EFFECTS OF EDGE AND INTERNAL PATCHINESS
ON HABITAT USE BY BIRDS IN A FLORIDA HARDWOOD FOREST
REED F. NOSS
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
I thank the members of my supervisory committee, Drs.
Ronald F. Labisky (Chairman), Michael W. Collopy, and John G.
Robinson in the Department of Wildlife and Range Sciences,
and Drs. Peter Feinsinger and Brian K. McNab in the
Department of Zoology, University of Florida, for guidance
and comments throughout this study. Drs. T.C. Edwards and
K.M. Portier and Mr. Tim O'Brien provided statistical advice,
and Dr. L.D. Harris provided helpful dialogue throughout the
course of this study. Ms. Candy Hollinger drew all figures.
Funding was provided by the Alachua Audubon Society, the
Frank M. Chapman Memorial Fund, the Florida Ornithological
Society, the Josselyn Van Tyne Memorial Fund, and the
Department of Wildlife and Range Sciences, University of
I owe much gratitude to my wife, Myra, and daughter,
April, for their steadfast patience during the countless
hours I spent in the field and behind the PC. This
dissertation is dedicated to my mother, Margaret Johnson Noss
(1923-1987), who always encouraged my interest in natural
history, and to my father, James Frederick Noss, who
continues to help and encourage me today.
TABLE OF CONTENTS
STUDY AREA AND METHODS......................................5
Study Area.............................................. 5
Edge Effects ..........................................26
Bird Densities in Edge versus Interior Plots.........43
Internal Patchiness and its Relation to Bird
Densities and Edge Effect..........................43
Patterns at the Between-Plot Scale...................56
Responses of Individual Species to
Bird Responses to Forest Edge and Internal
Patchiness: A Matter of Scale?....................64
A Final Comment on Scale and Observation.............80
I PROPORTIONAL RELATIVE ABUNDANCES OF TREE AND
SHRUB-LEVEL WOODY SPECIES.........................95
II BIRD SPECIES OBSERVED IN STUDY PLOTS AND THEIR
EDGES IN SAN FELASCO HAMMOCK, 1985-86 ............ 101
BIOGRAPHICAL SKETCH... ..................................108
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
EFFECTS OF EDGE AND INTERNAL PATCHINESS
ON HABITAT USE BY BIRDS IN A FLORIDA HARDWOOD FOREST
Reed F. Noss
Chairman: Dr. Ronald F. Labisky
Major Department: Wildlife and Range Sciences
(Forest Resources and Conservation)
Effects of edge and internal patchiness on habitat-use
by birds were studied in an old-growth hardwood forest in
northern Florida. Registrations of birds during the breeding
seasons of 1985 and 1986, fall 1985, and winter 1986 were
mapped in 12, 5.0-ha plots that abutted edge and in 3, 5.0-ha
interior plots >700 m from edge. Habitat-use patterns of 27
avian species were analyzed at within-plot (0.5-ha subplots)
and between-plot observation scales. Forest edges, canopy
gaps, and shrubby seepage areas ("bayheads") exhibited high
densities of birds within plots, but with seasonal and
species-specific variations. Density of birds at 0-50 m from
edge was significantly (P < .05) greater than in distance
zones farther from edge in 6 of 12 edge plots, for all
seasons combined. All 27 species used both interior and edge
habitat, but 12 species were generally attracted to edge, 11
were indifferent, and 4 avoided edge. Edge effects were
greatest during the breeding seasons in plots with high-
contrast edges, and in east-facing edges on sunny winter
mornings (suggesting a microclimatic influence). Attraction
to edge in the breeding seasons was stronger in subplots
lacking gaps and bayheads than in more heterogeneous
subplots. In interior plots and in all plots combined,
regression models containing indices of patchiness
(especially gap/bayhead area) explained 5-71% of the
variation in bird density, depending on season; strongest
relationships occurred during the breeding seasons.
In between-plot analyses, neither bird density nor
richness differed between edge plots and interior plots, or
between patchy plots and more homogeneous plots (except that
richness was correlated with CV of shrub density, one measure
of patchiness). Birds used gaps, bayheads, and edges
extensively within plots, but did not concentrate in areas of
the forest near edge or with more gap/bayhead area. Hence,
both edge effects and "patch effects" may be dependent on
observation scale. Allowing forests to mature to naturally
patchy old-growth may be a more prudent management strategy
than maintaining artificial openings or edges.
No natural community is homogeneous. Yet, a tradition
in community ecology has been to select the most uniform and
undisturbed areas as study sites (Forman and Godron 1981,
Noss 1987a). Although recognition of natural disturbances
and spatiotemporal mosaics is not new (e.g., Cooper 1913,
Watt 1925, 1947), selection of heterogeneous systems as units
of study only recently has become popular in North American
ecology (Forman and Godron 1981, 1986, Risser et al. 1984,
Pickett and White 1985, Urban et al. 1987). The increasing
acceptance of landscape ecology follows the recognition that
many organisms depend on environments that are frequently
disturbed and composed of multiple habitats (White 1979, den
Boer 1981, Karr and Freemark 1983, May 1986, Noss 1987a,
Habitat heterogeneity occurs at many spatial scales, as
an expression of environmental gradients and resource
patchiness, and disturbances that are patchy in time and
space. Not only does landscape pattern affect population
distribution, but the scale at which patterns are sought will
determine what patterns are detected (Wiens 1985, 1986, Wiens
et al. 1987). Coarse-grained landscape heterogeneity
(macroheterogeneity; Forman and Godron 1986) is generated by
large natural disturbances as well as by anthropogenic
habitat modification and fragmentation. Patches in
macroheterogeneous landscapes, e.g., interspersed woodlots,
crop fields, pastures, ponds, and housing developments, are
easily recognized as separate habitats. Boundaries between
habitats at this scale are usually called "edges." In
contrast to the mosaic of habitats at a landscape scale,
fine-grained patchiness (microheterogeneity; Forman and
Godron 1986) occurs within what are usually recognized as
distinct habitats or community-types.
Habitat selection, best studied for birds (Cody 1985),
involves responses to habitat cues at several spatial scales
(Hilden 1965, James 1971, Hutto 1985, Wiens 1985). Elements
of habitat heterogeneity at different scales determine the
distribution of bird populations, territories, and activity
within territories. Over a range of habitats, bird species
diversity is positively correlated with complexity of the
vertical foliage profile (MacArthur and MacArthur 1961,
Recher 1969, Karr and Roth 1971). Horizontal complexity or
patchiness of vegetation, however, may be more important than
vertical profiles in determining bird species composition and
diversity within study sites (MacArthur et al. 1962, Roth
1976, 1977, Wiens 1985, Freemark and Merriam 1986). An
important element of within-habitat patchiness is treefall
gaps. Gaps often contain high densities of resources such as
insects and fruits, and therefore attract an abundance of
birds (Blake and Hoppes 1986, Martin and Karr 1986).
Patchiness at a coarser scale may affect bird
distribution within habitats, i.e., the well-known "edge
effect." Many bird species are attracted to edges, but a few
species are repelled and some are indifferent (Whitcomb et
al. 1981). Although maximizing edge habitat has been a
common wildlife management prescription (Thomas et al. 1979),
this policy has been criticized in recent conservation
biological literature because of documented deleterious
effects on forest interior species (Whitcomb et al. 1976,
1981, Gates and Gysel 1978, Noss 1981, 1983, Brittingham and
Temple 1983, Lovejoy et al. 1986, Wilcove et al. 1986).
Unfortunately, studies of edge effects on birds generally
have ignored habitat patchiness within forests. Gates and
Gysel (1978) hypothesized that edges attract many passerines
because they contain structural cues similar to those of the
mixed life-form habitats in which these species evolved.
Before European settlement, the eastern deciduous forest
landscape was old-growth (best understood as a mosaic of
developmental stages, rather than just the mature stage). In
contrast to the generally even-aged, close-canopied,
secondary forests in which edge effects usually have been
studied, old-growth is horizontally patchy with numerous gaps
in various stages of post-disturbance regeneration (Bormann
and Likens 1979, Runkle 1985, Whitney 1987).
Does edge effect diminish in forests with increased
internal patchiness such as that associated with treefall
gaps? Or conversely, do artificial edges disrupt the normal
distribution of bird activity with respect to internal
patchiness? This study simultaneously considered effects of
edge and internal patchiness on habitat use by birds in an
old-growth hardwood forest in north central Florida.
Specific objectives were (1) to test the null hypothesis of
uniform distribution of birds with distance from edge for
different species, seasons, and edge types; (2) to
investigate the relation between bird density and habitat
heterogeneity (as measured by several indices) in 5.0-ha
plots that abutted edge, and in 5.0-ha "control" plots
located at distances > 700 m from edge; (3) to identify bird-
habitat heterogeneity relationships at within-plot and
between-plot observation scales; and (4) to determine
interacting effects of internal habitat heterogeneity and
edge on bird habitat-use.
STUDY AREA AND METHODS
San Felasco Hammock State Preserve (29045' N, 82030' W)
is located 8 km northwest of Gainesville, Alachua County,
Florida. Over half of the 2500-ha preserve is mesic hammock
(a mixed-species, predominantly hardwood forest), considered
the climax community of northern Florida by many authors
(Harper 1905, Quarterman and Keever 1962, Monk 1965). In
Florida, fire, interacting with slope-moisture gradients, is
a primary determinant of plant distribution; hammocks are
naturally restricted to ravines and sinkhole slopes, islands,
peninsulas, and other sites of reduced fire frequency (Harper
1911, Clewell 1981, Platt and Schwartz in press). The high
species richness (probably the largest stand-scale number of
woody species in the continental United States) and
structural heterogeneity of north Florida hammocks derive
from a complex disturbance regime which creates numerous gaps
and maintains nonequilibrium populations of plants (Platt and
Schwartz in press).
The study site was chosen for its size (with gradients
from edge to deep interior habitat) and old-growth character.
San Felasco Hammock is probably the largest remaining mesic
hammock in the hammock belt that extends along the central
ridge of the Florida peninsula (Ewel and Simons 1976, Platt
and Schwartz in press). The vegetation and flora of San
Felasco Hammock were described by Ansley (1952) and Dunn
(1982), and summarized by Skeate (1987). The woody flora of
this mesic hammock is extremely rich, with as many as 20 tree
species per ha (Noss, unpublished data) and 50 "canopy
potential" tree species (Dunn 1982). Dominant canopy trees
include Ouercus hemisphaerica, Carya glabra, Magnolia
grandiflora, Quercus nigra, Quercus virginiana, Pinus glabra,
Liquidambar styraciflua, Quercus michauxii, Quercus austrina,
Quercus shumardii, Quercus falcata, Acer barbatum, Fraxinus
americana, Carva tomentosa, and Tilia caroliniana. San
Felasco is near the southern range limit for many plant
species (R. Simons, personal communication), and for several
forest-interior bird species that are known elsewhere to be
sensitive to forest fragmentation and edge effects (cf.
Whitcomb et al. 1981).
The portion of San Felasco Hammock in which study plots
were selected was mature mesic hammock, as mapped by Dunn
(1982). This forest has been selectively logged (Dunn 1982),
but is essentially old-growth, with an uneven age composition
of trees, a pit-and-mound microtopography, and a horizontal
mosaic pattern enriched by treefall gaps, sinkholes, seepage
areas with dense broad-leaved evergreen trees and shrubs
(bayheads), and other patches. The mesic hammock in San
Felasco abruptly borders several other habitat types. Some
of these sharp edges are natural (e.g., with adjoining
sinkhole marshes and pinelands), whereas others are
artificial (e.g., with oldfields and cleared powerline
Five major edge types were identified: hammock/
oldfield; hammock/open pine plantation; hammock/longleaf
pine-turkey oak (sandhill); hammock/herbaceous marsh; and
hammock/open powerline right-of-way. These edge types were
the strata from which 12, 5.0-ha (200 m x 250 m) plots were
selected randomly, with the restriction that the central
transects of plots be at least 300 m apart to avoid double-
counting of birds. Two plots were selected to represent each
edge type, except that 4 plots (2 with southwest-facing edges
and 2 with northeast-facing edges) were selected from the
powerline right-of-way. Hence, each of 12, 5.0-ha plots
abutted an open habitat along a 200-m edge, and extended 250
m into the hammock (Fig. 1, Table 1). In addition, 3
"control" plots of 5.0 ha each were selected randomly in
interior forest >700 m from any abrupt edge (Table 1). Each
5.0-ha plot was divided into 10, 0.5-ha subplots for data
compilation and analysis (Fig. 1). A central transect line
for bird surveys extended 250 m through each plot, with
observation posts flagged every 25 m. Prominent habitat
features such as major gaps and sinks, streams, trails, and
co o CO
tO I Q)-
,o Cn ,
O 4 r-4
C Q) C
t3 U *
. U~- 4Ji-
TABLE 1. Individual 5.0-ha study plots and their transect
orientations, San Felasco Hammock, Florida. All edge
plots are perpendicular to edge,whereas interior plots
are perpendicular to narrow trails.
