Avifaunal responses to fire in Southwestern montane forests along a burn severity gradient

Article (PDF Available)inEcological Applications 17(2):491-507 · April 2007with58 Reads
DOI: 10.1890/06-0253 · Source: PubMed
Abstract
The effects of burn severity on avian communities are poorly understood, yet this information is crucial to fire management programs. To quantify avian response patterns along a burn severity gradient, we sampled 49 random plots (2001-2002) at the 17351-ha Cerro Grande Fire (2000) in New Mexico, USA. Additionally, pre-fire avian surveys (1986-1988, 1990) created a unique opportunity to quantify avifaunal changes in 13 pre-fire transects (resampled in 2002) and to compare two designs for analyzing the effects of unplanned disturbances: after-only analysis and before-after comparisons. Distance analysis was used to calculate densities. We analyzed after-only densities for 21 species using gradient analysis, which detected a broad range of responses to increasing burn severity: (I) large significant declines, (II) weak, but significant declines, (III) no significant density changes, (IV) peak densities in low- or moderate-severity patches, (V) weak, but significant increases, and (VI) large significant increases. Overall, 71% of the species included in the after-only gradient analysis exhibited either positive or neutral density responses to fire effects across all or portions of the severity gradient (responses III-VI). We used pre/post pairs analysis to quantify density changes for 15 species using before-after comparisons; spatiotemporal variation in densities was large and confounded fire effects for most species. Only four species demonstrated significant effects of burn severity, and their densities were all higher in burned compared to unburned forests. Pre- and post-fire community similarity was high except in high-severity areas. Species richness was similar pre- and post-fire across all burn severities. Thus, ecosystem restoration programs based on the assumption that recent severe fires in Southwestern ponderosa pine forests have overriding negative ecological effects are not supported by our study of post-fire avian communities. This study illustrates the importance of quantifying burn severity and controlling confounding sources of spatiotemporal variation in studies of fire effects. After-only gradient analysis can be an efficient tool for quantifying fire effects. This analysis can also augment historical data sets that have small samples sizes coupled with high non-process variation, which limits the power of before-after comparisons.

Figures

Ecological Applications, 17(2), 2007, pp. 491–507
Ó 2007 by the Ecological Society of America
AVIFAUNAL RESPONSES TO FIRE IN SOUTHWESTERN MONTANE
FORESTS ALONG A BURN SEVERITY GRADIENT
NATASHA B. KOTLIAR,
1,3
PATRICIA L. KENNEDY,
2
AND KIMBERLY FERREE
1
1
U.S. Geological Survey, 2150 Centre Avenue, Building C, Fort Collins, Colorado 80526 USA
2
Eastern Oregon Agricultural Research Center and Department of Fisheries and Wildlife, Oregon State University, P.O. Box E,
372 South 10th Street, Union, Oregon 97883 USA
Abstract. The effects of burn severity on avian communities are poorly understood, yet
this information is crucial to fire management programs. To quantify avian response patterns
along a burn severity gradient, we sampled 49 random plots (2001–2002) at the 17 351-ha
Cerro Grande Fire (2000) in New Mexico, USA. Additionally, pre-fire avian surveys (1986–
1988, 1990) created a unique opportunity to quantify avifaunal changes in 13 pre-fire
transects (resampled in 2002) and to compare two designs for analyzing the effects of
unplanned disturbances: after-only analysis and before–after comparisons. Distance analysis
was used to calculate densities. We analyzed after-only densities for 21 species using gradient
analysis, which detected a broad range of responses to increasing burn severity: (I) large
significant declines, (II) weak, but significant declines, (III) no significant density changes,
(IV) peak densities in low- or moderate-severity patches, (V) weak, but significant increases,
and (VI) large significant increases. Overall, 71% of the species included in the after-only
gradient analysis exhibited either positive or neutral density responses to fire effects across all
or portions of the severity gradient (responses III–VI). We used pre/post pairs analysis to
quantify density changes for 15 species using before–after comparisons; spatiotemporal
variation in densities was large and confounded fire effects for most species. Only four
species demonstrated significant effects of burn severity, and their densities were all higher in
burned compared to unburned forests. Pre- and post-fire community similarity was high
except in high-severity areas. Species richness was similar pre- and post-fire across all burn
severities. Thus, ecosystem restoration programs based on the assumption that recent severe
fires in Southwestern ponderosa pine forests have overriding negative ecological effects are
not supported by our study of post-fire avian communities. This study illustrates the
importance of quantifying burn severity and controlling confounding sources of spatiotem-
poral variation in studies of fire effects. After-only gradient analysis can be an efficient tool
for quantifying fire effects. This analysis can also augment historical data sets that have small
samples sizes coupled with high non-process variation, which limits the power of before–after
comparisons.
Key words: after-only, before–after comparisons; bird communities; breeding densities; Cerro Grande,
New Mexico, USA; distance sampling; fire effects; gradient analysis; historical data.
INTRODUCTION
In May of 2000, the Cerro Grande fire in New
Mexico, USA, made national headlines (e.g., Janofsky
2000) when it burned over 17 000 ha, threatened Los
Alamos National Laboratory, and destroyed 235 homes
(information available online).
4
Beginning as a pre-
scribed fire at Bandelier National Monument, the Cerro
Grande fire was a harbinger of a particularly severe fire
season; 3.4 million ha burned in 2000, exceeding the 10-
year average (1992–2002) by .1.5 million ha (informa-
tion available online).
5
The severity of the 2000 and 2002
fire seasons (collectively 6.8 million ha burned) has been
used as evidence to support the assertion that decades of
fire suppression, logging, and grazing practices have led
to increased size and severity of wildland fires and
prompted major fuel reduction initiatives including the
National Fire Plan and Healthy Forest Restoration Act
(Graham et al. 2004, Schoennagel et al. 2004). Primary
targets of such programs are montane forests that
historically burned frequently, but with low severity
(Keane et al. 2002, Graham et al. 2004). In particular,
Southwestern ponderosa pine forests (like those that
burned in the Cerro Grande fire) have been character-
ized as unnaturally dense due to fire exclusion since
Euro-American settlement (Moore et al. 1999, Allen et
al. 2002). Consequently, open, park-like forests main-
tained by frequent surface fires have been replaced in
many areas by dense stands capable of supporting
crown fires (Allen et al. 2002, Schoennagel et al. 2004).
A primary goal of many ecosystem restoration pro-
Manuscript received 15 February 2006; revised 14 June 2006;
accepted 23 June 2006. Corresponding Editor: T. R. Simons.
3
E-mail: kotliart@usgs.gov
4
hwww.nps.gov/cerrogrande/i
5
hwww.nifc.go vi
491
grams is to reduce tree densities, thereby reducing the
risk of severe wildland fire and allowing the reintroduc-
tion of prescribed understory fire (Covington et al.
1997).
Recent research, however, has begun to challenge the
generality of the ‘‘Southwest paradigm,’ and an
alternative view is beginning to emerge of more variable
historical fire regimes, which included crown fire in low-
elevation montane forests (Baker and Ehle 2001,
Schoennagel et al. 2004). Yet, this variation is often
ignored in ecosystem restoration programs (Schoennagel
et al. 2004). There is historical precedence for severe fires
in montane systems (Pierce et al. 2004), consequently,
the relatively limited time frame used as a reference for
restoration of historical fire regimes may not adequately
reflect current climatic conditions (Tiedemann et al.
2000, Wagner et al. 2000, Pierce et al. 2004). Thus, the
relative importance of climate and fuels in dictating
recent fire activity is uncertain (Veblen et al. 2000).
A corollary of the Southwest paradigm is the
assumption that both fire exclusion and the greater
severity of recent fires have predominantly negative
ecological effects. This assumption, although largely
untested, is a fundamental justification for ecosystem
restoration programs (e.g., Covington et al. 1997, Fule
´
et al. 1997, Moore et al. 1999, Keane et al. 2002,
Graham et al. 2004). Because fire ecology has focused on
vegetation structure and dynamics, our current under-
standing of fire effects on avifauna in western forests is
quite limited. Recent reviews found a limited number of
studies comparing burned and unburned forests, most of
which sampled single burns and lacked suitable replica-
tion (Hutto 1995, Finch et al. 1997, Kotliar et al. 2002,
Saab and Powell 2005); in the Rocky Mountains, many
of the studies sampled burns ,500 ha and burn severity
was not quantified (Kotliar et al. 2002, Saab et al. 2005).
Thus, the general avifaunal response patterns suggested
by such qualitative reviews are preliminary and over-
simplified because burn severity has typically been
treated as a binomial variable.
Wildfires are unplanned events; consequently, pre-fire
data rarely exist and fire effects are typically evaluated
using an impact–reference approach (Wiens and Parker
1995; also called ‘after-only,’ Osenberg et al. 1994) in
which burned sites are compared to unburned reference
areas. We used this approach by conducting post-fire
surveys at Cerro Grande. In addition, the existence of
pre-fire data on avian populations and forest types from
the area of the Cerro Grande fire created a unique
opportunity to evaluate post-fire changes in avian
communities by resampling the pre-fire study area. The
comparison of before and after data in burned and
unburned sites approximates a Before After Control
Impact (BACI) design for planned impacts (Stewart-
Oaten et al. 1986, Wiens and Parker 1995). Before–after
comparisons have advantages over impact–reference
designs because temporal and pre-impact spatial varia-
tion in populations can be quantified, whereas randomly
located samples using the impact–reference approach
avoids pseudoreplication problems inherent in before–
after comparisons (Wiens and Parker 1995).
Our primary objective was to characterize avifaunal
responses to fire by examining how densities changed
along a burn severity gradient and by comparing
community composition pre- and post-fire. We also
characterized the spectrum of observed avian response
patterns, which we present as a comparative framework
for studies on the effects of burn severity. Our second
objective was to compare the advantages and disadvan-
tages of after-only and before–after designs in charac-
terizing avifaunal response to fire. We include a recent
study of fire effects using the before–after design for a
mixed-severity regime in the northern Rocky Mountains
(Smucker et al. 2005) as a part of this evaluation. We
discuss the implications of our results for fire research
and management practices.
M
ETHODS
Study area
The study area is in the Jeme z Mountains and
adjacent Pajarito Plateau in north-central New Mexico,
ranging in elevation from 1645 m to 3200 m (Fig. 1). The
Jemez Mountains were formed by volcanic activity and
are dissected by steep-walled canyons (Allen 1989).
Mean annual precipitation is 45 cm. Ponderosa pine
(Pinus ponderosa), Douglas-fir (Pseudotsuga menziesii),
and white fir (Abies concolor) are the most prevalent
forest cover types found over the study area. Engelmann
spruce–subalpine fir (Picea engelmanniA. lasiocarpa),
pinyon–juniper (Pinus edulisJuniperus spp.), juniper–
grassland, and riparian habitats are also present (Siders
and Kennedy 1996).
Pre-fire sampling
Pre-fire sampling of montane forests (1901–2493 m
elevation) included areas that subsequently burned, as
well as nearby areas that did not burn and served as
reference areas. In 1986–1988 and 1990, Kennedy
conducted avian surveys and qualitatively classified the
vegetation in canyon bottoms and adjacent upland
forests in 23 transects on the Los Alamos National
Laboratory, Santa Fe National Forest, Bandelier
National Monument, San Ildefonso Pueblo, and Santa
Clara Pueblo (Fig. 1, and Fig. A1 in Appendix A). The
nonrandom sampling design was developed to estimate
prey availability within the home range of radio-tagged
accipiters nesting in the area (Morrison and Kennedy
1989, Kennedy 1991). Sampling points were located at
200-m intervals along these transects; the number of
points surveyed in these transects varied from 12 to 48.
