Forest Health Monitoring Program Overview - USDA Forest Service
Area-sensitive forest birds move extensively among forest patches
Transcript of Area-sensitive forest birds move extensively among forest patches
BIOLOGICAL
CONSERVATION
Biological Conservation 118 (2004) 377–387
www.elsevier.com/locate/biocon
Area-sensitive forest birds move extensively among forest patches
Gail S. Fraser *, Bridget J.M. Stutchbury
Biology Department, York University, Toronto, Ont., Canada M3J 1P3
Received 1 November 2002; received in revised form 2 May 2003; accepted 19 June 2003
Abstract
Habitat fragmentation is thought to create a barrier to individual movements particularly for area-sensitive species which, by
definition, prefer to breed in large tracts of forest. For two breeding seasons, we radio-tracked an area-sensitive species, the scarlet
tanager Piranga olivacea, in a fragmented landscape in northeastern PA. We found that scarlet tanagers made extensive and frequent
movements among fragments. Paired males were less likely to leave their capture fragment, and traveled shorter distances. Unpaired
males in fragments had two distinct tactics we labeled: ‘‘Sedentary’’ and ‘‘Mobile’’. Sedentary males stayed at one fragment and sang
at high rates, while mobile males spent 58% of the total time tracked off their capture fragment and traveled over a kilometer away
from the capture site over open fields and through forests. Mobile males were not floaters per se because they were territorial (i.e.,
singing) in multiple sites. Habitat structure of a fragment did not correlate with the percent time a male spent off the capture
fragment. Males in fragments experienced lower pairing success and were more likely to be first time breeders compared to males in
continuous forest. Our results suggest that movements by scarlet tanagers in fragmented landscapes are not restricted during the
breeding season, and that these movements are related directly to pairing status and indirectly to population density.
� 2003 Elsevier Ltd. All rights reserved.
Keywords: Forest fragmentation; Neotropical migrant; Pairing success; Corridors; Radiotelemetry; Space-use; Scarlet tanager; Piranga olivacea
1. Introduction
Many North American songbird populations are
declining as result of habitat fragmentation on their
breeding grounds (e.g., Robinson et al., 1995). Many
studies suggest that Neotropical migrant species that
prefer larger tracts of forests (‘‘area-sensitive’’ birds; see
Villard, 1998) are especially sensitive to forest frag-mentation (e.g., Askins et al., 1987; Freemark and
Collins, 1992; Lee et al., 2002). Area-sensitive birds are
expected to (1) experience lower reproductive and pair-
ing success with a decrease in forest area (e.g., Gibbs
and Faaborg, 1990; Porneluzi et al., 1993; Villard et al.,
1993; Robinson et al., 1995; Roberts and Norment,
1999); (2) frequently be absent in small forest fragments
(e.g., Askins et al., 1987; see also Brown, 1984); and (3)prefer to use corridors and frequently avoid crossing
open, non-forested areas (Sieving et al., 1996; Desro-
chers and Hannon, 1997; B�elisle et al., 2001; B�elisle and
* Corresponding author.
E-mail address: [email protected] (G.S. Fraser).
0006-3207/$ - see front matter � 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.biocon.2003.06.006
St. Clair, 2001). While the effects of fragmentation on
the distributions of Neotropical migrant species (e.g.,
Lynch and Whigham, 1984; Blake and Karr, 1987;
Freemark and Collins, 1992), and factors influencing
reproductive success (e.g., Robinson et al., 1995; Friesen
et al., 1999) have been extensively studied, little work
has been done on how individuals move within and
among fragments.Community-level, temperate breeding studies have
driven concepts on how populations are affected by
forest fragmentation in Neotropical migrant research
(Hunter, 1992), but an important gap in our under-
standing concerns how individuals of different species
move through a mosaic environment (Wiens, 1994).
Individual movements and habitat selection are key
components in understanding the effects of forest frag-mentation on population dynamics (Wiens, 1994). For
example, edge avoidance, or a reluctance to cross open
areas, are considered typical traits of area-sensitive and/
or forest interior Neotropical migrants breeding in the
temperate eastern forests of North America, yet for
most species we have little or no empirical data to
378 G.S. Fraser, B.J.M. Stutchbury / Biological Conservation 118 (2004) 377–387
support these ideas (Villard, 1998). Although edge
avoidance or a reluctance to cross open areas appear
uncharacteristic for migratory bird species, some evi-
dence suggests that migratory birds, upon arriving at the
breeding grounds, do not disperse or colonize new areasas rapidly as resident bird species (Bensch, 1999), and
many of these species are particularly sensitive to habi-
tat fragmentation (e.g., Freemark and Collins, 1992).
Low breeding dispersal rates or colonization into new
areas may be linked to behaviors such as site fidelity
(Bensch, 1999) or higher predation risks when crossing
open areas (Wiens, 1994; Grubb and Doherty, 1999).
Recent radio-tracking studies on two area-sensitivebird species demonstrate that movements between
fragments are not only possible, but can occur regularly
(Bayne and Hobson, 2001a; Norris and Stutchbury,
2001). Norris and Stutchbury (2001) found that male
hooded warblers Wilsonia citrina leave their small forest
fragments on a daily basis through the incubation and
nestling stages. They crossed gaps up to 460 m in search
of extra-pair copulations, despite being paired andhaving an active nest. Bayne and Hobson (2001a) ob-
served no differences in the amount of movement made
by male ovenbirds Seiurus aurocapillus in continuous
forest compared to fragmented habitat in the post-
fledging period. However, males with fledglings moved
less compared to those without young and rarely crossed
open areas and consequently those on forest fragments
were restricted in their movements compared to males incontinuous forest. These studies suggest that moderate
fragmentation and isolation of forest patches on the
breeding grounds may not inhibit movement among
patches by birds during certain periods of the breeding
season. Clearly, more data on habitat use are required to
understand how area-sensitive species behave in frag-
mented landscapes and to facilitate better strategies for
their conservation.We chose to study scarlet tanager Piranga olivacea
behavior in forest fragments because this species is
negatively influenced by forest fragmentation (Rosen-
berg et al., 1999), and scarlet tanagers are canopy birds
that prefer mature forests (Mowbray, 1999). There are a
number of studies that describe a positive relationship
between forest area and the presence of tanagers (e.g.,
Robbins et al., 1989; Askins et al., 1987; Freemark andCollins, 1992). Rosenberg et al. (1999, Project Tanager)
found a positive relationship between forest cover at the
landscape scale and the occurrence of tanagers, although
the response of scarlet tanagers to forest fragmentation
varied among regions. Roberts and Norment (1999)
recorded a complete absence of scarlet tanagers in
fragments less than 10 ha in a highly forested area
(northwestern NY). Although tanagers can occur insmall woodlots (e.g., 3 ha, Galli et al., 1976; 6.5 ha,
Blake and Karr, 1984), it is unclear whether these birds
were breeders or vagrants (see Mowbray, 1999) because
none of these studies involved radio-tracking of indi-
viduals. Here, we present the results of a radio-tracking
study on scarlet tanagers in northwestern PA and de-
scribe the movements and pairing success of individuals
in a fragmented landscape.
