650
The Auk 118(3):650?664, 2001
FACTORS GOVERNING THE DISTRIBUTION
OF SWAINSON'S WARBLER ALONG A HYDROLOGICAL GRADIENT
IN GREAT DISMAL SWAMP
GARY R. GRAVES1
Department of Vertebrate Zoology, National Museum of Natural
History, Smithsonian Institution, Washington, D.C. 20560, USA
ABSTRACT.?Due to extensive clearing of bottomland forest in the southeastern United
States, Swainson's Warbler (Limnothlypis swainsonii) is restricted in many drainages to sea-
sonally inundated buffer zones bordering rivers and swamps. This migratory species is es-
pecially vulnerable to ?ooding because of its ground foraging ecology, but little is known
about patterns of habitat occupancy at wetland ecotones. I investigated the physiognomic
and ?oristic correlates of habitat use along a subtle hydrological gradient in the Great Dismal
Swamp, southeastern Virginia. Hydrology is the driving force in?uencing vegetation and
the distribution of Swainson's Warbler in that habitat. Foraging and singing stations of ter-
ritorial males were signi?cantly drier and more ?oristically diverse than unoccupied habitat.
There was scant evidence that the distribution and abundance of particular plant species,
including giant cane (Arundinaria gigantea), in?uenced habitat selection. Instead, Swainson's
Warbler seems to evaluate potential territories on the basis of multiscale physiognomic, hy-
drological, and edaphic characteristics. Territories were characterized by extensive under-
story thickets (median 5 36,220 small woody stems and cane culms per hectare; range,
14,000?81,400/ha), frequent greenbriar tangles, deep shade at ground level, and an abun-
dance of leaf litter overlying moist organic soils. Those sites occurred most frequently in
relatively well-drained tracts of broad-leaf forest that had suffered extensive canopy damage
and windthrow. Data suggest a preference for early successional forest in the current land-
scape or disturbance gaps in primeval forest. Because territories in otherwise optimal habitat
are abandoned when ?ooding extends into the breeding season, it is recommended that the
water table be maintained at subsurface levels from late March through September in natural
areas managed primarily for this species. Direct and indirect environmental factors that in-
?uence the breeding biology of the warbler are summarized in an envirogram. Received 16
February 2000, accepted 20 January 2001.
PATTERNS OF AVIAN DISTRIBUTION AND ABUN-
DANCE depend largely on how individual birds
perceive and use habitat at local, regional, and
continental scales (e.g. Cody 1985, Wiens 1989,
Orians and Wittenberger 1991). The explosive
growth of research on avian habitat selection
since Grinnell (1917) and Lack (1933) has gen-
erated considerable insight on the intertwined
in?uences of habitat physiognomy and ?orist-
ics as correlates of microhabitat preferences
(e.g. MacArthur 1958, Franzreb 1978, Holmes
and Robinson 1981, Robinson and Holmes
1984, Rotenberry 1985, Parrish 1995), the in-
nate behavior of habitat choice (e.g. Partridge
1976, Greenberg 1984, Morton 1990, Morton et
al. 1993), and intraspeci?c geographic variation
in habitat selection (e.g. Dow 1968, Collins
1983, James et al. 1984). Mounting evidence
1 E-mail: graves.gary@nmnh.si.edu
suggests that passerines choose among gross
habitat types on the basis of physiognomy but
may be sensitive to ?oristic cues in selecting
microhabitats (Rotenberry 1985). In exception-
al cases, the ecology of local populations may
be inextricably linked to a single tree species
(e.g. Benkman 1999). More commonly the phys-
iognomic and ?oristic factors governing the
distribution of wide-ranging avian species can
be characterized only in the most general
terms. Analysis of potential regulating in?u-
ences, always dif?cult without experimental
manipulation, is facilitated by a descriptive
species-centered analysis of habitat selection
along environmental or geographic gradients
(Grinnell 1917, James et al. 1984), an approach
that I embraced in this paper.
Swainson's Warbler (Limnothlypis swainsonii)
is an uncommon and locally distributed migra-
tory passerine that breeds in the southeastern
July 2001] 651Swainson's Warbler in Great Dismal Swamp
United States and winters in the Caribbean Ba-
sin (Meanley 1971, Brown and Dickson 1994,
Price et al. 1995, Graves 1996). Recent surveys
indicate that most (.90%) breeding popula-
tions occur in deciduous ?oodplain forest in
the lower Mississippi Valley and on the coastal
plain from eastern Texas to southeastern Vir-
ginia (G. Graves unpubl. data). Within the past
30 years, several populations have disappeared
along the northern periphery of the warbler's
historic breeding range in Delaware, Mary-
land, Missouri, and Illinois (Robbins and Eas-
terla 1992, Brown and Dickson 1994). Some au-
thorities suspect that it may be susceptible to
range-wide population declines because of
habitat destruction and a small wintering
range in the Caribbean Basin (Terborgh 1989,
Rappole 1995). The conversion of bottomland
forests for agriculture, reservoirs, pine planta-
tions, and suburban housing developments,
has substantially reduced the area of suitable
habitat above the mean vernal high-water level
in large parts of its breeding range. Swainson's
Warbler is now restricted to seasonally inun-
dated buffer zones along rivers and swamps in
several drainages. Ultimately, long term sur-
vival of the species may depend on its ability to
reproduce successfully in annually ?ooded
wetlands that are economically risky to devel-
op. Consequently, the discovery and manage-
ment of viable breeding and wintering popu-
lations of this ?agship species are urgent
conservation priorities (Hunter et al. 1993,
Smith et al. 1993, Thompson et al. 1993).
The breeding ecology of Swainson's Warbler
has been the subject of intense ornithological
interest since the nineteenth century (Brewster
1885a,b; Wayne 1886, Meanley 1945, 1966, 1971;
Graves et al. 1996). A clear understanding of its
habitat requirements has been slow to develop,
however, because of the species' rarity, secre-
tive behavior, large territory size, and the in-
hospitable, often impenetrable, nature of its
breeding habitat (Graves 1998). The frequent
association of Swainson's Warbler with giant
cane (Arundinaria gigantea) was noted as early
as the 1880s (Brewster 1885a, Beckham 1887).
Brewster (1885a:72) observed, ``four things
seem indispensable to his existence, viz., water,
tangled thickets, patches of cane, and a rank
growth of semi-aquatic plants.'' Despite pub-
lished exceptions (Widmann 1895), by the
1920s it was widely believed that breeding pop-
ulations were more or less restricted to cane-
brakes in deep swamps and bottomland forest
(e.g. Howell 1924, 1932; Pearson et al. 1919).
The surprising discovery of several popula-
tions of Swainson's Warbler in thickets of rho-
dodendron and mountain laurel in the Appa-
lachian Mountains (Brooks and Legg 1942)
considerably expanded the range of habitats in
which the species was known to breed. Proba-
ble breeding populations have now been doc-
umented in a perplexing variety of habitats that
lack canebrakes: for example, (1) fragments of
old growth bottomland forest, (2) early seral
stages (10?30 years) of deciduous bottomland
forest, (3) young pine plantations with a sig-
ni?cant deciduous component, (4) live oak
thickets and old growth live oak forests in
coastal Texas, (5) second growth bottomland
forest with scrub palmetto undergrowth, (6)
Rhododendron and Kalmia thickets in the Appa-
lachian Mountains, and (7) hardwood cove for-
ests in the Appalachians (Brooks and Legg
1942, Sims and DeGarmo 1948, Meanley 1966,
1971, Brown and Dickson 1994, W. Barrow
pers. comm.; G. Graves unpubl. data). Al-
though there can be little doubt that large cane-
brakes in bottomland forests provide prime
breeding habitat for Swainson's Warbler, it is
clear from the aforementioned examples that
giant cane, per se, is not required.