Edge Plots (N = 12)
PL1 (Powerline 1): 300
PL2 (Powerline 2): 2200
PL3 (Powerline 3): 2200
PL4 (Powerline 4): 400
MA1 (Marsh 1): 700
MA2 (Marsh 2): 2800
SA1 (Sandhill 1): 1100
SA2 (Sandhill 2): 3400
PP1 (Pine Plantation 1): 2700
PP2 (Pine Plantation 2): 2700
OF1 (Oldfield 1): 2700
OF2 (Oldfield 2): 2700
Interior Plots (N = 3)
IF1 (Interior Forest 1): 2700
IF2 (Interior Forest 2): 3200
IF3 (Interior Forest 3): 1800
big trees were mapped prior to the study to facilitate
accurate location of birds.
Birds were surveyed by use of a strip-map method (Emlen
1984; similar to the "plot mapping" of Christman 1984), which
is essentially a combination of transect and spot-mapping
techniques. The method is observer-specific, with plot width
dependent on the detection abilities of an individual
observer. The 12 edge plots were sampled in spring 1985
(S85: 18 censuses each from 5 March-18 June), fall 1985 (F85:
6 censuses each from 12 September-30 November), winter 1986
(W86: 6 censuses each from 11 January-27 February) and spring
1986 (S86: 12 censuses each from 2 March-19 May). The 3
interior (control) plots were censused in W86 and S86; these
censuses were conducted during the same calendar period and
at the same frequency as for edge plots. Demarcation of
seasons was necessarily arbitrary. Breeding seasons
(indicated by singing and other territorial behavior) of
permanent resident passerines and woodpeckers begin in this
region in January and February, at the same time when
wintering species reach peak abundance. Summer residents
j begin breeding in March and April, coinciding with peak
densities of migrants, often of the same species.
Daily censuses were confined to the period between
sunrise and 3 h post-sunrise. Three or 4 plots were censused
per census day, with plot order alternated to assure
equivalent temporal coverage. Censuses were not conducted if
weather conditions were adverse (i.e., rain or strong winds).
Auditory and visual observations of birds were mapped
during each census. A registration was defined as any
auditory/visual observation within the study plot, excluding
birds flying above the canopy. Density refers to the sum of
registrations recorded in a defined area (e.g., a subplot).
Birds using the edge (or plot boundary, for interior plots)
were mapped as each transect starting point was approached;
thereafter, registrations were recorded during 3-min periods
at each 25-m interval observation post. Registrations also
were recorded while walking slowly between observation posts.
Bird movements were recorded as separate registrations if
they were >25 m from the previous registration; however,
movements that appeared to be in response to the observer
(either attraction or avoidance) were not recorded.
Preliminary censuses indicated that most resident
(breeding, winter, and permanent) species could be detected
by ear up to 100 m from the central transect line; thus,
registrations were mapped within 100 m on each side of the
transect. The analysis assumed equivalent detectability in
all subplots of each plot, in accordance with equivalent
sampling effort. Because comparison of abundances among
species was not an objective, the analysis did not assume
that species were equally detectable, or that detectability
was constant from 0-100 m; thus, no coefficients of
detectability (Emlen 1971) were calculated. Resident species
with poor detectability (e.g., Blue-gray Gnatcatcher) were
excluded from analysis. Also excluded from analysis were
those species with sample sizes <30 in the 12 edge plots
(single-species analyses were performed for combined edge
plots and for all 15 plots combined), and one abundant winter
resident (Yellow-rumped Warbler) that usually occurred in
sporadic, itinerant flocks. Species with large territories
or ranges (raptors, jays, and crows) were not mapped or
analyzed. Breeding status was based on Florida Breeding Bird
Atlas criteria (Noss et al. 1985). Species richness of
probable and confirmed breeders with at least 25% of a
territory in a plot was recorded for each plot in 1985 and
The objective of habitat analysis was to measure
parameters of forest microheterogeneity proposed in previous
studies to be important to birds, including diversity of
vertical profiles, spatial variability in tree and shrub
dispersion, variation in shrub density and canopy openness,
floristic (tree and shrub) diversity, and proportion of plot
occupied by canopy gaps and bayheads (Table 2). These
attributes of heterogeneity were measured in summer and early
fall 1986 at 2 scales: 0.5-ha subplot (for within-plot
TABLE 2. Habitat variables measured for each 0.5-ha subplot
(from 10 point-quarter samples/subplot) and/or each 5.0-ha
plot, San Felasco Hammock, Florida. See text for further
Plot (P) or
Variable Measurement definition Subplot (SP)
median distance (m) from edge
for each subplot
GAB GA + GB
coefficient of variation (CV) of
distances to nearest trees in
CV of distances to nearest "shrubs"
in point-quarter samples
CV of distances to nearest trees or
shrubs in point-quarter samples
mean number (N) of shrub stems in
quarter-circles within 2 m of each
CV of shrub density (DENS)
mean canopy openness (%), or converse
of canopy density, as measured by
spherical densiometer at each
CV of canopy openness (OPEN)
Shannon diversity (H' logl0) of tree
species from point-quarter samples
H' of shrub species from point-
H' of vertical foliage profiles at
each sample point
proportion of area in canopy gaps
with shrub-sapling growth generally
< 2 m
proportion of area in canopy gaps
with shrub-sapling growth >2 5 m
SP and P
SP and P
SP and P
SP and P
SP and P
SP and P
SP and P
SP and P
SP and P
SP and P
SP and P
SP and P
SP and P
Plot (P) or
Variable Measurement definition Subplot (SP)
proportion of area in bayhead
GAB + BAY (sum)
largest single canopy gap in
number of tree species (S) in plot,
from point-quarter samples
number of "shrub" species (S) in
plot, from point-quarter samples
SP and P
S and P
analysis) and 5.0-ha plot (for between-plot analysis). A
grid of 100 uniformly-distributed sampling points was
superimposed on each plot, with 10 points per subplot.
Systematic sampling is preferable to random sampling when the
objective is to distinguish pattern or variability in the
vegetation (Greig-Smith 1964, Gauch 1982).
Trees and shrubs were sampled by use of the point-
centered quarter technique (Cottam and Curtis 1956). The
nearest tree and shrub in each of 4 quarter-circles around
each sampling point was recorded by species and distance from
the point. Trees (woody stems >5 m in height) were separated
into dbh categories (<10, >10-30, >30-50, >50-70, and >70
cm). Shrubs were defined as woody stems 0.3-5m in height.
The number of shrub stems was recorded in each quarter circle
within a 2-m radius from each sampling point for determina-
tion of mean shrub density. A coefficient of variation (CV)
was calculated for distances to nearest trees, shrubs, and
trees and shrubs combined in each subplot (Roth 1976), and
for shrub density. A mean CV for 10 subplots was equivalent
to the CV for a plot. The Shannon diversity index (H',
logl0) was calculated for trees and shrubs for each subplot
and separately for each plot. Presence or absence (+ or -)
of vegetation within a 0.5-m diameter circle centered on each
sampling point was recorded for 4 layers: herb (<0.3 m),
shrub (>0.3-5 m), understory (>5-10 m) and overstory (>10 m),
and H' was calculated for profile types (e.g., +++-, ++-+) in
each subplot and plot. Mean canopy openness (%, as
determined by a spherical densiometer over each sampling
point) and CV of canopy openness were determined for each
plot and subplot.
Canopy gaps and bayheads were mapped on each plot and
converted to proportions of area by use of a dot grid. A gap
was defined as a vertical hole in the canopy > 10 m in mean
diameter (a minimum area of 78.5 m2), and included canopy
openings caused by sinkholes as well as windthrow (treefall)
gaps. The proportionate gap area in each subplot and plot
was calculated in 3 categories: GA (gaps with shrub-sapling
growth averaging <2m in height), GB (gaps with shrub-sapling
growth >2-5 m in height) and GAB (GA + GB). The
proportionate area occupied by bayheads was calculated
separately (BAY), and added to gap area for a final patch
Local (edge-specific) sunshine data were collected
during censuses of east-facing edges in winter 1986 to
determine whether edges warmed by early-morning sun had
greater densities of birds. Edges were considered sunny if
sunlight was directly striking the edge face at the time of
census, and not sunny if cloudy, foggy, or before sunrise
illuminated the edge.
Data were analyzed with SYSTAT programs (Wilkinson 1987)
on an IBM PC-XT. Major analyses included least squares
regression modelling (simple and multiple), analysis of
covariance (ANCOVA), tests for homogeneity and goodness of
fit, and tests for difference between means. Correlation of
bird densities from subplots between seasons was a measure of
consistency of site use and thus indicated which seasonal
combinations were homogeneous. Bird densities had normal
distributions in all seasons, as determined from normal
probability plots. Some habitat variables (DIST, HT, HS,
HPD, ST, SS) were distributed normally without
transformation; the other variables were normalized and their
variances stabilized by logarithmic (In) transformation.
Zero values for GA, GB, GAB, BAY, and GAPBAY were converted
to -10 on the log scale (smallest log-transformed non-zero
values were > -5). Zero values for bird registrations in
subplots were converted to .001 for single-species regression
Edge effects in the 12 edge plots were determined by
analysis of registration densities in 5 zones 100 m wide
parallel to edge and 50 m perpendicular to edge, each
comprising 2 subplots (e.g., 01L and 01R in Fig. 1). A G-
test for goodness of fit (Sokal and Rohlf 1981) was used to
test the extrinsic null hypothesis that bird registrations
were distributed uniformly in the 5 distance zones, Ho:
Pi = P2 = P3 = P4 = P5- Overall G-tests for plots (all
species, all seasons), seasons (all species, all plots), and
species (all plots, all seasons) were partitioned into
separate G-tests to determine which individual distance zones
were significantly high or low in density relative to the
mean (expected frequency). G-tests also were performed to
test for uniformity of density in distance zones in the 3
Regression analysis was used to model response of bird
density to distance from edge (DIST) and to habitat
heterogeneity variables. Dependent variables were
registration densities of individual bird species and all-
species sums, totaled for each of the 4 seasons, for all
seasons combined, and for appropriate combinations of
seasons. Independent variables were distance from edge and
habitat heterogeneity variables (Table 2).
ANCOVA was used to test regression models for
differences in bird density among groups, e.g., between east
and non-east facing edges, between high- and moderate-
contrast edges, and between subplots with and without
gaps/bayheads, in response to the independent variable
(covariate) DIST. A preliminary ANCOVA model was first
applied to test for homogeneity of slopes among groups. If
slopes were homogeneous, ANCOVA was used to test for
homogeneity of Y-intercepts among groups (equivalent to a
test for homogeneity among group means, Sokal and Rohlf
1981). The null hypothesis of equal density of birds in edge
and interior plots was tested by comparing the mean number of
registrations per subplot in each group in winter and spring
1986, when all 15 plots were censused.
Simple linear regressions were performed for all-
species registrations in each season and in all seasons
combined for each habitat variable. Multiple regression
models were constructed using all possible subsets
(combinations) of variables. Criteria for optimal (i.e.,
best prediction) regression models were (1) the highest r2
value; (2) the fewest number of independent variables, each
with non-significant correlations (P > .05) with all others;
(3) each variable with a significant (P < .05) t-statistic
for regression; (4) homogeneous slopes among groups (e.g.,
individual plots), as indicated by P > .05 for group by
covariate interaction; (5) normal distribution of residuals;
and (6) homogeneity of variance of residuals across different
levels of independent variables, as indicated by a
homoscedastic plot of residuals against estimates (Sokal and
Rohlf 1981, Gutzwiller and Anderson 1987, Wilkinson 1987,
K.M. Portier, personal communication). Analysis of residuals
included correlating residuals with each independent
(habitat) variable not in the optimal model. Finally,
correlations with habitat variables were determined
individually for the 12 most abundant species observed during
winter-spring 1986 in all 15 plots.
