Enamel structure and microwear An experimental study of the response of enamel to shearing force.код для вставкиСкачать
AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 85:31-49 (1991) Enamel Structure and Microwear: An Experimental Study of the Response of Enamel to Shearing Force MARY c. MAAS Department of Biological Anthropology and Anatomy, Duke Uniuersity Medical Center, Durham, North Carolina 27710 KEY WORDS Dental microwear, Enamel microstructure, Abrasion ABSTRACT The anisotropic fracturing and differential wear properties of enamel microstructure represent factors that can obscure the predictive relationship between dental microwear and diet. To assess the impact of enamel structure on microwear, this in vitro experimental study examines the relative contributions to wear of three factors: 1) species differences in microstructure, 2) direction of shearing force relative to enamel prisms and crystallites, and 3) size of abrasive particles. Teeth oflemur, Ouis, Homo, and Crocodylus, representing, respectively, the structural categories of prismatic patterns 1,2,and 3 and nonprismatic enamel, were abraded by shearing forces (forces having a component directed parallel to abraded surfaces) and examined by scanning electron microscopy. Striation width increased with particle size for nonprismatic, but not for prismatic, specimens. Direction of shear relative to prism and crystallite orientation had a significant influence on striation width in only some prismatic enamels. The different responses of prismatic and nonprismatic enamels to abrasion reflect the influence of structure, but at the level of organization of crystallites rather than prisms per se. Such interactions explain in part the inability of striation width to discriminate among animals with different dietary habits. Heteroscedasticity and deviations from normality also may confound parametric analyses of microwear variables. Variation in crystallite orientation in prismatic enamels may contribute to optimal dental function through the property of differential wear in functionally distinct regions of teeth. Dental wear patterns offer abundant information about dental function and diet of living and fossil mammals. On a general level, for example, the well established relationship between dental wear facets and chewin patterns offers a reliable picture of the evo ution and diversity of mammalian mastication (see, e. ., Butler, 1952, 1973; Crompton, 1971; Mil s, 1955,1973; Kay and Hiiemae, 1974; Ka , 1977). To infer more detailed dietary in ormation, other studies have focused on microscopic features of dental wear (microwear) (e.g., Walker et al., 1978; Fine and Craig, 1981; Gordon, 1982; Teaford and Walker, 1984; Teaford, 1985, 1988; Grine, 1986, 1987; Kay, 1987; Grine and Kay, 1988; Harmon and Rose, 1988; Walker and Teaford, 1989). The notion that K K F 0 1991 WILEY-LISS, INC. microwear atterns can be used to reconstruct specir)ic dietary habits is based on two assumptions. The first is that microwear variation is controlled primarily b two factors, both of which pertain direct y to diet: the physical properties of foods and ma tude and direction of chewing force. second assumption is that microwear variation within functionally homologous wear surfaces is less than that between functionally distinct surfaces or between individuals with different diets. An important series of studies of living rimates with different feeding habits (i.e., Rard-object feeders and soft-object feeders) P !?& _ _ _ _ ~ Received J u n e 25,1990; accepted October 5 , 1990. 32 M.C. MAAS documented statistically significant associations between microwear and diet (Teaford and Walker, 1984; Teaford, 1985; Teaford and Oyen, 1989).With an eye to more precise documentation of dietary variation, other studies have uantified microwear patterns of species wit less marked dietary differences and of o ulations of the same species that occu y J f g r e n t habitats (e.g., Teaford, 1986; So ounias et al., 1988; Teaford and Robinson, 1989). The relationship between diet and microwear is, not surprisingly, less clear-cut in these cases; some, though not all, metrical parameters of microwear failed to discriminate among dietary groups. These more ambiguous results may indicate a limitation to the predictive resolution of microwear analysis; they also may reflect the failure of the predictive models to account for some critical determinant of microwear morphology. One potentially critical factor is the microstructure of tooth enamel (Rensberger, 1978, 1983; Walker et al., 1978; Puech and Albertini, 1983; Walker, 1984; Boyde and Martin, 1982, 198413; Fortelius, 1985; Puech et al., 1986; Walker and Teaford, 1989). If the structure and physical properties of enamel are variable, rather than constant as has been assumed, then structural variation in enamel may contribute to variation in microwear. 1 P ENAMEL MICROSTRUCTURE tion (Gwinnett, 1967). In nonprismatic enamels, enamel crystallites are arran ed with their long axes parallel to one anot er and roughly perpendicular to the tooth surface. Its structural organization is uniform and relatively homogeneous and therefore is expected to have little differential effect on wear. In contrast to nonprismatic enamels, prismatic enamels are structurally heterogeneous, reflecting both their prismatic organization and the arrangement of individual crystallites that make up the prisms. Enamel prisms are bundles of similarly oriented mineral cr stallites separated from areas of different y oriented crystallites by pronounced discontinuities in orientation (prism boundaries). Differences in crosssectional arrangements of prism boundaries represent one cause of variation in patterns of fracture and wear, since prism boundaries represent potential planes of weakness in enamel and therefore potential routes of referential fracture (Powers et al., 1973a; oyde, 1976a; Rasmussen et al., 1976; Hassan et al., 1981). Prism cross-sectional shape and packin differ among species but also within singe teeth. For example, a single primate tooth may contain not only nonprismatic enamel but three different prism packing patterns (Boyde and Martin, 1987). A second, and well established, source of differential wear concerns the orientation of prism long axes relative to the wear surface (see, e.g., Boyde, 1984a,b).Prism orientation also varies within teeth as well as among taxa. A third potential source of differential wear ertains to the orientation of enamel crystalites, which differ between prisms and interprismatic material, and also among enamel t pes (Fig. 1).The influence of all three of t ese structural factors will differ according to the plane of a wear facet or other functional surface. These microstructural (prism) and ultrastructural (crystallite) features represent an as ect of the enamel wear process that is in ependent of diet. An understanding of its influence may elucidate unanswered questions concerning the nature of abrasive wear and therefore ambiguities in the relationship between microwear and diet. a P li 7 Enamel, a composite tissue that is some 95% mineral by weight, is the hardest of the three dental tissues (enamel, dentine, and cementum) and is tough and resistant to brittle fracture (Boyde, 1976a). Enamel is structurally heterogeneous among taxa (e.g., Boyde, 1964) and within individual teeth (e.g., Boyde, 1976a; Boyde and Martin, 1984a, 1987;Martinet al., 1988; Skobeet al., 1985; Stern and Skobe, 1985; Stern et al., 1989). Importantly, the structural heterogeneity of enamel, both intra- and interspecifically, confers anisotropic properties of fracture and wear (Powers et al., 197313; Boyde, 1976b, 1984a,b; Rasmussen et al., 1976; Hassan et al., 1981). The two major structural categories of mammalian enamel are nonprismatic and prismatic (Fig. 1). Nonprismatic enamel is characteristic of the outer enamel of many EXPERIMENTAL DESIGN mammalian teeth as well as of the enamel of most reptiles. It is reported to be harder than The model of tooth abrasion presented prismatic enamel (Hodge and McKay, 19331, here differs from previous microwear analyperhaps because