Content uploaded by Yu Xu
Author content
All content in this area was uploaded by Yu Xu on Mar 16, 2014
Content may be subject to copyright.
Roosting Behavior and Roost Selection by Buff-throated Partridges
Tetraophasis szechenyii during the Breeding Season
Yu Xu1, Nan Yang1, Ying Wang2, Bi-Song Yue1, and Jiang-Hong Ran1,*
1College of Life Sciences, Sichuan University, Chengdu 610064, China
2Department of Life Science, National Taiwan Normal University, Taipei 106, Taiwan
(Accepted December 4, 2009)
Yu Xu, Nan Yang, Ying Wang, Bi-Song Yue, and Jiang-Hong Ran (2010) Roosting behavior and roost
selection by Buff-throated Partridges Tetraophasis szechenyii during the breeding season. Zoological Studies
49(4): 461-469. Studying roosting behavior and roost selection is very important for Galliformes conservation.
The breeding season roosting behavior and roost selection by Buff-throated Partridges Tetraophasis szechenyii,
a rare galliform species endemic to western China, were studied by radio tracking and direct sighting in
2006-2008. Partridges roosted in family groups, and group members almost always huddled together within
roosts except in the incubation period and the 1st 2 wks after the eggs hatched, when males remained separate
from females. The vegetation they selected for roosts was almost exclusively r forests with high tree density,
good canopy cover, and plenty of trees with large diameter at breast height (DBH). Trees typically used were
large-DBH (27.3 ± 1.1 cm S.E., n = 58) Abies squamata or Larix potaninii situated on sites with large degree
of slope, long distance to the nearest trail, and close proximity to other large-DBH trees. Branches occupied
were frequently (47.2%, n = 74) easterly oriented, and 7.4 ± 0.1 m S.E. (n = 95) in height and 4.2 ± 0.1 cm
S.E. (n = 95) in diameter. This suggests that the roosting behavior and patterns of roost selection by Buff-
throated Partridges are principally to maximize thermoregulatory benets and avoid predation. The relevance
for management of this species is mainly discussed in terms of forestry practices and exploitation restraints.
http://zoolstud.sinica.edu.tw/Journals/49.4/461.pdf
Key words: Group roosting, Behavior, Habitat selection, Buff-throated Partridge, Galliformes.
*To whom correspondence and reprint requests should be addressed. E-mail:rjhong-01@163.com
Night roosts play a crucial role in bird
biology because of the large amount of time birds
spend roosting during the night and their presumed
vulnerability to inclement weather and predation
when asleep (Cody 1985, Woltmann 2004).
Roosting behavior and roost selection are likely
to be important determinants of individual fitness
(Fisher et al. 2004), with particular consequences
for energy budgets and predator avoidance
(Walsberg 1983).
This is particularly true for the Galliformes,
an avian group characterized by a poor dispersal
ability and limited spatial movements. To our
knowledge, many galliform species show group
roosting behavior even in the breeding season,
which has been hypothesized to have benefits
in terms of reduced thermoregulatory costs,
decreased predation risk, and increased foraging
efficiency (Eiserer 1984, Beauchamp 1999);
and their selection of roosts is not random:
they prefer places that can provide benefits for
thermoregulation and predation avoidance. For
example, forest-dwelling species (e.g., Syrmaticus
ellioti, Ding et al. 2002; S. reevesii, Sun et al.
2002; Crossoptilon crossoptilon, Jia et al. 2005; S.
humiae, Jiang et al. 2006) usually roost in large-
diameter trees, and sites where roosting trees
occur have high tree densities (e.g., Sun et al.
2002, Thompson 2003, Jiang et al. 2006), high
percentages of canopy cover (e.g., Rumble 1992,
Zoological Studies 49(4): 461-469 (2010)
461
Ding et al. 2002, Lu and Zheng 2002, Thompson
2003, Jia et al. 2005, Sweringin 2007), great
amounts of large-diameter trees (e.g., Rumble
1992, Thompson 2003, Sweringin 2007), and
steep slopes (e.g., Liang 2000, Ding et al. 2002,
Sun et al. 2002, Jia et al. 2005, Jiang et al. 2006).
Acquiring knowledge of the roosting behavior and
roost selection by the Galliformes will undoubtedly
aid in the management and conservation of these
species (Fisher et al. 2004).
