Northern Prairie Wildlife Research Center
Lark Bunting: The final model for Lark Bunting included County (P < 0.0001), Year (P = 0.37), County-Year interaction (P < 0.0001), Grass (P = 0.0017), and Legume (P = 0.0058). Year was retained in the model despite its nonsignificance because the County-Year interaction was significant. Lark Buntings were most common in Hettinger and the other western counties but declined to the east. The County-Year interaction suggests that the distribution changed among years. In fact, compared with the overall distribution, the species in 1990 shifted from the most southwestern counties (Butte, Fallen, and Hettinger) northward to Sheridan County and eastward to McPherson County. In 1991 the major changes were decreases in Kidder and Hettinger counties and an increase in Fallen County. In 1992 densities were depressed in Sheridan and Fallen counties, and heightened in Hettinger and Kidder counties. Coefficients associated with both Grass and Legumes were 0.17 (representing the predicted change in birds/100 ha resulting from a 1 percent change in cover), indicating that, within a county and year, Lark Buntings tended to be less common in fields with large cover values for these plant forms.
The Lark Bunting is a bird of the western prairies, as indicated by the west-to-east decline in abundance we found. It also varies in distribution from year to year (Baumgarten 1968), so the County-Year interaction could be anticipated. In North Dakota Lark Buntings favor sage prairie or mixed-grass prairie with shrubs and, to a lesser extent, retired croplands are also used (Stewart 1975).
Grasshopper Sparrow: Grasshopper Sparrows varied geographically (P < 0.0001) with greatest densities in the central and eastern counties and lowest densities in Fallen and Butte counties. The main effect of Year was significant (P < 0.041), with somewhat greater densities (23.7 pairs per 100 ha) in 1992 than in 1991 (19.8) or 1990 (18.7). Again, the County-Year interaction was significant (P < 0.0004), indicating that spatial shifts occurred among years. The most notable effects were a relative change in 1990 from Butte and Fallen counties to Eddy and Kidder counties, followed by a near-reversal in 1991. Densities in the southeastern counties of Day and Grant increased in 1992, while densities declined in McPherson County. The only other explanatory variable that was retained in the model was Legume, with a coefficient of 0.23 (P < 0.0001), suggesting lower Grasshopper Sparrow abundance where legumes were especially common.
The breeding range of the Grasshopper Sparrow is centered in the plains of western Kansas and Nebraska (J. T. Price, personal communication). Annual variation in its numbers often has been reported (Wiens 1973). The species has been reported to favor grassy vegetation with dense low growth but sparse taller cover (Wiens 1969; Harrison 1974). In contrast, Smith (1963), Cody (1968), and Wiens (1973) indicated it was more plentiful in sparser vegetation. Frawley and Best (1991) found that Grasshopper Sparrows colonized alfalfa fields after mowing, but densities declined when alfalfa reached about 300 mm height. Stewart (1975) mentioned use by the species of retired cropland fields. The negative coefficient we found for Legume is consistent with the species' preference for grass.
Red-winged Blackbird: The Red-winged Blackbird was most common in the southeastern counties and least common in the two Montana counties. Densities varied annually, highest in 1991, lowest in 1990. Densities also differed by Conservation Practice (P < 0.0001), highest in CP-4, next highest in CP-1, and lowest in CP-2 and CP-10. Coefficients associated with cover of Grass and Bare were both negative (0.117, P = 0.0019; 0.244, P = 0.028, respectively), suggesting higher densities associated with forbs. The cover of Water was positively and strongly associated with Red-winged Blackbird densities (10.5, P < 0.0001).
Although once considered a wetland species (Bent 1965), the Red-winged Blackbird now nests regularly in uplands. We found a positive association with cover value of Water, indicating that the species certainly favors wetlands. Wiens (1969) found it abundant in clover and alfalfa fields and Stewart (1975) listed retired cropland as a prime habitat. The negative associations we found with percent Grass and percent Bare is consistent with redwing use of forbs for nesting and perching. Its high use of Conservation Practices 4 (permanent wildlife habitat) and 1 (introduced grasses and legumes) are also consistent with their propensity to use legumes or other forbs.
