INFLUENCE OF FIRE ON BACHMAN’S SPARROW, AN ENDEMIC
NORTH AMERICAN SONGBIRD
JAMES W.TUCKER, JR.,1, 2 Department of Biological Sciences, Auburn University, AL 36849, USA
W. DOUGLAS ROBINSON,3Department of Biological Sciences, Auburn University, AL 36849, USA
JAMES B. GRAND, USGS, Alabama Cooperative Fish and Wildlife Research Unit, School of Forestry and Wildlife Sciences,
Auburn University, AL 36849, USA
Abstract: Bachman’s sparrow (Aimophila aestivalis), a near endemic songbird of the longleaf pine (Pinus palustris)
ecosystem, is known to respond positively to prescribed fires. The influence of season (growing vs. dormant) and
frequency (1to ≥4 yr since burning) of fire on density of Bachman’s sparrows, however, is poorly understood. We
examined effects of fire on density of Bachman’s sparrows in longleaf pine forests at the Conecuh National Forest,
Alabama, and Blackwater River State Forest, Florida, USA. Density of Bachman’s sparrows was greater the first 3
years after burning than ≥4 years after burning, and season of burning had little effect on the density of Bachman’s
sparrows. Percent coverage by grass had a greater influence on density of Bachman’s sparrows than either season
or frequency of burning. Percent canopy cover had a strong negative effect on coverage of grass but had a weaker
effect on grass at stands burned frequently during the growing season. Growing-season fires (Apr–Sep) did not
adversely affect density of Bachman’s sparrows. Results from our study suggest that management and restoration
of longleaf pine communities probably can be accomplished best by burning on a 2–3-year rotation during the
growing season, when most fires historically occurred. Suppression of fire, or burning at intervals >4–5years, will
greatly reduce or eliminate habitat required by Bachman’s sparrows.
JOURNAL OF WILDLIFE MANAGEMENT 68(4):1114–1123
Key words: Aimophila aestivalis, Alabama, Bachman’s sparrow, fire ecology, Florida, habitat management, habitat
restoration, longleaf pine ecosystem, Pinus palustris, prescribed fire.
The longleaf pine ecosystem once dominated the
coastal plain of the southeastern United States and
extended from Virginia to Texas (Wahlenberg
1946). Forests within this ecosystem were charac-
terized by large, widely spaced longleaf pines with
a dense ground cover of grasses and forbs
(Chapman 1932). Longleaf pine forests are among
the most species-rich plant communities in North
America (Peet and Allard 1993) and are unusual
because most of the plant diversity is in the ground
cover (Simberloff 1993). Frequent fire resulting
from a high incidence of lightning strikes is the pri-
mary agent responsible for development and struc-
ture of longleaf pine forests (Chapman 1932), but
anthropogenic effects of burning by Native Ameri-
cans also might have had an effect (Krech 1999).
The longleaf pine ecosystem is among the most
heavily impacted of all forested ecosystems (Noss
1989, Simberloff 1993). Disruptions of natural
fire regimes and conversion to other land uses
have resulted in the loss of 95% of this once-
extensive ecosystem (Outcalt and Sheffield 1996).
Concurrent with the loss of longleaf pine forests,
populations of many species characteristic of
these forests (e.g., red-cockaded woodpecker
[Picoides borealis]) have declined and become
threatened with extinction. Previous studies
(e.g., Tucker et al. 1998, 2003) suggest that pre-
scribed burning benefits bird species associated
with longleaf pine forest, but information that
assesses the importance of frequency and season
of burning on bird populations in the longleaf
pine ecosystem is lacking. In particular, studies of
species utilizing the understory of longleaf pine
forests are needed because most sensitive species
occupy that stratum of the forest. Although the
understory structure of these forests also strongly
affects red-cockaded woodpeckers (James et al.
1997), the relatively strict habitat requirements of
Bachman’s sparrow are more directly related to
understory vegetation (Dunning and Watts 1990).
Bachman’s sparrow is 1of the few bird species
endemic to the continental United States and is
largely endemic to longleaf pine forests (Jackson
1988). Bachman’s sparrow is among the species
of highest management concern within the
southeastern United States (Hunter et al. 1994)
and is classified as threatened or endangered by
several states (Dunning 1993). Like most threat-
ened species associated with longleaf pine
1Present address: Archbold Biological Station,
APAFR Field Office, 475 Easy Street, Avon Park, FL
3Present address: Department of Fisheries and Wild-
life, 104 Nash Hall, and Oak Creek Laboratory of Biol-
ogy, Oregon State University, Corvallis, OR 97331, USA.
J. Wildl. Manage. 68(4):2004 1115
FIRE AND BACHMAN’S SPARROWS • Tucker et al.
forests, Bachman’s sparrows are adversely affect-
ed by the loss of herbaceous ground cover result-
ing from disruption of natural fire regimes (Dun-
ning and Watts 1990, Plentovich et al. 1998,
Tucker et al. 1998). Bachman’s sparrows nest and
forage on the ground, and they require a dense
ground cover of herbaceous vegetation (Dun-
ning 1993). Several studies (e.g., Dunning and
Watts 1990, Plentovich et al. 1998, Tucker et al.
1998) have documented the importance of under-
story vegetation and used evidence from botani-
cal studies (e.g., Platt et al. 1988, Waldrop et al.
1992, Streng et al. 1993) to suggest that frequent
fire, and particularly growing season (Apr–Sep)
fires, are needed to maintain the dense herba-
ceous ground cover required by Bachman’s spar-
rows. However, no studies have directly examined
the influence of frequency and season of fire.
