Does brain size variability provide evidence of multiple species in Homo habilis.код для вставкиСкачать
AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 84~385398(1991) Does Brain Size Variability Provide Evidence of Multiple Species in Homo Habilis? JOSEPH A. MILLER Department of Anatomy and Cell Biology, University of Southern California School of Medicine, Los Angeles, California 90033 Endocranial volume, Coefficient of variation, KEY WORDS Hominids, Paleoanthropology ABSTRACT Endocranial volume (ECV) variability as measured by the coefficient of variation (CV) has been important in supporting the view that more than one species is represented in Homo habilis. Supporters of this view used a CV of 10 as a standard to determine that 1)the H. habilis CV of 12.7 indicates multiple species and 2) there is a low probability of H. habilis specimens KNM-ER 1470 and KNM-ER 1813 being members of the same taxon. This study examines published data for ECVs of fossil and extant hominoids to determine whether CV yields any information regarding species number in H. habilis. Results indicate that there is no empirical basis for using a CV of 10 as a standard to detect multiple species in H. habilis. Also, geography, time, sample choice, sex ratio, and measurement technique are complicating factors that must be considered when interpreting CVs for fossil samples. Additionally, the broad 95% statistical confidence limits (5.1-20.3) indicate that the CV estimate of 12.7 for H. habilis is not sufficiently reliable to allow biologically meaningful interpretation. However, if the CV for H. habilis is actually 12.7, it still falls within the range of variation for single species of modern hominoids. The evidence from ECV variability does not support the argument for multiple species in H. habilis. Controversy surrounds the interpretation of the mor hological variability of specimens attribute to Homo habilis. Some interpret this variability as indicative of a variable, polymorphic species (Howell, 1978; Johanson et al., 1987; Tobias, 1987; Johanson and Shreeve, 1989). Others consider it as evidence of more than one species (Leakey and Walker, 1980; Walker, 1981; Wood, 1985; Stringer, 1986; Chamberlain and Wood, 1987; Lieberman et al., 1988; Groves, 1989). The argument for multiple s ecies is supported by two lines of morpYlological evidence l)craniofacial morphometrics and 2) brain size as estimated by endocranial volume (ECV).Evidence from craniofacial variability was presented by Stringer (19861, who examined the pattern of midsagittal cranial and transverse facial angles of H. habilis relative to extant hominoids. Lieberman et al. (1988) em loyed a probability model based on a samp e of gorillas to determine whether sexual dimorphism could ac- a P @ 1991 WILEY-LISS. INC count for the craniofacial difference between H. habilis specimens KNM-ER 1470 and KNM-ER 1813. Endocranial volume variability as measured by the coefficient of variation (CV) has been considered by some authors to be important evidence supporting the h pothesis that more than one species may e represented in H. habilis. Stringer (1986) argued that a CV of 12.4 for H. habilis indicates multiple s ecies. Wood (1985) used a CV of 10 to calc3ate a low robability that KNMER 1470 and KNM-EE 1813 are members of the same species. Both arguments are based on the assum tion that a CV of 10 can be used effective y to detect multiple species in H. habilis. This study reviews published data for ECVs of extant and fossil hominoids to answer the following questions. 1)Is there an z f Received March 15,1990; accepted September 13,1990 386 J.A. MILLER empirical basis for using a CV of 10 to detect mixed-species samples? 2) What factors confound the interpretation of CVs for paleontological samples? 3) How do the statistical consequences of small sample size affect the reliability of CV estimates? For this study, H. habilis refers to Stringer’s (1986) sample of nine specimens (see Appendix A). THE COEFFICIENT OF VARIATION The CV is a measure of relative variability that is calculated b dividing the standard deviation (SD) by t e sample mean (CV = SD/mean x 100). This provides a rough size correction that allows comparisons of variability to be made between different-sized animals or species. CVs are commonlyused to 1) document the relative variability of a given character, 2) examine the relative variability of homologous characters between subspecies or species (Sokal and Braumann, 1980; Sokal and Rohlf, 1981). 3) identify characters of low variability that might be useful taxonomically (Simpson et al., 1960), and 4) indicate heterogeneity in neontological samples (Simpson et al., 1960).The last practice was described by Mayr (1969:170):“If. . . the CV for a certain statistic fluctuates around 4.5 in a series of samples, but is 9.2 in one sample, such a sample should be reinvestigated.” Such heterogeneity might be due to the inclusion of different subspecies, an additional sibling species, wrongly sexed specimens, or some other source of variability (Ma r, 1969). Tze use of CV as an indicator of heterogeneity in neontological samples has been generalized by some investigators to detecting multiple species in fossil samples (e.g., H. habilis). This has been done using CVs of extant and fossil samples as a basis for determining what a “reasonable” CV “should be” for a given fossil taxon. If the fossil taxon in question has a higher than expected CV, then a mixed-species sample has been inferred by several investigators, but such an approach ignores other possible sources of higher