Wilson Bull., 110(3), 1998, pp. 375-383 NEST-SITE SELECTION AND NESTING SUCCESS OF WOOD THRUSHES JEFFREY P HOOVER’ ~*~’ AND MARGARET C. BRITTINGHAM’ ABSTRACT-We characterized nest sites and compared specific nest-site characteristics to nesting success for Wood Thrushes (Hylocichlu mustelina) nesting in southeastern Pennsylvania in 1991. We determined if nests were placed in areas that differed from randomly selected points within a given tract of forest and compared specific nest-site characteristics for successful nests (those that produced at least one fledgling) and nests that failed because of predation. Wood Thrushes selected nest sites non-randomly within a tract of forest, and female Wood Thrushes built nests in areas that had a higher density of trees, higher canopy, higher density of shrubs, and higher average shrub height than randomly selected points. Specific nest-site characteristics had little effect on the ultimate success or failure of nests. The only specific nest-site characteristic included in a stepwise logistic regression model comparing successful and failed nesting attempts was the concealment of the nest from above and below. The average concealment of successful nests was greater than unsuccessful nests, but the model that included nest concealment did not give good fit to the data. Rather, a landscape-level feature, size of forest tract, had the greatest influence on the success and failure of nests for Wood Thrushes in this region. Received 10 Feb. 1997, accepted 20 April 1998. The Wood Thrush (Hylocichla mustelina) is a neotropical migrant that has undergone sig- nificant population declines in recent decades (Sauer et al. 1996). Poor reproductive success, particularly as a result of high rates of nest predation, has been cited as one probable cause of the decline (Robinson 1992, Roth and Johnson 1993, Hoover et al. 1995). Iden- tification of specific habitat features associated with nest sites and nesting success, and a cal- culation of the probability of success given certain characteristics are needed in order to develop long-term strategies for reversing de- clines in populations of Wood Thrushes and other neotropical migrants (Martin 1992). Also, information on nest-site selection may be applied to management of habitat for this and other species of neotropical migrants. General characteristics of the forest habitat where Wood Thrushes are found during the breeding season have been described by other researchers (Bertin 1977, James et al. 1984, Roth 1987). In addition, other researchers have documented the influence of landscape features such as forest patch size and prox- imity to edge habitat on the probability of nest success (e.g., Robinson 1988, 1992; Hoover I School of Forest Resources, The Pennsylvania State Univ., University Park, PA 16802. 2 Present address: Illinois Natural History Survey, 607 E. Peabody Drive, Champaign, IL 61820; E-mail: j-hoover@staff.uiuc.edu. 1 Corresponding author. et al. 1995). In this study, we looked at nest- site selection by Wood Thrushes in two dif- ferent ways. We first determined whether or not Wood Thrushes, within the forest, used particular areas for nesting based on the struc- ture of the vegetation. We then measured mi- crohabitat characteristics of successful nests and nests that were lost to predators. Our spe- cific objectives were to: (1) determine the characteristics of the vegetation that influ- enced the probability that a site would be used for nesting by a Wood Thrush, and (2) deter- mine whether or not microhabitat character- istics at the nest site influenced the probability of nesting success. METHODS During the summer of 1991, data for this study were collected as part of a larger study of the nesting suc- cess of Wood Thrushes in a fragmented forest land- scape in Berks County, Pennsylvania (see Hoover et al. 1995 for a general description of the study sites). Characteristics of the vegetation were measured on nine tracts of forest (study sites) ranging from 16.4 ha to more than 500 ha (Table 1). Nine randomly-selected points were established on each tract of forest by plac- ing a 150 X 150 m scale grid over a map of each tract, assigning a number to each grid point, and using a random numbers table to select the points. Points were separated by a minimum of 150 m and, when possible, were located at least 60 m from the nearest forest edge. We collected vegetation data from each study site during July. We measured characteristics of the vege- tation within a 0.04 ha circle centered on each of the nine points by using a modification of the James and Shugart (1970) method. We recorded the number of 375