Palaeont. afr., 28, 53-59 (1991) ENAMEL THICKNESS IN SOUTH AFRICAN AUSTRALOPITHECINES: NONINV ASIVE EVALUATION BY COMPUTED TOMOGRAPHY by Glenn C. Conroy Department of Anatomy & Neurobiology, Department of Anthropology Washington University Medical School, St. Louis, MO.63110 USA ABSTRACT Until recently, it has not been possible to systematically study enamel thickness in fossil hominids except by physically sectioning the teeth. Because sectioning studies destroy original specimens, sample sizes will always be low. For this reason, anthropologists have had to devise other methods for acquiring these data such as by measuring enamel in naturally fractured teeth or where it is exposed in worn teeth. It is clearly important to develop and apply non-invasive techniques to augment and expand the data base of early hominid enamel thickness. This is a first attempt to provide such data for a sample of South African australopithecines by utilizing high-resolution computed tomography (CT). This study is based on over 130 CTscans taken at I mm slice thickness on a sample of22 originalAustralopithecus africanus andA. robustus lower molars from Sterkfontein, Kromdraai, Makapansgat, Swartkrans and Taung. Mean values of absolute and relative enamel thickness between A. africanus and A. robustus are significantly different, confirming that robust australopithecines have thicker enamel than their gracile counterparts. CT sections were taken in the buccolingual plane through the mesial cusps (protoconid, metaconid). While the mean value of enamel thickness at the buccal cusp (protoconid) is greater in A. robustus than in A. africanus, the difference is not statistically significant. The difference in enamel thickness at the lingual cusp (metaconid) is statistically significant, however. This study represents an important, albeit preliminary, first step in establishing a methodology for the non-invasive evaluation of enamel thickness in fossil hominids by computed tomography. It demonstrates the viability of the technique and the type of problem oriented approach that can be tackled using computed tomography in modem anthropological research. Measurements derived from CT cannot, of course, be expected to have the same degree of precision as those taken directly from sectioned teeth; nevertheless, important insights into the functional morphology of early hominid teeth are still easily decipherable from the CT data. Given that the alternative to CT is the physical destruction of original hominid fossils, the slight loss in mensurational accuracy seems well worth the price. INTRODUCTION "The most useful measurement of the amount of enamel on a tooth would be the volume of the tissue. This could be expressed as a thickness by dividing the volume of enamel by the surface area of the enamel/dentine junction over which the enamel formed; this area is proportional to the number of ameloblasts which formed the enamel" (L. Martin, 1985:260). Until recently, it has not been feasible to meet this goal in paleoanthropological studies. Determination of total enamel volume in fossil hominid teeth would require a complete set of sequential thin sections to be taken of original specimens, a procedure not enthusiastically en­ dorsed by many museum curators. For this reason anthro­ pologists ha ve had to devise other methods for measuring enamel thickness, for example by utilizing naturally fractured teeth or by using teeth in which the enamel was exposed through wear (Beynon and Wood 1986; Gantt 1985; Kay 1981). Due to its destructive nature, studies dependent upon sectioning techniques are only able to muster limited sample sizes for both extant and extinct primates (Molnar and Gantt 1977; Martin 1985; Grine and Martin 1988). For example, dental sections are available for only two Australopithecus africanus teeth (Stw 284, Stw 402), one Paranthropus robustus l tooth (KB 5223), and one P. c.rassidens tooth (SKX 21841) (Grine and Martin 1988). Thus it is important to develop and apply non-invasive, non-destructive imaging techniques to augment and expand the data base of early hominid enamel thickness. This contribution describes the feasability of providing such data through high-resolution computed tomography (CT). Although Robinson (1956) was the first to specifically I Grine and Martin (1988) place all species of "robust" australopithecines in the genus Paranthropus. 