samples, a sample of 52 individuals was selected randomly and with replacement. The data for each bootstrapped sample were then used in a standard multiple regression analysis to produce the regression coefficients for each predictor in the regression model. The resulting 2,000 sets of regression coefficients then constituted the bootstrapped sampling distribution on which the confidence intervals were based. The second resampling used to create the t-values was conducted on 100 samples from each bootstrapped sample. RESULTS Three components, which accounted for 28.64% of the variance, were judged as meaningful based on the Scree and random Scree tests. Details of the three components are shown in Table III. Component 1 contained many Extraversion behaviors and several Agreeableness behaviors. Sociable, affiliative, aloof, welcome, approachable, and willing to interact loaded positively, and solitary and slow loaded negatively at significant levels. Only one variableFagonistic–was significantly related to Component 2. Bold and cautious behaviors loaded positively on Component 3. As a matter of interest, the components from the R-type analysis are presented in Table IV. Bold, cautious, uninterested, or curious (bold + cautious) behaviors from the novel-objects study were not related to the P-type Component 3 scores for individuals. However, bold behaviors from the focal animal data were associated with novel-object bold behaviors (r=.30, Po.05) and focal animal cautious behaviors correlated with novel-object cautious behaviors (r=.41, Po.01). The results of the multiple regression analysis indicated that the mean component score for females was higher than that for males on the Extraversion-like Component 1 (95% confidence interval for the mean difference=0.732–1.661) and on Component 3 (95% confidence interval for the mean difference=0.105–0.699), after controlling the effects of age and group. These differences can be seen in Fig. 1, which displays the simple mean differences in component scores by sex prior to correction for age and social group. Figure 2, which is also based on the simple mean component scores, shows that the mean scores on Component 2, which was related to a ‘‘disagreeable’’ behavior (agonistic), were greater for high-status individuals relative to lower-ranked macaques (95% confidence interval for Lion-Tailed Macaque Personality Traits / 187 TABLE III. P Type Principal Components Component Behaviors Sociable (E) Solitary (E) Affiliative (E) Aloof (E) Interacts (A) Approachable (A) Slow (E) Welcome (A) Active (E) Persistent (P) Agonistic (A) Aggressive (A) Sensitive (A) Avoided (A) Reactive (N) Moves away (A) Intervenes (A) Relaxed (N) Submissive (E) Assertive (E) Anxious (N) Unreactive (N) Cautious (NS) (E) Bold (NS) (E) Insensitive (A) Eigenvalues for rotated components Percent of total variance explained 1 2 3 0.92 0.92 0.85 0.62 0.53 0.43 0.36 0.34 0.16 0.13 0.06 0.01 0.05 0.03 0.14 0.09 0.02 0.01 0.03 0.07 0.13 0.12 0.01 0.03 0.02 3.62 13.94 0.15 0.15 0.20 0.16 0.30 0.45 0.26 0.28 0.12 0.11 0.50 0.49 0.40 0.37 0.34 0.32 0.30 0.25 0.23 0.22 0.22 0.17 0.07 0.02 0.13 1.96 7.52 0.02 0.02 0.05 0.14 0.03 0.14 0.10 0.03 0.04 0.03 0.01 0.15 0.01 0.00 0.06 0.11 0.17 0.03 0.13 0.17 0.07 0.15 0.91 0.90 0.15 1.87 7.18 Po.05 in bold. E, Extraversion; N, Neuroticism; A, Agreeableness; NS, Novelty Seeking; P, Persistence. the mean difference for mid-ranked=–0.203 to –0.984; low-ranked=–0.636 to – 1.129). DISCUSSION Although the traits included in this study were drawn from previous nonhuman primate personality studies and human models, the components that resulted from the PCA reflected the relationships among the recorded behaviors of the lion-tailed macaques in the study. Three components were judged as meaningful, and bootstrapped confidence intervals provided the statistical criteria for judging the significance of the behavioral categories to the components. Component 1. Component 1 contained behaviors that are consistent with the sociable (gregariousness) and affiliative (warmth) facets of Extraversion. This component emerged from all four preliminary PCAs of the subgroups, the analysis of all animals, and the R-type analysis, which suggests that this component represents 188 / Rouff et al. TABLE IV. R-Type Principal Components Component Behaviors Sociable (E) Solitary (E) Affiliative (E) Slow (E) Aloof (E) Interacts (A) Approachable (A) Anxious (N) Agonistic (A) Submissive (E) Aggressive (A) Moves away (A) Cautious (NS) (E) Welcome (A) Active (E) Bold (NS) (E) Assertive (E) Reactive (N) Self-groom (E) Avoided (A) Unreactive (N) Sensitive (A) Persistent (P) Relaxed (N) Insensitive (A) Intervenes (A) Eigenvalues of rotated components Percent of total variance explained 1 2 3 4 0.86 0.86 0.83 0.77 0.70 0.69 0.66 0.38 0.18 0.24 0.14 0.32 0.09 0.51 0.01 0.22 0.32 0.02 0.31 0.13 0.27 0.22 0.06 0.23 0.37 0.02 5.36 20.63 0.03 0.03 0.07 0.10 0.20 0.11 0.13 0.23 0.80 0.69 0.68 0.54 0.35 0.14 0.32 0.09 0.52 0.02 0.08 0.45 0.11 0.26 0.24 0.09 