of its greater mineraliza- ses, which routinely handle force and abra- z B 33 ENAMEL STRUCTURE AND MICROWEAR A B Fig. 1. Three-dimensional block diagrams depicting cross-sectionalpacking arrangement of prisms and relative orientations of crystallites in prisms, interprismatic enamel, and nonprismatic enamel (not to scale). Apical and cervical (relative to the apicocervical axis of a tooth) are towards the top and bottom of each block, respectively. Prism boundaries are shown as solid lines. A. Nonprismatic enamel; crystallites are oriented parallel to one another. B: Pattern 1 enamel; prism boundaries appear as complete circles and prisms are packed in a hexagonal arrangement and separated by interprismatic material with differently oriented crystallites. C: Pattern 2 enamel; prism boundaries are incomplete cervically and are acked in longitudinally oriented rows, separated by sgeets of inter rismatic material. D: Pattern 3 enamel; prism bounfaries are incom lete cervically, but risms are packed in a hexagonay arrangement, anlinterprismatic material is restricted to the cervically located ‘tail” regions of prisms. sive particle properties as variables but have implicitly treated the physical properties of enamel as constants. This model includes enamel microstructure as a variable and treats wear as a mechanical process consisting of interaction among three factors: 1) physical properties of food, 2) vectors of chewing force, and 3) structural properties of teeth. To predict diet from microwear patterns, the influence of physical properties of food and vectors of chewing force must predominate and their interactlon must produce a consistent result (i.e., a characteristic wear pattern). The ex eriments described here detail the resultsnio! vitro abrasion by shearing force; experiments in which force was directed normal to the abrasion surface are reported elsewhere (Maas, 1988, 1989). The shearing experiments tested the influence of variation in particle size and enamel structure; article shape and hardness, magnitude of Force, and direction of force relative to the abrasion surface were “extrinsic” experimental constants (those variables not de endent on the structure of the enamel itse 0. The effect of particle size was tested by alternating abrasive grit size among experiments. The influence of enamel structure was tested at two levels, by varying two different factors. First, the effect of variation in crystallite arrangement was tested by comparing abrasion patterns among species representing the three different prism packing patterns and non rismatic enamel. Second, the effect of di erences in crystallite f K. 34 M.C. MAAS orientation was tested by abrading enamel specimens in different directions relative to an apicocervical axis. Tests of the effects of different directions of shear also addressed the influence of differences in prism pattern, since asymmetry in prism boundaries might produce different responses to abrasive force, depending on the direction of force. The experimental series consisted of 36 experiments, each defined by unique combinations of the three factors: differences in prism packing pattern (represented in these experiments by different species), differences in crystallite orientation relative to direction of shear, and particle size (Table 1). The three variables are designated species, direction, and grit, respectively. Striation width was selected to represent microwear variation. Striation width is of special interest, because, although it is routinely recorded in microwear studies, it typically fails to discriminate among different dietary groups (see, e.g., Teaford, 1986; Solounias et al., 1988; Ryan and Johanson, 1989; Teaford and Robinson, 1989). Even in cases in which mean striation widths differ between groups, within-group variation can obscure the statistical si ificance of between group differences. T is suggests that striation width re resents an attribute of microwear morpho ogy influenced by factors not generally considered in microwear analyses. The null hypothesis The hypothetical framework consisted of the null hypothesis and two alternative hypotheses. The null hypothesis predicts that the effects of the experimental variables, structural factors (species differences in enamel and orientation of prisms and crystallites relative to the direction of shear), and particle size should be random. According to the null hypothesis, there should be no statistically significant difference in striation width between experiments defined by different attributes of enamel structure or by different sizes of abrasive grit. The influence of particle size The first alternative hypothesis predicts that abrasive grit size is the primary determinant of striation width. According to this hypothesis, striation width should vary directly with the size of abrasive particles, given constancy of force vectors, although some variation in striation width would be P ?? TABLE 1 . Experimental design Grit size SDecies Direction of shear (urn) Crocodylus (nonprismatic) Lemur fuluus (pattern 1) Apicocervical 240 (73) Mesiodistal 400 (23) Ouis aries (pattern 2) 45' to apicocervical 600 (14) Homo sapiens (pattern 3) expected because of the irregularity of abrasive particle shape (and therefore variation in contact area between the abrasive and the enamel surface). In addition, because particle shape regularity increases with decrease in size (Wright, 1968), sample variance should decrease with decline in grit size, if the size of contact area is the prime determinant of striation width. The influence of enamel structure The second alternative hypothesis predicts that structural characteristics of enamel would have the predominant influence on striation width. Because of the structural complexity of enamel, there are four separate predictions associated with this hypothesis. 1.If enamel structure influences striation width, then only nonprismatic enamel (which is comparatively homo eneous in structure) would show a random tistribution of striation widths (the null hypothesis) or a clear correlation between abrasive particle size and striation width (the first alternative hypothesis). 2. If arrangement of prism boundaries is the most im ortant structural influence on striation wi th, enamel would be removed referentially in units of whole herefore, striation width distri utions would vary accordin to prism dimensions. This should occur on y when abrasive particle dimensions are greater than the dimensions of prism boundaries and therefore are likely to induce fracture alon prism boundaries. When particles are sma ler than prism diameters, striation widths would show either a random distribution or vary with particle size. 3. If arrangement of prism boundaries is the most important structural influence, the effect of direction of shear (relative to the B ! grisms. B P ENAMEL STRUCTURE AND MICROWEAR 35 apicocervical axis of the tooth) should var among enamel types for enamels in whic potential planes of weakness (prism boundaries) are asymmetric (patterns 2 and 3) and when particle diameters are greater than rism dimensions. This is because rism oundaries in pattern 2 and 3 ename s are arranged so that the most continuous and linear lanes of weakness occur in an apicocervica direction. The continuity and/or linearity of prism boundaries is interrupted in other directions (Fig. l),and such interruptions increase the amount of energy required to propagate a fracture and therefore reduce the ease of abrasion in those directions (Boyde, 1976a; Hassan et al., 1981; Pfretzschner, 1986). 