Patterns of tooth crown size and shape variation in great apes and humans and species recognition in the hominid fossil record.код для вставкиСкачать
AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 125:303–319 (2004) Patterns of Tooth Crown Size and Shape Variation in Great Apes and Humans and Species Recognition in the Hominid Fossil Record Jeremiah E. Scott1* and Charles A. Lockwood1,2 1 2 Department of Anthropology, Arizona State University, Tempe, Arizona 85287 Institute of Human Origins, Arizona State University, Tempe, Arizona 85287 KEY WORDS dental variation; taxonomy; Hominidae; Homininae; Ponginae ABSTRACT It has been suggested that patterns of craniodental variation in living hominids (Gorilla, Homo, Pan, and Pongo) may be useful for evaluating variation in fossil hominid assemblages. Using this approach, a fossil sample exhibiting a pattern of variation that deviates from one shared among living taxa would be regarded as taxonomically heterogeneous. Here we examine patterns of tooth crown size and shape variation in great apes and humans to determine 1) if these taxa share a pattern of dental variation, and 2) if such a pattern can reliably discriminate between samples that contain single species and those that contain multiple species. We use parametric and nonparametric correlation methods to establish the degree of pattern similarity among taxa, and randomization tests to assess their statistical signiﬁcance. The results of this study show that extant hominids do not share a pattern of dental size variation, and thus these taxa cannot be used to generate expectations for patterns of size variation in fossil hominid species. The hominines (Gorilla, Homo, and Pan) do share a pattern of shape variation in the mandibular dentition; however, Pongo is distinct, and thus it is unclear which, if either, pattern should be expected in fossil hominids. Moreover, in this case, most combined-species samples exhibit patterns of shape variation that are similar to those for single hominine species samples. Thus, although a common pattern of shape variation is present in the mandibular dentition, it is not useful for recognizing taxonomically mixed paleontological samples. Am J Phys Anthropol 125:303–319, 2004. © 2004 Wiley-Liss, Inc. Delineating species is a primary goal of paleoanthropology, as it is a necessary ﬁrst step in the generation and testing of phylogenetic and adaptationist hypotheses. Therefore, much research has been devoted to quantitative methods of identifying taxonomically heterogeneous fossil assemblages, particularly in the hominid1 fossil record (e.g., Lieberman et al., 1988; Kimbel and White, 1988; Kelley and Etler, 1989; Miller, 1991, 2000; Wood, 1991, 1993; Cope and Lacy, 1992; Cope, 1993; Martin and Andrews, 1993; Plavcan, 1993; Teaford et al., 1993; Kramer et al., 1995; Richmond and Jungers, 1995; Lockwood et al., 1996; Grine et al., 1996; Kelley and Plavcan, 1998; Donnelly and Kramer, 1999; Plavcan and Cope, 2001; Wood and Lieberman, 2001). Such investigations typically use extant species as models for interpreting levels of variation in fossil assemblages. In the case of fossil hominids, levels of variation in samples of Gorilla and Pongo, the most sexually dimorphic living hominids, are commonly considered to represent the upper limit of intraspeciﬁc variation, and thus fossil hominid samples that exceed the limits set by these taxa are some- times argued to contain multiple species (e.g., Stringer, 1986; Teaford et al., 1993). However, in some cases, degrees of variation by themselves may not correctly identify taxonomically heterogeneous samples. One problem with relying on the variational limits of extant species is the fact that some fossil taxa may have possessed levels of variation that exceed those observed in samples of living species. Kelley (1993; see also Kelley and Xu, 1991; Kelley and Plavcan, 1998) argued that there is no reason to expect that the maximum level of hominid sexual dimorphism is sampled among extant species (but for a contrary view, see Kay, 1982; Cope and Lacy, 1992; Martin and Andrews, 1993; Plav- Charles A. Lockwood is currently at the Department of Anthropology, University College London, Gower Street, London WC1E 6BT, UK. E-mail: firstname.lastname@example.org *Correspondence to: Jeremiah E. Scott, Department of Anthropology, Box 872402, Arizona State University, Tempe, AZ 85287-2402. E-mail: email@example.com Received 10 March 2003; accepted 22 July 2003. 1 The taxonomy employed in this paper follows that of Groves (1986), who included Gorilla, Homo, Pan, and Pongo in the family Hominidae, while separating Pongo from the African apes and humans into the subfamilies Ponginae and Homininae, respectively. © 2004 WILEY-LISS, INC. DOI 10.1002/ajpa.10406 Published online 12 May 2004 in Wiley InterScience (www. interscience.wiley.com). 304 J.E. SCOTT AND C.A. LOCKWOOD can, 1993; Teaford et al., 1993). Since sexual dimorphism is a major component of intraspeciﬁc variation (e.g., Plavcan, 1990; Wood et al., 1991; Albrecht and Miller, 1993), levels of dimorphism that exceed those observed in modern hominids could produce levels of variation in fossil species that may be incorrectly regarded as excessive for a single species. In other words, assumptions of process uniformitarianism may hold (e.g., the biological mechanisms that produce sexual dimorphism), but still result in cases that violate assumptions of pattern uniformitarianism (e.g., a level of sexual dimorphism in a fossil primate that exceeds that of all known living primates). Another factor complicating the use of magnitudes of variation is the fact that some fossil assemblages may contain multiple taxa without exhibiting excessive variation (Godfrey and Marks, 1991). Tattersall (1986) warned that there is a tendency among paleontologists, and paleoanthropologists in particular, to underestimate taxic diversity in the fossil record, a consequence of the fact that many closely related species do not exhibit marked skeletal or dental differentiation, making it difﬁcult to distinguish between them. Cope (1993) and Plavcan (1993) demonstrated this problem empirically, showing that levels of variation in dental dimensions for some pooled-species samples of sympatric cercopithecids do not exceed those of single-species reference samples. Plavcan (2002, p. 580) further noted that, because a fossil hominid sample generally must exceed Gorilla and Pongo in levels of character variation in order for the single-species hypothesis to be rejected, two species would have to “differ in body size by as much as a factor of 2 before a mixture of the two would be universally accepted as excessively variable.” Not all closely related sympatric primate species exhibit such a size difference (e.g., species of Cercopithecus and Hylobates), and thus, low levels of variation in a fossil sample cannot result in a strong rejection of a multiple-species hypothesis (Kimbel, 1991; Plavcan and Cope, 2001). Wood et al. (1991) suggested that an alternative to using the degree of variation is to use the pattern of variation. Whereas degree refers to magnitude and is generally character-speciﬁc (i.e., different characters possess different magnitudes of variation), the pattern of variation is the magnitude of variation of characters in relation to each other (Fig. 1). Wood et al. (1991) argued that if extant hominids share a pattern of variation, then it might be reasonable to expect fossil hominids to exhibit such a pattern as well. Thus, the presence of a pattern deviating from that of extant hominids should be taken as evidence for the presence of multiple taxa in a fossil sample (Wood, 1993; Kramer et al., 1995). The expected advantage of such an approach can be seen in Figure 1. For example, even though a fossil species that possesses an extreme level of sexual dimorphism may possess a relatively high level of variation, its pattern of variation may not be Fig. 1. Variability proﬁles illustrating difference between pattern of variation and degree of variation. a: Two taxa in which average degree of character variation (as measured by the coefﬁcient of variation, or CV) is different but pattern is similar. b: Two taxa which are similar in average degree of variation but which differ in pattern (adapted from Sokal and Braumann, 1980). unusual (Fig. 1a). Conversely, a fossil sample containing two taxa of similar size may not vary as much in degree as a single highly dimorphic extant species, but its pattern of variation may diverge considerably from the common pattern displayed by closely related extant taxa (Fig. 1b). The utility of the pattern approach in discriminating between fossil assemblages that contain a single species vs. those that contain multiple species depends on two factors. First, the extant comparative taxa must share a pattern of variation. Second, mixed-species samples must exhibit a pattern of variation that differs from that seen in single-species samples. In the case of living hominid species, several studies investigated the degree to which taxonomic patterns of craniodental variation within this clade are similar or different (Lieberman et al., 1985; Oxnard et al., 1985; Oxnard, 1987; O’Higgins et al., 1990; Plavcan, 1990, 2002; Wood et al., 1991; O’Higgins and Dryden, 1993; Kramer et al., 1995; Uchida, 1996, 1998a,b; Lockwood, 1999; Wood and Lieberman, 2001; Ackermann, 2002), while relatively little attention has been paid to how the patterns of variation for mixed-taxon samples compare to those for single-taxon samples. 305 PATTERNS OF DENTAL VARIATION IN HOMINIDS TABLE 1. Samples used in this study1 P. troglodytes4 2 G. g. gorilla Lower I1 I2 C P3 P4 M1 M2 M3 Upper I1 I2 C P3 P4 M1 M2 M3 3 P. paniscus4 P. p. pygmaeus troglodytes schweinfurthii verus 74 76 74 82 82 86 80 68 36 38 40 40 40 40 40 36 62 58 58 70 66 66 72 58 38 36 26 42 46 62 46 30 66 58 64 86 84 88 86 82 82 80 76 88 74 102 102 70 36 38 40 38 38 40 40 38 64 58 56 66 66 70 68 54 40 42 32 52 50 76 54 36 48 46 48 86 82 86 84 70 H. sapiens Nubians5 Fort Center6 30 36 46 42 44 76 46 36 30 30 30 40 40 44 46 42 22/27/19 55/57/49 64/69/57 72/76/57 82/84/81 113/117/111 92/92/91 49/48/48 34 30 42 48 42 72 46 28 267 34 42 44 44 44 58 40 52/54/43 55/56/48 104/103/91 118/117/115 127/120/116 162/165/155 127/127/122 34/34/33 1 All samples contain equal numbers of males and females except where noted. Data are from Mahler (1973). 