Discussion and Debate*

Transcription

1 Discussion and Debate* TELANTHROPUS AND THE SlNCLE SPECIES HYPOTHESIS: A REEXAMINATION VICKI J. GUTGESELL University of Wisconsin Wolpoff has rejected Telanthropus as a refutation of the single species hypothesis on the basis of the stratigraphy at Swartkrans, the morphological characters of the Telanthropus specimens, and a statistical test. The stratigraphic arid rnorphological evidence has been reexamined, and MI length and breadth dimensions of Telanthropus and Paranthropus have been compared statistically. These results are inconsistent with Wolpofs conclusions and the single species hypothesis of human evolution. Accepted for publication 14 November In this journal, M. H. Wolpoff (AA 70: ) has dealt with Telanthropus as a possible refutation of the single species hypothesis of human evolution. He presented an interpretation of culture as the primary hominid adaptive mechanism and its implications, discussed the stratigraphy at Swartkrans and the morphology of Telanthropus, made a statistical comparison of M, in Paranthropus, Telanthropus, Homo erectus, and Honio sapiens, and thereby concluded that not one piece of evidence stands to distinguish it [ Telanthropus ] validly from the other australopithecines (p. 490). This paper reexamines the stratigraphic and morphological evidence, evaluates the statistical procedures and conclusions, and then considers the basic assumptions underlying this interpretation of hominid adaptation and evolution. The terminology used follows that of Robinson. The Stratigraphic Evidence at Swartkrans Wolpoff believes that the stratigraphic evidence at Swartkrans is somewhat ambiguous (p. 483), that there is a lack of temporal ordering in the deposit, and that the evidence does not support Robinson s con- * Photographs in this section courtesy of Alan MaM. Photography by Dorothy Mann, with the permission of C. K. Brain and the financial assistance of the Wenner- Gren Foundation. 565 tention of coeval hominid genera. If one examines the works of Brain, Broom and Robinson, and Robinson, one can assess the reli- ability of these conclusions. Brain (1958) has shown that the Swartkrans cavern system may be divided into an inner, an outer, and a lower cave. The lower cave is not fossiliferous and will not concern us here. The inner cave was incornpletely filled with calcified brown breccia, is well stratified, and contains few fossils. Although a Paranthropus mandible (SK 23) was found near the base of this brown breccia far back in the inner cave, this fact does not affect the interpretation of the remains in the outer cave deposit. A well-marked, almost vertical unconformity separates the inner and outer caves. The outer cave was completely filled with highly calcified pink breccia, which showed only slight traces of stratification, and which contained almost all of the hominid remains. These deposits consolidate as they accumulate; thus younger material cannot be buried into older levels (Brain 1958; Robinson 1967). Mineralogically the upper levels of the pink breccia are practically identical with the lower levels of the brown, and the faunae of the two breccia are indistinguishable. However, the outer cave pink breccia is older than the brown breccia of the inner cave (Brain 1958). At the outer edge of the outer cave deposit was a 4 ft by 4 ft by 1 ft pocket of consolidated chocolate-brown breccia (Broom and Robinson 1952). On the basis of grading analyses, sand grain angularity, and the chert-quartz ratio, Brain ( 1958) concluded that the material in this pocket was mineralogically identical to the outer cave pink breccia, except that its carbonate content was lower. Regions of breccia that have been partially decalcified by leaching are fairly common in the Transvaal cave breccias. In this brown pocket (in the outer cave pink breccia) Robinson found a mandible containing all left molars and M, and M, on the right, an isolated P,, and the proximal end of a radius. These remains were considered to belong to one individual and were

2 566 A merican A nthropologist [72, INS. 2 I 2 3 I FIGURE 1. (a) SK 45 (Telanthropus 11), (b) SK 15 (Telanthropus I), and (c) SK 80 (Telanthropus 111). described by Broom and Robinson as specimens have been found throughout the Telunthropus cupensis (and are referred to vertical height of the deposit; the manas Telanthropus I) (Robinson 1953). In dibular fragment Telanthropus I1 and the the outer cave pink breccia, Parunthropus maxillary fragment Telanthropus I11 (see

3 Fig. 1) were found intermingled with, and within inches of, Paranthropus specimens about one-half to two-thirds of the way down from the surface. The stratigraphic evidence at Swartkrans appears to be explicit. Telanthropus remains in the brown pocket in the pink breccia, which is mineralogically identical to the pink breccia itself (except in carbonate content), and the direct association of Telanthropus and Paranthropus material at the same levels in the uniform, highly consolidated pink breccia deposit establishes conclusively the coevality of these two forms at Swartkrans. Wolpoff has apparently confused the brown pocket of the outer cave with the brown breccia deposit of the inner cave, and has consequently misinterpreted the temporal relationships between the Paranthropus and Telanthropus remains. With regard to the manner of deposition, Wolpoff states: There is no guaranty of temporal uniformity along horizontal layers. Any horizontal area may contain deposits from the entire span of deposition. An examination of Figure 1 shows how this type of unstratified and mixed deposition could have occurred. An object falling into the cave could lodge at any point along the slope. As a result, the deposit of pinkish breccia could have as easily