Identification of Picea pollen of Late Quaternary age in eastern North America: a numerical approach

Abstract

Five collections of modern Picea glauca pollen, four of P. mariana, and two of P. rubens were examined in an attempt to distinguish the pollen of the species. Three morphological variables were recorded for individual grains in each collection. The sculpturing of the furrow membrane is not diagnostic, but the presence of an undulating margin to the cap and of irregular reticulation in the sacci characterize P. rubens pollen. Seven size variables were also measured for each grain. No simple combination of morphological and size criteria provides effective discrimination between P. glauca and P. mariana pollen.Linear discriminant analysis was applied to the size data for P. glauca and P. mariana. The mathematical assumptions of the method were tested, and to satisfy them, the discriminant analysis was confined to six of the seven variables measured. The discriminant analysis resulted in 91.5% of the pollen being correctly distinguished.The same six variables were measured on fossil Picea pollen from two Late Wisconsin sites in Minnesota and one Holocene sequence in Labrador. The individual fossil grains were assigned to either P. glauca or P. mariana by means of discriminant analysis. The Late Wisconsin spectra consist of both species, a result supported by macrofossil evidence. Picea glauca predominates in the early Holocene spectra from Labrador. It is replaced by P. mariana after about 5000 years before present, reflecting paludification and the spread of muskeg.The limitations of the procedure are discussed, and its potential is emphasized.