I investigated the environmental factors cor-
related with the distribution of Swainson's
Warbler along a subtle hydrological gradient in
unfragmented forest in Great Dismal Swamp,
southeastern Virginia. I addressed four princi-
pal questions: (1) Is the habitat physiognomy of
male territories, particularly of foraging micro-
habitats, distinguishable from that of contigu-
ous but unoccupied habitat? (2) Is microhabitat
selection linked to the density and patchiness
of giant cane? (3) Are there ?oristic differences
between occupied and unoccupied sites? (4) Is
habitat selection of Swainson's Warbler in Great
Dismal Swamp related to spatial distribution of
pooled or standing water? I then graphically
summarized the environmental factors that in-
?uence the distribution and abundance of
Swainson's Warbler in an ``envirogram'' (An-
drewartha and Birch 1984, Van Horne and
Wiens 1991, James et al. 1997). I conclude the
paper with a discussion of hydrological man-
agement in Great Dismal Swamp.
652 [Auk, Vol. 118GARY R. GRAVES
FIG. 1. Study area and census transects (hatched
area) east of Suffolk Scarp in Great Dismal Swamp
National Wildlife Refuge, Virginia and North Caro-
lina. Location of study site (black area) is depicted in
the inset map.
METHODS
Site description. Great Dismal Swamp, an 850 km2
palustrine wetland in southeastern Virginia and
northeastern North Carolina, is one of the largest re-
maining tracts of unfragmented forest on the Atlan-
tic Coastal Plain of North America. Present bound-
aries of forest in Great Dismal Swamp in Virginia are
largely coincident with those of the Great Dismal
Swamp National Wildlife Refuge (GDSNWR), which
probably represents the heart of the original swamp.
The study was conducted in the northwestern corner
of GDSNWR (Fig. 1), southeast of the city of Suffolk,
and east of the largely deforested Suffolk Scarp, a
Pleistocene shoreline that rises abruptly to form the
western border of the swamp at 18?21 m above sea
level (Oaks and Coch 1973, Whitehead 1972, White-
head and Oaks 1979, Lichtler and Walker 1979). The
face of the scarp slopes abruptly eastward at rates
approaching 25 m/km, to elevations of 7?9 m above
sea level. The swamp slopes eastward from the base
of the scarp at ;0.2 m/km. Topographic relief within
the study area is ,1 m. The swamp is underlain by
the impermeable Yorktown Formation of early Plio-
cene age. Ground water discharges from the Tabb
Aquifer at the western margin of the swamp and
?ows eastward. For reference, vegetation sampling
transect ` A`'' of Carter et al. (1988) and Carter et al.
(1994) intersects the Suffolk Scarp on the western
boundary of the study area.
The hydrology of Great Dismal Swamp has been
greatly altered by the construction of the Feeder
Ditch (constructed in 1812), connecting Dismal
Swamp Canal to Lake Drummond, and dozens of
other ditches and their accompanying spoil banks
that crisscross the swamp. The ?ow of water in the
study area (Fig. 1) is governed by spoil banks bor-
dering Jericho Ditch (constructed around 1810) and
Lynn Ditch (constructed prior to 1930) and several
control structures and ?xed level weirs in ditches.
Water levels are typically high during winter and
early spring and decrease during the growing season
(April?May) as transpiration increases.
Mature timber has been cut several times since the
shareholders of Dismal Swamp Land Company, in-
cluding George Washington, received claim to the
western half of Great Dismal Swamp in 1764. When
logging began, the study area was dominated by
swamp black gum (Nyssa sylvatica var. bi?ora), water
tupelo (Nyssa aquatica), bald cypress (Taxodium dis-
tichum), and Atlantic white cedar (Chamaecyparis
thyoides) (Kearny 1901, Levy 1991). Older culls and
snags left by previous episodes of logging are
sparsely distributed (,1/ha) throughout the north-
western corner of the GDSNWR. Swamp forest in the
study area, now 40?80 years old, is unfragmented ex-
cept for narrow breaks in the canopy along the ditch-
es and scattered windthrow gaps. The impenetrable
character of the swamp's understory was pointedly
observed by William Byrd (1967:65) during a survey
of the swamp in 1728: ``The reeds [giant cane], which
grew about 12 feet high, were so thick, and so inter-
laced with bamboo-briars [Smilax laurifolia], that our
pioneers were forced to open a passage.'' Today as in
the past, passage through much of the study area can
be accomplished only by cutting a pathway.
The forest canopy is currently dominated by red
maple (Acer rubrum) and swamp black gum, with
lesser amounts of red bay (Persea borbonia), sweetbay
magnolia (Magnolia virginiana), American holly (Ilex
opaca), sweetgum (Liquidambar styraci?ua), tulip pop-
July 2001] 653Swainson's Warbler in Great Dismal Swamp
lar (Liriodendron tulipifera), water oak (Quercus nigra),
laurel oak (Q. laurifolia), bald cypress, Atlantic white
cedar, and loblolly pine (Pinus taeda) (Levy and
Walker 1979, Carter et al. 1994). A rich shrub ?ora
includes coastal pepperbush (Clethra alnifolia), sweet
spires (Itea virginica), swamp sweetbells (Leucothoe
racemosa), coastal sweetbells (L. axillaris), fetterbush
(Lyonia lucida), maleberry (Ly. ligustrina), gallberry
(Ilex glabra, I. coriacea), highbush blueberry (Vaccin-
ium corymbosum), wild raisin (Viburnum nudum),
huckleberry (Gaylussacia frondosa), sweetleaf (Symplo-
cos tinctoria), and pink azalea (Rhododendron nudi?o-
rum). Giant cane (Arundinaria gigantea) is patchily
distributed, but forms dense brakes where it occurs.
Common woody vines in the study area include poi-
son ivy (Rhus radicans), Virginia creeper (Parthenocis-
sus quinquefolia), cross vine (Anisostichus capreolata),
yellow jessamine (Gelsemium sempervirens), climbing
hydrangea (Decumaria barbara), and muscadine
grape (Vitus rotundifolia) (taxonomy follows Mussel-
man et al. 1977). Nearly impenetrable stands of
greenbriar (S. laurifolia, S. rotundifolia) are abundant-
ly distributed throughout the study area, particular-
ly in the vicinity of windthrow gaps.
Climate of the study area is temperate with long
warm summers and mild winters. The growing sea-
son varies from ,174 days (10% of years) to .209
days (10% of years), averaging 192 days (Reber et al.
1981). Annual rainfall averages 1,280 mm with most
precipitation falling in July and August (U.S. Weather
Bureau 1965).
Soil characteristics of the study area are of signif-
icant interest in view of the terrestrial foraging ecol-
ogy of Swainson's Warbler (Graves 1998). The Suffolk
Scarp is crowned by well-drained sandy loams with
low organic content. Water-retaining capacity and
organic content of soils increase rapidly with de-
creasing elevation along the east-facing ?ank of the
scarp. Soils along the transects are poorly drained,
moderately to extremely acidic, and classi?ed in the
Torhunta-Deloss and Pungo-Belhaven associations
(Reber et al. 1981). Both soil associations are subject
to regular ?ooding in winter and spring. Scattered
to extensive pools of water may persist during the
nesting season (April?July), particularly in the
southwestern corner of the study site.
Spatial distribution of warblers. The Great Dismal
Swamp supports the northernmost viable popula-
tion of Swainson's Warbler on the Atlantic Coastal
Plain. Males arrive and establish territories during
the third week of April (Meanley 1966, 1971) when
the water table is at the highest level during the
breeding cycle. The earliest record of nest building is
23 April (Meanley 1971).
Motorized access to the study area was limited to
roads built atop spoil banks (0.5?1.0 m above ground
level) that parallel ditches. I mapped territorial
boundaries each year (17 May?3 June, 1989?1994)
with the aid of ``playback-and-follow'' trials along 18
km of transects along spoilbanks. A full discussion
of census techniques will be published elsewhere.
Brie?y, loop cassette tapes (60 s) were prepared from
song recordings of two or more males interspersed
with 5 s of call notes. Recordings were broadcast
from a dual-speaker portable tape player. Power out-
put was variable due to battery strength and tem-
perature, but the audio output was adjusted before
every census run so that songs were faintly audible
to me at a distance of 100 m (unobstructed by vege-
tation). The estimated effective census width was
;250 m based on the response of territorial males to
the playback of songs. Approximately half of the
males was color-banded each year.
