See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/277683494

Factors influencing occurrence of a freshwater turtle in an urban landscape: A

resilient species?

Article  in  Wildlife Research · January 2014

DOI: 10.1071/WR13205

CITATIONS READS

24 1,435

5 authors, including:

Danielle Stokeld Andrew Hamer

20 PUBLICATIONS   528 CITATIONS   
Balaton Limnological Research Institute
42 PUBLICATIONS   1,435 CITATIONS   
SEE PROFILE
SEE PROFILE

Rodney van der Ree Vincent Pettigrove

University of Melbourne University of Melbourne
134 PUBLICATIONS   3,839 CITATIONS    109 PUBLICATIONS   3,104 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Elucidating drivers of change in bird, reptile and mammal communities of tropical savanna View project

Handbook of Road Ecology View project

All content following this page was uploaded by Rodney van der Ree on 03 November 2015.

The user has requested enhancement of the downloaded file.

CSIRO PUBLISHING
Wildlife Research, 2014, 41, 163–171
http://dx.doi.org/10.1071/WR13205

Factors influencing occurrence of a freshwater turtle

in an urban landscape: a resilient species?

Danielle Stokeld A,D,E, Andrew J. Hamer A, Rodney van der Ree A, Vincent Pettigrove B
and Graeme Gillespie C,D
A
Australian Research Centre for Urban Ecology, Royal Botanic Gardens Melbourne c/o School of Botany,
University of Melbourne, Parkville, Vic. 3010, Australia.
B
Centre for Aquatic Pollution Identification & Management, Department of Zoology, University of Melbourne,
Parkville, Vic. 3010, Australia.
C
Department of Zoology, University of Melbourne, Parkville, Vic. 3010, Australia.
D
Present address: Flora and Fauna Division, Department of Land Resource Management, Northern Territory
Government, PO Box 496, Palmerston, NT 0831, Australia.
E
Corresponding author. Email: dstokeld@gmail.com

Abstract
Context. Species vary broadly in their ability to adapt to urbanisation. Freshwater turtles are vulnerable to the loss and
degradation of terrestrial and aquatic habitat in urban environments. There have been few publications investigating
impacts of urbanisation on freshwater turtles in Australia.
Aims. We investigated the effects of urbanisation on the distribution and abundance of the eastern long-necked turtle
(Chelodina longicollis) in greater Melbourne.
Methods. We examined occurrence and relative abundance of C. longicollis at 55 wetlands across an urban–rural gradient
in relation to site- and landscape-level factors. Occupancy was modelled using the program PRESENCE, and incorporated
landscape and habitat covariates. A negative binomial regression model was used to examine the influence of landscape and
habitat factors on relative abundance by using WinBUGS.
Key results. C. longicollis occupied 85% of the 55 wetlands we surveyed, and we found no evidence that wetland
occupancy was influenced by the variables we measured. However, relative abundance was highest at wetlands with low
water conductivity and heavy metal pollution, and in wetlands furthest from rivers.
Conclusions. C. longicollis appears to be resilient to urbanisation and is likely to persist in urban landscapes, possibly
because of the creation of new wetlands in Australian cities. However, long-term studies focussed on demographic
parameters, or survivorship, may elucidate as yet undetected effects of urbanisation. Although no specific management
recommendations may be necessary for C. longicollis in urban areas at this time, this species may be in decline in non-urban
areas as a result of climatic changes and wetland drying.
Implications. Our findings suggest that caution is required before drawing generalised conclusions on the impacts of
urbanisation on turtles, as the effects are likely to be species-specific, dependent on specific ecology and life-history
requirements. Further studies are required to ascertain these relationships for a wider array of species and over longer time
spans.

Additional keywords: Chelodina longicollis, landscape ecology, occupancy, reptile, urbanisation, urban wetlands.