Our objective was to examine the influence of
season and time since burning on density of
breeding Bachman’s sparrows. We took advan-
tage of a landscape-level fire management plan
executed by the adjacent Conecuh National For-
est in Alabama and Blackwater River State Forest
in Florida that included both dormant- and grow-
ing-season fires. Historically, active management
by fire had been used during winter to avoid
impacts on breeding birds, yet natural fires
occurred most often during late spring and sum-
mer, coinciding with the season of electrical
storms (Robbins and Myers 1992). Recent shifts
to burning during the “natural fire season” have
again raised concerns that nesting birds may be
threatened by such activities. Our aim was to eval-
uate whether such concerns for numbers of
breeding Bachman’s sparrows are warranted.
During April 1999, we established counting
points at 180 stands, 110 stands in the Conecuh
National Forest (hereafter, Conecuh) and 70 in
the Blackwater River State Forest (hereafter,
Blackwater; Table 1). Stands were forest manage-
ment units ranging in size from 6.2to 247.5ha
(median = 68.2ha, 95% of stands ≥13.4ha). All
stands were dominated by mature (>50 yr old)
longleaf pines. Prevalent grasses included wire-
grass (Aristida stricta) and bluestems (Andropogon
spp. and Schizachyrium spp.), and prevalent
shrubs included gallberry (Ilex glabra) and yaupon
(I. vomitoria). All stands were selected based on
season (growing season, Apr–Sep; and dormant
season, Oct–Mar) and frequency (1–5yr since
last burned; i.e., post-burn age) of burning. We
had few opportunities to examine stands imme-
diately after growing-season fires; therefore, post-
burn age of stands burned in the growing season
correspond to number of full growing seasons
after burning (i.e., not including season when
We used stand maps to identify potentially suit-
able stands, and a single counting point was
established in each stand meeting the following
criteria: (1) point-count center located ≥100 m
from a stand edge, and (2) point-count center
≥150 m from the edge of any other treatment
group (i.e., another season–frequency combina-
tion included in our study). Thus, all points were
≥300 m from any other counting point. In addi-
tion, all points within a treatment were ≥800 m
apart during 1999, but burning between years con-
verted some stands initially in different treatment
groups into equivalent treatments for the field
season of 2000. However, all points within a treat-
ment group during 2000 were still ≥400 m apart.
Initially, we attempted to locate 15–20 stands
within 10 treatment groups (i.e., 2seasons within
each of 5post-burn age groups). However, few
Table 1. Number of stands examined by season of burning and
number of growing seasons since burning (burn group) for the
study of breeding Bachman’s sparrows during 1999 and 2000
at Conecuh National Forest (CNF), Alabama, USA, and Black-
water River State Forest (BSF), Florida, USA.
1999 Field season 2000 Field season
Season Burn Burn
Location burneda group Stands group Stands
CNF Growing 1 15 1 30
CNF Growing 2 14 2 15
CNF Growing 3 25 3 2
CNF Growing ≥4 7
CNF Dormant 1 15 1 13
CNF Dormant 2 22 2 15
CNF Dormant 3 11 3 20
CNF Dormant ≥4 8 ≥4 8
BSF Growing 1 5 1 1
BSF Growing 2 5 2 3
BSF Growing 3 4 3 4
BSF Growing ≥4 28 ≥4 26
BSF Dormant 1 5 1 15b
BSF Dormant 2 5 2 5
BSF Dormant 3 6 3 4
BSF Dormant ≥4 12 ≥4 12
aGrowing season = Apr–Sep; Dormant season = Oct–Mar.
bOne stand burned 17 Apr 2000 treated as dormant season
2000 burn, because vegetative growth most similar to that of
stands burned in dormant season 2000.
J. Wildl. Manage. 68(4):20041116 FIRE AND BACHMAN’S SPARROWS • Tucker et al.
stands >3 years post-burn were available at
Conecuh, and no stands burned during the dor-
mant season of 1996 were available at either
Conecuh or Blackwater. Thus, sample sizes with-
in treatment groups were unequal and the sam-
pling design was unbalanced (Table 1). To exam-
ine the influence of fire history in more detail, we
reviewed the management histories of each stand
and recorded the year and season for each fire
dating to 1990. We used these data to derive 3
variables summarizing the number of fires within
the last 10 years for the growing season, the dor-
mant season, and the 2seasons combined.
We conducted point counts at each of the 180
stands between 16 April and 26 June 1999 and
between 20 April and 31 May 2000. Point counts
were 7min in duration, beginning with 3min of
recording all birds seen or heard, followed by 1
min of broadcasting a tape recorded song of Bach-
man’s sparrow, and ending with another 3-min lis-
tening period. We recorded the distance and
direction to each bird detected during the point
counts but restricted the analysis to Bachman’s
sparrows detected within 100 m of the counting
points. Plastic flagging tied at 50- and 100-m inter-
vals from counting points aided in estimating dis-
tances to birds. We made a complete rotation of all
180 stands before returning to a stand for an addi-
tional count. Within a rotation of counts, we
grouped stands into routes containing stands
located in the same general area and randomized
the order routes were surveyed to reduce potential
bias resulting from seasonal effects. The order
stands within each route were surveyed was alter-
nated such that counts at each stand were distrib-
uted equally among the first 4hr of daylight. Two
observers visited the points: J. Tucker visited each
point 3times in 1999 and 2times in 2000, and D.
Robinson visited each point once in both years.
Statistical comparisons revealed that the number
of Bachman’s sparrows per count did not differ
between observers in either year (paired sample t
≤0.659, P≥ 0.511, df = 179).
Thirty-one of the stands (30 at Conecuh and 1
at Blackwater) were burned during the growing
season of 1999. Before these stands were burned,
we had completed 2point counts at 11 stands, 3
point counts at 2stands, and 4point counts at 18
stands. Twenty-eight other stands were burned
during winter 1999–2000, and no stands were
burned during breeding season 2000 because of
a severe regional drought.
We sampled vegetation at 86 stands during July
and August 1999, using a random stratification to
select stands for vegetation sampling. When pos-
sible, 5stands were selected at random from each
treatment group at both Conecuh and Blackwa-
ter. For the older burn groups (≥4 yr), we
attempted to select 10 stands at random from
Blackwater because all the older sites at Conecuh
had been burned.