variability as indicated by Mayr (1969). When calculating CV, Sokal and Braumann (1980) recommended a correction for bias introduced by small sample size (N): i CV* = (1+ 1/4N) x CV. All CVs in the present study were adjusted using this correction. Thus for H. habilis, a corrected value of 12.7 is used rather than Stringer’s (1986) value of 12.4. IS CV = 10 A REASONABLE STANDARD TO DETECT MULTIPLE SPECIES? Stringer (1986) argued that a CV of ECV greater than 10 indicates multiple species in H. habilis. This contention was based on CVs for small samples of extant and fossil hominoid taxa considered to be analogs for H. habilis. These values were 9.7 for pygmy chimps (N = 40), 8.9 for common chimps (N = 331, 11.0 for gorillas (N = 39), 10.3 for modern humans (N = 131, 9.0 for H. erectus (N = 51, and 10.3 for Neandertals (N = 6). Stringer (1986:270) claimed “it is evident from these data that a coefficient of variation of about 10% is a reasonable figure for a hominoid species.”Because the CV of 12.7for H. habilis was larger than that for any of the hominoid samples, he suggested that more than one taxon was present. Published data for other small samples of hominoids, however, often indicate CVs greater than 10. The gorilla sample (N = 40) of Lieberman et al. (1988) has a CV of 13.02. Smaller series of modern humans have CVs ranging from 6.1 to 13.5 (Table 1).My ECV data for all the measurable adult orangutans (N = 10) in the Museum of Comparative Zoology (Harvard University) yield a CV of 18.3. More importantly, published data for large, combined-sexSam les of extant hominoids indicate that the 8V of ECV is often greater than 10 (Fig. 1). Tobias’s (1971) CVs are 9.7 for common chimps (N = 3631, 11.0 for orangutans (N = 402), 13.1 for gorillas [N = 668; or 12.9 if the extreme ECV of 752 cc discussed by Tobias (1971:40)is omitted], and 12.6-14.9 for modern humans ( N = l,OOOs, males only). The CV for Randall’s (1943) gorilla sample (N = 151) is 13.2. The CVs for four large series of modern humans [Naqada (N = 2061, Hythe (N = 1931, Whitechapel (N = 154)and Farringdon (N = 13211 range from 8.1 to 11.6 (Table 1).These CVs are more reliable indicators of true population CVs since they are based on larger Sam les. Thus values for both lar e and sma 1 samples of extant hominoids emonstrate that there is no empirical basis for using a CV of 10 to detect multiple species in H. habilis. Additionally, these data show that the H. habilis CV of 12.7 is within the range of variation for single species of modern hominoids. Employing a different method, Wood P % 387 BRAIN SIZE VARIABILITY IN HOMO HABILIS TABLE 1. Endocranial volumes [mi) and CVs for some modern human populations N Mean SD CV 206 154 132 52 27 63 52 64 68 92 33 54 15 12 11 85 60 28 38 24 193 30 9 7 1,330.0 1,385.3 1,369.9 1,409.6 1,423.7 1,251.3 1,418.8 1,378.4 1,301.3 1,310.9 1,340.8 1,337.4 1,280.8 1,319.2 1,473.5 1,258.2 1,354.2 1,245.0 1,447.7 1,472.2 1,396.8 1,335.0 1.441.1 1;455.7 108.0 147.3 158.6 134.9 118.3 118.4 137.7 141.9 137.3 139.3 148.6 127.0 127.6 137.6 128.8 102.3 131.1 112.9 113.6 89.0 128.9 107.8 188.8 179.9 8.1 10.6 11.6 10.9 8.4 9.5 9.8 10.3 10.6 10.7 11.2 9.5 10.1 10.6 8.9 8.2 9.7 9.1 7.9 6.1 9.2 8.1 13.5 12.8 Series Naqada Whitechapel Farringdon Moorfields Easter isle Teita Nepalese Spitalfield Congo Gaboon Gaboon Burmese A Burmese B Burmese C Bidford-on-Avon New Britain Ninth Dynasty Tasmanian Maiden Castle Scottish Hythe First Dynasty Timor-Laut Esquimaux Region Source of raw data Africa Europe Europe Europe Oceania Africa Asia Europe Africa Africa Africa Asia Asia Asia Europe Oceania Africa Australia Europe Europe Europe Africa Australia North America Fawcett and Lee (1902) MacDonell (1904) Hooke (1926) MacDonell (1906) van Bonin (1931) Kitson (1931) Morant (1924) Morant and Hoadley (1931) Benington (1912) Benington (1912) Benington (1912) Tildesley (1921) Tildesley (1921) Tildesley (1921) Brash and Young (1935) van Bonin (1936) Woo (1930) Wunderly (1939) Goodman and Morant (1940) Reid and Morant (1928) Stoessiger and Morant (1932) Morant (1925) Garson (1884) Duckworth (1896) range for males only H. HABlLlS CV = 12.7 P E3 4 K 0 9 14: 12 L 0 132 L 0 39 154 t- z w 0 LL 402 lo 0 13 0 40 363 0 193 0 206 LL w 0 0 Fig. 1. A comparison of CVs of ECV for Homo habilis and large samples of extant hominoids. Open rectangles, Stringer (1986); solid rectangles, Tobias (1971); stipled rectangle, Randall (1943); black circles, Miller (see Table 1).Numbers beside symbols indicate sample size. Horizontal line indicates CV = 10. (1985) used a CV of 10 to calculate a low probability that KNM-ER 1813(510 ml) and KNM-ER 1470 (770 ml) are members of the same population. He used the ECVs of these specimens to calculate a hypothetical population mean of 640 ml. Using a CV of 10, Wood (1985) solved the formula CV = SD/ mean x 100 to obtain a standard deviation of 64 ml. He then determined the approximate 95% opulationlimits to be 512-768 ml (calculate as 2 2 SD from the mean). Since the ECVs for KNM-ER 1813 (510 ml) and B 388 J.A. MILLER KNM-ER 1470 (770 ml) fall outside these limits, the probability of either belonging to the same population would be 5%. The probability of both being members of the same population would be 0.25%. Disregarding the possibility that these two specimens might be members of two different local populations of the same species, Wood (1985: 212) concluded that KNM-ER 1813 and KNM-ER 1470 “should not be uncritically regarded as belonging to the same taxon.” One problem is that Wood (1985) used an ECV of 770 ml for KNM-ER 1470 (a preliminary