The Buff-throated Partridge Tet raophasis
szechenyii is a galliform species endemic to
western China. It is considered to be vulnerable
according to the Red Book of China (Wang and Xie
2004), and is legally listed as a category I species
of nationally protected animals in China, because
it is threatened by habitat loss and degradation,
and a reportedly high rate of hunting (Mackinnon
et al. 2000, Klaus et al. 2003, Li 2004, Wang and
Xie 2004). As this species inhabits coniferous
forests, alpine shrublands, and tundra above the
tree line at 3,350- 4,600 m in elevation (Mackinnon
et al. 2000), where environmental factors can be
severe (e.g., low ambient temperatures), predator
pressure can be high, and individuals spend a
large amount of time roosting during the night (Xu
et al. 2008), their roosting behavior is probably
unique, and roost selection is likely very strictly
controlled. However, to date, nothing is known
about these aspects of this species. We therefore
examined their night roosting behavior and roost
selection during the breeding season. The major
objectives were to (1) determine whether group or
communal roosting occurs in this species and if
there are any temporal changes, and (2) identify
habitat factors associated with the roost selection
by this species.
MATERIALS AND METHODS
Study area
The study area was located in Pamuling
Mountain (30°06'N, 101°11'E), Yajiang County,
Ganzi Tibetan Autonomous Prefecture, Sichuan,
China (Fig. 1). The area of the study site was
about 340 ha, and the elevation ranged 3,900-
4,200 m. The climate is the sub-humid type of the
Qinghai-Tibetan plateau. The mean temperatures
at sunrise (sunset) were 2.7 (7.0) and 4.7 (8.4)°C
in Apr.-May and June-Oct., respectively.
Fieldwork
Buff-throated Partridges lived in groups in
the study area, and 15 groups were observed in
Apr.-Oct. 2006 (n = 5) and Apr.-May 2007 (n = 8)
and 2008 (n = 2). Eight birds from 3 groups
were captured and fitted with colored leg bands,
and 2 adult males were equipped with necklace-
transmitters (model RI-2D, weight 18.5 g,
frequency 164-165.999 MHz, lifespan 36 mos;
Holohil Systems, Carp, Canada). But other groups
(n = 12) could be recognized by their exclusive
territories and the unique plumage differences of
individual birds. Groups were located by radio
tracking or direct sighting (for more details see Xu
et al. 2008). Once a group was located, it was
followed without disturbing it (because a monastery
was located on the mountain, the partridges were
particularly tame and they could be approached
within 10 m without affecting their behavior), and
its roosting site was observed until all members
had entered the roost, at which point a visual
location was obtained. During this process, the
general behavior of the partridges was recorded
(e.g., spatial distance between members, and
roosting height). The roosting site was flagged,
and the coordinates were marked with a GPS
(model Etrex Venture; Garmin, Taipei, Taiwan),
after which we immediately left. The next morning,
we revisited the roosting site 30 min before sunrise
to determine whether the roost was abandoned
by the group or not. Habitat characteristics were
measured after the partridges had departed.
Habitat description
We determined the roosting habitat of Buff-
throated Partridges from macro- and microhabiat
scales. In the study, macrohabitats were identied
by distinct changes in vegetation type and were
described based on the dominant species of
vegetation (Rumble 1992). Four macrohabitat
types were discriminated in the study area: (1)
fir forests (FF), dominated by Abies squamata
and Larix potaninii (the proportion of occurrence
between these 2 tree species was 2.19: 1, n = 58),
which occurred on northern slopes, (2) oak thickets
(OT), dominated by Quercus aquifolioides, which
occurred on southern slopes, (3) rhododendron
shrubs (RS), dominated by Rhododendron
nitidulum, Salix spp., and Dasiphora fruticosa,
which occurred on northern slopes, and (4)
alpine meadows (AM), dominated by Kobresia
setchwanensis, Polygonum viviparum, Potentilla
Zoological Studies 49(4): 461-469 (2010)
462
discolor, and Hemiphragma heterophyllum, which
occurred on the plateau of the study area and were
generally at (Table 1).
Microhabitats were evaluated at 2 scales
(roosting tree and roosting branch). On the
roosting tree scale, habitat variables of roosting
trees measured included diameter at breast
height (DBH), distance to the nearest tree, DBH
of the nearest tree, distance to the nearest trail,
distance to the nearest forest edge, and degree
of slope (Table 1); surrounding each roosting
tree, habitat variables, including tree density, tree
cover, average DBH of trees, and shrub cover
(Table 1), were measured by employing 10 × l0 m
plots (0.01 ha) with the roosting tree at the center.