4. If crystallite orientation influences enamel abrasion, then, for prismatic enamels, striation width would differ according to prism packing pattern and according to direction of shear because crystallite orientation varies between prism types and between risms and interprismatic material (Fig. 1). !his variation would be expressed for abrasion by particles smaller than prism diameter. E E P P MATERIALS AND METHODS Materials Abrasive particles consisted of three sizes of commerciallyproduced silicon carbide grit (Buehler, Ltd., Waukegan, IL). Silicon carbide is hard enough t o abrade enamel (Wright, 1968) and has previously been used in experimental studies of enamel abrasion (Wright, 1968; Ryan, 1979a,b).The grit sizes were selected to maximize size-related differences: 240 grit (maximum particle diameter = 73 Fm), 400 grit (maximum particle diameter = 23 km), and 600 grit (maximum particle diameter = 14 m) (Fig. 2). The abrasives are irregular in s ape, but it is important to emphasize that the degree of shape variability in these abrasive grits is comparable to that found in naturally occurring abrasives that have been implicated in the microwear process (Kay and Covert, 1983). Size of particles relative to prism diameters was an important aspect of the test of structural influence. Particles greater than rism diameter test the effects of prism goundary arrangements on striation width and particles smaller than prism diameter test the effects of crystallite orientation. On t Fig. 2. SEM micrographs of silicon carbide abrasive grits. A 240 grit; B: 400 grit; C: 600 grit. Bar = 100 pm. 36 M.C. MAAS average, particles abrade enamel with a contact area approximately one-tenth their maximum dimension (Wright, 1968). Consequently, although the maximum diameter of all three grits is greater than that of the average dimension of enamel prisms for the experimental species (3-8 pm), it was expected that the abrasive contact area of 600 (14 pm) grit would most frequently be smaller than average prism diameter and that such small contact areas would be less frequent for 400 (23 pm) grit and infrequent for 240 (73 pm) grit. The experimental specimens represented four species, each characterized by one of the four structural categories of enamel found in extant and extinct mammals. In addition t o the differences in prism morphology, the enamel types represent differences in crystallite orientation (Fig. 1).Cheek teeth of Lemur fulvus (brown lemur), Ovis aries (sheep),and Homo sapiens correspond to the prismatic enamel patterns 1, 2, and 3, res ectively. Teeth of Crocodylus rhombifer uban crocodile) represent nonprismatic enamel. Although all four structural types ma be found in single teeth of some mamma1ys, a single pattern redominates in each of the four species se ected here, ensuring control of maximum structural differences between experiments. Specimen preparation Molars of Lemur, Ouis, and Homo extracted and stored in 70% ethanol and drystored shed teeth of Crocodylus were cleaned ultrasonically with ethanol, rinsed, rehydrated in distilled water for 24 hr, and allowed to air dry. Preliminar experiments demonstrated that the dif? erent storage methods had no effect on response to abrasion (Maas, 19881,a conclusion that supports the results of other hardness and abrasion tests that utilized rehydrated air-dried and alcohol-stored specimens (Miihlemann, 1964; Boyde, 1984a,b). After rehydration, each specimen was mounted with epoxy cement on an aluminum scanning electron microscope (SEMI stub. Flat facets were prepared on horizontally oriented tooth surfaces using a series of increasingly fine grades of abrasives (600,800, 1,200grade polishing aperand 1.0-0.05 pm alumina oxide polis ing powder). In all cases, the abrasive facet was prepared on an undamaged, nonocclusal surface. The polished facet was maintained in a level orientation during preparation and abrasion by (8 f 1 means of a specially designed specimen holder (Fig. 3). The horizontal orientation of the facet ensured that magnitude and direction of force were consistent amon all ex eriments, with no variation due to t e ang e of the specimen surface. The horizontal facet orientation also eliminated the potential problem of distortion of microwear dimensions during SEM viewing that results when wear facets are not oriented perpendicular to the electron beam (see, e.g., Grine, 1986; Gordon, 1988). An abrasive sliding device (Fig. 3) ensured consistenc of direction and magnitude of shearini&-ce for all of the experimental trials. T is device consisted of a glass plate resting on flexible wheels above the specimen holder. A fixed amount of abrasive rit was applied to a glass slide and attache to E f 2 specimen holder with a sin le stroke. Abrasion facets (the areas of t e specimens in contact with the abrasive late) were roughly equivalent in size for a specimens; therefore, magnitude of force (kglmm') was considered constant among experiments. To test the factor of direction of shear, the apicocervical axis of the tooth relative to the orientation of the plate was adjusted for each experiment. Each of the four enamel types was tested at three orientations relative to the apicocervical axis of the tooth: parallel to the apicocervical axis, 45" to the apicocervical axis, and 90" to the apicocervical axis (mesiodistal). These directions are respectively designated A-C, 45", and M-D in subse uent discussions. !'ourteen of the 36 experiments were replicated. Following each experimental trial, the abraded s ecimen was removed from the specimen hol er and cleaned, and the abrasion facet was marked with a diamond stylus adjacent to the densest occurrence of microwear features. Vinyl polysiloxane molds Regular Set; 3-M Go., casts (Araldite hardener; CibaGeigy Corp., Hawthorne, NY)were made of each specimen. The epoxy casts were mounted on SEM stubs and, by comparison with camera lucida drawings of the originals, positioned identically to the enamel specimens. Following the abrasion experiments, the fl K rs ENAMEL STRUCTURE AND MICROWEAR 37 Fig. 3. Abrasion device for shearing experiments. A Tooth in specimen holder; B: plate containing abrasive grit; C: weights. See text for explanation of the abrasion procedure. enamel specimens were etched with 0.5% H3P04for 10-15 sec to highlight microstructural features. Epoxy replicas and etched enamel s ecimens were examined in a JEOL 35C SE , in secondary electron mode at 10 kV accelerating volta e. SEM examination of specimens include photo a hic documentation of abrasion facets or 0th epoxy replicas and original enamel specimens. Four areas of microwear on each replica were photographed at x 400 for measurement of microwear features. Identical areas were documented on the original specimens for evaluation of the structural features underlying the microwear. The stylus mark facilitated consistent orientation of enamel s ecimens and epoxy resin replicas in the EM chamber. Data collection For each specimen, the orientation of prisms relative to the enamel surface was determined by examination of micrographs and specimens classified in one of three nominal prism orientation categories: 1) transverse, 2) oblique, and 3) mixed (fields including prisms with transverse and with oblique orientations). Classification of prism orien- d fY !i tation provided a means of evaluating the relationship of striation width to prism long axis orientation, by comparing replicate experiments that differed only in prism orientation. The objective of collection of metrical data from experimentally produced microwear features was to evaluate the variability of feature dimensions within and between experiments. Width of each striation (based on a single site) was recorded with a Graf-Pen sonic digitizer (Science Accessories) and computer from tracings of enlarged projections of micrograph negatives. Striation width was defined as a line perpendicular to the lon axis of a striation. In addition, micrograp s were evaluated visually to assess size and distribution of wear features and structural features, including prism boundaries and orientation of prism and crystallite long axes. StatisticaE analysis Most microwear studies have ado ted nonparametric (e.g., Teaford and WaR( er, 1984) or parametric analyses of individual sample means (e.g., Grine, 1986; Teaford, 1988; Solounias et al., 1988; Teaford and a 38 M.C. MAAS Robinson, 19891, because microwear data tend not to meet the assum tions of parametric statistical tests (e.g., omogeneity of variance, normality, and independence). However, in this stud , an important goal is evaluation of the inf uence of sample variability on among-group differences, and parametric tests of equality