High mate and site fidelity in cunningham's skinks (Egernia cunninghami) in natural and fragmented habitat.

Abstract

While habitat alteration has considerable potential to disrupt important within-population processes, such as mating and kin structure, via changed patterns of dispersal, this has rarely been tested. We are investigating the impact of anthropogenic habitat alteration on the population biology of the rock-dwelling Australian lizard Egernia cunninghami on the Central Tablelands of New South Wales, Australia, by comparing deforested and adjacent naturally vegetated areas. The novel analyses in this paper, and its companion, build on previous work by adding a new replicate site, more loci and more individuals. The additional microsatellite loci yield sufficient power for parentage analysis and the sociobiological inferences that flow from it. Genetic and capture-mark-recapture techniques were used to investigate mate and site fidelity and associated kin structure. Analyses of the mating system and philopatry using 10 microsatellite loci showed high levels of site fidelity by parents and their offspring in natural and deforested habitats. Parentage assignment revealed few individuals with multiple breeding partners within seasons and fidelity of pairs across two or more breeding seasons was typical. Despite reduced dispersal, increased group sizes and significant, dramatic increases in relatedness among individuals within rock outcrops in deforested areas, no significant differences between deforested and natural areas were evident in the degree of multiple mating or philopatry of breeding partners within and across seasons. With the exception that there was a significantly higher proportion of unmated males in the deforested area, the social and mating structure of this species has so far been surprisingly robust to substantial perturbation of dispersal and relatedness structure. Nonetheless, approximately 10-fold elevation of mean pairwise relatedness in the deforested areas has great potential to increase inbred matings, which is investigated in the companion paper.

Full text

Page 1

Molecular Ecology (2004)
13
, 419–430doi: 10.1046/j.1365-294X.2003.02061.x
© 2004 Blackwell Publishing Ltd
Blackwell Publishing, Ltd.
High mate and site fidelity in Cunningham’s skinks
(
Egernia cunninghami
) in natural and fragmented habitat
A. J. STOW
*
Department of Biological Sciences, Macquarie University, NSW 2109 Australia,
3086, Australia
* and P. SUNNUCKS†
†
Department of Genetics, La Trobe University, VIC
Abstract
While habitat alteration has considerable potential to disrupt important within-population
processes, such as mating and kin structure, via changed patterns of dispersal, this has
rarely been tested. We are investigating the impact of anthropogenic habitat alteration on
the population biology of the rock-dwelling Australian lizard
Central Tablelands of New South Wales, Australia, by comparing deforested and adjacent
naturally vegetated areas. The novel analyses in this paper, and its companion, build on
previous work by adding a new replicate site, more loci and more individuals. The addi-
tional microsatellite loci yield sufficient power for parentage analysis and the sociobiolo-
gical inferences that flow from it. Genetic and capture–mark–recapture techniques were
used to investigate mate and site fidelity and associated kin structure. Analyses of the mat-
ing system and philopatry using 10 microsatellite loci showed high levels of site fidelity by
parents and their offspring in natural and deforested habitats. Parentage assignment
revealed few individuals with multiple breeding partners within seasons and fidelity of
pairs across two or more breeding seasons was typical. Despite reduced dispersal, increased
group sizes and significant, dramatic increases in relatedness among individuals within
rock outcrops in deforested areas, no significant differences between deforested and natu-
ral areas were evident in the degree of multiple mating or philopatry of breeding partners
within and across seasons. With the exception that there was a significantly higher propor-
tion of unmated males in the deforested area, the social and mating structure of this species
has so far been surprisingly robust to substantial perturbation of dispersal and relatedness
structure. Nonetheless, approximately 10-fold elevation of mean pairwise relatedness in
the deforested areas has great potential to increase inbred matings, which is investigated
in the companion paper.
Egernia cunninghami
on the
Keywords
: family, habitat fragmentation, lizard, mate fidelity, microsatellite, site fidelity
Received 24 February 2003; revision received 1 September 2003; accepted 15 October 2003
Introduction
Habitat fragmentation and the rationale of the
cunninghami
programme
Egernia
Habitat fragmentation has been identified as a distinct
threatening process in conservation biology (e.g. WCMC
1992; Bjørnstad
et al
. 1998; Boudjemadi
siderable research has shown that restricted gene flow
owing to fragmentation may elevate genetic differentiation
et al
. 1999). Con-
among populations, and can increase the rate of popu-
lation extinction through genetic factors (Sarre 1995, 1996;
Cunningham & Moritz 1998; Saccheri
demographic factors (e.g. Boudjemadi
theless, the influences of habitat fragmentation on within-
group processes (such as mating opportunities and kin
structure) are much less well understood (Peacock & Smith
1997; Amos & Balmford 2001; Luck 2003). If within-group
processes are affected by habitat alteration, the nature of
residual populations may be changed deleteriously in ways
that are potentially cryptic and cumulative. For example,
human-induced habitat fragmentation could lead to post-
poned or impeded dispersal. This could lead to increased
et al
. 1998) as well as
et al
. 1999). None-
Correspondence: Adam Stow. Fax: + 61 29850 8245;
E-mail: astow@rna.bio.mq.edu.au