Vegetation sampling in Swainson's Warbler territo-
ries.?Swainson's Warbler territories in Great Dismal
Swamp were large (3?18 ha) and physiognomically
complex. The sheer size of territories presented a
vegetation sampling problem. First, when sampling
sites were randomly distributed within a territory,
many plots were located in patches of habitat that
were seldom used by warblers (e.g. areas with sparse
undergrowth) or avoided entirely (e.g. ?ooded
swales). Furthermore, the labor needed to quantify
the vegetation on just 5% of a typical territory (10 ha)
was estimated at 45 man hours (;4 h/sampling
plot). Given the limitations in amount of help, I cen-
tered a single circular sampling plot (0.045 ha; di-
ameter 5 24 m) at the ?rst terrestrial site in each ter-
ritory at which the undisturbed color-banded male
was observed to both sing and forage. Sampling at
dual purpose singing?foraging sites ensured that
physiognomic and ?oristic data actually correspond-
ed to microhabitats used by Swainson's Warblers. Al-
though this paper focuses on habitat selection of
males, a substantial body of data indicates that males
and females use similar foraging microhabitats (G.
Graves 1998 unpubl. data). Foraging observations
and the consequent placement of sampling plots
were made 40?120 m from roadsides to minimize
``ditch'' effects in vegetation sampling. In hindsight,
that cautionary procedure was deemed unnecessary
because understory thickets bordering ditches did
not differ physiognomically or ?oristically from
those associated with frequent canopy disturbances
in the forest interior.
All trees (diameter at breast height [DBH] .5 cm)
occurring in sampling plots (0.045 ha) were mea-
sured (diameter to nearest centimeter) and identi?ed
to species. Diameters were later converted to size
classes for comparison, whereas basal area was com-
puted from raw ?eld measurements. Woody vines
supported by trees were also counted (1.4 m above
ground) and identi?ed.
Small woody stems (DBH ,5 cm; i.e. shrubs, cane,
vines, and tree saplings) were counted and identi?ed
to species in four subplots (12.6 m2, circle diameter
5 4 m) positioned at the cardinal compass coordi-
nates on the perimeter of the larger primary plot. The
654 [Auk, Vol. 118GARY R. GRAVES
typical method of estimating the density of small
stems by counting the number intercepted by an ob-
server's outstretched arms (e.g. James and Shugart
1970) was unfeasible, if not impossible, to perform
because of the prevalence of thorny greenbriar and
the immense number of stems in the forest understo-
ry. I obtained exact counts by clipping all small
stems within each circular subplot at a height of 0.5
m above ground. Patchiness of cane was estimated by
the coef?cient of variation (CV) of culm counts
among the four shrub subplots. Habitat data were
collected on warbler territories, 23 June to 5 July 1993
(n 5 15) and 22 June to 4 July 1994 (n 5 15).
Characterization of unoccupied habitat. Patterns of
habitat selection in passerine birds are often corre-
lated with intraspeci?c population density (e.g. Klu-
ijver 1951, Fretwell and Lucas 1970, Wiens and Ro-
tenberry 1981, Van Horne 1983, Wiens 1989). For
instance, low-quality habitat is often occupied only
during high-density years. Characterization of hab-
itat selection thus has limited relevance in the ab-
sence of population density estimates. Swainson's
Warbler territories in Great Dismal Swamp are large
relative to those of other migratory wood warblers
(Nolan 1978, Morse 1989). Annual censuses revealed
that a majority of territories appeared to abut one an-
other, whereas a minority were separated from their
nearest neighbors by signi?cant gaps (.750 m). Al-
though occupancy patterns ?uctuated from year to
year, the census population was relatively stable (G.
Graves unpubl. data). Cumulative overlays of terri-
tory maps prepared after the fourth annual census
(1992) showed that virtually all habitat traversed by
census transects had been occupied during one or
more breeding seasons. However, an unoccupied
area (;1.2 km in length) was identi?ed along both
sides of Lynn Ditch immediately north of Middle
Ditch (Fig. 1). My puzzlement about the underlying
cause of the chronic distributional hiatus?the only
such gap detected along 18 km of census transects in
unfragmented forest?provided the motivation for
this paper.
The unoccupied zone was sampled in a strati?ed
randomized design. The 1.2 km segment was grid-
ded into 12 blocks (200 m long 3 100 m deep), six on
each side of the ditch. Four sampling plots, identical
to those used in warbler territories, were randomly
placed in each block such that plots did not overlap
or occur within 10 m of the ditch or adjacent road. A
total of 48 plots were sampled in the unoccupied
zone, 2 June to 25 July 1992. Data from four plots
were omitted from the analyses after a territorial
boundary narrowly overlapped the southeastern
corner of the unoccupied zone in 1994.
Pooled water. Flooded territories are abandoned
presumably because terrestrial leaf litter is a critical
foraging resource (G. Graves unpubl. data). Accord-
ingly, the presence of pooled water was recorded at
20 standardized locations within each sampling plot
(?ve points within each of the four shrub subplots).
Statistics and hypothesis testing. The original suite
of 60 habitat variables was reduced to 24 by combin-
ing some variables and eliminating others whose fre-
quency of occurrence was too limited to be useful in
analyses or that were highly correlated with other
variables (Table 1). Ranges and medians of the re-
tained variables were presented for comparison with
previously conducted studies (Eddleman et al. 1980,
Thomas et al. 1996). I tested variables for goodness
of ?t to a normal distribution with Lilliefors test. Ten
of 24 variables exhibited signi?cant deviation from
normality, even after being subjected to variance-sta-
bilizing transformations (Table 1). Consequently,
nonparametric one-way analysis of variance (Krus-
kal-Wallis ANOVA) was used to examine intergroup
differences between occupied and unoccupied hab-
itat, and between warbler territories sampled in 1993
and 1994. Signi?cance values for each suite of hy-
pothesis tests were Bonferroni adjusted for the num-
ber of simultaneous tests (P 5 0.05/n).
The relationship between pooled water and habitat
variables was explored with bivariate scatterplots
and Spearman rank correlation coef?cients. The cu-
mulative size-class distribution of trees (rows 3 col-
umns) in occupied and unoccupied habitat was an-
alyzed with a chi-square test of homogeneity.
I used logistic regression (Hosmer and Lemeshow
1989) of transformed data as a supplementary meth-
od of identifying the physiognomic and ?oristic var-
iables most closely associated with microhabitat se-
lection of Swainson's Warbler. The dependent
variable was binary (WARBLER: 0,1), designating
unoccupied or occupied habitat, respectively. Those in-
dependent variables exhibiting signi?cant P-values
(,0.05, n 5 15) in univariate logistic regression anal-
yses were entered into a multiple logistic regression
analysis employing a stepwise backward elimination
process (P 5 0.10). Backward elimination removed
variables one by one that did not signi?cantly en-
hance the model's ability to correctly classify site oc-
cupancy. In order to identify less complex models
that performed reasonably well, I conducted addi-
tional simple multiple logistic regression analyses on
all possible subsets of variables drawn from the re-
duced list produced by the stepwise procedure. All
analyses were performed with SYSTAT Version 8
(SPSS, 1998).
RESULTS
Habitat physiognomy. Between-year differ-
ences in habitat variables recorded in Swain-
son's Warbler territories in 1993 and 1994 were
not statistically signi?cant (Kruskal-Wallis
ANOVA, P . 0.05 for all variables). Conse-
Ju
ly
2001]
655
Sw
ainson's
W
arbler
in
G
reat
D
ism
al
Sw
am
p
TABLE 1. Medians (ranges) of physiognomic and ?oristic variables measured on 0.045 ha plots in 30 territories of Swainson's Warblers and in 44 unoccupied
areas in the Great Dismal Swamp.