of variance are the most appro riate. Striation wi th distributions for each experiment were tested for deviations from normality by G tests (Sokal and Rohlf, 19811, and the entire experimental test was examined for homoscedasticity (homogeneity of variance) (Bartlett’s test and Fmax test; Sokal and Rohlf, 1981). Because no single transformation succeeded in normalizing 95% of the experiments or reducing heteroscedasticity to an acceptable limit, two alternative methods of parametric analysis were adopted. The first alternative, the Games and Howell method for multiple comparisons, is appropriate for inherently heteroscedastic data (Games and Howell, 1976; Sokal and Rohlf, 1981). The second alternative is appropriate for both heteroscedastic and nonnormally distributed data. It entailed transformation of individual variates to ranks, and the application of the analysis of variance (ANOVA)procedure to the ranked variates (Iman, 1974; Conover and Iman, 1976,1981). R P B Factorial ANOVA of the rank-transformed data was used to evaluate the contribution of each factor ( it size, species differences, and direction o shear) and the interaction among factors to the overall variance of the experiments. The design was a model I ANOVA testing main effects (the contribution of each factor to the overall variance) and interaction (the effect of the dependence of one factor on the level of another factor); model I ANOVA is appropriate because all factors re resented fixed treatments rather than ran om effects (Sokal and Rohlf, 1981). A series of planned comparisons tested which ex erimental means or groups of means di fered from others. The planned comparisons include tests for significant differences among subgroups defined by each experimental factor (species, grit size, direction of shear) and tests for significant differences among groups defined by unique combinations of factors (e.g., among grit sizes for each direction of shear and each species, among species for each grit size and each direction of shear). Table 2 lists the questions addressed by each planned comparison. ff cp f! RESULTS Observations The experiments produced discrete and nonoverlapping elongate microwear fea- TABLE 2. Planned comparisons among experimental subsets Hypotheses First-order interactions 1. Does microwear width differ according to grit size, for each species (all directions combined)? 2. Does microwear width differ according for each grit size - to soecies (all directions combined)? Does microwear width differ according to direction of shear for each species (all grit sizes combined)? Interactions Species X grit Species X direction Does microwear width differ according to direction of shear for each grit size (all species combined)? Does microwear width differ according to grit size, for each direction of shear (all species combined)? Direction X grit Does microwear width differ according to grit size for each direction of shear (all species combined)? Second-order interactions 7. Does microwear width differ according to grit size for each direction of shear, for each species? 8. Does microwear width differ according to species for each direction, within each grit size? Species X grit X direction 9. Does microwear width differ according to direction of shear for each grit size, within each species? 39 ENAMEL STRUCTURE AND MICROWEAR tures (striations). Striations coursed freely across prism cores (Fig. 4). For pattern 1 there was no discernible pattern of difference in striation dimensions between those portions crossing prism cores and those crossing interprismatic regions (Fig. 4A). For pattern 2, enamel striations generally were better defined where they passed across prism heads than where crossing in- terprismatic regions, but they did not deviate along prism boundaries (Fig. 4B). Prism boundaries apparently exercised no influence on striation width for pattern 3 enamel specimens (Fig. 4C); striations were no more variable than those viewed on etched nonprismatic enamel specimens (Fi . 4D).The relationship of striation boun aries and prism boundaries was the same for both s Fig. 4. Micrographs of abraded enamel (etched for 10 sec with 0.5% H,PO,). A Lemur (pattern 1 enamel); B: Ouis (pattern 2 enamel); C: Homo (pattern 3 enamel); D: Crocodylus (nonprismatic enamel), Bar = 10 km. 40 M.C. MAAS large (greater than prism diameter) and small (less than rism diameter) striations; there was no evi ence that larger striations were produced by fracture along prism boundaries and removal of whole prism fragments. Variation in striation width is therefore not a direct reflecton of differences in prism size. As expected from previous descriptions of prism orientation in surface enamel ofHomo and Ouis (e.g., Boyde, 1964, 1969, 1976b), almost all the human specimens exhibited transverse prisms (relative to the facet plane), whereas most Ovis specimens were characterized by combinations of oblique and transverse prisms (mixed), with considerable variation in the obliquity of the prism long axes. For Lemur specimens, prism orientations were either transverse or oblique. Consequently, comparisons between prism orientation categories for a single species, direction of shear, and grit size were limited. Five experiments (Lemur 400 grit/M-D and 600 grit/M-D, Ouis 240 g r i m - D and 400 grit/M-D, Homo 600 gritI45") included replicates re resenting different prism orientations. ean striation width was greatest within each replicate pair for the specimen with more obliquely oriented prisms, with the exception of the Ouis 240 grit/M-D experiment. Analysis of variance The complex relationship between striation width and the factors considered in this experimental stud is illustrated by the plot of mean -+ 1 stan ard deviation of natural log-transformed striation widths for each experiment (Fig. 5). Any obvious correlation between striation widths and any single factor (grit size, species differences, or direction of shear) is obscured by the degree of overlap among ex eriments. The in uence of interactions among factors on striation width is clear from the results of the overall ANOVA (Table 3). Not only was the effect of each factor significant, but the second-order interaction among all three factors (species X grit x direction) was highly significant. In other words, the effect of each factor varied according to the expression of the other factors. Consequently, any statement about the effect of a single factor must be qualified by its dependence on the other factors (Kempthorne, 1975; Sokal and Rohlf, 1981). For these experiments then, the conclusion that each main effect (species, B 600 M-D - c - H - 0 - L - c 600 600 H c A-C 4 0 I L I - c, H - 0 - L 450 t o I A-C U 240 4 i 450 I -c I 0 L U I s B R Ln STRIATION WIDTH Iuml * Fig. 5. Mean 1 standard deviation natural logtransformed striation width, by ex eriment. L, Lemur attern 1);0, ouis (patFern 2!; H,Romo ( attern 3); C, 8rocodylus (nonprismatic). Grit sizes (240flargestl,400 [intermediate], 600 [smallest]) and direction of shear relative to the apicocervical tooth axis (45,', A-C [apicocervical], M-D [mesiodistal]) are indicated in the column on the left. grit, and direction) contributed significantly to differences between experiments is not very meaningful, since the effect of each was dependent on the other factors. Nevertheless, the presence of significant interaction among factors is an im ortant characteristic of the wear process. nteraction, tested by planned comparisons, includes first-order interactions (interactions between two factors, summed over all levels of a third factor) and second-order interactions (interactions among all three factors). First-order interactions: Species and grit, species and direction, direction and grit First-order interactions usually are not tested when second-order interactions are P 41 ENAMEL STRUCTURE AND MICROWEAR TABLE 3. Summary ANOVA for shearing experiments for rank-transformed data Source of variation Main effects Species Grit Direction First-order interactions Species X grit Species X direction Grit X direction Second-order interactions Species X grit X direction Error D.F. F 