Identifying forest stand types using pollen from forest hollows

Methods of interpreting pollen assemblages in sediment were examined using surface samples from 66 small forest hollows in Michigan and Wisconsin. All canopy trees in the surrounding 50 m were measured to provide detailed information about the source vegetation of each surface pollen assemblage. Basal area of trees in each forest sample was used to classify them into sis stand types: hemlock-dominated. sugar maple/hemlock mixed, sugar maple-dominated, and ash-, oak-. and pine-dominated stands. Various statistical procedures were tested to learn which was most successful in sorting the pollen assemblages into appropriate vegetation groups. Two ordination techniques - detrended correspondence analysis (DCA) and canonical variate analysis (CVA) - give similar overall results, although CVA more successfully separated assemblages from hemlock stands from those of sugar maple-dominated stands. Squared chord distance <0.05 also successfully identified samples from the same forest type. After stand dominants have been identified fi-om pollen assemblages using multivariate methods. a further determination of stand composition is sometimes possible using ratios of pollen counts of individual taxa. Ratios can be calibrated by comparison with species abundances around surface samples. For instance, ratios of pine to hemlock pollen can indicate the abundance of pine within a stand dominated by hemlock, whereas pine pollen percentages alone are affected by variable abundance of other species.