POLLEN SOURCE AREA AND POLLEN PRODUCTIVITY - EVIDENCE FROM FOREST HOLLOWS

1 Although fossil pollen from forest hollows (approximate to 5m in diameter) is often interpreted as a record of stand-scale forest changes, this assumption has not previously been tested by comparing pollen percentages in modern sediment to vegetation from a range of sampling radii. I compared pollen in surface sediment with distance-weighted basal area of trees in radii from 10 to 100m, to test a model (Sugita 1994) that predicts the 'relevant pollen source area' for forest hollows is 50-100m, and that about 40% of the pollen comes from trees growing within this radius. 2 Maximum likelihood estimates of pollen productivity, based on the relevant source area, are used to test for differences in species pollen productivity between a pine-dominated region in north-western Wisconsin, and a hemlock-northern-hardwoods-dominated region of upper peninsula Michigan. 3 Likelihood function scores, which measure the goodness-of-fit of the pollen-vegetation relationship based on the maximum likelihood method, show that estimates of pollen productivity (alpha) and relative background pollen loading (omega) improve as radius of distance-weighted vegetation sampling increases from 10 to 50m. There is little further improvement from 50 to 100m. Linear regression analysis of pollen percentages on distance-weighted basal area also shows little improvement in pollen-vegetation correlation when vegetation samples from beyond 50m are included. Predictions of pollen percentages therefore are not improved by increasing the vegetation sampling radius beyond 50-100m. These results support Sugita's (1994) model prediction and confirm that fossil pollen in hollows records stand-scale vegetation heterogeneity at a scale of 50-100m. 4 Estimates of background pollen (w) indicate that on average, only 40-50% of the pollen in forest hollows comes from trees growing within 50-100m of the hollow. However, this is sufficient to record the stand scale vegetation heterogeneity because larger areas of vegetation are recorded as homogeneous regional background, even in patchy vegetation. 5 Relative pollen productivity of three monospecific pollen taxa (red maple, sugar maple, basswood) is similar for two regions, supporting the implicit assumption in palynology that pollen productivity is a species-specific constant. Productivity estimates of birch are similar between regions, although yellow birch dominates in one region and paper birch in the other.