3 2 2 65.14*** 21.03*** 35.51*** 6 6 6 5.78*** 12.40*** 8.55*** 12 3.013 6.75*** *** P < 0.001. significant. In this case, however, due to the fact that the critical value for the secondorder interaction was so small (because of the large number of error degrees of freedom in this analysis), further analysis was expected to be of biological, if not statistical, significance. The first-order interactions are illustrated graphically in Figure 6, where each point represents the mean value of a subset of experiments, defined by unique combinations of two factors summed for all levels of the third factor. Each species in the plot of mean striation widths for the species x grit interaction (for each species and each grit size, averaged for all directions of shear) shows different trajectories (Fig. 6A).Nonparallel trajectories denote significant interaction among factors (Sokal and Rohlf, 1981).Therefore, the plot illustrates how the effect of grit size on striation width differed according to enamel structure at the species level. The trajectories differ most markedly between prismatic enamels and nonprismatic enamel. Surprisingly, for the prismatic enamels mean striation widths for the 240 grit experiments (all shear directions combined) were smaller than mean striation widths for the smaller grit experiments (400 and 600grits). Only the nonprismatic Crocodylus enamel showed the relationship predicted by the first alternative hypothesis (particle size as the main determinant of striation width) of decrease in striation width with decrease in grit size. In contrast, for the three prismatic enamels the relationship between mean striation width and grit size was op osite that predicted. Although the plannei compari- sons were not significant for an single species (unlike the significance o the overall anova), the contrast between prismatic and nonprismatic enamels suggests that the effect of particle size differs for the two enamel ty es. !?he significance of the apparent dichotom in abrasion response between rismatic an nonprismatic enamels is clari ied when the effects of direction of shear are considered. The influence of shear direction was tested in two of the first-order interactions, s ecies x direction and direction x grit. &en all grit sizes were combined, the species X direction interaction (Fig. 6B) resulted in greater mean striation width for A-Cshear for all four of the species, although the differences were significant only for Lemur. Plots of mean striation width for each direction and each grit size, summed over all species (direction x grit) (Fig. 6C),show that for 45"shear and M-D shear mean striation width decreased with decrease in grit diameter. However, the plot for mean striation width by it size for A-C shear differed dramatical y from the other two: Striation width increased with decrease in grit size. The differences in striation width were significant only for the 600 grit oup, summed over all species, but the very ifferent trajectories of the plots suggest stron ly that striation width was influenced by actors other than grit size when shear was directed a icocervically. In sum, analysis of first-or er interactions indicates that the effects of grit size on striation width are mitigated by species differences (expressed in the dichotomy between prismatic and nonprismatic enamels) and by direction of shear (for prismatic enamels). Second-order interactions: Species, grit, and direction The interaction among all three factors (second-order interactions) is de icted from a single perspective, grouped y species, each point representing the mean value for a single experiment (Fig. 7). If there were no significant interactions among factors, then the tra'ectories of the plots of mean striation width direction of shear for each species would e arallel. Instead, each species shows a dif erent pattern for mean striation widths according to grit size and direction of shear. Only the nonprismatic enamel (Crocodylus)specimens approach the pattern expected if grit size were the prime determi- P B P Y f. B cp E ii P 42 M.C. MAAS SPECIES X DIRECTION SPECIES X GRIT fi I I- Lemur 0 3 z 2000- 2 4 a I- cn 0 1600W z 2 cn z d 1200- z '\i z w 4 c o d y l u \ s a --- 800 240 A 400 8 GRIT SIZE 0 45 600 A-C M-D DIRECTION OF SHEAR DIRECTION X GRIT I I- P 3 2000 0 z U I- cn eool 240 C 400 600 GRIT SIZE Fig. 6. Plots oftwo-way (first-order)interactions. A: Species and grit (all directions of shear);B: species and direction (all grit sizes); C: direction and grit (all species). 43 ENAMEL STRUCTURE AND MICROWEAR 0 VIS LEMUR /450 ,."* 1800 M-D 1300 M-D A-C Y z z 2 2 800 240 BOO 400 000 B GRIT SIZE 240 400 800 GRIT SIZE CROCODYLUS HOMO A-C 0 W z B v) z 450 M-D 1300 U a Lz 2 k z A e c o o 400 240 l 800 GRIT SIZE D 4 00 240 800 GRIT SIZE Fig. 7. Plot of three-way (second-order) interactions among factors (species, grit, and direction), grouped by species. A Lemur; B: Ouis; C: Homo; D: Crocodylus. nant of striation width (decrease in striation width with decrease in grit size). For the three prismatic enamels, the apicocervical shear experiments show an increase in mean striation width with decrease in mean grit size. Furthermore, for the three prismatic enamels mean striation width is greatest for apicocervical shear of all the shearing directions, but only for the two smaller grit sizes (400 grit and 600 grit). Multiple comparisons Multiple comparison tests (total experimentwise error rate of 0.05 for 630 possible comparisons) general1 sup ort the interpretations based on d O V A . q h e most striking aspects of the multiple comparisons are the absence of significant differences amon the nonprismatic enamel experiments an the frequent occurrence of significant differ- % 44 M.C. MAAS ences between nonprismatic and prismatic enamel experiments. Amonf prismatic enamels, there are few signi icant differences between pattern 2 and pattern 3 experiments, although both differ significantly from pattern 1experiments. DISCUSSION The Null Hypothesis Based on the significant results of the analysis of variance and multiple comparisons, the null hypothesis that striation widths are random according to the experimental factors of particle size, species differences in structure, and direction of shearing force can be rejected. However, the significant interaction among all three factors demonstrates that the effect of any one experimental factor was dependent on the expression of the other factors. Therefore, the alternative h potheses of microstructural influence an particle size influence outlined above must be reevaluated. The influence of particle size The clear lack of association between average particle size and average striation width for all the prismatic enamel experiments demonstrates no support for the conjecture that width of microwear features reflects the size of abrasive particles (see, e.g., Ryan, 1979a; Wells et al., 1982; Solounias et al., 1988; Teaford and Robinson, 1989). However, nonprismatic enamel experiments, in contrast to prismatic enamel experiments, did show a strong association between average width and average particle size. The silicon carbide grit particles used in these experiments are, like naturally occurring abrasives, irregular in shape. Kay and Covert (1983) sug ested that particle shape and hardness, ratfler than size per se, may be the important determinant of striation morphology, and Walker et al. (1978) explicitly noted that the area of contact between grit and enamel will vary for irregularly shaped particles. As predicted, variance in striation width declined with decrease in average particle size, presumably because of the greater regularity in sha e of smaller abrasives (400 and 600 grit a rasive particles are more uniform in sha e than 240 grit silicon carbide grit particles! Therefore, although particle shape variability may contribute to the lack of correlation between avera e striation width and average particle size, t is factor should be least important for B E a very small particles. In these experiments, however, it was the width of striations produced b the smallest sized particles that deviate the most from