USGS Home
Contact USGS
Search USGS

We established 30 random pairs (treatment and control) of 2.25 m² sample plots (hereafter referred to as plots) each year in 1988 and 1989. Before selecting the 30 pairs of plots, we conducted vegetation frequency counts on 30 randomly selected potential plots to obtain mean percent frequencies for all plant species. We used a 1.5 m × 1.5 m sampling frame containing 25 sampling points spaced 0.3 m apart. At each sampling point, we recorded either the plant species present or bare ground. Sample plot pairs were accepted for study only if the percent frequency for the most common vegetation species (prickle grass, Crypsis vaginiflora [Forsk.] Opiz; swamp timothy, Crypsis schoenoides [L.] Lam.; cudweed, Gnaphalium sp. L.; bearded sprangle-top, Leptochloa fascicularis [Lam.] Gray) was ± 2 SD (Hicks 1992).

Each year, the 30 pairs of plots were randomly separated into 3 groups of 10 pairs. All plots were fenced with covered 2.25 m² wire enclosures. A duck carcass was added to each treatment plot; its control plot located 10 m away did not receive a carcass. Duck carcasses were either frozen wild ducks (mallard, Anas platyrhynchos L.; northern pintail, A. acuta L.; northern shoveler, A. clypeata L.) that had been collected for a previous food habits study or frozen game-farm mallards. Esophagi of wild ducks had been removed, but the carcasses were otherwise intact. Fifteen of 49 game-farm mallards had been exposed to botulism during a concurrent study (Reed and Rocke 1992). One group of 10 pairs was established in each of 3 consecutive 29-day time periods from 10 August 1988 to 3 November 1988 and 8 August 1989 to 2 November 1989. The August to November time period was selected because it is during this time period that most waterfowl deaths due to avian botulism occur (Wobeser 1987). We used a 29-day sampling period because carcass decomposition was nearly complete in this time (only skin, bones, and feathers remained) and 29 days exceeds the generation time of many aquatic invertebrates under summer and fall conditions. For example, a chironomid can develop to an adult in as little as 2 weeks (Merritt and Cummins 1984).

Invertebrates were sampled at 7-day intervals during the 29-day period. During the initial sampling of each period (day 1), we sampled the center of the plots in each pair using a pelagic and benthos sampler (Euliss et al. 1992). The 4 acrylic tubes of the sampler collected a profile sample of water, vegetation, and substrate. The volume of the water column fraction ranged from 0.0 to 1,196 cm³ (including vegetation on the soil surface), and the benthic fraction volume was 98 cm³ per tube; the area sampled was 78.5 cm². Each sample consisted of the pooled data from the water column and benthic fractions of all 4 tubes. After sampling, a duck carcass was added to the center of the treatment plots. The duck carcass was removed on Day 22 after the sample was collected. Samples were collected approximately 0.4 m from the carcass/center of plot on the north, south, east, and west sides of the plots on Days 8, 15, 22, and 29, respectively.

Invertebrate samples were screened through 0.5 mm mesh in the field (Euliss and Swanson 1989) and frozen until processed in the laboratory, where samples were thawed and invertebrates were sorted into taxonomic groups, counted, and weighed. Invertebrates were identified to the lowest practical taxonomic level, usually genus, according to keys of Usinger (1956), Grodhaus (1967), Borror et al. (1981), Merritt and Cummins (1984), Pennak (1989), and Lauck (Humboldt State Univ., pers. comm.). Wet weights were determined on an analytical balance (±0.1 mg) after blotting to remove excess surface moisture.

We calculated the percent occurrence of aquatic invertebrate taxa in all samples (Appendix). Seventy-six of 83 taxa occurred in <2.5% of the samples, while the remaining seven taxa were present in >8.2% of the samples. We focused our analysis on the taxa present in >8.2% of the samples with the exception of the unidentified Chironomidae taxon. We omitted this taxon from the analysis because we questioned the validity of comparing a taxon containing a potentially large group of unidentified chironomid genera with chironomids identified to the generic level.

Counts and weights were analyzed with repeated-measures analysis of variance (ANOVA) after transformation (log₁₀(X + 1)) to determine whether these taxa differed among sampling days at the treatment and control plots (Milliken and Johnson 1984). An alpha level of 0.05 was used to determine statistical significance.