Elevation and aspect of slope were not included
in the measurement, because there were no
obvious variations in them due to the range of the
study area and the method of choosing random
trees (see below). At the roosting branch scale,
habitat variables measured included roosting
branch height and diameter (of the trunk end) and
orientation (recorded as 4 categories: 316°-360°N
and 0°-45°N; 46°-135°E; 136°-225°S; and
226°-315°W). Diameter variables were measured
with calipers; physiographical variables were
measured with a GPS and a clinometer; distance
variables were measured with a tape measure; and
cover variables were visually estimated following
Prodon and Lebreton (1981).
Simply documenting habitat characteristics
in and around roosting trees does not allow for
an explicit statement of how these trees differ
from other available ones (Fisher et al. 2004). To
quantitatively assess the roost selection by the
birds, we chose random trees for comparison with
roosting trees. A random tree (if there was no tree
at the location, the nearest tree to the location
was viewed as the random tree) was chosen in
a randomly determined direction for a distance
Fig. 1. Map showing the study area (30°06'N, 101°11'E) with roost locations.
N
Roost locations
Monastery
km10.500.5
4000 m
4100 m
4100 m 3900 m
3700 m
Study area
Sichuan Province
Xu et al. – Group Roosts of Buff-throated Partridges 463
of 100 m from the roosting tree (e.g., Thompson
2003, Jiang et al. 2006), but still within the general
macrohabitat boundaries in order to insure that key
influences on habitat selection were not missed
(Johnson 1980, Orians and Wittenberger 1991,
Jones 2001). The same measurements were
taken as those at the roosting tree (Table 1).
Data analysis
As the roosting behavior by Buff-throated
Partridges varied with periods of breeding (see
“RESULTS”), we used the following categories
in the analyses: incubation (about 23 ds), early
post-hatching (the 1st 2 wks of chicks’ lives),
and other breeding periods (including the pairing
period and the mid-to-late post-hatching period up
to when chicks turned 5 mo old). The statistical
analyses used included Chi-square tests, principal
component (PC) analysis (PCA), and generalized
linear mixed models (GLMMs). Chi-square tests
were used to determine whether roosting tree
species were selected disproportionately from
expected use, and whether the selection of the
orientation of roosting branches differed from
a random distribution (SPSS vers. 11.0, SPSS
2001). The use of macrohabitats (i.e., vegetation
types) was not examined by Chi-square tests,
because the use was almost exclusive (Table
2). GLMMs were used to analyze the probability
of finding a roosting tree, by examining habitat
differences between roosting trees and random
trees. But before beginning the models, to obtain
habitat gradients and summarize patterns of
covariance in habitat variables (Iamsiri and Gale
2008), a PCA on the correlation matrix was first
performed in SPSS vers. 11.0 (SPSS 2001);
note that habitat variables were transformed for
normality prior to the PCA using log and root-
square transformations. Then the GLMMs were
implemented using PROC GLIMMIX in SAS vers.
9.1.3 (SAS Institute 2005), with the binomial error
Table 1. Descriptive statistics of habitat variables measured at the roosting tree scale of Buff-throated
Partridges on Pamuling Mountain, Sichuan, China, 2006-2008a
Habitat variable
Roost (n = 58) Random (n = 58)
Mean ± S.E. Range Mean ± S.E. Range
Degree of slope (°) 39.5 ± 0.4 29.0 - 47.0 31.4 ± 1.2 10.0 - 43.0
Distance to the nearest trail (m) 30.8 ± 2.4 9.0 - 100.0 21.3 ± 2.3 1.1 - 75.0
Distance to the nearest forest edge (m) 33.4 ± 4.0 0 - 120.0 42.5 ± 5.4 0 - 130.0
Tree density (stems/m2) 0.15 ± 0.01 0.04 - 0.56 0.13 ± 0.02 0.01 - 0.75
Average DBH of trees (cm) 16.8 ± 0.6 7.5 - 31.0 15.3 ± 0.5 8.0 - 31.0
Tree cover (%) 32.2 ± 1.5 18.0 - 80.0 32.1 ± 1.8 15.0 - 86.0
Shrub cover (%) 40.0 ± 1.7 13.0 - 65.0 50.3 ± 2.3 10.0 - 85.0
Roosting tree DBH (cm) 27.3 ± 1.1 15.0 - 60.0 20.2 ± 0.7 15.0 - 40.0
The nearest tree DBH (cm) 15.0 ± 0.8 6.0 - 32.0 15.2 ± 0.8 7.0 - 32.0
Distance to the nearest tree (m) 1.3 ± 0.1 0.1 - 2.9 2.3 ± 0.2 0.1 - 10.0
aAll habitat variables but “Distance to the nearest tree” were measured within the 10 × 10 m plot. DBH, diameter at breast height; S.E.,
standard error.