the predicted pattern of the particle size hypothesis. In any event, whether particle shape, enamel microstructure, or some other variable was the mitigating factor, there is no evidence from these experiments that striation width is directly correlated with particle size when prismatic enamel is abraded by shearing force. A variety of substances potentially produce microwear. Among the most frequently mentioned are exogenous mineral particles and plant phytoliths; both are characterized by properties of shape and hardness similar to the silicon carbide used in these experiments. Given the results of this study, similarity in striation width cannot be used uncritically as evidence of similarity in size of abrasive particles. The influence of enamel microstructure Differences in striation width distributions did not correspond directly with prism packing patterns, but did differ according to prismatic or nonprismatic structure. The distinction between the two enamel types appears to lie in their responses to the different directions of shear. Direction of shear had a differential effect on striation width for prismatic enamels and not for nonprismatic enamel. The structural attributes of prismatic enamel that potentially contribute to anisotropic wear properties are 1) the arrangement of groups of similarly oriented crystaland inter rismatic lites into material, 2) t e orientation of the ong axes of risms, and 3) the orientation of individua enamel crystallites or subparallel crystallite groups (Ronnholm, 1962; Boyde, 1964; Boyde and Fortelius, 1986; Maas, 1988) (Fig. 1).The first two characteristics, prism packing atterns and prism orientation, appl to t e enamel at a microstructural leve , whereas the third (enamel crystallites) is an expression of structure on the smaller, ultrastructural scale. All three structural attributes form a regular rather than random structural attern, because enamel structure reflects t e mor holo enamel-secreting cells during toot deveY opO f ment. Although the correspondence was not perfect, the hypothetical model did explain some of the differences in wear between prismatic B irisms P P P K R 45 ENAMEL STRUCTURE AND MICROWEAR and nonprismatic enamels observed in these experiments. First, the model predicted that the effects of nonstructural factors would be expressed most clearly for nonprismatic enamels and that, for prismatic enamels, structure would mitigate the effect of particle size. The experimental results, which showed different patterns of striation width for prismatic and nonprismatic enamels, were consistent with this prediction. Second, the model identified two potential structural influences on striation width, the arrangement and orientation of prism boundaries and the orientation of cr stallites. The difference in scale between t e microstructural features (prisms) and ultrastructural features (crystallites) provided a means of testing the relative contribution of each aspect of enamel structure to striation width. The lack of a consistent attern of sigf for nificant between-species c ifferences prismatic enamels, demonstrated by the multiple comparison tests, sup orts the conclusion that rism size and pac ing arrangement per se $o not influence striation width. This statistical evidence is consistent with observations that striations coursed across prism cores and through interprismatic material and were not constrained by arrangements of prism boundaries. It is possible that differences in the orientations of prism long axes among specimens of the same species obscured between-species differences and precluded a fair test of the prismatic structural influence hypothesis. However, the pairwise multiple com arison results indicate that this was not t e case. The observation that the three prismatic enamels, and, in particular, pattern 2 and pattern 3 enamels, showed larger than expected striation widths for apicocervical shear for the experiments with small abrasives (400 and 600 grits) but not the experiments with large abrasives (240 grit) supports the hypothesis that orientation of crystallites, rather than attributes of prism boundaries, is the important structural constraint on microwear. These abrasion experiment results also are consistent with experimental studies of fracture that have documented the propensity for enamel to fracture parallel to rather than perpendicular to crystallite long axes (Rasmussen et al., 1976) and are consistent with suggestions that enamel is preferentially removed in units smaller than enamel prisms (Boyde and Fortelius, 1986). The mechanical expla- t !i K nation lies in the orientation of enamel crystallites within risms and inter rismatic material: Crysta lite long axes wit in prism bodies are parallel to the prism long axis but have a more apicocervical orientation in prism tails or interprismatic regions, particularly for attern 2 and attern 3 prisms (see Fig. 1). his preferre orientation allows greater ease of abrasion, and therefore greater striation width, when shear is directed apicocervically. When the long axis of a prism is oriented obliquely, the net orientation of crystallite long axes also is more oblique, for all three prism patterns. In this study, experiments in which the net orientation of prisms was oblique to the surface were characterized by greater mean striation widths, for any given grit size, than for otherwise identical experiments in which the net orientation of prisms (and therefore crystallite long axes) was ierpendicular to the direction of force. Bot these observations, the larger striation width for shear in an apicocervical direction and the larger striation widths when prism long axes were oriented more oblique to the wear surface, are consistent with the hypothesis that crystallite orientation influences abrasive wear. P 8 g a CONCLUSIONS This analysis specifically evaluated the im act of within-individual variability on differences between microwear samples and found that in most cases within-sample differences obscured those between groups. Analysis of individual means, the approach commonly adopted in microwear studies whose goal is discrimination among groups, avoids this problem by ignoring withinsample variance. Although arguably a propriate in such cases, ignoring varia ility within individuals also ignores an inherent property of microwear distributions, a property that reflects the multiplicity of factors that influence microwear morphology. In this study, variability in enamel microstructure was in part exemplified by differences among four species. In fact, the same degree ofvariation (three prism packing patterns and nonprismatic enamel) can be found within single teeth (e.g., Boyde and Martin, 1987). Likewise, differences in prism orientation, demonstrated here between different specimens of the same species, also may characterize individual wear surfaces. Therefore, the most careful control of phylogenetic and functional consider- E 46 M.C. lMAAs ations may not be enou h t o eliminate the potentially confounding actor of microstructural variation from comparative wear studies. The causal complexity of the tooth wear rocess demonstrated by this study is the [kely explanation for the inability of parametric analyses of striation width to discriminate among some dietary variables. Likewise, variation in other microwear parameters also may be explainable in part in terms of a complex causal relationship that includes both dietary factors and factors independent of diet. If that is the case, there may indeed be a limit to the explanatory ower of microwear pattern analysis. That Emit can be defined on1 by more rigorous analyses of the process o tooth wear, as well as its pattern (see, e.g., Teaford, 1988; Teaford and Oyen, 1989). Other factors whose influence on microwear morphology has et to be assessed fully include magnitude o force, variation in particle shape and hardness, and cumulative effect of wear on morphometrics of individual features. One study (Peters, 1982)has suggested that particle hardness may influence resence or absence of microwear produce by compression but indicated no relationship between