This study examined molar teeth of gorillas, G. gorilla, housed at the Field Museum of Natural History in Chicago, Illinois, USA. This taxon was chosen because of the large body of work focused on primate dental functional anatomy and because gorilla teeth evince substantial visible morphological change with wear. Kay (1981) has speculated that high cusps and thin tooth enamel in primate folivores, such as the gorilla, would lead to sharp edges at the sites of dentine exposure, which would improve shredding and slicing abilities of the tooth with wear. Methods described here were designed to examine such aspects of the morphology of worn teeth.

Five variably worn lower second molars were selected to include a range of wear from unworn to extremely worn (Figure 1). High-resolution replicas were prepared as follows. First, dental impressions were taken using a polyvinylsiloxane putty (President's Jet, Coltene, Inc.). Molds were allowed to harden, and casts were prepared using Epotek 301 resin and hardener (Epoxy Technologies, Inc.). This procedure has been demonstrated to produce casts with resolutions to less than one micron (Beynon 1987; Ungar 1996), more than sufficient for detailed analyses of the occlusal surfaces. Tooth replicas were then coated with a thin layer of Magniflux Spotcheck SD-S2 Developer (Illinois Toolworks, Inc.) to mitigate cast translucency.

Data collection and preparation for analysis involved several steps. First, three-dimensional point data representing the occlusal surface of each tooth were collected. Data points for individual teeth were then aligned and scaled and imported into the GRASS 4.1 (U.S. Army Construction Engineering Laboratory) GIS package as a digital elevation model (DEM). Each DEM was then regridded and cropped for analysis.

We collected point data using a modified Surveyor 500 laser scanner with an RPS 450 laser (Laser Design, Inc.) (Figure 2). The scanner has a maximum resolution of 0.0254 mm in x, y, and z dimensions and a maximum work envelope of 152.4 mm x 152.4 mm x 304.8 mm. The laser can scan the occlusal table of a gorilla molar and record hundreds of thousands of data points representing that surface in just a few minutes. In this study, we created a DEM from points sampled at an interval of 0.0508 mm. This resolution preserves subtle aspects of occlusal morphology, yet limits data files to an easily manageable size.

This approach to data collection differs from those described by other authors. Reed (1997) suggested using a reflex microscope (Reflex Measurement Ltd.) to collect coordinate data for teeth. The reflex microscope is, however, an impractical tool for dental topographic analysis. It requires the researcher to identify each individual point on a tooth's surface, an extremely tedious and time-consuming endeavor when hundreds if not hundreds of thousands of points are needed to adequately characterize a tooth's surface. In another study, Zuccotti et al. (1998) suggested that a 3Draw Pro (Polhemus Corp.) electromagnetic digitizer might be used to collect such data. In that case, a stylus is passed over the occlusal surface, and points are collected at a rate of 70 per minute. This procedure is impractical for smaller mammalian teeth (including most primates) because the resolution of this digitizer is only 0.13 mm. The MicroScribe-3D (Immersion Corp.) would have a similar limitation, with a published accuracy of 0.38 mm (0.23 mm for the 3DX model). Finally, Jernvall and Selänne (1999) suggested the use of a confocal microscope. These authors used a Zeiss Axiovert 135M microscope with the BioRad MRC-1024 confocal system and an American Laser Corporation 60WL argon/krypton laser . This approach is very effective for digitizing small teeth, but is currently of limited use in dental topographic analyses because the maximal working envelope for the system described by Jernvall and Selänne restricts tooth sizes to less than 10 mm in diameter, somewhat smaller than many mammal teeth. The laser scanner presents a good compromise between work envelope and resolution because it is capable of collecting data for all but the smallest mammal teeth. (We have resolved occlusal morphology on bat teeth less than 1 mm in diameter.) Laser scanning can quickly generate a large dataset of points representing the occlusal surfaces of most mammalian molar teeth.

Once data were collected, individual DEMs had to be aligned and scaled so that measurements would be consistent among specimens. This was accomplished using three points: the lowest points on the anterior and posterior foveae and the junction between crests connecting the metaconid and entoconid (Figure 3). These landmarks were chosen because they are the lowest points consistently identifiable on the occlusal surface, thereby allowing inclusion of the most worn teeth in this study. Data Sculpt software (Laser Design, Inc.) was used to align these three points to the x-y plane, with the x-axis passing through the two foveae. Each tooth was then scaled to the mean distance separating pairs of landmarks on that tooth. The aligned and scaled DEMs were exported in ASCII data files as x, y, z data points to the GRASS 4.1 (U.S. Army Construction Engineering Laboratory) GIS software package.

Because rotational alignment of data results in an irregular matrix, inverse-distance weighting was used to re-grid the coordinates to a regularly spaced surface model. In order to further assure the comparability of specimens, each DEM was cropped to the horizontal plane intersecting the lowest point on the occlusal surface. This also eliminated the problem of surface overhang (more than one z-value for an x-y pair), such as often occurs down the buccal and lingual sides of the tooth as the enamel cap approaches the cervix. Individual cusps were delineated on the basis of contour lines (isometric lines connecting points of identical elevation). The areas defined by individual cusps were used as masks to filter out data from other parts of the tooth. This allowed separate consideration of data for each cusp.

Once the DEM was aligned, scaled, and cropped, we used standard GIS algorithms to compute descriptive statistics for each cusp. These included average slope, average delta slope (the second derivative of elevation), topographic aspect or surface orientation (in thirty-degree increments), surface area, and a relief index. The relief index is a ratio of three-dimensional surface area to two-dimensional (x, y) surface area. We also calculated the volume of fluid required to "fill" the central basin of each tooth to the point that it would overflow. This is a useful and repeatable measure of basin volume that reflects the morphology of the tooth.