Page 2

420
A. J. STOW and P. SUNNUCKS
© 2004 Blackwell Publishing Ltd,
Molecular Ecology
, 13, 419–430
competition, unnaturally elevated philopatry and the dis-
ruption of potentially adaptive kin and social structures
and breeding opportunities (Peacock & Smith 1997; Bjørnstad
et al
. 1998; Boudjemadi
et al
. 1999; Stow
et al
. 2001). Certainly reduced juvenile dispersal can elevate
genetic differentiation among local groups (Baker
As with other phenomena in this context, several impacts
of reduced dispersal on the mating system are possible. In
the case of
E
.
cunninghami
, the subject of the present study,
constrained dispersal (Stow
promote the formation of long-term pair bonds by holding
individuals together, or could hold together individuals
that would not otherwise be likely mating partners. Second,
it has been suggested that females may facultatively use
polyandry to counter increased risk of inbreeding, which
may occur under increased natal philopatry (Petrie &
Kempenaers 1998; Hosken & Blanckenhorn 1999; Tregenza
& Wedell 2000). Third, increased aggregation of females
may promote harem defence polygyny (Clutton-Brock
1989). Thus disruption to the mating system of
hami
following dramatic decreases in dispersal and increases
in density of close relatives (Stow
and even likely.
To enhance our understanding of the functions and evo-
lutionary processes involved with long-term pair bonds,
family group structure, and the potential impacts of habi-
tat alteration, it is necessary to study such behaviours in
phylogenetically and ecologically diverse taxa (Petrie &
Kempenaers 1998; Bull 2000). We have been investigating
the effects of habitat alteration on
secretive scincid lizard that utilizes mainly rocky retreat
sites throughout its distribution in southeastern Australia.
This convenient model lizard species was chosen because
reptiles are under-represented in this type of research, yet
they may be expected to show responses quite different
from those of birds and mammals for which more data are
available (Stow
et al
. 2001). Initial data on
from one site (Tarana) suggested that deforestation around
rock outcrops dramatically reduced dispersal, particularly
of females, and led to elevated local relatedness (Stow
2001). However, the full implications of those findings
could not be further investigated at that time because (i)
a replicate site had not yet been identified and sampled,
(ii) some loci in the microsatellite array were problematic
and after their removal the genetic assay lacked sufficient
resolution, e.g. for confident parentage estimation, and
(iii) sample sizes of lizards were small for some analyses. A
replicate site (Bathurst) has now been sampled and more
samples and field data have been collected at Tarana. In
addition, more microsatellite loci have been isolated (Stow
2002) and all sampled individuals have been typed at
these. We can now report much more fully on the social
structure and mating system of
impacts of deforestation. We present the data in a pair of
et al
. 2001; Templeton