Canopy gaps caused by treefall and sinkholes were a
prominent feature of most plots (Table 3). A mean of 3.4% of
plot area was occupied by canopy gaps (> 78.5 m2). Sinkhole
gaps were prominent in plots PP1, OF1 (an outlier in gap
area, with one canopy opening of 3208.02 m2), and IF1.
Treefall gaps averaged 3.1% of plot area; the largest single
treefall gap (in SA1) was 864.66 m2. Bayhead vegetation
occurred in only 2 plots, PP2 and IF1, where it occupied
18.3% and 9.8% of plot area, respectively. Ninety-one (61%)
of the 150 total subplots had gaps and/or bayheads, which
covered 5.3% of the total plot area.
None of the habitat heterogeneity variables (Table 3)
was related significantly to distance from edge. The
strongest association was a positive correlation between tree
species diversity (HT) and distance from edge (r = .163, P =
.075). Plots were rich in tree species, ranging from 17 to
27 (x = 22.33) and in "shrubs" (which included tree saplings
in the shrub layer), ranging from 24 to 39 species (x =
29.73) (Appendix I).
TABLE 3. Summary of habitat measurements from 15, 5.0-ha
plots (see Table 2 for definition of variables), San
Felasco Hammock, Florida, 1986. HT, HS, HPD, LG, ST, and
SS are measured at the 5.0-ha scale; other values are means
from the 10, 0.5-ha subplots in each plot.
PL1 55.8 68.9 87.1 6.8 62.9 5.6
PL2 56.8 75.2 96.3 13.8 71.6 7.9
PL3 58.1 75.2 95.9 11.2 67.9 9.2
PL4 57.8 74.0 88.5 8.7 103.9 8.6
MA1 59.0 76.3 91.1 8.7 92.7 6.3
MA2 58.0 80.6 90.4 7.8 86.9 6.1
SA1 58.2 71.5 93.5 9.4 80.4 5.6
SA2 55.6 90.4 88.2 9.0 91.2 6.2
PP1 58.3 83.1 101.8 12.0 60.5 5.9
PP2 56.8 67.3 100.3 12.6 61.7 5.3
OF1 68.5 87.9 109.0 10.2 70.2 12.4
OF2 56.0 77.4 93.4 10.9 68.1 4.9
IF1 59.9 74.0 99.0 15.2 71.3 5.3
IF2 56.1 61.0 87.4 7.8 83.5 5.9
IF3 57.3 80.9 89.3 9.1 93.3 5.6
x 58.1 76.2 94.1 10.2 77.7 6.7
SD 3.1 7.7 6.3 2.4 13.5 2.0
Plot VCO HT HS HPD GA GB
PL1 57.2 1.15 1.15 .66 .010 .003
PL2 37.2 0.97 0.93 .61 .005 .030
PL3 44.2 0.91 0.87 .52 0.0 .060
PL4 37.8 0.99 1.17 .39 .008 .032
MA1 57.7 1.08 1.17 .52 .004 .048
MA2 53.8 1.07 1.15 .69 .005 .048
SAl 48.2 1.02 1.16 .54 .005 .037
SA2 61.9 1.04 0.98 .59 .010 .008
PP1 51.5 1.02 0.88 .45 .022 .009
PP2 43.0 1.00 1.04 .47 .005 .002
OF1 67.3 1.10 1.14 .62 .068 .010
OF2 35.8 1.05 1.08 .50 0.0 .003
IF1 48.5 1.20 1.21 .48 .020 .004
IF2 50.4 1.00 1.17 .56 .010 .030
IF3 40.1 1.10 1.35 .60 0.0 .008
x 49.0 1.05 1.10 .55 .011 .022
SD 9.5 .07 .13 .08 .174 .020
Plot GAB BAY GAPBAY LG (m2) ST SS
PL1 .013 0.0 .013 500.00 25 33
PL2 .035 0.0 .035 319.55 19 26
PL3 .060 0.0 .060 620.30 17 24
PL4 .040 0.0 .040 839.60 20 30
MA1 .052 0.0 .052 651.63 25 26
MA2 .053 0.0 .053 363.41 27 31
SA1 .042 0.0 .042 864.66 23 31
SA2 .018 0.0 .018 200.50 26 26
PP1 .031 0.0 .031 914.79 22 31
PP2 .007 0.183 .190 125.31 22 28
OF1 .078 0.0 .078 3208.02 22 32
OF2 .003 0.0 .003 150.00 22 30
IF1 .024 0.098 .122 870.93 27 39
IF2 .040 0.0 .040 651.63 17 27
x .034 .019 .053 701.65 22.33 29.73
SD .022 .052 .048 746.95 3.24 3.75
One hundred twenty-nine bird species were observed on
the San Felasco Hammock study plots in 1985 and 1986
(Appendix II). Quantitative analyses were conducted on 27
species; 102 species were excluded because of insufficient
sample sizes (n < 30 for 12 edge plots over all seasons) or
sampling biases. Breeding species richness (S) was
significantly higher in 1985 than in 1986 (Table 4; t = 3.89,
P = .003). In 1986, when all plots were censused, Sidid not
differ between edge and interior plots (t = 0.681, P = .508).
For all plots combined, S was significantly related to only
one habitat variable, variation in shrub density (VSD; r =
0.517, P = .048).
Correlations of bird densities in subplots between
seasons indicated that site use was not always consistent.
For the 12 edge plots, site use in the 1985 and 1986 breeding
seasons (S85 and S86) was consistent (r = 0.56, P < .001).
Site use for edge plots in fall 1985 (F85) was not consistent
with that in any other season (P > .05). Site use for edge
plots in winter 1986 (W86) was consistent with S86 (r =
0.272, P = .003), but not with S85 (r = 0.126, P = .17).
Site use in W86 and S86 was consistent for the 3 interior
plots (r = 0.51, P = .004) and for all 15 plots combined (r =
0.30, P < .001).
TABLE 4. Breeding bird species richness in 15, 5.0-ha
plots, San Felasco Hammock, Florida, 1985-86. A species
was included only if > 25% of the territory of one
breeding pair was within plot.
Number of species
Plot 1985 1986 Mean
a Not censused in 1985
G-tests of the null hypothesis that bird density was
uniform with respect to distance from edge, in the 12 edge
plots, yielded diverse results (Table 5). For all seasons
combined, 6 of the 12 edge plots had a significant
concentration of bird registrations within 50 m from edge; 4
of these 6 plots also had reduced density in the 2 zones
farthest from edge (150-200 m and 200-250 m). Of the 6 plots
that did not show a positive edge effect, 3 exhibited a low
density at 200-250 m, 1 a low density at 50-100 m, 1 a high
density at 150-200 m, and 1 a low density at 0-50 m and a
high density at 100-150 m.
For all 12 edge plots combined, a positive edge effect
was evident for all seasons combined, for the 1985 and 1986
breeding seasons, and for winter 1986--but not for fall 1985
(Table 5). Although the density of registrations in the 0-50
m zone for fall 1985 was not significantly high, densities in
the 2 zones farthest from edge were low. In all other
seasons, densities in the 0-50 m zone were significantly
high, indicating an edge effect, and those in the 200-250 m
zone were low.
Bird densities in none of the 3 interior plots were
higher in the 0-50 m zone; hence, no "pseudo-edge effect" was
evident (Table 6). Although 1 interior plot had low density
in the 200-250 m zone, the other 2 had high densities in the
central zones (significant in one case). Densities in all 3
*H 0 0
4 00 I
w0 w 00
V 4) 0)a 0n
En r -H (1
4) 0 k4 4)
4) N O
U 0) ( *
'oin o Q
UIn i i
)0 0 C
S-H ) 0
44 ,9 0 (0
4 0- Q)
*.4 0 in
3 0 *4J
v -1q 0 4)
01 in rr-4
-1 c l -
* P (0 0) *
k nI H +rd
M 0 4.) 0
-H t71 (1)
>4 z z >4 >4 Z Z >4 Z >4
rHl (N |
al (O ril N N
H N m H NHC -l M ( N0
a4l 134 l04 D4l Mf 0f a4i rt4 0 0
NI LO M o NM (n V 0 nl
col o m n m k < 0 0 q|
H r- i- Hil H t-4 i-I r-I Hi i-4 r-l
>4 >4 >4
co o m -:r
r-i| V-i| VIr 'V
r- co co i
N Io co o i
H- M N !
> Q) 0
-4 ts 0)
4-1) f 4-4
*ro 0 41
u 0 0
II r- 4
0 z 9
Or >C I
1I t c
I C 0)
0 C C I
I > C
H p H L
W 0 0
4-'> 4.J S
.H 4-w 0
(0 :1 :
*H U r
4f- 4-4 *H(40-
0 *'-c I
41 .0 C
0C XI O
H > C
Q) W 9
) C (0
v> u i
pj C 0
M m 11 0
in o o co0
co r% '-I
co 0 LA m
ID0 in %0
a ua v e
interior plots combined were highest in the 50-100 m and 100-
150 m zones.
Individual species responded differently to edge but
were classified into 3 general categories (Table 7). Twelve
species were edge-attracted (i.e., they showed a significant
concentration in the first distance zone), 11 species were
edge-indifferent (i.e., they had neither higher nor lower
than expected densities near edge), and 4 species were edge-
avoiding (i.e., they showed lower than expected densities
near edge). All 4 edge-avoiding species and 10 of the 12
edge-attracted species (exceptions were the Ruby-crowned
Kinglet and Gray Catbird) bred on the study plots. In
contrast, 7 of the 11 indifferent species were wintering
The directional exposure of edge plots influenced the
distribution of birds with respect to edge, but only in
winter (Fig. 2). The relationship between bird density and
distance from edge for east-facing and non-east facing edge
plots, in the breeding seasons, did not differ in slope (Fig.
2, A and D; F = 0.242, P = .623), or in intercept (ANCOVA, F
= 1.275, P = .261). Regressions for fall (Fig. 2, B and E)
also did not differ in slope (F = 0.919, P = .340) or in
intercept (ANCOVA, F = 0.003, P = .958). In winter, however,
the slope for east-facing edges was significantly steeper
than for non-east-facing edges (Fig. 2, C and F; F = 6.723, P
= .011). Total winter registrations in subplots averaged
4) t 4.
'44V'44 0 >
ao I w u
0 r 0
4t-4 rO 44 0 3
U -4 UW 0
(n 4) u ) >
v -P kQ) 4-)
0 0) -o
o o 4 -4
00 0 tr
4) ail > .iq
a) CO w m
b) I CP r.
sid V co V -'-
.Q 0( k-i
CO O 4
3 (Ua -v
-H W3 0 $4 3
tH O >
0 -r4 (a
C (* (U &
o o NA
0 ( 0 w)V
0 r-14) u
W 4J 4J U)
0 0 CO 4-) -1
cr 11 e
0 001 0
r o 0
N .4 CM
NI .' (
S I v %D co lrI
Oil Ni N CN HI H i
Pc r-P 4
A ( 0
(0 0 C 0) 4
3: 4- TI a aM
Q H 4 0 4) V- 4-
PI V ". 0
&40 $i 0 0 04 > a)
W 4 0 4 4- 4 I U
0 ( 2 M44 U 4 V v C C
S(U E- c > r o 0
5 l C G 0 HO 4)
a O ( -4 0 aI 0o 00 o S
I Z A H Z Z IH X
M 4. 0 +. 4.) 4. IV >ib' 0
u '4-4 5 -4H VO AC 0
Q 0 10 0 0 3 00 3-H 3
M Z u E Z s 1s MM C
N v (1S
0| co o col r| ml col
SCl r-i HI
IO0 %D q M N n LO
'0D W.0 H 0 cO r-1l c
IO Cl l m r-Ai
N C CO In I ) H- co L
i m w inL m mn 04 M
m cO c o0 0
V) a a a aC
M, 4 41
UI HO 5
0 to 0 0 0 0 r-
f H 0 H- 0 0 0 0
A A A V V V V C
NN HM 1 HL o0
0 1 '4-i
-H 0 N U 0 r 01 CO
44 cN H H N N n w rq>
) I N H- Q 0
0 0 Om
C A o N N 01 ul LO 0t
o H H H H4 m LA LO HI
N I IN H
> 0 %0>
C O -4
-o t H N > %
H 4 H O-
4a u $4 in C4
0) 0IA H1 HA LA
v0 00 C
o q > a N H p 0 iu
0 4 0 H ) A 0 H P4
r 4 4 (. 1 > P r- o Ir_
00 C C.4 -
4 0 .0 to
40 H 0) H -,I
U 4 4 U w I v v v
( ) I 0 co 0 54 4.