3 Data are from Plavcan (1990). 4 Data provided by D.C. Johanson. 5 Sample measured by ﬁrst author. 6 Data are from Cucina and Iscan (2000). Sex ratio is unknown. Sample sizes for each tooth are listed as follows: mesiodistal/ buccolingual/crown area. 7 Male to female ratio ⫽ 9:17 (see text for discussion). 2 The present study examines the dentition of living hominids to determine whether or not there is a shared pattern of variation among these taxa, and evaluates the utility of using such a pattern to identify mixed-species samples. While previous studies of pattern have focused on sexual dimorphism, this study examines intraspeciﬁc variation as a whole. This difference in focus is important because, although there is a relationship between sexual dimorphism and intraspeciﬁc variation, particularly when levels of dimorphism are high and intrasexual variation is low (Fleagle et al., 1980; Plavcan, 1994), a measure of sexual dimorphism is not an exact proxy for a measure of variation, particularly when levels of dimorphism are low (Plavcan, 1994; Josephson et al., 1996; Rehg and Leigh, 1999), as they are in the dentition. MATERIALS AND METHODS Samples and measurements The species and sample sizes used in this study are listed in Table 1. Dental metrical data include mesiodistal (MD) and labiolingual (LL)2 or buccolingual (BL) dimensions for each maxillary and mandibular tooth. In general, the left tooth is preferred unless it is missing, in which case the antimere is substituted. Eight single-taxon samples are examined: Gorilla gorilla gorilla, Pongo pygmaeus pygmaeus, Pan troglodytes troglodytes, P. t. schweinfurthii, P. t. verus, P. paniscus, and 2 For the sake of simplicity, the breadth measurement of the incisors and canines will be referred to as its postcanine equivalent, the buccolingual diameter, for the remainder of this paper. two geographically restricted populations of Homo sapiens: a population of Meroitic Nubians from Sudan, and a precontact Florida Indian population. Dental metrical data for G. g. gorilla and P. p. pygmaeus are from Mahler (1973) and Plavcan (1990), respectively. D.C. Johanson provided the data sets for the species and subspecies of Pan (see Johanson, 1974). The Nubian sample, measured by the ﬁrst author, derives from a collection housed at Arizona State University, while the Florida Indian population data set is from Cucina and Iscan (2000). Each sample contains an equal number of males and females, except for the Florida human population, in which the sex ratio is unknown. In the Nubian human population, the sex ratio is balanced for all teeth except for I1. In this case, the male to female ratio is approximately 1:2. However, balancing the sex ratio does not signiﬁcantly alter the level of variation in MD and BL, and thus all teeth are included in the analyses. In measuring the Nubian sample, the measurement deﬁnitions follow those laid out in Mahler (1973) and Plavcan (1990). For the postcanine teeth, the MD dimension is deﬁned as the “greatest dimension of the tooth in the mesiodistal axis” (Plavcan, 1990, p. 40). The BL dimension for the upper premolars, upper molars, and lower premolars is taken as the greatest breadth across the paracone (or paraconid) and protocone (or protoconid) (Plavcan, 1990). This procedure differs slightly from those outlined in Mahler (1973) and Johanson (1974) in that the BL dimension does not necessarily lie perpendicular to the MD dimension. 306 J.E. SCOTT AND C.A. LOCKWOOD For the lower molars, both Johanson (1974) and Plavcan (1990) took two BL measurements, one across the trigonid and one across the talonid. In apes, maximum breadth (the BL dimension used in this study) usually corresponds to the latter measurement. However, in humans, the position of maximum breadth is much more variable, and thus the BL dimension in the Nubian sample is taken as the maximum breadth perpendicular to the MD dimension. For the incisors and canines, the MD dimension is as deﬁned for the postcanine teeth. The BL (⫽ LL) dimension is the maximum width of the tooth perpendicular to the MD dimension. Both Mahler (1973) and Plavcan (1990) commented on the irregular morphology of the canine and P3 in apes, which makes the MD and BL dimensions somewhat difﬁcult to deﬁne for these teeth. Since this analysis focuses on variation, differences in measuring technique should not affect this study, as long as the researchers from whom the data sets employed in this study derive were consistent in their methodology. Furthermore, the canines and P3 should be among the most variable teeth in the ape samples. The difﬁculties encountered in measuring ape canines and P3s do not represent a problem for the Nubian sample. The MD measurements for the Nubian sample are not corrected for interstitial wear. In order to minimize the effect of wear on levels of dental variation, heavily worn teeth are excluded from the analysis. Mahler (1973), Plavcan (1990), and Cucina and Iscan (2000) also did not estimate corrected MD length, but Johanson (1974) did. The effects that both interstitial wear and correcting for interstitial wear have on magnitudes and patterns of variation is unclear, although presumably one of the goals of correcting for wear is to reduce irrelevant variation. The effect that interobserver error has on this analysis is addressed below (see Discussion). Each dimension of each tooth in the Nubian sample was measured three times in order to determine the degree of intraobserver error. The dimension used in the ﬁnal analysis is an average of these three measurements. Measurement error in the Nubian sample is comparable to that for other studies, i.e., less than 2%. Patterns of variation in single-taxon samples Analyses are conducted on patterns of variation in size and shape: MD and BL dimensions together, MD dimensions only, BL dimensions only, tooth crown areas (calculated as the product of MD and BL), and tooth crown shape indices (calculated as the natural logarithm of the ratio of MD to BL). Cucina and Iscan (2000) presented only the summary statistics for MD and BL dimensions and crown areas, and thus this sample is not included in the analyses for patterns of shape variation. Patterns are examined for all maxillary and mandibular teeth together and for each jaw separately. Thus, there are 15 pattern comparisons between each pair of taxa, except for those including the Fort Center human population, for which there are only 12 intertaxic comparisons due to the fact that patterns of shape variation for this sample are not examined. Size variation is quantiﬁed using the coefﬁcient of variation (CV), a measure of relative variation that expresses the standard deviation of a sample as a percentage of the mean (CV ⫽ 100s/X), facilitating comparisons of variability between taxa or characters of different sizes (Simpson et al., 1960; Yablokov, 1974; Sokal and Braumann, 1980; Sokal and Rohlf, 1995). The CV is widely used in studies of variation in both extant and fossil taxa, particularly in the primate fossil record (Gingerich, 1974; Fleagle et al., 1980; Kimbel and White, 1988; Cope and Lacy, 1992; Plavcan, 1994; Lockwood et al., 1996, 2000; Arsuaga et al., 1997; Kelley and Plavcan, 1998). For crown shape, because it is a dimensionless index, the standard deviation (s) is used as the measure of variation. Pearson’s product-moment correlation coefﬁcient (r) and Spearman’s coefﬁcient of rank correlation (rs) are used to express the degree of similarity among the patterns of variation for each taxon. The former is used to compare the patterns produced by the raw CVs (and standard deviations, in the case of shape variation) for each taxon, while the latter is used to compare the patterns generated by the rank orders of those CVs, i.e., the CVs are ranked from 1 to p for p variables, 1 being the least variable and p being the most variable (Lockwood, 1999; Plavcan, 2002). There are two reasons for using both r and rs. First, using the raw CVs has the advantage that similarities and differences in magnitude can be incorporated into the analysis of pattern. However, the disadvantage to using raw CVs is that high magnitudes for a particular character or small group of characters can obscure subtler patterns of variation (Fig. 2). In hominid dentitions, high variation in the canines and P3 may represent a possible confounding factor, i.e., high product-moment correlations may be due to the common difference in the level of variation between these teeth and the others. Thus, the rank correlation test is also performed in order to negate the effect that the absolute magnitude of variation has on the product-moment correlations. However, it is still possible that the canines and P3 will affect the rank correlations. Therefore, for heuristic purposes, all comparisons of patterns of size variation are also carried out with the canines and P3 excluded. These teeth are not excluded for comparisons of patterns of shape variation, as they are not consistently the most variably shaped teeth (see Results). The null hypothesis for all analyses is no pattern similarity between taxa, i.e., no correlation (r or rs ⫽ 0). Developing criteria for what constitutes “similarity” or “difference” is somewhat problematic. Clearly, a high correlation for a comparison between two taxa would suggest the presence of a shared pattern; however, placing a boundary between cor- PATTERNS OF DENTAL VARIATION IN HOMINIDS Fig. 2. Variability proﬁles illustrating difference between using (a) raw CVs and (b) a rank order of those CVs. In a, taxon A’s pattern is more highly