Code Variable Territories Unoccupied habitat x2(df 5 1)?
WATa
BASb
ACERb
NYSSc
Percent coverage of water
Total basal area (m2) of trees (DBH . 5 cm)/ha
Basal area (m2) Acer rubrum (DBH . 5 cm)/ha
Basal area (m2) Nyssa sylvatica (DBH . 5 cm)/ha
0.0 (0?0)
37.3 (26.7?63.5)
23.8 (14.6?46.5)
5.7 (0?29.5)
35.0 (0?100)
45.0 (33.1?78.9)
22.6 (7.1?33.8)
16.3 (6.7?50.0)
26.78***
15.28**
0.08
34.73***
BOTHc
ONEc
TWOb
THREb
Basal area (m2) all other tree species (DBH . 5 cm)/ha
Trees (DBH 5 5?14.9 cm)/ha
Trees (DBH 5 15?24.9 cm)/ha
Trees (DBH 5 25?39.9 cm)/ha
4.6 (0.4?21.7)
166.7 (0?556)
244.4 (44?556)
144.4 (44?467)
1.3 (0?23.4)
77.7 (0?267)
300.0 (111?711)
322.2 (156?600)
7.54
18.62***
3.10
60.58***
FOURb
FIVEc
SIXc
TREEb
TSPEb
Trees (DBH 5 40?59.9 cm)/ha
Trees (DBH 5 60?79.9 cm)/ha
Trees (DBH . 80 cm)/ha
Trees (DBH . 5 cm)/ha
Tree species (DBH . 5 cm)/0.045 ha
111.1 (22?244)
44.4 (0?133)
0.0 (0?89)
777.8 (267?1,333)
5.0 (3?7)
111.1 (44?289)
44.4 (0?133)
0.0 (0?67)
900.0 (578?1,444)
4.0 (2?6)
0.12
0.04
0.82
5.46
11.19*
VINEd
VSPEb
CANEe
CVCAc
Vines (including Smilax spp.) supported by trees/ha
Vine species/0.045 ha
Cane culms/ha
Cane patchiness: coef?cient of variation
of cane culms among four shrub plots
555.6 (22?2,889)
3.0 (1?5)
4,400 (0?50,400)
0.55 (0?2.0)
666.7 (111?2,755)
2.0 (1?5)
23,300 (0?8,000)
0.82 (0?2.0)
0.63
0.25
22.60***
3.15
SHRUd
SSPEb
CLETb
LEUCc
PERSc
SMILc
Woody stems (DBH , 5 cm)/ha, excluding Smilax spp./ha
Shrub species/50.3 m2 (including tree saplings)
Clethra alnifolia stems/ha
Leucothoe racemosa stems/ha
Persea borbonia stems (DBH ,5 cm)/ha
Smilax spp. stems/ha
27,000 (9,200?45,600)
6 (4?9)
14,100 (3,600?29,000)
0 (0?4,600)
3,700 (0?8,800)
3,300 (0?41,000)
38,800 (9,800?73,600)
6.5 (4?10)
33,800 (2,800?70,600)
500 (0?4,000)
800 (0?5,000)
0 (0?6,600)
12.23*
0.16
23.42***
6.15
15.37***
28.49***
? P-values adjusted for number of tests (n/24): *P , 0.05; P , 0.01; ***P , 0.001.
Data transformations for principal components and logistic regression analyses: aarcsine; bsquare-root(x); csquare-root(x 1 1); dlog10(x); elog10(x 1 1).
656 [Auk, Vol. 118GARY R. GRAVES
FIG. 2. Parallel coordinate plots of tree diameter
classes in Swainson's Warbler territories (n 5 30) and
in unoccupied habitat (n 5 44) in the Great Dismal
Swamp. Size classes (DBH): (1) 5?14.9 cm; (2) 15?24.9
cm; (3) 25?39.9 cm; (4) 40?59.9 cm; (5) 60?79.9 cm;
and (6) . 80 cm.
TABLE 2. Relative abundance and frequency of occurrence of tree species (DBH . 5 cm) in Swainson's War-
bler territories and in nearby unoccupied habitat in Great Dismal Swamp.
Tree species
Territories (n 5 30)
% stemsa % basal areaa No. plots
Unoccupied habitat (n 5 44)
% stems % basal areaa No. plots
Acer rubrum
Nyssa sylvatica
Persea borbonia
Ilex opaca
44.1
27.0
14.4
5.3
59.4
22.2
2.9
1.1
30
27
27
16
47.1
44.1
4.2
2.3
50.2
42.1
0.6
0.4
44
44
28
20
Liquidambar styraci?ua
Taxodium distichum
Liriodendron tulipifera
Other species
4.4
2.1
1.2
1.6b
5.0
6.0
2.5
1.0
10
10
8
NA
0.7
0.2
0.5
0.6c
1.8
0.5
0.9
3.5
10
4
8
NA
a Percentages do not add to 1.00 due to rounding error.
b Pinus taeda, Quercus nigra, Q. laurifolia, Asimina triloba, Pyrus spp., Vaccinium spp.
c Pinus taeda, Fraxinus spp.
quently, territory data from both years were
combined for statistical analyses.
Univariate analyses indicated that 13 of 24
habitat variables exhibited intergroup differ-
ences between warbler territories and random-
ly sampled plots in unoccupied habitat (Table
1). The size class distribution of trees (ONE?
SIX) varied substantially among sampling
groups (x2 5 172.15, df 5 5, P , 0.0001) (Fig.
2). Territories contained signi?cantly more
trees in the smallest diameter class (5?14.9 cm
DBH, ONE) but fewer medium-sized trees (25?
39.9 cm DBH, THRE). The surfeit of medium-
sized trees in unoccupied habitat was due pri-
marily to even-aged stands of black gum
(NYSS). The density of trees in large diameter
classes (.40 cm DBH, FOUR?SIX) and the
overall density of trees (TREE) were similar in
territories and unoccupied habitat. However,
total basal area (BAS) averaged ;16% less in
territories. The density of small woody stems
was lower in territories (median 5 27,000 stems
per hectare; range, 9,200?45,600) than in un-
occupied habitat (median 5 38,800 stems per
hectare; range, 9,800?73,600). There were no
differences between territories and unoccupied
habitat in the number of vines supported by
trees.
Habitat ?oristics. There was little evidence
that Swainson's Warbler selects habitat on the
basis of speci?c ?oristic cues. Red maple
(ACER) was the dominant tree species in war-
bler territories (44.1% stems; 59.4% basal area)
and in unoccupied habitat (47.1% stems; 50.2%
basal area). Swamp black gum (NYSS) was
nearly as abundant, comprising 27.0% of stems
and 22.2% of basal area in territories, and
44.1% of stems and 42.1% of basal area in un-
occupied habitat. No other tree species com-
posed more than 14.4 and 4.2%, respectively, of
the stems in territories and unoccupied habitat
(Table 2). Territories supported a higher diver-
sity of tree species (TSPE), but there were no
differences between occupied and unoccupied
habitat in the number of vine (VSPE) or shrub
species (SSPE) (Table 1). Three of the four most
common shrubs exhibited signi?cant density
differences among sampling groups: coastal
pepperbush (CLET) was more abundant in un-
occupied habitat, whereas fetterbush (LYON)
and redbay (PERS) were prominent understory
components in warbler territories. Thorny
greenbriar (SMIL) tangles were conspicuous
components of understory thickets in warbler
territories (median 5 3,300 stems per hectare;
range, 0?41,000), but were far less common in
unoccupied habitat (median 5 0 stems per
hectare; range, 0?6,600).
July 2001] 657Swainson's Warbler in Great Dismal Swamp
FIG. 3. Scattergrams of selected habitat variables
in Swainson's Warbler territories (?lled circles 5 30)
and in unoccupied habitat (open circles 5 44) in the
Great Dismal Swamp. Circle diameters are propor-
tional to the percentage coverage of pooled water in
sampling plots.