In all four years, one observer surveyed birds using 50-m
radius point counts of six-minute duration. In 1988 and
1990, each detection was assigned to a distance category
(0–25 m, 26–50 m). Sampling periods varied across
years; in 1986–1988 there were three sampling periods
(May/June, July, Aug/Sept), whereas in 1990 sampling
NATASHA B. KOTLIAR ET AL.492
Ecological Applications
Vol. 17, No. 2
occurred in July. Cover type at each point was classified
as canyon bottom [(1) ephemeral stream with a ponder-
osa pine overstory, (2) ephemeral stream with a mixed-
conifer overstory, or (3) perennial stream with a mixed-
coniferous and deciduous overstory] or upland forest
[(1) ponderosa pine or (2) mixed coniferous].
Post-fire sampling
Hand-drawn maps of the pre-fire transects were
digitized from USGS topographic maps at 1:24 000. In
2001, the starting point of each transect was located and
marked using a GPS. All subsequent post-fire sample
points were spaced ;200 m from the initial point. Due
to access restrictions at Los Alamos National Labora-
tory, San Ildefonso Pueblo, and Santa Clara Pueblo,
four of the 23 pre-fire transects were not resampled.
Three unburned transects at considerable distances from
the Cerro Grande Fire and three transects that had been
altered by anthr opogenic activities were also n ot
resampled. The remaining 13 transects were surveyed
by two observers in June 2002 (hereafter ‘‘before–after’’;
Fig. 1; Appendix A). Each point was surveyed for six
minutes. To en hance the precision of the density
estimates using distance sampling, we expanded the
number of distance categories sampled from two to five:
0–10 m, 11–25 m, 26–50 m, 51–75 m, 76–100 m
(Buckland et al. 2001).
FIG. 1. Sampling locations in relation to the burn-severity map based on the D Normalized Burn Ratio. Only plots and
transects ,1 km from the burn perimeter are shown (see Fig. A1 in Appendix A for unburned transects .1 km from burn
perimeter).
March 2007 493AVIFAUNAL RESPONSES TO BURN SEVERITY
In June 2001 and 2002, bird surveys were conducted
at 49 plots at Cerro Grande (hereafter ‘after-only’’).
Plots were randomly located on burned and nearby
unburned areas within Santa Fe National Forest and
Bandelier National Monument (Fig. 1, Fig. A1). Plot
elevations ranged from 2270 m to 2970 m. Each 200 m 3
500 m plot contained four overlapping 100-m radius
survey points, with centers spaced 100 m apart. All plots
were visited twice each year; during the first visit, the
first and third points were surveyed by one observer,
during the second visit the second and fourth points
were surveyed by a second observer (four observers total
over two years). Each after-only point was surveyed for
10 minutes and used the same distance categories as the
post-fire transects. Cover t ype (upland on ly) was
classified as ponderosa or mixed-coniferous.
Canopy and understory burn severity was ranked in 15-
m radius plots centered at each before–after (2002) and
after-only (2001) survey point (canopy ranks, 0 ¼ no
crown scorch, 1 ¼ some trees partial crown scorch, 1.5 ¼
all trees partial scorch, 2 ¼all trees complete scorch, 2.5 ¼
all needles consumed, 3 ¼ all needles and small branches
consumed; understory ranks, 0 ¼ no burn, 1 ¼ needles
scorched, 2 ¼ needles consumed, 3 ¼ needles and small
branches consumed). Intermediate values were assigned if
conditions included characteristics of adjacent ranks.
Field ranks were used to categorize burn severity into four
levels: unburned (canopy 0, understory 0), low (canopy 0–
1.50, understory 1–1.75), moderate (canopy 1.51–2.50,
understory .1.75), and high (canopy .2.50, understory
.1.75). Burn severity levels were used for categorical
summaries and analysis, whereas burn severity ranks were
used in the analysis of a continuous gradient.
To quantify differences in burn severity among the
before–after and af ter-only sampling areas and to
compare burn severity across spatial scales, we used a
burn severity map (Fig. 1) developed using the
Normalized Burn Ratio (DNBR; Kotliar et al. 2003,
Key and Benson 2005). The average DNBR score was
quantified within 90-m radius plots centered on survey
points. Burn severity ranks for 15-m plots and DNBR
scores for 90-m buffers were highly correlated (Pearson
correlation ¼ 0.912). To quantify burn severity at larger
scales, we classified DNBR scores into four severity
classes using a K means clustering algorithm called
ISODATA (using ERDAS Imagine; Leica Geosystems
2004); canopy burn severity ranks were used to classify
clusters created by the ISODATA algorithm (N. B.
Kotliar, unpublished data). The proportion in each
NBR severity class within 500-m buffers surrounding a
center line running the length of each transect or plot
was used to compare burn severity among sampling
locations.
Statistical analysis
Density estimation by burn severity and cover type.—
Sample units varied by study design. For after-only
densities, plots were used as sample units and canopy
burn severity rank was averaged by plot. Before–after
densities were calculated by transect, stratified by burn
severity level and cover type. We classified each point
within transects by burn severity level. Because fire
behavior varied within transects, more than one burn
severity level was represented in each burned transect
(Fig. 1, Table A1 in Appendix A). To eliminate undue
emphasis on very small samples when estimating
densities, a minimum of four points per severity level
for each transect was established. Isolated points within
transects were switched to the burn severity level of
adjacent points; all nine (3%) points that switched levels
fell along the breakpoint between levels.
We used distance analysis to estimate densities while
controlling for variation in detectability among species,
cover types, and observers (Buckland et al. 2001). In
studies of fire effects, it is critical to test the assumption
of constant detectability because removal of the forest
canopy may increase the probability of detecting birds
and potentially inflate abundance estimates in unadjust-
ed counts. Likewise, in before–after comparisons, it is
necessary to control for pre- and post-perturbation
observer differences (Wiens and Parker 1995).
We used all visual and aural observations of adult
birds. Birds designated as family groups were counted as
a single detection, where distance was estimated to the
center of the group, and birds detected as flocks or flying
over the canopy were eliminated from analysis (Buck-
land et al. 2001). To increase the number of species with
adequate sample sizes (.40; Buckland et al. 2001) for
calculating detection probabilities, we augmented the
Cerro Grande data set with detections from 40 plots in a
companion study at the 2000 Pumpkin fire in Arizona
(see Appendix B for details and number of detections).
Density estimates, log-normal confidence intervals, and
detection probabilities were obtained for pre- and post-
fire data using program DISTANCE, version 4.0
(Thomas et al. 2002). Detection probabilities were
generated for pre-fire and post-fire data separately (see
Appendix B for details on calculation of detection
probabilities). For all detection probabilities, compari-
sons of Akaike’s Information Criterion corrected for
small sample sizes (AIC
c
), chi-square goodness-of-fit
statistics, and visual inspections of probability density
and detection probability plots were used to select the
most parsimonious model and assess overall model fit
(Appendix B; Buckland et al. 2001, Buckland and
Anderson 2002). Because the lowest AIC
c
values were
obtained with the half-normal cosine model, this model
was used to estimate densities.
We conducted preliminary statistical analyses on the
pre-fire data (three-way ANOVA, PROC GLM) to
determine if densities could be pooled over covariates
that might cause significant heterogeneity in the data
(sampling period, year, and cover type). We applied
pre-fire detection probabilities to the full pre-fire data
set (23 transects) to test the null hypothesis that mean
densities for each covariate level were equal; we used a
NATASHA B. KOTLIAR ET AL.494
Ecological Applications
Vol. 17, No. 2
significance level of P ¼ 0.05 for this analysis. Four
transects with pre-fire sampling solely in August/Sep-
tember were dropped because densities differed signif-
icantly from the other two sampling periods and we
lacked a corresponding sampling period post-fire. Thus,
only nine transects could be used to calculate before–
after densities (Appendix A). Because seven of the
remaining transects were sampled only one year pre-fire
and two were sampled two years pre-fire, we pooled
samples across years for all species except Steller’s Jays
(see Appendix C for scientific names) and Western
Bluebirds, which had significant year effects. Finally,
we tested for differences in densities among cover types
for each species and pooled cover types within transects
if densities were not significantly different. We per-
formed pairwise multiple comparisons (Fisher’s least
significant difference [LSD]; PROC LSD) to determine
if densities could be calculated for broader cover type
groups.
Statistical comparison of densities by burn severity.—
There are several techniques to analyze the effects of
perturbations using after-only and before–after data
sets, depending on the study design and objectives
(Wiens and Parker 1995). We used gradient analysis for
the after-only data, which provides greater power when
time series data are not available (Wiens and Parker
1995, Day et al. 1997). For the before–after data, we
used the pre/post pairs analysis to quantify and control
potential sources of temporal and spatial variation,
which can confound or obscure differential responses
to burn severity (Wiens and Parker 1995, Murphy et al.
1997). Densities were log transformed prior to analy-
ses.
For the gradient analysis, we tested for the effects of
burn severity on avian densities by treating mean burn
severity rank per plot as a continuous variable (values
ranged from 0 to 3). We also tested for the effects of
annual variation, cover type (upland ponderosa pine or
mixed coniferous), and all two-way interactions on
transformed de nsities using a general linear model
(PROC GLM; Appendix D) with repeated measures.
A quadratic term, (burn severity)
2
, was included in the
model to test for threshold effects and other nonlinear
response patterns. The general form of the model for the
gradient analysis was
lnðdensityÞ¼b
0
þ b
1
ðburn severityÞþb
2
ðburn severityÞ
2
þ b
3
ðyearÞþb
4
ðcover type Þ
þ b
5
ðburn severity 3 year Þ
þ b
6
ðburn severity 3 cover typeÞ
þ b
7
ðyear 3 cover typeÞ: ð1Þ
For the pre/post pairs analysis, we evaluated differ-
ences in densities by burn severity level while controlling
for cover type variation using a two-way analysis of
variance (ANOVA) with fixed effects (Type III sums-of-
squares F test; PROC MIXED; Appendix E). Burn
severity was treated as a categorical variable because of
sample size limitations. For each spe cies, we used
differences in densities among pre-post pairs (i.e., Eq.
1 in Murphy et al. 1997) at each burn severity level. The
general form of the model for the pre/post pairs analysis
was
lnðdensity
postfire
Þlnðdensity
prefire
Þ
¼ b
0
þ b
1
ðburn severityÞþb
2
ðcover typeÞ
þ b
3
ðburn severity3cover typeÞ: ð2Þ
To compare each level of burn severity, we performed
post hoc, pairwise multiple comparisons tests (Fisher’s
LSD) for species with significant main effects. This
allowed us to examine the sources of variation in more
detail. All statistical tests were conducted using SAS 8.02
(SAS Institute 2001) and SYSTAT 10 (SPSS 2002). The
significance level of all tests of burn severity main effects
and interactions were set a priori at P ¼ 0.1 to better
balance the probabilities of committing Type I and Type
II error. Burn severity was retained in the final models,
while all other covariates were dropped when P 0.10.