At each sampled stand, we generated 10 sets of
random coordinates to dictate the location of
sampling points within a 100-m radius of the
counting point. At each sampling point, we used
a 10-factor cruising prism to measure basal area
and a spherical densiometer to measure canopy
cover. In addition, a random compass bearing was
used to orient a 10-m transect at each sampling
point. At each 1.0-m interval along the transect,
we held a 4.0-m pole in a vertical position and
recorded the presence of each plant species that
contacted each 0.1-m height interval of the pole.
Percent ground cover by vegetation types (e.g.,
grasses, forbs, shrubs) was estimated by the per-
centage of 1.0-m intervals along the transect
where the vegetation type contacted the pole.
We used distance sampling (Buckland et al.
2001) to estimate probability of detection and
density (birds/ha) of Bachman’s sparrows. Data
were truncated to include only birds detected
within 100 m of the counting points. In addition,
we included only the counts conducted before
stands were burned during 1999, and sampling
effort was accounted for in our estimates of den-
sity. We used program Distance (Thomas et al.
2003) to model the influence of time since burn-
ing (TSB; average number of growing-season
days since burning), percent shrub cover (arcsine
square root transformed), and basal area of trees
(log transformed) on probability of detection at
the 86 stands where vegetation data were collect-
ed during 1999. To maximize our power of
detecting significant effects among post-burn
groups and season of burning, we also estimated
probability of detection and density using data
from all 180 stands post-stratified by year of study
and included TSB as a covariate. Based on pre-
liminary examinations of the data (Buckland et
al. 2001), we used a hazard-rate function with no
adjustments to model probability of detection.
In all analyses involving vegetation data, we
used arcsine square root transformations (Zar
J. Wildl. Manage. 68(4):2004 1117
FIRE AND BACHMAN’S SPARROWS • Tucker et al.
1984) to normalize the distributions of variables
measured as percentages (e.g., canopy cover)
and used the natural logarithm for basal area of
trees. We used 2-way analysis of variance
(ANOVA) to examine the influence of season of
fire and post-burn age on dependent variables.
Although densities of Bachman’s sparrows were
not normally distributed, subjecting ranks of the
data to ANOVA (Zar 1984) yielded equivalent
results to parametric ANOVA. Thus, violations of
assumptions for parametric analysis did not influ-
ence results, and only results from parametric
ANOVA are presented. Repeated measures
ANOVA could not be used because burning
between years of study caused inconsistent
changes in season of fire and post-burn age, so we
analyzed years separately. Including location (i.e.,
Conecuh and Blackwater) in 3-way ANOVA
resulted in missing and/or small group sizes (see
Table 1); therefore, Tucker (2002) analyzed these
data separately by location. Results from those
analyses were consistent between locations (Tuck-
er 2002) and nearly identical to results with loca-
tions pooled. We therefore present only results
with locations pooled to simplify the presenta-
tion. We used Tukey’s HSD test for multiple com-
We used standard least-squares linear models
(PROC REG; SAS Institute 1999) to examine the
influence of vegetation structure, season of burn-
ing, and TSB on density of Bachman’s sparrows.
Only data collected in 1999 at stands where vege-
tation was measured (n =86) were included in
this analysis. We used an information–theoretic
approach to direct model selection and parame-
ter estimation (Burnham and Anderson 2002,
Anderson et al. 2000). Model selection was based
on using Akaike’s Information Criterion correct-
ed for small sample size (AICc), which allows
selection of the most parsimonious model from a
candidate set of models (Burnham and Anderson
2002). We used AICcweights to calculate parame-
ter estimates and their precision across all models
(i.e., model averaging and unconditional vari-
ances), which incorporated uncertainty of model
selection and reduced bias compared to parame-
ter estimates from a single best model (Burnham
and Anderson 2002). We calculated AICcusing
residual sum of squares from the least-squares
models (Burnham and Anderson 2002).
We used knowledge and experience in devel-
oping an a priori set of candidate models (Burn-
ham and Anderson 2002). Because previous stud-
ies (see Plentovich et al. 1998, Tucker et al. 1998,
and especially Haggerty 2000) have found per-
cent coverage of grass to be a pervasive factor
influencing the presence and abundance of
Bachman’s sparrows, percent coverage of grass
(arcsine square root transformed) was the only
vegetation variable considered in the models. We
tested for colinearity between percent coverage
of grass and TSB using Spearman’s rank correla-
tion and found that these 2variables were not
correlated (rs= 0.045, P= 0.680, n =86).
We perceived percent coverage by grass to be
the most important vegetation variable influenc-
ing Bachman’s sparrows, and we therefore also
modeled percent coverage by grass as a function
of canopy cover and burn history to examine the
influence of management on grass. We also used
AICcto evaluate these models. We expected many
of our vegetation variables to be correlated, and
we expected that canopy cover, an index of sun-
light reaching the forest floor, would be the veg-
etation variable having the most direct effect on
percent coverage by grass. The 3variables for
burn history (number of fires in the previous 10
yr during the growing season, dormant season,
and both seasons combined) were not indepen-
dent, so we included only 1of each in individual
models. We did not include TSB in modeling per-
cent coverage of grass because TSB is largely
dependent on burn history; that is, stands with
small values of TSB are more likely to have been
burned more times than stands with larger values
of TSB, and vice versa.
Density of Bachman’s Sparrows
Bachman’s sparrows were common at Conecuh
and Blackwater. We detected Bachman’s sparrows
(at unlimited distances) in 156 of the 180 stands
in both years (91  and 94  of the 110
stands at Conecuh and 65 and 62 of the 70 stands
at Blackwater). Restricting detections of Bach-
man’s sparrows to ≤100 m of counting points
resulted in 147 and 149 of the stands (87 and 88 at
Conecuh and 60 and 61 at Blackwater) having ≥1
detection for analysis in 1999 (568 total detections)
and 2000 (505 total detections), respectively.
