MULTIPLE-CORRESPONDENCE ANALYSIS OPTICAL MICROSCOPY FOR DETERMINATION OF STARCH GRANULES

Raw starch is composed botanically of characteristic granules of various sizes and shapes, so that each kind of starch may be characterized by the population of its granules. In the present study ten commercial starch species were studied: wheat, rice, manioc, potato, arrowroot, amylomaize, normal maize, waxy maize and two different banana species. Six variables measuring the size and shape of granules were obtained by image analysis. The objective was to find a method to describe and compare the granule populations of the ten species. For such a study, multiple-correspondence analysis (MCA) was applied. MCA makes it possible to draw similarity maps of categories and objects. For each starch species the frequency distributions (histograms) of the six variables were assessed and each granule was characterized by its species and the classes of histograms to which it belonged. MCA was applied to the granule table and a description of the histogram classes and the granules was obtained. From the variables description a general typology of the granules was deduced. The similarity maps showed considerable scatter of the granules for all species except rice. A particular species could therefore not be identified by a single granule, but the granule distribution seemed to be characteristic. MCA was an appropriate method to analyse these data because it points out non-linear relationships between quantitative and qualitative variables.