4 U H0 0 0 0 r-l
o H 0 M I W r- 3 I >
4 0 H *l 1 0 0) M I ) r>l -HH
C 0 5- 1 ia >1 V> 0) E- CO t zr 4-4
H U 4.)I 4.i a F Ia V' U
I C.4 ( C I 0 5O CD V 0 + 0
0) i H4 c l a) 0 U 0 -rH
E -4 jC 0) CI ) I ) 0 c4 In
rI en 2 M 3 CQ s: M s s ui
* N *
N NqC (0N **
0 0 0 0 0
0 CO ( C\j
suo!JDj si6e 9
o* c C c p
0* m *
0U 0 C
.^ CO .v-
E C 0w 41
0 CO u
P 4-J V 0W)
4-4 0 -
Co bo 4-1 C
41 1U 0 4-I
M U *rl
O r0-4 u
.C *r4 M4
4W U Co
*r a Q) C
C Co *r4
0 (0 u Z
Co CO *
CL o I -I
aU C C m
P 0 0 4-4
p-4 a) .- 0C
*,4 CO 9Q 4-
4.4 C Q.
0 *-r H
c W 41 ba
0 4) C V
*r- r-4 )
41 0 0 ..CO
V V .
0 0 WC
C14 (U- a)
C C C
* N 0 1 m
-- : o 0 rJl r-
o .1 0
0 o C\
S* -N r E
P i'-- 1
ro e *
I I I 2I I- 0
O O O O O G
r--l ON c
C> 34 0 *
03 0 O c>
S / 0 .\
0) 0 c
qq- I- 4-
Fr) CU 0
O o o oJCl 0
0 0 0
II ro E
00 %J c
S suo0 o qe
C 0 03
c o 1 I
0 ) 0 0 0 l0 C0
I 1 -0
~O ~h. M (
(/) cuj o
o o r
ID o o to N
i o aN 0II
I- e L*
SUo IS 8
10.15 (SD = 6.61) in east-facing plots and 5.15 (SD = 4.21)
in non-east facing plots. The difference in means was
significant (t = 4.942, P < .001), indicating that birds used
east-facing edge plots more than they used non-east facing
edge plots in winter.
Sunlight (and presumably a warmer microclimate for birds
and their insect prey) apparently attracted birds to east-
facing edges in winter. Of 36 "edge-mornings" (each of the 6
east-facing edge plots was censused 6 times in winter), 16
were sunny and 20 were not. The mean number of registrations
in the first distance zone (0-50 m from edge) was 3 times
higher on sunny mornings (x = 9.94, SD = 5.20) than on
non-sunny mornings (x = 3.60, SD = 2.42), a significant
difference (t = 4.851, P <.001).
All of the edge-types sampled in this study were abrupt
edges between mesic hammock and relatively open habitat. Of
the 2 natural edge-types sampled, however, the marsh edge
(MA1 and MA2) had higher structural contrast with hammock
than did the sandhill edge (SA1 and SA2). Each of these
edge-types was represented by 1 east-facing and 1 non-east
facing plot. In the breeding seasons, the slopes for high-
contrast marsh edges and moderate-contrast sandhill edges
were significantly different, the former being positive and
the latter negative (Fig. 3; F = 13.463, P = .001). Neither
slopes nor intercepts were different between these 2 edge-
types in fall or winter (P > .05).
- E o sc
0 0 0, C
0) CO O4
0 0 o
I I I, t 1- r
CO | 1- CCO i:l
U) 0 .C-
O 0 0
0 0 ) 0) 4)
C- 0 co (D 44J
mC *'- 4 C
." 01 / -0
(D T3 r
00 J 0 0
OE ** o E
0 O \. 1 .
o --- 0 -- 0 -- 0 -- 1 --0 0 -- 1 0 QJ
Bird Densities in Edge versus Interior Plots
Birds generally were attracted to edge within edge
plots, although seasonal trends and variation among species
and edge-types were evident. Edge effect was not apparent,
however, when bird densities were compared between edge and
interior plots (> 700 m from edge). Mean densities in
subplots of edge and interior plots were almost identical in
the seasons when all plots were censused (Table 8).
Variation in density was greater among subplots of edge plots
than among subplots of interior plots. Thus, edge affected
the distribution of bird activity within plots (and within
territories) that abutted edge; between plots, there was no
apparent attraction of birds to parts of the forest near edge
as opposed to deep in the interior.
Internal Patchiness and Its Relation to Bird
Densities and Edge Effects
Internal patchiness supplemented distance from edge as a
predictor of bird density in edge plots, but was a more
important predictor in interior plots. Correlations of bird
density with distance and with habitat variables varied among
plots and seasons in strength and, in a few cases, direction
(Table 9). Distance from edge (DIST), variation in distance
to nearest shrub (VS), shrub density (DENS), canopy openness
(OPEN), shrub species diversity (HS), and proportion of plot
area in gaps and bayheads (GAPBAY) were the variables with
the largest numbers of significant correlations with bird
TABLE 8. Number of bird registrations in 0.5-ha subplots of
5.0-ha edge and interior plots during the seasons when all
plots were censused, San Felasco Hammock, Florida, 1986.
N is the number of subplots in each group (edge and
Seasons/Registrations Edgee Interior z P
P-value is for 2-tailed z-test. Group variances were
homogeneous for Spring 1986 and TOTAL, as determined by
Bartlett's test, but not for Winter 1986.
TABLE 9. Significant correlations of bird registrations in
subplots with distance from edge and habitat variables for
15, 5.0-ha plots in each season, San Felasco Hammock,
Florida, 1985-86. Variables that produced no significant
correlations (HPD, GA, and BAY) are omitted.
PLOT DIST VT VS VTS
PLOT DENS VSD OPEN VCO
PLOT HT HS GB
PLOT GAB GAPBAY
PL3 F85:.768* F85:.768*
PL4 S85:.959*** S85:.959***
PP1 S86:.972*** S86:.972***
* P < .05
** P< .01
*** P < .001
registration density. Multiple regression models explained
5-71% of the variation in bird density within edge plots,
interior plots, and all plots combined (Table 10).
GAPBAY, the proportion of subplot area in gaps/bayheads,
was the single most important variable in multiple
regressions, although other variables were more important in
some seasons; DIST and HS were the most important variables
in subplots that lacked gaps or bayheads (-G). For those
subplots that contained gaps and/or bayheads (+G), GAPBAY
explained 14% of the variation in bird density in winter and
spring 1986 in all plots combined (Fig. 4). Bird density in
gap/bayhead subplots was significantly higher than in
subplots without gaps/bayheads in spring (P <.001), and in
winter and spring combined (P <.001), but did not differ in
winter (P = .26) (Table 11). Gap area alone (GAB) was nearly
as good a predictor of bird density in all cases, but does
not appear in the optimal equations (Table 10) because it was
highly correlated with GAPBAY (e.g., r = .88, P <.001 for all
plots combined); in fact, GAB is equivalent to GAPBAY in all
but the 2 plots (PP2 and IF1) that contained bayheads.
Bayheads, however, were major attractors of birds in these 2
Distance from edge (DIST) was the most important
variable in fall and winter in the 12 edge plots, and in the
breeding seasons in subplots of edge plots that lacked
gaps/bayheads (Table 10). All correlations with DIST were
H I ~ i) ti (
) 0-4J C 0 0 r- 0
r. co)t 0 r, Z -H (d ,-4
) Ln) -,i M(a > +I
o to o a) a(ri o
*- M 0 M 0 04 M 0) I + I + + + + + + +
Jtr a) 'i 4 or
C *-I ,0 ) E -40
, to H %-m U) + .
-4i t OH la) o o r-0
J > 4'U-(0 0 0 0 0 0 0
HHH00 \ *WW O. O V V
40 +JCO H 0 CO O -C
in k e os ai o N I ++ O + +i
-40H -40 H- 0 i 0
0 (0 ) -C 00
C' ( O-V I> r.0A 9 o4
O H 041> r- i- i -- -0 O-
S.,n O ou Q*- 0 o o
o o' M a O a >i .
n O (*n Oi E -P O O -, ,- *< I VI
m 4004c I4 W -E-4
04 0 0 0* 0 -
0 u o oo 0- -0
, (X H p o H
v I I r. r- o U co o Q 4 r
*HJ HrI r( rO M^ m-l H H^ OU)- c- N
9 -H 0 (13*-H 14-4 H 00
043 (04-J 4JP ( 0- C0 % l 0O H 0 0 0-
3 t-o'04.) 0i-H>4-) En +Q + E
UO 0 V to AS (0 >4 >
0i OM 4 O4-) 0 l '- I I I I
SJ N4J 00-. M )i O m
iH (0 6 C 3 l4-).-i c(O HlO uO H( C- n
* w .- O 0 OH *r *4 M y in o
0) to 06> :00 p 0 o
(03o04 J o I0-H0 CO I 0 04 (k (
W 4) r. FA OD :J eU-4 -
4) H! Q -C p04 ((d ( H H H
pn )o a coO
-HM 4 r-i O M+ P N N Nr +
W >4 4- W 0- (1) U ) H
rzl w^l CA Ca' U COH + I H +
0 r-A O1 A- +0 4+
j Vo 0 (U Bo M to v O Kr Ch
(13 0 ; 4 1 c4 *
*-iOrO0-)0H0 0 t 0 L A L
0 0 0 M 4-) C0
En wc z
E-4H E- E-4
> >0 >
++ ++ I I
EnE-i p )t U M
> > >+ >+
I + + 1 +
w 0 0
0> >> > 0
++ 1 + ++
*O c> t
in + cc
I i-4 + 0
H r 4
+ + I,
0 Mu ar
* C (NJ
1H H -
m M o
CO I r
0 0 0
En C MO C I
+ > >
+ +1 +1
0 0 i 0 0
++ +E U+
+ + +1 + +
rp aIm c A
> > > > 0 Q
++ 1+ +
Lo >L L
.I hI ()
M E-4 MW
01 co +
S 0 O
X u ur T
N C CQ)
S* 0 ,I
o( il ~l .T3
II r- a
c' a cO
0. cc 0
No CD 0
0 I C r
(M* \ r I c 1
N N N4N *n
* N 0 r- i-4 cO
0CM N CO .,
p 0 0
"a 0 p
C* in 0\0 0 M 0 V
\n aof. U U)
o Oa > *
0C0 44 0- 00
S6D m ro m o-
0 r. a
S .0 .*0
P Q91 'a
\t C >
\r- -m X 0C
0ul 0I 0 0 00 m E-4
TABLE 11. Number of bird registrations in subplots with
(+G) and without (-G) area in gaps/bayheads for the 15,
5.0-ha plots, winter and spring 1986, San Felasco Hammock,
Florida. N is the number of subplots in each group. P-
value is for 2-tailed z-test.
Seasons/Registrations caps gaps z P
11 69 7 55
negative, indicating a positive edge effect. DIST was not
important when interior plots (calculated as DIST = 700 m for
all subplots) were considered with edge plots. HS was more
important than DIST or GAPBAY in winter for subplots lacking
gaps/bayheads, and was a significant variable in interior
plots and in all plots combined. HS was negatively related
to bird density, probably because the dense shrub-level
vegetation that was used heavily by birds often was dominated
by just 1 or 2 woody species.
Regression analyses revealed that subplots without gaps
or bayheads (Fig. 5A) exhibited a stronger edge effect than
those with gaps/bayheads (Fig. 5B), in the breeding seasons.
The slopes of these 2 regressions did not differ (F = 1.241,
P = .268), but their intercepts did (ANCOVA, F = 18.381, P
< .001). The same was true when subplots with and without
gaps alone were compared, but the intercepts differed less
than in the comparison with bayheads included (ANCOVA, F =
10.527, P = .002). There were no differences in slope or
intercept in either case in fall or winter (ANCOVA, P > .25).
Patterns at the Between-Plot Scale
Relationships between bird density and habitat
heterogeneity at a between-plot observation scale (Table 12)
were different from those determined at a within-plot scale
and were less often significant. For spring 1985 and both
NO ** *
0 0 0
CO (D I-
a 0- CO
w w 00
1.0 C 00
C 1-4 0 r-4
Q) cc* c
4-41 w CO C
w) < ow 0
'-' 0 Co
0 m a
o E (U
C 0 C a
.C 44 .-r W)
co u wu
0 0 r4
C0 CO .