et al
. 2000).
et al
. 2001) could potentially
E. cunning-
et al
. 2001) are plausible
E. cunninghami
, a large,
E. cunninghami
et al
.
E. cunninghami
and any
papers: this one focuses on mate and site fidelity and their
impact on spatial organization of genetic relatives, while
the other focuses on whether the coexistence of highly
related individuals leads to inbred matings (Stow and
Sunnucks 2002). In each paper, natural and adjacent defor-
ested habitat is compared at two sites (Tarana and the
recently added Bathurst).
Lizard social group structures and mating systems
The most commonly reported mating systems in lizards
involve multiple mating partners, with polygyny being
most prevalent (Bull 2000). Although several studies
show social monogamy in lizard species within a season
(Toxopeus
et al
. 1988; Olsson & Shine 1998), research demon-
strating the persistence of pair bonds across seasons is rare,
being known in
Tiliqua rugosa
(Bull
et al
. 1998; Gardner
et al
. 2002; O’Connor & Shine 2003).
Presently, genetic evidence for family group structure in
lizards has been published for only three lizard species,
all in the genus
Egernia
:
E. cunninghami
saxatilis
(Gardner
et al
. 2001; Stow
Shine 2003).
For offspring to delay dispersal and remain with par-
ents may reflect a selective advantage (Emlen 1994). Poten-
tial benefits include extended parental care, access to
resources, and altruistic benefits of social behaviours
(e.g. Clutton-Brock 1991; Emlen 1994; Jones & Parker 2002;
Schneider 2002). Social group organization has fitness
consequences and is subject to evolutionary optimization
(Stacey & Ligon 1991; Roberts 1996; Brown & Brown 2000).
Thus it is important to understand the extent of the impact
of habitat alteration/fragmentation on the social group
structure and mating systems. In this study we investigate
mate and site fidelity and kin structure in
and the impact of deforestation (habitat fragmentation) on
these aspects of the species’ biology. We are not aware of
another study that does this for a reptile.
Egernia cunninghami
forms aggregations of up to 27
individuals or more, comprising adults of each sex and
immatures of several age-cohorts (Barwick 1965; Stow
2001). The species reaches sexual maturity at approxi-
mately 190 mm snout–vent length (SVL), is viviparous and
usually has four to eight young per litter (Barwick 1965).
Captive specimens can survive in excess of 35 years (P
Harlow personal communication). Previous genetic work
suggested that
E. cunninghami
close kin, with elevated levels of relatedness among adults
within groups, and lower levels of inferred dispersal asso-
ciated with deforestation (Stow
these inferences were yet to be demonstrated by direct
parentage analysis. In addition, more than one adult male
or female are often captured together, therefore it was
unknown whether the mating system involves long-term
,
E. stokesii
and
E. saxatilis
,
E. stokesii
. 2001; O’Connor &
and
E.
et al
E. cunninghami
,
et al
.
lives in groups consisting of
et al
. 2001). Nonetheless