correlated with taxon B’s pattern than is taxon C’s pattern using Pearson’s product-moment correlation coefﬁcient, even though A has a divergent pattern of variation in characters 7–12 in comparison to B and C. This is due to fact that A and B possess similar magnitudes of variation in characters 5 and 6. In b, the patterns of B and C are more highly correlated with each other than either is with pattern for A using Spearman’s coefﬁcient of rank correlation because they share the same rank order pattern. Note that in b, the proﬁles of B and C overlap completely, despite fact that they differ in overall magnitudes of variation. relations that indicate similarity vs. those that indicate difference is rather subjective. For example, it is reasonable to assert that two taxa share a pattern when a comparison between the two yields a correlation of r ⫽ 0.85, but a moderate correlation of r ⫽ 0.60 is more difﬁcult to interpret. In this study, we employ two criteria with which to judge pattern similarity. First, we use statistical signiﬁcance of the correlation coefﬁcient. However, as discussed below, some correlations (e.g., r ⫽ 0.42) have the potential to achieve statistical signiﬁcance without being very high due to the large number of data points in the comparison. Therefore, we arbitrarily recognize correlations greater than or equal to r or rs ⫽ 0.67 as indicative of pattern similarity. Randomization tests are used to determine the statistical signiﬁcance of the correlation coefﬁcients, generally following methods outlined in Leigh (1992) and Lockwood et al. (2000). The raw CVs or rank-ordered CVs of one taxon in each pairwise comparison are randomly reordered 1,000 times and the 307 correlation coefﬁcients for each permutation are calculated, creating a distribution of correlation coefﬁcients with which to evaluate the probability of obtaining the r and rs values generated from the actual pairwise comparisons. Probabilities are expressed as (P ⫹ 1)/(R ⫹ 1), where P equals the number of randomized correlation coefﬁcients that exceed the values generated by comparisons between extant taxa, and R is the total number of correlation coefﬁcients calculated from the randomized CVs and ranks. In this study, correlations with a probability of 0.05 or less are considered statistically signiﬁcant. All tests for signiﬁcance are one-tailed, as there is no biological reason to expect any two species in this study to exhibit patterns of variation that are highly negatively correlated, i.e., the opposite of one another. It should be noted that the number of data points included in each comparison determines the magnitude at which correlation coefﬁcients achieve statistical signiﬁcance. For example, in a comparison that includes 16 data points, a correlation coefﬁcient of approximately3 r ⫽ 0.42 represents the lowest statistically signiﬁcant value using a randomization test with 1,000 permutations and an alpha of 0.05 or less. However, reduction of the number of data points from 16 to 8 raises the minimum statistically signiﬁcant correlation to approximately r ⫽ 0.62. Thus, correlations for comparisons with relatively few data points might be high (i.e., r ⬎ 0.60), but not statistically signiﬁcant. Inspection of the magnitudes of correlation coefﬁcients is conducted to reveal cases where this might affect interpretation of results. Patterns of variation in combinedspecies samples In cases where some or all taxa are found to share a pattern, combined-species samples are created in order to evaluate whether taxonomically mixed samples exhibit patterns that allow them to be recognized as such. Combined-species samples are composed of two species that share a pattern; they contain equal numbers of each species and equal numbers of each sex of each species. Clearly, most taxonomically mixed fossil assemblages are neither species- nor sex-balanced. However, this situation is the one most likely to result in a divergent pattern in a mixed-species sample. If these samples do not result in unusual patterns, there would be little reason to simulate alternative sex and species sample compositions. Patterns of variation for combined-species samples are compared to those of single taxa, as outlined above. For patterns of variation to be considered useful for detecting mixed-species fossil assemblages, correlation coefﬁcients for comparisons be- 3 The word “approximately” is used because each randomization test generates a unique set of permutations. Consequently, the minimum value needed to achieve signiﬁcance may be slightly more or less than the actual values given in the example above. 308 J.E. SCOTT AND C.A. LOCKWOOD TABLE 2. Pearson product-moment correlations between CV tooth crown area patterns of variation for each nonhuman taxon (maxillary and mandibular dentitions)1 G. g. gor. G. g. gor P. p. pyg. P. pan. P. t. trg. P. t. sch. P. t. ver. All teeth No canines or P3 0.93*** 0.93*** 0.91*** 0.89*** 0.89*** 0.63* 0.57* 0.29 0.10 0.34 P. p. pyg. All teeth No canines or P3 0.91*** 0.93*** 0.91*** 0.93*** 0.48* 0.09 0.08 0.42 P. pan. All teeth No canines or P3 0.96*** 0.95*** 0.92*** 0.37 0.20 0.38 P. t. trg. P. t. sch. All teeth No canines or P3 All teeth No canines or P3 0.98*** 0.96*** 0.74** 0.66** 0.93*** 0.54* 1 G. g. gor., G. g. gorilla; P. pan., P. paniscus; P. p. pyg, P. p. pygmaeus; P. t. sch., P. t. schweinfurthii; P. t. trg., P. t. troglodytes; P. t. ver., P. t. verus. *P ⬍ 0.05. **P ⬍ 0.01. ***P ⬍ 0.001. tween combined-species samples and single-species samples must be consistently lower than those for comparisons between single-species samples. RESULTS The effect of the canines and P3 The most striking result for the intertaxic pattern comparisons using the raw CVs is how strongly canine size variation affects the product-moment correlations. Table 2 displays the product-moment correlations for crown area patterns of variation (maxillary and mandibular teeth together) in the nonhuman taxa. There is a large discrepancy between the correlations that include the canines and P3 and those that exclude them. For example, when the canines and P3 are included in the comparison between P. t. troglodytes and Pongo, the correlation is r ⫽ 0.93, indicating that the two taxa share a pattern. However, when the canines and P3 are excluded, r ⫽ 0.09, indicating no similarity in pattern. This result occurs because the disparity between the levels of variation in the canines and the rest of the teeth produces a correlation that is essentially based on two data points (Fig. 3). The combination of high variation in the canines and P3 and low variation in the remaining teeth does represent a pattern of variation that is shared among the nonhuman taxa (Wood et al., 1991). However, the statistical signiﬁcance is unreliable. Also, knowing that the canines are always more variable than the other teeth in sexually dimorphic species is not particularly useful for identifying taxonomically mixed fossil hominid assemblages, as the canines in such samples are likely to be the most variable teeth if the constituent taxa are sexually dimorphic in canine size. The primary goal of this study is to determine the degree of overall pattern similarity throughout the dentition among living hominids. In this respect, the product-moment correlations that include the canines and P3s are uninformative, as they obscure the degree of pattern similarity in the noncanine dentitions. For example, the high product-moment correlation for the comparison between P. t. troglodytes and Pongo conceals the lack of a Fig. 3. Crown area coefﬁcients of variation (maxillary and mandibular dentitions together) for P. p. pygmaeus plotted against those for P. t. troglodytes. Each point represents the CV for the crown area of a particular tooth in one taxon plotted against the CV for the same tooth in the other taxon (e.g., the CV for M1 in P. p. pygmaeus vs. the CV for M1 in P. t. troglodytes). Note marked separation of canines from all other teeth. Pearson product-moment correlation for all data points is r ⫽ 0.93. With canines (and P3s) removed, r ⫽ 0.09. shared pattern of crown area variation in the noncanine teeth between these taxa, a signiﬁcant result of this study. All other comparisons of size variation using product-moment correlations produce similar results, i.e., when the canines are included, the correlations are high, but when the canines are excluded, the correlations generally drop, often considerably (e.g., Table 2). While the canines and P3 also affect the rank correlations (see Tables 3– 6), the effect is not as marked, as it is in the product-moment correlations. Returning to the example above, the rank correlation between the crown area patterns of variation for 309 PATTERNS OF DENTAL VARIATION IN HOMINIDS TABLE 3. Summary of Spearman rank correlations among groups of taxa for patterns of size variation in mesiodistal and buccolingual diameters1 Excluding canines and P3 All teeth Maxillary Mandibular All teeth Maxillary Mandibular 0.49** 0.23–0.76 0.47* 0.18–0.76 0.53* 0.23–0.82 0.49** 0.09–0.65 0.42 ⫺0.04–0.64 0.48 ⫺0.03–0.78 0.50** 0.32–0.76 0.49* 0.18–0.76 0.56* 0.23–0.78 0.54** 0.28–0.65 0.50* 0.16–0.64 0.58* 0.12–0.78 0.49** 0.32–0.76 0.44* 0.28–0.69 0.52* 0.23–0.76 0.52** 0.30–0.65 0.50* 0.27–0.60 0.58* 0.12–0.78 0.72*** 0.58–0.76 0.66** 0.62–0.69 0.75** 0.50–0.76 0.58** 0.34–0.65 0.50* 0.44–0.54 0.64* 0.12–0.64 0.82*** 0.72–0.84 0.79*** 0.72–0.92 0.80*** 0.68–0.90 0.68*** 0.60–0.77 0.70** 0.59–0.88 0.59* 0.57–0.91 2 All taxa (24) Median Range Hominines (17) Median Range Pan ⫹ Homo (11) Median Range Pan (3) Median Range P. troglodytes3 (3) Median Range Homo sapiens3 (1) Value4 Nonhuman (12) Median Range 0.48** 0.66*** 0.42–0.76 0.43* 0.62** 0.30–0.76 0.53* 0.58* 0.34–0.82 0.49** 0.35 0.49** 0.09–0.65 0.42 ⫺0.04–0.64 0.57* 0.35 ⫺0.03–0.74 1 Comparisons are interspeciﬁc only, unless otherwise noted. Indicates number of within-group comparisons. Comparisons are subspeciﬁc. 