2. Methods
2.1. Study area
We studied scarlet tanagers at the Hemlock Hill Bi-
ological Research site (41�460N, 79�560W) and sur-rounding region (approximately 740 km2) in Crawford
County, northwestern PA in 2000 and 2001. Male
scarlet tanagers were located on 10 small isolated
woodlots (<10 ha, mean¼ 5.3� 2.5 ha, range 3.2–9.6
ha), 2 mid-sized fragments (44 and 75 ha), and 2 rela-
tively large forest plots (150, 165 ha), hereafter referred
to as continuous forest. All fragmentation in this area
was caused by agricultural activities and eight fragmentswere patches completely isolated by fields or roads,
while four woodlots were connected to other forest
habitat by small (<40 m wide) corridors. Mean distance
to nearest forests for the 12 fragments was 138� 21.9 m
(range 10–325 m).
We calculated the area and perimeter–area ratio of
forest woodlots with a planimeter (Planix 7) from 1993,
1:40,000 aerial photographs. Perimeter–area ratio wascalculated as the ratio of edge (m) to area (ha) for each
fragment. Percent forest cover within 1 km of each
woodlot was also estimated from aerial photographs.
Based on USGS topographical maps, total forest cover
for the region (740 km2) was approximately 39%. Al-
though forest cover in some areas has increased slightly
(<1%) since the production of these maps (personal
observation), all of the fragments used in the study ap-peared to be the same size as shown on the 1993 aerial
photographs. Furthermore, no major logging has oc-
curred in any of the fragments except for one that was
selectively logged (i.e., the fragment is still the same size)
in the past ten years. We concluded that there has been
little change in forest cover in the region since the early
nineties.
The four most predominant tree groups in the forestsfragments were maple Acer spp., hemlock Tsuga
canadensis, American beech Fagus grandifolia, and
hickory Carya spp. In the continuous forest plots the
mixture was similar except elm Ulmus spp. was more
common than American beech.
2.2. Radio-telemetry
We captured tanagers with song playbacks, a tanager
decoy (taxidermic mount), and one 12 m long mist-net
placed at standard height. Each bird was banded with a
Table 1
A summary of telemetry on scarlet tanagers in Northwestern PA
2000 2001
Continuousa Fragment Continuousb Fragment
Total tracking (h) 110.8 114.2 30.1 104.1
Mean tracking time (h� SD) 11.1� 3.6 12.7� 4.5 10.0� 3.5 10.4� 4.1
Range of tracking time (all males) 6–14.7 6–19.5 6–12.1 6–18
Number of males tagged 10 9c 3c 9
aMales tracked on 150 ha plot only.bMales tracked on 150 and 165 ha plots.cOne male each year was captured in a fragment and then settled in continuous forest.
G.S. Fraser, B.J.M. Stutchbury / Biological Conservation 118 (2004) 377–387 379
US Fish and Wildlife aluminum band, and an individ-
ually distinct color band combination. We measured
wing chord, tarsus length, and body mass (using a 50 g
Pesola scale) of each adult. We aged tanagers based on
plumage coloration; second-year (SY) males have brown
on their primaries, whereas after second-year males
(ASY) have all black primaries (Mowbray, 1999). Males(n ¼ 31; Table 1) were fitted with a small (1.4 g), BD-2G
radio transmitter (Holohil Systems Ltd, Carp, Ont.,
Canada) attached with a figure-8 cotton embroidery
thread harness (see Rappole and Tipton, 1991). The
transmitter and harness together weighed about 5% of
adult body mass (average �30 g) and the harness was
designed to fall off by the end of the field season.
Transmitters lasted up to 8 weeks and the range wasapproximately 1.5 km.
While we attempted multiple recaptures of all tana-
gers at the end of each season to remove the transmitter,
we had a low success rate (n ¼ 3). Nine transmitters (5 in
2000; 4 in 2001) fell off before the end of the study. In
the continuous forest, males with transmitters left on at
the end of the 2000 season experienced a similar return
rate (67%, n ¼ 6) to males with transmitters removed(57%, n ¼ 7). Surprisingly, 3 males with transmitters
attached in 2000 returned with transmitters still in place
upon arrival on the breeding grounds in 2001 (contin-
uous forest plot, all ASY males in 2000). These three
males were the first to return in 2001, attracted females
within a few days, and lost the transmitters within a
month of arriving. Also, all males that were recaptured
at the end of each study year had a mass that was within1 g of initial capture mass (mean initial capture
mass¼ 27.8� 1.0; mean recapture mass¼ 27.8� 0.29,
n ¼ 3). From these data, we concluded that scarlet
tanagers were not negatively affected by wearing a radio
transmitter (see also Powell et al., 2000).
We radio-tracked tanagers in 2 h bouts between 06:00
and 14:00 h, usually on foot 20–30 m away from the
focal bird, however on a few occasions (2.6% of totaltracking time) tanagers were tracked by car. Observa-
tions were not made during wet weather. We had a total
of 359 tracking hours and each male used in our anal-
yses had a minimum of 6 hours of tracking time (Table
1). For both years, birds were tracked between 13 May
and 30 June.
Because many studies have described scarlet tanagers
as forest interior or area-sensitive birds, we were par-
ticularly interested in documenting whether males
moved off the isolated woodlots. We report these results,
for paired and unpaired males, in percent time off cap-ture fragment (minutes off fragment divided by total
minutes tracked), and the rate off capture fragment
(number of trips off fragment per hour tracked). Percent
time off fragment for paired males was not divided into
different breeding stages because only two of 11 pairs
fledged young during our study and nests were difficult
to locate, so stage of nesting was most often unknown.
