Northern Prairie Wildlife Research Center
Plat books were used to determine changes in the number of occupied farmsteads, number of different landowners, and average farm size since the early-1950s. In December 1991 and January 1992, we visited landowners on or near the study area to obtain permission to conduct research on their lands. Landowners living some distance from the study area were contacted by telephone.
Each quarter section of land was given an identification number from 1 to 42 (Fig 3). Upland habitats on the study area were mapped during 1992-93 to show current land use practices and were compared with land use during the early 1950s. Land use on each quarter section was verified and delineated on field maps. Landowners were contacted to verify certain practices and habitat components.
Annually tilled land consisted of corn (Zea mays), soybeans (Glycine max), wheat (Triticum), barley (Hordeum), oats (Avena sativa), rye (Secale cereale), buckwheat (Fagopyrum esculentum), alfalfa (Medicago sativa), and fallowed land. Scientific names of domestic grains follow Scott and Wasser (1980).
Other habitat types included Conservation Reserve Program (CRP) grasslands (highly erodible land taken out of crop production for 10 years and planted to native grasses), pastures, annually hayed grasslands, trees and shrubs, miscellaneous rock piles, junk piles, abandoned buildings, and wetlands. Field sizes were measured with a digitizing planimeter.
|Figure 3. The Waubay study area, Day County, South Dakota, divided into 42 quarter sections, 1992-1993.|
|Figure 4. Wetland numbering system, section 14, quarter 1, Waubay study area, Day County, South Dakota, 1992-1993. For details of Figures 3 and 4, see Evans and Black (1956).|
Wetlands in the study area were classified in 1992 according to Stewart and Kantrud (1971). Individual wetlands were identified and referred to by the same wetland numbering system that Evans and Black (1956) used (Fig 4).
Wetland vegetation was mapped according to emergent plant species that occupied 5% or more of a wetland basin. Species lists of aquatic macrophytes that occupied 5% or more of a basin were made for each wetland. Cover types (Stewart and Kantrud 1971) were also assigned to each wetland.
Wetlands were categorized as completely drained, partially drained, undrained, restored, or tilled.
Total drained area in each wetland class was measured from total original wetland area. Records of drainage dates were available for all basins drained on or before 1968 from earlier studies and surveys (Evans and Black 1956; USFWS files, Waubay National Wildlife Refuge, Waubay, S.D.). These data were used to analyze wetland losses over time.
We also noted wetland-basin condition (wet or dry), location, and size (hectares). Wetlands were considered wet if 10% or more of the basin contained standing water at least 2.5 cm deep.
Four surveys of wetland-basin condition were conducted annuallyin early May, late May, late June, and late July. Wetland size was measured with a digitizing planimeter on aerial photographs and field maps, and different sizes were grouped according to Evans and Black (1956). Frequency histograms were made of wetland classes and drainage histories.
Breeding Pair Counts. Two counts of breeding waterfowl pairs were made in May of each field season to provide an index of the number of ducks nesting on and around the study area. If properly timed, pair densities can be accurately estimated from only two counts (Higgins et al. 1992).
The first pair count was timed to coincide with onset of nesting by blue-winged teal (Anas discors), and the second pair count was timed to coincide with onset of nesting by gadwalls (A. strepera).
Breeding pairs, hereafter called pairs, were counted by a person walking around the perimeter of each wetland. Pairs in large wetlands choked with vegetation were counted by two people walking on opposite sides of the wetland until they met at the far end, where they compared notes and eliminated duplicate counts. Pairs on large, open-water wetlands were surveyed from a distant vantage point.
We attempted to avoid duplicate counts later in the survey by noting the flights of flushed ducks to other wetlands. In quarter sections with large wetlands and large numbers of ducks, smaller wetlands were counted first.
Only ducks flushed from counted wetlands were tabulated. Ducks flying over or landing in a wetland were not counted.
Pairs were counted between 0630 and 1800 hours. All counts were conducted by walking. Two to four people completed a count of ducks on all wetlands in the study area in 3 to 5 days.
During pair counts, all ducks were recorded on maps by specific location, species, and sex. At final tabulation, groups were segregated from pairs. Pairs, lone drakes, and lone hens of all species were tabulated for comparison with data from Evans and Black (1956). Groups of males and mixed groups of males and females were each tabulated as pairs when occurring in groups of "five or fewer males" except American wigeon (Anas americana) and northern shovelers (A. clypeata) for which only pairs and lone drakes were tabulated. Pairs were tabulated according to Hammond (1969).
An average of the two annual pair counts was used for analyses of pair densities by species. Relative use of different wetland classes and basin sizes by pairs was only calculated with data from 1993, which also were compared with similar data obtained by Evans and Black (1956).
Brood Counts. Two brood counts were conducted annually to obtain an index of duck production.
The first count was initiated when the first class IIa (Gollop and Marshall 1954) ducklings were seen in the immediate area, on approximately 25 June in both years. The second brood count was initiated approximately 24 July in both years.
All wetlands classified as wet were searched for broods. Broods in large wetlands choked with vegetation were counted by two or more people wading through the emergent cover in a zigzag pattern to drive broods or brood hens toward an observer at the opposite side of the wetland (Evans and Black 1956). Smaller wetlands were searched by one person. Broods in large, open-water wetlands were counted from a distant vantage point with binoculars and spotting scopes during early morning and evening hours.