Received 3 December 2013, accepted 26 June 2014, published online 8 August 2014

Introduction In general, urbanisation decreases species richness, with

Urbanisation is one of the most intensive and irreversible forms
of habitat loss, fragmentation and degradation (McKinney 2002;
Fahrig 2003). Urban sprawl is increasing, with more than five
billion people predicted to reside in urban areas by 2030 (UNFPA
2007). Urbanisation alters habitat structure and interconnectivity
(Alberti 2005), and results in broad and spatially variable
environmental and biotic effects across the urban landscape
(McDonnell and Pickett 1990).
Journal compilation
CSIRO 2014
biological homogenisation increasing along a rural–urban
gradient (McKinney 2008). Infrastructure development for
human settlements increases the percentage of impervious
surfaces, reducing and degrading terrestrial and aquatic
habitats. Increased water volume and land-based contaminants
(e.g. heavy metals, pesticides, oils) flow into aquatic systems,
modifying hydrological and sedimentation regimes of rivers and
wetlands (Walsh et al. 2005; Lee et al. 2006). Aquatic toxicants
www.publish.csiro.au/journals/wr
164 Wildlife Research
can have both immediate and long-term impacts on species and
ecosystems (Closs et al. 2004). Built infrastructure impedes
movement by terrestrial species and reduces population
viability (Clevenger and Wierzchowski 2006). Despite these
effects, species vary widely in their tolerance and ability to
adapt to urban environments (McKinney 2002), including
variation in stress and immune responses (French et al. 2008),
population structure (Patrick and Gibbs 2010), survival and
recruitment (Eskew et al. 2010), and ultimately distribution
and abundance (Riley et al. 2005).
Turtles are among the world’s most endangered vertebrates,
with over 50% of species threatened (Turtle Taxonomy Working
Group 2010). Freshwater turtles depend on aquatic and terrestrial
habitats for different parts of their life cycle. As such, they are
vulnerable to the loss and alteration of, and the levels of
connectivity among, both habitats (Semlitsch and Bodie 2003;
Roe and Georges 2007). Terrestrial habitats are utilised for
nesting, aestivation and dispersal (Grgurovic and Sievert 2005;
Harden et al. 2009). Increases in urban road networks can
impede overland movement and increase mortality rates
(Gibbs and Shriver 2002; Eskew et al. 2010). Moreover,
degradation of stream and wetland systems through reduced
water quality and increased concentrations of contaminants
can affect the occurrence and abundance of freshwater turtles
(DeCatanzaro and Chow-Fraser 2010), and may contribute to
changes in prey availability (Luiselli 2004). As such, urbanisation
significantly alters pertinent habitat that may adversely affect
turtle populations.
Urbanisation impacts on turtles have been extensively
examined in North America (see Mitchell and Brown 2008;
Eskew et al. 2010; Patrick and Gibbs 2010); however, little is
known about such impacts in Australia. Research on freshwater
turtles in Australia has primarily focussed on relatively natural
systems, with very little attention on the urban environment
(Roe and Georges 2008a; Chessman 2011; Bower et al. 2012).
Only five studies have examined urban impacts on turtles in
Australia (Burgin 2008; Giles et al. 2008; De Lathouder et al.
2009; Rees et al. 2009; Roe et al. 2011). Importantly, none of
these studies examined distribution or abundance over adequate
spatial scales for a robust investigation of responses of turtles to
urbanisation. These studies (i) had small sample sizes and spatial
scales, (ii) did not use explicit metrics to define urbanisation and
(iii) did not assess species–habitat relationships. As urban areas
expand, more Australian freshwater turtles will become affected
by urbanisation; however, the impacts remain largely unknown.
To address this knowledge gap, we examined the relationships
between urbanisation and the distribution of C. longicollis in a
large Australian city. This species is widely distributed across
south-eastern Australia (Kennett et al. 2009), encompassing the
most densely populated and urbanised regions of the continent
(Fig. 1). Although currently considered secure in the wild, there is
some evidence of population declines in the Murray–Darling
Basin (Chessman 2011). We hypothesised that this species
would be adversely affected by increased levels of urbanisation.
Materials and methods
This study was conducted within the metropolitan area of
Melbourne, Victoria, south-eastern Australia (37
480 S,
D. Stokeld et al.
144
550 E). Melbourne was once covered with extensive
natural wetlands, many of which were drained or filled in for
development in the early to mid-1900s (Melbourne Water 2012).
Some remnant wetlands still remain, but the landscape is now
largely dominated by man-made wetlands, which are generally
included in the design of open spaces and new housing
developments for stormwater management, flood abatement
and public recreation (Melbourne Water 2012).
Discrete wetlands were selected from a dataset of 120
wetlands analysed for heavy metal pollutants (V. Pettigrove,
Melbourne University, unpubl. data). Sites were excluded if
they were dry, inaccessible for survey, or less than 2 km apart,
so as to minimise the effects of spatial auto-correlation (Wintle
and Bardos 2006). This distance was chosen as 95% of
the maximum terrestrial straight-line distance travelled by
C. longicollis has been less than 2 km (Roe and Georges
2008b; Rees et al. 2009). This selection process resulted in
selection of 55 sites spread across a wide urban gradient, from
inner Melbourne to its outer suburbs (Fig. 1).
Trapping was undertaken at each wetland on three separate
occasions between October 2010 and February 2011, coinciding
with the active season for freshwater turtles in Victoria. Turtles
were captured using fyke and cathedral traps, which have
previously been used with equal success (De Lathouder et al.
2009). Fyke traps were set in water up to a depth of 125 cm,
ensuring that part of the funnel was above the surface of the
water providing breathing space. Cathedral traps were placed
in water >150 cm deep to ensure that turtles could pass through
the internal funnel to the holding chamber that provided access
to the water surface. All traps were baited with diced beef heart.
Three fyke traps and one cathedral trap were placed in each
wetland, with the following exceptions. Cathedral traps were
precluded from a site if accessible water depth was <150 cm deep,
and the number of fyke traps set at a site was reduced if the
water body was too small to accommodate all three. Trap
placement was determined using randomly generated cardinal
points and a compass. Traps were set in the morning, between
0700 hours and 1000 hours Australian eastern daylight time,
and removed after 4.5–6 h. The number and species of turtle
captured per trap was recorded. So as to identify previously
captured animals, turtles were individually marked by
notching the marginal scute, or combination of scutes, of the
carapace (Cagle 1939) by using a cordless rotary tool with a
1.5-inch metal cut-off wheel (DREMEL®, Racine, WI, USA).
To assess the influence of local- and landscape-level
environmental variables on wetland use by turtles, the
following parameters were recorded for each site: water
physico-chemistry variables were measured at three randomly
selected locations around the edge of the wetland; water pH
(0.01 pH) and electrical conductivity (0.01 mS cm–1) were
measured with a handheld digital meter (LaMotte Tracer
Pocketester, Chestertown, MD, USA), and water turbidity was
estimated using a turbidity tube (Waterwatch, Melbourne Water).
Measurements were recorded during all trapping occasions, with
an exception of a subset of occasions as a result of equipment
failure. Mean values for physico-chemical variables were used in
all analyses.
The structure of the riparian vegetation within 5 m of the
wetland edge was assessed by walking four 50-m transects
Factors influencing occurrence of a freshwater turtle
Surveyed wetlands
Roads
Kilometres
Fig. 1. Location of randomly selected wetlands surveyed for Chelodina longicollis, in the greater
Melbourne area, 2010–11. Inset map of Australia, shaded area represents the natural range of
C. longicollis.
around the wetland, starting at cardinal points, or the entire
perimeter of a wetland if its circumference was <200 m. The
presence of vegetation structure components (trees, shrubs,
herbaceous ground cover and tussocks/rushes) was recorded
every 10 m along the transect(s), so as to derive a repeatable
estimate of the proportion of various components. The proportion
of emergent and submerged wetland vegetation was estimated
visually, and the presence or absence of islands within the
wetland was also recorded.
Surface area (km2) of each wetland was calculated using
ArcGIS (ESRI 2010). Road density (km km–2) was calculated
for each wetland by summing the total length of roads in a 1-km
buffer around the wetland (Data Source: ‘Road Network
1 : 25 000 (version 1 December 2011)’, The State of Victoria,
Wildlife Research 165
Department of Sustainability and Environment, Melbourne,
Australia). The straight-line distance from a surveyed site to
the nearest neighbouring wetland (km) and nearest river or
stream (km) was estimated using Google Earth (Google Inc.
2011) and Melways 2011 (Ausway Publishing Pty Ltd,
Melbourne, Vic., Australia). A sediment quality quotient
(SQQ) was derived for each wetland by using previously
collected data on heavy metal pollutants and the interim
sediment quality guidelines (ISQG) identified in the National
Water Quality Management Strategy (ANZECC and
ARMCANZ 2000), as follows:
X
SQQ ¼ ðobserved heavy metal X Þ=ðISQG-high X Þ;
166 Wildlife Research D. Stokeld et al.

