CAN PALEOECOLOGICAL PROXIES ADEQUATELY FORECAST SURFACE COVER ACROSS NORTH CENTRAL NORTH AMERICA? (A CAUTIONARY TALE)

Pollen stratigraphy provides the core body of paleoecological perspective across north central North America. Pollen based methodologies have matured over nearly a century and are now joined by a vast and vetted literature focused on paleolimnology, wetland dynamics, soil stratigraphy and landscape ecology. The COHMAP project of the 1970s and 80s placed these data sets in spatial context across the continent and more recent updates and refinements of these maps have been developed. The utility and importance of paleoecological models are more apparent given recognition of anthropogenic climate forcing. Concurrent with their adoption and broad use in discussions of such forcing, however, caveats and limitations of paleoecological data, prominently voiced by their authors, must be honored. Linda Brubaker, a pioneer in the application of pollen stratigraphy to Upper Michigan, recognized in 1975 that patchy habitats often impose permanent control on vegetation pattern on varying spatial scales. The scale of patchiness varies across landscapes and greatly complicates their classification as well as any assessment of their projected behavior. In broad and roughly uniform landscapes these complications may be minimal; in patchy landscapes, they may defy intuition and interpretation. As Brubaker pointed out, some of the more obvious drivers of patchiness in the upper Midwest are differences in soil texture and development and in the nature of regolith. These differences are invisible in broad pattern but detectable on a local level. Pollen-based maps of surface cover in eastern Upper Michigan prove inadequate in detecting the major climatic anomaly that drove hydrologic closure of the Upper Great Lakes ~10-8 ka. In contrast, an intensive study of 2 ponds within a small sandy tract in Alger County clearly identifies this and other major droughts at that time, albeit on different scales. Maps of Quaternary vegetation at sub-continental scales are useful across uniform environments but may fail to portray important local and even regional differences. This has long been recognized and expressed by pollen practitioners but may be glossed over by modelers who need to summarize projected change across the continent.