Brooks and Stouffer • GRASSLAND BIRDS IN PINE SAVANNAS 
73 
year of sampling (Brooks 2010). Patch size could 
be one reason why birds did not remain in RCW 
stands. Henslow’s Sparrows occupy small habitat 
patches in high-quality bogs but could be more 
area sensitive in lower-quality habitats. 
Sedge Wrens occurred in all stand types, and 
densities did not differ among stand types. We 
found only one Sedge Wren in one RCW stand; 
this species was not detected in two other studies 
of winter bird communities in RCW clusters 
(Conner et al. 2002, Provencher et al. 2002). Our 
modeling of stand occurrence based on vegetation 
variables weakly suggests Sedge Wrens in DSNF 
may prefer woody understory vegetation. Woody 
understory vegetation is abundant in all stand 
types in DSNF but is lowest in RCW clusters. Our 
results do not show statistically fewer Sedge 
Wrens and lower woody structure in RCW 
clusters, perhaps because of our small sample 
size, but we suspect that further sampling may 
reveal differences. 
Sedge Wren site occurrence was best predicted 
by decreasing tree basal area and increasing 
woody understory vegetation structure. The many 
habitat types used by Sedge Wrens across the 
Southeast show they are habitat generalists in 
wintering areas (Lowery 1974, Imhof 1976, 
Baldwin et al. 2007). Thus, it is not surprising 
the best model explained only 17% of the relative 
proportional variation. Our finding of Sedge Wren 
preference for woody understory contrasts with 
Baldwin et al. (2007), the only other quantitative 
study of Sedge Wren winter habitat preference. 
Site occupancy and abundance in their study were 
not associated with shrub densities but with dense 
herbaceous vegetation. However, they worked in 
Texas costal prairies, an entirely different ecosys¬ 
tem than pine savannas. Sedge Wrens are 
insectivorous (Herkert et al. 2001), and insect 
abundance also could be an important driver of 
habitat selection. 
Caution must be used when interpreting trends 
observed during this study because they are based 
on only two seasons of observations. We believe, 
however, the conditions in DSNF during the study 
influenced the observed habitat-type preferences 
of the study species. These habitats change rapidly 
between years and after fire; thus, quality of 
habitat types will be in flux, and specific habitats 
may appeal to birds differently between years. 
Our best habitat models were not substantially 
better than null models, and the parameter 
estimates had large confidence intervals. These 
models may not be suitable for prediction, but are 
suggestive of potentially important associations 
warranting further research. 
We did not include weather data in our analyses. 
We observed a gradual decline in Henslow’s 
Sparrow and Sedge Wren densities over sampling 
events in the first sampling year (Brooks 2010). We 
speculate this was due to lower-than-average 
precipitation the previous growing season, which 
could lead to increased predation and fewer seed 
resources (Pulliam and Parker 1979, Thatcher et al. 
2006). Precipitation in DSNF in 2007 was 4- 
159 mm below average every month from March to 
November except for October (85 mm above 
average), while the 2008 growing season received 
an even mixture of above- and below-average 
precipitation (USDC 2009). 
CONSERVATION IMPLICATIONS 
Many areas that appear suitable for Henslow’s 
Sparrows, particularly upland longleaf pine sa¬ 
vannas, are unsuitable because of the patchy 
distribution of herbaceous vegetation. Restoring 
the herbaceous component of longleaf pine 
savannas to a continuous layer should be one of 
the principal goals for those interested in grass¬ 
land bird conservation. Upland stands are suitable 
for Bachman’s Sparrows and Sedge Wrens in 
DSNF and, if restored, may become suitable for 
Henslow’s Sparrows. This sparrow occasionally 
occupies isolated patches of upland habitat, but 
the general trend in many areas is a preference for 
bogs. Bogs can be maintained with tree thinning 
and prescribed fire, but upland stands require 
reduction in shrub cover, which will not happen 
with fire alone. It may require a combination of 
fire, mechanical removal, and herbicide applica¬ 
tion (Boyer 1992, Olson and Platt 1995, Drewa et 
al. 2002b). Small-scale bog restoration for Hen¬ 
slow’s Sparrow is likely the most effective 
management strategy. Many practices currently 
used in DSNF benefit grassland birds via 
ecosystem restoration (e.g., prescribed fire, a shift 
towards more growing-season fires, and slash pine 
removal in bogs). Increasing the size and number 
of RCW clusters also will increase the area of 
potentially suitable habitats for all three grassland 
bird species. 
ACKNOWLEDGMENTS 
We thank the USDI, Fish and Wildlife Service for 
funding and the USDA, Forest Service for support, 
particularly M. E. Moody for providing field housing, C. 