Table 2. Selection of macrohabitats for roosting by Buff-throated Partridges on Pamuling Mountain,
Sichuan, China, 2006-2008
Macrohabitat
(percentage)
Incubation period
(n = 5 groups)
Early post-hatching period
(n = 4 groups)
Other breeding periods
(n = 13 groups)
Female Male Female with chicks Male
FF (38.8%) 26 26 8 10 57
OT (49.5%) 0 0 2 0 1
RS (10.5%) 0 0 0 0 0
AM (1.2%) 0 0 0 0 0
FF, r forests; OT, oak thickets; RS, rhododendron shrubs; AM, alpine meadows.
Zoological Studies 49(4): 461-469 (2010)
464
and logit link function; scores of habitat variables
extracted from the PCA were included as factors
in the models, and family group was used to a
random factor. In the analysis, p < 0.05 was
interpreted as being statistically significant, and
means are presented as ± standard error (S.E.),
unless otherwise stated.
RESULTS
Roosting behavior
Buff-throated Partridges were cooperative
breeders in the study area, because partridge
groups were frequently comprised of 1 female and
multiple males (the mean size of groups excluding
broods was 2.67 ± 0.16 birds, n = 15), and each
group only had a single set of brood if it had any.
Once a group formed and occupied an exclusive
territory, the members foraged together during the
daytime. At night, they huddled together on the
same branch (height = 7.2 ± 0.2 m, diameter =
4.2 ± 0.1 cm, n = 95 observations from 13 groups)
of Abies squamata or Larix potaninii (DBH = 27.3
± 1.1 cm, n = 58 trees used by 13 groups), and
individuals normally remained close to the tree
trunk and were perpendicular to the branch 100%
of the time, with their heads pointing downslope.
Compared to a uniform distribution, the branches
used were more frequently pointed in an easterly
direction (χ2 = 31.061, d.f. = 3, p < 0.001, n = 74
observations from 13 groups, Fig. 2).
However, when females began to incubate
eggs, males did not accompany them, and at night,
the distance between males and females was 24.8
± 2.4 m (range = 7.0-55.0 m, n = 26 observations
from 5 groups). If eggs successfully hatched,
females (with chicks) were joined by males during
the day, but at night they were not accompanied
by males until chicks were > 15 d old. Within this
2 wk period, because it was difficult for chicks to
y up into the tree, females brooded chicks under
dense undergrowth (e.g., dense rhododendron
shrubs, logs, and stumps, but not at old nest sites),
or in Q. aquifolioides or Rh. calophytum thickets
where the perching height was about 2 m above
the ground, while males traveled 14.0 ± 2.4 m
(range = 6.0-30.0 m, n = 10 observations from 4
groups) to return to their permanent roosting trees.
Roost selection
In total, 110 roosts were found (Fig. 1), some
of which were repeatedly used. If repeatedly used
roosts were not distinguished among the 3 periods
of breeding, there would be 130 roosts, of which
52 were used during the incubation period (all of
which were located in fir forests), 20 in the early
post-hatching period (18 of which were located in
r forests), and 58 in the other breeding periods (57
of which were located in fir forests) (Table 2). It
seemed that regardless of the period, Buff-throated
Partridges almost exclusively used r forests.
Of 58 roosting trees used during the other
breeding periods, 41 were A. squamata and 17
were L. potaninii. The proportional use of these
2 tree species by Buff-throated Partridges did
not differ from the expected (χ2 = 0.081, d.f. = 1,
p = 0.777). The PCA of habitat variables yielded
4 components which collectively accounted for
67.73% of the total variance (Table 3). The 1st
component (PC1) represented tree and shrub
communities of the plots, with heavy loading
by tree density (positive score), tree cover
(positive score), average DBH of trees (negative
score), and shrub cover (negative score). The
2nd (PC2) and 3rd components (PC3) mostly
reected physiographical characteristics or spatial
positions of the target trees, with heavy loading by
degree of slope (positive score), distance to the
nearest forest edge (positive score), distance to
the nearest trail (positive score), distance to the
nearest tree (negative score), and the nearest tree
DBH (positive score). The 4th component (PC4)
was heavily loaded by roosting tree DBH (positive
score), which defined vegetation features of the
target trees.
