A Multi-Scale Analysis of Grassland Bird Habitat Relationships in the St. Lawrence River Valley

Abstract

I used a combination of 10 vegetation variables and 10 landscape variables to model abundance and occurrence of six grassland bird species in agricultural grassland (n=55) throughout Jefferson County, NY during the 2004 and 2005 field seasons. Landscape composition was quantified by classifying the proportion of land use within a 1 km radius from the center of all survey fields. Land use classification was based on 2003 aerial photo interpretation. Bobolinks (Dolichonyx oryzivorous) and Savannah Sparrows (Passerculus sandwichensis) were the most abundant species, followed by Grasshopper Sparrows (Ammodramus savannarwn), Eastern Meadowlarks (Sturnella magna), Upland Sandpipers, (Bartramia longicauda) and Henslow's Sparrows (Arrmzodramus henslowii). Bird habitat models generated through best subsets regression and stepwise multiple regression indicated that perimeter-area ratio and variables associated with area, such as distance to nearest edge and distance to forest edge, generally explained most of the variance in grassland bird species richness and abundance, and individual species abundance. Vegetation variables, including grass cover, legume cover, litter depth and the number of plant species, also entered into the grassland bird-habitat models. Bobolink and Savannah Sparrow abundance increased in fields high in forb cover and plant diversity. A significant proportion of the variance in Grasshopper Sparrow and Savannah Sparrow abundance was explained by a decrease in vegetation cover, while an increase in vegetation cover explained a significant amount of the variance in Eastern Meadowlark abundance. As with the Bobolink, the most important predictor variables for Grasshopper Sparrow and Savannah Sparrow were related to area. Henslow's Sparrow abundance increased as the proportion of development in the surrounding landscape decreased. In contrast, Upland Sandpiper abundance increased as the proportion of development in the surrounding landscape increased. My models differed between years and also produced some results that differed from those of other grassland bird habitat-selection studies from the Midwest and Northeast, thus suggesting that grassland bird habitat selection is dynamic among years, and that habitat requirements are broad across regions. Average obligate grassland bird density in the agricultural grasslands of Jefferson County ranged from 0.04 to 3.77 birds/ha across both years. Grassland bird densities in Jefferson County compared favorably to grassland bird densities at 13 U.S. Fish and Wildlife Service National Wildlife Refuges in Region 5, thus suggesting that anthropogenic grasslands planted with non-native, cool season grasses are a valuable resource for grassland bird conservation in the Northeast.