We also calculated percent time off territories for malesin continuous forest and compared them to paired males
in forest fragments with a t test. The distance traveled
was calculated from topographical maps (fragments) or
from grid marks (continuous forest; see below) and re-
ported as average values. When relevant, we also re-
ported corridor use as a percent of time tracked. All
averaged values are reported with �1 SE.
We determined whether a male was paired based onrepeated sightings of a female with the focal male, or
seeing her carry nest material, on a nest, or carrying
food for young. We used a v2 test to compare age dis-
tribution and pairing success between fragments and
continuous forest. We measured song rates (number of
song bouts during a tracking period) during the radio-
telemetry observations from males on forest fragments
and compared paired and unpaired male rates with a ttest.
We had two different methods for recording the lo-
cation of a male during a tracking session and deter-
mining subsequent territory size, depending on which
plot a male was tracked in. The 150 ha forest had con-
spicuous grid marks every 50 m; the location of a male
tracked on this plot was recorded by noting the closest
grid mark whenever the bird moved. We measured ter-ritory size with a planimeter for each tracking session
and reported the average territory size for each male.
The location of males in forest fragments or in the 165
ha plot were recorded by noting their position relative to
380 G.S. Fraser, B.J.M. Stutchbury / Biological Conservation 118 (2004) 377–387
obvious landmark features. At the end of each season,
we mapped all movements for males, measured the ex-
tent of these movements with a 50 m tape and used a
planimeter to determine territory size (i.e., area defended
with song). Movements off territories for continuousforest males were assumed when males moved greater
than twice the distance away from the perimeter of their
territory (i.e., usually 100 m or more) and were in other
males� territories. Movements off fragments for males in
forest fragments were considered time off territory and
these movements were not included in the area estimate.
Four males used and defended an entire woodlot,
therefore we considered the whole fragment to be theirterritory.
We found that some males in fragments not only
gained a partner mid-season, but that they also switched
territories mid-season (i.e., from a fragment to contin-
uous forest); therefore, we report the final status (e.g.,
paired or not, and in fragment or continuous forest) of
male tanagers as they were by the end of the telemetry
work (30 June for both years, or sooner if the trans-mitter fell off).
2.3. Habitat structure and movements off territories
Habitat structure, which can influence prey abun-
dance (Smith and Shugart, 1987), may also determine
how often males leave their fragments or territories. We
examined (1) whether there was a relationship betweenamount of movement off a forest fragment and the
fragment�s habitat features, (2) whether movements in
continuous forest were related to the habitat features of
territories, and (3) whether paired males� habitat differedbetween continuous forest and forest fragments. We
measured habitat characteristics with the James and
Shugart (1970) method. For each territory we used 5–
0.02 ha plots. The initial plot was positioned at thecenter of the territory, while the other 4 plots were po-
sitioned 50 m from the center, in each cardinal direction.
For each territory we determined tree density (number
of trees per ha), average diameter of tree at breast height
(cm), number of tree species, total log basal area, %
canopy cover, canopy height (m), and shrub density
(number of shrub stems per ha). Many of our variables
were correlated, therefore we log transformed all thevariables and used a Principle Components Analysis
(PCA; correlations matrix) to collapse our dataset
(McGarigal et al., 2000). We used principle components
(PC) I and II, % forest cover within 1 km, and perime-
ter–area ratio, in a backwards step-wise regression with
percent time off fragment (arcsin transformed; Zar,
1996) as the dependent variable for males on 12 frag-
ments (Mobile males not included, see Section 3). If afragment had more than one male in both study years
(n ¼ 3), we averaged the percent time off territory for all
males in the fragment. In addition, if a fragment had
more than one vegetation plot (n ¼ 1), we averaged the
values for this analysis.
We repeated the PCA and backwards step-wise re-
gression for 13 territories in continuous forest, excluding
perimeter–area ratio as a variable with the dependentvariable as percent time off territory. Percent forest
cover was included in the regression because some males
were on the edge of the forest (n ¼ 5). We included edge
and interior territories in one step-wise regression be-
cause we found no difference in percent time off territory
between males in the two areas (p ¼ 0:8) and because of
our limited sample sizes. Finally, we compared habitat
features (canopy cover and height, log total basal area,and shrub and tree density) between continuous forest
(n ¼ 12) and fragment plots (n ¼ 10) for paired males in
a series of t tests.
2.4. Occupancy of forest fragments
We used a playback system to survey 13 woodlots in
May 2000 (mean area� SD¼ 10.1� 19.6, range 3.0–75ha) and 15 woodlots in May 2001 (mean area¼11.9� 19.6, range 3.0–75 ha) to estimate percent occu-
pancy of isolated forest fragments. Fragments with
tanagers (either banded or unbanded) observed more
than once were considered occupied. In all but two cases
each year, the males with transmitters were the only
color-banded individuals present in the fragments.
3. Results
3.1. Radio-telemetry
3.1.1. Movements of males in forest fragments: paired
versus unpaired behavior
Paired males were less likely to leave their capturefragment than were unpaired males (t test, n ¼ 11; 8; t ¼�2:8; p ¼ 0:01) and traveled shorter distances than un-
paired males (t test, n ¼ 6; 8; t ¼ �3:15:0; p ¼ 0:01; Ta-ble 2). Paired males also had significantly lower song
rates than unpaired males (paired: mean¼ 0.4� 0.1
songs/min; unpaired: mean¼ 2.5� 0.3 songs/min; (t test,
n ¼ 11; 6; t ¼ �7:4; p < 0:001). We found no differences
in territory size between paired and unpaired males(Mobile males not included-see below) in fragments
(t test, n ¼ 11; 4; t ¼ �2:0; p ¼ 0:06; Table 2).
