I studied how breeding and wintering forest bird communities across Connecticut responded to variation in habitat characteristics and particularly such landscape attributes as forest fragmentation. I surveyed birds at 1,815 points along 121 transects that traversed ca. 400 km of forest. I also made 12705 habitat measurements at survey points and computed areas of forest, non-forest, core forest and perimeter/area ratios of forest for 31,550 ha of study area. I computed sampled species richness and community density as well as individual species’ population densities for each transect. Moreover, I classified species encountered as to their nest site selection, macrohabitat use, microhabitat use, migratory strategy and trophic affiliation. Based on observations of 36,702 summering individuals of 123 species and 13,742 wintering individuals of 63 species, declines in community density occurred with increasing fragmentation although species richness was often more closely associated with habitat measures. Among landscape measures, forest fragmentation had the closest association with summer community measures 67% of the time, strongly suggesting that fragmentation effects were the predominant driver of such community patterns. However, short-distance migrant density and richness, foraging generalist density and richness, edge/successional species density and richness, habitat generalist density, and Brown-headed Cowbird density showed little relationship to landscape measures. The effects of fragmentation appeared to predominate over those of simply forest extent in predicting summer and winter bird community characteristics even in the comparatively extensive forests of southern New England. Despite the importance of fragmentation effects, community and individual species measures often tended to be more closely associated with habitat measures than with those of fragmentation. In addition, few summer or winter community measures or species patterns showed any significant relationship to natural forest breaks.  Winter community and species density patterns showed little relationship to any landscape measures, with particularly elevation appearing to be a principal driver of winter patterns.

I established 121 bird survey transects, with each traversing 3.2−4 km of forest depending upon terrain and other local conditions.  Each transect had 15 survey points—the maximum I could visit during the peak of morning bird activity.  Routes began at first light (05:15 in summer, 07:00 in winter) and lasted 3.5−4 hr.  I used the Variable Circular Plot (VCP) technique to survey. I computed population densities with Distance 7.3 software.

I visually evaluated habitat to a 70 m radius from each sampling station for: 1) forest type, 2) moisture regime, 3) diameter of canopy trees at breast height (dbh), 4) canopy cover, 5) understory density and 6) elevation at the location of the survey point. I also summed canopy and understory measures to provide a measure of 7) vertical vegetation complexity. To characterize the landscapes within which the survey transects were situated, I employed QGIS 3.16 geographic information systems software to analyze aerial orthophotos. I plotted the survey points from each transect onto the photos and then constructed a 400 m wide buffer around the transect, thereby producing an 800 m wide corridor (averaging 260.7 ha area and covering 31,550 ha in sum for all transects) through which the transect passed. I chose this width because nearly all bird detections were within this corridor. In the corridor, I constructed polygons outlining each habitat present. Although pixel classification tools are frequently used for delineating habitats at this scale, I found that although doing so was more labor intensive, by visually comparing non-growing and growing season images and well as by relying on my familiarity with the study sites, I could construct digitized habitat polygons that maximized accuracy, precision and detail.