0.06 0.33 3.02 11.62 0.09 0.09 0.01 0.06 0.40 0.24 0.21 0.06 0.08 0.19 0.15 0.26 0.30 0.71 0.69 0.62 0.61 0.59 0.51 0.50 0.05 0.33 0.12 0.05 0.07 0.18 3.27 12.58 0.36 0.36 0.35 0.25 0.18 0.29 0.63 0.02 0.14 0.11 0.13 0.20 0.18 0.22 0.09 0.03 0.27 0.00 0.35 0.14 0.73 0.65 0.61 0.59 0.57 0.53 3.61 13.89 Po.05 in Bold. E, Extraversion; N, Neuroticism; A, Agreeableness; NS, Novelty Seeking; P, Persistence. a robust dimension of lion-tailed macaque personality. The higher mean score for females on Component 1 is consistent with the social organization of this femalebonded species. The results of this study suggest that the sociable and affiliative personality traits of female lion-tailed macaques predispose them to engage in interactions that establish and reinforce these social bonds. However, males disperse from the natal group, and strong social bonds, fostered by social and affiliative personality traits, may deter a male from leaving and/or interfere with social factors that encourage emigration. A predisposition to socialize may also distract resident males from directing vigilance behavior toward predators and other rival males. Intermale aggression is not uncommon for this species, and avoiding close proximity to others may reduce inadvertent conflicts and wounds. Component 2. The lower limits of the confidence interval for agonistic, the sole significant loading, was close to zero with both the percentile and normal approximation Lion-Tailed Macaque Personality Traits / 189 P Type Component Score Profiles By Sex 1.5 95% Confidence Intervals for the Mean 1.0 .5 0.0 Component 1 -.5 Component 2 -1.0 Component 3 -1.5 N= 32 32 32 20 Females 20 20 Males Sex Fig. 1. P-type component score profiles by sex, based on the simple mean scores for males and females without correcting for age or group membership. The multiple regression indicated that the mean score for females was significantly higher than the mean score for males on Components 1 and 3. methods. This provides weaker support for considering agonistic a significant loading, and component 2 a significant component, compared to the evidence for the other components. This component might be interpreted as the negative pole of agreeableness, because agonistic and aggressive had the highest loadings on the component (.50 and .49, respectively). It would not be surprising if a replication of this study indicates that aggressiveness is also significantly related to this component. Personality traits may be organized differently in different primate species, and the negative aspects of Agreeableness may form a separate personality dimension in some. Two other behavior-based personality studies also found a separate aggressive/agonistic component (in chimpanzees [van Hooff, 1970] and rhesus monkeys [Chamove et al., 1972]). However, the relationships of all the variables contribute to a component, not just the significant loadings. All but one of the nine highest-loading variables on component 2 were behaviors representing aspects of the Big Five Agreeableness factor. It was an unexpected finding that all of these agreeable-like behaviors loaded positively on this component, since this is not the bipolar pattern predicted by the Big Five Model. It is possible that the behaviors representing these traits were not truly bipolar opposites. However, the multiple regression procedure indicated a relationship between this component and dominance status, with the mean score of high-status individuals being greater than the mean scores of mid- or lowranking individuals. The pattern of higher agonistic, aggressive, sensitivity to 190 / Rouff et al. P Type Component Profiles By Dominance Rank 95% Confidence Intervals for the Mean 1.5 1.0 .5 0.0 Component 1 -.5 Component 2 -1.0 Component 3 -1.5 N= 18 18 High 18 16 16 16 Mid 18 18 18 Low Dominance Rank Fig. 2. P-type component score profiles by dominance rank, showing that the simple uncorrected mean score for high-ranking individuals was higher on Component 2 than the mean scores for lower-ranking individuals. group aggression, intervention, and reactivity behaviors seems consistent with the behavior of more-dominant individuals, as does the attraction of the group to them and the tendency of other group members to avoid dominant individuals when approached by them. A dominance-related factor defined by assertiveness, physical aggression, and low fearfulness was reported in seven out of 19 studies of nonhuman animals [Gosling & John, 1999]. Neither assertive nor submissive behaviors were related to this component, but these behaviors were narrowly defined by priority of access. The highest loadings on Component 2 from the Rtype analysis were assertive, submissive, aggressive, moves away, and agonistic; however, only submissive and agonistic were judged as significant. Clearly, additional studies including more behavioral measures are required to determine the exact nature of this component and whether it is best characterized as an ‘‘agreeable’’ component of