The GLMMs showed that the probability
of Buff-throated Pa rtridges using a roost tr ee
increased along PC2, PC3, and PC4 (Table
4), indicating that trees which were suitable for
roosting by Buff-throated Partridges were large-
DBH trees located in sites with large degree of
slope, long distances to the nearest forest edge
Frequency (%)
50
40
30
20
10
0
East South West North
Orientation
Fig. 2. Orientations (n = 74) of roost branches of Buff-throated
Partridges on Pamuling Mountain, Sichuan, China, 2006-2008.
Xu et al. – Group Roosts of Buff-throated Partridges 465
(but this was contrary to our measurements, Table
1) and the nearest trail, and close to other large-
DBH trees. Other components failed to enter the
models.
DISCUSSION
Roosting behavior
Group m embers huddle d toge ther on t he
same branch when roosting. This is not known
in other galliform species that roost in groups
during the breeding season, the group members
of which roost in neighboring trees (e.g., Ithaginis
cruentus, Jia et al. 1999; A. rufipectus, Liao et
al. 2008), or different branches even in the same
tree (e.g., Chrysolophus amherstiae, Kang and
Zheng 2007; C. harmani, Lu and Zheng 2007).
Herein, we present 2 alternative, but not mutually
exclusive, reasons for huddled roosting by Buff-
throated Partridges: cooperative breeding and
thermoregulatory benets.
In our study area, Buff-throated Partridges
wer e coo pera t ive breed ers. Sku tch (1989 )
suggested that in cooperatively breeding birds,
group members might be particularly prone to
huddle together when roosting. The mechanisms
for this, although not yet determined, are possibly
asso c i a t ed with th e sp e cific li f e hi story of
cooperatively breeding birds. Yet, further research
is needed, because we found that I. cruentus was
a cooperative breeder in the study area, but it was
uncommon for this species to huddle together
when roosting.
On cold days, diurnal birds can choose
Table 3. Factor loading, total and cumulative percent variance explained by the principal component
analysis (Varimax normalized rotation) of habitat characteristics measured at the roosting tree scale of Buff-
throated Partridges (n = 58). Factor loadings with absolute values ≥ 0.56 are shown in bold, based on our
search for the highest absolute value for each factor
Habitat variable
Component
1 2 3 4
Degree of slope (°) -0.029 0.810 0.051 0.155
Distance to the nearest trail (m) 0.267 0.241 0.636 -0.040
Distance to the nearest forest edge (m) 0.137 0.608 -0.202 -0.262
Tree density (stems/m2)0.855 0.303 0.151 0.136
Average DBH of trees (cm) -0.635 -0.023 0.523 0.343
Tree cover (%) 0.936 0.034 -0.113 0.020
Shrub cover (%) -0.563 -0.068 -0.053 -0.541
Roosting tree DBH (cm) 0.009 0.090 -0.058 0.867
The nearest tree DBH (cm) -0.216 -0.136 0.803 -0.032
Distance to the nearest tree (m) -0.185 -0.570 -0.161 -0.183
Eigenvalue 2.865 1.698 1.154 1.055
Percent (%) of variance 28.654 16.984 11.541 10.548
Cumulative percent (%) 28.654 45.637 57.178 67.726
DBH, diameter at breast height.
Table 4. Generalized linear mixed models (binomial error and logit link function) analyzing the probability of
nding a Buff-throated Partridge roosting tree along 4 habitat gradients
Source Estimate ± S.E. d.f. χ2p
Intercept -0.078 0.270
PC1 0.326 0.291 1, 99 1.26 0.265
PC2 1.752 0.399 1, 99 19.32 < 0.001
PC3 1.038 0.344 1, 99 9.12 0.003
PC4 2.163 0.439 1, 99 24.24 < 0.001
PC, principal component; S.E., standard error; d.f., degree of freedom.
Zoological Studies 49(4): 461-469 (2010)
466
between reducing their activity to minimize heat
loss and increasing it in the hope of generating
higher heat production (Morse 1980, Doucette and
Reebs 1994). Species that use the 1st strategy
would have to adopt a particular roosting behavior
(Morse 1980). In our study area, Buff-throated
Partridges chose the 1st strategy when coping with
cold days (Xu et al. 2008); thus, huddling while
roosting may be a particular behavior creating a
microclimate that ameliorates energy losses.