Community patterns.—To evaluate the assumption
that recent severe fires like Cerro Grande have negative
effects on avian communities, we compared pre- and
post-fire species richness and community similarity
based on the before–after data. To include all species
in this analysis, we used frequency of occurrence (i.e.,
percentage of points occupied) for all 13 transects
(Appendix F). Because richness is strongly dependent
on sample size, which varied among burn severity levels,
we restricted comparisons within levels. To standardize
sampling efforts, we used only one pre- and one post-fire
visit for each point, and truncated all samples at 50 m.
Pre- and post-fire community similarity was evaluated
by comparing the percentage of species occurring: (1)
both pre- and post-fire, (2) pre-fire only, and (3) post-fire
only.
R
ESULTS
Post-fire density patterns across a burn severity gradient
Collectively, the 21 species analyzed using gradient
analysis represent a continuum of positive and negative
responses across the burn severity gradient (Table 1).
We classified these into six potential response patterns
basedonparameterestimatesofslopesfromthe
gradient analysis (Table 2, Fig. 2). In both years,
densities of Cordilleran Flycatchers, Warbling Vireos,
Mountain Chickadees, Hermit Thrushes, Yellow-
rumped Warblers, and Dark-eyed Juncos decreased
significantly with increasing burn severity (Tables 1, 2,
and Fig. 2a). In 2002 only, American Robins and
Chipping Sparrows demonstrated smaller, but signifi-
cant, decreases in density with increasing burn severity
(Tables 1, 2, and Fig. 2b). Mourning Doves, Northern
Flickers, White-breasted Nuthatches, American Robins
(2001), Western Tanagers, and Virginia’s Warblers
March 2007 495AVIFAUNAL RESPONSES TO BURN SEVERITY
(2001) had similar, but often variable, densities across
the burn severity gradient (Tables 1, 2, and Fig. 2c).
Many species exhibited peak densities in low- or
moderate-severity patches, as indicated by a significant
quadratic term, including Steller’ s Jays, Virginia’s
Warblers (2002), Spotted Towhees (2002), and Black-
headed Grosbeaks (Tables 1, 2, and Fig. 2d). Broad-
tailed Hummingbirds (2002), Hairy Woodpeckers
(2001), and House Wrens (2001) increased moderately
with increasing burn severity (Tables 1, 2, and Fig. 2e).
Western Wood-Pewees and Western Bluebirds were
uncommon in all but the highest burn severity level
(Tables 1, 2, and Fig. 2f). In total, 18 species exhibited
significant burn severity or (burn severity)
2
effects
(Table 2; Appendix D).
The gradient analysis also detected significant tempo-
ral variation (i.e., year main effects or interactions) for
13 species (Table 2; Appendix D). In several cases, the
overall direction of changes was the same across years,
but the magnitude was different. House Wrens shifted
from weak to strong positive responses (Table 2, and
Fig. 2e, f ). Broad-tailed Hummingbirds exhibited more
pronounced increases in 2002 compared to 2001,
whereas Hairy Woodpeckers showed the reverse pattern
(Tables 1, 2). Several spe cies exhib ited signi fican t
patterns for one year only. Virginia’s Warblers and
TABLE 1. Density (means, with SE in parentheses) by burn severity level for after-only data the first year (2001) and the second
year (2002) post-fire at the Cerro Grande Fire, New Mexico, USA.
Species
Density (no. birds/ha)
2001 2002
Unburned
[n ¼ 7]
Low
[n ¼ 12]
Moderate
[n ¼ 13]
High
[n ¼ 17]
Unburned
[n ¼ 7]
Low
[n ¼ 12]
Moderate
[n ¼ 13]
High
[n ¼ 17]
Mourning Dove 0.00 0.14
(0.10)
0.00 0.25
(0.12)
1.23
(0.59)
0.57
(0.22)
0.93
(0.33)
1.67
(0.41)
Broad-tailed Hummingbird 0.27
(0.17)
0.55
(0.14)
0.58
(0.20)
0.83
(0.19)
0.94
(0.29)
1.25
(0.39)
1.37
(0.38)
2.04
(0.27)
Hairy Woodpecker 0.35
(0.24)
0.61
(0.29)
1.07
(0.23)
1.10
(0.27)
1.52
(0.45)
1.70
(0.39)
2.01
(0.38)
2.26
(0.50)
Northern Flicker 0.85
(0.24)
0.00 0.20
(0.09)
0.27
(0.10)
0.47
(0.19)
0.44
(0.12)
0.66
(0.17)
0.62
(0.12)
Western Wood-Pewee 0.00 0.07
(0.07)
0.34
(0.16)
2.03
(0.37)
0.00 0.66
(0.29)
0.95
(0.38)
1.98
(0.40)
Cordilleran Flycatcher 2.44
(0.63)
0.85
(0.25)
0.17
(0.12)
0.40
(0.24)
2.11
(0.68)
1.14
(0.42)
0.26
(0.14)
0.27
(0.12)
Warbling Vireo 3.82
(0.79)
2.65
(0.73)
1.57
(0.52)
0.07
(0.07)
7.64
(1.47)
3.61
(0.77)
2.25
(0.83)
1.20
(0.56)
Steller’s Jay 0.99
(0.69)
1.66
(0.22)
0.99
(0.28)
0.53
(0.15)
0.43
(0.20)
1.66
(0.41)
1.53
(0.52)
0.53
(0.19)
Mountain Chickadee 2.81
(0.76)
0.94
(0.30)
0.58
(0.27)
0.00 2.01
(0.88)
1.64
(0.38)
1.22
(0.42)
0.00
White-breasted Nuthatch 0.32
(0.21)
0.19
(0.19)
0.61
(0.21)
0.40
(0.22)
1.14
(0.00)
0.85
(0.32)
0.79
(0.30)
1.07
(0.30)
House Wren 0.31
(0.16)
0.23
(0.14)
0.17
(0.10)
0.65
(0.21)
0.71
(0.31)
0.50
(0.24)
0.34
(0.22)
1.19
(0.35)
Western Bluebird 0.00 0.00 0.00 1.08
(0.25)
0.12
(0.12)
0.21
(0.21)
0.26
(0.15)
2.41
(0.32)
Hermit Thrush 1.73
(0.26)
0.82
(0.14)
0.43
(0.19)
0.08
(0.03)
1.17
(0.28)
1.07
(0.21)
0.73
(0.19)
0.04
(0.04)
American Robin 0.91
(0.40)
0.68
(0.28)
1.26
(0.30)
0.59
(0.25)
1.30
(0.65)
0.61
(0.26)
0.49
(0.22)
0.21
(0.12)
Virginia’s Warbler 0.00 0.27
(0.27)
0.12
(0.12)
0.00 0.91
(0.32)
2.52
(0.69)
2.08
(0.71)
0.28
(0.15)
Yellow-rumped Warbler 5.17
(1.22)
4.70
(1.11)
3.12
(0.67)
0.68
(0.31)
5.37
(0.75)
5.31
(0.90)
3.78
(0.67)
0.94
(0.30)
Western Tanager 0.56
(0.21)
0.29
(0.15)
0.74
(0.23)
0.41
(0.10)
0.81
(0.22)
1.27
(0.25)
1.27
(0.15)
0.67
(0.19)
Spotted Towhee 0.24
(0.24)
0.70
(0.48)
0.77
(0.65)
0.00 0.00 0.84
(0.56)
3.61
(1.31)
0.69
(0.35)
Chipping Sparrow 1.04
(0.70)
0.15
(0.15)
0.07
(0.07)
0.11
(0.07)
1.04
(0.37)
0.76
(0.19)
0.84
(0.30)
0.32
(0.15)
Dark-eyed Junco 2.81
(0.76)
2.03
(0.47)
1.08
(0.35)
0.83
(0.26)
2.41
(0.53)
3.04
(0.55)
1.94
(0.49)
0.88
(0.36)
Black-headed Grosbeak 0.00 0.50
(0.20)
0.53
(0.18)
0.30
(0.13)
0.49
(0.26)
1.15
(0.34)
0.53
(0.25)
0.30
(0.15)
Notes: Plot densities are stratified by burn severity and pooled over cover type.
See Appendix C for scientific names and list of species.
NATASHA B. KOTLIAR ET AL.496
Ecological Applications
Vol. 17, No. 2
Spotted Towhees exhibited significant density peaks in
low/moderate severity in 2002 only (Table 2). Chipping
Sparrows and American Robins shifted from no
significant trends in 2001 to a weak negative response
in 2002 (Table 2).
Spatial variation, as indicated by significant cover
type main effects or interactions, was exhibited by six
species (Table 2; Appendix D). For example, Cordilleran
Flycatchers and Warbling Vireos exhibited more pro-
nounced declines in upland mixed-coniferous stands
than in upland ponderosa pine stands. Significant
increasing slopes for House Wrens were demonstrated
for upland mixed-coniferous stands, but not ponderosa
pine stands.
Pre- and post-fire changes in density
The pre/post pairs design allowed us to analyze
density patterns for 15 species. Only four species
exhibited significant burn severity effects based on this
analysis (Table 3; Appendix E). Mourning Doves had
significantly higher densities across all burn severities
compared to unburned forest, reflecting s ignificant
increases from pre-fire densities (F
3,17
¼ 2.64, P ¼
0.083; Table 3). American Robins were highly variable
pre-fire, but nevertheless exhibited significant post-fire
declines in unburned forest and significant post-fire
increases in high-severity patches (F
3,17
¼ 6.39, P ¼
0.004; Table 3). However, post-fire densities were similar
among burn severity levels, as reflected by a lack of
TABLE 2. Parameter estimates from the repeated-measures GLM for after-only data at the Cerro Grande fire, New Mexico.
Species
Significant burn
severity effects (P) Year
Burn
severity
(Burn
severity)
2
r
2
Response
class#
Cordilleran Flycatcherà ,0.01 pooled 0.23 NS 0.21 I
Warbling Vireoà ,0.001 2001 0.51 NS 0.5 I
2002 0.52 NS 0.36 I
Mountain Chickadee ,0.001 pooled 0.34 NS 0.37 I
Hermit Thrush ,0.001 pooled 0.28 NS 0.56 I
Yellow-rumped Warbler ,0.001 pooled 0.49 NS 0.39 I
Dark-eyed Junco ,0.001 pooled 0.29 NS 0.23 I
Chipping Sparrow ,0.05 2001 0.07 NS 0.03 III
2002 0.12 NS 0.05 II
American Robin ,0.1 2001 0.04 NS 0.01 III
2002 0.13 NS 0.08 II
Steller’s Jay ,0.05 pooled 0.44 0.17 0.10 IV
Virginia’s Warbler ,0.05 2001 0.12 0.04 0.01 III
2002 0.60 0.27 0.25 IV
Spotted Towhee NS 2001 0.07 0.05 0.01 III
2002 0.92 0.27 0.06 IV
Black-headed Grosbeakà ,0.05 pooled 0.25 0.10 0.04 IV
Mourning Dove ,0.1 2001 0.05 NS 0.04 III
2002 0.11 NS 0.02 III
Northern Flicker NS 2001 0.06 NS 0.02 III
2002 0.04 NS 0.00 III
White-breasted Nuthatchà NS 2001 0.05 NS 0.00 III
2002 0.004 NS
0.02 III
Western Tanager NS 2001 0.002 NS 0.02 III
2002 0.09 NS 0.04 III
Broad-tailed Hummingbird ,0.01 2001 0.08 NS 0.02 III
2002 0.19 NS 0.14 V
Hairy Woodpecker ,0.05 2001 0.15 NS 0.10 V
2002 0.10 NS 0.02 III
House Wrenৠ,0.001 2001 0.14 NS 0.10 V
2002 0.26 NS 0.24 VI
Western Wood-Pewee ,0.001 pooled 0.32 NS 0.34 VI
Western Bluebird ,0.001 2001 0.22 NS 0.32 VI
2002 0.42 NS 0.58 VI
Notes: Estimates and adjusted r
2
are provided by year if there was a significant (P , 0.1) year or year 3 burn severity effect;
otherwise, estimates are pooled across years. Only significant (P , 0.1) terms for (burn severity)
2
are provided; NS denotes not
significant. For species with a nonsignificant (P . 0.1) quadratic term, burn severity represents the slope. For species with a
significant (P , 0.1) term for (burn severity)
2
, negative quadratic terms indicate a concave downward curve.