Page 3

MATING AND KIN STRUCTURE IN LIZARDS
421
© 2004 Blackwell Publishing Ltd,
Molecular Ecology
, 13, 419–430
pair-bond fidelity or multiple breeding partners. Here we
attempt to fill these gaps in knowledge and estimate the
effects of habitat fragmentation on social structuring and
mating system in
E. cunninghami
.
Materials and methods
Study sites and sampling
Individuals of
study locations ‘Tarana’ (33
on in Stow
et al
the Central Tablelands of New South Wales, Australia. The
two sites are separated by approximately 45 km and are
similar in climate, geology and biotic factors. Each site
contains some naturally vegetated and some deforested
grazed habitat, with discrete rock outcrops. The minimum
time since deforestation at each site is 70 years. The over-
storey of the naturally vegetated habitat is largely
with a shrub understorey. The range of distances between
rock outcrops from which lizards were sampled was 27 m
to 3096 m at Bathurst and 22 m to 1375 m at Tarana (Fig. 1).
The a
priori
groups for
E. cunninghami
caught at the same rock outcrop.
ally uses deep crevices within rock outcrops as retreat sites,
with aggregations of lizards often observed within one
crevice. All lizards captured at a rock outcrop were captured
within 15 m of one another. Strenuous efforts were made to
sample all lizards observed, and capture–mark–recapture
(CMR) findings indicate that this was largely successful
(
ca
. 90% of adult male and female lizards captured). At
Egernia cunninghami
were sampled from two
S, 149
°
55
′
E; the site reported
°
27
′
S, 149
°
33
′
. 2001) and ‘Bathurst’ (33
°
24
′
E) on
Eucalyptus
,
comprised individuals
Egernia cunninghami
typic-
Bathurst, 198 individuals were sampled from 31 rock
outcrops between September 2000 and January 2002, and
at Tarana, 167 lizards were sampled from 23 rock outcrops
between March 1998 and December 2001. (For the Tarana
site, 143 of the lizards comprised the sample base for Stow
et al
. 2001.) The average number of sampled individuals
within each group (rock outcrops with two or more
individuals sampled) was eight lizards at each site (range
2–26 at Bathurst and 2–16 at Tarana). Group sizes were not
significantly different between habitats, although groups
were on average larger in deforested habitats (mean
9.08
±
5.78) than the naturally vegetated habitats (7.06
4.64, Mann–Whitney
U
-test;
2003). Individuals were captured, sexed, marked uniquely,
measured and sampled for tissue as described in Stow
(2001).
±
SD,
±
Z
=
−
1.11,
P
= 0.254) (Stow
et al
.
Genotyping
DNA extraction and genotyping methods were essentially
as described previously (Stow
However, locus Tr5.20 exhibited high-frequency putative
null allele(s) at Bathurst and was excluded from these
analyses. Mistyping and recording errors were minimized
through multiple polymerase chain reaction and electro-
phoretic runs of samples and repeated checks for input
errors. Characteristics of the 10 microsatellite loci used are
given in Table 1. Expected heterozygosity, exclusion prob-
abilities and null allele estimates were calculated using
the program
cervus
(Marshall
ative estimate for the probability of genotypic identity (
amongst full siblings (Waits
using the software
gemini
(Valière
et al
. 2001; Stow 2002).
et al
. 1998). The conserv-
P
ID
)
et al
. 2001) was calculated
et al
. 2002).
Lizard age
Groups of
ing to discrete size classes.
maximally once a year, with litters produced usually during
late January and February (Barwick 1965). The timing of
fieldwork allowed direct capture of newborns and thus
yielded their year of birth. For individuals not caught as
newborns, growth curves estimated from SVL measure-
ments obtained through CMR were used to estimate age
(Stow 2003).
Lizards in their first and second years of growth were
allocated an age-cohort and year of birth. However, to
accommodate the possibility of accumulated individual
differences in growth rates and incorrect age determina-
tion, lizards in their third or more year of growth but prior
to maturity were placed in the category ‘subadults’. There-
fore only lizards captured in their first two years of growth
are used to examine breeding-pair fidelity within and
across two consecutive seasons. The subadult category
E. cunninghami
often contain individuals belong-
Egernia cunninghami
breeds
Fig. 1 Sampling locations of Egernia cunninghami at Bathurst and
Tarana, in two different habitats, naturally vegetated and defor-
ested. Individuals were sampled from a total of 54 rock outcrops,
from which groups of lizards were sampled in 41 of these (single
lizards were sampled from 13 rock outcrops).