both positive and negative traits, a ‘‘disagreeable’’ component, or perhaps even a dominance-related component. Component 3. Component 3 was unanticipated, because bold and cautious behaviors (response to novelty) were relatively uncommon in the focal data. Bold and cautious behaviors from the focal data both loaded positively on this component, suggesting the trait of curiosity. Although Components 1 and 2 from the R-type analysis appeared similar to the first two P-type Components, none of the R-type components resembled P-type Component 3. P-type Component 3 is an example Lion-Tailed Macaque Personality Traits / 191 of the different type of information gained when within-animal variance is included with between-animal variance. Component 3 scores for individuals were not correlated with bold, cautious, uninterested, or curious (bold + cautious) behaviors in the novel-objects study, even though focal bold behaviors were associated with novel-object bold behaviors, and focal cautious behaviors were related to novel-object cautious behaviors. This suggests that the component scores carry important (but as yet unspecified) additional information from other variables beyond interest in novelty. The Big Five Factor Model places a preference for novelty over familiarity and intellectual curiosity on the openness to Experience factor. Factors similar to the openness dimension were identified in studies of seven nonhuman species [Gosling & John, 1999]. Curiosity is associated with impulsiveness and recklessness in Cloninger’s  Novelty Seeking factor and Zuckerman et al.’s  Impulsive Sensation Seeking. Impulsivity was not included in this study, because no behaviors that covered all age/sex classes and would be feasible in a zoo setting could be found to represent this trait. Adding a measure of impulsivity might clarify whether Component 3 resembles either the Openness to Experience Factor or Novelty/Sensation Seeking. After the effects of age and group on Component 3 were controlled for, the mean score for females was still significantly higher than the mean score for males. The idea that males and females have different levels or types of curiosity makes sense for this species in the wild. In the novel-objects study, the males were interested in potentially threatening objects (the plastic owl, poster of a felid head, and large mirror) but were less interested than the females in nonthreatening items, such as pinwheels or small stuffed animals. Wild groups should benefit if males, with their larger size and sharper canines, are predisposed to remain vigilant and attend to potentially dangerous objects, without being distracted by trivial novel items. However, higher levels of curiosity would predispose females to investigate nonthreatening novel objects as well, fostering the discovery of food resources to meet the high nutritional demands of pregnancy, lactation, and infant care. Critique of the method. The method used in this study (i.e., behavioral observations used as data, a P-type PCA, and bootstrapped confidence intervals as criteria for judging the significance of factor loadings) provided an informative description of lion-tailed macaque personality dimensions, and shows promise for both intra- and interspecific personality studies. P-type Component 3 indicates that incorporating both within- and between-individual variation in the P-type analysis, as used in this study, can produce additional important information on personality. The confidence intervals provided a more precise estimate of significance compared to arbitrary rules of thumb. For example, a factor loading as low as 0.34 (welcome: P-type Component 1) was judged as significant, while a loading as high as –0.69 (active: R-type Component 3) was rejected, and a 0.43 loading of approachable on Component 1 was accepted, while a loading of 0.45 on Component 2 for this behavior was not. Seven behaviors with loadings over 0.60 on the R-type components would have been incorrectly reported as significant if an arbitrary 0.60 criterion for significance had been applied. Granted, requiring significance by both the normal approximation and percentile methods, as done in this study, is a stringent and conservative criterion; however, it is not inappropriate for PCA, which capitalizes on chance. Most, but not all, of 192 / Rouff et al. the excluded loadings over 0.60 were deemed significant by the normal approximation method, which assumes that the data are normally distributed. The results suggest that when a 0.60 cutoff is arbitrarily applied to non-normal data, the reported results can be seriously misleading. Bootstrapped confidence intervals for factor loadings appear to be an effective method for decreasing the possibility that chance factor loadings will be considered significant for both P- and R-type analyses, which is especially important when sample sizes are small. However, the choice of a significance level is also a somewhat arbitrary decision, and the expert eye of the researcher