USGS Home
Contact USGS
Search USGS

Field Samples.  We identified 46 categories of invertebrate remains in our field samples (Table 1), 6 of which were used in our simplified taxonomy (Table 2). Average values of taxon richness and LogCount for field samples differed markedly among the three habitat types (upland, temporary wetland, seasonal wetland) regardless of whether or not weighting was done or if the simplified taxonomy was used (Table 3). When unweighted, taxon richness (F_2,60 = 26.21; P < 0.0001) differed among the three habitat types with both wetland types being similar to each other (P = 0.34) but not (P < 0.0001) to the upland. The unweighted LogCount (F_2,60 = 17.85; P < 0.0001) also differed by habitat type with wetland types differing from upland (P < 0.0001) and marginally from each other (P = 0.058). Taxon richness for the weighted samples also differed among habitat types (F_2,60 = 32.41; P < 0.0001), with both wetland types being fairly similar to each other (P >0.09) but not to the upland (P < 0.0001). The weighted LogCounts also differed by habitat type (F_2,60 = 21.06; P < 0.0001), but each was statistically unique (P < 0.03). Using the simplified taxonomy, each of the habitat types were statistically unique, with both taxon richness (F_2,60 = 26.93; P < 0.0001) and LogCount (F_2,60 = 26.04; P < 0.0001) differing among the three habitat types.

Incubated Samples.  We successfully incubated invertebrates representing 15 taxa (Table 4), which we grouped into 5 taxa in our simplified taxonomy (Table 5). For our incubated samples, we found that the differences in taxon richness and LogCount among the three habitat types (upland, temporary wetland, seasonal wetland) were even greater than found for the field samples, regardless of whether or not weighting was done or the simplified taxonomy used (Table 6). Using unweighted data, the three habitat types differed from one another for both taxon richness (F_2,60 = 60.23; P < 0.0001) and LogCount (F_2,60 = 42.33; P < 0.0001). Similarly for the weighted samples, we found that both taxon richness (F_2,60 = 60.56; P < 0.0001) and LogCount (F_2,60 = 42.19; P < 0.0001) differed among habitat types. In all cases for both weighted and unweighted samples, each of the habitat types was statistically unique (P < 0.05). The results for the analysis using the simplified taxonomy were similar for taxon richness and LogCount, with a large difference among habitat types for LogCount (F_2,60 = 48.26; P < 0.0001) and each of the habitat types being distinctive (P < 0.05).

Figure 1.   Relationship of LogCount and the number of invertebrate taxa, weighted by habitat affinity, for visually derived field data using complete taxonomy for 62 intensively farmed temporary and seasonal wetlands in the Prairie Pothole Region. The line, weighted LogCount = 3-4(weighted taxon richness), separated wetland and upland sites with the minimal number of classification errors (21.8%). Thirty upland observations are hidden in the lower, lefthand corner.

Figure 2.   Relationship of LogCount and the number of invertebrate taxa, weighted by habitat affinity, for visually derived field data using complete taxonomy for 62 intensively farmed temporary and seasonal wetlands in the Prairie Pothole Region. Twenty-five upland sites where samples were incorrectly collected from within wetland areas are identified as wetland sites. Thirty upland observations are hidden in the lower, lefthand corner.

Classification of Sites

For the aquarium data, we found satisfactory the simple rule that classified sites as wetland if either number of taxa > 0 or LogCount > 0. This rule, with the complete taxonomy, yielded seven wetland sites (six temporary and one seasonal) classified as upland, and seven upland sites classified as wetland, for an overall error rate of 14/124 = 11.3%. With the simplified taxonomy, the rule misclassified the same sites as with the complete taxonomy, as well as an additional seasonal wetland site that was misclassified as upland; the overall error rate was 15/124 = 12.1%.