See Appendix C for scientific names and list of species.
à Significant (P , 0.1) cover type effects in overall model.
§ Mixed-coniferous cover type only.
# Response classes were based on slopes and significant quadratic terms. Response class I, slope 0.2; response class II, 0.2
, slope , 0.1; response class III, slopes not significantly different from 0; response class IV, significant quadratic term; response
class V, 0.1 , slope , 0.2; response class VI, slope 0.2.
March 2007 497AVIFAUNAL RESPONSES TO BURN SEVERITY
significant post-fire contrasts (Table 3). Post-fire densi-
ties of Broad-tailed Hummingbirds and Western Blue-
birds increased with increasing burn severity and were
significantly higher in high-severity patches compared to
unburned patches (F
3,18
¼ 3.33, P ¼ 0.043; F
3,18
¼ 3.49, P
¼ 0.037, respectively; Table 3). Thus, all significant
differences in densities indicated a positive numerical
response to burned forests.
Although fewer significant differences in densities
were dete cted with the pre/post pairs, many nonsignif-
FIG. 2. Representative species for each response class corresponding to variation in density patterns along the burn severity
gradient: (a) response class I represents species with strong declines in density with increasing burn severity; (b) response class II
represents species with weak declines; (c) response class III represents species that show no significant differences in densities across
the burn severity gradient; (d) response class IV represents species that reach peak densities at low or moderate severity; (e) response
class V represents species with weak positive responses across the burn severity gradient; (f ) response class VI represents species
with strong increases in density with increasing burn severity (see Table 2 for the full set of species within each response class).
NATASHA B. KOTLIAR ET AL.498
Ecological Applications
Vol. 17, No. 2
icant post-fire trends (Table 3) were consistent with the
patterns detected by the after-only gradient results
(Table 1). Mountain Chickadee and Virginia’s Warbler
exhibited density declines with increasing burn sev erity;
Black-headed Grosbeak, Chipping Sparrow, and Spot-
ted Towhee exhibited peak densities at low severity;
and Hairy Woodpecker and Western Wood-Pewee
exhibited increasing d ensities with inc reasing burn
severity. For several of the species (Hairy Woodpecker,
Virginia’s Warbler, Western Tanager, and Chipping
Sparrow) showing nonsignificant trends; treatment
effect P values were between 0.1 and 0.2, which is
sometimes used as evidence of impacts in pre/post pairs
analysis (e.g., Murphy et al. 1997). There were t wo
exceptions : Mourning Doves and American Robins
exhibited significantly higher densities post-fire based
on before–after data, but were similar or slightly lower
based on after-only data (Tables 1, 3) . Thus, the
response patterns for these species re main equivocal.
Our ability to detect significant fire effects was limited
by underlying spatiotemporal variation. Five species had
significant cover type effects (Appendix E), although our
power to test for cover type effects was low. In addition,
densities of many species were much higher in unburned
reference areas pre-fire as compared to post-fire (Table
3). Although pre/post pairs analysis can help to control
these sources of variation, low sample sizes prevented
adequate partitioning of such confounding variation.
Community patterns
A total of 49 species was observed with the before–
after data. Species richness was similar pre- vs. post-fire
in unburn ed (46 vs. 41 species), low- (42 vs. 49),
moderate- (35 vs. 33), and high-severity patches (27 vs.
28). Likewise, community similarity (represented by
species occurring both pre- and post-fire) was relatively
high and consistent across unburned, low-, and moder-
ate-severity patches (Fig. 3). Community similarity was
lowest in high-severity areas, yet richness remained
TABLE 3. Density (means, with SE in parentheses) for species for pre- and for post-fire (before–after transects) by burn severity
level at the Cerro Grande Fire, New Mexico.
Species
Density (no. birds/ha)
Pre-fire Post-fire
Unburned
[n ¼ 7]
Low
[n ¼ 6]
Moderate
[n ¼ 4]
High
[n ¼ 5]
Unburned
[n ¼ 7]
Low
[n ¼ 6]
Moderate
[n ¼ 4]
High
[n ¼ 5]
Mourning Doveà 0.54
(0.41)
0.19
(0.14)
0.59
(0.37)
0.17
(0.17)
0.49
a
(0.19)
1.82
ab
(0.84)
2.04
ab
(1.44)
2.26
b
(0.75)
Broad-tailed Hummingbirdà 4.07
(1.59)
7.18
(1.73)
4.67
(0.89)
4.40
(1.89)
0.76
a
(0.35)
2.08
a
(0.93)
3.26
ab
(1.39)
5.52
b
(0.87)
Hairy Woodpecker 0.75
(0.35)
0.81
(0.43)
0.76
(0.76)
2.59
(2.06)
0.00 0.33
(0.18)
0.56
(0.38)
1.18
(0.47)
Northern Flicker 1.57
(0.38)
0.58
(0.29)
1.78
(0.93)
3.32
(1.71)
0.33
(0.17)
0.07
(0.07)
0.29
(0.19)
0.10
(0.06)
Western Wood-Pewee 3.91
(1.16)
2.27
(0.71)
2.02
(0.85)
2.17
(0.90)
0.36
(0.14)
0.53
(0.27)
1.01
(0.39)
1.06
(0.44)
Steller’s Jay 1.16
(0.56)
0.62
(0.45)
0.74
(0.27)
1.15
(0.69)
0.10
(0.10)
0.38
(0.20)
0.70
(0.48)
0.10
(0.10)
Mountain Chickadee 1.32
(0.43)
0.40
(0.26)
1.17
(0.99)
0.57
(0.23)
0.51
(0.19)
0.55
(0.25)
0.50
(0.50)
0.00
Pygmy Nuthatch 1.13
(0.41)
1.22
(0.46)
1.43
(0.22)
2.41
(1.96)
0.55
(0.37)
0.50
(0.26)
0.85
(0.52)
0.18
(0.18)
Western Bluebirdà 0.62
(0.44)
1.62
(0.83)
1.69
(0.99)
0.29
(1.38)
0.30
a
(0.23)
0.51
a
(0.22)
1.14
b
(0.6)
3.44
c
(1.31)
American Robinà 1.77
a
(0.67)
0.66
b
(0.36)
1.36
b
(1.06)
0.15
c
(0.15)
0.47
(0.17)
0.88
(0.32)
0.76
(0.50)
1.09
(0.45)
Virginia’s Warbler 3.07
a
(1.16)
2.27
a
(0.75)
0.53
b
(0.53)
0.00
b
0.42
(0.16)
0.41
(0.28)
0.41
(0.41)
0.00
Western Tanager 0.47
(0.13)
0.42
(0.27)
0.66
(0.47)
0.09
(0.09)
0.32
(0.13)
1.33
(0.42)
0.59
(0.30)
0.91
(0.33)
Spotted Towhee 3.74
a
(0.58)
3.11
a
(1.15)
3.21
a
(1.92)
0.65
b
(0.65)
1.69
(0.69)
2.94
(0.71)
2.04
(1.03)
0.56
(0.56)
Chipping Sparrow 1.47
a
(0.90)
2.66
a
(1.64)
5.20
b
(1.51)
1.31
a
(0.90)
0.47
(0.14)
1.02
(0.38)
0.59
(0.36)
0.33
(0.24)
Black-headed Grosbeak 1.03
(0.51)
0.46
(0.29)
0.51
(0.51)
0.09
(0.09)
0.24
(0.16)
0.54
(0.29)
0.31
(0.31)
0.00
Notes: Transect densities are stratified by burn severity and pooled over cover type (Table A1 in Appendix A). Significantly
different densities (based on post hoc contrasts) among burn severity levels (based on post-fire classifications) are designated by
different superscript letters.
See Appendix C for scientific names and list of species.
à Significant burn severity effects (ANOVA; see Appendix E for details).
March 2007 499AVIFAUNAL RESPONSES TO BURN SEVERITY
unchanged. This was because the number of species only
observed pre-fire in areas that subsequently burned at
high severity (e.g., Brown Creeper, Grace’s Warbler,
Hermit Thrush) was similar to the number of species
observed only post-fire in these areas (e.g., Green-tailed
Towhee, Rock Wren, Western Tanager; Appendix F).
Burn severity patterns among sampling locations
Burn characteristics of landscapes sampled by before–
after and after-only designs differed in several respects.
Before–after samples were located in an area of
moderate fire behavior, whereas after-only samples were
distributed in an area dominated by a wind-driven
crown fire (Fig. 1). The distribution of after-only
samples corresponded to that of the entire burn, whereas
severely burned areas were under represented by the
before–after design (Fig. 4). Many before–after samples
fell along the burn perimeter in proximity to large
contiguous areas of unburned forest (Fig. 1). Crown fire
patches were generally small in the area sampled by
before–after transects, whereas most after-only plots
were within or near large patches of severely burned
forest where unburned or less severely burned areas were
often small and isolated (Fig. 1).
D
ISCUSSION
Species response patterns
A broad spectrum of responses to the burn severity
gradient was evident in the after-only density patterns
(Fig. 2). Although these responses represent a continu-
um of positive and negative responses, the separation of
this continuum into discrete response classes provides a
framework for comparing past and future studies, as
well as proposing mechanisms for the observed patterns.
Negative response patterns are represented by re-
sponse classes I and II. Species in response class I
generally demonstrated pronounced declines over the
entire severity gradient, including low-severity patches.
This group included sub-canopy aerial insectivores
(Cordilleran Flycatc her), ground forage rs (Hermit
Thrush), and foliage gleaners (Mountain Chickadee,
also a cavity nester; Jones and Donovan 1996, McCal-
lum et al. 1999, Lowther 2000), which were likely
sensitive to foliar volume in the understory and canopy.
Within this response class, there was variation in the
magnitude of response at high severity. Mountain
Chickadees were not observed in high-severity patches,
whereas Yellow-rumped Warblers and Dark-eyed Jun-
cos were occasionally detected (Table 1). Some of the
variation among species may correspond to differential
sensitivity to the size of forest gaps (Kotliar et al. 2002).
For example, Yellow-rumped Warblers and Dark-eyed
Juncos may use small forest openings or edges, but
avoid expansive areas of crown fires where most trees
have died. Because these species prefer a mixed-open
canopy (Hunt and Flaspohler 1998, Nolan et al. 2002),
the spatial patterning of burn severity may play a role in
the magnitude of their response (Kotliar et al. 2002).