filled from the sides to the middle as from the bottom to the top [1968:482]. Wolpoff s Figure 1, supposedly adapted from Brain 1958, Fig. 70, is so different from Brain s original drawings that one s interpretation of the manner of cave deposition is altered significantly; it appears to be unrelated to Brain s Fig. 70 and at best is a misrepresentation of Fig. 71. From Brain s Fig. 71 (and especially Fig. 5 in Brain 1967) one can see that material falling into the cave would have fallen to the bottom and formed a talus slope. Wolpoff does not take into account the fact that the deposit consolidated as it accumulated; that later material could not be incorporated into earlier levels. Thus, Paranthropus and Telanthropus material found only inches apart must have fallen into the cave and become consolidated into the breccia during the same time period. Because Paranthropus remains were also found both above and below the Telanthropus fragments, Paranthropus was living in the Swartkrans re- Discussion and Debate 567 gion both before and after the period in which the two forms lived there simultaneously. Furthermore, it appears that Wolpoff is confused about geological uniformity. He states: The argument that (1) the Swartkrans deposits are uniform, and thus (2) the different hominids mixed within them are sympatric, appears circular. If, on the one hand, the hominids within the deposit are assumed sympatric, it is then possible to say that with respect to the fauna, the deposits are uniform. On the other hand, one cannot take the claim of uniform deposits based on this assumption and use this claim to support the conclusion that the hominids found in the uniform deposit are therefore coeval [ 1968 :48243]. Wolpoff implies that someone (apparently Robinson) has assumed that the hominids are sympatric, and has therefore concluded that the deposit is uniform, and has further concluded that the remains in this uniform deposit are therefore coeval. Robinson never assumed sympatry. Since nonconformities, which would indicate time gaps in the deposition, are lacking in the pink breccia at Swartkrans, Brain and Robinson determined on the basis of the geological evidence that continuous deposition had occurred and that the outer cave deposit was uniform (a concept unrelated to the number of taxa involved). They then concluded that the remains within the deposit had to be sympatric. The argument that (1) the Swartkrans outer cave deposit is uniform and thus (2) the different hominids mixed within it are sympatric is logical and legitimate and not circular. Wolpoff has stated that if there are truly two hominid genera within the deposit, then the deposits are not uniform... [1968:4821. If.. the hominids within the deposit are assumed sympatric, it is then possible to say that with respect to the fauna, the deposits are uniform [1968:483]. In other words, he is saying that the existence of two sympatric hominid genera in a deposit is an impossibility-if the hominids belong to two genera, they cannot be sympatric; and if the hominids are sympatric, they cannot belong to two genera. Although Wolpoff purports to examine the Telanthropus

4 568 American A nthropologist [72, material in order to determine whether it can validly refute the single species hypothesis, he has prejudged the situation and reached his conclusion on the basis of an unproven assumption even before looking at the morphological and statistical differences between Paranthropus and Telanthropus. From the above discussion one can see that the stratigraphic evidence at Swartkrans demonstrates unequivocally the sympatric coexistence of hominid forms, attributed to Paranthropirs and Telanthropus. One must now examine the morphological and statistical differences between them in order to determine if they belong to the samc taxon. Nondental Morphology of Telanthropus The material attributed to Telanthropus consists of a mandible, a mandibular fragment, a maxillary fragment, and the proximal end of a radius (which Wolpoff refers to as the distal extremity of a radius on pp. 483, 484). The radius has not been described in detail since its features are hominid and can be matched among modern human specimens (Robinson 1953). While detailed discussions of the Telanthropus nondental cranial and mandibular morphology are contained in the literature (Broom and Robinson 1950, 1952; Robinson 1953), Wolpoff passes over this evidence lightly and does not consider its import. Although Telanthropus has some characters in common with both Aiistralopithecits and Ptrranthropus, it differs from the latter form and resembles the hominines ( euhominids ) in many features (Robinson 1953). Maxillary evidence indicates that the Telanthropus C- root length and maximum mesiodistal and buccolingual dimensions are smaller than those of Puranthropirs, while the root length is only slightly longer than that of H. sapiens. Although Telanthropus and Paranthropus probably had canines of similar crown size (the C- of Telanthropus is not preserved), the C- root of Paranrhroprrs was much larger. In Telanthropus and H. sapiens the anterior region of the palate is vaulted; thus the palate runs nearly parallel to the occlusal plane of the teeth. The Parantliropus (in South Africa) palate forms a sharp angle with the occlusal plane and is very shallow anteriorly (see Fig. 2). The right incisive canal is smaller than 1, I LO 30 mm FIGURE 2. Sagittal section through snout, slightly to one side of midline. Vomer is shaded. A and B-Paranfhropus crasdens (from Swartkrans) ; C- Telanthropus ; D- Honio snpiens (Bush). Redrawn from Robinson 1953: Fig. 6. the left canal in Telanthropus, as in Pithecanthropus and H. sapiens; the incisive fossa is smaller than that in Paranthropirs. There is a ridge from the