Observations of paired males off territories (i.e., off
fragment) can provide insights into the reasons for these
movements. Only six of 11 paired males moved off their
forest fragments and we observed considerable variation
in the number of times these movements occurred. Three
paired males only left once out of the total time tracked(24.2 h; mean per male¼ 8.1 h): two males (with no
young) left for unknown reasons (no visual observations
of them) and the third (nesting status unknown) was
Table 2
Movements of radio-tracked male scarlet tanagers in fragments and continuous forest
Continuous Fragment
Paired Unpaired Paired Unpaired
Stay and sing Move and
sing
Total
No. of males 12 1 11 4 4 8
Percent time off fragment or territory 4.7� 4.3n:s: 1.1 5.7� 2.2 6.7� 2.6 57.8� 8.5 32.3� 10.5
No. times off fragment or territory per hour 0.2� 0.2 0.17 0.2� 0.07 0.2� 0.09 N.A.c 0.2� 0.09
(n ¼ 4)
Distance traveled (m)a 106� 16.5��
(n ¼ 10)
25 173� 27
(n ¼ 6)
467.0� 144.8 1120� 204 793.4� 169.1
Territory size (ha)b 0.9� 0.2� 0.6 1.6� 0.3 3.2� 0.9 N.A.c 3.2� 0.9
(n ¼ 4)
Values are mean� SD.aCorridors were not used in movements. Distances are calculated as one-way trips. Sample sizes differ from number of males tagged because mean
distance traveled is only from males that left their territory.b Territory size calculated from tagged males only.c Because Move and Sing males were off their capture fragment more than 50% of the time tracked and did not make discrete movements related to
their capture fragment we did not calculate the number of times off fragment per hour or territory size. t Tests between paired males in fragments
versus paired males in continuous forest.* p ¼ 0:05.** p ¼ 0:04.
G.S. Fraser, B.J.M. Stutchbury / Biological Conservation 118 (2004) 377–387 381
observed foraging off territory. One paired male (with
no young) left twice out of total time tracked (12.1 h),
and on one of these occasions he was engaged in a male-
male chase. The last two paired males left multiple times
(>3) over the course of the total time tracked (33.7 h;
mean per male¼ 16.9 h). One male spent 17% of his time
in a neighboring fragment (200 m away) and was also
observed there with a female. This pair also spent 52%of their time in an adjacent corridor (40� 10 m) where
on one occasion, they were observed foraging with
fledglings on ripe elderberry berries (Sambucus nigra).
The second male increased his time off the capture
fragment from 2% during incubation stage to 28%
during the fledgling stage. He was observed singing
during the incubation stage in a neighboring fragment
and foraging during the fledgling stage in this samefragment and once more in another nearby forest.
We found that tanager pairs (4/11) on forest frag-
ments sometimes moved territories mid-season. Two
males disappeared immediately after nest predation
events, and they were not subsequently observed (either
visually through playbacks or by radio transmitter sig-
nals) in their fragment or in nearby forests for the re-
mainder of the field season. Two other pairs movedthree times to different territories within a season. Final
territories were 300 and 400 m away from the original
territories and were entirely off the capture fragments;
one pair moved to a neighboring fragment (approxi-
mately twice as large as the capture fragment), the other
pair moved to continuous forest.
We found that unpaired males (all SY; n ¼ 8) adop-
ted one of two distinct tactics that we describe as: (1)Sedentary and (2) Mobile. Sedentary males (n ¼ 4) spent
the majority of their time in their capture fragment
singing at high rates (Table 2). Mobile males (n ¼ 4)
moved over open fields and through forests and were off
their capture fragment more than 50% of the time
tracked (e.g., Fig. 1(a); Table 2). Unlike secretive float-
ers (e.g., Zack and Stutchbury, 1992), Mobile males
advertised by singing on the original capture fragment
(territorial behavior that allowed us to initially capturethem). Mobile males were difficult to follow because we
did not know where they would be from one day to the
next and they were often out of transmitter range (ap-
proximately 1.5 km) at the time of a scheduled tracking
session. For example, one male was out of transmitter
range 11 out of 18 days we attempted tracking and he
was located in the capture fragment on only three out of
the remaining seven days. Yet when he was on thecapture fragment he would respond to playbacks (i.e.,
was territorial). A conservative estimate of the total area
utilized by Mobile males (n ¼ 3) averaged 122� 99 ha.
The fourth Mobile male likely had a similar range as the
other males, but we only located him once off the cap-
ture fragment and consequently could not estimate the
total area traveled.
The mating status of unpaired males sometimeschanged comparatively late in the breeding season.
Three males gained females mid-season (two Sedentary
– 14 June and 7 July; one Mobile – 18 June). The two
Sedentary males (both the sole occupants of each
fragment) were counted as unpaired in the data anal-
ysis because one male�s transmitter fell off before
pairing occurred and the other male gained a female
after the end of our study. The one Mobile male wewere able to track after he was paired behaved
Ferris unpaired
+*53.8%
+9.5%
+12%
24.6%+
1 km
+100%+
Ferris paired(b)
(a)
Fig. 1. An example of (a) an SY, unpaired male who adopted the Move
and Sing strategy, tracked from 13 May to 13 Jun 2001 for 14 h. Song
rate was 1.2 songs/min. The boxes represent available forest. The star
represents the capture fragment (3.6 ha) and the crosses are the dif-
ferent areas in which this male was tracked. The percent time is the
time spent in each area out of the total time tracked (14 h). The dashed
line encompasses the total area of movements by male. (b) The same
male later in the breeding season after establishing a territory in con-
tinuous forest and attracting a female. Tracked from 18 to 26 Jun 2001
for six hours. Song rate was 0.008 songs/min. Percent time is spent in
territory. Note the marked reduction movements and the decline in
song rate after pairing.
Table 3
Comparisons of scarlet tanager pairing success, age structure, and
territory size, between continuous forest and forest fragments
Continuous forest Forest fragments
No. malesa 30 17
Age ratio (SY:ASY) 3:27 14:3
No. paired males
(SY, ASY)
27 (2; 25) 10 (7; 3)
% Pairing success 90% 59%
One SY unpaired male was captured in a fragment, but settled with
a female in continuous forest (see Fig. 1). Another ASY paired male
was initially in a fragment for two weeks (20 May–3 June) then moved
to continuous forest with, presumably, the same partner. Both of these
males were placed in the continuous forest category.a Includes all birds intensively followed from both study years, in-
cluding individuals that were not banded. For radio-tagged birds lo-
cation and pairing status of males is that at end of study (30 June).
382 G.S. Fraser, B.J.M. Stutchbury / Biological Conservation 118 (2004) 377–387
similarly to paired males observed at the beginning of
the season (Fig. 1).
3.1.2. Age structure and pairing success: fragments versus
continuous forest
SY males were more likely to be in forest fragments(14/17), whereas ASY males were more likely to be
found in the continuous forest (27/30; v21 ¼ 24:6; p <0:0001; Table 3). SY males (9/17) also experienced lower
pairing success than ASY males (28/30; v21 ¼10:6; p ¼0:001; Table 3). All three ASY males in forest fragments
were paired, while only one SY male in continuous
forest was unpaired.
