Northern Prairie Wildlife Research Center
|Figure 1. Counties selected for surveys of breeding birds in Conservation Reserve Program fields. B - Butte County, SD; D - Day County, SD; E - Eddy County, ND; F - Fallon County, MT; G - Grant County, MN; H - Hettinger County, ND; K - Kidder County, ND; M - McPherson County, SD; S - Sheridan County, MT.|
We counted birds on 240 fields totaling 4654 ha in 1990, 335 fields encompassing 6181 ha in 1991, and 371 fields covering 6985 ha in 1992 (Table 1). Many of the same fields were surveyed in two or three years, so that future analyses could consider successional changes in the habitat. Fields surveyed more than once are multiply represented in Table 1.
From files in the ASCS offices we also sought information on the field size, type of vegetation planted, date of planting, the crop that had been planted before being enrolled in the CRP (base crop), soil types, and the conservation practice employed. Conservation practices encountered and their frequencies were:
CP1 Establishment of permanent introduced grasses and legumes (70%) CP2 Establishment of permanent native grasses (17%) CP3 Tree planting (<1%) CP4 Permanent wildlife habitat (4%) CP10 Already-established grass (8%) CP12 Wildlife food plots (<1%)
A plurality (36%) of the fields had been planted in CRP cover in 1987, but many were planted in 1988 (23%), 1986 and 1989 (16% each year), and in 1990 (10%). Most fields were surveyed 2-4 years after planting.
During our surveys, we visually estimated the areal coverage of the most common plant taxa. Grasses, legumes, and annual weeds were the most common plants on CRP fields, but variation was considerable from county to county and among conservation practices (Table 2), as well as among fields within counties and conservation practices. When grouped by life form, grasses constituted 50% cover, on average, compared with 23 percent for legumes, and 19% for other forbs (Table 3). The percentage of bare ground ranged from nearly 0 to 16%. A few fields had appreciable coverage of water in wetlands.
We surveyed fields for breeding birds using a minor modification of the procedures followed by Stewart and Kantrud (1972), which allow a fairly rapid assessment of the breeding bird community of a field. One or two observers on foot searched each field once each year. All indicated breeding pairs were tallied, based on singing or calling males, females (for Brown-headed Cowbirds), observed pairs, or presence of an active nest. Since some males may not have been paired, the number of actual pairs might be somewhat lower than determined. Although a single observer surveyed small fields, two observers surveyed large fields, each covering about half. Care was taken to avoid double-counting birds. Surveys began about dawn and continued until midafternoon. Although surveys extended beyond the time of most active bird vocalization, Stewart and Kantrud (1972) concluded that activities of open-country birds were not appreciably affected by time of day other than early morning and late evening. Surveys were conducted from late May to about 1 July each year. Counts in adverse weather conditions (precipitation or strong winds) were avoided. During the survey, observers recorded date and time, and weather conditions (temperature, wind, and precipitation).
For 12 of the most common species of birds, we used the SAS GLM (General Linear Model) procedure (SAS Institute, Inc. 1987) to statistically model the density in a field as a function of County (a categorical variate), Year (also categorical), County-Year interaction, Conservation Practice (CP, also categorical), and estimated areal cover of vegetation by life form (Grass, Legume, Other Forb, Bare, and Water). Because of small samples, Conservation Practices 3 and 12 were excluded from the model-building. The model contained all explanatory variables and successively eliminated variables that were not significant at the P = 0.05 level. Population marginal means (also known as least-squares means) were computed from the final model fitted to account for the imbalance in the design (e.g., unequal numbers within each category of County, Year, and Conservation Practice). If County or Year was significant, Fisher's Least Significant Difference method was employed to determine which groups of counties or years differed, and to draw general inferences from those patterns.
We compared the geographical variation in densities of birds to maps of their breeding distributions. The maps were prepared from North American Breeding Bird Survey (Robbins et al. 1986) data by Jeff T. Price (U.S. Fish and Wildlife Service, Northern Prairie Wildlife Research Center, Jamestown, ND), who applied a kriging technique to spatially interpolate data collected during 1985-89.