Error rates in dental microwear quantification using scanning electron microscopy

There is a degree of correlation between dietary habits and dental microwear in extant primates, and this has enabled inferences to be made about prehistoric diets. Several techniques have been used to quantify microwear, but the comparability of results derived from each has not been demonstrated. Moreover, neither intra- nor interobserver error rates in microwear quantification have been documented to date. We here assess intra- and interobserver error using Microware 4.0, and evaluate intertechnique comparability using the three methods that have been most widely employed in the field. This study documents an overall intraobserver error rate of about 7%, and an overall interobserver error rate of some 9%. Both intra- and interobserver error appears to be influenced substantially by the nature of the micrograph being measured. In no instance did the results obtained by different observers using Microware 4.0 differ significantly, and there was a reasonable degree of interobserver consistency in the rank ordering of micrographs in relation to any given parameter. The results obtained through the use of different quantitative techniques differed significantly, with an overall intertechnique error rate of approximately 19%. Several variables, including differences in magnification factor, scanning electron microscope kV settings, and specimen-detector relationships undoubtedly contribute to the differences among the three methods, but we were not able to assess their relative importance. Microwear quantification permits distinctions between broad dietary categories, but the margin of intra- and interobserver error should be taken into account when defining pattern differences between populations (or species) or when documenting seasonally mitigated differences within a taxon. In view of the error introduced by the use of different methods, we suggest that a consistent technique, such as offered by the Microware software package, be adopted by current researchers to establish a common microwear database.