Hazard-rate models with no adjustments and
no covariates were the most parsimonious mod-
els for estimating probability of detection. At the
86 stands where vegetation data were collected
during 1999, a model without any covariates
(AICcweight = 0.562) was ≥5.5times more likely
to be the best model for estimating probability of
J. Wildl. Manage. 68(4):20041118 FIRE AND BACHMAN’S SPARROWS • Tucker et al.
detection than models that included TSB, shrub
cover, or basal area of trees (individually or in
combination) as covariates (AICcweights ≤0.103).
Probability of detection at the 86 stands during
1999 was estimated to be 0.76 (95% CI: 0.70 to
0.83). Probability of detection by year of study
using data from all 180 stands was estimated better
by a model that did not include TSB (AICcweight
= 0.756) than a model that included TSB (AICc
weight = 0.244). Probability of detection at the
180 stands was similar in 1999 (0.60, 95% CI: 0.50
to 0.70) and 2000 (0.65, 95% CI: 0.50 to 0.84).
The effect of season of fire and post-burn age
on density of Bachman’s sparrows during 1999 was
significant (F7, 172 = 3.084, P= 0.004). Season of
last burning did not affect density of Bachman’s
sparrows (F1, 172 = 0.092, P= 0.762) but post-burn
age did (F3, 172 = 6.156, P= 0.001; Table 2A). Den-
sities of Bachman’s sparrows were greater in
stands during the first 3growing seasons after
burning than they were in stands ≥4 growing sea-
sons after burning (P≤0.044; Table 2A).
Similarly, the ANOVA model for breeding sea-
son 2000 was significant (F7, 172 = 3.948, P<0.001).
Density of Bachman’s sparrows did not differ be-
tween seasons of burning (F1, 172 = 0.168, P= 0.683)
but did differ among post-burn ages (F3, 172 =
7.037, P<0.001). Density of Bachman’s sparrows
was greater at stands 2years after burning than at
stands 1, 3, and ≥4 years after burning (Table 2B).
A contrast of stands 1, 2, and 3years since burn-
ing with stands ≥4 years since burning revealed
that the results from breeding season 2000 were
consistent with results
from breeding season
1999, and stands ≤3
years since burning had
greater densities of
Bachman’s sparrows than
stands ≥4 years since
burning (F1, 172 = 4.634,
The 2-way ANOVA
examining the influence
of season of burning and
post-burn age found dif-
ferences (P<0.05) in
percent coverage of bare
ground, standing dead
shrubs, forbs, live shrubs,
and vines (Table 3).
None of the models suggested interactions
between season of last burning and post-burn age
(F3, 78 ≤2.419, P≥ 0.073).
Standing dead shrub was the only vegetation
variable to differ (P<0.05) between seasons of
burning (Table 3A), but percent coverage of bare
ground, standing dead shrubs, forbs, live shrubs,
and vines all differed (P<0.05) among post-burn
ages (Table 3B). Bare ground, standing dead
shrubs, and forbs tended to decrease with post-
burn age, and live shrubs and vines tended to
increase with post-burn age (Table 3B).
The linear model containing variables describ-
ing percent coverage of grass (GRASS) and TSB
was identified as the best model for explaining the
variation in density of Bachman’s sparrows (Table
4). The 95% confidence intervals (±2 SE based on
unconditional variances) around the model-aver-
aged parameters suggested that GRASS (β= 1.26,
95% CI: 0.53 to 1.98) had a large influence on den-
sity of Bachman’s sparrows, but TSB (β = –0.0005,
95% CI: –0.0018 to 0.0008) had a relatively weak
influence compared to grass (Fig. 1).
The best model for explaining variation in per-
cent coverage of grass included percent canopy
coverage (CANOPY COVER) and number of
growing-season fires in the previous 10 years
(Table 5). The model that included canopy
COVER, number of growing-season fires, and the
interaction between the 2variables also was a
plausible model (Table 5). The 95% confidence
Table 2. Resultsafrom 2-way analysis of variance comparing mean density (number/ha) of
breeding Bachman’s sparrows by season of burn and years since burning (burn groups) at
stands in the Conecuh National Forest, Alabama, USA, and Blackwater River State Forest,
Florida, USA, during (A) 1999 and (B) 2000.
No. of Burn No. of
Seasons stands Means (±SE) groups stands Meansb(±SE)
(A) Breeding season 1999
Growing (Apr–Sep) 96 0.425 (0.033) 1 40 0.481 (0.050) A
Dormant (Oct–Mar) 84 0.439 (0.035) 2 46 0.529 (0.047) A
3 46 0.453 (0.048) A
≥4 48 0.264 (0.046) B
(B) Breeding season 2000
Growing (Apr–Sep) 88 0.447 (0.045) 1 59 0.449 (0.044) A
Dormant (Oct–Mar) 92 0.471 (0.035) 2 38 0.670 (0.054) B
3 30 0.352 (0.076) A
≥4 53 0.364 (0.047) A
aInteractions between season and burn groups were not significant (
> 0.1), and those
statistics are not presented.
bMeans followed by different letters differed (
< 0.05) in Tukey’s (HSD) multiple compar-
isons. An independent contrast of burn groups 1, 2, and 3 against ≥4 for breeding season
2000 revealed that abundance was greater ≤3 years after burning than ≥4 years after burning
J. Wildl. Manage. 68(4):2004 1119
FIRE AND BACHMAN’S SPARROWS • Tucker et al.
intervals around the
ters indicated that
canopy cover had a large
influence on grass (ß =
–0.960, 95% CI: –1.749
to –0.172), but number
of growing-season fires
(β= –0.016, 95% CI:
–0.164 to 0.132) and the
interaction of canopy
cover and number of
growing-season fires (β=
0.071, 95% CI: –0.094 to
0.236) was relatively weak
compared to canopy
cover. Akaike weights
indicated 3.3and 5.3
times greater support for
the model including the
number of growing-sea-
son fires in the last 10
years versus the top-
ranked models contain-
ing number of dormant-
season fires and total
number of fires, respec-
tively (Table 5). A plot of
regression lines for per-
cent coverage of grass by
percent canopy cover at
the minimum and maxi-
mum number of grow-
Table 3. Means (±SE)aand test statistics from 2-way analysis of variance comparing vegeta-
tion variables by (A) season of burning and (B) years since burning (burn group) at Conecuh
National Forest, Alabama, USA, and Blackwater River State Forest, Florida, USA, during 1999.