.- C CQ
C o0 0rl
C1 r4 C ^
t 0 CO4
CO ca V C
CU)l 00 CU
C W0 Co
**C CQ to
X Lr L, "
0 II II
+ 0 o CU ~,p
N O* ** E
0 0 0
*- s0l 0
0* C\I Y~ N1e
0 o 0 0 0 0
s uo ID pip E 0
C r- c ( 0
00 My 0 41 M mM ME4
*HF- MH* ri = > c En
(L ) tYr- A> -u + + II +
S0 X 0
*n rc a a0 c()
0e1 01 ( 4-q 10
>i0 rH k U Q) U]
M -4 (0 N
So (amr C(O *
U 0 -4 > (0 +
SV ) q m to
O* 0m .D t -
H -0H M W d n v
ON 0 4) O
cO 0 l *
Oi QH) ( N <
c i 0c o M -o
4) 0 0 0 04
OW* C -4 0
S) Ico 44. 0 + I4
0 M* o WC r H Hc4)p H4 (N
04 V m*- O 0 U
S0 H 4 *H q *
0 o M r. C 0
<4) 00 0 H
4O0 4 4 0 0 0
-riHWO) H 4nV -H 0 V 1 H
w c cO co 9 0 9 o H w (
-H 4aJo -H J-n4J
$4 (3 (4p 00 u 4 o (1)0o *v
$4 > r--1 Q P rQ) M r4
tH a0f 0 -H0 U) N
0M o l n -4 o ) o
k r-- 4 4 int U
0 4)) UN0 0 0H
E0 a) P M) U) *t 0
E-4 rn u) 44
0 r Ea
IO G)M 0 0
4 U)O a a c G oIC
springs (breeding seasons) combined, tree species diversity
(HT) explained most of the variation in bird density among
plots, the relationship being inverse. In fall, a positive
relationship with canopy openness (OPEN) was most important.
Gap and bayhead area (GAPBAY) and other variables indicating
patchiness were not important at the between-plot scale
(except, as noted above, breeding species richness was
significantly associated with variation in shrub density,
Responses of Individual Species to Habitat Heterogeneity
Responses of the 12 most abundant bird species to
distance and habitat variables revealed that, for most
species, density was associated significantly with habitat
patchiness (Table 13). Edge effects observed within edge
plots (Table 7) for many species were swamped by high
abundances in interior plots >700 m from edge; distance from
edge was an important variable for only the Red-bellied
Woodpecker (which was attracted to edge) and Hooded Warbler
(which avoided edge) when all 15 plots were considered (Table
Gap and/or gap/bayhead area (GB, GAB, or GAPBAY) showed
significant correlations with density for 5 species (Northern
Parula, Carolina Wren, White-eyed Vireo, Hooded Warbler, and
Downy Woodpecker). Variations in spacing of trees and/or
shrubs (VT, VS, or VTS) correlated positively with density
TABLE 13. Responses of individual species to edge and
habitat heterogeneity, as indicated by significant
correlations for the 12 most abundant bird species in
the 12 edge and 3 interior 5.0-ha plots combined, San
Felasco Hammock, Florida, 1986.
Species Variable r
Species Variable r
Acadian Flycatcher (none significant)
Summer Tanager (none significant)
* P < .05
** P < .01
*** P < .001
for 4 species (Carolina Wren, White-eyed Vireo, Red-bellied
Woodpecker, and Hooded Warbler). Density of 4 species
(White-eyed Vireo, Red-bellied Woodpecker, and Ruby-crowned
Kinglet) was related negatively to diversity of trees and/or
shrubs (HT, HS). Two species (Carolina Wren and Red-bellied
Woodpecker) were positively associated with an open canopy
(OPEN), whereas the Red-eyed Vireo was negatively associated
with OPEN. Two species (White-eyed Vireo and Hooded Warbler)
were positively associated with shrub density (DENS), and 1
(Red-bellied Woodpecker) with variation in canopy openness
(VCO). Finally, 4 species (Tufted Titmouse, Northern
Cardinal, Acadian Flycatcher, and Summer Tanager) showed no
significant correlations with any of the variables.
Bird Responses to Forest Edge and Internal
Patchiness: A Matter of Scale?
A major objective of this study was to assess the
simultaneous and interacting effects of edge and internal
patchiness on bird habitat-use. Both forest edges and
internal patchiness (especially in the form of gaps and
bayheads) attracted high densities of birds within 5.0-ha
plots. Edge effect during the breeding season was
significantly stronger in 0.5-ha subplots lacking gaps or
bayheads than in patchy subplots. Forest edges, particularly
in fall and winter, appeared to "distract" birds from gaps
and bayheads; i.e., birds were attracted more strongly to
gaps and bayheads in interior plots (> 700 m from edge) than
in edge plots. In interior plots, patchiness was the best
predictor of bird density in all seasons, and was especially
important during the breeding season, when nesting, foraging,
and territorial display for many species were concentrated in
gaps and bayheads.
Attraction to edge was strongest seasonally in winter,
whereas attraction to gaps and bayheads was strongest during
the breeding season. East-facing and high-contrast edges
attracted higher densities of birds than did edges that faced
other directions or exhibited less structural contrast
between adjoining habitats. Of 27 bird species analyzed, 12
were attracted to edge, 11 were apparently indifferent, and 4
avoided edge. A common null hypothesis in edge-effect
studies is that animal densities at different distances from
edge are equivalent. In this study, bird densities across
distance zones were rarely uniform, and a clear edge effect
(concentration of registrations 0-50 m from edge) occurred in
6 of 12 edge plots, for all seasons combined.
Despite indications of attraction to edge within edge
plots, edge and interior 5.0-ha plots had equivalent
densities of birds and equivalent breeding species richness.
The lack of any difference in density between edge and
interior plots may be a consequence of habitat selection
mechanisms differing at different spatial scales (see below),
or may be related to the lower densities observed in the most
distal (200-250 m) zone of edge plots. Although sampling
problems cannot be ruled out, attraction to edge may produce
a "vacuum effect," where animal activity or density is
depressed at an intermediate distance from edge compared to
deeper forest (Bider 1968).
Similarly, although birds were attracted to gaps and
bayheads within plots, between-plot comparisons indicated
that bird density did not differ between patchy plots and
more homogeneous plots. Breeding species richness, however,
was positively associated with one indicator of habitat
patchiness at a between-plot scale, the coefficient of
variation of shrub density.
The empirical generalization that birds are abundant
near openings, discussed in the ornithological literature
since Lay (1938), was supported by this study. But this
simplistic interpretation is incomplete because it derives
from only one scale of resolution. Habitat associations of
birds differ depending on the scale at which they are
examined (Gutzwiller and Anderson 1987, Wiens et al. 1987).
The edge effect and the "patch effect" may be scale-
dependent in 2 ways: (1) birds respond to different sets of
habitat cues when selecting forests in which to settle, when
establishing territories or home ranges within forests, and
when selecting nesting, foraging, singing, and roosting sites
within territories and home ranges (Hutto 1985, Wiens 1985);
and (2) different human observation scales lead to detection
of different patterns (Allen and Starr 1982, O'Neill et al.
1986, Wiens et al. 1987).
Beginning with Shelford (1913, 1927), who described an
abundance of animals at the forest margin, and Leopold
(1933), who advanced the idea that "game is a phenomenon of
edges," the attraction of wildlife to openings has been one
of the best-studied phenomena of habitat selection. Numerous
studies since Lay (1938) have documented higher bird species
richness near forest edge (Johnston 1947, Johnston and Odum
1956) and/or increased densities of birds near edge (Beecher
1942, Good and Dambach 1943, Johnston 1970, Gates and Gysel
1978). Although documentation of edge effects has been
ample, a lack of standardized methodologies and definitions
has produced some confusion. Sometimes the area of habitat
in various distance zones varies widely and is not specified;
hence, density-distance correlations can be spurious and edge
"effects" are really artifacts (Nelson et al. 1960, Harris
and McElveen 1981).
Much of the confusion in the edge literature reflects
the many ways in which edge, edge species, and edge effects
have been defined. Some studies of edge effects on birds
(e.g., Kendeigh 1944, Johnston 1947, Whitcomb et al. 1981)
have used qualitative criteria to classify species according
to their edge affinities. Typically, bird registrations are
mapped, territory boundaries are defined, and species are
classified as "forest-edge" if territories are concentrated
along forest margins, as "forest-interior" if territories are
located primarily inside forest, or sometimes as "interior-
edge" if territories are found both in edge and interior
Other studies have used quantitative criteria to
determine the edge-interior affinities of birds. Galli et
al. (1976) defined edge width according to structural
characteristics of vegetation determined in a different study
nearby (Wales 1972). Gates and Mosher (1981) took a
"functional approach" to estimating edge width, based on
dispersion of nests of bird species associated with edge
habitat. Kroodsma (1982, 1984) mapped territories along an
edge-interior transect, and defined edge species as those
with highest densities (fractions of territories) within 60 m
of edge and/or a significant negative slope of density
against distance from edge. Strelke and Dickson (1980) and
Helle and Helle (1982) used densities of registrations within
arbitrary distance zones to determine strength of edge effect
and to classify species by edge affinities. My study was
modeled in part after those of Kroodsma (1982, 1984), Strelke
and Dickson (1980), and Helle and Helle (1982). Because edge
effects were tested among a series of replicated, well-
dispersed plots, statistical inferences could be made that
were inappropriate in previous studies.
Researchers in the eastern deciduous forest region have
often described a characteristic forest-edge avifauna (e.g.,
Kendeigh 1944, Johnston 1947, Johnston and Odum 1956, Forman
et al. 1976, Whitcomb et al. 1981). In contrast, this study
at the southern extreme of the deciduous forest biome (Braun
1950) revealed few distinct forest-edge birds. Five breeding
species, the Brown Thrasher, Common Yellowthroat, Blue
Grosbeak, Indigo Bunting, and Brown-headed Cowbird, were
confined to edges but were uncommon in the study area and
were not analyzed. The Rufous-sided Towhee, considered an
edge species in the northern studies mentioned above,
inhabits deciduous forest edges in Florida, but is more
abundant as a breeding species in pine flatwoods and
plantations (Repenning and Labisky 1985). Other breeding
species classified as forest-edge birds in many northern
studies, such as the Yellow-billed Cuckoo, White-eyed Vireo,
and Northern Cardinal, were found in San Felasco in all
distance zones but most abundantly near edge. In forest
interior, these latter species concentrated their activity in
gaps, bayheads, and other patches of dense shrub growth. The
Gray Catbird, a typical edge species that wintered at San
Felasco, was found at all distance zones but was strongly
associated with edge. Thus, almost all of the bird species
analyzed in this study used both forest interior and edge
habitat, although species differed in degree of attraction
to, or avoidance of, edge. Several species generally
associated with edge also inhabited the interior of San
Felasco Hammock, apparently due to its patchiness. This
observation leads to a question, as yet unanswered by any
study -- at what point does forest "interior" become so
patchy that it is no longer interior?
Qualitative aspects of edges may influence faunal
distribution (Harris and Smith 1978, Harris 1980, 1984,
Harris and McElveen 1981). Exposure and contrast are
potentially important qualities of edge in terms of bird
attraction. At San Felasco, a greater positive edge effect
was observed in high-contrast hammock-marsh edge plots than
in moderate-contrast hammock-sandhill edge plots during the
breeding seasons. This finding agrees with previous
conclusions that the magnitude of edge effect increases with
greater structural contrast between abutting communities
(Thomas et al. 1979, Harris and McElveen 1981, Harris 1984).
This study also documented a greater attraction of birds
to east-facing edges in winter, as compared to edges that
faced other directions. Densities of birds on east-facing
edges were significantly higher when morning sunlight was
striking the edge, suggesting thermoregulatory behavior, or
more likely, a response to increased insect activity. In
contrast, Helle and Helle (1982) found a peak in bird density
50-100 m from forest edge on Finnish islands and suggested
that avoidance of climatic extremes may keep many birds
(especially tropical migrants) away from edge in colder
climates. Carpenter (1935) found higher bird densities on
edges on the "lee" side of Illinois forests, sheltered from
prevailing winds; density differences between windward and
leeward edges were more pronounced in winter and early spring
than in late spring. In north-central Florida, winds are
multi-directional during the day but generally northerly at
night (Dohrenwend 1978), so wind should not produce
consistent differences in diurnal use of edges. Hence,
climatological effects of edge appear to influence bird
activity in many regions, but differ according to regional
and local environmental conditions.