Western Meadowlark: Western Meadowlark densities varied by County (P < 0.0001) and declined markedly from west to east. No other explanatory variable was significant, although Legume was close (0.057, P = 0.052).
Although all our study areas are within the breeding range of the Western Meadowlark, it is more abundant in the west. The species is eclectic in its use of grassland habitats (Stewart 1975), which is compatible with our finding no significant explanatory variables other than County. Frawley and Best (1991) found that densities of Western Meadowlarks did not vary during a season, despite large-scale mowing. They attributed this constancy to the species' large territories, which encompass a variety of habitats. Wiens (1969) indicated that forb height was lower within Western Meadowlark territories than outside them, a finding consistent with our (nearly significant) negative association with Legume cover.
Horned Lark: The Horned Lark also showed a strong trend with decreasing densities from west to east (P < 0.0001), although the County-Year interaction was also significant (P < 0.0001). Conservation Practice was significant (P < 0.0006), with densities greater in Conservation Practice 2 (native grasses) than in the other categories.
Stewart (1975) also noted a gradient in Horned Lark abundance declining from west to east within North Dakota. The species is known to favor open areas with low and sparse vegetation (Wiens 1973; Stewart 1975) Skinner (1975) found an inverse relation between abundance of Horned Larks and grass height in Missouri. Owens and Myres (1973) mentioned that it was the only passerine to use cultivated fields to any degree. In our study, Horned Larks favored Conservation Practice 2 (native grasses), but no vegetation variable was significant.
Savannah Sparrow: Although Savannah Sparrow densities varied by County (P < 0.0001), there was no obvious gradient. Highest densities were in the southeastern counties of Day and Grant, but McPherson County, in north central South Dakota, had densities almost as low as those in the southwestern counties of Fallen and Butte. Savannah Sparrows tended to be more common in fields with high coverage of Water (3.37, P < 0.0005) and low coverage of Legumes (0.056, P = 0.033).
The breeding distribution of the Savannah Sparrow is widespread but with no single center of abundance (J. T. Price, personal communication). The species requires grassy habitat with dense low vegetation (Wiens 1969; Harrison 1974). Owens and Myres (1973) found it fairly evenly distributed across habitats in their Saskatchewan study area, but with a preference for undisturbed grass. Vickery et al. (1993) found lower reproductive success for Savannah Sparrows in territories with greater coverage of tall forbs and shrubs in a Maine study area. Favoring of grassy vegetation is consistent with the negative association we observed with Legumes. The positive association with Water is not easily explicable, although they do nest in dry wetland basins and use emergent vegetation for song perches.
Brown-headed Cowbird: Densities of Brown-headed Cowbirds were highest in the northwestern county of Sheridan, and lowest in the southwestern counties of Fallen and Butte, as well as in Grant County, Minnesota. Year was significant (P < 0.0007), with greater densities in 1991 than in the other two years. The County-Year interaction was also significant (P < 0.0021), mostly the result of a large decline in Sheridan County in 1992. Cowbird densities were negatively associated with both Grass (0.095, P = 0.0039) and Legume (0.071, P = 0.044).
The cowbird is primarily a species of open or edge habitats. Its distribution is centered in the northern Plains states (J. T. Price, personal communication). Their densities vary inversely with percent cover of Grass and Legume, possibly because they forage on the ground, which is difficult in heavy cover.
Clay-colored Sparrow: Clay-colored Sparrows were less common in the southwestern counties than elsewhere (P < 0.0001). Densities were positively associated with cover of Legume (0.11, P = 0.0021). The continental distribution of the Clay-colored Sparrow is centered in central Saskatchewan, and densities decline to the south (J. T. Price, personal communication). In grassland areas, the species is usually found in brushy or shrubby habitat. Western snowberry (Symphoricarpos occidentalis), silver-berry (Eleagnus argentea), and Woods rose (Rosa woodsii) are favored nest substrates in the region. It was thus surprising to find the species as common as it was in CRP fields, which generally lack brushy vegetation. It was, however, positively associated with the coverage of Legumes, primarily alfalfa (Medicago sativa) and sweetclover (Melilotus spp.), which may provide a viable substitute.