incisive fossa to the posterior margin of the I1 alveolus, and the surface between these lines is concave. In Paratittiropits the fossa is wider with a thickened border on either side; halfway to the alveolar margin, the border turns sharply away from the midline. The palatal midline posterior to the incisive foramen is slightly depressed in Telanthropus but is raised into a small torus in Puranthropus. The midline from the pyriform aperture to the alveolar margin of Telanthropus and the nasal spine can be seen easily in sideview. The canine eminences are robust and the region between them depressed in Parunthropus; neither midline nor the only equivalent of an anterior nasal spine, which is situated relatively far back in the floor of the external nasal aperture, is visible from the side. The lower margin of the pyriform aperture is clearly defined in Telanthropus but not in Paranthropzrs. About 1 cm sep-- arates the nasal spine and vomer in Telanthropus ; in Paranthropus the vomer inserts against the back of the spine. All ten Paranthropus specimens that have the snout region preserved show remarkable \ \

5 uniformity in the above features, and all contrast sharply with the Telanthropus condition (Robinson 1953). There are no intermediate forms, and in all features in which Telanthropus and Paranthropus differ, Telanthropus closely resembles the hominine condition. In mandibular features Telanthropus and Paranthropus differ in size, proportions, and structure. (It should be noted that the quotation on p. 484 of Wolpoff is from Robinson 1953: and not pp as given.) The corpus and ramus of Telanthropus are relatively low; the bicondylar width is greater than the maximum mandibular length, measured in the sagittal plane. The opposite condition is true of Paranthropus. Although the anteroposterior dimension of the ramus at the height of the tooth row is only slightly greater in Paranthropus, the ramus height is up to two times greater. This high ramus indicates that the glenoid fossa and base of the skull were relatively high above the occlusal plane of the teeth. In Telanthropus, however, the base of the skull was probably much lower and therefore allowed for greater endocranial volume (Robinson 1953:498). At MI the minimum internal distance between halves of the mandible is about two times the corpus width in Telanthropus, while in Paranthropus that distance is less than the labiolingual diameter of the corpus. The thick symphysis and bodies of Paranthropus and Australopithecus result in a V-shaped internal mandibular contour, reflecting an early stage in human evolution compared to the thinner bodies and U-shaped contour of Telanthropus and H. sapiens. Although this feature cannot be used as a diagnostic character for sorting taxa, it does serve to differentiate the Telanthropus I mandible from the sufficiently complete Paranthropus mandibles at Swartkrans. The front of the Telanthropus symphysis most closely resembles that of the Heidelberg mandible and appears to be an intermediate stage of evolution between Paranthropus and H. sapiens (Broom and Robinson 1952). The mandibular foramenmylohyoid groove complex of Telanthropus is more nearly typical of modern hominines than is that of Paranthropus. Paranthropus has a much larger retromolar fossa; the cristae endocoronoidea and en- Discussion and Debate 569 docondyloidea, and the planum triangularis and genioglossal fossa are more strongly defined than in Telanthropus. These cristae in Paranthropus indicate heavier use of the masticatory muscles than in Telanthropus and H. sapiens. In Paranthropus the endocoronoid buttress runs anterior to the tip of the coronoid; the coronoid curves backward and is flattened on the top for attachment of the posterior fibers of m. temporalis, which exert a backward force. The anterior fibers of this muscle attach to the anterior edge of the coronoid and here exert an upward force. Thus in Paranthropus there are both upward and more obvious backward components of motion, which would be helpful in heavy grinding and chewing. In H. sapiens the endocoronoid buttress runs up to the tip of the coronoid, where m. temporalis attaches and thus exerts an upward force. Although the area of muscle attachment is not preserved in Telanthropus, it was probably similar to that in H. sapiens and unlike that of Paranthropus since the buttressing is like that of modern man. Thus it is clear that Paranthropus and Telanthropus differ considerably in nondental cranial and mandibular morphology. Since the above features are components of the masticatory complex, and since mastication is an integral part of a mammal s basic adaptation, these differences are significant. Wolpoff attempts to discount the maxillary distinctions (which are qualitative rather than quantitative as stated on p. 484) by stating: the distance from the nasal spine to the alveolar point [in Telanthropus 1111 is the same as in the other Swartkrans australopithecines... and is greater than the corresponding distance in euhominids (p. 486). Thus, all of these euhominid distinctions occur on a fragment with a face as large as that of the largest australopithecines, and one of the major trends in human evolu- tion between the australopithecines and their successors is the reducrion of the face! [ 1968:485]. He therefore concludes that these features, on a maxilla far within the australopithecine range of variation with respect to the most evolutionary significant characteristics (p. 485), justify extension of the australopithecine range of variation rather than creation of a new genus. It is noteworthy that

