Northern Prairie Wildlife Research Center
In the summer of 1994, we revisited each of the sites misclassified by our techniques. In all 25 situations where we falsely classified upland sites as wetland, we found that errors in our interpretation of the surrounding landscape were responsible for the misclassifications. In 23 sites, we found specific topographic features (e.g., road beds functioning as dams) that, during periods of high water, would allow the wetland to flood the area where we collected our samples; thus these "upland" samples were actually collected from within the wetlands' boundaries. In the other two sites, adjacent wetlands would periodically backflood the wetlands sampled by backing water up and over the elevation at which the sample wetland would normally overflow. Wetlands throughout much of the semi-arid PPR lack surface outlets (Eisenlohr 1972), but adjacent basins often do coalesce, especially during periods of high water as occurred following the historic 1993 flood throughout the PPR. However, most of the misclassifications of upland sites (i.e., upland samples containing aquatic invertebrate remains) were due to secondary ponding from elevated road beds, which are numerous throughout the PPR.
Our 1994 visit to the misclassified sites occurred after the historic precipitation event in 1993, which refilled wetlands to record depths throughout the PPR and provided us with a much better view of wetland boundaries, the topographic relationships between adjacent basins that caused them to coalesce, and the impact of elevated road beds that functioned as dams and impounded water above historic pool levels. Had we correctly interpreted the topographic surroundings of each of the sites we sampled, our error rate would have been much lower. Hence, we believe that our methods would be more accurate than other available wetland indicators. The technique should work better in less disturbed habitats and in wetlands with more permanent water regimes because they contain more perennial and diverse invertebrate communities (Driver 1977, Euliss et al. 1999). While other currently available indicators are usually adequate for identifying wetlands when they contain water, invertebrate remains provide additional information, which should strengthen identification decisions.
The study also suggests that simple visual examination of soil samples collected from basins using a simplified taxonomy is often sufficient to identify most temporary and seasonal wetlands in agricultural fields. However, in cases where wetlands cannot be positively identified, incubation of samples or the use of a more detailed taxonomy may be used to corroborate or refute results obtained with other methods. Incubation of the samples in aquaria, although logistically more challenging than visual inspection, provided the clearest and most straightforward separation of wetland and upland sites. The rule was extremely simple: if aquatic invertebrates were cultivated in the aquarium incubations, assign the site as wetland. Using this rule, we failed to identify only 7 out of 62 wetlands. However, even those apparent misclassifications provided valuable information on the classification and appear to reflect accurately the history of each specific site. The absence of viable aquatic invertebrate eggs in the soil samples may suggest that the wetland has ceased to function as a wetland. Four of the wetland sites that were classified as upland (i.e., wetland samples from which we were unable to culture aquatic invertebrates in the aquarium incubations) turned out to be drained wetlands from which we found recalcitrant remains of aquatic invertebrates in the visual evaluation. However, these seemingly contradictory results actually contrast the current versus historic condition and function of each of the drained wetlands; they contained invertebrate signatures detected in the visual field data indicating that they were once wetlands, but they apparently no longer contained viable invertebrate eggs and hence appear to function currently as uplands. In this study, we found no evidence to suggest that intensive land use reduced our ability to identify wetlands within agricultural landscapes, but Euliss and Mushet (1999) found that intensive farming reduced numbers of aquatic invertebrate eggs in temporary wetlands.
Incubation of samples in aquaria failed to separate only 2 sites where obvious depressions existed. Both basins were dry during the 1994 revisit despite the unusually wet conditions throughout the PPR; wetlands adjacent to one of the misclassified sites were fully ponded, whereas the wetlands adjacent to the other dry site that was misclassified were also dry. While not evaluated in our study, sand lenses can provide natural drains in basins. Sand lenses are stringers of sandy material deposited from glacial melting that provide an underground conduit for water flow. Sand lenses are common in glacial till and are known to contribute to differences in ground-water recharge among wetlands (Swanson 1990). The remaining misclassified wetlands had been drained (either by subsurface tile or by shallow ditching), which had not been noted by the field crew when sites initially were sampled. Interestingly, all of these wetlands were classified as wetlands using the field data.
Since definitions of wetlands involve the delineation of distinct cutoff points of variables that occur along gradients (e.g., soils never saturated to soils always saturated, 0% hydrophytes to 100% hydrophytes) (U.S. Army Corps of Engineers 1987), the wetlands most difficult to identify will be those with features falling closest to these cutoff points; in the PPR these are the temporary and seasonal wetlands. Because conventional wetland indicators are of limited value when basins are intensively farmed, aquatic invertebrates may offer the most practical and cost-effective tool to identify farmed wetlands.