However, males did not roost close to
incubating females. This behavior was also
reported in other forest-dwelling galliform species,
e.g., I. cruentus (Jia et al. 1999), C. harmani (Lu
1997), and A. rupectus (Liao et al. 2008). Liao et
al. (2008) suggested this was a strategy to reduce
the possibility of nest predation. After all, huddling
behavior may lead to greater conspicuousness
of aggregating individuals to predators, and it
would make a nest susceptible to predation,
although adults can better protect them via mutual
defense against predators (Alcock 2005). If eggs
successfully hatched, male partridges’ roosts were
still not close to brooding females when chicks
were < 15 d old, and this differs from I. cruentus (Jia
et al. 1999) and A. rufipectus (Liao et al. 2008),
in which males roosted in places very close to
brooding females.
Roost selection
Buff-throated Partridges almost exclusively
used r forests, since the forests can provide high
tree density, good canopy cover, and plenty of
large-DBH trees (pers. observ.). Not only can fir
forests provide Buff-throated Partridges protection
from avian predators (e.g., Gyps himalayensis
and Strix aluco), but more importantly provide
thermoregulatory benets in terms of reduced wind
velocities (Eiserer 1984, Cody 1985) and reduced
loss of long-wave radiation (Campbell and Norman
1998). Moisture-heat conditions in fir forests
should also be better than those of other available
macrohabitats, and decreases in air temperature
are expected to be relatively slow at night.
Yet, in our study, high tree density, good
canopy cover, and large average-DBH of trees
did not successfully predict the probability of Buff-
throated Partridges using a roosting tree as in
other studies of forest-dwelling galliform species
(see “Introduction”). As an explanation, we
suggest that this is because both roosting trees
and random trees occurred in areas with similar
tree density, canopy cover, and average DBH of
trees. It is possible that the sampling method,
for which random trees were selected to be
100 m from the roosting trees, caused all forest
measurements to be somewhat unbiased.
Nevertheless, trees in which Buff-throated
Partridges roosted were predicted to have a large
DBH and be located in sites with large degree of
slope, long distance to the nearest trail, and close
proximity to other large-DBH trees. Roosting
trees had large DBHs and were located in sites
with large degree of slope; these results are in
agreement with the general view of forest-dwelling
galliform species (see “Introduction”). Large-DBH
trees provide spread-out canopies containing
many nearly horizontal limbs for perching, also
more easily allow birds to isolate themselves
from ground-dwelling predators, and provide a
better vantage point for predator detection prior to
leaving the roost in the morning (Sweringin 2007).
As members of the Galliformes are poor fliers,
the choice of steep slopes makes it easier for
individuals to escape threats and y farther. As a
matter of fact, it is not surprising that Buff-throated
Partridges always roosted facing downslope,
and often glided in a downslope direction when
suddenly disturbed.
Few studies documented that the distance
to the nearest trail was a significant predictor
of roosting trees. Perhaps, for Buff-throated
Partridges, a long distance to a trail is a result of a
long-term process of natural selection of habitat,
as in the past this species was highly hunted
especially at night when it is easy for hunters to
catch more individuals each time, due to their
specic huddled roosting behavior.
On trees, groups often used easterly pointed
branches, which allows individuals to receive the
earliest morning light (Boeker and Scott 1969),
and provides protection against the prevailing
wind from the west (Rumble 1992). However, it
is also possible that tree characteristics affect the
directional selection of roosting branches, e.g.,
there may be thicker branches with an easterly
orientation. Unfortunately, we have no data to
indicate this was the case.
Relevance for management
We reported huddled roosting by Buff-throated
Partridges during the breeding season, and
suggested 2 possible mechanisms (cooperative
breeding and thermoregulatory benefits) for this
rare behavior in the Galliformes, which can be
attributed to protection from harsh environmental
Xu et al. – Group Roosts of Buff-throated Partridges 467
conditions. This rare behavior itself is of interest
and worthy of conservation (Sutherland 1998).
Our study also showed that Buff-throated
Partridges obviously selected roosting habitat,
which has important consequences for thermal
protection and predator avoidance. We
acknowledge that the small sample size may limit
the ability to generalize our conclusions. Further
investigations, using more samples would be
useful. Furthermore, it is suggested that random
trees should be sampled further away (but they
should still be constrained by the boundaries of
the general macrohabitat and the territory), to try
to make all forest measurements biased between
sites where roosting and random trees occur.
