Roberts et al. • WILDFIRE AND GRASSLAND BIRDS 
25 
is characterized by rolling hills leading to flat 
plains interspersed with ephemeral wetland 
depressions known as playas (Williams and 
Welker 1966). Elevation ranges from 420 to 
>600 m. The climate is characterized by hot 
summers and cold to mild winters, but temper¬ 
atures fluctuate extensively within seasons (Wil¬ 
liams and Welker 1966). The mean summer 
temperature (May-Jul) for 2007 and 2008 was 
22.3 C (U.S. Department of Commerce 2009). 
Historically the Texas panhandle received an 
average of 53 cm of precipitation a year 
(Williams and Welker 1966); the study area 
received ~61 cm of precipitation annually 
during the study period. 
Study Plot Selection .—We selected 20 survey 
plots for study based on access to private property, 
bum history', and vegetation community type of 
either short-grass or mixed-grass prairie (cen¬ 
tered at 14 S 342427 E, 3928890 N). The latter 
distinction was based on soil type and dominant 
vegetation. Burn type, either burned or unburned 
area, was established by talking with landowners 
and local officials, and visual examination of any 
woody vegetation in the area such as prickly pear 
cactus (Opuntia spp.), catelaw mimosa {Mimosa 
aculeaticarpa), or sand sagebrush (Artemisia 
fdifolia). We were unable to initiate this study 
and select sites until the fall following the EAC 
tires and were unable to examine avian commu¬ 
nities during the breeding season directly after 
the fires occurred. The 20 plots were equally 
distributed among areas identified as short-grass 
burned, short-grass unburned, mixed-grass 
burned, and mixed-grass unbumed. Individual 
plots were at least I km apart and 0.5 km from 
a known fire boundary, roads, or other vegetation 
or topographic changes. We analyzed each 
vegetation type separately to examine within-type 
differences among five replicates each of burned 
and unburned plots. All study sites were on 
private property and we were unable to alter 
grazing regimes; sites ranged from no grazing to 
moderate slocking levels. 
Breeding Season Surveys .—We used fixed- 
radius point counts to survey avian populations 
within the 20 plots, May-June, 2007 and 2008. We 
conducted surveys using a three by three grid of 
nine 75-m radius point counts for a total survey 
area of ~|6 ha at each plot. We used three 
observers throughout the study, all of w'hom had 
either prior experience with grassland bird 
identification or were trained prior to surveys. 
Point-count centers were placed 200 m apart to 
minimize, risk of recounting individuals (Ralph 
et al. 1993). We conducted surveys between 0.5 hrs 
before and 3 hrs after sunrise. We did not survey 
points during inclement weather such as rain, or in 
winds >16 km/hr (Ralph et al. 1993). Observers 
recorded all birds flushed within 75 in of the point 
when walking to a point. All birds heard or seen 
in the 75-m radii within a 7-min window' were 
recorded and the distance from the observer was 
estimated (Reynolds et al. 1980). Birds seen while 
walking between points and birds that flew 
overhead but not using the area within the radii 
were not recorded. 
Data Analysis.—We used both the May and 
June surveys to create a measure of average 
abundance and species diversity measures over 
the breeding season. Average abundance was 
calculated by averaging the count of each species 
between the two survey periods. We calculated 
species diversity using the Shannon diversity 
index (H'; Shannon and Weaver 1963). Taylor 
(1978) suggested that diversities calculated over 
a variety of samples arc normally distributed, 
but we used the more robust /-test suggested by 
Hutcheson (1970) to compare H' across burn 
conditions. Hutcheson's /-test uses the number of 
individuals to calculate degrees of freedom, often 
resulting in large numbers. We derived evenness 
(E) from H' using the ratio of observed to 
maximum diversity as described by Magurran 
(1988). 
We only used singing males from June surveys 
to calculate density estimates. We were only 
interested in the number of breeding territories 
supported in each habitat type in contrast to 
richness and diversity measures. June surveys 
were during the height of the breeding season 
when territories were established, and indicative 
of the density of breeding pairs. We used actual 
counts of singing males rather than estimates of 
density derived using detection probabilities. 
Species-specific estimates of detection probabili¬ 
ties introduce additional sources of variability 
(Johnson 2008), and we believed detection 
probability in a grassland landscape is close to 
one and similar across bum conditions, especially 
given our survey was based on territorial males 
that are announcing their presence (Thompson 
et al. 2009). Density estimates are an average of 
the five plots in each treatment. We compared 
densities among years and burn treatments using 
a t-test with an alpha level of 0.05. Our interest 
