Testing methods to produce landscape-scale presettlement vegetation maps from the US public land survey records

The U.S. Public Land Survey (PLS) notebooks are one of the best records of the pre-European settlement landscape and are widely used to recreate prese ttlement vegetation maps. The purpose of this study was to evaluate the rel ative ability of several interpolation techniques to map this vegetation, a s sampled by the PLS surveyors, at the landscape level. Field data from Syl vania Wilderness Area, MI (U.S.A.), sampled at the same scale as the PLS da ta, were used for this test. Sylvania is comprised of a forested landscape similar to that present during presettlement times. Data were analyzed usin g two Arc/Info interpolation processes and indicator kriging. The resulting maps were compared to a 'correct' map of Sylvania, which was classified fr om aerial photographs. We found that while the interpolation methods used a ccurately estimated the relative forest composition of the landscape and th e order of dominance of different vegetation types, they were unable to acc urately estimate the actual area occupied by each vegetation type. Nor were any of the methods we tested able to recreate the landscape patterns found in the natural landscape. The most likely cause for these inabilities is t he scale at which the field data (and hence the PLS data) were recorded. Th erefore, these interpolation methods should not be used with the PLS data t o recreate pre-European settlement vegetation at small scales (e.g., less t han several townships or areas < 10(4) ha). Recommendations are given for w ays to increase the accuracy of these vegetation maps.