54 comment on enamel thickness in australopithecines, it was not until Jolly's (1970) classic paper on the "seed­ eating hypothesis" that attention was focused on enamel thickness as an important functional adaptation of early hominid teeth. This theme was laterreinforced by several other studies that related enamel thickness to occlusal wear patterns (Pilbeam 1972; Simons and Pilbeam 1972). By the end of the 1970's it had become arthropological dogma that australopithecine teeth were characterized by "hyper-thick" enamel even though actual measurements of enamel thickness remained few and far between. The limited number of direct enamel thickness measures available were based solely on naturally broken tooth fragments (Robinson 1956; Gantt 1983). More recently, Beynon and Wood (1986) completed a study of enamel thickness in 47 East African australopithecine and early Homo teeth using naturally broken tooth fragments. They concluded that absolute enamel thicknesses atcusp tips and occlusal surfaces were significantly greater in robust australopithecines compared to earl y Homo even after correcting for overall tooth-size differences. In a more controlled study using fmely polished sections, Grine and Martin (1988) were able to measure enamel thickness on four South African australopithecine molars by carefully sectioning each tooth buccolingually through the mesial two cusps (protoconid and metaconid). They also sectioned two australopithecine specimens from Omo, one buccolingually through the entoconid and hypoconid and the other mesiodistally through the protoconid and hypoconid. Although sample sizes were necessarily small given the destructive nature of such procedures, they concluded that relative enamel thickness values in A. africanus fell within the 99% confidence limits for the means of their combined upper and lower molar sample of modem Homo sapiens, whereas the Paranthropus robustus, P. crassidens, and P. boisei values fell well above those same limits. Conventional radiographic techniques have also been used to assess enamel thickness in australopithecines but their limitations are well recognized because of relatively low contrast and geometric errors due to inherent magni­ fication errors and/or X-ray beam angulation variations (Sperber 1985). Computed tomography does not have such inherent geometric distortions because of the fixed geometry thinly collimated X-ray beam (typically 1- 2mm) and has great potential for non-invasive studies of dental structures (Zonneveld and Wind 1985; Conroy 1987; Conroy and Vannier 1987, 199Ia,b; but see Grine 1991). METHODS This study of enamel thickness is based upon approximately 130 high-resolution CTscans of specimens 1 isted in Table 1. All scans were taken at slice thicknesses of 1 mm. Pixel edge length in the plane of section varied between 0,1 - 0,3 mm. CT sections were taken in the buccolingual plane through the mesial cusps (protoconid and metaconid) in order to be as comparable as possible to the data reported by Martin (1985) and Grine and Martin (1988). Each specimen was properly aligned in the CT scanner by first producing a topogram (digital radiograph) of the specimen from which the desired CT slice could be automatically selected by using the built-in light-pen integrated into the CT console. In selected cases, contiguous sequences of 1 mm thin CT sections were taken through the entire tooth in order to determine: 1) total enamel volume; and 2) total dentine/enamel surface area over which the enamel formed (see below). All scans were taken on a Siemens Somatom DR3 CT Scanner in the Radiology Department, Hillbrow Hospi­ tal, Johannesburg (operating at 125 KVP and 160 to 520 mAs; with a scan file diameter of 552 mm). The Somatom DR3 can scan objects with density values from air (-1000 Hounsfield Units or H.D.) to cortical bone (+ 1000 H.D. typically). By definition, water is 0 R.U. The maximum range of X-ray attenuation measured by the DR3 is ordinarily +3072 R.D., and under softward control, we can extend this to approximately +7000 H.D. Extended bone range window settings were used since the density of the fossils was well beyond + 1 000 Hounsfield units, the upper limit of cortical bone in normal CT viewing. The CT scans (5122 or 2562 matrices) and digital scan projection radiographs used for localization of CT sec­ tions (topograms) were stored on 8" floppy disks (DEC RXOI format, 0,5 byte/disk) and carried back to the Mallinckrodt Institute of Radiology, Washington Uni­ versity Medical School, St. Louis. There they were copied from the floppy disks onto magnetic tape using a Siemens Evaluscope-DR reviewing console and transferred from the tape to a DEC Micro Vax 3600. The images were sent from the Vax to a Macintosh IIfx workstation via a thin­ wire Ethernet network with a DECNET/pACER proto­ col. There they were automatically converted into an image (PICT format) which could be stored, scaled, and displayed in an image analysis