4 Note that comparison is between only two populations of H. sapiens, and thus there is no median or range. *P ⬍ 0.05. **P ⬍ 0.01. ***P ⬍ 0.001. 2 3 TABLE 4. Summary of Spearman rank correlations among groups of taxa for patterns of size variation in mesiodistal diameters1 Excluding canines and P3 All teeth Maxillary Mandibular All teeth Maxillary Mandibular 0.43* ⫺0.12–0.68 0.33 ⫺0.10–0.86 0.35 ⫺0.24–0.95 0.38 ⫺0.005–0.81 0.36 ⫺0.07–0.79 0.49 ⫺0.03–0.94 0.41 0.003–0.68 0.38 ⫺0.10–0.86 0.31 ⫺0.24–0.76 0.40 0.005–0.81 0.43 0.07–0.79 0.49 ⫺0.03–0.89 0.33 0.006–0.63 0.33 0.19–0.69 0.26 ⫺0.10–0.62 0.40 0.005–0.81 0.46 0.07–0.71 0.49 ⫺0.03–0.89 0.59** 0.44–0.63 0.57 0.38–0.69 0.60 0.26–0.62 0.40 0.32–0.55 0.36 0.07–0.54 0.37 0.03–0.66 0.80*** 0.68–0.81 0.90*** 0.83–0.98 0.74* 0.55–0.76 0.64* 0.63–0.84 0.86* 0.75–0.96 0.54 0.43–0.94 0.49* 0.24 0.59** 0.42–0.68 0.48 0.29–0.86 2 All taxa (24) Median Range Hominines (17) Median Range Pan ⫹ Homo (11) Median Range Pan (3) Median Range P. troglodytes3 (3) Median Range Homo sapiens3 (1) Value4 Nonhuman (12) Median Range 0.86** 0.68* 0.26–0.95 0.35 0.37 0.12–0.57 0 0.23 ⫺0.07–0.79 0.83* 0.48 ⫺0.03–0.94 1 Comparisons are interspeciﬁc only, unless otherwise noted. Indicates number of within-group comparisons. 3 Comparisons are subspeciﬁc. 4 Note that comparison is between only two populations of H. sapiens, and thus there is no median or range. *P ⬍ 0.05. **P ⬍ 0.01. ***P ⬍ 0.001. 2 P. t. troglodytes and Pongo is rs ⫽ 0.47 with the canines and P3, and rs ⫽ 0.01 without them. Thus, the difference between the product-moment correlation and the rank correlation in the comparison that includes the canines and P3 is large (r ⫽ 0.93 vs. rs ⫽ 0.47), while the difference between the two types of correlations without the canines and P3 is very small (r ⫽ 0.09 vs. rs ⫽ 0.01). In both cases, the canines (and P3 to a lesser extent) are driving the correlations up. However, in the case of the rank correla- 310 J.E. SCOTT AND C.A. LOCKWOOD TABLE 5. Summary of Spearman rank correlations among groups of taxa for patterns of size variation in buccolingual diameters1 Excluding canines and P3 All teeth Maxillary Mandibular All teeth Maxillary Mandibular 2 All taxa (24) Median Range Hominines (17) Median Range Pan ⫹ Homo (11) Median Range Pan (3) Median Range P. troglodytes3 (3) Median Range Homo sapiens3 (1) Value4 Nonhuman (12) Median Range 0.62** 0.33–0.82 0.55 0.29–0.86 0.65* 0.14–093 0.45 ⫺0.08–0.70 0.35 ⫺0.07–0.79 0.40 ⫺0.37–1.00 0.64** 0.44–0.76 0.52 0.33–0.69 0.69* 0.40–0.93 0.46 0.23–0.61 0.32 0–0.57 0.49 0.09–1.00 0.65** 0.44–0.76 0.52 0.33–0.69 0.71* 0.40–0.93 0.46 0.35–0.61 0.32 0–0.57 0.49 0.20–1.00 0.75*** 0.66–0.76 0.52 0.33–0.55 0.86** 0.74–0.93 0.53* 0.37–0.56 0.29 0–0.32 0.71 0.37–1.00 0.80*** 0.65–0.84 0.62 0.60–0.88 0.86** 0.67–0.93 0.63* 0.38–0.71 0.43 0.39–0.82 0.71 0.37–0.89 0.66** 0.81** 0.63** 0.33–0.82 0.57 0.29–0.86 0.62 0.56 0.14–0.93 0.62* 0.36 ⫺0.08–0.70 0.75* 0.35 ⫺0.07–0.79 0.37 0.23 ⫺0.37–1.00 1 Comparisons are interspeciﬁc only, unless otherwise noted. Indicates number of within-group comparisons. 3 Comparisons are subspeciﬁc. 4 Note that comparison is between only two populations of H. sapiens, and thus there is no median or range. *P ⬍ 0.05. **P ⬍ 0.01. ***P ⬍ 0.001. 2 TABLE 6. Summary of Spearman rank correlations among groups of taxa for patterns of size variation in crown areas1 Excluding canines and P3 All teeth Maxillary Mandibular 0.51* 0.12–0.76 0.38 ⫺0.17–0.79 0.57 0.12–0.86 0.51* 0.12–0.71 0.38 ⫺0.17–0.71 0.51* 0.12–0.71 All teeth Maxillary Mandibular 0.37 ⫺0.07–0.75 0.29 ⫺0.18–0.75 0.40 ⫺0.26–0.94 0.55 0.12–0.86 0.40 0.09–0.75 0.29 ⫺0.18–0.75 0.54 ⫺0.26–0.94 0.38 ⫺0.17–0.71 0.55 0.12–0.86 0.47 0.09–0.75 0.50 ⫺0.18–0.75 0.54 ⫺0.26–0.94 0.59* 0.51–0.64 0.38 0.21–0.52 0.74* 0.52–0.76 0.35 0.10–0.40 0.07 ⫺0.18–0.29 0.37 ⫺0.14–0.60 0.71** 0.62–0.86 0.76* 0.69–0.95 0.69* 0.57–0.79 0.59* 0.47–0.73 0.64 0.54–0.93 0.49 0.31–0.54 0.54* 0.38 0.62 0.43 0.60 0.57* 0.43–0.76 0.40 0.21–0.79 0.67* 0.40–0.83 2 All taxa (24) Median Range Hominines (17) Median Range Pan ⫹ Homo (11) Median Range Pan (3) Median Range P. troglodytes3 (3) Median Range Homo sapiens3 (1) Value4 Nonhuman (12) Median Range 0.56* 0.36 ⫺0.07–0.63 0.10 ⫺0.18–0.68 0.40 ⫺0.14–0.77 1 Comparisons are interspeciﬁc only, unless otherwise noted. Indicates number of within-group comparisons. 3 Comparisons are subspeciﬁc. 4 Note that comparison is between only two populations of H. sapiens, and thus there is no median or range. *P ⬍ 0.05. **P ⬍ 0.01. ***P ⬍ 0.001. 2 tion test, the effect is not as great, and the difference in pattern between the two taxa is more apparent. While the noncanine/P3 product-moment correlations provide a clearer picture in terms of overall degree of pattern similarity among taxa than do the product-moment correlations that include the canine and P3, these results are comparable to those for the noncanine/P3 rank correlation tests. For this reason, and those discussed above, the remainder of this paper will focus on rank correlations. 311 PATTERNS OF DENTAL VARIATION IN HOMINIDS TABLE 7. Summary of Spearman rank correlations among groups of taxa for patterns of shape variation1 Patterns of variation in size Tables 3– 6 summarize the results of the rank correlation tests for patterns of variation in size. Each analysis for a given set of measurements is presented separately, and each table includes the correlations for each pattern of dental variation both with and without the canines and P3. The median rank correlations and rank correlation ranges are presented for: 1) all hominids, 2) all hominines (⫽ African apes and humans), 3) Pan and Homo, 4) Pan, 5) the subspecies of P. troglodytes, 6) the two human populations, and 7) the nonhuman taxa. Groups 1– 6 are clades, and group 7 is a grade. Comparisons summarized for groups 1– 4 and 7 are interspeciﬁc only, while those for groups 5 and 6 are intraspeciﬁc only. For example, in the rank correlation summary for all hominines, comparisons between subspecies of P. troglodytes are not included, but comparisons between P. paniscus and each subspecies of P. troglodytes are. This is done because correlations between subspecies of P. troglodytes are generally high, and thus including them with the interspeciﬁc comparisons would bias the median rank correlations upward. The most important conclusion to be drawn from Tables 3– 6 is that, in general, the median rank correlations for each group of taxa are low and the ranges are wide, indicating that extant hominids as a group do not share a stable pattern of size variation in the dentition. This conclusion is not affected by the presence or absence of the canines and P3 in the analysis. Moreover, the nonhuman comparisons that include the canines and P3 demonstrate that low rank correlations are not restricted to comparisons that include Homo, as might be predicted based on the fact that humans and apes differ in canine dimorphism (Wood et al., 1991). The highest group medians come from comparisons within the genus Pan; however, the rank correlations for these comparisons are not as high as one might expect, ranging from rs ⫽ 0.21 to rs ⫽ 0.98. Signiﬁcantly, removal of the canines and P3 from the analysis results in surprisingly low rank correlations between subspecies of P. troglodytes. Likewise, the two human populations are not very similar to each other in most comparisons. In fact, the Nubian population shares more patterns of variation with some nonhuman samples than it does with the Fort Center population. Although relatively high correlations are found between some taxa (usually nonhuman) in each group, other correlations within that group can be quite low. For example, in a comparison between patterns of variation in BL dimensions (maxillary and mandibular teeth together) for Gorilla and Pongo on one hand and Gorilla and P. t. schweinfurthii on the other, the rank correlations are rs ⫽ 0.82 and rs ⫽ 0.76, respectively. However, a comparison between Pongo and P. t. schweinfurthii yields a rank correlation of rs ⫽ 0.49, indicating that there is All teeth Maxillary Mandibular 0.59** 0.22–0.88 0.46 0.07–0.90 0.71* 0.33–0.95 0.70** 0.53–0.88 0.56 0.24–0.90 0.76* 0.67–0.95 0.76*** 0.53–0.84 0.57 0.33–0.88 0.76* 0.67–0.90 0.76*** 0.54–0.84 0.57 0.38–0.79 0.86** 0.67–0.90 0.70** 0.61–0.78 0.64* 0.57–0.71 0.69* 0.62–0.79 0.55* 0.22–0.84 0.46 0.07–0.83 0.70* 0.33–0.90 2 All taxa (18) Median Range Hominines (12) Median Range Pan ⫹ Homo (7) Median Range Pan (3) Median Range Pan troglodytes3 (3) Median Range Nonhuman taxa (12) Median Range 1 Comparisons are interspeciﬁc only, unless otherwise noted. Fort Center human population is not included in analysis of patterns of shape variation. See Materials and Methods. 2 Indicates number of within-group comparisons. 