Northern Prairie Wildlife Research Center
Interpreting biological significance of effect sizes like ours is important for future studies but requires consideration of minimally significant biological differences (MSBD95) (Muller and Benignus 1992, Gerard et al. 1998). Ideally, any sampler that increases information gain would be favored. Yet, use of our SATs is likely to be more expensive and time-consuming than conventional ATs. Thus, in some situations, small information gains with SATs may be biologically unjustified. We suggest that increased expense and effort required to use SATs may be warranted when relative abundance estimates are likely to be at least double that of ATs; that is, the lower bound on the interval estimates of the improvement ratio must be ≥2.0. Detection errors (i.e., declaring a taxon absent when it is not), on the other hand, are likely to have severe implications; resulting errors may be more serious than underestimating relative abundance. Thus, we suggest that investigators consider a more conservative value, perhaps 5%, for differences in non-detection rates. For example, we recommend that for assessing presence/absence of Dytiscidae and Haliplidae, ATs may serve as well as our more expensive SATs (Table 1; lower limit of confidence intervals <5.0%). For other taxa we considered, we believe that SATs are more useful for determining presence/absence. When comparing relative abundance of Dystiscidae (improvement ratios), ATs apparently perform nearly as well as SATs (Table 2; lower limit of confidence intervals ≤2.0). For all other taxa we assessed, relative abundance was best determined using SATs. We emphasize that these recommendations are based on our subjective MSBD95 values. Depending on requirements of specific studies, investigators should apply their own MSBD95 criteria. However, data in Tables 1 and 2 will be useful in deciding which trap to use, no matter what level is specified.
Aquatic invertebrate abundance in wetlands remains difficult to assess, even though interest in these organisms as food for waterfowl has been a key issue guiding development of wetland sampling methods. Due to their small size and high buoyancy, young (age 1-5 days) ducklings forage mostly on surface-associated invertebrates such as emerging insects and pupae (chironomids, culicids, etc.). Later (age approximately 5-15 days), duckling diets broaden to include subsurface forms such as various crustaceans, insect larvae, gastropods, and others (Chura 1961, Perret 1962, Bartonek and Hickey 1969, Bartonek and Murdy 1970, Swanson and Sargent 1972, Sugden 1973, Swanson 1977, Danell and Sj÷berg 1980, Sj÷berg and Danell 1982). We believe that our SAT will be useful for sampling the zone where most ducklings forage and for simultaneously collecting surface-associated, nektonic, and planktonic invertebrates, thus providing improved estimates of invertebrates available to foraging ducklings. Because timing of duckling foraging may be difficult to predict (Ringelman and Flake 1980), opportunity for collecting time-integrated samples using SATs (e.g., over a 24-hr period) may also be advantageous for some study objectives.
Invertebrate distribution and movements in prairie wetlands are not well understood. Corkum (1984) documented seasonal horizontal movements of invertebrates in a semipermanent wetland in central Alberta, Canada, but we are not aware of published reports assessing potential vertical distributions of invertebrates. Our SAT may be useful for assessing distribution and movements of invertebrates in prairie wetlands and other lentic habitats, even though we observed differences in vertical capture rates only occasionally and only for Dytiscidae, Hydrophilidae, and Notonectidae. Alternatively, we suggest that future investigators consider modifying the SAT such that each trap consists of a single chamber (in contrast to three strata). Resulting traps would be easier to use, less expensive (construction costs similar to ATs), and would yield nearly as much information as traps constructed after our original stratified design. In either case, it is plausible that the surface portion of our SAT may capture insects from uplands or adjacent wetlands; thus, investigators should be mindful of this when interpreting catches from the SAT.
Finally, we expected that differences in trap performance would be less evident in wetlands with fathead minnows since predation by minnows sharply reduces abundance, biomass, and taxon richness of many aquatic invertebrates (Hanson and Riggs 1995, Hanson et al. 1995). Suppression of cladocera in wetlands with high densities of fish probably made it more difficult to detect differences in improvement ratios based on data from our study wetlands (Table 2). Yet, for most ecological studies, SATs would still be advantageous because they are much more likely to detect presence of cladocera even when density of these organisms is relatively low in wetlands with fish (Table 1).