TABLE 3. Correlation between pooled water (WAT)
and habitat variables. Signi?cant Spearman rank
correlation coef?cients (P , 0.05/23 5 0.002, one-
tailed) are given in boldface.
Habitat
variable Pooled water
BAS
ACER
NYSS
BOTH
ONE
0.35
0.36
0.47
20.12
20.11
TWO
THRE
FOUR
FIVE
SIX
0.25
0.37
0.22
20.12
0.03
TREE
TSPE
VINE
VSPE
0.29
20.22
20.04
20.03
CANE
CVCA
SHRU
SSPE
0.44
0.23
0.05
0.08
CLET
LEUC
LYON
PERS
SMIL
0.13
0.37
20.16
20.36
20.31
The density of giant cane was lower on ter-
ritories (median 5 0 culms per hectare; range,
0?50,400/ha) than in unoccupied habitat (me-
dian 5 23,300 culms per hectare; range, 0?
88,000). Moreover, cane was only recorded on
57% of the vegetation sampling plots (17 of 30)
in territories. Patchiness of cane (CVCA) was
similar in territories and in unoccupied habitat.
Pooled water. Hydrology appears to be the
driving force in?uencing vegetation and the
distribution of Swainson's Warbler in Great
Dismal Swamp (Fig. 3, Table 3). Standing or
pooled water (WAT) was logged on 61% (43%
of 880 reference points) of the plots in unoc-
cupied habitat. By comparison, water was un-
recorded on vegetation plots (0% of 600 refer-
ence points) in warbler territories, although
scattered pools or water-?lled ditches occurred
within the boundaries of every territory.
Pooled water was positively correlated with to-
tal basal area (BAS), basal areas of red maple
(ACER) and swamp black gum (NYSS), the
number of medium-sized trees (THRE), and
the densities of cane (CANE) and swamp
sweetbells (LEUC), but negatively correlated
with the density of small redbay stems (PERS)
(Table 3). Several signi?cant ?oristic correla-
tions appear to be directly attributable to con-
trasting hydrological adaptations?for exam-
ple, ACER versus NYSS (rs 5 20.43, Fig. 3),
CLET versus LEUC (rs 5 20.46), CLET versus
PERS (rs 5 20.38), CLET vs. SMIL (rs 5 20.59).
Multivariate perspective. From the logistic re-
gression (log likelihood, x2 5 73.07, df 5 5, P ,
0.00001), the estimated probability (P) of oc-
currence of Swainson's Warbler was:
ln(P/[1 2 P]) 5 21.13 ? 0.01 (NYSS)
2 0.61 (THRE)
1 0.50 (TREE)
2 0.55 (CANE)
1 0.05 (SMIL)
This model correctly classi?ed the presence
or absence of Swainson's Warbler on 90.4% of
the 74 plots (Table 4). Habitat variables strongly
associated with the distribution of Swainson's
Warbler include the basal area of swamp black
gum (NYSS), the number of medium-sized
trees (THRE), the total number of trees (TREE),
and the densities of cane (CANE) and green-
briar (SMIL). However, the unwieldy number
of variables retained in the model does little to
658 [Auk, Vol. 118GARY R. GRAVES
TABLE 4. Multiple logistic regression models with two or three habitat variables that correctly classify .80%
of vegetation plots. The presence or absence of Swainson's Warbler was the dependent variable.
Regression model
Log
likelihood
% correctly
classi?ed P
1. Warbler 5 6.28 ? 0.01 (NYSS) 2 0.28(THRE) 1 0.04(SMIL)
2. Warbler 5 4.17 ? 0.01 (NYSS) 2 0.60(CANE) 1 0.05(SMIL)
3. Warbler 5 21.02 ? 0.02 (NYSS) 1 0.14(TREE) 1 0.06(SMIL)
4. Warbler 5 2.61 ? 0.01(NYSS) 1 0.06(SMIL)
5. Warbler 5 2.92 ? 0.01(NYSS) 2 0.71(THRE) 1 0.36(TREE)
63.90
62.30
60.67
59.46
58.64
86.6
85.4
84.7
84.4
83.6
,0.00001
,0.00001
,0.00001
,0.00001
,0.00001
6. Warbler 5 6.75 ? 0.40(THRE) 2 0.65(CANE) 1 0.04(SMIL)
7. Warbler 5 10.42 ? 0.01(NYSS) 2 0.71(THRE) 2 0.52(CANE)
8. Warbler 5 8.96 ? 0.01(NYSS) 2 0.44(THRE)
9. Warbler 5 0.57 ? 0.61(THRE) 1 0.25(TREE) 1 0.03(SMIL)
58.78
54.53
51.70
54.33
83.3
82.1
81.2
81.0
,0.00001
,0.00001
,0.00001
,0.00001
facilitate the rapid assessment of habitat suit-
ability. How well do less complex models per-
form? I conducted simple multiple logistic re-
gression analyses on all possible subsets of the
variables drawn from the ?ve retained by the
initial stepwise regression. Nine simpler mod-
els provided reasonably good (.80%) classi?-
cation success (Table 4). In particular, pairwise
combinations of NYSS (basal area of swamp
black gum), THRE (number of medium-sized
trees), and SMIL (number of greenbriar stems)
correctly classi?ed 80% or more of the habitat
plots. NYSS (rS 5 0.47) and THRE (rS 5 0.37)
were signi?cantly correlated with the presence
of standing water (Table 3), and with one an-
other (rS 5 0.52). Quanti?cation of two key var-
iables, NYSS and SMIL (84% correctly classi-
?ed), may provide a rapid and reasonably
reliable assay of habitat suitability in the study
area.
DISCUSSION
The envirogram summarizes the environ-
mental factors that directly (centrum) or indi-
rectly (web) in?uence or control the distribu-
tion and abundance of Swainson's Warbler (Fig.
4). Environmental factors are partitioned into
four compartments or submodels according to
their effects (negative/positive) on the warbler
and the reciprocal effects (negative/positive/
neutral) of the warbler on the factor. Resources
are de?ned as factors that have a positive in?u-
ence on the warbler but that are unchanged or
decreased by the actions of the warbler (e.g. ar-
thropods). Hazards negatively in?uence the
warbler but receive no reciprocal advantage or
bene?t from the interaction (e.g. ?ooding). In
contrast, predators negatively affect the war-
bler and acquire or receive some bene?t from
the association. The fourth compartment con-
tains factors or components related to repro-
duction. Because many aspects of the warbler's
breeding biology are poorly known, the result-
ing ?ow diagram must be regarded as a hy-
pothetical working model. Nevertheless, by fo-
cusing attention on the interrelationship
between habitat elements and environmental
in?uences that are believed crucial for the
maintenance of viable populations, the envi-
rogram serves as a heuristic model for habitat
managers and as a guide for future research.
The supplementary narrative that follows, as
distinguished from logic equations (e.g. Niven
and Stewart 1987, Niven and Abel 1991) or en-
virogram itself, summarizes the best docu-
mented components of the web and centrum in
nontechnical language.
In the resource compartment of the enviro-
gram, the availability of arthropods is linked to
the characteristics of soil and leaf litter and to
a cascade of indirect effects. Swainson's War-
bler is a terrestrial dead-leaf specialist with a
limited repertoire of foraging behaviors direct-
ed toward litter arthropods (Meanley 1970,
Graves 1998). The quality and quantity of leaf
litter depend directly on the taxonomic diver-
sity and basal area of tree species, which in
turn depends on hydrology, topography, and
soil characteristics. Breeding territories are re-
stricted to moist soils throughout the species'
breeding range but periodic ?ooding can have
a profound in?uence on the year-to-year oc-
cupancy of otherwise optimal sites (Meanley
1966, 1971; Graves 1998).