3 Comparisons are subspeciﬁc only. *P ⬍ 0.05. **P ⬍ 0.01. ***P ⬍ 0.001. a substantial difference between these two taxa, and thus, each is similar to Gorilla in a different way. While the canines, P3, and ﬁrst and second molars of the nonhuman taxa are generally stable in their ranks relative to each other, the other teeth exhibit wide variation in their distributions in different taxa. In fact, removal of the canines and P3 eliminates any suggestion of a shared pattern in many of the nonhuman grade comparisons. Thus, patterns of size variation in living hominids do not appear to be consistent. Patterns of variation in shape Table 7 summarizes the results for comparisons between patterns of variation in shape. The results are presented as in Tables 3– 6 with the following differences: 1) the canines and P3 are not excluded from the analysis, and 2) the Fort Center human population is not included in the comparisons (see Materials and Methods). Table 7 reveals that comparisons among the hominine patterns of shape variation are characterized by relatively high median rank correlations with narrow ranges when all teeth are examined together and when the mandibular dentition is considered separately, indicating the presence of a shared pattern among these taxa. All rank correlations among hominine taxa are statistically signiﬁcant at P ⬍ 0.05. When Pongo is included in the comparisons (Table 7: all taxa, all nonhuman taxa), the median rank correlations are lower and the ranges are relatively wider, indicating a difference between the orangutan and the hominines. Thus, there are two patterns of shape variation in the dentition of extant hominids: a hominine 312 J.E. SCOTT AND C.A. LOCKWOOD TABLE 8. Sexual dimorphism in mandibular canine shape1 ISD t-test P. p. pyg. G. g. gor. P. pan. P. t. trg. P. t. sch. P. t. ver. H. s. (N) 0.896 * 0.984 0.98 0.982 0.984 1.007 0.989 1 Canine shape is calculated as MD/BL. G. g. gor., G. g. gorilla; H. s. (N), H. sapiens (Nubian population); P. pan., P. paniscus; P. p. pyg., P. p. pygmaeus; P. t. sch., P. t. schweinfurthii; P. t. trg., P. t. troglodytes; P. t. ver., P. t. verus. ISD, index of sexual dimorphism ⫽ (male mean/female mean). *Signiﬁcance level of a t-test of difference between males and females: P ⬍ 0.001. parison with Pongo produces a rank correlation that achieves statistical signiﬁcance is Gorilla (rs ⫽ 0.69; P ⬍ 0.05). The rank correlations between the orangutan and all other taxa are not statistically significant, and range from a high of rs ⫽ 0.60 to a low of rs ⫽ 0.33, with a median of 0.45 (n ⫽ 5 comparisons). One factor contributing to the difference between the two subfamilies is shown in Table 8: in comparison to the hominines, the mandibular canine of the orangutan exhibits a relatively high level of sexual dimorphism in shape, with females having relatively narrow canines in comparison to males. Fig. 4. Patterns of crown shape variation in mandibular dentition of (a) all extant hominines (⫽ African apes and humans) and (b) all extant hominids. Rank 1 is least variable; rank 8 is most variable. Note difference between Pongo and hominines in rank of mandibular canine (b). pattern and a pongine pattern. Furthermore, the low median correlations exhibited by the hominine maxillary dentition pattern comparisons imply that the pattern exhibited when both jaws are examined together is largely a product of the pattern in the mandibular dentition. Therefore, only the mandibular pattern is considered hereafter. Figure 4 illustrates the patterns of mandibular tooth crown shape variation for each subfamily. The hominine pattern (Fig. 4a) is characterized by high variation in incisor and premolar shapes, and low variation in canine and molar shapes. This pattern can be further subdivided based on the rank of P3: P. paniscus, P. t. troglodytes, and P. t. schweinfurthii form one group (rs ⱖ 0.79 among all taxa) in which the P3 is the most variable tooth, while Gorilla, Homo, and P. t. verus form a second group (rs ⱖ 0.90 among all taxa) which is characterized by a P3 that is intermediate in its level of shape variation. Intertaxic comparisons between these two subgroups have rank correlations that range from rs ⫽ 0.62 to rs ⫽ 0.76. The orangutan pattern is similar to that of the second group of hominines, with the major distinction being in the rank of the canine (Fig. 4b). In Pongo, the canine is the most variable tooth, whereas in hominines it is among the least variable teeth. The only hominine taxon with which a com- Similarities and differences among individual taxa Tables 9 and 10 list the patterns of variation that are the most similar and most different, respectively, among individual taxa. Only the results for analyses that include the canines and P3 are reported, although those for the analyses in which the canine and P3 are excluded are generally comparable. The most similar taxonomic patterns of variation represent the top 25% of all 342 interspeciﬁc comparisons (in this case, rs ⱖ 0.67). The least similar taxonomic patterns represent the bottom 25% of all interspeciﬁc comparisons (rs ⱕ 0.34). Intraspeciﬁc comparisons with rank correlations that fall within these ranges are also listed. The most obvious conclusion to be drawn from Tables 9 and 10 is that the two human samples are very distinct from the nonhuman taxa in the majority of their patterns, particularly in comparison to Gorilla, Pongo, and P. paniscus. While this is partly due to the difference in sexual dimorphism in the canines and P3 between humans and apes, the human patterns of variation are not more similar to those for the ape samples when these teeth are removed from the analyses. In addition, as mentioned above, the human populations share very few patterns with each other. Table 9 shows that only two of the comparisons between the Nubians and the Fort Center population have rank correlations that are greater than or equal to rs ⫽ 0.67. Conversely, Table 10 shows that only a single comparison between the human populations falls at or below rs ⫽ 0.34. Thus, the human populations are moderately correlated. Interestingly, the Nubian population shares more patterns 313 PATTERNS OF DENTAL VARIATION IN HOMINIDS TABLE 9. Taxonomic patterns of variation that are most similar1 G. g. gor. P. p. pyg. P. pan. P. t. trg. P. t. sch. P. p. pyg. MBA, MBL, ML, BA, BU, BL, CAA, CAU, CAL, CSL MBA, MBU, MU, BU, CAA, CAU, CAL, CSA, CSU, CSL MBA, MBL, MA, ML, BL, CAL, CSL MBU, BU, CSA, CSL P. pan. P. t. trg. P. t. sch. P. t. ver. H. s. (N) CAA, CAL ML ML MBA, MBL, BA, BL, CAL, CSA, CSL MU, BL, CSA, CSU, CSL P. t. ver. MBL, ML, CAL, CSL ML MBA, MBL, MBU, BA, BL, CAL, CSL H. s. (N) CSA, CSU, CSL BU BU, BL, CSA, CSU, CSL H. s. (F) BU MBA, MBU, MBL, MA, MU, ML, BA, BU, BL, CAA, CAU, CAL, CSA, CSU, CSL MBA, MBU, MBL, MA, MU, ML, BA, BL, CAA, CAU, CAL MBL, BA, BL, CAU, CSL MBA, MBU, MBL, MA, MU, BL, CAU, CSA, CSL BL, CAA, CAL, CSA, CSL BA, BU, BL BL, CSL CAL ML, BU 1 Only rank-order comparisons are shown; comparisons using Pearson’s product-moment correlation coefﬁcient are not listed (see text for discussion). Only comparisons having Spearman rank correlations of rs ⫽ 0.67 or greater are reported. Comparisons of patterns of size variation include canines and P3. G. g. gor., G. g. gorilla; H. s. (F), H. sapiens (Fort Center, Florida); H. s. (N), H. sapiens (Nubian population); P. pan., P. paniscus; P. p. pyg., P. p. pygmaeus; P. t. sch., P. t. schweinfurthii; P. t. trg., P. t. troglodytes; P. t. ver., P. t. verus. BA, buccolingual, both jaws; BL, buccolingual, lower jaw; BU, buccolingual, upper jaw; CAA, crown area, both jaws; CAL, crown area, lower jaw; CAU, crown area, upper jaw; CSA, crown shape, both jaws; CSL, crown shape, lower jaw; CSU, crown shape, upper jaw; MA, mesiodistal both jaws; MBA, mesiodistal ⫹ buccolingual, both jaws; MBL, mesiodistal ⫹ buccolingual, lower jaw; MBU, mesiodistal ⫹ buccolingual, upper jaw; ML, mesiodistal, lower jaw; MU, mesiodistal, upper jaw. TABLE 10. Taxonomic patterns of variation that are the most distinct1 G. g. gor. P. p. pyg. P. pan. P. t. trg. P. t. sch. P. t. ver. H. s. (N) P. p. pyg. MU, CSU P. pan. P. t. trg. P. t. sch. P. t. ver. H. s. (N) H. s. (F) MBL, MU, BL, CSA MBU, MU, BA, BU, CAU, CSA, CSU MU, BL, CAU, CSA, CSL CSU BL, CSA, CSU MBA, MBU, MA, MBA, MBU, MBL, MA, MU, ML, CAA, MU, ML, CAA, CAU, CAU, CAL CAL, CSU MA, ML, CAU MBA, MBU, MA, MU, ML, CAA, CAU, CAL ML BU, CAU MBU, MA, MU, MA, MU, ML ML MBU, MU ML, CAA, CAL MBL, MA, MU, MA, MU, ML MBA, MBL, MA, MA, ML, CAU ML, CAA, MU, ML, CAU CAU, CAL MA 1 Only rank-order comparisons are shown; comparisons using Pearson’s product-moment correlation coefﬁcient are not listed (see text for discussion). Spearman rank correlations range from a high of rs ⫽ 0.34 to a low of rs ⫽ ⫺0.24. Comparisons of patterns of size variation include canines and P3. G. g. gor., G. g. gorilla; H. s. (F), H. sapiens (Fort Center, Florida); H. s. (N), H. sapiens (Nubian population); P. pan., P. paniscus; P. p. pyg., P. p. pygmaeus; P. t. sch., P. t. schweinfurthii; P. t. trg., P. t. troglodytes; P. t. ver., P. t. verus. BA, buccolingual, both jaws; BL, buccolingual, lower jaw; BU, buccolingual, upper jaw; CAA, crown area, both jaws; CAL, crown area, lower jaw; CAU, crown area, upper jaw; CSA, crown shape, both jaws; CSL, crown shape, lower jaw; CSU, crown shape, upper jaw; MA, mesiodistal both jaws; MBA, mesiodistal ⫹ buccolingual, both jaws; MBL, mesiodistal ⫹ buccolingual, lower jaw; MBU, mesiodistal ⫹ buccolingual, upper jaw; ML, mesiodistal, lower jaw; MU, mesiodistal, upper jaw. of variation with Gorilla (3), P. paniscus (5), P. t. troglodytes (5), and P. t. schweinfurthii (5) than it does with the Fort Center population. The nonhuman species that share the most patterns are Gorilla and Pongo (10) and Gorilla and P. paniscus (10). The number of patterns that each subspecies of P. troglodytes shares with Gorilla and P. paniscus varies from 4 –7. Thus, overall, P. paniscus is more similar to Gorilla than it is to P. troglodytes. Note also that P. t. troglodytes shares all of its patterns with P. t. schweinfurthii and 11 with P. t. verus, while only 9 comparisons between the latter two taxa have rank correlations greater than or equal to rs ⫽ 0.67. 