Page 4

422
A. J. STOW and P. SUNNUCKS
© 2004 Blackwell Publishing Ltd,
Molecular Ecology
, 13, 419–430
was included to determine the number of cases wherein an
individual has the same breeding partner in more than two
seasons.
Parentage analysis
Parentage was estimated to provide an indicator of mate
fidelity within and across breeding seasons and to assess
the mobility of parents and offspring. Parentage analysis
was conducted separately for each site, using 10 microsatel-
lite loci (Table 1), with parentage assigned only for immature
lizards (< 190 mm SVL). Lizards that could possibly have
matured during the study were tested as potential parents
and offspring.
Parentage was assigned using the software
(Marshall
et al
. 1998) which determines the logaritum of
the odds (LOD) score for all candidate parents and carries
out simulations to estimate the critical difference in LOD
score between the most likely and second most likely
candidate parent, at 80% and 95% confidence levels.
Simulations for both sites were run with the following
parameters: 10 000 cycles, 60% of candidate parents sam-
pled, 100% of loci typed and a genotyping error rate of 1%
(approximated from multiple assays). It was assumed that
60% of candidate parents had been sampled, an extremely
conservative estimate, far lower than the likely real value
(Results). Parental assignments were accepted if the can-
didate was not genetically incompatible at more than one
locus and could be the parent with 80% or 95% confidence.
In addition, the level of genetic compatibility in parent-
offspring assignments made at 80% confidence had to be
higher than that obtained between the offspring and any
other sampled candidate parent. Parentage assignment at
less than 80% confidence was accepted for 12 adults that
cervus
presented no genetic mismatches with offspring, and had
the highest positive LOD score. This was considered reason-
able because no other sampled candidate parents were
genetically compatible with these offspring, and the major-
ity of parents (10) were assigned with at least 80% con-
fidence to other offspring that most probably belonged
to their clutch (based on genotypes, age and capture
location). According to these criteria, parentage was
assigned to each sex in a stepwise manner (Marshall
et al
. 1998). Offspring paternity was determined first, then
maternity was assigned with some offspring already
having an identified father.
The estimates for null alleles segregating at each locus
(Table 1), suggests that nonamplifying alleles may be
present at some loci. These could result in homozygote
mismatches between parent and offspring. Null alleles
tend to underestimate LOD scores and result in a more
conservative parental assignment (Marshall
However, it was important to ensure that estimates of
breeding-pair fidelity were not biased by the false exclu-
sion of one parent because of null alleles. Using the assign-
ment criteria above, genotypes of all offspring that were
assigned to only one parent were compared with the geno-
types of the most likely candidate as second parent. If off-
spring were found to be incompatible with this candidate
parent solely because of homozygous mismatches at loci
that potentially have null alleles, the single parent–offspring
assignment was excluded from analysis.
et al
. 1998).
Breeding-pair fidelity and single parent assignments
The procedures outlined in this section were applied within
and among seasons to test for fidelity on those two time
scales. Estimation of mate fidelity was most straightforward
Table 1 Number of alleles (Na), heterozygosity estimates (HE), exclusion probabilities (Ex), cumulative genotypic identity for first degree
relatives (PID(Sib)) and null allele estimates (Null) of 10 microsatellite loci used for parental assignment
Bathurst site Tarana site
Locus
Na
HE
Ex
PID(Sib)
NullLocus
Na
HE
Ex
PID(Sib)
Null
Est 13
Est 1
Tr5.21
Est 9
Tr3.2
Ecu 3
Ecu 5
Ecu 4
Ecu 2
Ecu 1
22
23
20
9
25
11
17
31
13
17
0.906
0.898
0.935
0.630
0.916
0.821
0.895
0.924
0.843
0.906
0.674
0.660
0.761
0.239
0.705
0.467
0.646
0.733
0.521
0.676
3.035 × 10−1
8.998 × 10−2
2.573 × 10−2
1.230 × 10−2
3.765 × 10−3
1.344 × 10−3
4.157 × 10−4
1.218 × 10−4
4.141 × 10−5
1.252 × 10−5
0.000
0.000
0.000
0.000
0.000
0.044
0.000
0.054
0.000
0.028
Est 13
Est 1
Tr5.21
Est 9
Tr3.2
Ecu 3
Ecu 5
Ecu 4
Ecu 2
Ecu 1
19
26
18
15
27
12
13
38
13
20
0.925
0.927
0.871
0.516
0.915
0.867
0.808
0.955
0.861
0.921
0.728
0.734
0.587
0.160
0.703
0.574
0.466
0.825
0.553
0.717
2.914 × 10−1
8.472 × 10−2
2.737 × 10−2
1.525 × 10−2
4.533 × 10−3
1.483 × 10−3
5.325 × 10−4
1.463 × 10−4
4.828 × 10−5
1.420 × 10−5
0.000
0.001
0.000
0.000
0.000
0.001
0.000
0.087
0.000
0.018
Total exclusionary power for first parent Tarana site = 0.999962.
Total exclusionary power for first parent Bathurst site = 0.999954.