estimate attributed to Holloway by Day et al., 1975) rather than Holloways (1978) later estimate of 752 ml. When the latter value is used, the mean becomes 631-ml and the 95% PO ulation limits are 505-757 ml. Now, both&M-ER 1813and KNM-ER 1470 can be accommodated in a single PO ulation. The conclusions of Wood (19857 should also be reexamined using CVs other than 10. Employing Wood’s (1985)method, if Tobias’s (1971)gorilla CV of 13.1, and if the corrected mean of 631 ml are used, then the 95% population limits are 466-796 ml (or 472808 ml using Wood’s mean of 640 ml). If, as Tobias (personal communication) suggests, the CV of 11.3 for his H. erectus pekinensis sample (Tobias, 1987)is used, then the limits are 488-774 ml (or 495-785 ml using Wood’s mean). If a CV of 10.8 for a sample of H. erectus erectus (Tobias, 1987) is used, then the limits are 495-767 ml (or 502-778 ml usin Wood’s mean). In all the above cases, the ECVs of KNM-ER 1813 and KNM-ER 1470 fall within the 95% population limits. Thus, Wood’s (1985) approach provides no evidence that H. habilis comprises more than one taxon. A potential justification for using a CV of 10 to detect multiple species comes from Simpson et al. (19601,who observed that the majority of CVs for linear measurements tend to fall between 4 and 10. However, this does not imply that a CV of 10 is a useful standard for area or volumetric measurements. Lande (1977) and Yablokov (1974, citing Schmalhausen, 1935) observe that CVs for traits of different dimensionalities are not directly comparable and that a CV for a volumetric measurement may be up to three times as large as the average CV of its component linear dimensions. Data from two samples of modern H. sapiens illustrate this point (Table 2). In the first sample (N = 11) from Bidford-on-Avon, England (Brash and Young, 19351,the respective CVs TABLE 2. A comparison of CVs for endocranial volume (ml) and linear cranial dimensions lmmi for two samwles of modern humans Bidford-on-Avon (N = 11) Length Breadth Height ECV (Brash and Young, 1935) Mean SD CV Timor-Laut (N = 9) (Garson, 1884) Mean SD CV 191.0 8.4 141.1 6.3 136.7 6.3 1,473.5 128.8 166.3 144.7 135.6 1,441.1 4.5 4.6 4.7 8.9 7.1 7.8 5.3 188.8 4.4 5.5 4.0 13.5 for cranial length, breadth, and height are 4.5, 4.6, and 4.7. Thus the average CV for cranial linear dimensions is 4.6. However, for endocranial volume, the CV of 8.9 is almost twice as great. In the second sample (N = 9) from Timor-Laut (Garson, 18841,the CVs for cranial length, breadth, and height are 4.4, 5.5, and 4.0, respectively, with an average CV of 4.6, yet, for endocranial volume, the CV of 13.5 is almost three times as great. For H. habilis, this means that if a CV of 10 is an upper threshold for cranial linear dimensions (after Simpson et al., 1960),and, if CVs for volumetric measurements are up to three times as great as those for linear dimensions (after Lande, 1977; Yablokov, 1974), then the “threshold CV for detecting multiple species using ECV should be about 30. Alternatively, the CV for the cube root of ECVs for H. habilis can be calculated (see Yablokov, 1974). The resulting CV of 4.3 is comparable to those for the linear dimensions of the human cranium and is below the threshold CV of 10 for linear measurements. WHAT FACTORS CONFOUND THE INTERPRETATION OF CVS? Although CVs are intended to measure the relative variability of a given trait, they are subject to various biasing factors that can confound their interpretation. These factors include geogra hic variation, tem oral variation, sample c oice, sex ratio, anEpmeasurement technique. The inability to control for these factors in paleontological samples makes it difficult to compare CVs of fossil samples with those for extant samples. Simpson et al. (1960) warned that CVs for fossil samples are usually higher than those for extant samples. Geographic variation Elevated CVs for a sample can result from combining geographically varying popula- R 389 BRAIN SIZE VARIABILITY IN HOMO HABILIS tions of the same species (Mayr, 1969). Samples of modern humans from Tasmania and Scotland (Fig. 2) differ in mean ECV (Tasmanian 1,245.0 ml, Scottish 1,472.2 ml) and have CVs under 10 (9.2 and 6.1, respectively). However, the CV for the combined samples is 11.4. This elevated CV represents a complex interaction of within-group and between-group variability. A similar effect on CV might apply to the H. habilis sample, since it consists of specimens from localities that are geographically widely separated. Of course, it is also possible that the combined CV for two geographically varying samples could be intermediate or lower than that of the individual samples. This argues for caution in uncritically accepting hi h CVs as evidence of heterogeneity or low Vs as evidence of homogeneity. Temporal variation An elevated CV can result for a paleontological sample that includes specimens from different time periods. This point cannot be demonstrated convincingly with fossil samples, since the temporal continuity and taxonomic distinctiveness of fossil samples are roblematical. The data presented above Often or geographic variation, however, can be reformulated into a hypothetical example to demonstrate the effect of temporal changes in ECV and its variability between ancestral and descendant populations. The Tasmanian sample (Fig. 2) can be treated as a hypothetical ancestral population at time A that evolves into a descendent population (represented b the Scottish Sam le) a t a later time B. T us, at time A, the V would be 9.2; at time B, it would be 6.1. A paleontological