314 J.E. SCOTT AND C.A. LOCKWOOD TABLE 11. Combined species samples used in this study1 Lower I1 I2 C P3 P4 M1 M2 M3 Upper I1 I2 C P3 P4 M1 M2 M3 P. pan. ⫹ P. t. trg. P. pan. ⫹ P. t. sch. P. pan. ⫹ P. t. ver. G. g. gor. ⫹ P. t. ver. 60 72 92 84 88 132 92 72 60 60 48 68 76 112 76 56 60 72 92 84 88 152 92 72 132 116 112 164 164 172 160 136 60 60 60 80 80 88 92 72 68 60 84 96 80 140 92 56 64 56 56 80 84 136 92 56 68 60 84 96 84 144 92 56 96 92 88 172 148 172 168 140 52 60 84 88 84 88 96 56 P. pan. ⫹ H. s. (N) 1 All samples are species- and sex-balanced. Refer to Table 1 for data sources. G. g. gor., G. g. gorilla; H. s. (N), H. sapiens (Nubian population); P. pan., P. paniscus; P. t. sch., P. t. schweinfurthii; P. t. trg., P. t. troglodytes; P. t. ver., P. t. verus. The nonhuman taxa that have the least number of patterns in common are P. paniscus and Pongo (2) and P. troglodytes and Pongo (1 for each subspecies of P. troglodytes). In fact, comparisons between these taxa yield more low rank correlations (rs ⬍ 0.34) than high correlations (rs ⬎ 0.67), indicating that the Pan clade is very different from Pongo. Patterns of shape variation in combinedspecies samples The hominine pattern of mandibular tooth crown shape variation represents the only pattern that is stable across species and genera in this study, and thus it is the only one that is examined in terms of its ability to discriminate between single- and multiple-species samples. The combined-species samples used in this study (Table 11) simulate three types of taxonomically mixed fossil assemblages: 1) two sister species that overlap in dental size and are similar in tooth shape (P. paniscus ⫹ one of each subspecies of P. troglodytes), 2) two species that do not overlap in size but are similar in shape (P. t. verus ⫹ G. g. gorilla), and two species that overlap in size but differ in shape (P. paniscus ⫹ H. sapiens). In each case, the combined-species sample contains equal numbers from each species and sex. The latter two assemblages are extreme cases, and it is likely that they would be recognized as taxonomically heterogeneous on grounds independent of pattern of variation (e.g., the high magnitude of variation produced by the P. t. verus ⫹ G. g. gorilla sample would immediately mark it as taxonomically mixed, while the obvious differences in canine and P3 morphology between H. sapiens and P. paniscus would similarly indicate the presence of two taxa in the assemblage). However, they are included here in order to repre- sent a range of possible interspeciﬁc combinations, and to determine how the patterns for mixed-species samples behave in extreme circumstances. The results for the comparisons between singleand combined-species patterns of mandibular tooth crown shape variation are presented in Table 12, and the patterns themselves are illustrated in Figure 5. Most comparisons between single- and combined-species samples have rank correlations that are within the range of those derived from singlespecies comparisons. Perhaps surprisingly, the G. g. gorilla ⫹ P. t. verus sample exhibits a pattern that does not differentiate it from single-species samples. This result is due to the fact that, although there is a large disparity in dental size between the two taxa, differences in canine and molar shapes are only slight. The one combined-species sample that does exhibit an unusual pattern is the P. paniscus ⫹ H. sapiens sample. In this case, 4 of the 6 comparisons fall below the low end of the range for single-taxa comparisons and fail to achieve statistical signiﬁcance, thus rejecting the single-species hypothesis (Table 12). Figure 5 shows that the pattern for this sample is similar to the orangutan pattern (rs ⫽ 0.81 between these two samples), with the canine being the most variable tooth; however, with a standard deviation of s ⫽ 0.203, canine shape variation in the P. paniscus ⫹ H. sapiens sample is much more than canine shape variation in the orangutan sample (s ⫽ 0.094). As noted above, this result should be expected due to the marked difference in canine morphology between humans and other hominines, i.e., humans have canines that are longer labiolingually than they are mesiodistally, whereas in the other taxa, the opposite is true. Nevertheless, the rank of the canine is the only real difference between this sample and the single-taxon samples, and thus, correlations for two of the comparisons fall within the range of single-taxon comparisons. DISCUSSION Patterns of variation in extant hominids The results of this study are consistent with previous studies that emphasized the distinctiveness of taxonomic patterns of dental sexual dimorphism within the hominid clade (Lieberman et al., 1985; Oxnard et al., 1985; Oxnard, 1987; Uchida, 1996, 1998a,b). In general, the differences among taxa in patterns of dental variation outweigh the similarities. Of the 15 patterns examined, only one, the hominine pattern of mandibular tooth crown shape variation, is shared across a clade containing multiple genera. Signiﬁcantly, patterns of variation in size are not stable among hominids as a group, among the apes, or even among all subspecies of P. troglodytes. The latter ﬁnding is similar to the observation by Uchida (1998a) that subspecies of G. gorilla exhibit pattern differences in dental sexual dimorphism, indicating that some patterns of vari- 315 PATTERNS OF DENTAL VARIATION IN HOMINIDS TABLE 12. Spearman rank correlations between rank-order mandibular tooth crown shape patterns of variation for single taxon vs. combined-taxa samples1 P. pan. ⫹ P. t. trg. P. pan. ⫹ P. t. sch. P. pan. ⫹ P. t. ver. G. g. gor. ⫹ P. t. ver. P. pan. ⫹ H. s. (N) G. g. gor. P. pan. P. t. trg. P. t. sch. P. t. ver. H. s. (N) Single-species hypothesis rejected2 0.76* 0.81** 0.76* 0.86** 0.50 0.95** 0.90** 0.95** 0.76* 0.64* 0.95** 0.90** 0.93** 0.79* 0.67* 0.81** 0.93** 0.90** 0.86** 0.43 0.64* 0.69* 0.74* 0.81** 0.31 0.76* 0.74* 0.76* 0.86** 0.40 0 0 0 0 4 1 G. g. gor., G. g. gorilla; H. s. (N), H. sapiens (Nubian population); P. pan., P. paniscus; P. t. sch., P. t. schweinfurthii; P. t. trg., P. t. troglodytes; P. t. ver., P. t. verus. 2 Indicates number of combined-species/single-species comparisons in which rank correlation falls below low end of range for single-taxon comparisons (see Table 6). *P ⬍ 0.05. **P ⬍ 0.01. ***P ⬍ 0.001. Fig. 5. Patterns of crown shape variation in mandibular dentition of combined-hominine species samples. Note that most samples exhibit patterns of variation that are similar to those of single hominine taxa (compare to Fig. 4). The exception is P. paniscus ⫹ H. sapiens sample, which has a pattern of variation similar to that for P. p. pygmaeus. ation are speciﬁc to populations rather than species. While the human samples predictably have patterns that are distinct from those exhibited by the ape samples, there also appear to be substantial differences between Pan and Pongo, demonstrating that Homo is not exceptional in possessing a “unique” pattern (see also Oxnard, 1987; Lieberman et al., 1985; Oxnard et al., 1985). Thus, the results of this study indicate that patterns of dental size variation are not consistent among hominids. Wood et al. (1991, p. 199) found “two basic patterns” among hominids, one characterized by the absence of sexual canine dimorphism (Homo), and the other by its presence (Gorilla, Pan, and Pongo). With regard to the patterns of dental dimorphism in the other teeth, Wood et al. (1991, p. 199) described the taxonomic differences as “relatively minor compared to the presence or absence of canine dimorphisms.” While such a conclusion is true, i.e., high variation in the canines is the most prominent feature of the ape patterns of dental variation (as it is among most anthropoids; Plavcan, 1990), the results of this study and others (e.g., Oxnard, 1987) show that patterns of size variation in the other teeth have the potential to vary dramatically among species and even populations, and thus they are not likely to be useful for assessing variation in fossil samples of unknown taxonomic composition.4 Moreover, the pattern of high canine variation/low postcanine tooth variation by itself is probably not useful for evaluating the taxonomic composition of a fossil sample. This is because the canines are also likely to be the most variable teeth in a sample that contains multiple highly dimorphic species. Conversely, if a fossil sample exhibits a pattern in which canine size variation is exceeded by one or more of the postcanine teeth (i.e., a pattern that is the opposite of the one observed in great apes), the pattern is still consistent with one species, if that species follows the pattern seen in Homo. The absence of a stable pattern of size variation in the dentition of extant hominids may be surprising in light of the fact that some studies suggest that there is a shared pattern of cranial variation among these taxa (Wood et al., 1991; Kramer et al., 1995; Wood and Lieberman, 2001). For example, Wood and Lieberman (2001) and Plavcan (2002) showed that different regions of the skull (e.g., the basicranium, face, mandible) exhibit different levels of average variation, i.e., basicranial traits tend to be less variable then palatal and mandibular traits, suggesting that extant taxa can be used to model cranial dimorphism in fossil species. However, it should be noted that intertaxic differences have been documented in hominid cranial variation as well (e.g., O’Higgins et al., 1990; O’Higgins and Dryden, 1993; Lockwood, 1999; Ackermann, 2002). For example, Lockwood (1999) demonstrated that, while P. troglodytes and G. g. gorilla are very similar in their patterns of cranial dimorphism, these taxa differ substantially from G. g. beringei. Moreover, O’Higgins et al. (1990) noted that signiﬁcant differences can be found among modern human populations. In any event, while both similarities and dif- 4 These results should not be confused with patterns of magnitude, i.e., patterns in tooth size. 316 J.E. SCOTT AND C.A. LOCKWOOD ferences in patterns of variation can be identiﬁed among taxa, and each may have implications for components of variation such as sexual dimorphism, the differences imply that patterns of variation may not be useful for detecting mixed-species samples. It should be noted that, because the differences between levels of variation in the noncanine teeth are more subtle than the difference between the canines and noncanine teeth, the samples used in this study may not be large enough to detect a shared pattern. In other words, intertaxic differences in pattern may result from sampling error, i.e., patterns of variation may ﬂuctuate depending on sample size. While it is possible that this may be the case, it is nonetheless clear that, for practical purposes (e.g., species recognition in the fossil record), extant hominids lack a shared pattern of noncanine dental size variation. One might ask whether interobserver error is a possible explanation for some of the differences documented in this study, as the