Growing Dormant Statisticsd
= 44) (
Grass 72.8 (3.4) 66.3 (3.3) 2.32 0.132
Bare 0.4 (0.2) 0.5 (0.2) 0.14 0.711
Dead 3.2 (0.7) 5.3 (0.7) 4.12 0.046
Fern 11.8 (1.7) 13.2 (1.7) 0.35 0.555
Forb 27.5 (1.9) 23.3 (1.9) 2.33 0.131
Litter 24.9 (2.5) 30.3 (2.4) 3.00 0.087
Shrub 58.7 (2.1) 61.1 (2.0) 0.74 0.393
Vine 10.4 (1.9) 16.5 (1.9) 3.51 0.065
Canopy 37.7 (1.8) 39.1 (1.7) 0.30 0.586
Basal area 54.9 (2.5) 54.1 (2.5) 0.06 0.808
Group 1 Group 2 Group 3 Group 4 Statisticsd
= 20) (
= 16) (
= 29) (
Grass 72.8 (4.7) 67.7(5.5) 69.9 (3.9) 67.7 (4.8) 0.19 0.899
Bare 1.2 (0.2) A 0.2 (0.3) B 0.2 (0.2) B 0.3 (0.2) B 4.93 0.003
Dead 9.4 (1.096) A 4.8 (1.1) B 2.2 (0.8) B 0.6 (1.0) C 19.53 < 0.001
Fern 13.9 (2.4) 11.7 (2.7) 9.6 (2.0) 14.9 (2.4) 0.91 0.440
Forb 32.1 (2.7) A 20.8 (3.1) BC 23.1 (2.3) BC 25.6 (2.7) AB 3.34 0.023
Litter 22.6 (3.5) 26.2 (4.0) 29.0 (2.9) 32.6 (3.5) 1.63 0.190
Shrub 46.0 (2.9) A 56.5 (3.3) B 67.2 (2.4) C 69.8 (2.9) C 14.27 < 0.001
Vine 6.3 (2.7) A 12.3 (3.1) AB 14.6 (2.2) B 20.7 (2.7) B 5.08 0.003
Canopy 39.1 (2.5) 41.6 (2.9) 38.1 (2.1) 34.9 (2.5) 1.12 0.347
Basal area 54.8 (3.5) 59.4 (4.0) 54.9 (2.9) 48.8 (3.5) 1.36 0.261
aMeans are for untransformed data. Arcsine square root of proportions used in analyses of
percentages, and natural logarithm used for basal area.
b Variables are percent cover, except for basal area (feet2/acre).
cGrowing season = Apr–Sep; Dormant season = Oct–Mar.
dInteraction terms were not significant (
> 0.05) for any analysis and are not presented.
e Within rows, means not sharing a common letter differ (
< 0.05) in Tukey’s (HSD) multi-
Table 4. Variables, number of parameters (
), coefficient of determination (
2), Akaike’s Information Criterion adjusted for small
sample size (AIC
, and AIC
) for the a priori set of candidate models considered in the analysis examining
the influence of percent coverage of grass, season of burning, and number of growing season days since burning (TSB) on den-
sity of Bachman’s sparrows in longleaf pine stands (
= 86) in the Conecuh National Forest, Alabama, USA, and Blackwater River
State Forest, Florida, USA, during 1999. All models include intercept and parameter for residual variance.
TSB Grass 4 0.306 –260.999 0.000 0.466
TSB Grass TSB*Grass 5 0.308 –258.926 2.073 0.165
TSB Grass Season 5 0.307 –258.761 2.238 0.152
TSB Grass Season Season*Grass 6 0.308 –256.668 4.331 0.053
TSB Grass Season TSB*Grass 6 0.308 –256.638 4.361 0.053
TSB Grass Season Season*TSB 6 0.307 –256.448 4.551 0.048
TSB Grass Season TSB*Grass Season*Grass 7 0.311 –254.658 6.341 0.020
TSB Grass Season Season*TSB Season*Grass 7 0.308 –254.295 6.703 0.016
TSB Grass Season TSB*Season TSB*Grass 7 0.308 –254.265 6.734 0.016
TSB Grass Season TSB*Grass Season*Grass Season*TSB 8 0.311 –252.224 8.775 0.006
Grass 3 0.202 –251.125 9.873 0.003
Grass Season 4 0.207 –249.455 11.543 0.001
Grass Season Grass*Season 5 0.208 –247.359 13.639 0.001
TSB 3 0.102 –241.008 19.991 <0.001
TSB Season 4 0.108 –239.326 21.673 <0.001
TSB Season TSB*Season 5 0.108 –237.071 23.927 <0.001
Season 3 0 –231.742 29.257 <0.001
J. Wildl. Manage. 68(4):20041120 FIRE AND BACHMAN’S SPARROWS • Tucker et al.
ing-season fires revealed that the interaction of
canopy cover and growing-season fires resulted
from a greater influence of canopy cover at
stands that were burned fewer times in the grow-
ing season (Fig. 2A).