sample that combined specimens from these two hypothetical time periods together would have a CV of 11.4. Such an elevation of CV due to temporal variability is, in part, why Simpson et al. (1960) cautioned that CVs for fossil taxa are often higher than those for extant taxa. For H. habilis, however, the brief time duration of the taxon, fragmentary nature of the specimens, and small sample size prevent a reliable assessment of its temporal variability (if any). Sample choice The CV for a given sample depends on the specimens chosen to represent that sample. Specimen choice differs among investigators and depends, in large art, on one’s taxonomic preference. The e fect of sample choice 8 P K 8 P on CV is demonstrated with the following two examples. Neandertals. The five samples shown in Figure 3 are of six specimens each taken from a group of 13 Neandertal specimens that man consider to represent a single species. T ese samples were chosen to illustrate the remarkable range in CV that may potentially occur with sample choice. A CV of 3.8 is calculated for the group consisting of the Spy 11, Neandertal, Tabun I, Ganovce, Gibraltar, and La Quina specimens. A CV of 10.2 is calculated for the group consisting of the Spy 11, Neandertal, Monte Circeo I, La Chapelle, Tabun 1,and La Quina specimens. A CV of 16.9 results for the group consisting of the La Ferrassie I, La Chapelle, Saccopastore I, Tabun I, Amud I, and Gibraltar specimens. These extraordinary differences in CV are merely a consequence of sampling and document the potential variability inherent in CVs of small fossil samples. Homo erectus. Figure 4 shows a more complicated example of 10 different geographic, temporal, and taxonomic groupings of specimens that might conceivably be used to represent H. erectus. The CVs for these groups are as follows: East Africa (15.41, all H . erectus including Solo (14.5), all H . erectus excluding Solo (13.2), all Africa (13.6), East Africa excluding OH 12 (13.51, East Africa plus Java and China (13.31, Zhoukoudian (12.71, all Asia excluding Solo (12.41, Java excluding Solo (10.21, and Solo (8.1). This wide spectrum of values demonstrates how CVs can vary with sample choice. Given this variability and the difficulty in controlling for the effects of geography, time, and other factors on CV, caution should be employed when the variability of one fossil or extant sample is used as an analog for that of another. Sex ratio The relative numbers of males and females in a sample may affect the CV when sexual dimorphism is present. The gorilla sample (N = 668) from Tobias (19711, for example, consists of 414 (62%)males (mean ECV 534.6 ml, SD 57.05 ml) and 254 (38%) females (mean ECV 455.6 ml, SD 45.13 ml; combined-sex CV 13.1). This Sam le can be manipulated to demonstrate the e fect of sex ratio on CV. This is done by first recalculating the sum of squares for hypothetical combined-sex samples that contain varying roportions of males and females while holling constant the male and female mean ECVs, K P 390 J.A. MILLER GEOGRAPHIC VARIATION c TASMANIAN SCOTTISH COMBINED 11.4 Z 12r TIME A TIME B I COMBINED + I TEMPORAL VARIATION Fig. 2. Elevation of CV of ECV for combined geographic samples of modern humans (Tazmanian and Scottish). The same data illustrate elevated CV for the combined tem oral sample if the samples at times A and B are treated as a single hypothetical population evoking through time. CVs are listed above bars. z Q *O[ 16.9 i- a 5 15 LL O 10 Iz W g 5 LL W 0 O O 2, 4, 9, 11,12&13 1 , 3, 4 , 5,6&7 2,4, 6, 7,9813 1, 3, 5, 9,11&12 3 , 7 , a, 9,10&12 NEANDERTAL SAMPLES Fig. 3. Range of CVs of ECV for five samples of Neandertals. CVs are listed above bars. Bar labels refer to specimens listed in Appendix A. Horizontal line indicates CV = 10. SDs, and total sample size as given above. The resulting sum of squares is then used to calculate a new combined-sex SD, which, when divided by the weighted ECV mean for that sample, yields the new CV. Using this method, the following CVs are obtained for the hypothetical gorilla samples (Fig. 5): 13.1 (50% female), 13.0 (60% female), 12.7 (70%female), 12.2 (80%female), 11.3 (90% female), and 9.9 (100% female). 391 BRAIN SIZE VARIABILITY IN HOMO HABILIS E. Africa I All H. erectus including Solo I All H. erectus excluding Solo 16 E. Africa excluding OH 12 E.Africa, Java, China ( 1 .O- 0.8mya ) Zhoukoudian All Asia excluding Solo I Java excluding Solo 14 I 12 10 8 Specimens 1-5 1-28 Time Span 1.06 1.58 1.52 1-22 1-5 & 14 1-4 5 - 1 3 15-21 1.38 0.45 0.20 0.23 6 - 1 3 6 - 1 2 23-28 15 22 - 0.74 0.20 0.20 ( m y ) Fig. 4. CVs of ECV for 10 samples of Homo erectus. Samples were chosen on the basis of different geographic, temporal, and taxonomic criteria. Bar labels refer t o specimens listed in Appendix A. Numbers below bar labels indicate approximate time spans (durations in millions of years) for each sample calculated from data in Day (1986)and Feibel et al. (1989).Horizontal line indicates CV = 10. males 5 0 % females 5 0 % 40% 60% 30% 70% 20% 80% 10% 90% 0% 100% SEX RATIO Fig. 5. Effect of sex ratio on CV of ECV for a gorilla sample from Tobias (1971). Numbers above bars indicate CV. Bar labels indicate sex ratio for each hypothetical sample. Horizontal line indicates CV = 10. 