Response class II represents a weak negative response
and both species (American Robin, Chipping Sparrow)
exhibited significant declines with increasing severity in
2002, but no significant trends in 2001 (Tables 1, 2, and
Fig. 2b).
Response class III (Fig. 2c) included species that were
common across the entire burn severity gradient and,
thus, represent species that do not respond to fire in
terms of abundance. Species in this group prefer mixed-
open canopies (Pravosudov and Grubb 1993, Mirarchi
and Baskett 1994, Moore 1995, Hudon 1999, Sallabanks
and James 1999) and include cavity-nesting species
(Northern Flicker, White-breasted Nuthatch), aerial
insectivores (Western Tanager), and ground foragers
(Mourning Dove, American Robin ). Two s pecies
(Mourning Dove, American Robin) exhibited signifi-
cantly higher densities in burned forests with the before–
after data. The conflicting results between the before–
after and after-only datasets for these species illustrate
the limitations of both sampling approaches. Uncon-
trolled pre-fire variation potentially masks post-fire
FIG. 3. Community similarity pre- and post-fire across a
burn severity gradient at the Cerro Grande fire. Pre- and post-
fire community similarity was evaluated by comparing the
percentage of species detected pre-fire only, post-fire only, and
both pre- and post-fire. The number of species for each burn
severity class are as follows: unburned, n ¼ 75; low, n ¼ 50;
moderate, n ¼ 20; and high, n ¼ 27.
FIG. 4. Burn severity based on the D Normalized Burn
Ratio within 500-m buffers surrounding each transect (before–
after) or plot (after-only) located within the burn perimeter and
for the entire Cerro Grande fire.
NATASHA B. KOTLIAR ET AL.500
Ecological Applications
Vol. 17, No. 2
density changes. Alternatively, the lack of significant
post-fire differences across the severity gradient may
derive from the overriding re-structuring of forests by
fire, which may swamp pre-fire variation.
Response classes IV, V, and VI represent species that
demonstrated a positive response to burns. Species in
response class IV exhibited peak densities at low or
moderate severities, whereas densities in high-severity
patches were similar to unburned areas (Tables 1, 2, and
Fig. 2d). Most species in this group, such as Spotted
Towhees and Virginia’s Warbler, are associated with the
shrub layer (Greenlaw 1996, Olson and Martin 1999),
which may resprout vigorously following fire. Addition-
ally, Steller’s Jay is a relatively wide-ranging species
characteristic of open-canopy forests (Greene et al.
1998). Response class V included species that were
common across the entire burn severity gradient, but
reached peak densities in high-severity patches. This
class includes species associated with snags (Hairy
Woodpecker, House Wren; Johnson 1998, Jackson et
al. 2002) or vegetation, such as aspen and flowering
plants (Broad-tailed Hummingbirds; Calder and Calder
1992), which may rapidly increase immediately post-fire.
Indeed, the magnitude of response for House Wrens, a
secondary cavity nester associated with the shrub layer,
was greater in 2002 (mixed-conifer stands), presumably
in response to aspen resprouting and increased avail-
ability of nest cavities excavated by primary cavity
nesters. Response class VI includes two aerial insecti-
vores (Western Wood-Pewee and Western Bluebird, also
a cavity nester; Bemis and Rising 1999, Guinan et al.
2000). These species exhibited the strongest positive
response to burned forest and were largely restricted to
high-severity patches (Table 1). Although the responses
for all three classes were positive, there were markedly
different responses across the burn severity gradient
(Fig. 2d–f ).
Temporal variation suggested additional distinctions
between positive and negative responders. Strong
negative responses were immediate and consistent across
the first two years post-fire, whereas the magnitude of
the positive response increased in 2002 for many species
(Tables 1, 2, and Fig. 2). However, such changes in
density are difficult to interpret for two years of data.
Likewise, Smucker et al. (2005) observed an increase in
abundance for many positive responders in the first
three years post-fire. They did not test this statistically,
however, and the third year coincided with the use of an
untrained observer, confounding the interpretation of
their patterns because they only present abundance data
unadjusted for detection probabilities. Thus, the tem-
poral dynamics of avian populations post-fire needs
additional study (Kotliar et. al. 2002).
Overall, 71% of the species in the gradient analysis
exhibited either positive or neutral density responses to
fire effects in at least one year post-fire (response classes
III–VI). Two additional species (Yellow-rumped War-
bler and Hermit Thrushes) showed minimal differences
in low-burn compared to unburned reference areas in
2002 (Table 1). This suggests that the majority of species
may tolerate or benefit from many of the ecological
changes that occur across the severity gradient immedi-
ately post-fire. Most species occurred across all burn
severities, consequently, pre- and post-fire species
richness was simi lar. The gre atest species turnover
occurred in the high-severity patches, but the number
of species absent post-fire was balanced by the number
of species that only occurred post-fire. Community
changes in high-severity patches also derived from
pronounced density declines in response classes I and
II that were offset by density peaks in response classes V
and VI. Thus, positive effects of fire essentially balanced
the negative effects for this species pool. Because
approximately two-thirds of Cerro Grande burned at
low or moderate severity (Kotliar et al. 2003), our results
suggest the avifaunal community composition was
largely unchanged over much of the burn.
Most studies of avifaunal response to burns in western
forests have not quantified burn severity but simply
compared severe burns to nearby unburned reference
areas (but see Smucker et al. 2005). In a recent review
(Kotliar et al. 2002), 11 studies (,10 years post-fire)
were used to classify the abundance patterns of 41
species into three response classes: 22% of species were
more abundant in bu rned forests, 32% were more
abundant in unburned forests, and 44% were similar in
burned and unburned forests, or the responses varied
among burns (i.e., mixed/neutral). Because severe burns
contain complex mixtures of burn severities (Turner et
al. 1994, Kotliar et al. 2003), these studies undoubtedly
included more than just severely burned patches (Kotliar
et al. 2002). Our study, in conjunction with the recent
results of Smucker et al. (2005), illustrates how diverse
post-fire density patterns may be exhibited when burn
severity is quantified and included in the analysis. For
most species that exhibited clear positive or negative
associations with burned forests, the patterns we
observed are consistent with past studies. This includes
Warbling Vireos, Mountain Chickadees, H ermit
Thrushes, Hairy Woodpeckers, Western Wood-Pewees,
and House Wrens (Table 2). Many species categorized
by Kotliar et al. (2002) as mixed/neutral responses (e.g.,
Mourning Dove, White-breasted Nuthatch, American
Robin, Chipping Sparrow) also exhibited highly variable
densities, weak responses, a nd lack of significant
differences in our study (Table 2). The consistency of
our density patterns with more qualitative assessments
of fire effects suggests that many previously reported
response patterns are fairly robust, especially for species
falling into our response classes I, III, and VI.
There are important distinctions between our results
and past studies that may reflect, in part, the influence of
burn severity on avian response patterns. For example,
Steller’s Jay was previously classified as having a
primarily negative association with burned forest
(Kotliar et al. 2002), but we observed peak densities at
March 2007 501AVIFAUNAL RESPONSES TO BURN SEVERITY
low- and moderate-severity patches (Table 2, Fig. 2d).
Yellow-rumped Warblers and Dark-eyed Juncos were
previously classified as having mixed/neutral responses
(Kotliar et al. 2002), whereas we found pronounced
declines with increasing burn severity (Table 2),
although both of these species were observed in high-
severity areas (Table 1). Because Cerro Grande was
much larger than most of the burns reviewed in Kotliar
et al. (2002), spatial patterning of burn severity may
contribute to differences observed across studies.
Even when burn severity is quantified, design differ-
enc es among studies can hinder comparisons. For
example, Smucker et al. (2005) compared avifaunal
response one to five years before, and one to three years
after, a 55 000-ha fire in a mixed-coniferous forest in the
northern Rocky Mountains. Several species (Hairy
Woodpecker, Northern Flicker, Yellow-rumped War-
bler) exhibited similar response patterns as observed in
this study. There were several species that exhibited
significant differences in our after-only design (Warbling
Vireo, Mountain Chickadee, House Wren) that showed
nonsignificant trends with their before–after design.
Finally, several species exhibiting peak abundances at
low or moderate severity in Smucker et al.’s study
(2005), exhibited either a decline in abundance with
increasing severity (e.g., Hermit Thrush, Chipping
Sparrow, Dark-eyed Junco), and/or a mixed response
(e.g., American Robin, Western Tanager) in our after-
only data. Because Smucker et al. (2005) used a different
burn severity classification, unadjusted counts, smaller
size of sample units (individual points within transects),
and did not stratify by cover type (ponderosa pine–
Douglas-fir, mixed-coniferous, and lodgepole pine
stands), design differences as well as geographic and
landscape differences, may have contributed to the
varied results among our studies.
Study design considerations
The variation among studies illustrates the impor-
tance of quantifying burn severity, cover type variation,
and adjusting counts for detection probabilities so that
results among studies can be more readily compared. In
particular, responses of species that consistently exhibit
both positive and negative responses across the severity
gradient could be misinterpreted when severity is not
evaluated (Kotliar et al. 2002, Smucker et al. 2005). In
addition, the relative magnitude of negative or positive
responses is more difficult to assess if comparisons are
not made across the full range of severities. Thus,
differential responses across the severity gradient cannot
be quantified with simple burn vs. unburned forest
comparisons (Kotliar et al. 2002, Smucker et al. 2005).
Burns vary widely in size and spatial patterning of
burn severity, which can affect species response patterns
(Kotliar et al. 2002). Because standards for quantifying
burn severity are lacking (Key and Benson 2005), we
provide some guidelines for quantifying burn severity to
facilitate future comparisons among studies. Arbitrary
breakpoints between burn severity levels can hinder
gradient analysis and inter-burn comparisons. For
example, moderate-severity burns have been variously
classified as 0–10% crown fire (Dwyer 2000), 20–80%
tree mortality (Smucker et al. 2005), and all trees partial
to complete needle scorch (as in this study). Ideally, field
measurement of burn severity should be continuous to
prevent such discrepancies. This is particularly critical
for the detection of nonlinear responses to burn severity,
which we observed in this study.
Because severity patterns are strongly scale dependent
(Cocke et al. 2005) and species can respond differentially
across scales, measurement scale can affect results.
Likewise, within-plot variation in burn severity can
contribute to variation among studies. In our study,
moderate severity plots could be comprised of uniformly
burned survey points or a combination of low- and high-
severity points. Within-plot variation was less pro-
nounced for low- or high-severity plots. Similarly,
variation in burn severity across 1-km transects likely
contributed to the relatively high proportion (69%)of32
species (summarized by Kotliar e t al. 2002) that
exhibited a mixed/neutral response to the La Mesa fire
(this 6250-ha fire occurred in 1977 near the area
subsequently burned by Cerro Grande; Johnson and
Wauer 1996). Remotely sensed characterizations of burn
severity, like DNBR (Cocke et al. 2005, Key and Benson
2005), can facilitate comparisons among studies and be
used to quantify burn severity and spatial attributes of
burn patches at multiples scales.