program (IMAGE) devel­ oped by Wayne Rasband at the National Institutes of Health (NIH). Original australopithecine specimens from Sterkfon­ tein (N=lO), Swartkrans (n=9), Kromdraai (n=I), Makapansgat (n=I), and Taung (n=l) were examined utilizing the methods described above (Table 1). These specimens represent both gracile (n= 12) and robust (n= 1 0) australopithecines and are currently housed in the Depalt­ ment of Anatomy, University of the Witwatersrand, Johannesburg, and the Transvaal Museum, Pretoria. The specimens from Swartkrans and Kromdraai are referred toA. robustus and those from Sterkfontein, Makapansgat and Taung to A. africanus. Only lower molars exhibiting little, if any, occlusal wear were used in this study. A number of measures of enamel thickness have been defined over the past several years and from these a number of indices of relative enamel thickness have been calculated (see Fig. 1.8 in Grine and Mattin 1988). I have followed Martin (1985) and Grine and Martin (1988) in 55 TABLE 1 (all measurements are in millimeters) Specimen tooth taxon enamel area (c) Taung 1 LMI A. africanus 23,25 MLD2 LM2 A. africanus 45,52 TM 1536 RMI A. africanus 14,63 STW 14 RM3 A. africanus 35,11 STW 269 RM3 A. africanus 33,61 STW 278 RM3 A. africanus 38,94 STW 308 RM2 A. africanus 33,84 STW 309 RMI A. africanus 32,18 STW 364 RMI A. africanus 25,48 STW 412 RM2 A. africanus 19,52 STW 419 LM2 A. africanus 25,43 SK6 RM2 A. robustus 52,22 SK6 RM3 A. robustus 48,20 SK 19 RM2 A. robustus 23,32 SK25 RM1 A. robusfuS 34,59 SK25 RM2 A. robustus 44,70 SK37 LM2 A. robustus 46,88 SK 61 RMI A. robustus 45,89 SK63 LM1 A. robusfuS 36,59 SK 1587 LM2 A. robustus 30,86 KB 5223 LMI A. robustus 31,24 defining relative (or average) enamel thickness as the ratio of enamel cap area in the plane of section (c) divided by the length of the enamel/dentine junction (EDJ) of the same section (e). The vertical thickness of enamel of the buccal and lingual cusps are measured from the cuspal tips perpendicular to a line drawn tangent to the apices of the dentine horns (measurements f and g in Fig. 1 of Grine and Martin 1988) (Figure 1). Figure 1 I mm thin CT slice of SK 6 (RM2) showing measure­ ments used in this study: CT sectioned enamel cap area (in white), enamel height at buccal and lingual mesial cusps (protoconid and metaconid) (white arrows), length of enamel/dentine junction (EDJ) (black arrow). As each scaled CT section was brought up into the IMAGE program on the workstation, the area of the sectioned enamel cap was determined by thresholding (level slicing) the image so that the gray scale levels of all pixels comprising the enamel were highlighted. The area of the highlighted pixels could then be automatically calculated by the IMAGE program. The length of the EDJ could also be calculated by manually tracing its contour on the computer screen with a built-in cursor. EDJ length (e) c/e buc. cusp ht Iing.cusp ht 17,17 1,35 2,17 2.17 20,06 2,27 3,10 2.91 14,40 1,02 1,90 1.67 20,64 1,70 2,39 1.74 17,89 1,88 2,20 2.20 20,35 1,91 2,56 3.02 17,95 1,89 2,62 2.62 19,13 1,68 2,22 2.50 14,62 1,74 2,67 2.22 16,28 1,20 1.96 1.74 15,34 1,66 2.17 1.96 21,47 2,43 3.48 3.26 23,55 2,05 3,04 3.26 13,86 1,68 2,86 2.71 16,24 2,l3 2,54 3.05 21,42 2,09 3,45 3.17 19,75 2,37 3,33 3.33 22,17 2,07 2,56 2.82 17,64 2,07 2,59 3.10 16,50 1,87 2,50 2.50 30,07 1,56 1,52 1.74 Detennination of enamel thickness of buccal and lingual cusps was a straightforward procedure using the built­ in ruler command of the IMAGE program. The CT scan data are associated with a header record that contains calibration information, most importantly the field of view, matrix size (256 or 512), magnification factor, and number of pixels per 10 cm. All measurements were recorded (in millimeters) and entered into a statistical program (Statview II) for analysis. The precision of any measurement calculated from a CT scan is of course, limited by the size of the pixels that make up the image (i.e. pixel edge length). The overall reliability of the measurements derived from the australo­ pithecine CT's was cross-checked in two independent ways. Two specimens, Sts 14 and Sts 96, were CT scanned at right angles to their naturally fractured enamel caps. Enamel thickne ses calculated by the IMAGE program were consistently within 0,2 - 0,3 mm of the true values measured directly on the original specimens (i.e., within the pixel edge length values of the CT images). It is worth empha izing that since this small mensurational imprecision is essentially random with respect to membership in either A. africa nus or A. rohustus, it does not significantly influence the results of compara­ tive differences