3.2. Habitat structure, movements off territories, and
corridor use
Principle component scores I and II explained 64% of
the variance in the data matrix for forest fragmenthabitat. We interpreted PCI to be representative of
canopy height and cover (magnitudes of loading 0.52,
0.49, respectively; all other variables<�0.42) and PCII
to be representative of tree density (loading 0.77; all
other variables<�0.34) for forest fragment habitat.
Canopy height and cover (PCI), tree density (PCII),
edge density, or the amount of forest cover within 1 km
did not influence the probability of a male leaving hisfragment (backward stepwise regression, no variables
remained in model, R2 ¼ 0:00; n ¼ 12; does not include
Mobile males).
Principle component scores I and II explained 65% of
the variance in the data matrix for continuous forest
territories. We interpreted PCI primarily as diameter at
breast height (DBH) (loading 0.61; all other vari-
ables<�0.42) and PCII was most influenced by canopyheight and the number of tree species (loading 0.51,
0.49, respectively; all other variables<�0.47). DBH
(PC I), the number of tree species and canopy height
(PC II), or the amount of forest cover within 1 kin did
not influence the probability of a male leaving his ter-
ritory (backward stepwise regression, no variables re-
mained in model, R2 ¼ 0:00, n ¼ 13).
Paired males in continuous forest resided in forestswith significantly taller canopy and increased total basal
area compared to paired males in fragments (Table 4). In
the two years of our study, we tracked scarlet tanagers in
12 fragments. In four fragments we captured both paired
and unpaired males in both study years. Because both
paired and unpaired males were captured on these frag-
ments,we suggest that there are no special habitat features
related to pairing success. In six fragments, we capturedonly single paired males, and in the last two fragments we
captured only single unpaired males (Table 4).
Table 4
Mean (�SD) habitat characteristics for fragments and continuous forest plots
Continuous forest Forest fragments
Paired (n ¼ 12)a Unpaired (n ¼ 1) Paired (n ¼ 10) Unpaired Sedentary (n ¼ 2)b
Canopy cover (%) 56� 12n:s: 60 50.1� 26.3 64� 31
Canopy height (m) 25.6� 5.2� 35.4 17.7� 7.7 28.4� 6.5
Tree density (trees per ha) 436.3� 100.3 525 454.5� 124.1 377.5� 17.7
Shrub density (shrub stems per ha) 29,343� 22,878n:s: 37,910 46,428� 36,324 28,098� 16,399
Log basal area 1.4� 0.08� 1.6 1.3� 0.1 1.4� 0.17
Edge density (m of edge per ha) NA NA 35.3� 10.4 58.7� 19.2
Fragment size (ha) NA NA 15.3� 24.5 7.5� 3.0
% Forest cover within 1 km 47.6� 3.8 45.5 28.3� 6.8 36.9� 14.2
a t Tests between paired males in fragments and continuous forest.bCharacteristics of capture fragments; n ¼ 3 (instead of n ¼ 4) because on two fragments we captured two males (one Sedentary and one paired) in
opposing study years (see Section 3 for further information).* p < 0:01.
G.S. Fraser, B.J.M. Stutchbury / Biological Conservation 118 (2004) 377–387 383
Four fragments had forest corridors available for
tanagers to use in their movements. Six males (two un-
paired, four paired) were tracked on these fragments
(Table 5). Two males were observed using (or flying inthe direction of) a corridor when leaving the same
fragment (different years). The first male just went into
the corridor and no further. The second male not only
used a corridor to go to an adjacent fragment, but he
also spent 52% of the time tracked in the corridor and
we considered this area part of his territory (Section
3.1.1). All other movements, on both the remaining
fragments with corridors and the eight fragments with-out corridors, during tracking sessions, were across
fields and roads (Table 5).
3.3. Occupancy of fragments
In May of 2000 and 2001, 77% (10/13) and 73% (11/
15) of the fragments surveyed were occupied. Eighty-one
Table 5
Number of males radio-tracked on each fragment and fragment characterist
Fragment Size Closest forest
(m)
No. males radio tagged
(yr 1, yr 2)
Corridor
(no.)a
ST 3.2 110 ð1; 1Þ No
RP 3.2 150 ð1; 0Þ No
FR 3.6 20 ð1; 1Þ No
PC 3.7 200 ð1; 1Þ Yes (1)
AR 3.7 125 ð0; 1Þ No
MY 3.8 130 ð1; 0Þ Yes (1)
CE 4.8 100 ð0; 1Þ No
FL 5.4 100 ð0; 1Þ Yes (1)
LT 8.0 110 ð2; 1Þ No
DC 9.6 325 ð0; 1Þ No
FG 44 100 ð1; 1Þ Yes (1)
GF 75 200 ð0; 1Þ No
aNumber of corridors attached to fragment.bMale went in direction of corridor or was observed using corridor.cMobile males, see text for more information.d This male spent 52% of his time tracked in the corridor.
percent (9/11) of fragments of <10 ha were occupied in
2000 and 83% (10/12) were occupied in 2001.
4. Discussion
Current theory predicts that area-sensitive, or forest-
interior birds, occur more frequently in larger habitat
patches, experience lower pairing success with a decrease
in forest area, and avoid crossing open gaps. Yet scarlet
tanagers in our study made extensive and frequent
movements among fragments without using corridorsand had a high occupancy rate in small forest fragments
<10 ha, observations not supportive of current predic-
tions for area-sensitive species. On the other hand, some
of the patterns we observed were consistent with the
negative effects of fragmentation: the proportion of SY
tanagers was greater in forest fragments than in con-
tinuous forest and males in fragments experienced lower
ics
? Total no. times
tracked off
fragment per male
No. times corridor
in or throughb
No. males in
fragment per
season
2,3 N/A 1
12 N/A 1
1,Mc N/A 1
2,8 l,7d 2,2
0 N/A 1
0 0 1
3 N/A 1+
3 0 1
M,0 N/A 2,1
M N/A 1
0,1 0 3,3
0 N/A 2+
384 G.S. Fraser, B.J.M. Stutchbury / Biological Conservation 118 (2004) 377–387
pairing success. Tanagers in large forests had higher
pairing success and were more likely to be ASY, but SY
males on forest fragments were not restricted in their
movements by the fragmented landscape.