A signi?cant fraction of the remaining
Swainson's Warbler habitat, particularly in the
lower Mississippi Valley (see Twedt and
July 2001] 659Swainson's Warbler in Great Dismal Swamp
FIG. 4. Envirogram for Swainson's Warbler (Limnothlypis swainsonii). Elements in the submodels (resourc-
es, hazards, predators, reproduction) refer speci?cally to factors relevant to the breeding biology of Swain-
son's Warbler in Great Dismal Swamp, but also apply generally to breeding populations on the Atlantic and
Gulf coastal plains and in the lower Mississippi Valley. Deforestation does not currently threaten populations
in Great Dismal Swamp National Wildlife Refuge. ``Token'' denotes a resource prerequisite for the establish-
ment of territories and for reproduction. Question marks designate hypothesized but unsubstantiated factors.
Footnoted references: (1) Brewster 1885a; (2) Brown and Dickson 1994; (3) Eddleman et al. 1980; (4) Graves
1998; (5) Graves et al. 1996; (6) Graves, this paper; (7) G. Graves unpubl. data; (8) Hamel 1981; (9) Meanley
1945; (10) Meanley 1966; (11) Meanley 1970; (12) Meanley 1971; (13) Meanley 1982; (14) Pashley and Barrow
1993; (15) Sims and DeGarmo 1948; (16) Strong 2000; (17) Wayne 1886.
660 [Auk, Vol. 118GARY R. GRAVES
Loesch 1999), is sandwiched between ?ood
mitigation levees and river channels, where the
duration and severity of annual ?ooding now
routinely exceeds annual baselines document-
ed in pre-1920 records (G. Graves unpubl.
data). On the short term, ?ooded territories are
vacated. Observations suggest that abandon-
ment is stimulated by the inundation of leaf lit-
ter, a critical foraging resource on the breeding
and wintering grounds, although nesting sites
(0.3?3.1 m above ground) are also affected by
severe ?ooding. Populations displaced by
?ooding disperse to higher terraces within
?oodplains or to the margins of swamps, but
there is no evidence that displaced populations
successfully reproduce in marginal habitats (G.
Graves unpubl. data). Much of the observed
variance in territory size in Great Dismal
Swamp (Meanley 1966, 1969, 1971) and other
locations may be explained by annual variation
in the water table.
Repeated censuses in several locations indi-
cate that catastrophic ?oods affect habitat oc-
cupancy for several years by scouring and
washing away accumulated leaf litter. Many
large tracts (.500 ha) of bottomland forest ad-
jacent to river channels and subject to annual
?ooding (e.g. Alabama River, Pascagoula Riv-
er) support vanishingly small populations of
Swainson's Warblers (G. Graves unpubl. data).
Hydroperiod thus emerges as the principal
management concern for this species in annu-
ally ?ooded habitats in the core of its geo-
graphic range.
Foraging and singing stations of territorial
males in Great Dismal Swamp were signi?-
cantly drier and more ?oristically diverse than
unoccupied sites distributed along a continu-
ous and subtle hydrological gradient. There
was insuf?cient evidence to indicate that ?oris-
tic cues, to the extent those can be disassociated
from physiognomic factors, were used to select
territories. Instead, this species seems to eval-
uate potential territories on the basis of multi-
scale physiognomic, hydrological, and edaphic
characteristics. Territories in Great Dismal
Swamp were characterized by extensive under-
story thickets (median 5 36,220 small woody
stems and cane culms per hectare; range
14,000?81,400/ha), frequent greenbriar tan-
gles, deep shade at ground level, and an abun-
dance of leaf litter overlying moist organic
soils. Stem densities reported here were rough-
ly comparable to those reported for geograph-
ically peripheral sites in Missouri (Thomas et
al. 1996) and Illinois (Eddleman et al. 1980),
and for populations within the core breeding
range in Georgia (Meanley 1966), South Caro-
lina (Hamel 1981), Louisiana (G. Graves un-
publ. data), Arkansas (G. Graves unpubl. data),
Mississippi (G. Graves unpubl. data), and Flor-
ida (G. Graves unpubl. data). Habitat optimal-
ity, however, does not appear to be linearly cor-
related with the abundance of small woody
stems (Fig. 3). Rather, a range of understory
stem densities provides the requisite patches of
dense cover associated with nest sites (Brew-
ster 1885b, Wayne 1886, Meanley 1945, 1966,
1970, 1971; Graves 1992, and unpubl. data;
Brown and Dickson 1994, Thomas et al. 1996),
as well as the sparsely vegetated but heavily
shaded glades in which the warblers preferen-
tially forage (Graves 1998). Data from Great
Dismal Swamp strengthens the conclusion that
giant cane, although often a correlate of habitat
occupancy, is not an essential component of op-
timal habitat. Additionally, canopy height ap-
pears to exert little in?uence in habitat selec-
tion beyond the extent to which understory
thickets are associated with disturbance gaps
in taller closed-canopy forests. Data from nu-
merous sites on the Atlantic and Gulf coastal
plains, from the lower Mississippi Valley, and
from Great Dismal Swamp indicate a prefer-
ence for early successional forest in the current
landscape or disturbance gaps in primeval for-
est (e.g. Delta National Forest, Sharkey County,
Mississippi; G. Graves unpubl. data). Accord-
ingly, frequent gap simulations should be in-
corporated in forest restoration efforts aimed at
this species (Eddleman et al. 1980, Pashley and
Barrow 1993).
Implications for hydrological management.
Current management strategies for Swainson's
Warbler call for regeneration and maintenance
of canebrakes (Eddleman et al. 1980, Thomas et
al. 1994), the creation of small canopy gaps by
cutting individual trees (Eddleman et al. 1980,
Pashley and Barrow 1993), and the generation
of larger gaps (up to 4 acres) through clearcut-
ting (Eddleman et al. 1980). On the other hand,
hydrology has not been not discussed as a
management concern, even though the devas-
tating effects of ?ooding on breeding popula-
tions have been known since Meanley's (1966,
1971) pioneering work along the Ocmulgee
July 2001] 661Swainson's Warbler in Great Dismal Swamp
River. Although a full exploration of hydrolog-
ical in?uences on Swainson's Warbler popula-
tions is beyond the scope of this paper, a few
key points are germane to the present
discussion.
Natural hydrological cycles in Great Dismal
Swamp were signi?cantly altered by the con-
struction of ditches and canals long before the
?rst biological and geological surveys were
made (e.g. Chickering 1873, Kearney 1901).
With the exception of periodic assessments of
surface levels of Lake Drummond and Dismal
Swamp Canal, there are no long-term data use-
ful for establishing hydroperiod baselines for
most reference sites in the GDSNWR. Although
the refuge is currently partitioned into six wa-
ter management units, the extensive network of
ditches (240 km) and spoil banks functionally
divides the swamp into a patchwork of 401 hy-
drological compartments. Current water man-
agement capacities are rudimentary, limited to
the manipulation of ?xed weirs, control struc-
tures, and locks on ditches. Hydrological res-
toration is an untenable goal without perma-
nently damming ditches at ground level
combined with the widescale construction of
culverts to breach spoilbanks, permitting the
natural ?ow of water.
Hydrological policy in the GDSNWR was
administered with limited personnel on a
largely ad hoc basis from the inception of the
refuge in 1974 through 1996, favoring the re-
tention of surface water near peak levels dur-
ing the growing season in order to provide
Wood Duck (Aix sponsa) breeding habitat and
to promote the growth of bald cypress and
water tupelo (L. Culp pers. comm.). An un-
intended consequence of that practice was the
?ooding of large tracts of optimal habitat for
the ground-foraging Swainson's Warbler. High
water levels during the growing season also
thwarted efforts to regenerate stands of Atlan-
tic white cedar, which requires exposed soil
for germination. Accordingly, concerted ef-
forts have been made since 1996 to permit sur-
face water levels to recede naturally to sub-
surface levels during the growing season in
some hydrological compartments. In any
event, the water table of natural areas man-
aged primarily for Swainson's Warbler should
be maintained at subsurface levels from
March through September.