The edge effect, as it is usually understood, is a
response of animals to major habitat interfaces
(macroheterogeneity). Internal forest dynamics, however,
produce openings that are structurally similar to forest
edge, and might be considered forest edge at a finer scale.
In this study, the proportion of subplot area in gaps (plus
limited bayheads) was the best predictor of bird density at a
within-plot scale of analysis. The simple measure of
gap/bayhead area was more important than various
heterogeneity indices or the floristic diversity of trees and
shrubs. When data from interior plots were combined with
those from edge plots, the proportion of area in
gaps/bayheads was a far better predictor of bird density in
subplots than was distance from edge. This conclusion
applied to most individual species and to all species
combined. For the 12 edge plots, edge effect in relatively
homogeneous subplots during the breeding seasons was
significantly stronger than in subplots containing gaps
and/or bayheads. These results suggest that openings (and
other shrubby areas) of all sizes are attractive to birds,
but that the relative attractiveness of major edges and
smaller, internal patches depends on the range of distances
from edge and degrees of patchiness sampled.
The attraction of birds to gaps is just one indication
of the ecological significance of these patches. Gap
dynamics are an important regeneration and diversifying
phenomenon in many types of forest (Bray 1956, Williamson
1975, Hartshorn 1978, White 1979, Runkle 1981, 1982, 1985,
Brokaw 1985). In southern hardwood forests mesicc hammocks),
a number of co-dominant tree species coexist, apparently as
non-equilibrium populations, and respond independently to
disturbances that produce gaps in the canopy (Platt and
Herman 1986, Platt and Schwarz, in press). Rates of gap
formation by treefall and canopy degeneration have not been
measured in these forests, but might be expected to fall
within 0.5-2.0% per year, a rate described for mixed
mesophytic forests and many other forest types (Runkle
In San Felasco Hammock, sinkhole formation supplements
wind and tree death as a source of canopy gaps. Several
recent or enlarging sinkholes have resulted in treefalls and
shrub-level enhancement similar to that associated with
windthrows. Deep sinkholes are usually water-filled (at
least seasonally); the largest sinkholes form persistent
ponds with fringing marshes and/or shrub swamps. The largest
sinkhole in San Felasco Hammock covers 9.6 ha; the largest
gap occurring within the study plots was a 3208 m2 sinkhole
gap. The mean gap area (3.4%) on study plots was within the
range recorded by Runkle (1982) in 15 old-growth mesic
forests in the eastern United States. Although our gap
definitions differed (see Runkle 1982), gap area in San
Felasco Hammock appears to be typical of old-growth forests
in the eastern United States.
Unlike larger habitat discontinuities (edges), natural
gaps and other aspects of habitat patchiness were not studied
by many wildlife biologists or ornithologists until recently.
Kendeigh (1944) noted that forest interiors were sometimes
"infiltrated" by forest-edge species such as the Northern
Cardinal and Rufous-sided Towhee. These species, he wrote,
may occur "where openings or thickets have been naturally
made by trees being blown over or by other disturbance.
However, such openings are 'wounds' in the community
structure, and since they occur to such a varying extent in
different sample plots, it seems best to eliminate them
altogether" (Kendeigh 1944:96).
Recent studies have considered gaps in a more favorable
light. Birds may respond to gaps because they provide cues
indicating a concentrated food supply (cf. Hilden 1965). In
Panama, Schemske and Brokaw (1981) found more bird species in
gaps than in undisturbed forest understory, and considered
several species "treefall specialists." In Costa Rican cloud
forest, flowering and fruit production are concentrated in
treefall gaps (Linhart et al. 1987). In Illinois, Blake and
Hoppes (1986) and Martin and Karr (1986) captured more
migratory frugivores (in fall), granivore-omnivores, and
certain (especially foliage-gleaning) insectivores in gaps
than in non-gap areas of forest understory. Bird abundance
in gaps was correlated with greater food abundance,
particularly fruits and foliage. Many insects favor plants
growing in sunlight over those growing in shade, and
lepidopteran larvae may attack plants in gaps preferentially
(Wolda and Foster 1978, White 1984, Harrison 1987).
In this study, association of birds with gaps and
bayheads was strongest during the breeding seasons. Although
birds may concentrate in gaps because they are richer in
food, nest-site selection may be just as important (Morse
1985, Martin 1988). Dense shrub-level foliage within gaps
provides abundant nesting substrates for birds that nest in
this stratum; concealing cover around nests also offers
protection from nest predators (Chasko and Gates 1982). Ten
of 18 breeding bird species that were analyzed in this study
are primarily shrub-level (below 5 m) nesters, and all but 1
(Pileated Woodpecker) of the remaining species nest at either
shrub or tree level (Appendix II, Harrison 1975).
Van Horne (1983) cautioned that animal density may be a
misleading indicator of habitat quality. If surplus,
socially-subordinate individuals (e.g., young, inexperienced
birds, many of which may be non-breeders or "floaters")
collect in habitat sinks, lower-quality habitat may actually
contain a higher density of individuals. Fretwell and Lucas
(1969) predicted that individuals select patches of highest
quality first, but that as population increases, a threshold
is reached where individual fitness is maximized by selecting
lower-quality but unoccupied habitat. Without long-term,
intensive studies in gap versus non-gap sites that measure
differences in habitat occupancy and fitness over a range of
population densities, the importance of gaps to birds remains
inferential. Life history information (for most of the bird
species here), high resource levels in gaps, and attraction
of birds to gaps within plots warrant the prediction that
patchy plots will be preferred by birds until increasing
population density makes less patchy plots more advantageous
in terms of fitness.
Reviews of diversity concepts in wildlife management and
conservation (Samson and Knopf 1982, Noss 1983) indicate that
maximization of local habitat diversity and edge effect has
been a guiding principle. This management emphasis appears
perfectly consistent with what ecology tells us about the
dependence of organisms on disturbance, successional patches,
and habitat mosaics, and the inferred relationship between
diversity and stability (Pickett and Thompson 1978, Hansson
1979, Gilbert 1980, Karr and Freemark 1983, Pickett and White
1985, Forman and Godron 1986). But what is beneficial at a
local scale and for edge-adapted organisms, such as many game
species, may be deleterious at larger spatial scales and for
sensitive organisms (Faaborg 1980, Samson and Knopf 1982,
Noss 1983, Wilcove et al. 1986). Furthermore, natural
disturbances and anthropogenic disturbances may have
qualitatively different effects on wildlife, a problem that
awaits detailed study.
One common method of enhancing horizontal habitat
diversity is through the construction of "wildlife openings."
Lay (1938:256) was one of the first to advocate "the
provision of clearings with extensive margins" for songbirds.
Because the edge effect he noted was confined to the first
100 m from an opening, and because the interiors of large
clearings were depleted of wildlife, Lay recommended small
but numerous clearings. Leopold (1938:3) made similar
recommendations with regard to deer, songbirds, and
wildflowers: "The smaller and more frequent the selective
cuttings, the greater the benefit to wildlife." In practice,
wildlife openings usually have been constructed for the
benefit of game rather than nongame species. Stoddard (1936)
recommended the maintenance of existing openings and
construction of new openings in heavily forested lands for
the benefit of Wild Turkey, whose poults forage in clearings.
Provision of forest openings for turkey, quail, grouse, deer,
and other game animals quickly became a dominant feature of
wildlife management on public lands (Larson 1967, McCaffery
and Creed 1969, Healey and Nenno 1983).
The optimum size of maintained openings for different
species has been debated (Patric 1966, McCaffery and Creed
1969, Segelquist and Rogers 1975). Because little research
has been conducted on optimum sizes of openings for nongame
birds, Taylor and Taylor (1979) recommended that a variety of
opening sizes be maintained. Recent research in Illinois
(Overcash and Roseberry 1987) determined that 0.1-0.2-ha
openings in mature deciduous forest were not large enough to
increase bird counts (species or individuals); edge effects
began with openings 0.3 ha in size. Brown-headed Cowbirds,
an indicator of deleterious edge effect (Brittingham and
Temple 1983), also appeared at this size, but were more
abundant in larger openings of 0.7-1.0 ha (Overcash and
Roseberry 1987). In Wisconsin, passerine nests within 35 m
of 0.01-0.2-ha openings were parasitized by cowbirds more
frequently than were nests further away, but the difference
was not significant; parasitism rates declined significantly
with distance from openings > 0.2 ha (Brittingham and Temple
In San Felasco Hammock, most bird species were
distributed throughout plots across the full range of opening
(gap) sizes, and bird densities were generally enhanced at
all gap sizes except for the very largest (a 0.32-ha
sinkhole-marsh opening), where densities were depressed.
There was no evidence of deleterious edge effects (e.g.,
cowbird parasitism or increased predation on artificial
nests) in or near even the largest gaps (Noss, in
preparation). Cowbirds presently are uncommon breeders in
this region, however, and only one successful parasitism (a
Red-eyed Vireo feeding cowbird young, B. Muschlitz, personal
communication) has been documented in the study area. More
research is needed to determine the size of opening at which
edge effects begin to occur, and the qualitative differences
in edge effects associated with natural versus artificial
Construction of openings in heavily stocked conifer
plantations and other close-canopied, even-aged forests may
be beneficial to nongame birds and other wildlife. In such
cases, small openings may simulate treefall gaps, which
because of young stand age are not occurring naturally, and
provide the herbaceous and shrub growth otherwise lacking in
the forest. If, however, the landscape is already heavily
fragmented with abundant clearcuts, other open areas, and
roads--all of which increase edge--constructing additional
openings could be counter-productive. Construction of large
openings may intensify deleterious edge effects on forest
interior species by further fragmenting the forest landscape
and favoring opportunistic, weedy species over species more
in need of protection (Robbins 1979, Noss 1983, Wilcove et
al. 1986). For guidance on management, a general rule is to
look to the regional landscape for context and to the needs
of the most sensitive species for specific direction (Noss
1983, 1987b, Harris 1984, Noss and Harris 1986).
The results of this study, which document attraction of
birds to edge and openings within plots, should not be
interpreted as supporting construction of artificial openings
or edge in natural forests. Artificially maintained openings
may differ from natural gaps in species composition and other
ecological properties (Denslow 1985). The structural
heterogeneity of San Felasco Hammock is due primarily to
natural processes. Natural gaps and bayheads were important
features of site heterogeneity and attracted high densities
of birds within plots. Because gaps provide concentrated
resources (food and nesting cover) for birds, management
strategies that maintain natural levels of horizontal
patchiness would be prudent.
Old-growth forests are naturally patchy, uneven-aged
systems that fractionate through natural disturbance into a
mosaic of developmental stages (Bormann and Likens 1979,
Oliver 1981, Whitney 1987); hence, they provide the
heterogeneity required by native species at no cost to
managers. Despite early impressions that old-growth forests
were wildlife-poor, contemporary wildlife ecologists
recognize the richness of this system (Meslow et al. 1981,
Schoen et al. 1981, Harris 1984). Because birds may respond
less to edge in patchy areas of forest than in more
homogeneous sites (this study), maturation of forests to old-
growth potentially could ameliorate deleterious edge effects
associated with fragmentation, i.e., edge could become less
of an "ecological trap" (sensu Gates and Gysel 1978).
Exceptions to the rule of maintaining unmanipulated stands of
maturing forest would occur when sites are too small to
incorporate the natural disturbance regime and maintain
habitat diversity (Pickett and Thompson 1978, White and
Bratton 1980, Shugart and West 1981, Noss 1987a, Urban et al.
1987) or if the needs of particular endangered species
dependent on successional habitat take precedence over
A Final Comment on Scale and Observation
Birds appear to evaluate habitat suitability at several
spatial scales (Ambuel and Temple 1983, Hutto 1985, Sherry
and Holmes 1985). Hence, there may be no fundamental scale
at which to assess bird-habitat relationships; the
appropriate scale for research depends on the specific
questions being asked. Conclusions are most meaningful when
they derive from several observational scales (Allen and
Starr 1982, Maurer 1985, Wiens 1985, 1986, O'Neill et al.
1986, Wiens et al. 1987).
Attraction of birds to edge and habitat patchiness at a
within-plot, but not at a between-plot, scale in this study
does not preclude reappearance of the relationship at a still
higher (e.g., between-forest) scale. In fact, a positive
association of both bird species richness and density with
habitat heterogeneity has been demonstrated repeatedly at
landscape, regional, and biogeographic scales (Williams 1964,
Wiens 1985, Boecklen 1986, Freemark and Merriam 1986).