should still determine when refinement and/or replication are needed to firm up confidence in findings that are intriguing but do not currently meet the traditional standard. Some suggestions can be made for future studies that will use this method of personality research. The traits included in the current study did not comprise an exhaustive list, and traits that are important descriptors of lion-tailed macaque personality may have been omitted. One could expand the trait list by consulting researchers who are familiar with the species under study as to which traits to add, and which behaviors represent the traits. A measure of impulsivity is definitely needed, and Fairbanks’  intruder challenge test might be tried when circumstances permit. With regard to the persistence behaviors, the different circumstances at each zoo resulted in different persistence opportunities for different groups, the ‘‘intermittent reward’’ may have had a different appeal for different monkeys, and individuals with a low dominance rank may have had less access to enrichment devices. Some traits, such as impulsivity and persistence, may be better assessed under more controlled conditions and with experimental measures. In this study, continuous variables were dichotomized into two bipolar opposites, such as sociable (within a body length) and solitary (farther than a body length). Some traits, such as insensitive, were represented by several different behaviors, some of which may have been stronger indicators of a trait than others. Measuring continuous variables or behavioral categories comprised of multiple behaviors on a five-point scale might better represent these traits. The definitions of several behaviors representing Extraversion traits should be redefined or removed from the ethogram in future studies. Aloof behaviors were defined as refusing to play, refusing to groom another, or breaking off an interaction. This required an individual to be sociable and possibly affiliative, which resulted in a positive association of aloof with other positive Extraversion behaviors, and not the negative association predicted by the Big Five Model. Welcome and avoided behaviors were intended as a measure of how agreeable the approached animal found the approaching individual. Avoided did not load significantly on any component, and the relationship of welcome to other extraversion behaviors most likely reflected an inclination to approach others. One researcher coded nearly all but 20 hr of the lion-tailed macaque behaviors observed in this study. The use of multiple coders would reduce the influence of any bias from an individual’s idiosyncratic interpretation of behavioral definitions or propensity to notice certain behaviors. The P-type analysis used in this study integrated both within- and betweenindividual variation in behaviors. However, it also incorporated more variation due to situation, relative to R-type analyses, which most likely explains why the P-type components in this study accounted for much less of the variance (28.6%) than the 58.7% in the R-type analysis from this study, and the 50–85% seen with Lion-Tailed Macaque Personality Traits / 193 R-type components in previous personality studies. When using the P-type analysis, it may be necessary to collect more sample periods for individuals than the 24 per individual used in this study. This should improve the investigator’s ability to estimate the P-type components, since similar situations would occur with sufficient frequency that regular patterns of response would be detected, increasing the amount of variance accounted for by the P-type components. Although the method used in this study is well suited for comparative research, specific comparisons can be problematic when analyses of the same behavioral measures produce dissimilar components. Collecting trait term data in the same manner used for behavioral data, using a defined list of trait terms and behavior as an ethogram may help to resolve this problem. Behavioral measures have been criticized as being unable to capture subtle, complex, or rarely expressed traits, because they are limited to specific ethogram behaviors. Single or yearly ratings with trait terms are vulnerable to memory biases. Combining the best of both methods may compensate for the weakness of each technique alone. When a researcher observes an animal running across a field, he/she relies on the context to decide whether to code this motor action pattern as flight behavior, aggression, or play. Coding a trait, such as gentle (responds in a kind way, not rough or threatening), from a predefined adjective list should not be any more subjective than collecting behavioral data. Trait term data should provide information over and above what can be obtained from behavioral observations alone. Adjective trait terms need not be converted into ‘‘personality’’ terms. Therefore, simple frequencies could be compared, or the trait term data could be subjected to a PCA or multiple regression