We acknowledge that our methodology for sam-
pling Bachman’s sparrows may have resulted in
overestimation of densities. We did not separate
detections of Bachman’s sparrows before and after
broadcasting the tape-recorded song, but we
believe that any bias resulting from luring birds
closer from beyond 100 m was equal across stands.
We did not find significant effects of covariates
on probability of detection, suggesting that
detection of the playback by Bachman’s sparrows
also was equal across stands. Thus, we believe that
our results are valid, but our estimates of density
should be viewed as relative rather than absolute.
The need for frequent fires to maintain the lon-
gleaf pine ecosystem has been well documented
(Engstrom et al. 1984). Although evidence sug-
gests that most naturally occurring fires would have
occurred during the growing season, the season of
greatest lightning activity (Robbins and Myers
1992), our results suggest that season of burning
had little effect on vegetation structure (Table 3) or
density of Bachman’s sparrows (Table 2). However,
the analyses focused on the season of last burning,
and most stands had burn histories that included
both dormant-season and growing-season fires.
Thus, results might have differed if stands had long
histories of burning exclusively in 1season or the
other. Results from modeling percent coverage
of grass by number of growing-season fires in the
previous 10 years (Fig. 2B) also suggest that
results for season of burning may have been con-
founded by variation in burning histories.
We did not examine density of Bachman’s spar-
rows immediately after growing-season fires. Thus,
future research should focus on examining the
influence of timing of fire within the growing sea-
son. For example, fires in late April and early May
could destroy a high percentage of reproductive
effort late in the nestling or early fledgling stages
of the first nesting cycle (Tucker 2002) and result
in low annual recruitment if a high percentage of
results from early-season
nesting attempts (see Sto-
ber 1996). Shriver et al.
(1996, 1999) found that
burning Florida dry
prairies in mid-June
resulted in an extended
breeding season for
Florida grasshopper spar-
rows (Ammodramus savan-
narum floridanus) whereas
burning in July did not.
Our analysis of TSB
suggested that density of
Bachman’s sparrows be-
gan to rapidly decline 3
years after burning
(Table 2). Although this
trend was clear during
Fig. 1. Regression lines for minimum (20%) and maximum
(98%) measures of percent coverage of grass recorded in lon-
gleaf pine stands in the Conecuh National Forest, Alabama,
USA, and Blackwater River State Forest, Florida, USA, as a
function of number of growing-season days since burning
(TSB) show the positive influence of grass and negative influ-
ence of TSB on density (birds/ha) of Bachman’s sparrows.
Table 5.Variables, number of parameters (
), coefficient of determination (
2), Akaike’s Infor-
mation Criterion adjusted for small sample size (AIC
, and AIC
) for the a
priori set of candidate models considered in the analysis examining the influence of percent
canopy cover and burn historyaon percent coverage of grass in longleaf pine stands (
in the Conecuh National Forest, Alabama, USA, and Blackwater River State Forest, Florida,
USA, during 1999.
Variables in modelb
Growing10 Canopy 4 0.3501 –273.540 0 0.399
Growing10 Canopy Grow10*Canopy 5 0.3601 –272.623 0.916 0.252
Dormant10 Canopy Dorm10*Canopy 5 0.3492 –271.169 2.371 0.122
Total10 Canopy Total10*Canopy 5 0.3417 –270.184 3.356 0.075
Total10 Canopy 4 0.3216 –269.855 3.685 0.063
Dormant10 Canopy 4 0.3187 –269.483 4.056 0.052
Canopy 3 0.2952 –268.763 4.776 0.037
Growing10 3 0.0928 –247.054 26.486 7 × 10–7
Dormant10 3 0.0832 –246.152 27.387 5 × 10–7
Total10 3 0.0206 –240.474 33.065 3 × 10–8
aBurn history variables were the number of times a stand was burned in the previous 10
years during the growing season (Growing10), dormant season (Dormant10), and both sea-
sons combined (Total10).
bAll models include an intercept and parameter for residual variance.
J. Wildl. Manage. 68(4):2004 1121
FIRE AND BACHMAN’S SPARROWS • Tucker et al.
the breeding season of
1999, the trend was not
as evident during breed-
ing season 2000. Tucker
(2002) found that repro-
ductive success appeared
to mirror abundance of
Bachman’s sparrows in
Conecuh. Studies of
Bachman’s sparrows on
Florida dry prairies found
both breeding densities
and reproductive success
were similar during the
first 3breeding seasons
son fires (Shriver and
Vickery 2001), but densi-
ties increased immediate-
ly following mid-June and
July fires relative to sites
≥2 years since burning
(Shriver et al. 1999). Gob-
ris (1992) also found that
densities of Bachman’s
sparrows in loblolly pine
(Pinus taeda) forests in
Georgia were greatest the
first 3years after burning.
As suggested by previ-
ous studies of Bachman’s
sparrows (e.g., Haggerty
2000), percent coverage
of grass appeared to have a large influence on the
presence of Bachman’s sparrows (Fig. 1). Hagger-
ty (2000) found that percent coverage of grass and
litter were the only vegetation measurements that
consistently appeared important across the geo-
graphic distribution of Bachman’s sparrows.
Although density of Bachman’s sparrows may be
correlated with several measures of vegetation
structure other than grass (e.g., shrub cover), only
percent coverage of grass was included in the a pri-
ori list of candidate models. This is because most
variables that are correlated with density of Bach-
man’s sparrows either influence coverage of grass
(e.g., canopy and shrub cover) or are influenced
similarly to grass by these variables (e.g., forb
cover). Litter cover was not considered in the mod-
els because it was predicted to increase with TSB;
that is, litter will be consumed by fire and then
begin accumulating after fire. Our analyses of the
influence of season and post-burn age on vegeta-
tion structure (Table 3) largely concur with this
logic. For example, coverage by grasses and forbs
tended to be greatest in the first few years after
burning and then began to decline, whereas cov-
erage by shrubs, vines, and litter tended to increase
with time since burning (Table 3B). Surprisingly,
we found few differences in vegetation measure-
ments between seasons of burning, but most
trends with TSB followed patterns that would be
predicted following fire (see Engstrom et al. 1984).