392 J.A. MILLER r 8.5 Mustard Seed Poppy Seed Manouvrier Formula ( modified ) Pearson Formula ( modified ) MEASUREMENT TECHNIQUE Fig. 6. Four different CVs for a single chimpanzee sample (N = 18) whose ECVs were measured four different ways (data from Protsch von Zieten et al., 1987). CVs are listed above bars. Thus, in this case, the difference between a CV of 9.9 and 13.1 is not indicative of interspecific variation but is, instead, a reflection of sex-ratio differences in a sexually dimorphic taxon. Measurement technique The CV is affected b how endocranial volumes are measured. $rotsch von Zieten et al. (1987)used four different techniques to measure ECVs of 18 common chimpanzees. Two methods involved different filling material (mustard or poppy seed) oured into the crania. The other two metho$s involved the use of different formulae to estimate ECV from linear dimensions. The CVs calculated for these techniques range from 5.5 to 8.5 (Fig. 6). This nonbiological variation in CV approximates the difference between a CV of 10 and the H. habilis CV of 12.7. The measurement effects described above may be compounded when estimating ECV for fragmentary fossil specimens. For example, from 1964 to the present, the ECV of H. habilis specimen OH 7 has been estimated by various workers using a variety of techniques. The fragmentary nature of the s ecimen (two patial parietals) has resulte in estimates ranging from 560 ml to over 800 ml: 642.7-723.6 (Tobias, 1964); 684 ml (Tobias, 1971); 560 and 650 ml, (Kochetkova, 1978); 700-750 ml (Holloway, 1980a), 580600 ml (Wolpoff 1981);674 ml (Tobias, 1983); t; 539-868 ml (Vaisnys et al., 1984).This range nearly equals the entire range of ECV estimates for all H. habilis specimens. As several other H . habilis specimens (e.g., OH 13 and OH 16)are nearly as fragmentary as OH 7, it follows that their ECV estimates may be as problematical. The uncertainty of ECV estimates for fragmentary fossil specimens translates into problems in estimating CV.The CV for the same group of fossil specimens can vary markedly between observers who use different measuring techniques. For example, Figure 7A shows two sets of ECV estimates for a single group of six Indonesian H. erectus specimens. A CV of 10.3 is calculated from Holloway’s (1981a)ECV estimates, whereas a CV of 13.1 is based on previous ECV estimates made by others including Weidenreich (1943),von Koenigswald (19621,Tobias and von Koenigswald (1964)) and Tobias (1971). Similarly, Figure 7B shows that for Australopithecus africanus, a CV of 5.1 is calculated from Holloway’s ECV estimates, whereas a CV of 10.3 is derived from previous ECV estimates by Sche ers (1950) and Dart (1962). Recently, Hol oway’s 435 ml estimate for MLD 37/38 was tested by Conroy et al. (1990) who used an independent technique to make an estimate of 425 ml. Although there is good agreement between these two estimates, Holloway (1972,1981a) himself warns that his technique of estimat- P 393 BRAIN SIZE VARIABILITY IN HOMO HABILIS (A) W 5 3 $ 950 di d 0 850 i 0 750 V = 13.1 4 w 54 S2 T2 S10 S12 517 HOMO ERECTUS SPECIMEN - E 550 W f6 500 > 4 2 450 d 0 0 W 400 1- CV = 5.1 Taung STS60 STSS STS 19 STS71 MLD37!38 AUSTRALOPITHECUS AFRICANUS SPECIMEN Fig. 7. CVs for different estimates of ECV for samples of Homo erectus (A) and Australopitizecus africunus (B).For both A and B, open squares indicate Holloway’s ECV estimates, black squares indicate previous estimates. See Appendix for sources of previous estimates. ing ECV may underestimate the actual variability between specimens, Clearly, CVs are dependent on determinations of ECVs which may vary, sometimes quite markedly, between investigators. WHAT ARE THE STATISTICAL CONSEQUENCES OF SMALL SAMPLE SIZE? When considering estimations of CV for small fossil sam les, the statistical consequences of samp e size must be addressed. P This may be accomplished b using the 95% confidence limits for CV. Tyhese are calculated as CV plus or minus the standard error multiplied by the appropriate t value ( a = 0.05 for a two-tailed test usingN - 1degrees of freedom; Sokal and Rohlf, 1981). The formula for calculating the standard error of a CV is given b Sokal and Braumann (1980). In general, t e confidence limits around a given CV may be used as a reliability indicator of that estimate. All other things being i 394 J.A. MILLER (A) N=5 LL 0 N=6 15 t N=9 ( males only ) N=668 N = 1000s I- ; - 10 0 LL L L 5 w 8 0 Fig. 8. Ninety-five percent confidence limits for CVs of ECV. A: Stringer's (1986) samples of extant and fossil hominoids. B: Homo habilis and Tobias's (1971) samples of extant hominoids. Horizontal bars indicate CV. Rectangles indicate 95% confidence limits. Horizontal line indicates CV = 10. equal, the larger the sample, the tighter the confidence limits and the better the sample estimate is likely t o be of the true population value. The reliability of the CVs for Stringer's (1986) small hominoid samples can be checked by computing the 95% confidence limits. These limits for his samples are as follows (Fig. 8A): Gorilla gorilla (8.4-13.4), modern Homo sapiens (5.6-15.0), H. erectus (Solo) (0-18.31, Neandertals (1.5-19.11, Pan troglodytes (6.6-11.21, and Pan paniscus (7.4-12.0). These broad confidence intervals support Andrew's (1978:203-204) claim that for smaller samples, the confidence limits are often so broad that they make CVuseless 395 BRAIN SIZE VARIABILITY IN HOMO HABILIS as a measure of variability. These broad What has been shown is that brain size and limits also contradict Stringer’s (1986)use of its variability do not seem to be useful for a CV of 10 to detect multiple species in H. distinguishing whether specimens from more than one species have been included in habilis. How reliable is the CV estimate for the H. H. habilis. Claims for multiple species in H. habilis sample (N = 9) itself? The 95% con- habilis have been made