size of wear features and particle hardness. The effects of variation in magnitude of force has been tested under conditions of com ression (Maas, 1988, 1989)but only to a imited extent for shearing force (Ryan, 1979a). Other potential influences on microwear that have yet to be tested empirically for enamel include the effects of multiple abrasion events (see, e.g., Powers et al., 1973b; Teaford, 1988). In this study it has been shown that experimentally controlled tests can refute previous assumptions of causation; further experimental studies should provide the necessary information for evaluating the relative contributions of all these factors in a rigorous and meaningful manner. It has recently been claimed that the causes of tooth wear, although a legitimate research roblem, are not of primary relevance to ietary reconstruction (Walker and Teaford, 1989). Walker and Teaford argued that demonstration of correspondence between wear pattern and diet is enough to establish a predictive relationship, regardless of their causal relationship. Certain1 the increasing number of comparative stu ies documenting significant differences in P F F B P cp B wear patterns between animals with major differences in diet attests to the utility of microwear for dietary reconstruction. Nevertheless, misinterpretations of the causes of diagnostic wear patterns, arguably irrelevant when comparing very different patterns of microwear, may be of considerable importance when attempting to assess more subtle differences in microwear and therefore in diet. An understanding of the determinants of tooth wear facilitates assessment of the potential of different parameters to discriminate dietary differences; if a parameter varies with nondietary factors, as is the case with striation width, there is little point in evaluating its morphology or distribution, On the other hand, analyses of attributes of microwear that are unequivocally controlled by factors related to diet should ield useful results. Although this study in&cates that striation width is not such an attribute, variables such as microwear feature shape, density, and erhaps orientation ap ear to be more close associated with diet. ven if the mechanica determinants of a wear pattern are not the primary concerns of dietary reconstruction studies, knowledge of their causes is necessary for appropriate and informed analysis of the data. In addition to the cautionary note regarding causal interpretation of microwear, this study raises questions about the relationshi of tooth wear and enamel structure. It is we known that tooth form changes with wear and that for many mammals changes due to wear serve to optimize the biological role of the dentition (see, e.g., Berry and Poole, 1974; von Koenigswald, 1982; Fortelius, 1985; Stern et al., 1989). Among some primates, tooth wear serves to maintain occlusal function by, for example, maintaining sharp shearing crests or enlarging crushin basins (Kinzey and Rosenberger, 1976). T is ontogenetic component of dental function, which is clearly related to differential wear, has yet to be investigated. As was pointed out previously, differential rates of wear are close1 related to enamel structure (see, e.g., Rens erger and von Koenigswald, 1980; von Koenigswald, 1982; Boyde, 1984a,b; Walker, 1984). For the most part, structurally constrained wear properties have been considered in relation to relatively substantial wear and in relation to prominent structural features such as decussation zones (groups of similarly oriented prisms). This study demonstrates that ultrastruc- E pr K a il ENAMEL STRUCTURE AND MICROWEAR tural variation, the arrangement and orientation of crystallites in prisms and interprismatic material, contributes to variation in size of wear features. Such differences may ultimately control the amount of wear in a particular region of a tooth. Prismatic enamel is characterized by regular patterns of microstructural and ultrastructural variation, including the distribution of prismatic and nonprismatic enamel within teeth, differences in proportions of prismatic and interprismatic enamel, and differences in prism orientation that ma contribute to regional differences in toot function. It has been suggested that the evolution of prismatic enamel is linked to the maintenance of functionally critical occlusal relations between diphyodont teeth (e.g., Stern et al., 1989). The relationship between differential wear and ultrastructural variation may prove to be an important as ect both of age-related wear and of dental aiaptation and evolution in mammals. K ACKNOWLEDGMENTS The material presented here represents an as ect of the author’s dissertation, completelat SUNY Stony Brook, and I thank the members of my committee, Drs. Fred Grine, John Gwinnett, Bill Jungers, Lawrence Martin, and es ecially Dr. David Krause for their advice an assistance during the course of this research. Drs. R.F. Kay, D.W. Krause, M.J. Ravosa, and Ms. S.G. Strait provided helpful comments on an earlier draft of the manuscript. Josef Muenni drafted Fi 1 and 5, and I also than him for p E : graphic and engineering assistance. The American Fellowship Program of the American Association of University Women generously provided financial su port during the completion of this researck. This research was supported by the Doctoral Program in Anthropologcal Sciences, SUNY at Stony Brook, and NSF grant BNS-8611198. B a LITERATURE CITED Berry DC, and Poole DFG (1974) Masticatory function and oral rehabilitation. J . Oral Rehab., 1:191-205. Boyde A (1964)The Structure and Development ofMammalian Enamel. PhD Thesis, University College, London. Boyde A (1969) Electron microscope observations relating to the nature and development of prism decussation in mammalian dental enamel. Bull. Group Int. Rech. Sci. Stomat. 12:151-207. Boyde A (1976a) Amelogenesis and the structure of enamel. In B Cohen and IRH Kramer (eds.): Scientific 47 Foundations of Dentistry. London: William Heinemann, pp. 335-352. Boyde A (1976b) Enamel structure and cavity margins. Operative Dentistry 1:13-28. Boyde A (1984a)Qrpolishing effects on enamel, dentine, cement and bone. Br. Dent. J. 156:287-291. Boyde A (1984b) Dependence of rate of physical erosion on orientation and density in mineralised tissues. Anat. Embryol. 170:57-62. Boyde A, and Fortelius M (1986)Development, structure and function of rhinoceros enamel. Zool. J. Linn. SOC. 87:181-214. Boyde A, and Martin, L (1982) Enamel microstructure determination in hominoid and cercopithecoid primates. Anat. Embryol. 165:193-212. Boyde A, and Martin L (1984a) The microstructure of primate dental enamel. In DJ Chivers, BA Wood, and A Bilsborough (eds.):Food Acquisition and Processing in Primates. New York: Plenum Press, pp. 341-367. Boyde A, and Martin L (198413)A non-destructive survey of prism packing patterns in primate enamel. In RW Fearnhead and S Suga (eds.): Tooth Enamel IV.Amsterdam: Elsevier Science Publishers, pp. 41 7 4 2 1 . Bo de A, and Martin L (1987) Tandem scanning reiected light microscopy of primate enamel. Scanning Microsc. 1:1935-1948. Butler PM (1952) The milk molars of Perissodact la with remarks on molar occlusion. Proc. Zool. %ci Lond. 121:777-817. Butler PM (1973) Molar wear facets of Early Tertiary North American Primates. InMRZingeser (ed.):Craniofacial Biology of Primates. Vol. 3. Symposium of the N t h International Congress on Primatology. Basel: Karger, pp. 1-27. Conover WJ, and Iman RL (1976) On some alternative procedures using ranks for the analysis of experimental designs. Commun. Statist. Ser. A. 5:134&1368. Conover WJ, and Iman RL (1981)Rank transformations as a bridge between parametric and nonparametric statistics. Am. Statist. 35:124-133. Crompton AW (1971) The origin of the tribosphenic molar. In DM Kermack and KA Kermack (eds.): Early Mammals. New York: Academic Press, pp. 65-87. Fine D, and Craig GT (1981) Buccal surface wear of human premolar and molar teeth: A potential indicator of dietary and social differentiation. J. Hum. Evol. 10:335-344. Fortelius M (1985) Ungulate cheek teeth: Developmental, functional and evolutionary interrelations. Acta Zool. Fenn. 18O:l-76. Games PA, and Howell JF (1976) Pairwise multiple comparison procedures with unequal N s and/or variances: A Monte Carlo study. J. Educ. Statist. 