6 570 A merican A n thropologist [72, (1) Wolpoff places more emphasis on one feature that Telanthropus and Paranthropus have in common-nasal spine-alveolar point distance-than on all the morphological differences combined, and (2) although increased cranial capacity is considered by Wolpoff as evolutionarily important as face reduction, the probability that Telanthropus had a larger endocranial volume (but was a smaller animal) than Paranthropus is ignored. Furthermore, if one refers to Robinson one finds the following: The distance from the anterior nasal spine to the alveolar point is greater in the latter specimen [ Telanthropus 1111 than is usual in euhominids, but it is only slightly greater than the corresponding distance in the Pithecanthropus ZV specimen and is exceeded by that of Rhodesian man [1953:486; italics mine]. Thus it appears that Wolpoff has taken part of the relevant text out of context to support his argument. On the basis of his reasoning, Rhodesian man would be considered a more primitive australopithecine than Paranthropus. And Pithecanthropus would have to be included in the australopithecine range of variation also. In attempting to discount the mandibular differences, Wolpoff quotes several statements from Dart, dealing mainly with Australopithecus, Telanthropus, and modern man. Most of these statements are not particularly relevant and do not invalidate the distinctions between Telanthropus and Paranthropus, based on size, proportion, and morphological structure. Dental Morphology Wolpoff ignores all morphological differences between the teeth of Telanthropus and Paranthropus discussed in Broom and Robinson 1952 and Robinson The Telanthropus C- and Pa roots fit well in size with those of H. sapiens, while those of Paranthropus are considerably larger. In Telanthropus and other hominines the P3 crown is narrower at the cervical line than at the occlusal surface, unlike the condition in Paranthropus; the root system tapers downward, whereas in Paranthropus the root is wider at the apex. The buccal faces of the molar crowns differ in the two forms with regard to grooves, pits, and tubercles. Wear- ing through of the cusps also differs; Telanthropus closely approaches the condition in modern man. The Paranthropus M, almost invariably has six cusps. In Telanthropus there is no sixth cusp on M,, a moderate sixth cusp on M2, and a large one on MZi. All Telanthropus teeth have fissures showing only slight traces of secondary folding of enamel; M, has a simple, smooth surface. Moderate wrinkling of the surface is common in Paranthropus premolars and molars, especially on M, and M,; there is a tendency to complication of the fissure pattern on M, and M3 and to multiplication of cusps in the talonid region. In Telanthropus I1 M, is appreciably reduced; the roots are directed sharply backward and are reduced and fused into a single, small root. All but one of the teeth of Paranthropus have an anterior buccal cingulum, a feature not present in Telanthropus. Since the differences in cranial, mandibular, and dental morphology are so numerous, and since there are no intermediate forms between Telanthropus and Paranthropus, these two forms are unlikely to be members of the same population. However, Broom and Robinson (1952) did statistically test the significance of the differences in MI size. Statistical A nalysis At Swartkrans one is faced with a sorting problem; the fossils from the cave must be separated into groups such that each group represents one, and only one, species. One does this on the basis of morphological characteristics, since this is the only evidence available from fossils. First one separates major categories such as reptiles, birds, and mammals; then within the mammals, primates must be distinguished from rodents, carnivores, artiodactyls, etc. Usually the primate material consists of cercopithecoids and hominids. Of this hominid material at Swartkrans, one wants to know whether all the specimens belong together in one group and therefore represent a single sample or whether they fall into different groups and therefore represent more than one sample. Until all specimens have been assigned to one or more samples, one cannot determine the characteristics of the sample(s) and hence of the local population(s)

7 from which the sample(s) came. If all of the hominid specimens were part of the same local breeding population, and therefore form one group, a single taxon is involved; if not, more than one taxon is involved. At this stage of analysis one is concerned only with the number of samples present. After this has been decided, one can examine the samples in order to determine what particular species or genera are represented. The hominid specimens at Swartkrans sorted into two distinguishable groups-one identified as Purunthropus and the other described as Telanthropus. Two local samples from two different localities may be different and yet belong to the same species or genus because they belong to different local divisions of the gene pool. But where all specimens are from the same age period of a single site, they either belong to the same local division of a gene pool, or else they belong to two different, genetically isolated, sympatric gene pools. Only by confining comparisons to the hominid material from Swartkrans, where the forms were contemporaneous and the possibility of interbreeding existed, can one determine whether the Telanthropus specimens could have been members of the same local population as the Paranthropus specimens. These comparisons are therefore the most biologically meaningful. After this problem has been settled, one can make general comparisons with hominid samples from other sites. One must keep in mind that the Puranthropus sample at Swartkrans comprises about three-quarters of the known Puranthropus material on the African continent. It is the single largest Purunthropus sample known and is unlikely to represent a