Where ISQG-high is associated with chemical concentrations
that result in adverse biological effects 50% of the time
(ANZECC and ARMCANZ 2000).
Analyses
Relationships among variables were examined using Pearson’s
correlation coefficients to ensure none was highly correlated
(r > 0.4). Electrical conductivity and wetland area were highly
skewed and log10-transformed to improve normality. All
continuous variables included in the models were either
centred about the mean (X* = X – mean(X)), or scaled before
analysis.
Site occupancy by C. longicollis was analysed using
maximum likelihood to estimate both occupancy and detection
probabilities concurrently, and a link function to model site- and
sampling-specific variables (MacKenzie et al. 2002). The
modelling approach assumes that sites are closed to changes in
occupancy within the survey season; however, if a species is
randomly moving in and out of a site during the survey season,
then the occupancy estimator is expected to be unbiased
(MacKenzie 2006). The occupancy (y) and detection (p) were
modelled using the program PRESENCE (Proteus Research and
Consultancy, Dunedin, New Zealand). A detection history was
constructed for each site by using trapping data.
A set of candidate models was developed to examine the
influence of both local site- and landscape-level factors on
wetland occupancy (Table 1). Because trapping effort was not
uniform across the wetlands, a set of detection models
accounting for this was run a priori to determine whether a
detection covariate should be included in the occupancy
modelling. Models with constant detection and those including
covariates for survey effort were equally probable; therefore,
occupancy models were run with constant detection and with
detection dependent on wetland area. Models that failed to
converge during the modelling procedure were discarded from
the results set, because they were inadequate in describing the
data (Crawley 2007).
An information-theoretic approach was used to identify the
most parsimonious model using the Akaike information criterion
(Burnham and Anderson 2002). A goodness-of-fit test was
performed on the most parameterised model to assess fit of the
model to the data. A parametric bootstrap procedure was used
to assess the structure of the model by estimating an over-
dispersion factor (^c) (MacKenzie and Bailey 2004). Because
the over-dispersion parameter was >1, the quasi Akaike
information criterion (QAIC) was used for model selection
procedures, with the parameter standard errors adjusted by ^c
(Burnham and Anderson 2002). Furthermore, because the ratio of
observations to model parameters was <40, the second-order
QAIC (QAICc) was used to rank the relative importance of
each model against the top-ranked model (Burnham and
Anderson 2002). DQAICc and the evidence weight for each
model were used to determine the most parsimonious model(s)
in the set, as follows:
DQAICi ¼ QAICi  QAICj ; and