1:113125. Gordon KD (1982) A study of microwear on chimpanzee molars: Implications for dental microwear analysis. Am. J . Phys. Anthropol. 59:195-215. Gordon KD (1988) A review of methodolo and quantification in dental microwear analysis.!%canning Microsc. 2:1139-1147. Grine FE (1986) Dental evidence for dietary differences in Australopithecus and Parunthropus:A quantitative analysis of permanent molar microwear. J. Hum. Evol. 15:783-822. Grine FE (1987) Quantitative analysis of occlusal microwear in Australopithecus and Paranthropus. Scanning Microsc. 1:647456. Grine FE, and Kay RF (1988) Early hominid diets from 48 M.C. MAAS quantitative image analysis of dental microwear. Nature 333:765-768. Gwinnett AJ (1967) The ultrastructure of the “prismless” enamel of permanent human teeth. Arch. Oral Biol. 12:381-387. Harmon AM, and Rose J C (1988) The role of dental microwear analysis in the reconstruction of prehistoric diet. In BV Kennedy and GM LeMoine (eds.):Diet and Subsistence: Current Archaeological Perspectives. Calgary; The Archaeological Association of the University of Calgary, pp, 267-272. Hassan R, Caputo AA, and Bunshah RF (1981) Fracture toughness ofhuman enamel. J . Dent. Res. 60:820-827. Hodge HC, and McKa H (1933) The microhardness of teeth. J . Am. Dent. lssoc. 20:227-233. Iman RL (1974) A power study of a rank transform for the two-way classification model when interaction may be present. Can. J. Statist. Sec. C Applic. 2:227239. Ka RF (1977) The evolution of molar occlusion in the Zercopithecidae and early catarrhines. Am. J. Phys. Anthropol. 46:327-352. Kay RF (1987) Analysis of rimate dental microwear using image processing teciniques. Scanning Microsc. 1:657-662. Kay RF, and Covert HH (1983) True grit: A microwear experiment. Am. J. Phys. Anthropol. 61:33-38. Kay RF, and Hiiemae KM(1974) Jaw movement and tooth use in recent and fossil primates. Am. J. Phys. Anthropol. 40:227-256. Kempthorne 0 (1975) Fixed and mixed models in the analysis of variance. Biometrics 31:473486. Kinzey WG, and Rosenberger AL (1976) Functional patterns of molar occlusion in platyrrhine primates. Am. J. Phys. Anthropol. 45:281-298. Koenigswald, W von (1982) Enamel structure in the molars of Arvicolidae (Rodentia, Mammalia), a key to functional morphology and ph logeny. In B Kurten (ed.):Teeth: Form, Function anlEvolution. New York: Columbia University Press, pp. 109-122. Maas MC (1988)The Relationship of Enamel Microstructure and Microwear: An Experimental Study of Cause and Effect. PhD Thesis, State University of New York at Stony Brook. Ann Arbor: University Microfilms International. Maas MC (1989) Enamel microwear: An experimental study of cause and effect. Am. J . Phys. Anthropol. 78:264-265 (abstract). Martin LB, Boyde A, and Grine FE (1988) Enamel structure in primates: A review of scanning electron microscope studies. Scanning Microsc. 2:1503-1526. Mills JRE (1955) Ideal dental occlusion in the Primates. Dent. Pract. [Bristol Dent. Rec.] 6:47-61. Mills JRE (1973) Evolution in mastication in primates. In RM Zingeser (ed.):Craniofacial Biology of Primates. Volume 3. Symposium of the N t h International Congress on Primatology. Basel: Karger, pp, 65-81. Muhlemann HR (1964) Storage medium and enamel hardness. Helv. Odont. Acta 8:112-117. Peters CF (1982) Electron-optical microscopic study of incipient dental microdamage from ex erimental seed and bone crushing. Am. J. Phys. AntIropol. 57:283301. Pfretzschner H-U (1986) Structural reinforcement and crack propa ation in enamel. In DE Russell, J-P Santoro, and D gigogneau-Russell (eds.):Teeth Revisited: Proceedings of the VIIth International SvmDosium on Dental MGphology, Paris 1986. Mem. M k N a t . Hist. Nat., Paris (Serie C) 53:133-143. Powers JM,Craig RG, and Ludema KC (1973a) Frictional behavior and surface failure of human enamel. J . Dent. Res. 52:1327-1331. Powers JM, Ludema KC, and Craig RG (197313) Wear of fluora atite single crystals: VI. Influence of multiplepass siding on surface failure. J. Dent. Res. 52:10321040. Puech P-F, and Albertini H (1983) Usure des dents chez Australopithecus ufarensis: Examen a u microscope du complexe canine superieurelpremiere premolaire inferieure. C.R. Acad. Sci. Paris Serie. I11 296:10831088. Puech P, Cianfarani F, and Albertini H (1986) Dental microwear features a s an indicator for d a n t food in early hominids: A preliminary study of enamel. Hum. Evol. 1:507-515. Rasmussen ST, Patchin RE, Scott DB, and Heuer AH (1976)Fracture properties of human enamel and dentine. J . Dent. Res. 55:154-164. Rensberger JM (1978) Scanning electron microscopy of wear and occlusal events in some small herbivores. In PM Butler and KA Joysey (eds.): Development, Function and Evolution of Teeth. New York: Academic Press, pp. 415-438. Rensberger JM (1983) Effects of enamel structure on wear. Am. J. Phys. Anthropol. 60:243-244 (abstract). Rensberger JM, and Koenigswald W von (1980) Functional and phyletic interpretation of enamel microstructure in rhinoceroses. Paleobiology 6:477495. Ronnholm E (1962) The amelogenesis of human teeth as revealed by electron microscopy. 11. The development of the enamel crystallites. J. Ultrastruct. Res. 6:249303. Ryan AS (1979a) Wear striation direction on primate teeth: A scanning electron microscope examination. Am. J. Phys. Anthropol. 50:155-168. Ryan AS (1979b)A preliminary scanning electron microscope examination of wear striation direction on primate teeth. J. Dent. Res. 51:535-550. Ryan AS, and Johanson DC (1989) Anterior dental microwear in Australopithecus ufarensis: Comparisons with human and nonhuman primates. J. Hum. Evol. 18:235-268. Skobe 2, Proztak KS, and Trombly PL (1985) Scanning electron microscope study of cat and dog enamel structure. J. Morphol. 184:195-204. Sokal RR, and Rohlf FJ (1981) Biometry: The Princi les and Practice of Statistics in Biological Research. kecond Edition. New York: WH Freeman. Solounias N, Teaford M, and Walker A (19881 Interpreting the diet of extinct ruminants: The case of a nonbrowsing giraffid. Paleobiology 14:287-300. Stern D, Crompton AW, and Skobe 2 (1989) Enamel ultrastructure and masticatory function in molars of the American opossum Didelphis uirginiunu. Zool. J . Linn. Soc. 93:311-334. Stern D, and Skobe Z (1985) Individual variation in enamel structure of human mandibular first premolars. Am. J. Phys. Anthropol. 68:201-213. Teaford MF (1985) Molar microwear and diet in the genus Cebus. Am. J. Phys. Anthropol. 66:363-370. Teaford MF (1986) Dental microwear and diet in two species of Colobus. In J Else and P Lee (eds.):Proceedings of the Xth Congress of the International Primatological Society, Vol. 2. Primate Ecology and Conservation. Cambridge: Cambridge University Press, pp. 63-66. Teaford MF (1988)A review of dental microwear and diet in modern mammals. Scanning Microsc. 2:1149-1166. ENAMEL STRUCTURE AND MICROWEAR Teaford MF, and Oyen OJ (1989) In vivo and in vitro turnover in dental microwear. Am. J. Phys. Anthropol. 80:447-460. Teaford MF, and Robinson JA (1989) Seasonal or ecological differences in diet and molar microwear in Cebus nigriuittatus. Am. J. Phys. Anthropol. 8@391-401. Teaford MF, and Walker A (1984) Quantitative differences in dental microwear between primate species with different diets and a comment on the resumed diet of Siuupithecus. Am. J . Phys. Anthropof 64:191200. Walker A (1984) Mechanisms of honing in the male baboon canine. Am. J . Phys. Anthropol. 6547-60. 49 Walker A, Hoeck HN, and Perez L (1978) Microwear of mammalian teeth as an indicator of diet. Science 201:908-910. Walker A, and Teaford MF (1989) Inferences from quantitative analysis of dental microwear. Folia Primatol. 53:177-189. Wells RT, Horton DR, and Rogers P (1982) Thylacoleo curnifex Owen (Thylacoleonidae): Marsupial carnivore? In M Archer (ed.):Carnivorous Marsupials. Melbourne: Royal Zoological Society of New South Wales, pp. 573-586. Wright KHR (1968) The abrasive wear resistance of human dental tissues. Wear 14263-284.