The before–after comparisons and after-only gradient
analysis we employed at Cerro Grande provided an
unprecedented opportunity to evaluate the relative
usefulness of these sampling designs for quantifying
avifaunal response to burns. Further, our ability to
employ distance analysis to quantify both observer
effects and detection differences among burn severity
levels is a unique but important aspect of our study
design. Although there was general correspondence in
trends for species common to both data sets, fewer
significant patterns were detected by the pre/post pairs
analysis. This is likely due in part to lower power
resulting from smaller sample sizes, particularly at high
burn severity, which is typical of historical data sets
associated with unplanned disturbances (Wiens and
Parker 1995). Smucker et al. (2005) also had limited
power to detect significant responses (50% of 32 species
at P , 0.05) using before–after comparisons. In
comparison, the gradient analysis at Cerro Grande
detected significant patterns for 81% (71% at P , 0.05)
of the species. The after-only data had sufficient samples
to treat burn severity as a continuous rather than
categorical variable, thereby increasing our ability to
detect significant trends. In general, low sample sizes and
high underlying variation in the before–after data
resulted in weak inferences because only very large
changes could be detected. Likewise, in a study of the
Exxon Valdez oil spill, Murphy et al. (1997) detected
NATASHA B. KOTLIAR ET AL.502
Ecological Applications
Vol. 17, No. 2
only large effects of oiling using pre/post pairs analysis
because of high population variability and small sample
sizes.
Several other factors likely contributed to the
discrepancies in results. Because the t ime elapsed
between pre- and post-fire surveys was .10 years,
temporal variation in populations may derive from long-
term trends in the landscape that are not related to fire
(e.g., successional changes within cover types, expansion
of human developments). Densities in unburned refer-
ence areas were much higher pre-fire than post-fire for
many species, suggesting potential long-term trends or
unresolved sampling differences that masked short-term
respo nse to fire. Pre-fire sampling occurred during
relat ively wet conditions compared to the drought
conditions during post-fire sampling (based on Palmer
Drought Index, available online).
6
Some of the discrep-
ancies among data sets may also be due to differences
between species pools; the pre/post analysis included a
higher prop ortion of species exhibiting n eutral or
positive responses, but did not include five species
classified as response class I (after-only). Finally, the
larger size of sample units and/or more moderate fire
conditions surrounding the before–after transects com-
pared to after-only plots (Figs. 1 and 4) may also
account for lack of significant negative responses to
burns in the pre/post analysis. Indeed, the magnitude of
differences between burned and unburned areas in the
before–after data was sometimes much less than that
observed with the after-only data (e.g., Mountain
Chickadee; Tables 1 and 3). Likewise, many before–
after transects were in more mesic conditions of canyon
bottoms, which can moderate fire behavior and facilitate
post-fire regrowth (Dwire and Kauffman 2003). Conse-
quently, post-fire for est conditions for before–after
transects may be more similar to pre-fire conditions
compared to after-only plots. As a result of the design
limitations in the before–after data, we have greater
confidence in the results from the gradient analysis.
Despite inherent difficulties in analyzing and inter-
preting the historical data set, our results demonstrated
significant spatiotemporal variation and the importance
of controlling such potentially confounding sources of
variation in before–after and after-only designs (Wiens
and Parker 1995). Ideally, before–after designs should
include a minimum of three years pre- and post-fire
sampling to quantify non-process, temporal population
variation (Stewart-Oaten et al. 1986). Controlling
temporal variation is especially critical for planned
BACI and after-only designs (Stewart-Oaten et al. 1986,
Hewitt et al. 2001, Stewart-Oaten and Bence 2001), used
to study the effects of prescribed fire or other fuels
treatments, because uncontrolled population variation
will increase the difficulty of detecting fire effects under
more moderate conditions characteristic of understory
prescribed fire.
The greatest value of our before–after data set was
quantifying post-fire changes in community composi-
tion, which cannot be evaluated with after-only data.
Design limitations of our pre/post pairs analysis
underscore the challenges of using historical data to
evaluate the effects of unplanned disturbances. Al-
though often viewed as an ideal study design because
of its use of ‘‘controls’ (e.g., Smucker et al. 2005), in
practice, the use of BACI designs is problematic even for
well-planned experiments. Resampling of historical data
sets with limited sample sizes will inevitably lack
sufficient power to detect population changes (Murphy
et al. 1997), and such unplanned studies rarely meet the
strict requirements for sound statistical design (Stewart-
Oaten et al. 1986, Hewitt et al. 2001, Weiss and Reice
2005). The inclusion of after-only data that spatially
overlap before–after samples can provide independent
tests of avifaunal response patterns. Careful consider-
ation of the statistical constraints imposed by historical
data sets should be evaluated prior to resampling post-
fire. Gradient analysis (Wiens and Parker 1995, Day et
al. 1997), particularly if cofactors and a time series
component are included, may be a more powerful and
efficient technique for quantifying response to burns
than control–impact studies (Ellis and Schneider 1997).
Neither design can fully resolve the inherent statistical
challenges in quantifying fire effects at a single wildland
fire (Kotliar et al. 2002). Further, true replication of
burns cannot exist because of the large scales at which
they occur (van Mantgem et al. 2001, Weiss and Reice
2005) and the variatio n among burns in spatial
patterning of burn severity. Thus, multiple studies are
needed to fully characterize species response patterns.
To facilitate cross-study comparisons and maximize the
power of the individual data sets, a priori development
of study protocols are necessary (Anderson et al. 1999,
Franklin et al. 2004). We have highlighted several
important design considerations, such as collecti ng
continuous, quantitative measures of burn severity and
controlling sources of spatiotemporal variation. In
addition, avian abundance responses should be based
on densities, rather than unadjusted counts, to control
for observer and cover type variability among studies.
Because species responses vary, even over short time
frames, short- and long-term studies across a broad
range of fire conditions, including areas of fire exclusion,
are essential to our understanding of fire effects (Kotliar
et al. 2002).
Management implications
Our results demonstrated that many species tolerate
or capitalize on the ecological changes resulting from
severe fires; thus, assumptions about overriding negative
ecological effects of recent crown fires in Southwest
ponderosa pine forests (e.g., Allen et al. 2002) are invalid
for the avifauna at Cerro Grande. The diverse response
6
hwww.ncdc.noaa.govi
March 2007 503AVIFAUNAL RESPONSES TO BURN SEVERITY
patterns and the high level of species turnover in high-
severity patches illustrate the importance of severe fires
to landscape and habitat dynamics in fire prone systems.
Although there is increasing evidence that crown fires
occurred historically even in montane forests, the role of
severe fires cannot be fully resolved given limitations of
the historical record (Baker and Ehle 2001). To develop
sound fire management programs and policies, it is
important to quantify the ecological consequences of fire
regime dynamics across a broad range of temporal and
spatial scales, reg ardless of whether changes are
anthropogenic, climate driven, or a combination of
both.
The paradigm that recent large fires are beyond the
range of historical variability gives rise to additional
untested assumptions that influence fire management
decisions. The extent to which ecosystem restoration
programs targeting forest conditions and fire regimes
characteristic of a specific location and time period
provides for a diverse avifauna has been poorly tested.
Yet, the validity of this assumption is fundamental to
the success of such restoration programs. Although
many species, particularly those preferring low-severity
burns or mixed-open canopies (e.g., response classes II,
III, I V) may benefit from, or tolerate, ecosystem
restoration programs, severe fires create forest structures
and ecological elements that cannot be readily created
by forest thinning and understory prescribed fire (Hutto
1995, Kotliar et al. 2002). Species preferring closed-
canopy forests (including species in response class I) may
benefit from fire exclusion (Bock and Block 2005). Our
results demonstrated that even in Southwest ponderosa
pine forests, high-severity patches, like th ose that
resulted from the Cerro Grande fire, can create
ecological conditions that benefit many species thereby
increasing species diversity at landscape scales.
Increasing evidence from other disciplines indicates
that even in montane systems, severe fires provide
additional ecological functions (e.g., Bisson et al. 2003,
Fule
´
et al. 2004). The tendency to focus on total area
burned rather than more detailed reporting by burn
severity (Stephens and Ruth 2005) further obfuscates the
broad range of ecological changes that can occur
following severe fires. In conclusion, the current
management emphasis on low-severity fires (e.g., pre-
scribed burns) and a narrow range of stand densities in
montane forests may negatively impact species associ-
ated with high-severity burns, as well as species
associated with dense forests. Fire management that
includes a broad range of natural variability (Allen et al.
2002), including areas of severe fire, is more likely to
preserve a broad range of ecological functions than
restoration objectives based on narrowly defined historic
fire regimes (Schoennagel et al. 2004).
ACKNOWLEDGMENTS
We are grateful to Deborah Finch and the USDA Forest
Service, Rocky Mountain Forest and Range Experiment
Station, for providing financial support for this work. Field
assistance was tirelessly provided by Dorothy Crowe, Elizabeth
Reynolds, Chip Clouse, Angela Newberry, Michele Kelly,
Michael Voyles, Paul Yoder, John Arnett, Cynthia Melcher,
and Laura Ellison. Andrea Lueders, Brian Cade, Aaron
Ellingson, and Joan Morrison assisted with statistical analyses.
Sandra Haire, Tammy Fancher, and Terry Giles assisted in
creating and analyzing NBR maps. Dale Crawford assisted in
preparation of the figures. Logistical support was provided by
Steve Fettig, Kay Beeley, Randy Belice, Jeanne Fair, Fairley
Barnes, Jerry Godbey, Denny Bohon, and the staff at the Los
Alamos and Jemez districts of the Santa Fe National Forest.
Jim McIver, Brian Cade, and three anonymous reviewers
provided helpful comments on a draft of this manuscript, and
Jana Dick assisted with manuscript editing. We thank Susan
Skagen and Craig Allen for valuable discussions.
L
ITERATURE CITED
Allen, C. D. 1989. Changes in the landscape of the Jemez
Mountains, New Mexico. Dissertation. University of Cal-
ifornia, Berkeley, California, USA.
Allen, C. D., M. Savage, D. A. Falk, K. F. Suckling, T. W.
Swetnam, T. Schulke, P. B. Stacey, P. Morgan, M. Hoffman,
and J. Klingel. 2002. Ecological restoration of Southwestern
ponderosa pine ecosystems: a broad perspective. Ecological
Applications 12:1418–1433.
Anderson, D. R., K. P. Burnham, A. B. Franklin, R. J.
Gutierrez, E. D. Formsan, R. G. Anthony, G. C. White, and
T. M. Shenk. 1999. A protocol for conflict resolution in
analyzing empirical data related to natural resource contro-
versies. Wildlife Society Bulletin 27:1050–1058.
Baker, W. L., and D. Ehle. 2001. Uncertainty in surface-fire
history: the case of ponderosa pine forests in the western
United States. Canadian Journal of Forest Research 31:
1205–1226.
Bemis, C., and J. D. Rising. 1999. Western Wood-Pewee
(Contopus sordidulus). In A. Poole and F. Gill, editors. The
birds of North America, number 451. Academy of Natural
Sciences, Philadelphia, Pennsylvania, USA; American Orni-
thologists’ Union, Washington, D.C., USA.
Bisson, P. A., B. E. Rieman, C. Luce, P. F. Hessburg, D. C.
Lee, J. L Kershner, G. H. Reeves, and R. E. Gresswell. 2003.
Fire and aquatic ecosystems of the western USA: current
knowledge and key questions. Forest Ecology and Manage-
ment 178:213–229.
Bock, C. E., and W. M. Block. 2005. Fire and birds in the
southwestern United States. Studies in Avian Biology 30:14–
32.
Buckland, S. T., and D. R. Anderson. 2002. Model selection
and inference: a practical information-theoretic approach.
Second edition. Springer, New York, New York, USA.
