Directory UMM :Data Elmu:jurnal:A:Applied Animal Behaviour Science:Vol70.Issue2.2000:
Applied Animal Behaviour Science 70 (2000) 157±165
Social interactions among captive brushtail
possums (Trichosurus vulpecula)
T.D. Daya,b,*, C.E. O'Connora, J.R. Waasb, L.R. Matthewsa
a
Animal Behaviour and Welfare Research Centre, AgResearch Ruakura, Private Bag 3123,
Hamilton, New Zealand
b
Department of Biological Sciences, University of Waikato, Private Bag 3105,
Hamilton, New Zealand
Accepted 1 May 2000
Abstract
A biological control strategy relying on a self-disseminating agent may provide the only
affordable long-term technique for reducing brushtail possum (Trichosurus vulpecula) numbers
throughout New Zealand. The objective of this study was to determine the frequency and patterns of
social interactions in single and mixed-sex groups of possums, in order to identify interactions that
may assist in the dissemination of biocontrol agents. Thirty-two sexually mature wild-caught
possums (16 female, 16 male) were housed in captive groups (of four) in enclosures during the
breeding and non-breeding seasons. Groups consisted of either all females, females and males or all
males. Four types of social interactions were de®ned: threats, ®ghts, af®liative and sexual
interactions. Interactions only occurred between two animals at one time and were most frequent in
the non-breeding season. Within all female groups, possums readily engaged in interactions that had
either an agonistic or af®liative nature. Mixed-sex groups interacted less frequently than females,
but also engaged in both agonistic and af®liative interactions. In contrast, possums in all male
groups rarely interacted, with only a few ®ghts and no af®liative interactions observed. Some
mixed-sex dyads appeared to `associate' during the breeding period. Young were produced by three
females that regularly associated with a male and one female that showed little associative
behaviour. Given the different interaction patterns observed in each group type, biological control
agents that rely on speci®c interaction patterns for dissemination, are likely to spread at different
rates among different possum groups and in different seasons. # 2000 Elsevier Science B.V. All
rights reserved.
Keywords: Brushtail possum; Social behaviour; Biological control
*
Corresponding author. Tel.: 64-7-838-5560; fax: 64-7-838-5727.
E-mail address: dayt@agresearch.cri.nz (T.D. Day).
0168-1591/00/$ ± see front matter # 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 1 6 8 - 1 5 9 1 ( 0 0 ) 0 0 1 4 4 - 1
158
T.D. Day et al. / Applied Animal Behaviour Science 70 (2000) 157±165
1. Introduction
Brushtail possums (Trichosurus vulpecula) are Australian marsupials that were introduced to New Zealand in the 1840s to establish a fur trade (Pracy, 1974). They are
nocturnal, mainly arboreal and omnivorous. Despite attempts to control them, possums
now occupy more than 90% of the country, occurring at densities up to 20 times that found
in Australia (Cowan, 1990). In New Zealand, possums seriously threaten the livestock
industry as they are important vectors in the spread of bovine tuberculosis (Mycobacterium
bovis; Tb) to cattle and deer (Livingstone, 1991). They also cause severe damage to native
plants by selective browsing (Nugent, 1995) and adversely affect native fauna through
competition, disturbance and predation (Innes, 1995). Biological control, relying on a selfdisseminating control agent, may be the only option for achieving an affordable long-term
reduction in possum numbers throughout New Zealand (Jolly, 1993). Some behaviours or
social interactions in the possum may facilitate the spread of biological control agents.
Free-living possums are usually solitary (Winter, 1976), have widely overlapping home
ranges (Green, 1984) and do not defend a territory within their range (Winter, 1976). Wild
possums are suggested to spend little time on active social interaction, except during the
breeding season (Winter, 1976). Interactions usually only occur between a pair of possums
at one time, with groups rarely acting as a cohesive social unit (Winter, 1976). After initial
interactions, dominance is maintained without further interaction by mutual avoidance of
co-dominants (Winter, 1976; Biggins and Overstreet, 1978). Female possums are usually
dominant to males, with older and larger possums dominating small young animals (Jolly,
1976; Biggins and Overstreet, 1978).
The social organisation and behaviour of possums have been studied in both wild and
captive situations. Typically research has focused on dominance relationships (e.g. Biggins
and Overstreet, 1978; Jolly et al., 1996), mating behaviour (e.g. Oldham, 1986) and
ecological aspects of the possum's social organisation (e.g. Jolly, 1976; Winter, 1976;
Green, 1984). No studies to date have quanti®ed the frequencies and patterns of social
interactions between single and mixed sex groups of possums in different seasons. The
objective of this study was to quantify the frequency and patterns of social interactions
among single and mixed-sex groups of captive possums, in order to identify interactions
that may facilitate or restrict the dissemination of biological control agents.
