Associations among modern pollen, vegetation, and climate in western North America

A compilation of 1884 modern pollen surface samples was analyzed to explore the relationships between spatial distributions of pollen percentage data vs. climate and vegetation in western North America. Modern pollen spectra capture many of the unique traits of regional vegetation patterns and reflect regional patterns of climate diversity. Large-scale vegetation differences were identifiable by their pollen signatures. At the coarsest scale, forested regions were dominated by arboreal pollen and Pinus pollen abundances were typically >30%. In contrast, non-forested regions were dominated by shrub and herbaceous pollen types with Pinus percentages typically <30%. At finer scales, indicator taxa were useful for identifying vegetation types. For example, Picea was a good first-order indicator of boreal spruce forest vegetation with relatively high median percentages (22%), whereas Quercus was a good first-order indicator of desert vegetation with low median percentages (1%). Pollen abundance also provided climatic information. High Betula percentages occurred over a relatively narrow climatic range corresponding with their high latitudinal distribution, high Artemisia percentages were registered in wintercold and dry climates of the interior basins. Arboreal pollen types were abundant in cool and wet climates, while non-arboreal pollen types dominate in climates that are warm and dry. Using pollen assemblages to predict local climate conditions shows that climate conditions are well predicted given knowledge of pollen spectra and climate conditions of the nearest analogues. Low accuracy was evident with pollen-based climate predictions in desert regions and temperate rainforests because of either poor sample density (in the case of the former) or extremes in temperature and/or precipitation (in the case of both). In other regions, vegetation and climate space were accurately predicted even when proximal samples, within 100 km, are excluded as the possible analogues. These results show that, within the vegetation and climate space of this region, the modern-analogue technique is useful for quantifying broad-scale vegetation and climate changes. 2008 Elsevier Ltd. All rights reserved.