Pollen in sediments is used extensively as a record of vegetation chance over time. A variety of methods have been developed that apply to different spatial resolutions, with a strong emphasis in recent years on modeling designed to provide quantitative estimates of regional vegetation. Our research, however, emphasizes spatial heterogeneity at landscape scales (10s of km), which requires different methods and assumptions. For instance, for regional reconstructions it is reasonable to assume that nearby sites contain similar vegetation, and information can be “borrowed” across time and space, essentially averaging out differences between adjoining sites (STEPPS model). When the goal is maximizing spatial resolution the variability between adjoining sites is assumed to be potential signal in identifying heterogeneity.
We use PLS vegetation data to classify 6 vegetation types surrounding 31 small (<10 ha), deep (>8 m) lakes on the NW Wisconsin sand plain. Presettlement pollen samples from the 31 sites were then used to develop an identification key for the 6 vegetation types using a combination of squared chord distance analogs and ratios of pollen types. We tested our key with comparisons of 8 pairs of sites ranging from 8 to 18 km apart to determine if the method could correctly identify vegetation types from nearby sites.
Results/Conclusions
Our key correctly identified the presettlement vegetation types of 28 of the 31 sites. Comparison of 8 pairs of sites within 18 km of each other successfully identified both sites in 4 pairs that had the same vegetation type, and also 4 pairs that had different vegetation types.
This result supports the hypothesis that lakes of less than 10 ha can be interpreted as local vegetation records when adequate local calibration data is available. We emphasize that calibration data must be from a network of similarly sized sites, and in similar regional vegetation. Pollen from regional vegetation dominated by well-dispersed, abundant pollen types like pine and oak would be expected to record less spatial resolution than any other forest type in eastern N. America, and finer spatial resolution may be possible in other regions. Our results demonstrate that there is more spatial resolution possible from pollen in sediments than is generally recognized.