Earlier studies argued that forest interior birds could be used as an indicator of forest loss and fragmentation. Although there are many studies that investigated a bird-habitat relationship at a local/patch scale, few examined the relationship at a broad, regional scale. The objectives of the study are to: (1) determine if there is a threshold amount of forest below which the abundance of the selected forest breeding bird species (i.e., Eastern Wood-Pewee, Wood Thrush, Black-and-white Warbler, Ovenbird, and American Redstart) declines significantly; (2) determine if the effect of spatial configuration of forest increases below the threshold amount, if identified. The North American Breeding Bird Survey and the 2001 National Land Cover Database were combined; for each bird species, all the bird survey routes that are within the breeding range and Metropolitan Statistical Areas were selected. Three buffers (i.e., 180 m, 2010 m, and 6000 m) were created around each bird survey route; % tree cover, % imperviousness, various landscape structure metrics on “forest” land cover, and % of each land cover type were calculated within each buffer. The mean route-level abundance of the selected forest birds over five years (2002-2006) was used as the response variable in regression analysis.

Results/Conclusions

At 2010 m buffer distance, simple linear regression of bird abundance against % forest shows that Ovenbird (OVEN) has the highest r² value (0.2956) and Eastern Wood-Peewee (EAWP), the lowest r² value (0.0397). Multiple regression results show that for the species (EAWP and Wood Thrush (WOTH)) with lower r² values, land mosaic diversity and contrast-weighted edge density are the most important variables. For the species (American Redstart, Black-and-white Warbler, and OVEN—belonging to the same Family) with higher r² values, % forest and similarity to forest cover in the buffer are the most important variables. The results are consistent with earlier studies’ findings in terms of each species tolerance to fragmentation, use of forest edge as well as interior, and % forest requirement to be “present.” Forest amount thresholds found by piecewise regression are between 42 and 56%. However, the thresholds have wide confidence intervals, indicating instability. Mixed-effects models will be used next to account for possible spatial autocorrelations in the data. In conclusion, some land mosaic diversity and forest edges can be beneficial to more fragmentation-tolerant forest birds, and this kind of species-specific requirements need to be taken into consideration for the conservation of forest birds.