data sets for each genus derive from different sources. However, the fact that only 19 of the 45 (42.2%) interspeciﬁc comparisons (canines and P3 included) within the genus Pan are greater than or equal to rs ⫽ 0.67 (Table 9), despite the fact that these samples were measured by a single researcher (see Materials and Methods; Table 1), and further, that the P. paniscus sample shares more patterns with the Gorilla sample than it does with any of the P. troglodytes samples, suggests that interobserver error is not responsible for the lack of a shared hominid pattern. On the other hand, the hominines do share a pattern of shape variation in the mandibular dentition characterized by low variation in canine and molar shapes and high variation in incisor and premolar shapes. In Pongo, the pattern is slightly different: the canine is the most variably shaped tooth, a consequence of the fact that orangutans are sexually dimorphic in canine shape, but hominines are not. While the hominine pattern could prove useful in evaluating variation in fossil assemblages, the fact that Pongo exhibits a different pattern implies that fossil hominids outside of the hominine clade might also have possessed different patterns, effectively limiting the application of this pattern to the hominin fossil record. Patterns of variation and species recognition in the hominid fossil record There are two ways in which patterns of variation might be used to suggest taxonomic heterogeneity in a fossil sample: a comparison of a fossil sample’s pattern of variation to a shared pattern observable in extant members of its clade, and comparison of a fossil sample’s pattern of variation to the patterns expressed by any extant analogue. In both cases, results are subject to some statistical concerns, especially sample size. Moreover, interpretation requires that mixed-species samples have unusual patterns of variation and that extant taxa thoroughly represent possible patterns of variation. Wood (1993) and Kramer et al. (1995) focused on patterns of metrical variation in the hominin fossil record in evaluating variation in Homo habilis. Various studies noted that this taxon, as broadly deﬁned, exhibits a high degree of craniodental variation (Stringer, 1986; Lieberman et al., 1988; Wood, 1992; Grine et al., 1996). Although Miller (1991, 2000) suggested that it is probable that the magnitude of variation observed in H. habilis could be accommodated within a single, highly dimorphic species, both Wood (1993) and Kramer et al. (1995) found that the pattern of metric variation exhibited by the more complete early Homo crania is unusual in comparison to modern hominid patterns, leading them to reject the single-species hypothesis. Miller (2000) criticized the conclusions of Wood (1993) and Kramer et al. (1995) on the grounds that the sample size of early Homo crania is too small to represent the actual pattern of variation for a population. Indeed, the pattern-based approach to evaluating variation is probably inapplicable to most fossil samples, which generally consist of sample sizes less than n ⫽ 10. Nevertheless, there do exist fossil assemblages of sufﬁcient size to allow for a relatively accurate approximation of their pattern of variation (e.g., the Lufeng and Pasalar Miocene hominoid samples, and the Hadar Australopithecus afarensis dental and mandibular series). In these samples, at least, examination of patterns of variation could prove useful in taxonomic debates. However, beyond the problem of insufﬁcient sample size, the reliability of such a method for identifying taxonomically heterogeneous fossil samples has never been adequately tested. Even if all hominid species possess a similar pattern of variation, it is unclear whether we should expect an assemblage that contains multiple hominid species to exhibit an unusual pattern. The results for the pattern comparisons between single- and combined-species samples in this study bear directly on this issue. First, the fact that the P. paniscus ⫹ H. sapiens sample is identiﬁed as taxonomically mixed based on its pattern of variation in 4 out of 6 comparisons to single-species samples conﬁrms that mixed samples can exhibit unusual patterns. In this case, the two taxa in the sample differ markedly in a character that typically exhibits low variation in single-species samples, i.e., canine shape. However, as noted above, there is probably little need to examine the pattern of variation in a sample containing two taxa as dissimilar as P. paniscus and H. sapiens, as differences in dental proportions and morphology would clearly indicate the presence of two species. On the other hand, the other combined-species samples examined in this study exhibit patterns that cannot be distinguished from those for singlespecies samples, suggesting that this approach has little power in terms of its ability to recognize fossil PATTERNS OF DENTAL VARIATION IN HOMINIDS assemblages that contain more than one species. As demonstrated by the patterns of variation for the combined-Pan samples, when two morphologically similar taxa that share a pattern are mixed together, the resulting sample is likely to exhibit a pattern that is similar to one for a single-species sample. Thus, just as “the presence of a single species in a fossil sample . . . cannot be proved through studies of relative variation” (Cope, 1993, p. 232), the presence of a single species in a fossil sample cannot be proved using patterns of dental variation. Conversely, the presence of an unusual pattern in a fossil sample does not provide unequivocal evidence of taxonomic heterogeneity. Although all hominines share a pattern of mandibular tooth crown shape variation, the possibility that a fossil member of this clade could have possessed a unique pattern cannot be dismissed. This conclusion is supported by the fact that the Pongo pattern of mandibular tooth crown shape variation is unique within the hominid clade. If the Pongo sample used in this study represented an extinct taxon and living hominids were used to evaluate its pattern of shape variation, one might be led to conclude that two species were present in the sample on the basis of “unusual” variation in canine shape. On the contrary, hominid paleontologists should be prepared to encounter novel patterns of mandibular tooth crown shape variation in fossil hominines. Certainly the patterns of shape variation in the maxillary dentition are highly variable within the Homininae, as are patterns of size variation among living hominid species. Kelley (1993) cautioned that imposing the variational limits observed in extant taxa on paleontological species may be a misapplication of the principle of uniformitarianism, i.e., it confuses pattern with process, and thus it restricts our ability to identify novel conditions in fossil taxa. This argument is based on the observation that the degree of postcanine tooth size dimorphism in the Lufeng fossil hominoid sample appears to exceed that of any extant primate, assuming that the sample represents one species, as argued by Kelley and Xu (1991), Kelley (1993), and Kelley and Plavcan (1998). The implication is “that the greatest sexual dimorphism in tooth size, and probably body size, found among extant anthropoid primates does not represent some upper limit during primate evolutionary history” (Kelley, 1993, p. 446). The extent to which paleontologists should use variational limits of extant taxa as guides to interpreting the fossil record is a matter of debate (e.g., Kay, 1982; Cope and Lacy, 1992; Kelley, 1993; Martin and Andrews, 1993; Teaford et al., 1993; Kelley and Plavcan, 1998). Regarding patterns of variation, the results of this study indicate that living taxa do not sample all possible patterns of dental variation, and thus, novel patterns should be expected in paleontological taxa (for a similar conclusion regarding patterns of cranial dimorphism, see Lockwood, 1999; Ackermann, 2002; Plavcan, 2002). Consequently, patterns of varia- 317 tion in fossil samples should be interpreted with caution. Phylogenetic information content of patterns of variation One of the underlying assumptions of this study is that patterns of variation are phylogenetically patterned. This issue was more explicitly addressed by others (e.g., Oxnard, 1987; Wood et al., 1991; Ackermann, 2002; Plavcan, 2002), with most studies concluding that the degree of pattern similarity among taxa is inversely proportional to degree of phylogenetic separation, i.e., as phylogenetic distance increases, pattern similarity decreases. The results presented here do not contradict these conclusions. While most of the patterns examined in this study are not shared across hominids as a group, the number of patterns shared among taxa appears to be related to phylogenetic distance to some degree (see Tables 9 and 10). For example, the number of patterns shared among taxa within the genus Pan is high, the number of patterns shared between species and subspecies of Pan and Pongo is low, and the number of patterns shared between Pan and Gorilla is intermediate. The difference in pattern between Pongo and the hominines in mandibular tooth crown shape variation represents another example of phylogenetic patterning of a pattern of variation. While this pattern could prove useful in distinguishing between fossil pongines and hominines, its polarity remains to be determined. Including Hylobates as an outgroup could clarify this issue; however, the fact that gibbons are sexually monomorphic suggests that a broader comparative sample is required (e.g., all catarrhines or all anthropoids). CONCLUSIONS This analysis of dental variation reveals that extant hominids lack a stable pattern of size variation. While the human patterns are the most distinctive, signiﬁcant differences also exist among the nonhuman taxa (e.g., Pan and Pongo). Some of these differences may result from interobserver error, as the samples for each genus were measured by different researchers. However, the fact