on the basis of other fidence limits for the CV of 12.7 for H. habilis characters and approaches (see Lieberman are 5.1-20.3 (Fig. 8B). Therefore, relative et al., 1988; Stringer, 1986). However, such variability in H. habilis cannot be statisti- lines of evidence are not unequivocal (see cally distinguished from comparative sam- Miller, 1990) and are in need of additional ples having a CV of 6, 10, 15, or 20. Indeed, consideration. Future fossil discoveries and the broad confidence limits for H. habilis analyses should clarify the situation. For encompass the entire range of CVs for large now, it is clear that the evidence from endosamples of extant hominoids (Fig. 8B). cranial volume variability does not support Clearly, even if a CV of 10 was an appropri- the argument for multiple species in H. haate upper limit for a single species, small bilis and that it may be premature to reject sample size would now allow statistically the hypothesis that H. habilis re resents a meaningful distinctions to be made between single, sexually dimorphic, a n 8 perhaps this threshold CV of 10 and the H. habilis CV even polytypic species. of 12.7. ACKNOWLEDGMENTS CONCLUSIONS A more complete, statistically rigorous examination of the coefficient of variation for endocranial volume does not support the hypothesis of multiple species in H. habilis. Published data for extant hominoids indicates that there is no empirical basis using a CV of 10 as a standard for detecting mixedspecies samples of fossil hominoids. The same data also indicate that the CV of 12.7 for H. habilis is within the range of CVs for single species of modern hominoids. Furthermore, interpretation of CVs for fossil samples may be confounded by sources of variability such as geographic variation, temporal variation, sample choice, sex ratio, and measurement technique. The difficulty in controlling for these factors suggests that CVs for fossil samples may not be comparable to, and may often be higher than, those for extant samples. Additionally, the broad 95% confidence limits (5.1-20.3) calculated for the small H. habilis sample indicate that its CV of 12.7 is not statistically significantly different from CVs of extant hominoid taxa. Neither is it statistically different from a CV of 10, the supposed threshold for detecting multiple species. The broad confidence limits also mean that the CV of 12.7 for H. habilis is insufficiently reliable for biolo ically meaningful interpretation. These Endings suggest that there is no information from the CV for endocranial volume regarding species number in H. habilis. This does not mean that it has been proved that H. habilis represents a single species. My sincerest thanks to Gene H. Albrecht, who provided invaluable guidance on all aspects of this paper. I also thank Phillip Tobias, Chris Strin er, Don Johanson, Bernard Wood, Bruce elvin, Eric Scott, and two anonymous reviewers for providing insightful comments on the manuscript. A special thanks is given to Donna Yankovich, Eleanor Bates, Doroth Libby, and the late Stewart Shermis for t eir support and encouragement in the early stages of this work. This research was funded by a National Science Foundation Graduate Fellowship and was first presented in preliminary form at the 58th Annual Meeting of the American Association of Physical “Anthropologists (Miller, 1989). (5 i REFERENCES Andrews PJ (1978)A revision of the Miocene hominoidea of East Africa. Bull. Br. Mus. Nat. Hist. Geol. Series 30:85a-224a. Benington RC (1912) A study of the negro skull with special reference to the Congo and Gaboon crania. Biometrika 8:292-337. Bode M (1912) L’homme fossile de la Chapelle-AuxSaints. Ann. Paleont. 7:65-208. Brash JC, and Young M (1935) The Anglo-Saxon skulls from Bidford-on-Avon, Warwickshire, and Bunvell, Cambridgeshire, with a comparison of their principle characters and those of the Anglo-Saxon skulls in London museums. I. The Bidford-on-Avon skulls. Biometrika 27:373-87. Chamberlain AT, and Wood BA (1987) Early hominid phylogeny. J. Hum. Evol. 26:119-133. Conroy GC, Vannier MW, and Tobias PV (1990) Endocranial features of Awtrulopithecus ufricunus revealed by 2- and 3-D computed tomography. Science 247:838-841. Dart RA (1962) The Makapansgat pink breccia australopithecine skull. Am. J. Phys. Anthropol. 20:119-26. 396 J.A. MILLER Day MH (1986)Guide to FossilMan. Chicago: University of Chicago Press. Day MH, Leakey REF, Walker AC, and Wood BA (1975) New hominids from East Rudolph, Kenya I. Am. J . Phys. Anthropol. 42:461-476. Duckworth WLH (1896)Notes on a collection of crania of Esquimaux, exhibited by Prof. A. Macalister, F.R.S. J . Anthropol. Inst. 25:72-74. Fawcett CD, and Lee A (1902) A second study of the variation and correlation of the human skull, with special reference to the Naqada crania. Biometrika 1:408-467. Feibel CS, Brown FH, and McDougall I (1989) Stratigraphic context of fossil hominids from the Omo group deposits: Northern Turkana Basin, Kenya and Ethiopia. Am. J. Phys. Anthropol. 78595-622. Garson J G (1884) On the cranial characters of the natives of Timor-Laut. J . Anthropol. Inst. 8:38M01. Goodman CN, and Morant GM (1940) The human remains of the Iron age and other periods from Maiden Castle, Dorset. Biometrika 31:295-312. Groves CP (1989) A Theory of Human and Primate Evolution. New York: Oxford University Press. Heim J-L (1974) L'encephale neandertalien de l'homme de La Ferrassie. Anthropologie 74.527-572. Holloway RL (1972) Australopithecine endocasts, brain evolution, in the Hominoidea, and a model of hominid evolution. In RTuttle (ed.): The Functional and Evolutionary Biology of Primates. Chicago: Aldine Atherton, pp. 185-203. Holloway RL (1973)New endocranial values for the East African early hominids. Nature 243:97-99. Holioway RL (1978)Problems of brain endocast interpretation and African hominid evolution. In CJ Jolly (ed.1: Early Hominids of Africa. New York: St. Martin's Press, pp. 379401. Holloway RL (1980a)The O.H. 7 (Olduvai Gorge, Tanzania) hominid partial brain endocast revisited. Am. J. Phys. Anthropol. 