Buckland, S. T., D. R. Anderson, K. P. Burnham, J. L. Laake,
D. L. Borchers, and L. Thomas. 2001. Introduction to
distance sampling. Estimating abundance of biological
populations. Oxford University Press, London, UK.
Calder, W. A., and L. L. Calder. 1992. Broad-tailed Hum-
mingbird (Selasphorus platycercus). In A. Poole and F. Gill,
editors. The birds of North America, number 16. Academy of
Natural Sciences, Philadelphia, Pennsylvania, USA. Ameri-
can Ornithologists’ Union, Washington, D.C., USA.
Cocke, A. E., P. Z. Fule
´
, and J. E. Crouse. 2005. Comparison of
burn severity assessments using differenced Normalized Burn
Ratio and ground data. International Journal of Wildland
Fire 14:189–198.
Covington, W. W., P. Z. Fule
´
, M. M. Moore, S. C. Hart, T. E.
Kolb, J. N. Mast, S. S. Sackett, and M. R. Wagner. 1997.
Restoring ecosystem health in ponderosa pine forests of the
Southwest. Journal of Forestry 97:23–29.
Day, R. H., S. M. Murphy, J. A. Wiens, and G. C. Hayward.
1997. Effects of the Exxon Valdez oil spill on habitat use by
birds in Prince William Sound, Alaska. Ecological Applica-
tions 7:593–613.
NATASHA B. KOTLIAR ET AL.504
Ecological Applications
Vol. 17, No. 2
Dwire, K. A., and J. B. Kauffman. 2003. Fire and riparian
ecosystems in landscapes of the western USA. Forest Ecology
and Management 178:61–74.
Dwyer, J. K. 2000. Response of secondary cavity-nesting birds
to wildfire in ponderosa pine forests of northern Arizona.
Thesis. Northern Arizona University, Flagstaff, Arizona,
USA.
Ellis, J. I., and D. C. Schneider. 1997. Evaluation of a gradient
sampling design for environmental impact assessment.
Environmental Monitoring and Assessment 48:157–72.
Finch, D. M., J. L. Ganey, W. Yong, R. T. Kimball, and R.
Sallabanks. 1997. Effects and interactions of fire, logging,
and grazing. Pages 103–136 in W. M. Block and D. M. Finch,
editors. Songbird ecology in Southwestern ponderosa pine
forests: a literature review. USDA Forest Service General
Technical Report RMRS-GTR-292. Rocky Mountain Forest
and Range Experiment Station, Fort Collins, Colorado,
USA.
Franklin, A. B., et al. 2004. Population dynamics of the
California Spotted Owl (Strix occidentalis occidentalis): a
meta-analysis. Ornithological Monographs 54:1–54.
Fule
´
, P. Z., A. E. Cocke, T. A. Heinlein, and W. W. Covington.
2004. Effects of an intense prescribed forest fire: Is it
ecological restoration? Restoration Ecology 12:220–230.
Fule
´
, P. Z., W. W. Covington, and M. M. Moore. 1997.
Determining reference conditions for ecosystem management
of Southwestern ponderosa pine forests. Ecological Applica-
tions 7:895–908.
Graham, R. T., S. McCaffrey, and T. B. Jain. 2004. Science
basis for changing forest structure to modify wildfire
behavior and severity. USDA Forest Service General
Technical Report RMRS-GTR-120. Rocky Mountain Forest
and Range Experiment Station, Fort Collins, Colorado,
USA.
Greene, E., W. Davison, and V. R. Muehter. 1998. Steller’s Jay
(Cyancitta stellarri). In A. Poole and F. Gill, editors. The
birds of North America, number 343. Academy of Natural
Sciences, Philadelphia, Pennsylvania, USA; American Orni-
thologists’ Union, Washington, D.C., USA.
Greenlaw, J. S. 1996. Spotted Towhee (Pipilo maculatus). In A.
Poole and F. Gill, editors. The birds of North America,
number 263. Academy of Natural Sciences, Philadelphia,
Pennsylvania, USA; American Ornithologists’ Union, Wash-
ington, D.C., USA.
Guinan, J. A., P. A. Gowaty, and E. K. Eltzroth. 2000. Western
Bluebird (Sialia mexicana). In A. Poole and F. Gill, editors.
The birds of North America, number 510. Academy of
Natural Sciences, Philadelphia, Pennsylvania, USA; Ameri-
can Ornithologists’ Union, Washington, D.C., USA.
Hewitt, J. E., S. E. Thrush, and V. J. Cummings. 2001.
Assessing environmental impacts: effects of temporal vari-
ability at likely impact scales. Ecological Applications 11:
1502–1516.
Hudon, J. 1999. Western Tanager (Piranga ludoviciana). In A.
Poole and F. Gill, editors. The birds of North America,
number 432. Academy of Natural Sciences, Philadelphia,
Pennsylvania, USA; American Ornithologists’ Union, Wash-
ington, D.C., USA.
Hunt, P. D., and D. J. Flaspohler. 1998. Yellow-rumped
Warbler (Dendroica coronata). In A. Poole and F. Gill,
editors. The birds of North America, number 376. Academy
of Natu ral Sciences, Philadelphia, Pennsylvania, USA;
American Ornithologists’ Union, Washington, D.C., USA.
Hutto, R. L. 1995. Composition of bird communities following
stand-replacement fires in Northern Rocky Mountain
(U.S.A.) conifer forests. Conservation Biology 9:1041–1058.
Jackson, J. A., H. R. Ouellet, and B. J. S. Jackson. 2002. Hairy
Woodpecker (Picoides villosus). In A. Poole and F. Gill,
editors. The birds of North America, number 702. Academy
of Natu ral Sciences, Philadelphia, Pennsylvania, USA;
American Ornithologists’ Union, Washington, D.C., USA.
Janofsky, M. 2000. Citing poor plan, U.S. takes blame in Los
Alamos fire. New York Times, 19 May 200, Section A1:3.
Johnson, L. S. 1998. House Wren (Troglodytes aedon). In A.
Poole and F. Gill, editors. The birds of North America,
number 380. Academy of Natural Sciences, Philadelphia,
Pennsylvania, USA; American Ornithologists’ Union, Wash-
ington, D.C., USA.
Johnson, T. H., and R. H. Wauer. 1996. Avifaunal response to
the 1977 La Mesa fire. Pages 70–94 in C. D. Allen, editor.
Fire effects in Southwestern forests: proceedings of the
second La Mesa Fire Symposium, Los Alamos, NM, March
29–31, 1994. USDA Forest Service General Technical Report
RMRS-GTR-286. Rocky Mountain Forest and Range
Experiment Station, Ft. Collins, Colorado, USA.
Jones, P. W., and T. M. Donovan. 1996. Hermit Thrush
(Catharus guttatus). In A. Poole and F. Gill, editors. The
birds of North America, number 271. Academy of Natural
Sciences, Philadelphia, Pennsylvania, USA; American Orni-
thologists’ Union, Washington, D.C., USA.
Keane, R. E., K. C. Ryan, T. T. Veblen, C. D. Allen, J. Logan,
and B. Hawkes. 2002. Cascading effects of fire exclusion in
Rocky Mountain ecosystems: a literature review. USDA
Forest Service General Technical Report RMRS-GTR-91.
Rocky Mountain Forest and Range Experiment Station,
Fort Collins, Colorado, USA.
Kennedy, P. L. 1991. Reproductive strategies of northern
Goshawks and Cooper’s hawks in north-central New
Mexico. Dissertation. Utah State University, Logan, Utah,
USA.
Key, C. H., and N. C. Benson. 2005. Landscape assessment:
ground measure of severity, the Composite Burn Index; and
remote sensing of severity, the Normalized Burn Ratio. In
D. C. Lutes, R. E. Keane, J. F. Caratti, C. H. Key, N. C.
Benson, and L. J. Gangi, editors. FIREMON: Fire effects
monitoring and inventory system. USDA Forest Service
General Technical Report RMRS-GTR-164-CD:LA1–LA51,
Rocky Mountain Forest and Range Experiment Station,
Ogden, Utah, USA.
Kotliar, N. B., S. L. Haire, and C. H. Key. 2003. Lessons from
the fires of 2000: post-fire heterogeneity in ponderosa pine
forests. Pages 277–279 in P. Omi and L. Joyce, editors. Fire,
fuel treatments, and ecological restoration: USDA Forest
Service Conference proceedings April 16–18, 2002, Fort
Collins, Colorado. RMRS-P-29. Rocky Mountain Forest
and Range Experiment Station, Fort Collins, Colorado,
USA.
Kotliar, N. B., S. J. Heijl, R. L. Hutto, V. A. Saab, C. P.
Melcher, and M. E. McFadzen. 2002. Effects of fire and post-
fire salvage logging on avian communities in conifer-
dominated forests of the western United States. Studies in
Avian Biology 25:49–64.
Leica Geosystems. 2004. ERDAS Imagine. Version 8.3. Leica
Geosystems, Norcross, Georgia, USA.
Lowther, P. E. 2000. Pacific-slope Flycatcher (Empidonax
difficilis) and Cordilleran Flycatcher (Empidonax occidenta-
lis). In A. Poole and F. Gill, editors. The birds of North
America, number 556 . Academy of Na tural Sciences,
Philadelphia, Pennsylvania, USA; American Ornithologists’
Union, Washington, D.C., USA.
McCallum, D. A., R. Grundel, and D. L. Dahlsten. 1999.
Mountain Chickadee (Poecile gambeli). In A. Poole and F.
Gill, editors. The birds of North America, number 453.
Academy of Natural Sciences, Philadelphia, Pennsylvania,
USA; American Ornithologists’ Union, Washington, D.C.,
USA.
Mirarchi, R. E., and T. S. Baskett. 1994. Mourning Dove
(Zenaida macroura). In A. Poole and F. Gill, editors. The
birds of North America, number 117. Academy of Natural
Sciences, Philadelphia, Pennsylvania, USA; American Orni-
thologists’ Union, Washington, D.C., USA.
March 2007 505AVIFAUNAL RESPONSES TO BURN SEVERITY
Moore, M. M., W. W. Covington, and P. Z. Fule
´
. 1999.
Reference conditions and ecological restoration: a South-
western ponderosa pine perspective. Ecological Applications
9:1266–1277.
Moore, W. S. 1995. Northern Flicker (Colaptes auratus). In A.
Poole and F. Gill, editors. The birds of North America,
number 166. Academy of Natural Sciences, Philadelphia,
Pennsylvania, USA; American Ornithologists’ Union, Wash-
ington, D.C., USA.
Morrison, J. L., and P. L. Kennedy. 1989. The use of line
transect to evaluate the abundance of diurnal mammalian
prey. Journal of Raptor Research 23:172–175.
Murphy, S. M., R. H. Day, J. A. Wiens, and K. R. Parker.
1997. Effects of the Exxon Valdez oil spill on birds:
comparisons of pre- and post-spill surveys in Prince William
Sound, Alaska. Condor 99:299–313.
Nolan, V., Jr., E. D. Ketterson, D. A. Cristol, C. M. Rogers,
E. D. Clotfelter, R. C. Titus, S. J. Schoech, and E. Snajdr.
2002. Dark-eyed Junco (Junco hyemalis). In A. Poole and F.
Gill, editors. The birds of North America, number 716.
Academy of Natural Sciences, Philadelphia, Pennsylvania,
USA; American Ornithologists’ Union, Washington, D.C.,
USA.
