722 
THE WILSON JOURNAL OF ORNITHOLOGY • Vol 124. No. 4. December 2012 
Robbins et al. 1998). The latter (1998, 2010) 
conducted river-based surveys in 1999-2006 in 
Missouri and northern Arkansas to address this 
lack of data and document occurrence of some 
species of concern, including Cerulean Warblers, 
in riparian habitats across the state. These surveys 
are not a representative sample of forest in the 
state but they do represent forests where Cerulean 
Warblers are known to occur and reach modest 
densities (Reidy et al. 2011). The study design 
provided an opportunity to assess how local- 
versus landscape-scale habitat composition affects 
abundance in this previously unstudied region. 
Our objective was to evaluate the relative 
support for the hypotheses that percentage of 
bottomland and upland forest had a positive 
effect, and developed land a negative effect on 
abundance of Cerulean Warblers in Missouri and 
Arkansas. We compared support for effects at: (I) 
a local-habitat scale (within a 250-m buffer), (2) a 
landscape scale (within a 10-km buffer), or (3) a 
combination of local-habitat and landscape scales. 
Chariton 
Big Piney 
Jack's Fork River 
James River 
North Fork 
White River 
r- Buffalo River, AR 
Black River 
HG. I Portions of rivers along which we measured effects 
of habitat aiul landscape factors on the number of singing 
Cerulean Warblers in Missouri and Arkansas, 1999-2006.” 
ivitiJ IIW LAS 
Study Area. We conducted surveys along 
rivers in Missouri and northern Arkan: 
(Fig. l).We selected rivers that had at least so 
riparian lorest along them, occurred across i 
region and along a gradient of landscapes fr< 
mostly forested to minimally forested, anil w. 
accessible by canoe to allow us to conduct surve 
across public and private land. Rivers in southe 
Missouri and northern Arkansas occurred 
landscapes of rolling hills covered primarily w, 
hardwood toresLs interspersed with glades a: 
woodlands and dissected by deep river valleys T 
two rivers in northern Missouri occurred 
landscapes dominated by cropland and pasta 
with narrow corridors of riparian forest. Forests 
the nver floodplains included oaks (Q,terms spp 
hickories Katya spp.). ash (Fraxinus spp.). mapl, 
indAm-' ha c kbeny ° r su S arbcr T (Cel,is spp. 
Forei i r SyC ; a T e {Plant ™“s Occident1 
Northern A t ^ W 80uther » Missouri an 
extremes (i.e.. after heavy rains which ZtJZ 
ow and no.se. or during low water levels) and onK 
under conditions of no precipitation and no or ve.y 
light wind. Wc surveyed river sections that did not 
exceed 20 km, with lew exceptions, to ensure 
surveys were completed by -1000 hrs. River width 
was generally <50 m (maximum 90 m) and birds on 
both sides of the river could be heard. Robbins etal. 
(2010) estimated that Cerulean Warblers could be 
heard 100 m in width on each side of the river 
hank. Each Cerulean Warbler was heard singing at 
least twice before being recorded to ensure it was 
not confused with Type B (= Song Type II). 
(Moldenhauer and Regelski 1996) of Northern 
Farula f.S. americana). We calculated coordinates 
°l each singing Cerulean Warbler by estimating the 
direction and shortest perpendicular distance to a 
bird from our position on the river with a laser 
rangefinder; we measured our position on the river 
w ith a global positioning system unit (Gannin 12, 
Map Datum WGS 84). Observations were directly 
marked on topographic maps on a limited number of 
surveys. Lour observers conducted surveys but (he 
same observer conducted all surveys on a river to 
eliminate multiple surveyor bias (Sauer et aJ. 1994). 
Habitat and Landscape Measurements .—We 
partitioned rivers into 5-km segments and created 
250-m and 10-km flat-ended buffers around each 
segment in a geographic information system so 
there was no overlap in buffered areas between 
•segments. We partitioned rivers into segments to 
account lor variation in land cover along the rivers 
and 5-km was sufficiently large to capture variation 