Page 5

MATING AND KIN STRUCTURE IN LIZARDS 423
© 2004 Blackwell Publishing Ltd, Molecular Ecology, 13, 419–430
when two parents per individual could be identified for
putative offspring arrays (i.e. groups of offspring belong-
ing to one or several age-cohorts that had at least one
parent in common). However, groups of siblings for which
all or some individuals had only a single parent assigned
were still useful, in the following ways. First, polygamy
was inferred where a pair of parents was identified as
contributing to a progeny array but one was excluded as
a parent for a subset of the array. Second, an assigned
parent’s genotype could be used to deduce allelic combina-
tions contributed by unsampled parents and the minimum
number of unsampled parents contributing (using gerud,
& Jones 2001). In a progeny array, the presence of three or
more alleles (at any locus) not attributable to the identified
parent points to multiple unknown parents. This assumes
no mutation or genotyping errors and requires three or
more offspring assigned to a single parent (Dewoody et al.
2000). Where many offspring can be explained by a single
unsampled genotype, that genotype can be reconstructed
(Jones 2001), and such ‘reconstructed genotypes’ can be
tested as parents of offspring or arrays. To be sufficiently
confident in this, a reconstructed genotype was accepted
only when it could account for a whole litter or for all
offspring assigned to a parent. We note the high prob-
ability of genotypic uniqueness and power of exclusion
(Table 1).
Movement among rock outcrop groups
Levels of mobility for E. cunninghami were inferred by the
location of parents and offspring and by direct observation
through CMR. To allow for any disturbance to movement
patterns resulting from capture, repeat location data from
CMR was included only for time periods of 7 days or more
between capture and recapture. For every lizard captured,
the date and rock outcrop at which it was located was
recorded. Distances were measured between all rock
outcrops at which lizards were captured. The geographical
extent of a rock outcrop site was defined to include 11 m
from the edge of the rock outcrop, calculated as half the
minimum distance between rock outcrops at the two sites.
Dispersal amongst rock outcrops was detected using re-
cords of individual capture–recapture locations.
Habitat comparisons
The sites (Bathurst and Tarana) were matched as closely as
possible for climate, structure of rock outcrops and time
since deforestation. To examine the impacts of habitat
alteration on breeding-pair fidelity and philopatry, data
were first examined at each site to ensure consistency, then
pooled across study sites to increase sample size and
statistical power. Unless noted otherwise, comparisons
were made using Fisher’s Exact Test as it is robust to small
expected frequencies (Siegel & Castellan 1988). All prob-
ability values reported are two-tailed.
Results
Parentage assignment
From both sites combined, 201 of 240 offspring sampled
were assigned to at least one sampled parent (Table 2). In
total, 94% of parenthoods were assigned with 80% or 95%
confidence, most (78%) at the 95% level. Three offspring
were assigned a parent but excluded from further analysis
because the second most likely parent was not incom-
patible if a null allele was invoked. Because the two most
likely parents for these three offspring were each assigned
as parents for other immature lizards of the same age-cohort,
their exclusion will not influence breeding-pair fidelity
estimates across seasons. There were 24 sampled adult
pairs identified as parents for a total of 129 offspring. The
mean, standard deviation and range of offspring num-
bers per sampled pair were 5.38 ± 4.17 (1–17). For the
remaining 72 offspring, only a single parent could be
assigned. A pool of 21 parents was identified, each with
1–12 offspring (mean ± SD, 3.43 ± 3.25). These 21 ‘single
parents’ include six individuals also identified as members
of the 24 breeding pairs. There was no significant habitat
difference observed in the proportion of progeny arrays for
which one or both parents could be assigned (deforested
n = 30, natural n = 14; P = 0.510).
Ten of the 21 ‘single parents’ were assigned three or
more offspring (5.90 ± 3.21), and from these 10 arrays it
was possible to assess the number of unidentified parents.
For eight of the 10, a single unsampled parental genotype
could generate all offspring. In the two cases where more
than one unsampled parent was deduced, three or more
alleles unattributable to the known parent were found at
two and five loci, indicating multiple unsampled parents
rather than mutation. One case indicated polygyny within
a breeding season, and the other suggested a female chang-
ing partner between seasons. In the latter case, there were
Table 2 Total number of sampled immatures and adults of each
sex for each habitat and study site
Site/habitatOffspring
Adult
Males
Adult
Females
Bathurst/deforested
Bathurst/natural
Tarana/deforested
Tarana/natural
88 (75)
52 (49)
68 (60)
32 (17)
12 (5)
13 (7)
22 (8)
13 (2)
20 (13)
13 (8)
21 (9)
11 (4)
The number of immature lizards assigned to at least one parent
and the number of male and female lizards assigned as parents are
given in parentheses.