analysis in the same manner used for the behavioral data in this study. Although trait and behavioral data could be analyzed separately, some questions might be best answered by a combination of the two. Examining the correlations between trait terms and behavioral data could clarify which behaviors best represent a trait, or, conversely, which behaviors might be predicted from trait information. Physiological measures, such as salivary cortisol or urinary hormone levels, should also be included in these trait/behavior studies, when feasible, to investigate the biological bases of behaviors and/or assessed traits. FUTURE DIRECTIONS AND CONCLUSIONS By establishing personality dimensions and then validating them by replication, one can establish ‘‘marker’’ variables, or traits that consistently load highly on the same factor. Once a personality dimension has been validated, component scores for new subjects can be calculated with unit weightings of standardized behavioral frequency scores, provided that the behavioral categories are defined in the same way [Figueredo & Ross, 1992]. These individual component scores contain information on covariation as well as frequency of behaviors. More narrowly focused questions could be investigated, such as the relationship of a particular personality dimension to sex (as illustrated by Fig. 1) or dominance status (Fig. 2), or to explore similarities and differences between different primate groups. Differences in component score patterns and the behavioral profiles that contribute to these scores could be examined on an individual level (Fig. 3). Not only could between-individual comparisons be made, but within-individual differences could be investigated with longitudinal studies to consider the stability of personality and how component scores change from infancy to adulthood or in response to changing social environments. Species could be compared by examining the similarities and differences in the 194 / Rouff et al. TOFU Response to Novelty and Persistence Behaviors P Type Component Scores 3.0 3.000 2.000 1.000 0.000 -1.000 -2.000 -3.000 Series1 Component 1 Component 2 Component 3 0.986 0.676 -0.087 95% Confidence Interval for the Mean/Sample 2.5 2.0 1.5 1.0 .5 Bold 0.0 Cautious -.5 Persistence -1.0 Tofu Extraversion-Like Behaviors Extraversion-like Behaviors Affiliativeness and Dominance Facets Proximity and Activity Facets 15.0 40 Affiliative 30 Sociable 20 Solitary 10 Active 0 Slow -10 95% Confidence Interval for the Mean/Sample 95% Confidence Interval for the Mean/Sample 13.0 9.0 Refuse to Play/Groom 7.0 3.0 Assertive 1.0 Tofu Agreeableness-Like Behaviors 7.0 13.0 6.0 11.0 Aggressive 7.0 Intervenes 5.0 Moves away 1.0 Approachable Tofu 95% Confidence Interval for the Mean/Sample 95% Confidence interval for the Mean/Sample Agreeableness-Like Behaviors -1.0 Displaced -1.0 15.0 3.0 Breaks Off 5.0 Tofu 9.0 Self-groom 11.0 Agonistic Interacts 5.0 4.0 Insensitive 3.0 Sensitive 2.0 1.0 Welcome 0.0 Avoided -1.0 Tofu Fig. 3. P-type component score profile and behavioral profiles for Tofu, a low-ranking female in the St. Louis Zoo family group, who was born in 1981. components that emerge from the analysis of behavioral or trait data, provided that equivalent species-specific measures are used. Comparative studies on primate personality traits can be achieved only through a long-term coordinated effort that combines the results from different Lion-Tailed Macaque Personality Traits / 195 sites and investigators, which makes the use of standardized measures essential [de Waal & Luttrell, 1989; Gosling, 1999, 2001; Moore, 1992, 1993]. Research on personality traits should benefit if researchers studying different topics and species would collaborate by adding standardized trait term data to their research design, and by collecting trait data in the same manner used for behavioral data. Therefore, the development of a taxonomy of standardized trait terms that is applicable to a variety of primate species, and considers the different ways a trait might be manifested by different species is critical [Gosling, 1999, 2001]. ACKNOWLEDGMENTS We thank Jane Phillips-Conroy, C. Robert Cloninger, Tab Rasmussen, Richard Smith, and Samuel Gosling for sharing their expertise and comments. We are most grateful to the reviewers of the manuscript for their valuable insights and suggestions. We also thank the Baltimore Zoo, CRES, the Mesker Park Zoo, and the St. Louis Zoo (especially Ingrid Porton, Nancy Harvey, Karen Barnes, Kathy de Falco, Mike Cranfield, Laura Campbell, Joanne Olivia-Purdy, Gwen Mullen, and Syd Dawson) for their assistance and warm hospitality. REFERENCES Allport GW. 1937. Personality: a psychological interpretation. New York: Holt, Rinehart & Winston. 