small, insignificant, aberrant deme. An estimated population range of variation based on the observed range of variation at Swartkrans should give a reasonable indication of the local Puranthropus population. At the time Broom and Robinson (1952) did their original statistical test on the Swartkrans material, 13 specimens of MI had been described as Paranthropus and one attributed to Telanthropus. Broom and Robinson began with the assumption that Telanthropus and Puranthropus at Swartkrans belonged to the same population. They calculated the mean and standard Discussion and Debate 571 deviation of M, length for the entire sample, and then calculated the difference between the Telanthropus value and the sample mean in standard deviation units; the difference was 2.8 sd. They repeated the procedure without including the Telanthropus MI value in the calculations; this gave a highly significant difference of 5.4 sd for length (significant at the level of p <.001; Simpson et al., 1960) and 3.8 sd for MI breadth (p <.01). Since the Telanthropus M, values fell outside the estimated population range (mean 22.5 sd), it was concluded that Telanthropus was not a member of the local Parunthropus population. Although Broom and Robinson s test results were highly significant, Wolpoff believes that the Swartkrans sample is not the best unit of comparison: a fossil species represents a more consistently-related and biologically-meaningful unit than do the specimens from a site (1968:487). (It is difficult to understand how the Swartkrans Puranthropus specimens could be related in a more consistent and biologically meaningful manner with forms in Java and East Africa than with each other, with whom they were sharing a common local gene pool.) Therefore, he continues: let us compare the Telanthropus MI values with the range of variations for all specimens that Robinson includes in Paranthropus (1968:488). He therefore includes Purunthropus specimens from Kromdraai (N = 3: two in Robinson 1956 and one in Tobias 1966), Lake Natron (N = 2, Tobias 1966) Java (N = 1, Tobias and von Koenigswald 1964), and Swartkrans (N = 18, Robinson 1956) in his comparison and statistically tests a sample of twenty specimens-eighteen Parunthropus and two Telanthropus. Since MI length and breadth in Telanthropus fall 2.17 and 1.62 sd respectively from the sample mean (when both Telanthropus values are included in the calculations for sample mean and standard deviation), compared to the 2.24 sd rejection criterion of Chauvenet for this size sample, Wolpoff concludes that Telanthropus represents an extension of the estimated range of australopithecine variability. Besides not confining his comparisons to the hominid material at Swartkrans, Wolpoffs statistical approach is inappropriate for taxonomic purposes. In experiments in the physical sciences one often obtains a

8 572 A merican A n thropologist [72, measurement that difers from the others by a relatively large amount (Parratt 1961). The experimenter must decide whether or not to reject the bad measurement, and if so, on what objective criteria to base his decision. Parratt (1961) points out that all objective criteria are arbitrary. There is no universal agreement among statisticians on how to cope with these outlying measurements (Johnson, personal communication). Wolpoff has chosen Chauvenet s rejection criterion-the most widely accepted (Parratt 1961 )-which includes the bad measurement with the others in calculating the sample mean and standard deviation. If the difference between the bad measurement and the sample mean falls outside a given range (depending on the sample size), the bad measurement is rejected. Parratt (1961) states that one may also apply Chauvenet s rejection technique to a second bad measurement. However, one cannot apply Chauvenet s statistical technique to taxonomic problems. By treating the Telanthropus values as bad measurements and by including the values for both specimens simultaneously in his calculations, Wolpoff is diluting his Paranthropus population significantly more than if he included only one Telanthropus value (which is, in effect, what Broom and Robinson did initially). If one follows Wolpoffs procedure and adds in a Telanthropus value for each specimen subsequently found, eventually the comparison will be rendered biologically meaningiess, no matter how significantly different the Paranthropus and Telanthropus populations are. At Swartkrans there are two distinct samples; one wants to know how different they are from each other and therefore whether they could have been drawn from the same local population. One cannot consider the Telanthropus values to be bad measurements in the Paranthropus sample, for they form a distinct sample and are as valid as the Paranthropus values. The usual technique (and the most powerful) for analyzing data from two independent samples is to apply a t test to the means of the two groups (Siege1 1956). One can also use the t test to compare one specimen to a population (Simpson et al. 1960). If one wishes to avoid the assumptions underlying the t test (e.g., the populations from which the observations are drawn are normally distributed and have the same variance), one can use the Mann- Whitney U test-the most useful nonparametric alternative to the parametric t testto determine whether the two groups have been drawn from the same population (Siegel 1956). Although the samples at Swartkrans are small, it is not unreasonable to accept the assumptions of the t test and to apply it to the data.2 In addition to the above objections, one can make further criticisms of Wolpoffs procedures and conclusions: (1) In his summary of Broom and Robinson s statistical results, he reports (p. 487) that Telanthropus M, length differs from the sample