Morphotypological analysis of starch granules through discriminant method and its application in plant archeological samples

Abstract

Starch granule analysis effectively recovers microbotanical residues of starchy plants from archaeological contexts. Morphometric analysis is a common method to identify starch granules. However, this technique is time-consuming and inaccurate. Morphotypological differences in plants may also cause inconsistent results during microscopic observations. To address this problem, we evaluated the morphotypological features of three types of starch granules, specifically those found in wheat, millet, and yam, through light microscopy. Moreover, we used morphotypological analysis and discriminant analysis by ImageJ and SPSS software to perform computer-assisted analysis on the data set of geometric characteristics of the three types of starch granules, as well as the starch granule fossils from the Xianrendong and Diaotonghuan archaeological sites. Results are detailed and comparable to the light microscopy findings. The improved method with ImageJ and SPSS software required less time by direct measurement with a light microscopy and enhanced the accuracy of starch granule identification with the reduction of subjectivity. Therefore, the combination of ImageJ and SPSS software is a promising technique for the morphotypological analysis and identification of starch granules in archaeological starch granule analysis.