588p. Bahr NI. 1995. Environmental factors and hormones: their significance for maternal behavior in captive gorillas. In: Pryce CR, Martin RD, Skuse D, editors. Motherhood in human and nonhuman primates. Basel: Karger. p 94–105. Bard KA. 1995. Parenting in primates. In: Bornstein M, editor. Handbook of parenting. Mahwah, NJ: Lawrence Erlbaum Associates. p 27–58. Bard KA, Gardner KH. 1996. Influences on development in infant chimpanzees: enculturation, temperament, and cognition. In: Russon AE, Bard KA, Parker ST, editors. Reaching into thought: the minds of great apes. New York: Cambridge University Press. p 235–256. Bernstein IS, Shusterman RJ, Sharpe LG. 1963. A comparison of rhesus monkey and gibbon responses to unfamiliar situations. J Comp Physiol Psychol 56:914–916. Boinski S. 1999. Geographic variation in behavior of a primate taxon: stress responses as a proximate mechanism in the evolution of social behavior. In: Foster SA, Endler JA, editors. Geographic variation in behavior: perspectives on evolutionary mechanisms. New York: Oxford University Press. p 95–120. Bolig R, Price CS, O’Neill PL, Suomi SJ. 1992. Subjective assessment of reactivity level and personality traits of rhesus monkeys. Int J Primatol 13:287–306. Borries C, Sommer V, Srivastava A. 1994. Weaving a tight social net: allogrooming in free-ranging female langurs (Presbytis entellus). Int J Primatol 15:421–443. Boysen ST. 1994. Individual differences in the cognitive abilities of chimpanzees. In: Wrangham RW, McGrew WC, de Waal FBM, Heltne PG, editors. Chimpanzee cultures. Cambridge: Harvard University Press. p 335–350. Brock D. 1998. Social interactions of a reintegrated hand-reared juvenile chimpanzee (P. troglodytes) at the Little Rock Zoo. In: Landau V, editor. Proceedings of the Chimpanzoo Conference, Tucson, 1997. p 1–5. Budeav SV. 1997. Personality in the guppy. J Comp Psychol 111:399–411. Burton FD. 1992. The social group as information unit: cognitive behavior, cultural processes. In: Burton FD, editor. Social processes and mental abilities in non-human primates: evidences from longitudinal field studies. Lewiston: Edwin Mellen Press. p 31–60. Capitanio JP, Mendoza SP, Mason WA, McChesney MB, Lerche NW. 1994. Personality assessments of adult male rhesus macaques and immune system measures. Am J Primatol 33:200. Capitanio JP, Mendoza SP. 1995. Tetanusspecific antibody response is related to behavioral response to stable or unstable group formations in adult male rhesus monkeys (Macaca mulatta). Am J Primatol 36:115. Capitanio JP. 1999. Personality dimensions in adult male rhesus macaques: prediction of 196 / Rouff et al. behaviors across time and situation. Am J Primatol 47:299–320. Capitanio JP, Mendoza SP, Baroncelli S. 1999. The relationship of personality dimensions in adult male rhesus macaques to progression of simian immunodeficiency virus disease. Brain Behav Immunol 13: 138–154. Chamove A, Eysenck H, Harlow H. 1972. Personality in monkeys: factor analyses of rhesus social behavior. Q J Exp Psychol 24:496–503. Champoux M, Shannon C, Suomi SJ. 1997. Variability in response to novelty in surrogate-reared rhesus monkey infants. Am J Primatol 42:100. Clark LA, Watson D. 1999. Temperament: a new paradigm for trait psychology. In: Pervin LA, John OP, editors. Handbook of personality: theory and research. New York: Guilford Press. p 399–423. Clarke AS, Mason WA, Mendoza SP. 1994. Heart rate patterns under stress in three species of macaques. Am J Primatol 33: 133–148. Clarke AS, Boinski S. 1995. Temperament in nonhuman primates. Special issue: temperament, social cognition, and the evolution of social behavior. Am J Primatol 37:103–125. Clarke AS, Czekala NM, Lindburg DG. 1995. Behavioral and adrenocortical responses of male cynomolgus and lion-tailed macaques to social stimulation and group formation. Primates 36:41–56. Cloninger CR, Przybeck TR, Svrakic DM, Wetzel RD. 1994. The temperament and character inventory (TCI): a guide to its development and use. St. Louis: Center for Psychobiology of Personality. 184p. de Waal FBM, Luttrell LM. 1989. Toward a comparative socioecology of the genus Macaca: different dominance styles in rhesus and stumptail monkeys. Am J Primatol 19:83–109. Dutton DM, Clark RA, Dickins DW. 1997. Personality in captive chimpanzees: use of a novel rating procedure. Int J Primatol 18:539–552. Fairbanks LA. 1996. Individual differences in maternal style: causes and consequences for mothers and offspring. In: Rosenblatt JS, Snowden CT, editors. Parental care. San Diego: Academic Press. p 579–611. Fairbanks LA. 2001. Individual differences in response to a stranger: social impulsivity as a dimension of temperament in vervet monkeys (Cercopithecus aethiops sabaeus). J Comp Psychol 115:22–28. Figueredo AJ, Ross DM. 1992. The quantitative ethology of the zebra finch: a study in comparative psychometrics. Multivar Behav Res 27:435–458. Fleeson W. 2001. Toward a structure-andprocess-integrated view of personality: traits as density distributions. J Pers Soc Psychol 80:1011–1027. Fleeson W. 2004. Moving personality beyond the person-situation debate: the challenge and opportunity of withinperson variability. Curr Dir Psychol Sci 13: 83–87. Garcia-Sevilla L. 1984. Extraversion and neuroticism in rats. Pers Individ Dif 5: 511–532. Gillespie NA, Cloninger CR, Heath AC, Martin NG. 2003. The genetic and environmental relationship between Cloninger’s dimensions of temperament and character. Pers Individ Dif 35:1931–1946. Gold KC, Maple TL. 1994. Personality assessment in the gorilla and its utility as a management tool. Zoo Biol 13:509–522. Goodall J. 1986. The chimpanzees of Gombe. Cambridge: Harvard University Press. 