mean by 2.8 sd, and M, breadth by 3.8 sd. He does not mention that the 2.8 sd difference was obtained by including the Telanthropus value in the sample mean calculations, whereas the 3.8 sd difference was the result when Telanthropus was not included in the mean calculations. More important, he ignores the highly significant difference of 5.4 sd. (2) The M, of Telanthropus I1 is badly worn and the measurements arc only estimates. Although Wolpoff includes the values for both Telanthropus specimens in his calculations, he bases his hypothesis on the differences between Telanthropus I and the sample mean. The differences between Telanthropus 11 and the sample mean are 2.25 and 1.96 sd for M, length and breadth (rejection criterion = 2.24 sd). If he had based his hypothesis on these results, one wonders what he would have concluded. Since the values for Telanthropus I1 are estimates, it would be more proper to distinguish the Telanthropus mean values from the sample mean; these results are 2.21 and 1.79 sd for length and breadth respectively. (3) Wolpoff explicitly states that Telanthropus must be compared to all specimens that Robinson includes in Paranthropus, but omits five of the specimens from Swartkrans and one from Kromdraai in his calcula- tion~.~ The Telanthropus mean values for M, length and breadth differ from the sample mean by 2.41 and 2.01 sd respectively if all twenty-four Paranthropus and the two Telanthropus values are included in the calculations (N = 26, rejection criterion sd). If Wolpoff had followed his own

9 instructions, one wonders what conclusions he would have drawn, since the differences straddle the rejection criterion. (4) On the basis of one statistical comparison, Wolpoff concluded that the Swartkrans population was homogeneous. Simpson points out that although samples can frequently prove beyond reasonable doubt that the population is heterogeneous.... they can never strictly prove that it is homogeneous (1961 : 102). With living organisms, numbers cannot and do not tell the whole story. Although one s conclusions may appear logical on the basis of numbers, it is abhorrent biologically (Simpson : 174) to reject a hypothesis on minute (e.g., 0.07 sd) statistical differences. (5) One can calculate the mean and standard deviation for M, length and breadth in Paranthropus specimens alone and then compute the difference between the Telanthropus values and the Paranthropus mean values. Broom and Robinson (1952) made this comparison. Although Wolpoff refers three times to the paragraph in Broom and Robinson (1952:117) in which this procedure is described he neither makes this kind of comparison nor refers to it. (It should be noted that the reference to Broom and Robinson 1952:118 on p. 489 in Wolpoff s paper should be 1952:117.) The Telanthropus mean MI length and breadth differ from the respective mean values for the twenty-four Pnranthropus specimens by 3.55 and 2.60 sd (rejection criterion 2.30 sd for N = 24) in this comparison. These results are significant at the level of p <.01 and.02 respectively. Therefore, even when compared to an estimated range of variation based on all the Paranthropus samples com bined, Telanthropus is significantly different from Paranthropus. Compared with the eighteen Paranthropus specimens from Swartkrans, the Telanthropus mean M, length and breadth fall 2.54 and 2.35 sd respectively from the sample mean, if both Telanthropus values are included in the calculations (N = 20, rejection criterion = 2.24 sd), and 5.35 and 4.15 sd from the mean, if the calculations are done without including the Telanthropus values. The latter results are significant at the level of p <.001. Thus the differences are significant even using Wolpoffs method. Discussion and Debate 573 Since these results are in accord with the previous conclusions deduced from the morphology, it is reasonable to conclude that Telanthropus and Paranthropus do not belong to the same species and thercfore represent two coexisting hominid lineages. Wolpoff states that: A more serious objection can be raised to the relevance of Broom and Robinson s statistical test. This concerns the use of MI in distinguishing hominid species [ I968 :487]. From his statistics Wolpoff concludes: Indeed, according to Tables 1 and 2, neither the two Hotno erectus specimens [ Pithecanthropus B and MNK II] nor the Australian aborigine mean and extreme value can be used to differentiate these hominids from Paranthropus on the basis of MI dimensions.... Thus, the taxonomic value of MI dimensions is questionable. We surely do not wish to base our taxonomic determinations on criteria that cannot be used to distinguish Homo sapiens from the larger australopithecines [1968:4901. But Wolpoff does base taxonomic detcrminations on the M, criterion. Broom and Robinson primarily used morphological evidence for their conclusions. One can make criticisms of this procedure also: (1) Wolpoff did not compare either H. erectus or H. sapiens to Paranthropus. Rather, two individual H. erectus specimens and one race of H. sapiens were compared to a mixed population sample consisting of ten percent Telanthropus (=H. erectus) and ninety percent Paranthropus. He thus violates his own dictum that all available specimens should be used for comparison. (2) Of the twenty known specimens of H. erectus, one of the two used for comparison was Pithecanthropus B-one of the largest. Had Wolpoff arbitrarily chosen Sinanthropus 38 (M, length = 9.9 mm, M, breadth = 10.1 mm) for comparison instead, he would have found differences of 5.88 and 4.43 sd between this specimen and the respective means for the population of twenty-four Paranthropus specimens. Consequently, he would have concluded that M, is an excellent criterion for distinguishing hominids.