4.1. Movements off fragments
Tanagers� movements were not restricted to their
capture fragment. Movements off fragments seemed to
have a variety of causes including foraging, male-male
chases, and mate searching. Paired males were less likely
to leave their capture fragment than were unpaired
males, which suggests that a primary motivation forunpaired male movements off fragments was to find a
mate. The extreme of this behavior was the Mobile male
strategy. Because there was no difference in the amount
of movement off territories for paired males in fragments
and continuous forest, we conclude that paired males
react similarly to the surrounding landscape indepen-
dent of whether they are in fragments or continuous
forest.Comparisons to other radio-tracking studies are
limited because very few exist for Neotropical migrants
on the temperate breeding grounds (Norris and Stut-
chbury, 2001; Norris et al., 2000 and Bayne and Hob-
son, 2001a). Compared to hooded warblers, paired
tanager males left their fragments less frequently. Norris
and Stutchbury (2001) found that both paired and un-
paired male hooded warblers left their fragments andreturned, as much as once every two hours, and crossed
gaps in search of extra-pair matings through the incu-
bation and nestling stage.
Some tanager movements off fragments were perma-
nent changes in territory location after nest failure. In
the two cases where we were able to follow the birds,
both moved to areas where other tanagers were present
and one moved into continuous forest. Changing terri-tory sites may be an attempt to increase the likelihood of
success of the second nesting attempt, and this behavior
has been noted in other species. For example, prairie
warblers (Dendroica discolor) frequently changed mates
and territories within a breeding season in association
with nest failure (Nolan, 1978).
We discovered a previously undescribed behavior
used by forest birds to search for mates, Mobile males.We found that Mobile males moved and advertised their
position among both fragments and continuous forest in
search of a mate, a form of continuous dispersal
throughout the breeding season. Both mate finding
tactics (i.e., Sedentary and Mobile) sometimes were
successful at attracting females mid-season and this
success was likely related to re-nesting attempts, al-
though only three of the nine unpaired males foundmates. The success of Sedentary males in attracting a
female suggests that females were also moving among
forest fragments. Males that pair later in the breeding
season might still have an opportunity to rear chicks
because we have preliminary evidence that at least some
scarlet tanagers in this area will double brood and re-
nest (unpublished data). Thus, scarlet tanager females
may be reproductively available for longer periods oftime than in single-brooded species. With some level of
double brooding and re-nesting, scarlet tanagers may be
able to maintain viable populations in some fragmented
habitats, as Friesen et al. (1999) concluded for wood
thrush (Hylocichla mustelina) breeding in small forest
fragments in southern Ontario. More work is needed to
understand the role that female tanagers have in nest site
selection in a fragmented landscape, and in mate choiceand retention within and between seasons.
Our study demonstrated variation in male scarlet
tanager behavior was mostly related to pairing status.
Unpaired males were much more likely to leave their
capture fragment and sing in multiple sites, which may
mean that during annual breeding bird surveys (Sauer et
al., 2002) tanager densities may be overestimated.
However, paired males had significantly lower songrates than unpaired males, so they may go undetected
more often during breeding censuses. These two effects
may partially cancel each other out.
4.2. Age structure and pairing success
Pairing success can be influenced by factors such as
age (e.g., Sherry and Holmes, 1989; Bayne and Hobson,2001b) and location of breeding territory (i.e., fragment
or continuous forest). In our study, 83% of scarlet tan-
ager males in forest fragments were SY birds that ex-
perienced lower pairing success compared to males in
the continuous forest plot. Although most males in the
fragmented habitat were unpaired SY males, ASY males
with territories in forest fragments (n ¼ 3) experienced
100% pairing success, suggesting that age is also animportant factor in female mate choice in scarlet tana-
gers. Roberts and Norment (1999) observed lower
pairing success for tanagers in smaller forest fragments
(areas <1000 ha), though pairing success was higher in
their forest plots compared to ours (see below). Differ-
ences in pairing success due to area effects has been
documented in ovenbirds (Villard et al., 1993; Van Horn
et al., 1995), but Villard et al. (1993) suggest that lowpairing success for males in forest fragments may also be
related to higher local population density.
4.3. Habitat structure, movements off capture fragments,
and corridor use
In their extensive study on tanagers, Rosenberg et al.
(1999) found that regional forest cover affects the sen-sitivity of scarlet tanagers to forest fragmentation. In
areas with higher amounts of regional forest cover, such
as the US northeast, tanagers may occur in smaller
G.S. Fraser, B.J.M. Stutchbury / Biological Conservation 118 (2004) 377–387 385
forest fragments, whereas in areas with less regional
forest cover, tanagers may occur only in larger forest
plots (Rosenberg et al., 1999). Our study area has a
moderate amount of hardwood forest cover, and is in
the center of this species� breeding range, a combinationthat may result in high population density and higher
occupation of available habitat (Brown, 1984). In our
study area, not only did we find high occupancy rates of
small forest fragments, but we also tracked tanagers
moving among fragments. Thus, our findings appear to
support Project Tanager�s (Rosenberg et al., 1999)
conclusions on the importance of regional forest cover,
and help explain why scarlet tanager area-sensitivityvaries across regions.
As population density increases SY males may have
less opportunity to establish a territory on preferred
larger forest patches through competitive interactions
with ASY males (see Sherry and Holmes, 1989) and
consequently experience lower pairing success. It follows
that as density decreases the occupation of smaller forest
fragments should also decrease, but the proportion ofpaired SY males should increase (Weins, 1976; Brown,
1984; Villard et al., 1993). The differences we observed in
scarlet tanager pairing success, patch size occupancy,
and territory size between our study and one conducted
less than 150 km away in upstate NY (Roberts and
Norment, 1999) implicates regional population density
as a major factor in determining these differences.
Roberts and Norment (1999) found that scarlet tanagerswere not present in forest fragments of less than 10 ha in
upstate NY, whereas we frequently found scarlet tana-
gers on forest fragments of <10 ha. Roberts and Nor-
ment (1999) also measured higher rates of pairing
success (>75%) on their forest fragments (10–150 ha)
compared to the 59% pairing success we observed for
tanagers in smaller forest patches (3.2–75 ha). Finally,
Roberts and Norment (1999) also reported larger terri-tory sizes compared to those estimated in our study;
however, differences in estimated territory size may also
be due to differences in methodologies.
Male scarlet tanagers did not require the presence of
a corridor to travel among forest fragments (Table 5).