We sug gest a h abitat man ageme nt plan
for the Buff-throated Partridge, based on the
present results on roosting habitat selection. The
government has been implementing reforestation
programs within this species’ distribution range,
since commercial deforestation was prohibited
with full implementation of a ban on logging in the
upper Yangtze basin in 1998. However, to our
knowledge, the selection of tree species used for
reforestation is often made by considering the ease
of growth, but lacking any ecological guidelines.
In many areas of this species’ distribution, only
1 or 2 tree species (e.g., Pinus densata) were
planted. We suggest that forestry practices should
include replanting fir forests preferred by Buff-
throated Partridges, by planting Abies squamata
and Larix potaninii. Furthermore, the exploitation
of r forests needs to be restrained; in particular,
the logging of large-DBH A. s quamata and L.
potaninii located on sites with large degree of
slope and long distance to the nearest trail should
be avoided. The cutting of dense rhododendron
shrubs or Quercus aquifolioides and Rhodoendron
calophytum thickets should also be discouraged,
as they are also important to the survival of
females and chicks within the 1st 2 wks after
hatching.
Acknowledgments: We are grateful to the
Forestry Bureau of Yajiang County for their
support, to the Pamuling Monastery for
accommodations, and to N. Baima, R. Gongbu,
W. Li, P. Que, and X. Zhang for their assistance
with fieldwork. We are indebted to B. Du and K.
Zhang for their graphic assistance, and to E. King,
S. Klaus, P. McGowan, and 2 anonymous referees
for their critical comments which improved this
manuscript. This study was supported by the
National Key Technology R&D Program of China
(grant 2008BADB0B04).
REFERENCES
Alcock J, ed. 2005. Animal behavior: an evolutionary
approach. Sunderland, MA: Sinauer Associates.
Beauchamp G. 1999. The evolution of communal roosting
in birds: origin and secondary losses. Behav. Ecol. 10:
675-687.
Boeker EL, VE Scott. 1969. Roost tree characteristics for
Merriam’s Turkey. J. Wildlife Manage. 33: 121-124.
Campbell GS, JM Norman, ed. 1998. An introduction to
environmental biophysics. New York: Springer-Verlag.
Cody ML, ed. 1985. Habitat selection in birds. Orlando, FL:
Academic Press.
Ding P, Y Yang, Z Li, S Jiang, Y Zhuge. 2002. Studies on the
selection of roosting sites of Elliot’s Pheasant. J. Zhejiang
Univ. (Sci. Ed.) 29: 564-568. (in Chinese with English
summary)
Doucette DR, SG Reebs. 1994. Influence of temperature
and other factors on the daily roosting times of Mourning
Doves in winter. Can. J. Zool. 72: 1287-1290.
Eiserer LA. 1984. Communal roosting in birds. Bird Behav. 5:
61-80.
Fisher RJ, QE Fletcher, CKR Willis, RM Brigham. 2004.
Roosting selection and roosting behavior of male
Common Nighthawks. Am. Midl. Nat. 151: 79-87.
Iamsiri A, GA Gale. 2008. Breeding season habitat use by
Hume’s Pheasant Syrmaticus humiae in the Doi Chiang
Dao Wildlife Sanctuary, northern Thailand. Zool. Stud.
47: 138-145.
Jia C, G Zheng, X Zhou, H Zhang. 1999. Social organization
of Blood Pheasant (Ithaginis cruentus) in Wolong Nature
Reserve. Acta Zool. Sin. 45: 135-142. (in Chinese with
English summary)
Jia F, N Wang, G Zheng. 2005. Analysis of roosting habitat
characteristics of wintering Crossoptilion crossoptlion
populations. Chin. J. Ecol. 24: 153-158. (in Chinese with
English summary)
Jiang A, F Zhou, Z Lu, X Han, R Sun, X Li. 2006. Roost-site
selection of Mrs. Hume’s Pheasant (Syrmaticus humiae)
in Guangxi, China. Zool. Res. 27: 249-254. (in Chinese
with English summary)
Johnson DH. 1980. The comparison of usage and availability
measurements for evaluating resource preference.
Ecology 61: 65-71.
Jones J. 2001. Habitat selection studies in avian ecology: a
critical review. Auk 118: 557-562.
