Detecting a species limit from pollen in sediments

Pollen percentages in presettlement lake sediments from Michigan, Wisconsin and Minnesota were compared with the range maps of eastern hemlock (Tsuga canadensis (L.) Carr.) and American beech (Fagus grandifolia Ehrh.) in the early nineteenth century, compiled from the bearing-tree records of the U.S. Public land Survey. The species limit shown in the land-survey records corresponds within a few kilometres to the limit of living trees. Pollen percentages for both species in presettlement lake sediment fall to low levels 15-20 km beyond the land survey range limit. Percentages are variable within 10 km of the species limit, and therefore the position of the limit cannot be determined from pollen percentages at a single site. With a grid of fossil sites, however, it is possible to identify an area 20 km wide that includes the past species limit. On the basis of our data from the Great Lakes region, the past existence of outlying colonies of hemlock and beech can be demonstrated if pollen percentages at a fossil site are consistently higher than at surrounding sites, but it is impossible to prove from pollen that outlying colonies do not exist. Although regression models lead to similar conclusions, information about pollen percentages outside the species limit provided by y-intecept values is less precise than information from a geographical grid of sites.