San Felasco Hammock is larger, more mature, and more
heterogeneous than other hammocks in the region, and has a
richer avifauna (B. Muschlitz, personal communication, Harris
and Wallace 1984, personal observation). Although birds used
the entire between-plot range of habitat heterogeneity with
equal frequency, long-term research is needed to determine if
more heterogeneous sites are selected preferentially (cf.
Fretwell and Lucas 1969, Van Home 1983). Within plots, bird
activity was concentrated in gaps and other areas of dense
shrub-level vegetation. Edges were also attractive, possibly
because they provide selection cues similar to those of gaps
(Gates and Gysel 1978). Although many species were attracted
to both edge and internal patchiness, edge birds were not
equivalent to gap birds. Some species (e.g., Indigo Bunting)
were attracted to edge but were not found in gaps, whereas
others (e.g., Hooded Warbler) were associated with gaps but
avoided edge. These results suggest that species respond
uniquely to opening size and other habitat features.
The research reported here was correlative rather than
experimental because the site is protected as a preserve.
Therefore, mechanisms that controlled habitat selection could
only be inferred. Observed relationships ideally should be
tested by more rigorous experimental methods, but it is
premature to conclude that "further descriptive and
correlative studies can tell us little more that is new"
(Morse 1985:153). Although manipulations that create
different levels of heterogeneity may be useful,
observational studies spanning a breadth of spatial and
temporal observation scales can answer many questions about
habitat selection without modifying the small amount of
natural area that remains.
Allen, T.F.H., and Starr, T.B. 1983. Hierarchy: Perspectives
for ecological complexity. University of Chicago Press,
Chicago, Illinois, USA.
Ambuel, B., and S.A. Temple. 1983. Area-dependent changes in
the bird communities and vegetation of southern
Wisconsin forests. Ecology 64: 1057-1068.
American Ornithologists' Union (AOU). 1983. Check-list of
North American birds. Sixth edition. American
Ornithologists' Union, Washington, D.C., USA.
Ansley, C.C. 1952. An ecological comparison of the mesic
hardwood forests of central Florida. Thesis.
University of Florida, Gainesville, Florida, USA.
Beecher, W. 1942. Nesting birds and the vegetation
substrate. Chicago Ornithological Society, Chicago,
Bider, J.R. 1968. Animal activity in uncontrolled
terrestrial communities as determined by a sand
transect technique. Ecological Monographs 38: 269-308.
Blake, J.G., and W.G. Hoppes. 1986. Influence of resource
abundance on use of treefall gaps by birds. Auk 103:
Boecklen, W.J. 1986. Effects of habitat heterogeneity on the
species-area relationships of forest birds. Journal of
Biogeography 13: 59-68.
Bormann, F.H., and G.E. Likens. 1979. Pattern and process in
a forested ecosystem. Springer-Verlag, New York, New
Braun, E.L. 1950. Deciduous forests of eastern North
America. Macmillan, New York, New York, USA.
Bray, J.R. 1956. Gap phase replacement in a maple-basswood
forest. Ecology 37: 598-600.
Brittingham, M.C., and S.A. Temple. 1983. Have cowbirds
caused forest songbirds to decline? BioScience 33: 31-
Brokaw, N.V.L. 1985. Treefalls, regrowth, and community
structure in tropical forests. Pages 53-69 in S.T.A.
Pickett and P.S. White, editors. The ecology of natural
disturbance and patch dynamics. Academic Press,
Orlando, Florida, USA.
Carpenter, J.R. 1935. Forest edge birds and exposures of
their habitats. Wilson Bulletin 47: 106-108.
Chasko, G.G., and J.E. Gates. 1982. Avian habitat
suitability along a transmission-line corridor in an
oak-hickory forest region. Wildlife Monographs 82: 1-
Christman, S.P. 1984. Plot mapping: Estimating densities of
breeding bird territories by combining spot mapping and
transect techniques. Condor 86: 237-241.
Clewell, A.F. 1981. Natural setting and vegetation of the
Florida Panhandle. United States Army Corps of
Engineers, Contract No. DACWO1-77-C-0104. Mobile,
Cody, M.L. 1985. Habitat selection in birds. Academic Press,
Orlando, Florida, USA.
Cooper, W.S. 1913. The climax forest of Isle Royale, Lake
Superior, and its development. Botanical Gazette 55: 1-
44, 115-140, 189-235.
Cottam, G., and J.T. Curtis. 1956. The use of distance
measures in phytosociological sampling. Ecology 37:
den Boer, P.J. 1981. On the survival of populations in a
heterogeneous and variable environment. Oecologia
(Berlin) 50: 39-53.
Denslow, J.S. 1985. Disturbance-mediated coexistence of
species. Pages 307-323 in S.T.A. Pickett and P.S.
White, editors. The ecology of natural disturbance and
patch dynamics. Academic Press, Orlando, Florida, USA.
Dohrenwend, R.E. 1978. The climate of Alachua County,
Florida. IFAS Agricultural Experiment Station Bulletin
796, University of Florida, Gainesville, Florida, USA.
Dunn, W.J. 1982. Plant communities and vascular flora of San
Felasco Hammock, Alachua County, Florida. Thesis.
University of Florida, Gainesville, Florida, USA.
Emlen, J.T. 1971. Population densities of birds derived from
transect counts. Auk 88: 323-342.
Emlen, J.T. 1984. An observer-specific, full-season, strip-
map method for censusing songbird communities. Auk 101:
Ewel, J.J., and R.W. Simons. 1976. The vegetation of San
Felasco. Pages 20-28 in A plan to place San Felasco in
public ownership. Unpublished, Gainesville, Florida,
Faaborg, J. 1980. Potential uses and abuses of diversity
concepts in wildlife management. Transactions of the
Missouri Academy of Science 14: 41-49.
Forman, R.T.T., A.E. Galli, and C.F. Leck. 1976. Forest size
and avian diversity in New Jersey woodlots with some
land use implications. Oecologia (Berlin) 26: 1-8.
Forman, R.T.T., and M. Godron. 1981. Patches and structural
components for a landscape ecology. BioScience 31: 733-
Forman, R.T.T., and M. Godron. 1986. Landscape ecology. J.
Wiley, New York, New York, USA.
Freemark, K.E., and H.G. Merriam. 1986. Importance of area
and habitat heterogeneity to bird assemblages in
temperate forest fragments. Biological Conservation 36:
Fretwell, S.D., and H.L. Lucas. 1969. On territorial
behavior and other factors influencing habitat
distribution in birds. I. Theoretical development. Acta
Biotheoretica 19: 16-36.
Galli, A.E., C.F. Leck, and R.T.T. Forman. 1976. Avian
distribution patterns in forest islands of different
sizes in central New Jersey. Auk 93: 356-364.
Gates, J.E., and L.W. Gysel. 1978. Avian nest dispersion and
fledging success in field-forest ecotones. Ecology 59:
Gates, J.E., and J.A. Mosher. 1981. A functional approach to
estimating habitat edge width for birds. American
Midland Naturalist 105: 189-192.
Gauch, H.G. 1982. Multivariate analysis in community
ecology. Cambridge University Press, New York, New
Gilbert, L.E. 1980. Food web organization and the
conservation of neotropical diversity. Pages 11-33 in
M.E. Soule and B.A. Wilcox, editors. Conservation
biology: An evolutionary-ecological perspective.
Sinauer Associates, Sunderland, Massachusetts, USA.
Good, E.E., and C.A. Dambach. 1943. Effect of land use
practices on breeding bird populations in Ohio. Journal
of Wildlife Management 7: 291-297.
Greig-Smith, P. 1964. Quantitative plant ecology, 2nd
edition. Butterworth, London, England.
Gutzwiller, K.J., and S.H. Anderson. 1987. Multiscale
associations between cavity-nesting birds and features
of Wyoming streamside woodlands. Condor 89: 534-548.
Hansson, L. 1979. On the importance of landscape
heterogeneity in northern regions for the breeding
population densities of homeotherms: A general
hypothesis. Oikos 33: 182-189.
Harper, R.M. 1905. "Hammock," "hommock," or "hummock?"
Science, New Series 22: 400-402.
Harper, R.M. 1911. The relation of climax vegetation to
islands and peninsulas. Bulletin of the Torrey
Botanical Society 38: 515-525.
Harris, L.D. 1980. Forest and wildlife dynamics in the
southeast. Transactions of the North American Wildlife
and Natural Resources Conference 45: 307-322.
Harris, L.D. 1984. The fragmented forest: Island
biogeographic theory and the preservation of biotic
diversity. University of Chicago Press, Chicago,
Harris, L.D., and J.D. McElveen. 1981. Effect of forest
edges on north Florida breeding birds. School of Forest
Resources and Conservation, University of Florida,
Intensive Management Practices Assessment Center Report
Harris, L.D., and W. Smith. 1978. Relations of forest
practices to non-timber resources and adjacent
ecosystems. Pages 28-53 in T. Tippen, editor.
Principles of maintaining productivity on prepared
sites. USDA Forest Service, New Orleans, Louisiana,
Harris, L.D., and R. Wallace. 1984. Breeding bird species in
Florida forest fragments. Proceedings of the Annual
Conference of the Southeastern Association of Fish and
Wildlife Agencies 38: 87-96.
Harrison, H.H. 1975. A field guide to birds' nests. Houghton
Mifflin, Boston, Massachusetts, USA.
Harrison, S. 1987. Treefall gaps versus forest understory as
environments for a defoliating moth on a tropical
forest shrub. Oecologia (Berlin) 72: 65-68.
Hartshorn, G.S. 1978. Treefalls and tropical forest
dynamics. Pages 617-628 in P.B. Tomlinson and M.H.
Zimmerman, editors. Tropical trees as living systems.
Cambridge University Press, New York, New York, USA.
Helle, E., and P. Helle. 1982. Edge effect on forest bird
densities on offshore islands in the northern Gulf of
Bothnia. Annals Zoologica Fennici 19: 165-169.
Healey, W.M., and E.S. Nenno. 1983. Minimum maintenance
versus intensive management of clearings for wild
turkeys. Wildlife Society Bulletin 11: 113-120.
Hilden, O. 1965. Habitat selection in birds. Annals
Zoologica Fennici 2: 53-75.
Hurlbert, S.H. 1984. Pseudoreplication and the design of
ecological field experiments. Ecological Monographs 54:
Hutto, R.L. 1985. Habitat selection by nonbreeding,
migratory land birds. Pages 455-476 in M.L. Cody,
editor. Habitat selection in birds. Academic Press,
Orlando, Florida, USA.
James, F.C. 1971. Ordinations of habitat relationships among
breeding birds. Wilson Bulletin 83: 215-236.
Johnston, D.W. 1970. High density of birds breeding in a
modified deciduous forest. Wilson Bulletin 82: 79-82.
Johnston, D.W., and E.P. Odum. 1956. Breeding bird
populations in relation to plant succession on the
piedmont of Georgia. Ecology 37: 50-62.
Johnston, V.R. 1947. Breeding birds of the forest edge in
east-central Illinois. Condor 49: 45-53.
Karr, J.R., and K.E. Freemark. 1983. Habitat selection and
environmental gradients: dynamics in the "stable"
tropics. Ecology 64: 1481-1494.
Karr, J.R., and R.R. Roth. 1971. Vegetation structure and
avian diversity in several New World areas. American
Naturalist 105: 423-435.
Kartesz, J.T., and R. Kartesz. 1980. A synonymized checklist
of the vascular flora of the United States, Canada, and
Greenland. University of North Carolina Press, Chapel
Hill, North Carolina, USA.
Kendeigh, S.C. 1944. Measurement of bird populations.
Ecological Monographs 14: 67-106.
Kroodsma, R.L. 1982. Edge effect on breeding forest birds
along a power-line corridor. Journal of Applied Ecology
Kroodsma, R.L. 1984. Effect of edge on breeding forest bird
species. Wilson Bulletin 96: 426-436.
Kurz, H., and R.K. Godfrey. 1962. Trees of northern Florida.
University of Florida Press, Gainesville, Florida, USA.
Larson, J.S. 1967. Forests, wildlife, and habitat manage-
ment--A critical examination of practice and need. USDA
Forest Service Research Paper SE-30.
Lay, D. 1938. How valuable are woodland clearings to
birdlife? Wilson Bulletin 50: 254-256.
Leopold, A. 1933. Game management. Charles Scribners Sons,
New York, New York, USA.
Leopold, A. 1938. Report on Huron Mountain Club. Huron
Mountain Club, Huron Mountain, Michigan, USA.