ACKNOWLEDGMENTS
This paper is dedicated to Brooke Meanley. I owe
special thanks to Kamal Mohamed for collecting
vegetation data under physical conditions that few
people would voluntarily endure. I thank the staff of
Great Dismal Swamp National Wildlife Refuge
(Lloyd Culp, David Brownlee, Ralph Keel, Bryan
Poovey), Lytton Musselman (Old Dominion Univer-
sity), and Don Schwab (Virginia Department of
Game and Inland Fisheries) for ?eld support and as-
sistance. The manuscript bene?ted from the insight-
ful reviews of Wylie Barrow, Jeff Brawn, Raymond
Brown, Lloyd Culp, William Eddleman, Todd Engs-
trom, Paul Hamel, Frances James, Gerald Niemi, Tom
Sherry, and Kevin Winker. Brian Schmidt helped
with the ?gures. Banding permits were issued by the
U.S. Fish and Wildlife Service and the Virginia De-
partment of Game and Inland Fisheries. The project
was funded by the Smithsonian Research Opportu-
nities Fund and the U.S. Fish and Wildlife Service
(14-48-0005-92-9013 and 14-48-0009-946).
LITERATURE CITED
ANDREWARTHA, H. G., AND L. C. BIRCH. 1984. The
Ecological Web: More on the Distribution and
Abundance of Animals. University of Chicago
Press, Chicago.
BECKHAM, C. W. 1887. Additions to the avi-fauna of
Bayou Sara, Louisiana. Auk 4:299?306.
BENKMAN, C. 1999. The selection mosaic and diver-
sifying coevolution between crossbills and lod-
gepole pine. American Naturalist 153(Supple-
ment):75?91.
BREWSTER, W. 1885a. Swainson's Warbler. Auk 2:65?
80.
BREWSTER, W. 1885b. The nest and eggs of Swainson's
Warbler (Helinaia swainsoni). Forest and Stream
24:468.
BROOKS, M., AND W. C. LEGG. 1942. Swainson's War-
bler in Nicholas County, West Virginia. Auk 59:
76?86.
BROWN, R. E., AND J. G. DICKSON. 1994. Swainson's
Warbler (Limnothlypis swainsonii). In The Birds of
North America, no. 126 (A. Poole and F. Gill,
Eds.). Academy of Natural Sciences, Philadel-
phia, and American Ornithologists' Union,
Washington, D.C.
BYRD, W. 1967. Histories of the dividing line betwixt
Virginia and North Carolina. (Reprint.) Dover
Publications, New York.
CARTER, V., P. T. GAMMON, AND M. K. GARRETT. 1994.
Ecotone dynamics and boundary determination
in the Great Dismal Swamp. Ecological Appli-
cations 4:189?203.
CARTER, V., M. K. GARRETT, AND P. T. GAMMON. 1988.
Wetland boundary determination in the Great
662 [Auk, Vol. 118GARY R. GRAVES
Dismal Swamp using weighted averages. Water
Resources Bulletin 24:297?306.
CHICKERING, J. W. 1873. The ?ora of the Dismal
Swamp. American Naturalist 7:521?524.
CODY, M. L., ED. 1985. Habitat Selection in Birds. Ac-
ademic Press, Orlando, Florida.
COLLINS, S. L. 1983. Geographic variation in habitat
structure of the Black-throated Green Warbler
(Dendroica virens). Auk 100:382?389.
DOW, D. D. 1968. Home range and habitat of the car-
dinal in peripheral and central populations. Ca-
nadian Journal of Zoology 47:103?114.
EDDLEMAN, W. R., K. E. EVANS, AND W. H. ELDER.
1980. Habitat characteristics and management of
Swainson's Warbler in southern Illinois. Wildlife
Society Bulletin 8:228?233.
FRANZREB, K. E. 1978. Tree species used by birds in
logged and unlogged mixed-coniferous forests.
Wilson Bulletin 90:221?238.
FRETWELL, S. D., AND H. L. LUCAS, JR. 1970. On ter-
ritorial behavior and other factors in?uencing
habitat distribution in birds. I. Theoretical de-
velopment. Acta Biotheoretica 19:16?36.
GRAVES, G. R. 1992. A case of aggregated nest place-
ment and probable polygyny in the Swainson's
Warbler. Wilson Bulletin 104:370?373.
GRAVES, G. R. 1996. Censusing wintering popula-
tions of Swainson's Warblers: Surveys in the Blue
Mountains of Jamaica. Wilson Bulletin 108:94?
103.
GRAVES, G. R. 1998. Stereotyped foraging behavior of
the Swainson's Warbler. Journal of Field Orni-
thology 69:121?127.
GRAVES, G. R., M. B. SIMPSON, Jr., AND L. D. STE-
PHENS. 1996. Joseph LeConte and the discovery
of the nest of Swainson's Warbler (Limnothlypis
swainsonii): A reconsideration. Archives of Nat-
ural History 23:287?290.
GREENBERG, R. 1984. The role of neophobia in the for-
aging site selection of a tropical migrant bird:
An experimental study. Proceedings of the Na-
tional Academy of Sciences USA 81:3778?3780.
GRINNELL, J. 1917. Field tests and theories concern-
ing distributional control. American Naturalist
51:115?128.
HAMEL, P. B. 1981. A hierarchical approach to avian
community structure. Ph.D. dissertation, Clem-
son University, Clemson, South Carolina.
HOLMES, R. T., AND S. K. ROBINSON. 1981. Tree spe-
cies preferences of foraging insectivorous birds
in a northern hardwood forest. Oecologia 48:31?
35.
HOSMER, D. W., JR., AND S. LEMESHOW. 1989. Applied
Logistic Regression. John Wiley and Sons, New
York.
HOWELL, A. H. 1924. Birds of Alabama. Bureau of Bi-
ological Survey, U.S. Department of Agriculture
and Department of Game and Fisheries, Brown
Printing Company, Montgomery, Alabama.
HOWELL, A. H. 1932. Florida Bird Life. Department
of Game and Fresh Water Fish, Coward-McCann
Inc., New York.
HUNTER, W. C., D. N. PASHLEY, AND R. E. F. ESCANO.
1993. Neotropical migratory landbird species
and their habitats of special concern with the
southeast region. Pages 159?171 in Status and
Management of Neotropical Migratory Birds (D.
M. Finch and P. W. Stangel, Eds.). U.S. Forest Ser-
vice General Technical Report RM-229, Fort Col-
lins, Colorado.
JAMES, F. C., C. A. HESS, AND D. KUFRIN. 1997. Spe-
cies-centered environmental analysis: Indirect
effects of ?re history on Red-cockaded Wood-
pecker. Ecological Applications 7:118?129.
JAMES, F. C., R. F. JOHNSTON, N. G. WAMER, G. J. NIE-
MI, AND W. J. BOECKLEN. 1984. The Grinnellian
niche of the Wood Thrush. American Naturalist
124:17?30.
JAMES, F. C., AND H. H. SHUGART, JR. 1970. A quan-
titative method of habitat description. Audubon
Field Notes 24:727?736.
KEARNEY, T. H. 1901. Report on a botanical survey of
the Dismal Swamp region. Contributions from
the U.S. National Herbarium 5:321?550.
KLUIJVER, H. N. 1951. The population ecology of the
Great Tit, Parus m. major L. Ardea 39:1?135.
LACK, D. 1933. Habitat selection in birds. Journal of
Animal Ecology 2:239?262.
LEVY, G. F. 1991. The vegetation of the Great Dismal
Swamp: A review and an overview. Virginia
Journal of Science 42:411?417.
LEVY, G. F., AND S. W. WALKER. 1979. Forest dynam-
ics in the Dismal Swamp of Virginia. Pages
101?126 in The Great Dismal Swamp (P. W.
Kirk, Jr., Ed.). University of Virginia Press,
Charlottesville.
LICHTLER, W. F., AND P. N. WALKER. 1979. Hydrology
of the Dismal Swamp, Virginia-North Carolina.
Pages 140?187 in The Great Dismal Swamp (P.
W. Kirk, Jr., Ed.). University of Virginia Press,
Charlottesville.
MACARTHUR, R. H. 1958. Population ecology of some
warblers of northeastern coniferous forests.
Ecology 39:599?619.