Our study suggests that optimal habitat for
Bachman’s sparrows in longleaf pine forests is
maintained by burning on a 2- or 3-year rotation;
density of Bachman’s sparrows rapidly declined
with burning rotations >3 years. Season of burn-
ing appeared to have little influence on density
of Bachman’s sparrows, but additional research
focusing on timing of fire within the growing sea-
son is needed. In particular, future research
should address the influence of timing of fire
Fig. 2. Regression lines for minimum and maximum values of (A) number of growing-season
fires in the previous 10 years, and (B) percent canopy cover suggest that percent coverage
by grass (arcsine square root transformed) was affected more strongly by canopy cover when
stands were burned fewer times during the growing season and/or by number of fires in the
growing season at stands with greater canopy cover.
J. Wildl. Manage. 68(4):20041122 FIRE AND BACHMAN’S SPARROWS • Tucker et al.
within the growing season on the reproductive
success of Bachman’s sparrows.
Our conclusions have strong implications for
the management and restoration of longleaf pine
communities. Similar to most species associated
with the longleaf pine ecosystem, including many
rare and endemic species, Bachman’s sparrows
depend on the condition of herbaceous ground
cover. Thus, Bachman’s sparrows probably serve
as a good indicator for evaluating the influence
of management activities on much of the diversi-
ty within longleaf pine forests.
Because evidence suggests that most naturally
occurring fires would have occurred during the
growing season, and we did not find evidence for
adverse effects of growing-season fires on density
of Bachman’s sparrows, the optimal strategy for
burning longleaf pine forests appears to be grow-
ing-season fires on a 2- or 3-year rotation. If con-
straints (e.g., drought) prevent burning during the
growing season, burning during the dormant sea-
son probably will be more beneficial than post-
poning until suitable conditions occur during a
future growing season. Considerations of the pos-
sible effects on species-of-concern wintering in
longleaf ecosystems should be dealt with on a
case-by-case basis (Tucker and Robinson 2003).
Regardless of season of burning, longleaf pine
forests should be burned every 2or 3years to
maintain high-quality habitat for Bachman’s
sparrows. As a conservative measure, annual burn-
ing—especially during the growing season—should
be discouraged until information is available
addressing long-term effects. An exception to
annual burning might be during the initial stages
of restoration. Finally, our research was in the
region containing the greatest concentration of
longleaf pine that remains (Outcalt and Sheffield
1996), and the large-scale fires (≥400 ha) that typ-
ically occur in this region did not appear to have
negative effects. However, spatial scale of burning
might be a greater concern for fragmented land-
scapes where suitable unburned stands do not
occur in close proximity (Seaman 1998).
This research was supported by grants from the
U.S. Geological Survey, Biological Resource Divi-
sion, Species at Risk Program; a Frances M. Pea-
cock Scholarship from the Garden Club of Amer-
ica administered through the Cornell Laboratory
of Ornithology; and the Walter F. Coxe Research
Fund of the Birmingham Audubon Society. The
research has contributed greatly from partner-
ships among the following organizations and
agencies: Department of Biological Sciences,
Auburn University; Solon Dixon Forestry Educa-
tion Center, School of Forestry and Wildlife Sci-
ences, Auburn University; Alabama Cooperative
Fish and Wildlife Research Unit, Auburn Univer-
sity; U.S. Geological Survey, Biological Resource
Division; U.S. Department of Agriculture, Forest
Service, Conecuh National Forest; Florida Depart-
ment of Agriculture and Consumer Services,
Blackwater River State Forest; Alabama Depart-
ment of Conservation and Natural Resources,
Division of Wildlife and Freshwater Fisheries; and
the Longleaf Alliance. R. Johnson, R. Lint, T.
Arrington, P. Brinn, C. Cook, J. Klempa, and R.
Mullins provided logistic support during field
work. Assistance of J. Doherty in sampling vege-
tation is gratefully acknowledged.
ANDERSON, D. R., K. P. BURNHAM, AND W. L. THOMPSON.
2000. Null hypothesis testing: problems, prevalence,
and an alternative. Journal of Wildlife Management
BUCKLAND, S. T., D. R. ANDERSON, K. P. BURNHAM, J. L.
LAAKE, D. L. BORCHERS, AND L. THOMAS. 2001. Intro-
duction to distance sampling: estimating abundance
of biological populations. Oxford University Press,
New York, New York, USA.
BURNHAM, K. P., AND D. R. ANDERSON. 2002. Model selec-
tion and multimodel inference: a practical informa-
tion–theoretic approach. Second edition. Springer-
Verlag, New York, New York, USA.
CHAPMAN, H. H. 1932. Is the longleaf type a climax?
DUNNING, J. B. 1993. Bachman’s sparrow. Number 38 in
A. Poole, P. Stettenheim, and F. Gill, editors. The birds
of North America. The Academy of Natural Sciences,
Philadelphia, Pennsylvania, USA, and The American
Ornithologists’ Union, Washington, D.C., USA.
———, AND B. D. WATTS. 1990. Regional differences in
habitat occupancy by Bachman’s sparrow. Auk
ENGSTROM, R. T., R. L. CRAWFORD, AND W. W. BAKER.
1984. Breeding bird populations in relation to chang-
ing forest structure following fire exclusion: a 15-year
study. Wilson Bulletin 96:437–450.
GOBRIS, N. M. 1992. Habitat occupancy during the
breeding season by Bachman’s sparrow at Piedmont
National Wildlife Refuge in central Georgia. Thesis,
University of Georgia, Athens, Georgia, USA.
HAGGERTY, T. M. 2000. A geographical study of the veg-
etation structure of Bachman’s sparrow (Aimophila
aestivalis) breeding habitat. Journal of the Alabama
Academy of Science 71:120–129.