Scarlet tanagers are strong fliers and can cross 100 m
gaps in seconds (personal observation). Therefore, we
conclude that absence of corridors did not alter themovements of adult tanagers, although tanagers may
have moved more frequently if more corridors were
available. While it appears that corridors are not nec-
essary for tanager movement among forest fragments,
this study was not rigorously designed to test corridor
use (see Nicholls and Margules, 1991), nor is it known
whether corridors may be important for other aspects of
tanager reproductive success, such as juvenile dispersal(see Lens and Dhondt, 1994). This study supports the
view of Hunter (1992) and Wiens (1994), who argued
that a long-distance migrant like the scarlet tanager
should not be restricted in their movements in a frag-
mented landscape despite their preference for large
forests.
4.4. Conclusions
Our results show that a species exhibiting an ‘‘area-
sensitive’’ pattern of fragment occupancy (e.g., Roberts
and Norment, 1999) cannot be assumed to be limited to
forest-interior habitat. Based on Project Tanager results
(Rosenberg et al., 1999), and the fact that scarlet tana-
gers are canopy birds that prefer mature forests (Mow-
bray, 1999) we would never have predicted that scarlettanagers would move around extensively once they es-
tablished a territory in a fragment. This kind of empir-
ical work challenges long-held assumptions about how
birds behave in fragmented landscapes, how far they
move, and to what extent gaps represent a barrier to
daily movements. To date, three of three ‘‘area-sensi-
tive’’ birds studied (hooded warbler, ovenbird, and
scarlet tanager; Norris and Stutchbury, 2001; Bayne andHobson, 2001a, this study, respectively) are actually not
area-sensitive in terms of movements. Nonetheless,
species may still be considered area-sensitive if their
probability of occurrence increases with habitat patch
size. However, area-sensitive species should not be as-
sumed to avoid edge habitat, rely on corridors, or avoid
crossing open fields (Villard, 1998).
Many researchers emphasize the importance of coreforest area, or forest interior, when assessing how bird
populations and nesting success are affected by frag-
mentation. We suggest that small fragments are more
important to these forest birds than previously appre-
ciated. In our study area, smaller forest patches located
amongst larger forested areas appeared to allow SY
unpaired males to readily move around and prospect for
breeding sites or mates in different locations and theymay be particularly important in regions with higher
population densities. If regional density can be esti-
mated, managers may be able to make informed deci-
sions about how important smaller fragments are for
tanagers and potentially use tanagers as a model for
other Neotropical migrants (Wiens, 1994). In our study
area, all but one of the fragments was privately owned,
which emphasizes the importance of private land in re-serve networks (Wiens, 1994). Landowners educated
about the importance of these woodlots could choose to
make informed decisions on tree harvesting. For ex-
ample, one landowner in the area harvested two large
trees off 3.7 ha fragment, but removed them in a rela-
tively non-destructive manner (using a horse) leaving the
canopy fairly intact.
Rosenberg et al. (1999) stress caution in extrapolatingresults from one study to the whole species because of
the variation in tanager response to fragmentation
among different regions. Observed variation in breeding
386 G.S. Fraser, B.J.M. Stutchbury / Biological Conservation 118 (2004) 377–387
behavior just within one region (US northeast; Roberts
and Norment, 1999, this study) is further support for
caution in suggesting broad recommendations on habi-
tat management for scarlet tanagers. From this study,
we concluded that many of the reasons for movementsin a fragmented habitat are related to breeding adult
dispersal. However, our study also demonstrates the
need to understand: (1) the array of behavioral re-
sponses to fragmentation in various regions with dif-
ferent levels of forest cover and population densities; (2)
whether smaller forest fragments are reproductive
‘‘sinks’’ rather than ‘‘sources’’ (see Pulliam, 1988); and
(3) how fledgling dispersal may be affected by frag-mented habitat.
Acknowledgements
We thank Hemlock Hill Biological Research Area
and all of the land owners for permission to conduct
research. This project was funded by awards to BJMSfrom the Molson Foundation, the Premier�s Research
Excellence Award, and the Natural Sciences and Engi-
neering Research Council of Canada. Bianca Capuano,
Kevin J. Cooper, Ryan Norris, and Mari Ortwerth all
provided excellent field assistance. A special thanks to
M. Ortwerth and K. Cooper for their hard work.
Thanks to Steve Woolfenden for his help on calculating
territory and fragment sizes, and to Eugene Morton andBonnie Woolfenden for lively tanager discussions. Bev
Brown, Laurence Packer, Scott Robinson, Scott Rush,
and two anonymous reviewers contributed useful com-
ments on an earlier draft.
References
Askins, R.A., Philbrick, M.J., Sugeno, D.S., 1987. Relationship
between the regional abundance of forest and the composition of
forest bird communities. Biological Conservation 39, 129–152.
Bayne, E.M., Hobson, K.A., 2001a. Movement patterns of adult male
ovenbirds during the post-fledging period in fragmented and
forested boreal landscapes. The Condor 103, 343–351.
Bayne, E.M., Hobson, K.A., 2001b. Effects of habitat fragmentation
on pairing success of ovenbirds: importance of male age and floater
behavior. The Auk 118, 380–388.
B�elisle, M., Desrochers, A., Fortin, M.J., 2001. Influence of forest
cover on the movements of forest birds: a homing experiment.
Ecology 82, 1893–1904.
B�elisle, M.A., St. Clair, C.C., 2001. Cumulative effects of barriers on
movements of forest birds. Conservation Ecology 5 (2) [online]
URL:. Available from <http://www.consecol.org/vol5/iss2/art9>.
Bensch, S., 1999. Is the range size of migratory birds constrained by
their migratory program? Journal of Biogeography 26, 1225–1235.
Blake, J.G., Karr, J.R., 1984. Species composition of bird communities
and the conservation of large versus small forests. Biological
Conservation 30, 173–187.
Blake, J.G., Karr, J.R., 1987. Breeding birds of isolated woodlots: area
and habitat relationships. Ecology 68, 1724–1734.
Brown, J.H., 1984. On the relationship between abundance and
distribution of species. American Naturalist 124, 255–279.
Desrochers, A., Hannon, S.J., 1997. Gap crossing decisions by forest
songbirds during the post-fledging period. Conservation Biology 11
(5), 1204–1210.
Freemark, K., Collins, B., 1992. Landscape ecology of birds breeding
in temperate forest fragments. In: Hagan III, J.M., Johnson, D.W.
(Eds.), Ecology and Conservation of Neotropical Migrant Land-
birds. Smithsonian Institute Press, Washington, DC, pp. 443–454.
