Palynological applications of principal component and cluster analyses

Two multivariate statistical methods are suggested to help describe patterns in pollen data that result from changes in the relative frequencies of pollen types produced by past climatic and environmental variations. These methods, based on a geometric model, compare samples by use of the product-moment correlation coefficient computed from data subjected to a centering transformation. If there are m samples and n pollen types, then the data can be regarded as a set of m points in an n-dimensional space. The first method, cluster analysis, produces a dendrograph or clustering tree in which samples are grouped with other samples on the basis of their similarity to each other. The second method, principal component analysis, produces a set of variates that arc linear combinations of the pollen samples, are uncorrelated with each other, and best describe the data using a minimum number of dimensions. This method is useful in reducing the dimensionality of data sets. A further transformation known as varimax rotation acts on a subset of the principal components to make them easier to interpret. Both methods offer the advantages of reproducibility of results and speed in pattern description. Once the patterns in the data have been described, however, they must be interpreted by the palynologist. An application of the methods in palynology is shown by using data from Osgood Swamp, Calif.