Olson, C. R., and T. E. Martin. 1999. Virginia’s Warbler
(Vermivora virginiae). In A. Poole and F. Gill, editors. The
birds of North America, number 477. Academy of Natural
Sciences, Philadelphia, Pennsylvania, USA; American Orni-
thologists’ Union, Washington, D.C., USA.
Osenberg, C. W., R. J. Schmitt, S. J. Holbrook, K. E. Abu-
Saba, and A. R. Flegal. 1994. Detection of environmental
impacts: natural variability, effect size, and power analysis.
Ecological Applications 4:16–30.
Pierce, J. L., G. A. Meyer A., and J. T. Jull. 2004. Fire-induced
erosion and millennial scale climate change in northern
ponderosa pine forests. Nature 432:87–90.
Pravosudov, V. V., and T. C. Grubb, Jr.. 1993. White-breasted
Nuthatch (Sitta carolinensis). In A. Poole and F. Gill, editors.
The birds of North America, number 54. Academy of
Natural Sciences, Philadelphia, Pennsylvania, USA; Ameri-
can Ornithologists’ Union, Washington, D.C., USA.
Saab, V. A., and H. D. W. Powell. 2005. Fire and avian ecology
in North America. Studies in Avian Biology No. 30. Cooper
Ornithological Society, Ephrata, Pennsylvania, USA.
Saab, V. A., H. D. W. Powell, N. B. Kotliar, and K. R.
Newlon. 2005. Variation in fire regimes of the Rocky
Mountains: implications for avian communities and fire
management. Studies in Avian Biology 30:76–90.
Sallabanks, R., and F. C. James. 1999. American Robin
(Turdus migratorius). In A. Poole and F. Gill, editors. The
birds of North America, number 462. Academy of Natural
Sciences, Philadelphia, Pennsylvania, USA; American Orni-
thologists’ Union, Washington, D.C., USA.
SAS Institute. 2001. SAS/STAT guide for personal computer.
Version 6.12. SAS Institute, Cary, North Carolina, USA.
Schoennagel, T., T. T. Veblen, and W. H. Romme. 2004. The
interaction of fire, fuels, and climate across Rocky Mountain
forests. BioScience 54:661–676.
Siders, M. S., and P. L. Kennedy. 1996. Forest structural
char acteri stics of accipiter nesting habitat: Is there an
allometric relationship? Condor 98:123–132.
Smucker, K. M., R. L. Hutto, and B. M Steele. 2005. Changes
in bird abundance after wildfire: importance of fire severity
and time since fire. Ecological Applications 15:1535–1549.
SPSS. 2002. SPSS for Windows, release 11.5.0. SPSS, Chicago,
Illinois, USA.
Stephens, S. L., and L. W. Ruth. 2005. Federal forest-fire policy
in the United States. Ecological Applications 15:532–542.
Stewart-Oaten, A., and J. R. Bence. 2001. Temporal and spatial
variation in environmental impact assessment. Ecological
Monographs 71:305–339.
Stewart-Oaten, A., W. W. Murdoch, and K. R. Parker. 1986.
Environmental impact assessment: ‘pseudoreplication’ in
time? Ecology 67:929–940.
Thomas, L., J. L. Laake, S. Strindberg, F. F. C. Marques, S. T.
Buckland, D. L. Borchers, D. R. Anderson, K. P. Burnham,
S. L. Hedley, and J. H. Pollard. 2002. DISTANCE 4.0.
Release 1. Research Unit for Wildlife Population Assess-
ment, University of St. Andrews, St. Andrews, UK.
Tiedemann, A. R., J. O. Klemmedson, and E. L. Bull. 2000.
Solution of forest health problems with prescribed fire: Are
forest productivity and wildlife at risk? Forest Ecology and
Management 127:1–18.
Turner, M. G., W. W. Hargrove, R. H. Gardner, and W. H.
Romme. 1994. Fire effects on landscape heterogeneity in
Yellowstone National Park, Wyoming. Journal Vegetation
Science 5:731–742.
van Mantgem, P., M. Schwartz, and M. B. Keifer. 2001.
Monitoring fire effects for managed burns and wildfires:
coming to terms with pseudoreplication. Natural Areas
Journal 21:266–273.
Veblen, T. T., T. Kitzberger, and J. Donnegan. 2000. Climatic
and human influences on fire regimes in ponderosa pine
forests in the Colorado front range. Ecological Applications
10:1178–1195.
Wagner, M. R., W. M. Block, B. W. Geils, and K. F. Wenger.
2000. Restoration ecology : a new forest management
paradigm or another merit badge for foresters. Journal of
Forestry 98:22–27.
Weiss, J. M., and S. R. Reice. 2005. The aggregation of impacts:
using species-specific effects to infer community-level distur-
bances. Ecological Applications 15:599–617.
Wiens, J. A., and K. R. Parker. 1995. Analyzing the effects of
accidental environmental impacts: approaches and assump-
tions. Ecological Applications 5:1069–1083.
APPENDIX A
Sampling locations at the Cerro Grande fire, New Mexico (
Ecological Archives A017-018-A1).
APPENDIX B
Calculation of detection probabilities (Ecological Archives A017-018-A2).
APPENDIX C
Scientific and common names (Ecological Archives A017-018-A3).
NATASHA B. KOTLIAR ET AL.506
Ecological Applications
Vol. 17, No. 2
APPENDIX D
Results of after-only gradient analysis at the Cerro Grande fire, New Mexico (Ecological Archives A017-018-A4).
APPENDIX E
ANOVA results for pre/post pairs analysis at the Cerro Grande fire, New Mexico (Ecological Archives A017-018-A5).
APPENDIX F
Frequency of occurrence for before–after transects at the Cerro Grande fire, New Mexico (Ecological Archives A017-018-A6).
March 2007 507AVIFAUNAL RESPONSES TO BURN SEVERITY
    • "We lacked data necessary to analyze such shifts at the Coconino location. Other studies, however, have found broad consistency between burn severity relationships and population shifts from before to after wildfire (Smucker et al. 2005, Kotliar et al. 2007, Seavy and Alexander 2014). Examining changes in occupancy or abundance with time since disturbance is also necessary for a full understanding of species ecological relationships with wildfire ( Powell 2005, Sitters et al. 2014 ). "
    [Show abstract] [Hide abstract] ABSTRACT: Wildfire is a key factor influencing bird community composition in western North American forests. We need to understand species and community responses to wildfire and how responses vary regionally to effectively manage dry conifer forests for maintaining biodiversity. We compared avian relationships with wildfire burn severity between two dry forest locations of Arizona and Idaho. We predicted different responses to wildfire between locations due to regional differences in historical fire regime. We conducted point count surveys for 3 yr following wildfire (Arizona: 1997-1999; Idaho: 2008-2010) and used multispecies hierarchical models to analyze relationships of bird occupancy with burn severity. Consistent with our prediction for mixed-severity fire regimes characterizing the Idaho location, we observed proportionately more positive species occupancy relationships and, consequently, a positive species richness relationship with burn severity in Idaho. We also observed the opposite pattern in Arizona, which was congruent with our prediction for the low-severity fire regime characterizing that location. Cavity nesters and aerial insectivores occupied more severely burned sites following wildfire, corresponding with predicted increases in nesting substrate and foraging opportunities for these species. In contrast, canopy-nesting foliage gleaners and pine seed consumers exhibited negative relationships with burn severity. Our results were consistent with predictions based on species life histories and with patterns from the literature, suggesting generality of observed relationships and locational difference in relationships with wildfire. We therefore suggest that optimal management strategies for maintaining avian diversity could differ regionally. Specifically, intensive fuels management may be ecologically less appropriate for promoting biodiversity in areas such as the Idaho location where mixed-severity wildfires and dense forest stands were historically more common.
    Full-text · Article · May 2016
    • "In addition, it is possible that occupied sites in close proximity to infrastructure had combinations of characteristics that maximized benefits and minimized costs for raptors occupying developed areas, or were occupied by individuals tolerant of disturbance. We therefore suggest before-after control- impact [81] and/or patch-based occupancy studies [82] are needed to make strong inference about the relationship of ferruginous hawks to oil and gas development. Our results will be a useful baseline for comparison as density of energy development continues to increase in our study area. "
    [Show abstract] [Hide abstract] ABSTRACT: Grassland and shrubland birds are declining globally due in part to anthropogenic habitat modification. Because population performance of these species is also influenced by non-anthropogenic factors, it is important to incorporate all relevant ecological drivers into demographic models. We used design-based sampling and occupancy models to test relationships of environmental factors that influence raptor demographics with re-occupancy of breeding territories by ferruginous hawks (Buteo regalis) across Wyoming, USA, 2011-2013. We also tested correlations of territory re-occupancy with oil and gas infrastructure-a leading cause of habitat modification throughout the range of this species of conservation concern. Probability of re-occupancy was not related to any covariates we investigated in 2011, had a strong negative relationship with cover of sagebrush (Artemisia spp.) in 2012, was slightly higher for territories with artificial platforms than other nest substrates in 2013, and had a positive relationship with abundance of ground squirrels (Urocitellus spp.) that was strong in 2012 and weak in 2013. Associations with roads were weak and varied by year, road-type, and scale: in 2012, re-occupancy probability had a weak positive correlation with density of roads not associated with oil and gas fields at the territory-scale; however, in 2013 re-occupancy had a very weak negative correlation with density of oil and gas field roads near nest sites (≤500 m). Although our results indicate re-occupancy of breeding territories by ferruginous hawks was compatible with densities of anthropogenic infrastructure in our study area, the lack of relationships between oil and gas well density and territory re-occupancy may have occurred because pre-treatment data were unavailable. We used probabilistic sampling at a broad spatial extent, methods to account for imperfect detection, and conducted extensive prey sampling; nonetheless, future research using before-after-control-impact designs is needed to fully assess impacts of oil and gas development on ferruginous hawks.
    Full-text · Article · Apr 2016
    • "As suggested by earlier work that included fire severity as an independent variable (e.g. Smucker et al. 2005; Kotliar et al. 2007 Kotliar et al. , 2008 Vierling and Lentile 2008; Stephens et al. 2015), fire severity has a dramatic influence on the probability of occurrence of bird species. The same pattern was true here for nearly all species that were detected on at least 10 points. "
    [Show abstract] [Hide abstract] ABSTRACT: We conducted bird surveys in 10 of the first 11 years following a mixed-severity fire in a dry, low-elevation mixed-conifer forest in western Montana, United States. By defining fire in terms of fire severity and time-since-fire, and then comparing detection rates for species inside 15 combinations of fire severity and time-since-fire, with their rates of detection in unburned (but otherwise similar) forest outside the burn perimeter, we were able to assess more nuanced effects of fire on 50 bird species. A majority of species (60%) was detected significantly more frequently inside than outside the burn. It is likely that the beneficial effects of fire for some species can be detected only under relatively narrow combinations of fire severity and time-since-fire. Because most species responded positively and uniquely to some combination of fire severity and time-since-fire, these results carry important management implications. Specifically, the variety of burned-forest conditions required by fire-dependent bird species cannot be created through the application of relatively uniform low-severity prescribed fires, through land management practices that serve to reduce fire severity or through post-fire salvage logging, which removes the dead trees required by most disturbance-dependent bird species.
    Full-text · Article · Jan 2016
Show more