53:267-274. Holloway RL (1980b) Indonesian ''Solo; (Ngandong) endocranial reconstructions: Some preliminary observations and comparisons with neandertal and Homo erectus groups, Am. J . Phys. Anthropol. 53:285-295. Holloway RL (1981a) The Indonesian Homo erectus brain endocasts revisited. Am, J. Phys. Anthropol. 55:503-521. Holloway RL (1981b) Volumetric and assymetry determinations on recent hominid endocasts: Spy I and 11, Djebel Ihroud I, and the Sale Homo erectus specimens, with some notes on neandertal brain size. Am. J. Phys. Anthropol. 55:385-393. Holloway RL (1983)Human brain evolution:A search for units, models and synthesis. Can. J. Anthropol. 3:215230. Holloway RL (1988) Brain. In I Tattersal, E Delson, J Van Cowering (eds.): Encyclopedla of Human Evolution and Prehistory. New York: Garland Publishing, pp. 98-105. Hooke BGE (1926) A third study of the English skull with special reference to the Farringdon Street crania. Biometrika 18:l-55. Howell FC (1978) Hominidae. In VJ Maglio and HBS Cooke (eds.): Evolution of African Mammals. Cambridge: Harvard University Press, pp. 154-258. Johanson DC, Masao FT,Eck GG, White TD, Walter RC, Kimbel WH, Asfaw B, Manega P, Ndessokia P, and Suwa G (1987) New partial skeleton of Homo habilis from Olduvai Gorge, Tanzania. Nature 327:205-209. Johanson DC, and Shreeve J (1989) Lucy's Child: The Discovery of a Human Ancestor. New York: William Morrow and Company. Kitson E (1931) A study of the negro skull with special reference to the crania from Kenya colony. Biometrika 23:271-314. Kochetkova VI (1978) Paleoneurology. New York: John Wiley & Sons. Lande R (1977) On comparing coefficients of variation. Syst. Zool. 26:214-217. Leakey REF, and Wa1ker.A (1980) On the status of Australoptthecus afurensrs. Science 207: 1102-1 105. Lieberman DE, Pilbeam DR, and 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. Mayr E (1969) Principles of Systematic Zoology. New York: McGraw-Hill. MacDonell WR (1904) A study of the variation and correlation of the human skull, with special reference to English crania. Biometrika 2:191-244. MacDonell WR (1906) A second study of the English skull, with special reference to Moorfields crania. Biometrika 5:86-104. McCown TD, and Keith A (1939)The Stone Age of Monte Carmel, Vol. 2. The fossil human remains from the Levalloiso-Mousterain. Oxford: The Clarendon Press. Miller JA (1989)Homo habilis and endocranial volumes: The use and abuse of coefficients of variation. Am. J. Phys. Anthropol. 78:273. Miller JA (1980) Homo habilis: The sexual dimorphism hypothesis reconsidered. Am. J . Phys. Anthropol. 81:269. Morant GM (1924) A study of certain oriental series of crania including the Nepalese and Tibetan series in the British Museum (Natural History). Biometrika 16:1-105. Morant GM (1925) A study of Egyptian craniology from prehistoric to Roman times. Biometrika 17:l-52. Morant GM, and Hoadley MF (1931) A study of the recently excavated Spitalfields crania. Biometrika 23:191-248. Pope GG (1988) Sambungmachan. In I Tattersall, E Delson, and J Van Cowering (eds.): Encyclopedia of Human Evolution and Prehistory. New York: Garland Publishing, p. 503. Protsch von Zieten R, Gunkel F, and Welz B (1987) Cranial capacity estimations of the Frankfurt Pan troglodytes uerus collection. Hum. Evol. 2:365-372. Randall FE (1943) The skeletal and dental development and variability of the gorilla. Hum. Biol. 15.236-337. Reid RW, and Morant GM (1928)A study of the Scottish Short Cist crania. Biometrika 20:379-388. Schepers GWH (1946)The endocranial casts ofthe South African ape-men. Transvaal Mus. Mem., No. 2, Part 11. Schepers GWH (1950) The brain casts of the recently discovered Plesianthropus skulls. Transvaal Mus. Mem., No. 4., Part 11. Schmalhausen I1 (1935)Determination ofbasic concepts and methods of investigations of growth (In Russian). Sb. Rost zhivotnykh, pp. 8-60. Sergi S (1948)The palaeanthropi in Italy: The fossil men of Saccopastore and Circeo. Man 48:61-79. Simpson GG, Roe A, and Lewontin RC (1960) Quantitative Zoology. New York: Harcourt, Brace and Company. Sokal RR, and Braumann CA (1980) Sjgnlflcance tests for coefficients of variation and variability profiles. Syst. Zool. 29:50-66. 397 BRAIN SIZE VARIABILITY IN HOMO HABILIS Sokal RR, and Rohlf FJ (1981) Biometry. San Francisco: W.H. Freeman and Company. Stewart TD (1977) The Neandertal skeletal remains from Shanidar cave, Iraq: A summary of the findings to date. Proc. Am. Phil. SOC. 121:121-165. Stoessiger BN, and Morant GM (1932) A study of the crania in the vaulted ambulatory of Saint Leonard's church, Hythe. Biometrika 24: 135-202. Stringer CB (1986) The credibility of Homo habilis. In B Wood, L Martin, and P Andrews (eds.):Major Topics in Primate and Human Evolution. Cambridge: Cambridge University Press, pp 266-294. Suzuki H, and Takai F (1970) The Amud Man and his Cave Site. Tokyo: Keigaku Publishing Co. Tildesley ML (1921) A first study of the Burmese skull. Biometrika 13:17&262. Tobias PV (1964) The Olduvai bed I hominine with special reference to its cranial capacity. Nature 202: 3-4. Tobias PV (1971) The Brain in Hominid Evolution. New York: Columbia University Press. Tobias PV (1983)Recent advances in the evolution of the hominids with especial reference to brain and speech. Pontificiae Acad. Scient. Scripta Varia 50:85-140. Tobias PV (1987)The brain of Homo habitis: A new level of organization in cerebral evolution. J. Hum. Evol. 16:741-761. Tobias PV, and von Koenigswald GHR (1964)A comparison between the Olduvai hominines and those of Java and some implications for hominid phylogeny. Nature 204:515-518. Vaisnys JR, Lieberman D, and Pilbeam D (1984) An alternative method of estimating the cranial capacity of Olduvai Hominid 7. Am. J. Phys. Anthropol. 