MEANLEY, B. 1945. Notes on Swainson's Warbler in
central Georgia. Auk 62:395?401.
MEANLEY, B. 1966. Some observations on habitats of
the Swainson's Warbler. Living Bird 5:151?165.
MEANLEY, B. 1969. Pre-nesting and nesting behavior
of the Swainson's Warbler. Wilson Bulletin 81:
246?257.
MEANLEY, B. 1970. Method of searching for food by
the Swainson's Warbler. Wilson Bulletin 82:228.
MEANLEY, B. 1971. Natural history of the Swainson's
Warbler. North America Fauna, no. 69.
MEANLEY, B. 1982. Swainson's Warbler and the cow-
bird in the Dismal Swamp. Raven 53:47?49.
July 2001] 663Swainson's Warbler in Great Dismal Swamp
MORSE, D. H. 1989. American Warblers: An Ecolog-
ical and Behavioral Perspective. Harvard Uni-
versity Press, Cambridge, Massachusetts.
MORTON, E. S. 1990. Habitat segregation by sex in the
Hooded Warbler: Experiments on proximate
causation and discussion of its evolution. Amer-
ican Naturalist 135:319?333.
MORTON, E. S., M. VAN DER VOORT, AND R. GREEN-
BERG. 1993. How a warbler chooses its habitat:
Field support for laboratory experiments. Ani-
mal Behaviour 46:47?53.
MUSSELMAN, L. J., D. L. NICKRENT, AND G. F. LEVY.
1977. A contribution towards a vascular ?ora of
the Great Dismal Swamp. Rhodora 79:240?268.
NIVEN, B. S., AND D. E. ABEL. 1991. Logical synthesis
of environment of King Penguin, Aptenodytes pa-
tagonicus. Ecological Modelling 56:291?311.
NIVEN, B. S., AND M. G. STEWART. 1987. Logical syn-
thesis of an animal's environment: Sponges to
non-human primates. II. The freshwater sponge,
Spongilla lacustris. Australian Journal of Zoology
35:607?624.
NOLAN, V., JR. 1978. The ecology and behavior of the
Prairie Warbler Dendroica discolor. Ornithological
Monographs, no. 26.
OAKS, R. Q., JR., AND N. K. COCH. 1973. Post-Miocene
Stratigraphy and Morphology, Southeastern Vir-
ginia. Virginia Division of Mineral Resources
Bulletin 82.
OAKS, R. Q., JR., AND D. R. WHITEHEAD. 1979. Geo-
logic setting and origin of the Dismal Swamp,
southeastern Virginia and northeastern North
Carolina. Pages 1?24 in The Great Dismal
Swamp (P. W. Kirk, Jr., Ed.). University of Vir-
ginia Press, Charlottesville.
ORIANS, G. H., AND J. F. WITTENBERGER. 1991. Spatial
and temporal scales in habitat selection. Amer-
ican Naturalist 137(Supplement):29?49.
PARRISH, J. D. 1995. Effects of needle architecture on
warbler habitat selection in a coastal spruce for-
est. Ecology 76:1813?1820.
PARTRIDGE, L. 1976. Field and laboratory observa-
tions on the foraging and feeding techniques of
Blue Tits (Parus caeruleus) and Coal Tits (Parus
ater) in relation to their habitats. Animal Behav-
iour 24:534?544.
PASHLEY, D. N., AND W. C. BARROW. 1993. Effects of
land use practices on Neotropical migratory
birds in bottomland hardwood forests. Pages
315?320 in Status and Management of Neotrop-
ical Migratory Birds (D. M. Finch and P. W. Stan-
gel, Eds.). U.S. Forest Service General Technical
Report RM-229, Fort Collins, Colorado.
PEARSON, T. G., C. S. BRIMLEY, AND H. H. BRIMLEY.
1919. Birds of North Carolina. Edwards and
Broughton Printing Company, Raleigh, North
Carolina.
PRICE, J., S. DROEGE, AND A. PRICE. 1995. The Sum-
mer Atlas of North American Birds. Academic
Press, London.
RAPPOLE, J. H. 1995. The Ecology of Migrant Birds.
Smithsonian Institution Press, Washington, D.C.
REBER, E. J., M. A. BAILEY, P. J. SWECKER, J. S. QUE-
SENBERRY, AND D. BRADSHAW. 1981. Soil Survey
of the City of Suffolk, Virginia. Soil Conserva-
tion Service, U.S. Department of Agriculture,
Suffolk, Virginia.
ROBBINS, M. B., AND D. A. EASTERLA. 1992. Birds of
Missouri: Their Distribution and Abundance.
University of Missouri Press, Columbia.
ROBINSON, S. K., AND R. T. HOLMES. 1984. Effects of
plant species and foliage structure on the for-
aging behavior of forest birds. Auk 101:672?684.
ROTENBERRY, J. T. 1985. The role of habitat in avian
community composition: Physiognomy or ?o-
ristics? Oecologia 67:213?217.
SIMS, E., AND W. R. DEGARMO. 1948. A study of
Swainson's Warbler in West Virginia. Redstart
16:1?8.
SMITH, C. R., D. M. PENCE, AND R. J. O'CONNOR.
1993. Status of Neotropical migratory birds in
the Northeast: A preliminary assessment. Pages
172?188 in Status and Management of Neotrop-
ical Migratory Birds (D. M. Finch and P. W. Stan-
gel, Eds.). U.S. Forest Service General Technical
Report RM-229, Fort Collins, Colorado.
SPSS INC. 1998. SYSTAT Version 8. Chicago.
STRONG, A. M. 2000. Divergent foraging strategies of
two neotropical migrant warblers: Implications
for winter habitat use. Auk 117:381?392.
TERBORGH, J. 1989. Where Have all the Birds Gone?
Princeton University Press, Princeton, New
Jersey.
THOMAS, B. G., E. P. WIGGERS, AND R. L. CLAWSON.
1996. Habitat selection and breeding status of
Swainson's Warbler in southern Missouri. Jour-
nal of Wildlife Management 60:611?616.
THOMPSON, F. R. III, S. J. LEWIS, J. GREEN, AND D. EW-
ERT. 1993. Status of neotropical migrant land-
birds in the Midwest: Identifying species of
management concern. Pages 145?158 in Status
and Management of Neotropical Migratory
Birds (D. M. Finch and P. W. Stangel, Eds.). U.S.
Forest Service General Technical Report RM-
229, Fort Collins, Colorado.
TWEDT, D. J., AND C. R. LOESCH. 1999. Forest area and
distribution in the Mississippi alluvial valley:
Implications for breeding bird conservation.
Journal of Biogeography 26:1215?1224.
U. S. WEATHER BUREAU. 1965. Climatic survey of the
U.S., 1951?1960, Virginia. Washington, D.C.
VAN HORNE, B. 1983. Density as a misleading indi-
cator of habitat quality. Journal of Wildlife Man-
agement 47:893?901.
VAN HORNE, B., AND J. A. WIENS. 1991. Forest bird
habitat suitability models and the development
664 [Auk, Vol. 118GARY R. GRAVES
of general habitat models. United States Fish
and Wildlife Research 8:1?31.
WAYNE, A. T. 1886. Nesting of Swainson's Warbler in
South Carolina. Ornithologist and Oologist 11:
187?188.
WHITEHEAD, D. R. 1972. Developmental and environ-
mental history of the Dismal Swamp. Ecological
Monographs 42:301?315.
WHITEHEAD, D. R., AND R. Q. OAKS. 1979. Develop-
mental history of the Dismal Swamp. Pages 25?
43 in The Great Dismal Swamp (P. W. Kirk, Jr.,
Ed.). University of Virginia Press, Charlottesville.
WIDMANN, O. 1895. Swainson's Warbler an inhabi-
tant of the swampy woods of southeastern Mis-
souri. Auk 12:112?117.
WIENS, J. A. 1989. The Ecology of Bird Communities.
Cambridge University Press, Cambridge,
Massachusetts.
Associate Editor: J. Brawn