HUNTER, W. C., A. J. MUELLER, AND C. L. HARDY. 1994.
Managing for red-cockaded woodpeckers and
Neotropical migrants—is there a conflict? Proceed-
ings of the Annual Conference of the Southeastern
Association of Fish and Wildlife Agencies 48:383–394.
JACKSON, J. A. 1988. The southeastern pine forest ecosys-
J. Wildl. Manage. 68(4):2004 1123
FIRE AND BACHMAN’S SPARROWS • Tucker et al.
tem and its birds: past, present, and future. Pages
119–159 in J. A. Jackson, editor. Bird conservation 3.
International Council for Bird Preservation, Univer-
sity of Wisconsin, Madison, Wisconsin, USA.
JAMES, F. C., C. A. HESS, AND D. KUFRIN.1997. Species-
centered environmental analysis: indirect effects of
fire history on red-cockaded woodpeckers. Ecological
KRECH, S., III. 1999. The ecological Indian: myth and
history. W. W. Norton, New York, New York, USA.
NOSS, R. F. 1989. Longleaf pine and wiregrass: keystone
components of an endangered ecosystem. Natural
Areas Journal 9:211–213.
OUTCALT, K. W., AND R. M. SHEFFIELD. 1996. The longleaf
pine forest: trends and current conditions. U.S. For-
est Service, Research Bulletin SRS-9.
PEET, R. K., AND D. J. ALLARD.1993. Longleaf pine vege-
tation of the southern Atlantic and eastern Gulf Coast
regions: a preliminary classification. Proceedings of
the Tall Timbers Fire Ecology Conference 18:45–81.
PLATT, W. J., G. W. EVANS, AND M. M. DAVIS.1988. Effects
of fire season on flowering of forbs and shrubs in lon-
gleaf pine forests. Oecologia 76:353–363.
PLENTOVICH, S., J. W. TUCKER, JR., N. R. HOLLER, AND
G. E. HILL. 1998. Enhancing Bachman’s sparrow habi-
tat via management of red-cockaded woodpeckers.
Journal of Wildlife Management 62:347–354.
ROBBINS, L. E., AND R. L. MYERS.1992. Seasonal effects of
prescribed burning in Florida: a review. Miscella-
neous Publication 8. Tall Timbers Research Station,
Tallahassee, Florida, USA.
SAS INSTITUTE.1999. SAS OnlineDoc. Version 8. SAS
Institute, Cary, North Carolina, USA.
SEAMAN, B. D. 1998. Movements and survival of Bach-
man’s sparrows in response to prescribed summer
burns in South Carolina. Thesis, University of Geor-
gia, Athens, Georgia, USA.
SHRIVER, W. G., AND P. D. VICKERY.2001. Response of
breeding Florida grasshopper and Bachman’s spar-
rows to winter prescribed burning. Journal of Wildlife
———, ———, AND S. A. HEDGES. 1996. Effects of sum-
mer burns on Florida grasshopper sparrows. Florida
Field Naturalist 24:68–73.
———, ———, AND D. W. PERKINS. 1999. The effects of
summer burns on breeding Florida grasshopper and
Bachman’s sparrows. Studies in Avian Biology
SIMBERLOFF, D. 1993. Species-area and fragmentation
effects on old-growth forests: prospects for longleaf
pine communities. Proceeding of the Tall Timbers
Fire Ecology Conference 18:1–13.
STOBER, J. M. 1996. Territory dynamics and basic biolo-
gy of the Bachman’s sparrow (Aimophila aestivalis) at
the Savannah River Site, South Carolina. Thesis, Uni-
versity of Georgia, Athens, Georgia, USA.
STRENG, D. R., J. S. GLITZENSTEIN, AND W. J. PLATT. 1993.
Evaluating effects of season of burn in longleaf pine
forests: a critical literature review and some results
from an ongoing long-term study. Proceedings of the
Tall Timbers Fire Ecology Conference 18:227–263.
THOMAS, L., J. L. LAAKE, S. STRINDBERG, F. F. C. M ARQUES,
S. T. BUCKLAND, D. L. BORCHERS, D. R. ANDERSON, K. P.
BURNHAM, S. L. HEDLEY, J. H. POLLARD, AND J. R. B.
BISHOP. 2003. Distance 4.1. Release 2. Research Unit
for Wildlife Population Assessment, University of St.
Andrews, United Kingdom. [http:
TUCKER, J. W., JR. 2002. Influence of season and fre-
quency of fire on Bachman’s and Henslow’s sparrows
in longleaf pine forests of the Gulf Coastal Plain. Dis-
sertation, Auburn University, Auburn, Alabama, USA.
———, G. E. HILL, AND N. R. HOLLER. 1998. Managing
mid-rotation pine plantations to enhance Bachman’s
sparrow habitat. Wildlife Society Bulletin 26:342–348.
———, ———, AND ———. 2003. Longleaf pine
restoration: implications for landscape-level effects
on bird communities in the Lower Gulf Coastal Plain.
Southern Journal of Applied Forestry 27:107–121.
———, AND W. D. ROBINSON. 2003. Influence of season
and frequency of fire on Henslow’s sparrows winter-
ing on Gulf Coast pitcher plant bogs. Auk 120:96–106.
WAHLENBERG, W. G. 1946. Longleaf pine: its use, ecology,
regeneration, protection, growth, and management.
Charles Lathrop Pack Forestry Foundation, Washing-
ton, D.C., USA.
WALDROP, T. A., D. L. WHITE, AND S. M. JONES. 1992. Fire
regimes for pine–grassland communities in the
southeastern United States. Forest Ecology and Man-
ZAR, J. H. 1984. Biostatistical analysis. Prentice-Hall,
Englewood Cliffs, New Jersey, USA.
Received 4 March 2003.
Accepted 6 July 2004.
Associate Editor: Flaspohler.