Table 1. Suite of candidate models used for examining the influence of

evidence weighti ¼ weightj =weighti ;
site- and landscape-level factors on wetland occupancy by Chelodina
longicollis, and relative abundance of C. longicollis, in the greater
where j denotes the model selection weight of the most
Melbourne area
Variable terms: road = road density; river = distance to river; wetland = inter-
parsimonious model, and i is the model of interest. Models
wetland distance; area = wetland area; SQQ = heavy metal contamination; with a DQAICc of 2 have substantial support as an
NTU = turbidity; cond = electrical conductivity; shrubs = % riparian shrub approximating model (Burnham and Anderson 2002).
cover; herbs = % riparian herbaceous cover; island = presence/absence of To examine the influence of site- and landscape-level factors
an island; submerged = presence/absence submerged vegetation on relative abundance, catch-per-unit-effort (CPUE) was used
to minimise bias resulting from differences in detection
Model Variables probabilities among sites, which may be influenced by
Wetland-scale models trapping effort. CPUE was derived from the total number of
Vegetation structure (Shrubs, herbs, submerged) unique individuals captured at a site divided by the total
(Island, shrubs, herbs) number of traps used at the site. Detection probability of the
Wetland structure (Area, island, shrubs, herbs) species owing to unmeasured factors was assumed to be
(Area, submerged, shrubs, herbs) relatively constant across sampling sites. A negative binomial
Pond chemistry (Cond, pH, NTU) model was fitted to the CPUE data in a Bayesian framework
(Cond, pH) because there was a large amount of statistical dispersion,
Wetland pollution (Cond, pH, NTU, SQQ)
with the variance greater than the mean (see Hilbe 2011).
Landscape-scale models All candidate models (Table 1) were fitted using the free
Landscape connectivity (Road, river, wetland) software WinBUGS 1.4.3 (Spiegelhalter et al. 2003).
(River, wetland) Uninformative priors were used for all model parameters, and
Combined landscape–wetland models 100 000 iterations of three replicate Markov chains were used
Connectivity, wetland size (Road, river, wetland, area, SQQ) to estimate the posterior distribution for each parameter.
and pollution Convergence of the three Markov chains was assessed using
Connectivity and pollution (Road, river, wetland, SQQ) the Brooks–Gelman–Rubin statistic (Brooks and Gelman 1998)
Connectivity and wetland size (Road, river, wetland, area) and visual inspection of the chain histories. Convergence was
Wetland pollution (Road, SQQ) achieved within 10 000 samples. The deviance information
Connectivity and water chemistry (River, cond, pH, NTU)
criterion (DIC) was used to assess comparative model fit to
Connectivity and vegetation structure (River, wetland, shrubs, herbs)
the data (Spiegelhalter et al. 2002).
Factors influencing occurrence of a freshwater turtle Wildlife Research 167