in absolute or relative enamel thicknesse observed between these two fossil species. Indeed, ran­ dom measurement error would actually make it more difficult to obtain the statistically significant results reported below. The second cross-check on the reliability of the CT methods was quite fortuitous. One of the South African specimens actually sectioned through the mesial cu p by Grine and Martin (1988) was KB 5223, a slightly damaged unworn RM, from Member 3, Kromdraai (Grine 1982). Direct measurements of enamel cap area (c) and EDJ length (e) gave values of 31,71 and 19,56 respectively for a c/e ratio 1,62 (corrected values for the slightly damaged enamel cap were 36,39, 18,00 56 50,-------~~----------------~--------~ 45 40 'E 35 E 30 25 20~------~r_----------------_T---------L A. afrlcanus - ename l area (c l A rObust us - enamel area (cl Figure 2. Mean and one standard deviation bar for enamel area 2 .4 22 2 E 18 E 16 14 1.2.L---------r_----------------_T--------- A africanus - (IE A robustus - C/E Figure 4. Mean and one standard deviation bar for relative enamel thickness (CIE) 3.6,---------'-------------------'----------, 3.4 32 3 28 E 26 E 2 .4 22 2 18 16.L--------,------------------,-------~ A afrlcanus - I mgual cusp ht A. robuslus - lingual cusP ht Figure6. Mean and one standard deviation barforenamel thickness at Lingual Cusp and 1,02respectively) (Grine and Martin 1988). Evaluation of a CT section through the mesial cusps of the unworn LMI from the same individual gave reasonably similar results (31,24 and 20,07 for a c/e ratio of 1,56; Table 1). RESULTS AND DISCUSSION The raw data forenamel thickness measurements are presented in Table 1 and basic statistical data in Tables 2 - 6. Sample means and one standard deviation bars for each of the measurements are illustrated in Figures 2 - 6. 23~--------L-----------------~--------~ 22 21 20 E 19 E 18 17 16 15L-------~r_-----------------r--------~ A. afrlcanus - eo] length (el A robustus - eO] length (el Figure 3. Mean and one standard deviation bar for EDJ Length 36r---------'------------------~------~ 3 . 4 3.2 3 E 2 .8 E 26 24 2 .2 2L-------~------------------~------~ A. afrlcanus - buccal cusp hI A. robust us - buccal cusp ht Figure 5. Mean and one standard deviation barforenamel thickness at Buccal cusp The mean value of enamel cap area in CT sections taken through the mesial cusps of the lower molars is 30.15 mm2 inA. africanus and 39.45 mm2 inA. robustus, a statistically significant difference atthe 5% level (Table 2; p=,026). However, the mean ED] lengths of 17,63 mm in A. africanus and 19.27 mm in A. robustus are not significantly different (Table 3; p=,160). As discussed above, relative enamel thickness (or average enamel thickness) is measured as the ratio of the sectioned enamel cap area to the ED] length from the buccal to the lingual cervix of the same section. Mean values for this ratio of 1,69 mm in A. africanus and 2,03 mm in A. robustus are significantly different at the 5% level (Table 4; p=,019), confirming that robust australopithecines have relatively thicker enamel than their gracile counterparts. While the mean value of enamel thickness at the buccal cusp is greater in A. robustus (2,79 mm) than in A. africanus (2,42 mm), this difference approaches, but is not, statistically significant at the 5% level (Table 5; p=,092). However, the difference in enamel thickness at the lingual cusp between A. robustus (2,89 mm) and A. africanus (2,32 mm) is statistically significant (Table 6; p=,013). (Undoubtedly, differences between the two species would be even greater if the extremely small, rather thin enamelled specimen KB 5223 were not included in the A. robustus sample). TABLE 2: A. africanus A, robustus combined sample Enamel area (mm2) in South African australopithecines Mean SO 95% lower 95% upper Prob. (2-tail) 30,15 8,69 24,63 35,67 ,258* 39,45 9,41 32,72 46,19 34,38 10,00 29,95 38,81 TABLE 3: EDJ length (mm) in South African australopithecines A. africanus 17,63 2,17 16,25 ]9,01 ,160 A. robustus 19,27 3,09 17,06 21,48 combined sample 18,37 2,69 17,18 19,57 TABLE 4: Relative enamel thickness (c/e) in South African australopithecines A. africanus 1,69 ,35 1,47 1,91 ,0188* A. robustus 2,03,27 1,84 2, 23 combined sample 1,84 ,36 1,69 2,00 TABLE 5: Enamel thickness at buccal cusp (mm) in South African australopithecines A. africanus 2,42 ,39 2,17 2,66 ,0924 A. robustus 2,79 ,59 2,37 3,21 combined sample 2,59 ,51 2,36 2,81 TABLE 6: Enamel thickness at lingual cusp (mm) in South African australopithecines A. africanus 2,32 ,50 2,00 2,63 ,0129* A. robustus 2,89 ,49 2,55 3,24 combined sample 2,59 ,5 13 2,36 2,81 * indicates significane at P