10 5 74 American Anthropologist [72, It would be better to compare the mean length and breadth for all H. erectus specimens to the corresponding Paranthropus means. The results of this are 2.71 and 2.31 sd respectively (rejection criterion z 2.30 sd). If one compares the mean of means for seven modern races of H. sapiens to Puranthropus, rather than just comparing the aborigines (a very large-toothed form of modern man), the values are 4.12 and 3.61 sd from the mean Paranthropus M, length and breadth. Thus one can differentiate H. sapiens and Paranthropus on the basis of MI dimensions. The taxonomic value of M, for distinguishing hominid species and genera is questionable, however, since thcre is much overlap between the Paranthropus, H. erectus, and H. sapiens values. But this conclusion does not negate the results of Broom and Robinson. Rather, it strengthens their conclusions. They used MI only to show the likelihood of Telanthropus and Paranthropu.y belonging to the same local population, not as a taxonomic criterion used to assign specimens to one taxon or another. Since there is so much overlap among horninids with regard to M, dimensions, the large differences found between Telanthropus and Paranthropus are all the more significant. Hominid Adaptation and Evolution The essential features of Wolpoffs interpretation of hominid adaptation and evolution are that: reduced canines in the earliest hominids indicate that they regularly used defensive weapons; this early employment of defensive tools led to the differentiation of the hominid stock: any bipedal hominid population must have been dependent upon culture for its survival; since culture serves as man s primary means of adaptation and as the niche to which man has morphologically adapted ( 1968 :479), all hominids occupy the same adaptive niche; the competitive exclusion principle would therefore not allow two hom- inid species to coexist for any length of time. Although one can neither prove nor disprove this, or any other theory of human evolution, one can examine the validity of these assumptions in view of the available evidence and in terms of the evolutionary process. According to Wolpoff: The reduced canines found in even the earliest horninids similarly indicate an early replacement of the canine defensive function by the regular employment of weapons [ 1968 :479]. The implications of this st-t ement are twofold: (1) there is a direct correlation between canine size and tool use replacing the canine defensive function; and (2) the earliest hominids regularly used weapons. This argument, however, does not appear to be substantiated by any known evidence. If one assumes that the canines of Paranthropus, which are absolutely within the observed range for modern man and are proportionately (when compared to the great robusticity of the mandible as well as to the postcanine teeth) the smallest hominid canines known, are a consequence of tool use, how does one account for the proportionately and absolutely larger canines in A usfralopithecus, H. erectus, and modern man? If Wolpoff s premise were valid, one would expect the canine of H. sapiens to be extremely small, if not lost entirely. Even if canine reduction is a result of reduced canine defensive function, it does not follow that reduced canine defensive function is a result of regular tool use. There is much evidence for canine reduction and loss in mammalian evolution with no relation to tools or weapons. For instance, swiftness and agility can replace the canine defensive function. Another possible reason for canine reduction is the absence of natural predators. Since Puranthropus was apparently a rather large, robust creature, there is no reason to assume that it was frequently harassed by predators any more so than the modern gorilla. However, a mammal s teeth are most closely associated with its dietary adaptations and specializations. In mammalian evolution there are many examples of canine loss due to dietary adapta-

11 Discussion and Debate 575 tions. There is nothing, therefore, that correlates canine reduction to weapon use. At present there is no archeological evidence to suggest regular use of weapons by any of the earliest hominids. The oldest site in South Africa, the Lower Breccia of Sterkfontein, has yielded about three-quarters of all known A ustralopithecus remains but as yet no tools. At the slightly later Makapansgat Limeworks bone tools have been found associated with Australopithecus. The Middle Breccia of Sterkfontein, which is still later, contains A ustralopithecus remains and artifacts attributed by Mason to the African Acheulean culture (Mason 1961). During this time Puranthropus and Telanthropus were living less than a mile away at Swartkrans, where other more advanced artifacts have been found. At still later Kromdraai, where Paranthropus alone is found, no tools occur in the breccia. As for East Africa, lower Bed I at Olduvai has yielded very primitive early Oldowan artifacts with Australopithecus and Puranthropus. In upper Bed I the tools are more advanced but there are no hominid remains, and in Bed 11, where H. erectus is found with Paranthropus, there are African Acheulean hand axes. The majority of the artifacts at Olduvai consists of choppers, scrapers, and hammer- stones-tools used presumably for crushing long bones, cutting through tissue, and scraping meat from skin and bones (Leakey 1967). The evidence therefore indicates: (1) that some of the early hominids were making bone or very primitive stone tools, which cannot be classified specifically as defensive weapons, and (2) that there are no artifacts that can be specifically attributed to Purunthropus, since where tools are found with this form, a more advanced hominid is also present. No evidence supports Wolpoffs contention that any bipedal hominid population... must have been dependent upon culture for its survival (p. 479). That the early development of defensive tools led to the differentiation of the hominid stock is supposition. Since efficient bipedalism is the primary hominid morphological adaptation, the implication is that development of