The mean and standard deviations of the model coefficients
were calculated for each model, along with the 95% Bayesian
credible interval, akin to a frequentist 95% confidence interval
when an uninformed Bayesian prior is used (McCarthy 2007).
The relative importance of the explanatory variables was assessed
by calculating the multiplicative effect with the 95% credible
interval across the range of the variable. The multiplicative effect
(Ei) is calculated by taking the exponent of the parameter
coefficient (bi) that has been standardised by multiplying by
the range of the Variable I (Parris 2006), as follows:
E i ¼ expðbi  Rangei Þ:
Variables with a multiplicative effect that deviates from 1
are likely to have a biologically important effect on the relative
abundance of turtles at a site. Values <1 correspond to a negative
effect on the response variable, whereas values >1 correspond to
a positive effect on the response variable, and the 95%
credible interval represents the uncertainty in the effect size
(Parris 2006).
of C. longicollis at wetland sites. The candidate models were
largely indistinguishable from each other (DDIC < 2.0;
Table 3). However, there were three variables, namely
electrical conductivity, distance to river and heavy metal
pollution (SQQ), that had support across the model set. Three
of four models had support for a negative effect of SQQ on
relative abundance (Fig. 2). The mean multiplicative effect
of SQQ ranged from 0.11 to 0.12, corresponding to 11–12%
reduction in turtle abundance at sites with the greatest SQQ
value relative to the site with the lowest SQQ value. Likewise,
there was support for a negative effect of electrical conductivity
on relative abundance in three of four models (Fig. 2). The mean
multiplicative effect of electrical conductivity on abundance
ranged from 0.18 to 0.23. There was support for a positive
effect of the distance to the nearest river on relative abundance
in three of seven models (Fig. 2). The mean multiplicative
effect of river distance on abundance corresponded to a
3.17–3.35-fold increase in turtle abundance at wetlands
furthest from rivers, relative to wetlands closest to rivers.

Results Discussion
In total, 260 C. longicollis individuals, including 33 juveniles, Our study indicated that C. longicollis is common and widespread
were captured at 47 of 55 sites. Juveniles were captured at 21 across urban Melbourne. An analysis of atlas records by Hamer
sites. There was a 13% recapture rate, and the maximum and McDonnell (2010) inferred from historical trends that the
number captured at any one site was 27. Almost equal distribution of C. longicollis may be increasing in the Melbourne
numbers of each sex were trapped across the study area (112 area. This species has broad habitat preferences, using both
females, 113 males, 2 undetermined). still and slow-moving permanent water bodies, and ephemeral
Wetland occupancy by C. longicollis was high, with a naive wetlands, and feeds on a wide variety of prey items, including
estimate of the proportion of wetlands occupied of 0.85 (47/55). macro-invertebrates, crustaceans, fish and tadpoles (Kennett
The model-derived estimate of occupancy, accounting for et al. 2009). Although C. longicollis may be in decline in the
detection probability, was 0.90  0.05 (model estimate  s.e.), Murray–Darling Basin as a result of drought-induced loss of
with a detection rate of 0.64  0.05. The null model was the most wetlands (Chessman 2011), it may be benefitting from the
parsimonious of the candidate set, with 48% of the Akaike’s
weighting (Table 2), and there was considerably less support Table 3. Model selection results using the deviance information
for other models (DQAIC >2). criterion (DIC) for the negative binomial regression models of relative
Model selection using DIC values did not identify any single abundance of Chelodina longicollis
model that was clearly better at explaining relative abundance Variable terms: road = road density; river = distance to river; wetland = inter-
wetland distance; log(area) = wetland area; SQQ = heavy metal
contamination; NTU = turbidity; cond = electrical conductivity; shrubs = %
Table 2. Model selection results of wetland occupancy by Chelodina riparian shrub cover; herbs = % riparian herbaceous cover; island = presence/
longicollis absence of an island; submerged = presence/absence submerged vegetation
Models with an Akaike weight of <0.05 are not shown for clarity.
p = probability of detection; K = number of parameters; QAICc = quasi Model Variables DIC DDIC
Akaike’s information criterion; DQAICc = QAICc – minimum QAICc. K
1 Road, river, wetland, area, SQQ 241.48 1.16
parameter count includes y and p intercepts and ^c (dispersion factor).
2 Road, river, wetland, SQQ 241.40 1.08
QAICc values are based on the over-dispersion factor of the global model
3 Road, river, wetland, area 240.85 0.53
(^c = 2.29). Variable terms: road = road density; river = distance to river;
4 Road, river, wetland 240.68 0.37
wetland = inter-wetland distance; SQQ = heavy metal contamination;
5 River, wetland, shrubs, herbs 240.80 0.48
cond = log(electrical conductivity); shrubs = % riparian shrub cover;
6 Area, island, shrubs, herbs 240.99 0.67
herbs = % riparian herbaceous cover; submerged = presence/absence
7 Cond, pH, NTU, SQQ 241.20 0.88
submerged vegetation
8 Cond, pH, NTU 240.91 0.59
9 Island, shrubs, herbs 240.82 0.50
Models K Log- QAICc DQAICc Akaike
10 Shrubs, herbs, submerged 240.34 0.02
Likelihood weight
11 River, cond, pH, NTU 240.70 0.38
y(.),p(.) 3 219.91 102.63 0.00 0.480 12 Road, SQQ 240.91 0.60
y (shrub,herb,submerg),p(.) 6 212.13 106.51 3.88 0.069 13 River, wetland 240.66 0.34
y (river,wetland),p(.) 5 217.91 106.51 3.88 0.069 14 Area, submerged, shrubs, herbs 240.42 0.10
y (cond,pH),p(.) 5 218.01 106.55 3.92 0.067 15 Cond, pH 240.66 0.34
y (road,sqq),p(.) 5 218.76 106.88 4.25 0.057 Null model (intercept only) 240.32 0.00
168 Wildlife Research D. Stokeld et al.

