700 
THE WILSON JOURNAL OF ORNITHOLOGY • Vol. 123, No. 4. December 2011 
summed across all 50 stops; they represent a 
composite of the habitats encountered and stop- 
specific changes in habitat are overshadowed 
(Sauer 1999). A more in-depth approach to 
examine how habitat changes affect Cerulean 
Warbler population trends on BBS routes is to 
analyze data at each stop instead of across the 
entire route. Small scale habitat characteristics, 
including slope position, aspect, and microhabitat 
features such as canopy gaps, can affect Cerulean 
Warbler abundance (Weakland and Wood 2005, 
Perkins 2006); these features are lost when habitat 
is analyzed at the route level. 
We examined the association between change 
in the Cerulean Warbler population and habitat at 
stops along BBS routes in the core breeding range. 
We (1) analyzed the effects of land cover and 
forest fragmentation changes, measured from 
aerial photographs, on Cerulean Warbler popula¬ 
tions over three time periods at stops on a 
subsample of BBS routes to examine long-term 
changes; and (2) quantified land cover and forest 
fragmentation metrics at BBS stops over two lime 
periods using National Land Cover Data (NLCD) 
to examine effects of more recent habitat changes 
on Cerulean Warbler populations along a broader 
sample of BBS routes. 
detections of Cerulean Warblers within at least 
one time period. Our study examined how change 
in land cover and habitat may have affected 
detections of Cerulean Warblers, and routes that 
did not have a single detection lacked this 
information. (2) We included only BBS routes 
with stop-level Global Positioning System (GPS) 
coordinates collected in recent years by route 
observers so stops could be mapped as accurately 
as possible. Route observers at times adjust stop 
locations to safer or quieter stopping points; stop 
locations mapped by third parties may not be 
accurate. Stops were not included if the route path 
was changed between time periods. (3) We only 
used BBS routes that were surveyed at least three 
times within the 5 years centered on each year of 
available land cover data. Survey data over 
multiple years better identify stops at which 
Cerulean Warblers actually were present hut were 
missed in some years. Birds can be missed in a 3- 
min counting period and. if not detected, cannot 
be assumed to be absent. The 3-5 year period also 
may provide a more accurate measure of the 
response of Cerulean Warblers to habitat changes 
than data from only 1 year, which would be more 
of a snapshot in time. Using BBS data from years 
that bracket the land cover data ensured that land 
METHODS 
Study Area .—We used survey data from BBS 
routes within the West Virginia, Kentucky, and Ohio 
portions of BCR 28 (Fig. 1). The study area is within 
NLCD mapping zones 47, 53. 61. and 62 (Homer et 
al. 2004). The Ohio Hills, Northern Cumberland 
Plateau, and Mid Atlantic Ridge and Valley 
physiographic regions (www.partnersinflight. 
org/bcps/pi fplans.htm) of BCR 28 comprise most 
of the study area. The Ohio Hills (~8 million ha) is 
characterized by dissected plateaus ranging from 
150 to 450 m in elevation (Rosenberg 2000). The 
Northern Cumberland Plateau (~5.5 million ha) is 
a rolling hills tableland ranging from 300 to 580 in 
m elevation (Demurest 2003). The Mid Atlantic 
Ridge and Valley (~5 million ha) is dominated by 
long mountainous ridges and intervening valleys 
ranging from 100 to 1.100 m in elevation 
(Rosenberg 1999). Dominant land cover of each 
physiographic area is mixed mesophytic forests 
consisting primarily of oaks {Quercus spp.) and 
hickories (Carya spp.). 
Data Collection.—We used data from BBS 
routes within the study area that met three criteria. 
(1) Selected routes had at least one stop with 
routes were surveyed. 
Land cover data can be measured from several 
remotely-sensed data sources; we used aerial 
photographs and the NLCD in our study. Aerial 
photographs have been taken occasionally within 
the study area over the lifetime of the BBS and are 
most suited to investigate long-term changes in 
land cover along BBS routes. However, aerial 
photographs are spatially limited, do not provide 
complete coverage of many BBS routes, and 
hand-digitizing land cover from them is time- 
consuming. The NLCD allows land cover to be 
quickly assessed for a broader extent of BBS 
routes, but was limited to 1992 and 2001 at the 
time of our study. Thus, we used aerial photo¬ 
graphs to examine long-term land cover changes 
at a subset of BBS stops and NLCD data to 
quantily changes in land cover over a shorter time 
Period for BBS stops along 28 routes. 
Aerial photograph data were separated into 
three periods: 1967/1971 (early), 1982/1985 
(middle), and 2000/2003 (late) to correspond with 
years that aerial photographs were available. We 
used aerial photographs taken during leaf-off (Oct 
to early May) when it is possible to distinguish 