tools subsequently led to erect posture. However, one cannot assume that culture spontaneously appeared and that the form using it became bipedal. Since Wolpoff points out that availability at all times to use tools is crucial in distinguishing pongids and hominids, it is unlikely that culture could have developed in a form whose hands were not already free. Bipedality could offer distinct advantages to a form living at the edge of a woodland bordering on savanna. Therefore, a more logical explanation is that erect posture was attained first, thus freeing the hands and providing the opportunity for regular tool use. Subsequent development of tool use, however, was not necessarily a consequence of bipedality. Not until selective pressures were such that tools afforded definite survival advantage was there any need to develop culture. Wolpoff states, it is difficult to understand how different hominid species could either have arisen or have been maintained sympatrically (p. 480). No one, to the knowledge of the writer has ever suggested sympatric speciation of hominids, which is very different from the existence of sympatric species. However, that more than one hominid species evolved is consistent with the evidence. Although culture may serve as modern man s primary means of adaptation, it did not necessarily serve as u means of adaptation for Puranthropus two million years ago. The stratigraphic, morphological, and statistical evidence indicates that two different hominid forms lived sympatrically in the Swartkrans area. Parunthropus and H. erectus have also been found together in East Africa and Java. So long as their ecological adaptations remained sufficiently different, two hominid lineages could presumably coexist sympatrically for an indefinite period of time. Conclusions Direct association of Telanthropus and Paranthropus remains at the same levels in a uniform deposit of highly consolidated breccia at Swartkrans demonstrates unequivocally the sympatric coexistence of these two forms. There are numerous quantitative and qualitative differences between Telanthropus and other hominines on the one hand, and Paranthropus on the other, in dental and in nondental cranial and mandibular morphology. Statistical tests show that the Telanthropus mean MI length and breadth

12 576 American Anthropologist [72, values are highly significantly different from the respective Paranthropus values. The evidence therefore supports Telanthropus as a refutation of the single species hypothesis of human evolution. Notes 1 would like to express my sincere appreciation to Drs. L. Freedman, University of Western Australia, R. A. Johnson, University of Wisconsin. and G. P. Rightmire, State University of New York, Binghamton, for their helpful suggestions regarding statistics and J. T. Robinson, University of Wisconsin, for reading this paper in manuscript and for offering valuable suggestions and criticisms. 2 In attempting to solve another taxonomic problem, Wolpoff (1969) has applied Chauvenet s technique to endocranial volume measurements of hominids from East and South Africa. After examining the statistical references cited by Wolpoff in that paper, Pilbeam (1969) concluded that this method was inappropriate for dealing with the problem, since it should only be applied to a sample from one statistical population or universe. Thus samples from different geographical regions and time periods, and distinct samples from the same region and time period, should not be pooled into one large sample. Pilbeam suggested using a f test for statistical comparisons. aa Paranthropus M, from Omo has been reported (Howell 1969) but is not included in this paper since (1) it was not available to Wolpoff when he wrote his paper and (2) the specimen does not affect the relationship between Telanthropus and Paranthropus at Swartkrans. References Cited BRAIN, C. K The Transvaal ape-man bearing cave deposits. Transvaal Museum Memoir The Transvaal Museum s fossil project at Swartkrans. South African Journal of Science 63 : BROOM, R., and J. T. ROBINSON Man contemporaneous with the Swartkrans ape-man. American Journal of Physical Anthropology 8: Swartkrans ape-man. Transvaal Museum Memoir 6. HOWELL, F. CLARK 1969 Remains of hominidae from Pliocene/ Pleistocene formations in the Lower Omo Basin, Ethiopia. Nature 223: LEAKEY, M. D Preliminary survey of the cultural material from Beds I and 11, Olduvai Gorge, Tanzania. In Background to evolution in Africa. Walter W. Bishop and J. Desmond Clark, eds. Chicago 6r London: The University of Chicago Press. pp MASON, R. J The earliest tool-makers in South Africa. South African Journal of Science 57: 13-16, PARRATT, L. G Probability and experimental errors in science. New York & London: John Wiley and Sons. PILBEAM, DAVID 1969 Early hominidae and cranial capacity. Nature 224:386. ROBINSON, J. T Telanthropus and its phylogenetic significance. American Journal of Physical Anthropology 11: The dentition of the australopithecinae. Transvaal Museum Memoir Variation and the taxonomy of the early hominids. In Evolutionary biology. Th. Dobzhansky, M. K. Hecht. and Wm. C. Steere, eds. New York: Appleton-Century-Crofts. pp SIECEL, S Nonparametric statistics for the behavioral sciences. New York & London: McGraw-Hill Book Co. SIMPSON, G. G Principles of animal taxonomy. New York: Columbia University. SIMPSON, G. G., A. ROE, and R. C. LEWONTIN Quantitative zoology. New York: Harcourt-Brace. TOBIAS, P. V The distinctiveness of Homo habilis. Nature 209: TOBIAS, P. V., and G. H. R. VON KOENICSWALD A comparison between the Olduvai hominines and those of Java, and some implications for hominid phylogeny. Nature 204: WOLPOFF, M. H Telanthropus and the single species hypothesis. American Anthropologist 70: Cranial capacity and taxonomy of Olduvai Hominid 7. Nature 223: THE EVIDENCE FOR MULTIPLE HOMINID TAXAT SWARTKRANS MILFORD H. WOLPOFF Case Western Reserve University Gutgeseli s review of the single species hypothesis, arid the dismissal of its proposed refutation at Swartkrans, is cri!ically examined. Most of her claims are not supported by the available evidence at Swart-