that the rank correlations between P. paniscus and subspecies of P. troglodytes (measured by the same researcher) are not particularly high indicates that differences in measurement technique do not affect the conclusions of this study. Thus, patterns of dental metric variation cannot be used to identify taxonomically heterogeneous fossil assemblages, as fossil taxa are likely to exhibit patterns that are as unique as those for extant taxa. The hominines do share a pattern of shape variation in the mandibular dentition. However, the orangutan exhibits a pattern that sets it apart from the hominines, and thus the extent to which either pattern can be expected in the fossil record is un- 318 J.E. SCOTT AND C.A. LOCKWOOD clear. It is also likely that other patterns of shape variation in the mandibular dentition may characterize fossil taxa. Consequently, the presence of an unusual pattern in a fossil sample may not be indicative of the presence of multiple taxa. Furthermore, combined-hominine species samples do not consistently exhibit unusual patterns of mandibular tooth crown shape variation in comparison to single-species samples, underscoring the fact that the pattern approach has little power to identify mixed-species samples using dental variation. ACKNOWLEDGMENTS We thank Bill Kimbel, Diane Hawkey, and two anonymous reviewers for their invaluable comments and suggestions on numerous aspects of this study. We also thank Diane Hawkey and the Arizona State University Department of Anthropology for providing access to the Nubian skeletal collection. LITERATURE CITED Ackermann RR. 2002. Patterns of covariation in the hominoid craniofacial skeleton: implications for paleoanthropological models. J Hum Evol 42:167–187. Albrecht GH, Miller JMA. 1993. Geographic variation in primates: a review with implications for interpreting fossils. In: Kimbel WH, Martin LB, editors. Species, species concepts, and primate evolution. New York: Plenum Press. p 123–161. Arsuaga JL, Carretero JM, Lorenzo C, Gracia A, Martinez I, Bermúdez de Castro JM, Carbonell E. 1997. Size variation in middle Pleistocene humans. Science 277:1086 –1088. Cope DA. 1993. Measures of dental variation as indicators of multiple taxa in samples of sympatric Cercopithecus species. In: Kimbel WH, Martin LB, editors. Species, species concepts, and primate evolution. New York: Plenum Press. p 211–237. Cope DA, Lacy MG. 1992. Falsiﬁcation of a single species hypothesis using the coefﬁcient of variation: a simulation approach. Am J Phys Anthropol 89:359 –378. Cucina A, Iscan MY. 2000. Odontometric changes in pre-contact Florida Indians. Homo 51:220 –234. Donnelly SM, Kramer A. 1999. Testing for multiple species in fossil samples: an evaluation and comparison of tests for equal relative variation. Am J Phys Anthropol 108:507–529. Fleagle JG, Kay RF, Simons EL. 1980. Sexual dimorphism in early anthropoids. Nature 287:328 –330. Gingerich PD. 1974. Size variability of the teeth in living mammals and the diagnosis of closely related sympatric fossil species. J Paleontol 48:895–903. Godfrey L, Marks J. 1991. The nature and origin of primate species. Yrbk Phys Anthropol 34:39 – 68. Grine FE, Jungers WL, Schultz J. 1996. Phenetic afﬁnities among early Homo crania from East and South Africa. J Hum Evol 30:189 –225. Groves CP. 1986. Systematics of the great apes. In: Swindler DR, Erwin J, editors. Comparative primate biology, volume 1: systematics, evolution, and anatomy. New York: Alan R. Liss. p 187–217. Johanson DC. 1974. An odontological study of the chimpanzee with some implications for hominoid evolution. Ph.D. dissertation, University of Chicago. Josephson SC, Juell KE, Rogers AR. 1996. Estimating sexual dimorphism by method-of-moments. Am J Phys Anthropol 100: 191–206. Kay RF. 1982. Sivapithecus simonsi, a new species of Miocene hominoid, with comments on the phylogenetic status of the Ramapithecinae. Intl J Primatol 3:113–173. Kelley J. 1993. Taxonomic implications of sexual dimorphism in Lufengpithecus. In: Kimbel WH, Martin LB, editors. Species, species concepts, and primate evolution. New York: Plenum Press. p 429 – 457. Kelley J, Etler D. 1989. Hominoid dental variability and species number at the late Miocene site of Lufeng, China. Am J Primatol 18:15–34. Kelley J, Plavcan JM. 1998. A simulation test of hominoid species number at Lufeng, China: implications for the use of the coefﬁcient of variation in paleotaxonomy. J Hum Evol 35:577–596. Kelley J, Xu Q. 1991. Extreme sexual dimorphism in a Miocene hominoid. Nature 352:151–153. Kimbel WH. 1991. Species, species concepts, and hominid evolution. J Hum Evol 20:355–371. Kimbel WH, White TD. 1988. Variation, sexual dimorphism and the taxonomy of Australopithecus. In: Grine FE, editor. Evolutionary history of the “robust” australopithecines. New York: Aldine de Gruyter. p 175–192. Kramer A, Donnelly SM, Kidder JH, Ousley SD, Olah SM. 1995. Craniometric variation in large-bodied hominoids: testing the single-species hypothesis for Homo habilis. J Hum Evol 29: 443– 462. Leigh SR. 1992. Cranial capacity evolution in Homo erectus and early Homo sapiens. Am J Phys Anthropol 87:1–13. Lieberman DE, Pilbeam DR, Wood BA. 1988. A probabilistic approach to the problem of sexual dimorphism in Homo habilis: a comparison of KNM-ER 1470 and KNM-ER 1813. J Hum Evol 17:503–511. Lieberman SS, Gelvin BR, Oxnard CE. 1985. Dental sexual dimorphisms in some extant hominoids and ramapithecines form China: a quantitative approach. Am J Primatol 9:305–326. Lockwood CA. 1999. Sexual dimorphism in the face of Australopithecus africanus. Am J Phys Anthropol 108:97–127. Lockwood CA, Richmond BG, Jungers WL, Kimbel WH. 1996. Randomization procedures and sexual dimorphism in Australopithecus afarensis. J Hum Evol 31:537–548. Lockwood CA, Kimbel WH, Johanson DC. 2000. Temporal trends and metric variation in the mandibles and dentition of Australopithecus afarensis. J Hum Evol 39:23–55. Mahler P. 1973. Metric variation in the pongid dentition. Ph.D. dissertation, University of Michigan, Ann Arbor. Martin LB, Andrews P. 1993. Species recognition in middle Miocene hominoids. In: Kimbel WH, Martin LB, editors. Species, species concepts, and primate evolution. New York: Plenum Press. p 393– 427. Miller JA. 1991. Does brain size variability provide evidence of multiple species in Homo habilis? Am J Phys Anthropol 84: 385–398. Miller JMA. 2000. Craniofacial variation in Homo habilis: an analysis of the evidence for multiple species. Am J Phys Anthropol 112:103–128. O’Higgins P, Dryden IL. 1993. Sexual dimorphism in hominoids: further studies of craniofacial shape differences in Pan, Gorilla and Pongo. J Hum Evol 24:183–205. O’Higgins P, Moore WJ, Johnson DR, McAndrew TJ. 1990. Patterns of cranial sexual dimorphism in certain groups of extant hominoids. J Zool Lond 222:399 – 420. Oxnard CE. 1987. Fossils, teeth and sex: new perspectives on human evolution. Seattle: University of Washington Press. Oxnard CE, Lieberman SS, Gelvin BR. 1985. Sexual dimorphisms in dental dimensions of higher primates. Am J Primatol 8:127–152. Plavcan JM. 1990. Sexual dimorphism in the dentition of extant anthropoid primates. Ph.D. dissertation, Duke University. Plavcan JM. 1993. Catarrhine dental variability and species recognition in the fossil record. In: Kimbel WH, Martin LB, editors. Species, species concepts, and primate evolution. New York: Plenum Press. p 239 –263. Plavcan JM. 1994. Comparison of four simple methods for estimating sexual dimorphism in fossils. Am J Phys Anthropol 94:465– 476. Plavcan JM. 2002. Taxonomic variation in the patterns of craniofacial dimorphism in primates. J Hum Evol 42:579 – 608. Plavcan JM, Cope DA. 2001. Metric variation and species recognition in the fossil record. Evol Anthropol 10:204 –222. Rehg JA, Leigh SR. 1999. Estimating sexual dimorphism and size differences in the fossil record: a test of methods. Am J Phys Anthropol 110:95–104. PATTERNS OF DENTAL VARIATION IN HOMINIDS Richmond BG, Jungers WL. 1995. Size variation and sexual dimorphism in Australopithecus afarensis and living hominoids. J Hum Evol 29:229 –245. Simpson GG, Roe A, Lewontin RC. 1960. Quantitative zoology. New York: Harcourt, Brace, and Co. Sokal RR, Braumann CA. 1980. Signiﬁcance tests for coefﬁcients of variation and variability proﬁles. Syst Zool 29:50 – 66. Sokal RR, Rohlf FJ. 1995. Biometry. New York: W.H. Freeman and Co. Stringer CB. 1986. The credibility of Homo habilis. In: Wood B, Martin L, Andrews P, editors. Major topics in primate and human evolution. Cambridge: Cambridge University Press. p 266 –294. Tattersall I. 1986. Species recognition in human paleontology. J Hum Evol 15:165–175. Teaford MF, Walker A, Mugaisi GS. 1993. Species discrimination in Proconsul from Rusinga and Mfangano Islands, Kenya. In: Kimbel WH, Martin LB, editors. Species, species concepts, and primate evolution. New York: Plenum Press. p 373–392. Uchida A. 1996. Craniodental variation among the great apes. Cambridge, MA: Peabody Museum of Archaeology and Ethnology, Harvard University. 319 Uchida A. 1998a. Variation in tooth morphology of Gorilla gorilla. J Hum Evol 34:55–70. Uchida A. 1998b. Variation in tooth morphology of Pongo pygmaeus. J Hum Evol 34:71–79. Wood B. 1991. A palaeontological model for determining the limits of early hominid taxonomic variability. Paleontol Afr 28:71–77. Wood B. 1992. Origin and evolution of the genus Homo. Nature 355:783–790. Wood B. 1993. Early Homo: how many species? In: Kimbel WH, Martin LB, editors. Species, species concepts, and primate evolution. New York: Plenum Press. p 485–522. Wood B, Lieberman DE. 2001. Craniodental variation in Paranthropus boisei: a developmental and functional perspective. Am J Phys Anthropol 116:13–25. Wood BA, Li Y, Willoughby C. 1991. Intraspeciﬁc variation and sexual dimorphism in cranial and dental variables among higher primates and their bearing on the hominid fossil record. J Anat 174:185–205. Yablokov AV. 1974. Variability of mammals. New Delhi: Amerind Publishing Co.