65:7181. Vlcek E (1955) The fossil man of Ganovce, Czechoslovakia. J . R. Anthropol. Inst. 85:163-171. von Bonin G (1931) A contribution to the craniology of the Easter Islanders. Biometrika 23:249-270. von Bonin G (1936)On the craniology of Oceania. Crania from New Britain. Biometrika 28:123-148. vonKoenigswald GHR(1962)TheEvolution ofMan. Ann Arbor: University of Michigan Press. Walker A (1981) The Koobi Fora hominids and their bearing on the origins of the genus Homo. In BA Sigmon and JS Cybulski (eds.):Homo erectus Papers in Honor of Davidson Black. Toronto: University of Toronto press, pp. 193-215. Weidenreich F (1943) The skull of Sinanthropus pekinensis: A comparative study on a primitive hominid skull. Palaeontol. Sin. New Ser. D., No. 10. Wolpoff MN (1981) Cranial capacity estimates for Olduvai Hominid 7. Am. J. Phys. Anthropol. 56:297-304. Woo J K (1966)The skull of Lantian man. Curr. Anthropol. 7:83-86. Woo TL (1930) A study of seventy-one ninth dynasty Egyptian skulls from Sedment. Biometrika 22:65-93. Wood BA (1985) Early Homo in Kenya, and its systematic relationships. In E Delson (ed.): Ancestors: The Hard Evidence. New York: Alan R. Liss, Inc., pp. 206-214. Wu R, and Dong X (1982) Preliminary study of Homo erectus remains from Hexian, Anhui. Acta Anthropol. Sin. 1:2-13. Wunderly J (1939) The cranial and other skeletal remains of Tasmanians in collections in the Commonwealth of Australia. Biometrika 30:305-337. Yablokov AV (1974)Variability of Mammals. New Delhi: Amerind Publishing Co. APPENDIX A. Endocranial volumes (ml)for fossil h,ominids Specimen ECV' Region Source S. Africa Holloway (1972) S. Africa Holloway (1972) Schepers (1946) Holloway (1972) Schepers (1950) Holloway (1972) Schepers (1950) Holloway (1972) Schepers (1950) Holloway (1972) Dart (1962) Conroy et al. (1990) Australopithecus africanus 1. Taung 2. STS 60 (Ples. 1) 3. STS 5 (Ples. 5) 4. STS 71 (Ples. 7) 5. STS 19 (Ples. 8) 6. MLD 37/38 Homo habilis4 1. O H 7 7. KNM-ER 1805 8. KNM-ER 1813 9. KNM-ER 3732 Home erectus 1. KNM-ER 3733 2. KNM-WT 15000 440 562.52 428 435 485 480 428 500 436 560 435 480 4253 S. Africa S. Africa S. Africa S. Africa 582 510 (700) E. Africa E. Africa E. Africa E. Africa E. Africa E. Africa E. Africa E. Africa E. Africa Holloway (1978) Holloway (1973) Stringer (1986) Holloway (1973) Holloway (1978) Stringer (1986) Holloway (1978) Holloway (1983) Stringer (1986) 848 900 E. Africa E. Africa Holloway (1983) Holloway (1988) 687 650 6675 590 752 (750) (continued) 398 J.A. MILLER APPENDIX A. Endocranial uolumes (ml) for fossil hominids (continued) Saecimen 3. 4. 5. 6. KNM-ER 3883 OH9 OH 12 Sang2 7. S a n g 4 8. Sang 10 9. Sang12 10. Sang 17 11. Trinil 2 12. SambungI 13. Lantian 2 14. Sale 15. Zhoukoudian I1 16. Zhoukoudian 111 17. Zhoukoudian V 18. Zhoukoudian VI 19. Zhoukoudian X 20. Zhoukoudian XI 21. Zhoukoudian XI1 22. Hexian 23. Solo I 24. Solo V 25. Solo VI 26. Solo IX 27. Solo X 28. Solo XI Neandertals 1. S D V 1 2. s p y 2 3. La Ferrassie 1 4. Neandertal 5. Shanidar 1 6. Monte Circeo 1 7. La Chapelle 8. Saccopastore 1 9. Tabun 1 10. Amud 1 11. Ganovce 12. Gibraltar 13. La Quina ECV’ Region Source 804 1,067 727 813 775 908 750 855 975 1,059 915 1,004 1,029 940 850 E. Africa E. Africa E. Africa Java 1,035 780 880 1,030 915 1,140 850 1.225 11015 1,030 1,025 1,172 1,251 1.013 1;135 1,231 1,090 Java China N. Africa China China China China China China China China Java Java Java Java Java Java Holloway (1983) Holloway (1983) Holloway (1983) Holloway (1981a) Tobias (1971) Holloway (1981a) von Koenigswald (1962) Holloway (1981a) von Koenigswald (1962) Holloway (1981a) Tobias (1971) Holloway(l98la) Tobias (1971) Holloway (1981a) Tobias and von Koenigswald (1964) Pope (1988) Woo (1966) Holloway (1983) Weidenreich (1943) Weidenreich (1943) Weidenreich (1943) Weidenreich (1943) Weidenreich (1943) Weidenreich (1943) Weidenreich (1943) Wu and Dong (1982) Holloway (1980b) Holloway (1980b) Holloway (1980b) Weidenreich (1943) Holloway (1980b) Holloway (1980b) 1.553 11305 1,681 1,408 1,600 1,550 1,626 1.200 11271 1,740 1,320 1,296 1.367 Europe Europe Europe Europe Middle East Europe Europe Europe Middle East Middle East Europe Europe Euroue Holloway (1981b) Holloway (1981b) Heim (1974) Bode (1912) Stewart (1977) Sergi (1948) Boule (1912) Sergi (1948) McCown and Keith (1939) Suzuki and Takai (1970) Vlcek (1955) Boule (1912) B o d e (1912) Java Java Java Java Java ’Numbers in parentheses are Stringer’s (1986) estimates only and are not based on actual measurements. 2This value is the midpoint of the range of ECV estimatesfor this specimen as given by Holloway (1972). For a historical summary of the various ECV estimates for Taung, see Tobias (1971). jThis value is not used in Figure 7B. “Stringer’s(1986) sample of nine H. habdis specimens. The CV for this sample is 12.7. SStringer(1986) averaged the lowest (633 ml) and highest (700 ml) values mentioned by Tobias (1971)and Holloway (1978),respectively.