673p. Gosling SD. 1999. Personality dimensions in animals. Ph.D. dissertation, University of California–Berkeley, Berkeley, CA. 109p. Available from University Microfilms, Ann Arbor, MI; AAD99–22851. Gosling SD, John OP. 1999. Personality dimensions in nonhuman animals: a crossspecies review. Curr Dir Psychol Sci 8: 69–75. Gosling SD. 2001. From mice to men: what we can learn about personality from animal research. Psychol Bull 127:45–86. Gusnard D, Ollinger JM, Shulman GL, Cloninger CR, Price JL, Van Essen DC, Raichle ME. 2003. Persistence and brain circuitry. Proc Natl Acad Sci USA 1006:3479–3484. Hamilton LC. 1992. Regression with graphics: a second course in applied statistics. Belmont, CA: Brooks/Cole. 363p. Hesterberg T, Monaghan S, Moore DS, Clipson A, Epstein R. 2003. Bootstrapped methods and permutation tests: companion chapter to the practice of business statistics. New York: W.H. Freeman. 85p. http://bcs. whfreeman.com/pbs/cat_160/PBS.18.pdf. Higley JD, Linnoila M, Suomi SJ. 1994. Ethological contributions. In: Hersen M, Ammerman RT, Sisson LA, editors. Handbook of aggressive and destructive behavior in psychiatric patients. New York: Plenum Press. p 17–32. Higley JD, Bennett AJ. 1999. Central nervous system serotonin and personality as variables contributing to excessive alcohol consumption in non-human primates. Alcohol Alcohol 34:402–418. Horn JL. 1965. A rationale and test for the number of factors in factor analysis. Psychometrika 30:179–185. Lion-Tailed Macaque Personality Traits / 197 Huff AJ. 2002. Personality as a predictor of dominance in male vervet monkeys. Am J Primatol 57:69. Johnson PC. 1985. Notes on the ethogram of the captive lion-tailed macaque. In: Heltne PG, editor. The lion-tailed macaque: status and conservation. New York: Alan R. Liss, Inc. p 239–263. Keddy-Hector AC. 1992. Mate choice in nonhuman primates. Am Zool 32:62–70. Kemeny ME, Laudenslager ML. 1999. Beyond stress: the role of individual difference factors in psychoneuroimmunology. Brain Behav Immunol 13:73–75. King JE, Figueredo AJ. 1997. The five-factor model plus dominance in chimpanzee personality. J Res Pers 31:257–271. Laudenslager ML, Rasmussen KL, Berman CM, Lilly AA, Shelton SE, Kalin NH, Suomi SJ. 1999. A preliminary description of responses of free-ranging rhesus monkeys to brief capture experiences: behavior, endocrine, immune, and health relationships. Brain Behav Immunol 13:124–137. Lehner PN. 1996. Handbook of ethological methods. Cambridge: Cambridge University Press. 672p. Lilly AA. 1995. Behavioral and biological markers of stress susceptibility in adolescent female rhesus macaques experiencing separation from natal social groups. Ph.D. dissertation, Rutgers University, New Brunswick. 137p. Available from University Microfilms, Ann Arbor, MI; AAD96-18877. Mather JA, Anderson RC. 1993. Personality of octopuses (Octopus rubescens). J Comp Psychol 107:336–340. McCrae RR, Costa PT. 1997. Personality trait structure as a human universal. Am Psychol 52:509–516. McGuire MT, Raleigh MJ, Pollack DB. 1994. Personality features in vervet monkeys: the effects of sex, age, social status, and group composition. Am J Primatol 33:1–13. Mischel W. 1999. Personality coherence and dispositions in a cognitive-affective personality system (CAPS) approach. In: Cervone D, Shoda Y, editors. The coherence of personality: social-cognitive bases of consistency, variability, and organization. New York: Guilford Press. p 37–60. Mondragon-Ceballos R, Santillan-Doherty A, Chiappa P. 1991. Correlation between subjective assessment of individuality traits and ethological sampling of behaviour in stumptail macaques. In: Ehara A, editor. Primatology today. Proceedings of the XIIIth Congress of the International Primatological Society, Nagoya and Kyoto, 1990. New York: Elsevier Science Publishers. 351p. Mondragon-Ceballos R, Santillan-Doherty AM. 1994. The relationship between personality and age, sex and rank in captive stumptail macaques. In: Roeder JJ, Thierry B, Anderson JR, Herrenschmidt N, editors. Current primatology: social development, learning, and behavior. Strasbourg: Université Louis Pasteur. p 241–250. Mooney CZ, Duval RD. 1993. Bootstrapping: a nonparametric approach to statistical inference. Pacific Grove, CA: Sage Publications. 73p. Moore J. 1992. Dispersal nepotism and primate social behavior. Int J Primatol 13: 361–378. Moore J. 1993. Inbreeding and outbreeding in primates: what’s wrong with ‘‘the dispersing sex?’’ In: Thornhill NW, editor. The natural history of inbreeding and outbreeding: theoretical and empirical perspectives. Chicago: University of Chicago Press. p 392–426. Murray LE. 1996. Personality and individual differences in captive African apes. Primate Eye 59:40–41. Pavelka M. 1997. The social life of female Japanese monkeys. In: Morbeck ME, Galloway A, Zihlman AL, editors. The evolving female: a life-history perspective. Princeton: Princeton University Press. p 76–85. Pelham BW. 1993. The idiographic nature of personality: examples of the idiographic self-concept. J Pers Soc Psychol 64:665–677. Price EC. 1990. Infant carrying as a courtship strategy of breeding male cotton-topped tamarins. Anim Behav 40:784–786. Rhine RJ, Maryanski A. 1996. A twenty-oneyear history of a dominant stump-tail matriline. In: Lindburg DG, Fa JE, editors. Evolution and ecology of macaque societies. New York: Cambridge University Press. p 473–499. Rowe DC. 1994. The limits of family influence: genes, experience, and behavior. New York: Guilford Press. 232p. Saltzman W, Schultz-Darken NJ, Abbott DH. 1996. Behavioural and endocrine predictors of dominance and tolerance in female common marmosets, Callithrix jacchus. Anim Behav 51:657–674. Sapolsky RM, Ray JC. 1989. Styles of dominance and their endocrine correlates among wild olive baboons (Papio anubis). Am J Primatol 18:1–13. Sapolsky RM. 1990. Adrenocortical function, social rank, and personality among wild baboons. Biol Psychiatry 28:862–878. Sapolsky RM. 1991. Testicular function, social rank and personality among wild baboons. Psychoneuroendocrinology 16: 281–293. Sapolsky RM. 1999. Hormonal correlates of personality and social contexts: from non- 198 / Rouff et al. human to human primates. In: PanterBrick C, Worthman CM, editors. Hormones, health, and behavior: a socio-ecological and lifespan perspective. New York: Cambridge University Press. p 18–46. Saxton PM, Siegel J, Lukas JH. 1987. Visual evoked potential augmenting/reducing slopes in cats: correlations with behavior. Pers Individ Dif 8:511–519. Singh SD, Manocha SN. 1966. Reactions of the rhesus monkey and the langur in novel situations. Primates 7:259–262. Skinner SW, Lockard JS. 1979. An ethogram of the liontail macaque (Macaca silenus). Appl Anim Ethol 5:241–253. Small MF. 1989. Female mate choice in nonhuman primates. Yearb Phys Anthropol 32:103–127. Smuts BB. 1985. Sex and friendship in baboons. New York: Aldine. 303p. Smuts BB. 1987. Sexual competition and mate choice. In: Smuts BB, Cheney DL, Seyfarth RM, Wrangham RW, Struhsaker TT, editors. Primate societies. Chicago: University of Chicago Press. p 385–399. SPSS. 1997. SPSS base 7.5 applications guide. Chicago: SPSS Inc. 339p. Steen RG. 1996. DNA and destiny: nature and nurture in human behavior. New York: Plenum Press. 295p. Steklis HD. 1993. Primate socioecology from the bottom up. In: Almquist AJ, Manyak A, editors. Milestones in human evolution. Prospect Heights, IL: Waveland Press, Inc. p 39–74. Stevenson-Hinde J, Stillwell-Barnes R, Zunz M. 1980a. Individual differences in young rhesus monkeys: consistency and change. Primates 21:498–509. Stevenson-Hinde J, Stillwell-Barnes R, Zunz M. 1980b. Subjective assessment of rhesus monkeys over four successive years. Primates 21:66–82. Strube MJ. 1988. Averaging correlation coefficients: influence of heterogeneity and set size. J Appl Psychol 73:559–568. Strube MJ. 2001. Bootstrapped confidence intervals for a multiple regression procedure (unpublished computer program). Strube MJ. 2002. Bootstrapped confidence intervals for principal components (unpublished computer program). Suomi SJ. 1991. Uptight and laid-back monkeys: individual differences in the response to social challenges. In: Brauth SE, Hall WS, Dooling RJ, editors. Plasticity of development. Cambridge: MIT Press. p 27–56. Svrakic NM, Svrakic DM, Cloninger CR. 1996. A general quantitative theory of personality development: fundamentals of a self-organizing psychobiological complex. Dev Psychopathol 8:247–272. van Hooff JARM. 1970. A component analysis of the structure of the social behavior of a semi-captive chimpanzee group. Experientia 26:549–550. Virgin Jr CE, Sapolsky RM. 1997. Styles of male social behavior and their endocrine correlates among low-ranking baboons. Am J Primatol 42:25–39. Vitale AF, Visalberghi E, de Lillo C. 1991. Responses to a snake model in captive crabeating macaques (Macaca fascicularis) and captive tufted capuchins (Cebus apella). Int J Primatol 12:277–286. Watson D. 2000. Mood and temperament. New York: Guilford Press. 340p. Youn T, Lyoo IK, Kim JK, Park HJ, Ha KS, Lee DS, Abrams KY, Lee MC, Kwon JS. 2002. Relationship between personality trait and regional cerebral glucose metabolism assessed with positron emission tomography. Biol Psychol 60:109–120. Zuckerman M, Kuhlman DM, Joireman J, Teta P, Kraft M. 1993. A comparison of three structural models for personality: the big three, the big five, and the alternative five. Pers Individ Dif 65:757–768. Zumpe D, Michael RP. 1996. Social factors modulate the effects of hormones on the sexual and aggressive behavior of macaques. Am J Primatol 38:233–261.