Wearing down the constraints of low magnification tooth microwear analysis: reproducibility and variability of results based on extant ungulates

Abstract

Low magnification dental microwear analysis is a widespread dietary proxy for palaeoenvironmental analyses. The limitations of the method, such as observer bias or variation of microwear scars between different tooth positions, are still not quite understood. This study aims to reveal that reproducibility and variability of low magnification dental microwear is better, than it was previously thought. The main focuses of this study were differences between results produced by independent observers, and individual variability of the wear features on different teeth of the same specimen. To approach these issues, the microwear of 1944 0.4 × 0.4 mm areas on every right molar and premolar (144 teeth of 12 extant ungulate specimens) was quantified. Reproducibility and interobserver error was tested by calculating the intraclass correlation coefficients for the scores produced by the observers. The microwear features of each tooth were characterized by the mean, median, standard deviation, range, skewness and kurtosis. These statistical parameters were than compared. To test whether observed differences between the microwear patterns of different tooth positions are significant, ANOVA and Dunnett’s post hoc tests were performed. To calculate the minimal number of sampling sites required for characterizing a tooth, a computer-assisted bootstrap method was applied. As a result, it can be suggested that the low magnification microwear method is quite robust, with low interobserver error. The variance of microwear scars seems uniform throughout the dentition of the examined specimens. Some differences can be noted between tooth positions, however, some limitations could be lifted, at least in the case of ungulates.