Kang M, G Zheng. 2007. Roost-site selection of Lady
Amherst’s Pheasant. Acta Ecol. Sin. 27: 2929-2934. (in
Chinese with English summary)
Klaus S, R Ploner, Y Sun, Y Fang. 2003. The Chestnut-
throated Partridge (Tetraophasis obscurus) in the
Lianhuashan Nature Reserve, Gansu, China: ecological
observations and taxonomic questions. J. Ornithol. 144:
197-200.
Li X, ed. 2004. Game birds of China. Beijing: China Forestry
Publishing House. (in Chinese)
Liang W. 2000. The ecological adaptation and conservation of
Golden Pheasant Chrysolophus pictus. PhD dissertation,
Beijing Normal Univ., Beijing, China. (in Chinese with
English summary)
Zoological Studies 49(4): 461-469 (2010)
468
Liao W, J Hu, C Li, X Lu. 2008. Roosting behaviour of the
endangered Sichuan Hill-Partridge Arborophila rupectus
during the breeding season. Bird Conserv. Int. 18:
260-266.
Lu X. 1997. Study on habitat selection and behaviour of
Tibetan Eared Pheasant (Crossoptilon harmani). PhD
dissertation, Beijing Normal Univ., Beijing, China. (in
Chinese with English summary)
Lu X, G Zheng. 2002. Habitat use of the Tibetan Eared
Pheasant Crossoptilon harmani ocks in the non-breeding
season. Ibis 144: 17-22.
Lu X, G Zheng. 2007. Dominance-dependent microroost
use in flock-living Tibetan Eared-Pheasants. Ardea 95:
225-234.
Mackinnon J, K Phillipps, FQ He, ed. 2000. A eld guide to the
birds of China. Changsha: Hunan Education Press. (in
Chinese)
Morse DH, ed. 1980. Behavioral mechanisms in ecology.
Cambridge, MA: Harvard Univ. Press.
Orians GH, JE Wittenberger. 1991. Spatial and temporal
scales in habitat selection. Am. Nat. 137(Supplement):
29-49.
Prodon R, JD Lebreton. 1981. Breeding avifauna of a
Mediterranean succession: the Holm oak and cork oak
series in the eastern Pyrenees. Part 1. Analysis and
modeling of the structure gradient. Oikos 37: 21-38.
Rumble MA. 1992. Roosting habitat of Merriam’s Turkeys in
the Black Hills, South Dakota. J. Wildlife Manage. 56:
750-759.
SAS Institute. 2005. SAS/STAT 9.1.3 User guide. Cary, NC:
SAS Institute.
Skutch AF, ed. 1989. Birds asleep. Austin, TX: Univ. of Texas
Press.
SPSS. 2001. SPSS 11.0 User’s guide. Chicago, IL: SPSS.
Sun Q, Z Zhang, J Zhu, Z Gao. 2002. Roosting behavior and
factors affecting roost-site used by Reeve’s Pheasant
(Syrmaticus reevesii). J. Beijing Norm. Univ. (Nat. Sci.)
38: 108-112. (in Chinese with English summary)
Sutherland WJ. 1998. The importance of behavioural studies
in conservation biology. Anim. Behav. 56: 801-809.
Sweringin RM. 2007. Winter roosting ecology of Rio
Grande wild Turkeys in the rolling plains of Texas. MS
dissertation, Texas Tech Univ., Lubbock, TX.
Thompson DJ. 2003. Roosting habitat and poult survival of
Merriam’s Turkeys in the southern Black Hills of South
Dakota. MS dissertation, South Dakota State Univ.,
Brookings, SD.
Walsberg GE. 1983. Avian ecological energetics. In DS
Farner, JR King, KC Parkes, eds. Avian biology, Vol. 7.
New York: Academic Press, pp. 161-220.
Wang S, Y Xie, ed. 2004. China species red list. Beijing:
Higher Education Press. (in Chinese)
Woltmann S. 2004. Group roosting behavior of Yellow
Tyrannulets (Capsiempis flaveola). Wilson Bull. 116:
352-354.
Xu Y, J Ran, X Zhou, N Yang, B Yue, Y Wang. 2008. The
effect of temperature and other factors on roosting times
of Szechenyi Monal Partridges Tetraophasis szechenyii
during the breeding season. Ornis Fennica 85: 126-134.
Xu et al. – Group Roosts of Buff-throated Partridges 469