2. Materials and methods
2.1. Subjects and housing
Thirty-two sexually mature wild possums (mean age 4.10.2 years; 16 female, 16 male)
were caught in box traps from the Urewera National Park and the Marlborough Sounds,
New Zealand and transported to the Animal Behaviour and Welfare Research Centre
(AgResearch, Hamilton, New Zealand). Possums were initially housed in individual cages
and fed a daily diet of 200 g mash (Northern Rolling Mill possum pellets and water at a
1:1.5 ratio) and an apple. After a minimum of 4 weeks adaptation to captivity, groups of
four possums (two groups containing all females, two with all males and four groups with
T.D. Day et al. / Applied Animal Behaviour Science 70 (2000) 157±165
159
two males and two females) were housed together in grass enclosures during the nonbreeding (December 1996±February 1997) or breeding (March 1997±May 1997) seasons.
The four groups that were housed together in the non-breeding season remained in their
groups throughout the breeding season in addition to the four new groups. Each grass
enclosure (size range 50±130 m2) contained suf®cient wooden dens for all animals and had
two climbing logs and an ad libitum supply of pellet food and water. As far as possible,
possums within each group were from the same population and were of similar age and
body weight.
2.2. Procedure
Possums in each group were given a 3 cm1.5 cm eartag (yellow, red or blue; All¯ex1
New Zealand, Ltd.) to allow us to identify individual animals from a distance. Each group
was observed for a 3 h period from `civil twilight' (30 min after sunset) during their ®rst
night in an enclosure together, and on four further nights during the following 6 weeks. The
observer was positioned behind a glass window, in an elevated central observation tower,
approximately 5 m from the nearest part of the enclosures. Ten white spotlights (100 W)
were used to provide suf®cient illumination in the enclosures (about 20 lx) for possum
behaviour to be recorded. This light level was similar to bright moonlight and did not deter
the possums from using the enclosures.
All social behaviours shown by a possum that were directed towards another possum
(from less than 1 m away) were recorded. The behaviour de®nitions of Winter (1976),
Biggins and Overstreet (1978), Day (1996) and Hickling and Sun (2000) were used to
de®ne the observed behaviours. Each sequence of behaviours shown by a dyad (pair of
possums) while remaining within 1 m, was then classed into one of four interaction types:
(1) threat, which involved no physical contact (glare, vocalise, bipedal threat, swipe, lunge
and chase behaviours); (2) ®ght, which always involved physical contact (boxing, pounce
and ®ght behaviours); (3) af®liative (touch, food share and allogroom behaviours); and (4)
sexual (mating) interactions. Each interaction began with an `approach' by one possum and
ended with one animal `leaving' (Day, 1996). The frequency and patterns of interactions
were compared, during the breeding and non-breeding seasons, for the three group types:
(1) all female; (2) mixed-sex; and (3) all male. Within each group, the interactions of each
type of dyad (female±female, male±female, male±male) in both seasons were also
compared (six dyads per group, four groups in non-breeding season, eight groups in
breeding season, 72 dyads in total). All of the female possums were examined at 14-day
intervals for evidence of pouch young, to help determine if mating had occurred (i.e.
outside the observation period).
2.3. Statistical analysis
The frequencies of each interaction type were calculated for all dyads in each group,
during both seasons. The data were treated as Poisson counts (to test if interactions were
randomly distributed) and analysed using the generalised linear mixed model (GLMM),
with the type of group and the season of interaction as ®xed effects and the enclosure in
which the dyad was housed as a random factor. From the Poisson data, back-transformed
160
T.D. Day et al. / Applied Animal Behaviour Science 70 (2000) 157±165
mean interaction frequencies and their standard errors were calculated. For individual
interaction types, standard errors could not be calculated, as the individual cage variation in
interaction frequency was too large to average the standard error terms. Therefore, standard
errors of the differences between seasons or group types are presented. The Wald test for
®xed effects (Wald statistic; Genstat 5 Committee, 1993) was used to determine which
factors had signi®cant effects on interaction frequencies.
For each possum within each group type the probability of it being involved in an
interaction of a speci®c type in a speci®ed season was modelled. A quasi-independence test
was used for each dyad to determine whether that dyad performed speci®c interactions
more or less frequently in either season than would be predicted by random interactions
(Larntz and Weisberg, 1976). More interactions than predicted in the quasi-independence
test would suggest some non-random `association' between the possums in that dyad,
whereas fewer interactions would suggest some `separation' (although not necessarily
deliberate) between the dyad. All statistical analyses were performed using the Genstat 5
statistical analysis package (Release 3.1; Lawes Agricultural Trust, Rothamsted Experimental Station).