Friesen, L., Cadman, M.D., MacKay, R.J., 1999. Nesting success of
Neotropical migrant songbirds in a highly fragmented landscape.
Conservation Biology 13, 338–346.
Galli, A.E., Leck, C.F., Forman, R.T.T., 1976. Avian distribution
patterns in forest islands of different sizes in central New Jersey.
The Auk 93, 356–364.
Gibbs, J.P., Faaborg, J., 1990. Estimating the viability of ovenbird and
Kentucky warbler populations in forest fragments. Conservation
Biology 4, 193–196.
Grubb Jr., T.C., Doherty Jr., P.F., 1999. On home-range gap-crossing.
The Auk 116, 618–628.
Hunter, M.L.J., 1992. Paleoecology, landscape ecology, and conser-
vation of neotropical migrant passerines in boreal forests. In:
Hagan III, J.M., Johnson, D.W. (Eds.), Ecology and Conservation
of Neotropical Migrant Landbirds. Smithsonian Institute Press,
Washington, DC, pp. 511–523.
James, F.C., Shugart Jr., H.H., 1970. A quantitative method of habitat
description. Audubon Field Notes 24, 727–735.
Lee, M., Fahrig, L., Freemark, K., Currie, D.J., 2002. Importance of
patch scale vs landscape scale on selected forest birds. Oikos 96,
110–118.
Lens, L., Dhondt, A.A., 1994. Effects of habitat fragmentation on the
timing of crested titParus cristatus natal dispersal. Ibis 136, 147–152.
Lynch, J.F, Whigham, D.F., 1984. Effects of forest fragmentation on
breeding bird communities in Maryland, USA. Biological Conser-
vation 28, 287–324.
McGarigal, K., Cushman, S., Stafford, S., 2000. Multivariate statistics
for wildlife and ecology research. Springer, New York.
Mowbray, T.B., 1999. Scarlet tanager (Piranga olivacea). In: Poole,
A., Gill, F. (Eds.), The Birds of North America, vol. 479. The Birds
of North America, Inc, PA.
Nicholls, A.O., Margules, C.R., 1991. The design of studies to
demonstrate the biological importance of corridors. In: Saunders,
D.A., Hobbs, R.J. (Eds.), Nature Conservation 2: The Role of
Corridors. Surrey Beaty and Sons, Chipping Norton, Australia.
Nolan Jr., C., 1978. The ecology and behavior of the prairie warbler
Dendroica discolor. Ornithological Monograph 26.
Norris, D.R., Stutchbury, B.J.M., 2001. Effects of habitat isolation on
the extra-territorial movements of a Neotropical forest songbird.
Conservation Biology 15, 729–736.
Norris, D.R., Stutchbury, B.J.M., Pitcher, T.E., 2000. The spatial
response of male hooded warblers to edges in isolated fragments.
The Condor 102, 595–600.
Porneluzi, P., Bednarz, J.C., Goodrich, L.J., Zawada, N., Hoover, J.,
1993. Reproductive-performance of territorial ovenbirds occupying
forest fragments and a contiguous forest in Pennsylvania. Conser-
vation Biology 7, 618–622.
Powell, L.A., Lang, J.D., Conroy, M.J., Krementz, D.G., 2000. Effects
of forest management on density, survival, and population growth
of wood thrushes. Journal of Wildlife Management 64, 11–23.
Pulliam, H.R., 1988. Sources, sinks, and population regulation.
American Naturalist 132, 652–661.
Rappole, J.H., Tipton, A.R., 1991. New harness design for attachment
of radio transmitters to small passerines. Journal of Field
Ornithology 62, 335–337.
Robbins, C.S., Dawson, D.K., Dowell, B.A., 1989. Habitat area
requirements of breeding forest birds of the middle Atlantic states.
Wildlife Monographs No. 103.
G.S. Fraser, B.J.M. Stutchbury / Biological Conservation 118 (2004) 377–387 387
Roberts, C., Norment, C.J., 1999. Effects of plot size and habitat
characteristics on breeding success of scarlet tanagers. The Auk
116, 73–82.
Robinson, S.K., Thompson III, F.R., Donovan, T.M., White-
head, D.R., Faaborg, J., 1995. Regional forest fragmentation
and the nesting success of migratory birds. Science 267, 1987–
1990.
Rosenberg, K.V., Lowe, J.D., Dhondt, A.A., 1999. Effects of forest
fragmentation on breeding tanagers: a continental perspective.
Conservation Biology 13, 568–583.
Sauer, J.R., Hines, J.E., Fallen, J., 2002. The North American
breeding bird survey, results and analysis 1966–2001, Version
2002.1. USGS Patuxent Wildlife Research Center, Laurel, MD.
Sherry, T.W., Holmes, R.T., 1989. Age-specific social dominance
affects habitat use by breeding American redstarts (Setophaga
ruticilla): A removal experiment. Behavioral Ecology and Socio-
biology 25, 327–333.
Sieving, K.E., Willson, M.F., De Santo, T.L., 1996. Habitat barriers to
movement of understory birds in fragmented south-temperate
rainforest. The Auk, 944–949.
Smith, T.M., Shugart, H.H., 1987. Territory size variation in the
ovenbird: the role of habitat structure. Ecology 68 (3), 695–704.
Van Horn, M.A., Gentry, R.M., Faaborg, J., 1995. Patterns of
ovenbird (Seiurus aurocapillus) pairing success in Missouri forest
tracts. The Auk 112, 98–106.
Villard, M.A., 1998. On forest-interior species, edge avoidance, area
sensitivity, and dogmas in avian conservation. The Auk 115, 801–
805.
Villard, M.A., Martin, P.R., Drummond, C.G., 1993. Habitat frag-
mentation and pairing success in the ovenbird (Seiurus aurocapil-
lus). The Auk 110, 759–768.
Weins, J.A., 1976. Population responses to patchy environments.
Annual Review of Ecology and Systematics 7, 81–120.
Wiens, J.A., 1994. Habitat fragmentation: island versus landscape
perspectives on bird conservation. Ibis 137, S97–SI04.
Zack, S., Stutchbury, B.J.M., 1992. Delayed breeding in avian social
systems: the role of territory quality and ‘‘floater’’ tactics. Behav-
iour 123, 194–219.
Zar, J.H., 1996. Biostatistical Analysis. Prentice-Hall, Englewood
Cliffs, NJ.