Bobolink: Bobolink densities varied spatially (P < 0.0001), generally increasing from west to east, although Hettinger County had higher densities than the trend would suggest. Numbers varied by year (P = 0.033), greatest in 1992, lowest in 1990. The only habitat variable significantly associated with Bobolink density was Grass cover, which was positive (0.052, P = 0.0046).
The center of the Bobolink's breeding distribution is in the northeastern United States; densities decline to the west (J. T. Price, personal communication), as our results indicate. Wiens (1969) noted the preference of Bobolinks for lush grass and forb cover. Other favored attributes are height and density of the vegetation (Dalmbach and Good 1940; Bent 1965; Cody 1968). Wiens (1973) observed that Bobolink territories were nearly completely covered by grasses, which is in agreement with our findings.
Common Yellowthroat: Common Yellowthroats also varied by County (P < 0.0001), decreasing from east to west. They were more common in 1992 and 1991 than in 1990. The County-Year interaction was significant (P < 0.0001), mostly because numbers in Day County were lower in 1990 and higher in 1992 than the model without the interaction would have predicted. Yellowthroat densities varied by Conservation Practice (P = 0.019), being highest in Conservation Practice 4 (wildlife habitat) and lowest in Conservation Practice 2 (native grasses). The species was positively associated with the coverage of Water (2.9, P = 0.0004).
The continental distribution of breeding Common Yellowthroats is primarily in eastern North America (J. T. Price, personal communication), consistent with our west-to-east gradient. Yellowthroats favor lush stands of herbaceous vegetation (Stewart 1975). In this region they are most common in wetlands or brushy areas. The former association is compatible with the positive association we observed with cover of Water. The latter association explains the high densities found in Conservation Practice 4 (permanent wildlife habitat), which often includes the planting of shrubs.
Sedge Wren: Sedge Wrens varied spatially (County significant at P < 0.0001) in a pattern similar to the Bobolink, declining from east to west, with Hettinger County higher than the trend would indicate. Densities were higher in 1990 and 1991 than in 1992 (P = 0.0335). Sedge Wrens were more common in fields with greater coverage of Grass (0.00037, P = 0.0034).
Sedge Wrens breed primarily in eastern North America, with range centered in Minnesota and Wisconsin (J. T. Price, personal communication). They regularly use various wetland types, but are occasionally common in uplands such as retired cropland (Stewart 1975). They have been reported to favor idle grasses with tall and dense growth (Eddleman 1974; Skinner 1975), consistent with our observations.
Chestnut-collared Longspur: Although Chestnut-collared Longspurs varied by County (P < 0.0001), no trend was obvious. Densities were greatest in Sheridan and McPherson counties, and lowest in Fallen and Eddy counties. Longspurs varied annually (P < 0.0001), most abundant in 1990, intermediate in 1991, and least abundant in 1992. The County-Year interaction was significant (P < 0.0001). Densities were highest in Conservation Practice 10 (already established grass), intermediate in Conservation Practices 4 (wildlife habitat) and 1 (introduced grasses and legumes), and lowest in Conservation Practice 2 (native grasses). Longspurs were positively associated with Bare cover (0.039, P = 0.014) and negatively associated with Legumes (0.04, P < 0.0001).
The breeding range of the Chestnut-collared Longspur is centered in north-central North Dakota (J. T. Price, personal communication). Stewart (1975) indicated that the optimal habitat for the species was grazed or hayed mixed-grass or ecotonal mixed-grass and short-grass prairie. It occasionally occurs on cultivated areas. Owens and Myres (1973) found it only on grazed or mowed areas, and noted its requirement for sparse vegetation. We found it more common where Bare cover was high and Legume cover was low. It was also most common in Conservation Practice 10 (already established grass), which likely did not contain the tall, robust cover of early succession. The decline from 1990 to 1992 could reflect decreased areas of bare or sparse cover, due either to the increased establishment of planted vegetation or to greater precipitation in the latter years.