3. Results
All of the social behaviours were seen in each of the three types of possum dyad, except
for the `food share' and `allogroom' behaviours, which were never observed in male±male
dyads. All interactions occurred between only two possums at one time. A total of 726
interactions were observed in this study, of which 60% were threats, 21% were ®ghts, 19%
were af®liative and
Social interactions among captive brushtail
possums (Trichosurus vulpecula)
T.D. Daya,b,*, C.E. O'Connora, J.R. Waasb, L.R. Matthewsa
a
Animal Behaviour and Welfare Research Centre, AgResearch Ruakura, Private Bag 3123,
Hamilton, New Zealand
b
Department of Biological Sciences, University of Waikato, Private Bag 3105,
Hamilton, New Zealand
Accepted 1 May 2000
Abstract
A biological control strategy relying on a self-disseminating agent may provide the only
affordable long-term technique for reducing brushtail possum (Trichosurus vulpecula) numbers
throughout New Zealand. The objective of this study was to determine the frequency and patterns of
social interactions in single and mixed-sex groups of possums, in order to identify interactions that
may assist in the dissemination of biocontrol agents. Thirty-two sexually mature wild-caught
possums (16 female, 16 male) were housed in captive groups (of four) in enclosures during the
breeding and non-breeding seasons. Groups consisted of either all females, females and males or all
males. Four types of social interactions were de®ned: threats, ®ghts, af®liative and sexual
interactions. Interactions only occurred between two animals at one time and were most frequent in
the non-breeding season. Within all female groups, possums readily engaged in interactions that had
either an agonistic or af®liative nature. Mixed-sex groups interacted less frequently than females,
but also engaged in both agonistic and af®liative interactions. In contrast, possums in all male
groups rarely interacted, with only a few ®ghts and no af®liative interactions observed. Some
mixed-sex dyads appeared to `associate' during the breeding period. Young were produced by three
females that regularly associated with a male and one female that showed little associative
behaviour. Given the different interaction patterns observed in each group type, biological control
agents that rely on speci®c interaction patterns for dissemination, are likely to spread at different
rates among different possum groups and in different seasons. # 2000 Elsevier Science B.V. All
rights reserved.
Keywords: Brushtail possum; Social behaviour; Biological control
*
Corresponding author. Tel.: 64-7-838-5560; fax: 64-7-838-5727.
E-mail address: dayt@agresearch.cri.nz (T.D. Day).
0168-1591/00/$ ± see front matter # 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 1 6 8 - 1 5 9 1 ( 0 0 ) 0 0 1 4 4 - 1
158
T.D. Day et al. / Applied Animal Behaviour Science 70 (2000) 157±165
1. Introduction
Brushtail possums (Trichosurus vulpecula) are Australian marsupials that were introduced to New Zealand in the 1840s to establish a fur trade (Pracy, 1974). They are
nocturnal, mainly arboreal and omnivorous. Despite attempts to control them, possums
now occupy more than 90% of the country, occurring at densities up to 20 times that found
in Australia (Cowan, 1990). In New Zealand, possums seriously threaten the livestock
industry as they are important vectors in the spread of bovine tuberculosis (Mycobacterium
bovis; Tb) to cattle and deer (Livingstone, 1991). They also cause severe damage to native
plants by selective browsing (Nugent, 1995) and adversely affect native fauna through
competition, disturbance and predation (Innes, 1995). Biological control, relying on a selfdisseminating control agent, may be the only option for achieving an affordable long-term
reduction in possum numbers throughout New Zealand (Jolly, 1993). Some behaviours or
social interactions in the possum may facilitate the spread of biological control agents.
Free-living possums are usually solitary (Winter, 1976), have widely overlapping home
ranges (Green, 1984) and do not defend a territory within their range (Winter, 1976). Wild
possums are suggested to spend little time on active social interaction, except during the
breeding season (Winter, 1976). Interactions usually only occur between a pair of possums
at one time, with groups rarely acting as a cohesive social unit (Winter, 1976). After initial
interactions, dominance is maintained without further interaction by mutual avoidance of
co-dominants (Winter, 1976; Biggins and Overstreet, 1978). Female possums are usually
dominant to males, with older and larger possums dominating small young animals (Jolly,
1976; Biggins and Overstreet, 1978).
The social organisation and behaviour of possums have been studied in both wild and
captive situations. Typically research has focused on dominance relationships (e.g. Biggins
and Overstreet, 1978; Jolly et al., 1996), mating behaviour (e.g. Oldham, 1986) and
ecological aspects of the possum's social organisation (e.g. Jolly, 1976; Winter, 1976;
Green, 1984). No studies to date have quanti®ed the frequencies and patterns of social
interactions between single and mixed sex groups of possums in different seasons. The
objective of this study was to quantify the frequency and patterns of social interactions
among single and mixed-sex groups of captive possums, in order to identify interactions
that may facilitate or restrict the dissemination of biological control agents.