Recorded data included species, number of ducklings, duckling age class (Gollop and Marshall 1954), the presence or absence of a hen, and the section, quarter, and wetland in which the brood was seen.
All broods were recorded during each count; however, only broods that hatched since the completion of the first brood count were used for tabulation of the second count. Incidental sightings of broods known to have hatched since the completion of the second count were added to the final number of broods each year.
Estimates of brood densities are reported as the number of broods per square kilometer. The annual estimate of total broods was the sum of flightless broods, hens which by their actions and calling indicated the presence of a brood, and incidental sightings.
Locations of brood sightings were an index of the use of different wetland classes and sizes by broods. Hen success (the number of broods per 100 pairs) of all species found in the study area was calculated for 1992 and 1993 and compared with Evans and Black (1956).
|Figure 5. Over-water canvasback nest in a seasonal wetland in the Waubay study area, Day County, South Dakota, 1993.|
Nest Searches: Over-Water Nests. Nest searches were conducted on 15 class-III and 40 class-IV wetlands from mid May to mid July 1992 and 1993 by systematically wading through emergent vegetation and looking for nesting platforms or for hens that flushed from nests (Fig 5). Because redhead (Aythya americana) and ruddy duck (Oxyura jamaicensis) hens never flushed directly from nests, we had to locate their clutches by finding their nesting platforms.
When found, a nest was numbered and marked with a small strip of white cloth tied to emergent vegetation 4.6 m north of the nest. Additional data were recorded on cards similar to those described by Klett et al. (1986) and included location (section and quarter number), pond number, species, upland or over-water nest, dominant nest site vegetation, nest status (occupied by hen or terminated), date, time, number of eggs from the host and from parasitic hens, the age of a clutch of eggs, nest initiation date (determined by summing the number of host eggs and their age when the nest was found and counting backward on the calendar to the date the first egg was laid), estimated hatch date (determined by estimating age of the host eggs and counting forward the number of calendar days needed to complete incubation), water depth, and distance to nearest shoreline. Distances to nearest shorelines were estimated to avoid creating paths through emergent vegetation that predators could follow.
The age of a clutch of eggs of most species was determined by candling (Weller 1956), but because of eggshell thickness, the ages of ruddy duck and giant Canada goose (Branta canadensis maxima) eggs were determined by flotation (Westerskov 1950). The length of the incubation period needed to hatch a clutch of eggs of each species followed Klett et al. (1986).
Nests were revisited at 2-week intervals when possible to determine the fate of each clutch and to estimate the number of exposure days (number of days each clutch of eggs was under observation and vulnerable to loss to predators and other decimating factors). A different path was taken to nests on subsequent visits to avoid establishment of permanent trails.
Additional data recorded on subsequent nest site visits included date, time, number of host eggs, age of eggs, clutch fate (whether one or more eggs in a nest hatched or whether the eggs were destroyed and the nest terminated), cause of nest loss (predation, flooding, machinery, investigator disturbance, etc.), evidence of hen mortality, number of unhatched host eggs and condition of eggs, and evidence of nest parasitism.
A successful clutch was defined as a clutch in which one or more eggs hatched. A clutch was considered unsuccessful if destroyed by predators, abandoned, or flooded.
A clutch destroyed by predators was characterized by missing eggs, eggshell fragments (other than from hatching), and visible nest disturbance. Clutches were considered abandoned if on subsequent visits all eggs were cold and the embryos were dead. A flooded nest was characterized by fully or partially submerged eggs or by nest platforms and clutches that had been completely washed away.
Clutch success rates were calculated by the Mayfield 40% method (Miller and Johnson 1978, Johnson 1979). Only nests that were occupied by a hen during egg laying or incubation were used in clutch success calculations. The ages of clutches at hatching that were used for Mayfield calculations were adjusted for the average clutch sizes in the study area.
Nest Searches: Upland Nests. Incidentally found upland nests were marked, recorded, and monitored until they were terminated. Predators which destroyed upland nests were identified according to Rearden (1951).
Most data that we collected from surveys and inventories were descriptive in nature. Our use of inferential statistical analyses had to be limited because only summary data (mean, %) from the final publication by Evans and Black (1956) were available for statistical comparison. Consequently, statistical inference was used to evaluate differences in pair and brood densities between 1950-53 and 1992-93.
We used a two-sample t-test in which a single observation is compared with a mean of a sample (Sokal and Rohlf 1981) to compare pair densities of species between years with similar weather patterns and wetland conditions. Means from the early 1950s were treated as densities taken from only one breeding pair count because variances could not be calculated without the raw data. Data from 1951 were compared with data from 1992, and data from 1953 were compared with data from 1993.
Comparisons of total pair densities and brood densities between 1950-53 and 1992-93, 1951 and 1992, and 1953 and 1993 were made with a two-sample paired differences t-test (McClave and Dietrich 1988). Differences were deemed significant at α = 0.10 for all statistical tests. Differences were deemed marginally significant at α = 0.20 to minimize the chance of making a Type II error.