Linhart, Y.B., P. Feinsinger, J.H. Beach, W.H. Busby, K.G.
Murray, W.Z. Pounds, S. Kinsman, C.A. Guindon, and M.
Kooiman. 1987. Disturbance and predictability of
flowering patterns in bird-pollinated cloud forest
plants. Ecology 68: 1696-1710.
Lovejoy, T.E., R.O. Bierregaard, A.B. Rylands, J.R. Malcolm,
C.E. Quintela, L.H. Harper, K.S. Brown, A.H. Powell,
G.V.N. Powell, H.O.R. Schubart, and M.B. Hays. 1986.
Edge and other effects of isolation on Amazon forest
fragments. Pages 257-285 in M.E. Soule, editor.
Conservation biology: the science of scarcity and
diversity. Sinauer Associates, Sunderland,
MacArthur, R.H., and J.W. MacArthur. 1961. On bird species
diversity. Ecology 42: 594-598.
MacArthur, R.H., J.W. MacArthur, and J. Preer. 1962. On bird
species diversity--Prediction of bird censuses from
habitat measurements. American Naturalist 96: 167-174.
May, R.M. 1986. The search for patterns in the balance of
nature: Advances and retreats. Ecology 67: 1115-1126.
Martin, T.E. 1988. Habitat and area effects on forest bird
assemblages: Is nest predation an influence? Ecology
Martin, T.E., and Karr, J.R. 1986. Patch utilization by
migrating birds: Resource oriented? Ornis Scandinavica
Maurer, B.A. 1985. Avian community dynamics in desert
grasslands: Observational scale and hierarchical
structure. Ecological Monographs 55: 295-312.
McCaffery, K.R., and W.A. Creed. 1969. Significance of
forest openings to deer in northern Wisconsin.
Wisconsin Department of Natural Resources Technical
Bulletin No. 44.
Merriam, G. 1988. Landscape dynamics in farmland. Trends in
Ecology and Evolution 3: 16-20.
Meslow, E.C., C. Maser, and J. Verner. 1981. Old-growth
forests as wildlife habitat. Transactions of the North
American Wildlife and Natural Resources Conference 46:
Monk, C.D. 1965. The southern mixed hardwoods of north
central Florida. Ecological Monographs 35: 335-354.
Morse, D.H. 1985. Habitat selection in North American
parulid warblers. Pages 131-157 in M.L. Cody, editor.
Habitat selection in birds. Academic Press, Orlando,
Nelson, M.M., R.A. Chesness, and S.W. Harris. 1960.
Relationship of pheasant nests to hayfield edges.
Journal of Wildlife Management 24: 430.
Noss, R.F. 1981. The birds of Sugarcreek, an Ohio nature
reserve. Ohio Journal of Science 81: 29-40.
Noss, R.F. 1983. A regional landscape approach to maintain
diversity. BioScience 33: 700-706.
Noss, R.F. 1987a. From plant communities to landscapes in
conservation inventories: A look at The Nature
Conservancy (USA). Biological Conservation 41: 11-37.
Noss, R.F. 1987b. Protecting natural areas in fragmented
landscapes. Natural Areas Journal 7: 2-13.
Noss, R.F., and L.D. Harris. 1986. Nodes, networks, and
MUMs: preserving diversity at all scales. Environmental
Management 10: 299-309.
Noss, R.F., H.W. Kale, and C.W. Biggs. 1985. Florida
breeding bird atlas: handbook for cooperators. Florida
Audubon Society, Maitland, Florida, USA.
Oliver, C.D. 1981. Forest development in North America
following major disturbances. Forest Ecology and
Management 3: 153-168.
O'Neill, R.V., D.L. DeAngelis, J.B. Waide, and T.F.H. Allen.
1986. A hierarchical concept of ecosystems. Princeton
University Press. Princeton, New Jersey, USA.
Overcash, J.L., and J.L. Roseberry. 1987. Evaluation of
Shawnee National Forest wildlife openings. Cooperative
Wildlife Research Laboratory, SIU-C. Southern Illinois
University, Carbondale, Illinois, USA.
Patric, E.F. 1966. An evaluation of the size an shape of
forest clearings. New York Bureau of Game, Final
Report, Federal Aid Project W-105-R-6, Job 1-A.
Pickett, S.T.A., and J.N. Thompson. 1978. Patch dynamics and
the size of nature reserves. Biological Conservation
Pickett, S.T.A., and P.S. White. 1985. The ecology of
natural disturbance and patch dynamics. Academic Press,
Orlando, Florida, USA.
Platt, W.J., and S.M. Herman. 1986. Relationships between
dispersal syndrome and characteristics of populations
of trees in a mixed-species forest. Pages 309-321 in A.
Estrada, T.H. Fleming, C. Vasques-Yanes, and R. Dirzo,
editors. Frugivores and seed dispersal. Dr. W. Junk,
The Hague, The Netherlands.
Platt, W.J., and M. Schwartz. in press. Mixed species
southern hardwood forests. In Ecosystems of Florida.
Academic Press, Orlando, Florida, USA.
Quarterman, E., and C. Keever. 1962. Southern mixed hardwood
forests: climax in the southeastern coastal plain,
U.S.A. Ecological Monographs 32: 167-185.
Recher, H. 1969. Bird species diversity and habitat
diversity in Australia and North America. American
Naturalist 103: 75-80.
Repenning, R.W., and R.F. Labisky. 1985. Effects of even-age
timber management on bird communities of the longleaf
pine forest in northern Florida. Journal of Wildlife
Management 49: 1088-1098.
Risser, P.G., J.R. Karr, and R.T.T. Forman. 1984. Landscape
ecology: directions and approaches. Illinois Natural
History Survey Special Publication 2: 1-18.
Robbins, C.S. 1979. Effect of forest fragmentation on bird
populations. Pages 198-212 in R.M. DeGraaf and K.E.
Evans, editors. Proceedings of the workshop, Management
of north central and northeastern forests for nongame
birds. United States Forest Service General Technical
Roth, R.R. 1976. Spatial heterogeneity and bird species
diversity. Ecology 57: 773-782.
Roth, R.R. 1977. Vegetation as a determinant in avian
ecology. Proceedings of the Welder Wildlife Foundation
Symposium 1: 162-174.
Runkle, J.R. 1981. Gap regeneration in some old-growth
forests of the eastern United States. Ecology 62: 1041-
Runkle, J.R. 1982. Patterns of disturbance in some old-
growth mesic forests of eastern North America. Ecology
63: 1533- 1546.
Runkle, J.R. 1985. Disturbance regimes in temperate forests.
Pages 17-33 in S.T.A. Pickett and P.S. White, editors.
The ecology of natural disturbance and patch dynamics.
Academic Press, Orlando, Florida.
Samson, F.B., and F.L. Knopf. 1982. In search of a diversity
ethic for wildlife management. Transactions of the
North American Wildlife and Natural Resources
Conference 47: 421- 431.
Schemske, D.W., and N. Brokaw. 1981. Treefalls and the
distribution of understory birds in a tropical forest.
Ecology 62: 938-945.
Schoen, J.W., O.C. Wallmo, and M.D. Kirchhoff. 1981.
Wildlife-forest relationships: is a reevaluation of
old-growth necessary? Transactions of the North
American Wildlife and Natural Resources Conference 46:
Segelquist, C., and M. Rogers. 1975. Use of wildlife forage
clearings by white-tailed deer in the Arkansas Ozarks.
Proceedings of the Southeastern Association of Game and
Fish Commissioners 29: 568-573.
Shelford, V.E. 1913. Animal communities in temperate America
as illustrated in the Chicago region. University of
Chicago Press, Chicago, Illinois, USA.
Shelford, V.E. 1927. Animal communities in temperate America.
Bulletin of the Geographical Society of Chicago 5: 1-
Sherry, T.W., and R.T. Holmes. 1985. Dispersion patterns and
habitat responses of birds in northern hardwood
forests. Pages 283-309 in M.L. Cody, editor. Habitat
selection in birds. Academic Press, Orlando, Florida,
Shugart, H.H., and D.C. West. 1981. Long-term dynamics of
forest ecosystems. American Scientist 69: 647-652.
Skeate, S.T. 1987. Interactions between birds and fruits in
a northern Florida hammock community. Ecology 68: 297-
Sokal, R.R., and F.J. Rohlf. 1981. Biometry, 2nd edition.
W.H. Freeman, New York, New York, USA.
Stoddard, H.L. 1936. Management of wild turkey. Transactions
of the North American Wildlife Conference 1: 352-356.
Strelke, W.K., and J.G. Dickson. 1980. Effect of forest
clear-cut edge on breeding birds in east Texas. Journal
of Wildlife Management 44: 559-567.
Taylor, C.M., and W.E. Taylor. 1979. Birds of upland
openings. Pages 189-197 in R.M. DeGraaf and K.E. Evans,
editors. Proceedings of the workshop, Management of
north central and northeastern forests for nongame
birds. United States Forest Service General Technical
Thomas, J.W., C. Maser, and J.E. Rodiek. 1979. Edges. Pages
48-59 in J.W. Thomas, editor. Wildlife habitats in
managed forests: the Blue Mountains of Oregon and
Washington. United States Forest Service Agricultural
Handbook No. 553.
Urban, D.L., R.V. O'Neill, and H.H. Shugart. 1987. Landscape
ecology. BioScience 37: 119-127.
Van Home, B. 1983. Density as a misleading indicator of
habitat quality. Journal of Wildlife Management 47:
Wales, B.A. 1972. Vegetational analysis of north and south
edges in a mature oak-hickory forest. Ecological
Monographs 42: 451-471.
Watt, A.S. 1925. On the ecology of British beechwoods with
special reference to their regeneration. Journal of
Ecology 13: 27-73.
Watt, A.S. 1947. Pattern and process in the plant community.
Journal of Ecology 35: 1-22.
Whitcomb, R.F., J.F. Lynch, P.A. Opler, and C.S. Robbins.
1976. Island biogeography and conservation: strategy
and limitations. Science 193: 1030-1032.
Whitcomb, R.F., C.S. Robbins, J.F. Lynch, B.L. Whitcomb, K.
Klimkiewicz, and D. Bystrak. 1981. Effects of forest
fragmentation on avifauna of the eastern deciduous
forest. Pages 125-205 in R.L. Burgess and D.M. Sharpe,
editors. Forest island dynamics in man-dominated
landscapes. Springer-Verlag, New York, New York, USA.
White, P.S. 1979. Pattern, process, and natural disturbance
in vegetation. Botanical Review 45: 229-299.
White, P.S., and S.P. Bratton. 1980. After preservation:
Philosophical and practical problems of change.
Biological Conservation 18: 241-255.
White, T.C.R. 1984. The abundance of invertebrate herbivores
in relation to the availability of nitrogen in stressed
food plants. Oecologia (Berlin) 63: 90-105.
Whitney, G.G. 1987. Some reflections on the value of old-
growth forests, scientific and otherwise. Natural Areas
Journal 7: 92-99.
Wiens, J.A. 1985. Habitat selection in variable
environments: Shrub-steppe birds. Pages 227-251 in M.L.
Cody, editor. Habitat selection in birds. Academic
Press, Orlando, Florida, USA.
Wiens, J.A. 1986. Spatial scale and temporal variation in
studies of shrubsteppe birds. Pages 154-172 in J.
Diamond and T.J. Case. Community ecology. Harper & Row,
New York, New York, USA.
Wiens, J.A., J.T. Rotenberry, and B. Van Horne. 1987.
Habitat occupancy patterns of North American
shrubsteppe birds: The effects of spatial scale. Oikos
Wilcove, D.S., C.H. McClellan, and A.P. Dobson. 1986.
Habitat fragmentation in the temperate zone. Pages 237-
256 in M.E. Soule, editor. Conservation biology: The
science of scarcity and diversity. Sinauer Associates,
Sunderland, Massachusetts, USA.
Wilkinson, L. 1987. SYSTAT: the system for statistics.
SYSTAT, Evanston, Illinois, USA.
Williams, C.B. 1964. Patterns in the balance of nature.
Academic Press, New York, New York, USA.
Williamson, G.B. 1975. Pattern and seral composition in an
old-growth beech-maple forest. Ecology 56: 727-731.
Wolda, H., and R. Foster. 1978. Zunacetha annulata
(Lepidoptera; Dioptidae), an outbreak insect in a
neotropical forest. Geo-Eco-Trop 2: 443-454.