2. Materials and methods
2.1. Subjects and housing
Thirty-two sexually mature wild possums (mean age 4.10.2 years; 16 female, 16 male)
were caught in box traps from the Urewera National Park and the Marlborough Sounds,
New Zealand and transported to the Animal Behaviour and Welfare Research Centre
(AgResearch, Hamilton, New Zealand). Possums were initially housed in individual cages
and fed a daily diet of 200 g mash (Northern Rolling Mill possum pellets and water at a
1:1.5 ratio) and an apple. After a minimum of 4 weeks adaptation to captivity, groups of
four possums (two groups containing all females, two with all males and four groups with
T.D. Day et al. / Applied Animal Behaviour Science 70 (2000) 157±165
159
two males and two females) were housed together in grass enclosures during the nonbreeding (December 1996±February 1997) or breeding (March 1997±May 1997) seasons.
The four groups that were housed together in the non-breeding season remained in their
groups throughout the breeding season in addition to the four new groups. Each grass
enclosure (size range 50±130 m2) contained suf®cient wooden dens for all animals and had
two climbing logs and an ad libitum supply of pellet food and water. As far as possible,
possums within each group were from the same population and were of similar age and
body weight.
2.2. Procedure
Possums in each group were given a 3 cm1.5 cm eartag (yellow, red or blue; All¯ex1
New Zealand, Ltd.) to allow us to identify individual animals from a distance. Each group
was observed for a 3 h period from `civil twilight' (30 min after sunset) during their ®rst
night in an enclosure together, and on four further nights during the following 6 weeks. The
observer was positioned behind a glass window, in an elevated central observation tower,
approximately 5 m from the nearest part of the enclosures. Ten white spotlights (100 W)
were used to provide suf®cient illumination in the enclosures (about 20 lx) for possum
behaviour to be recorded. This light level was similar to bright moonlight and did not deter
the possums from using the enclosures.
All social behaviours shown by a possum that were directed towards another possum
(from less than 1 m away) were recorded. The behaviour de®nitions of Winter (1976),
Biggins and Overstreet (1978), Day (1996) and Hickling and Sun (2000) were used to
de®ne the observed behaviours. Each sequence of behaviours shown by a dyad (pair of
possums) while remaining within 1 m, was then classed into one of four interaction types:
(1) threat, which involved no physical contact (glare, vocalise, bipedal threat, swipe, lunge
and chase behaviours); (2) ®ght, which always involved physical contact (boxing, pounce
and ®ght behaviours); (3) af®liative (touch, food share and allogroom behaviours); and (4)
sexual (mating) interactions. Each interaction began with an `approach' by one possum and
ended with one animal `leaving' (Day, 1996). The frequency and patterns of interactions
were compared, during the breeding and non-breeding seasons, for the three group types:
(1) all female; (2) mixed-sex; and (3) all male. Within each group, the interactions of each
type of dyad (female±female, male±female, male±male) in both seasons were also
compared (six dyads per group, four groups in non-breeding season, eight groups in
breeding season, 72 dyads in total). All of the female possums were examined at 14-day
intervals for evidence of pouch young, to help determine if mating had occurred (i.e.
outside the observation period).
2.3. Statistical analysis
The frequencies of each interaction type were calculated for all dyads in each group,
during both seasons. The data were treated as Poisson counts (to test if interactions were
randomly distributed) and analysed using the generalised linear mixed model (GLMM),
with the type of group and the season of interaction as ®xed effects and the enclosure in
which the dyad was housed as a random factor. From the Poisson data, back-transformed
160
T.D. Day et al. / Applied Animal Behaviour Science 70 (2000) 157±165
mean interaction frequencies and their standard errors were calculated. For individual
interaction types, standard errors could not be calculated, as the individual cage variation in
interaction frequency was too large to average the standard error terms. Therefore, standard
errors of the differences between seasons or group types are presented. The Wald test for
®xed effects (Wald statistic; Genstat 5 Committee, 1993) was used to determine which
factors had signi®cant effects on interaction frequencies.
For each possum within each group type the probability of it being involved in an
interaction of a speci®c type in a speci®ed season was modelled. A quasi-independence test
was used for each dyad to determine whether that dyad performed speci®c interactions
more or less frequently in either season than would be predicted by random interactions
(Larntz and Weisberg, 1976). More interactions than predicted in the quasi-independence
test would suggest some non-random `association' between the possums in that dyad,
whereas fewer interactions would suggest some `separation' (although not necessarily
deliberate) between the dyad. All statistical analyses were performed using the Genstat 5
statistical analysis package (Release 3.1; Lawes Agricultural Trust, Rothamsted Experimental Station).
3. Results
All of the social behaviours were seen in each of the three types of possum dyad, except
for the `food share' and `allogroom' behaviours, which were never observed in male±male
dyads. All interactions occurred between only two possums at one time. A total of 726
interactions were observed in this study, of which 60% were threats, 21% were ®ghts, 19%
were af®liative and