100.00 metal sediment contamination and low electrical conductivity.

Although the relative abundance index we used may not
account for all factors affecting detectability, the estimates
Effect size - SQQ

10.00 provide some insights into relationships with site-specific factors.

Water electrical conductivity, a measure of salinity, is an
1.00
important factor in limiting the distribution of freshwater
turtles (Dunson and Mazzotti 1989; Bower 2011). Prolonged
exposure to saline waters by freshwater turtles may result in
0.10 adverse physiological effects on individuals, with negative
impacts on populations. In laboratory studies, salinity stress
reduces growth and alters respiratory patterns in Malaclemys
0.01 terrapin, a North American estuarine turtle species (Holliday
et al. 2009). Individuals of C. longicollis have previously been
1 2 7 12
observed in rural lakes with high salinity, in small numbers, and
100.00
have had elevated plasma concentrations of urea, uric acid,
sodium and chloride (Bower 2011). Human-induced changes
Effect size - Distance to river

to water catchments can increase salinity in urban wetlands

10.00 (Lee et al. 2006), potentially threatening urban populations of
C. longicollis.
Relative abundance of C. longicollis was generally lower at
1.00 sites with high levels of heavy metal sediment contamination.
Wetlands in urban catchments receive land-based contaminants
in stormwater runoff and become sinks for a range of pollutants,
0.10
in particular, heavy metals (Mason 1996). Although the impact
of heavy metals on turtles has been little studied, aquatic
0.01 species are exposed to contaminants through ingestion of
contaminated prey and sediment, direct contact with skin and
1 2 3 4 5 11 mucous membranes, maternal transfer, and exposure of eggs to
contaminated soils (Snodgrass et al. 2008). Heavy metals
bioaccumulate in the body tissues of turtles (Bishop et al.
100.00 2010), and have been associated with immunosuppression
(Day et al. 2007) and reduced juvenile survival (Burger et al.
Effect size - Conductivity

1998; Bishop et al. 2000). Heavy metal contamination of

10.00
1.00
0.10
0.01
7 8
Model number
Fig. 2. Multiplicative effects of variables on the relative abundance of
Chelodina longicollis, with 95% credible intervals. Points represent the
estimated effect of the variable on turtle relative abundance. Effect-size
values <1 indicate increasing negative effects, whereas values >1 indicate
increasing positive effects.
creation of wetlands in and around Melbourne as part of new
urban developments, despite wetland degradation and habitat
fragmentation.
Whereas we found no evidence that occupancy by
C. longicollis was influenced by the variables we measured,
relative abundance at wetlands was influenced by both
landscape and local habitat factors. Relative abundance was
greatest at wetlands more distant from rivers, with low heavy
waterways has also been associated with decreased
macroinvertebrate diversity and abundance (Walsh et al.
2001), an important component of the diet of C. longicollis
(Chessman 1984; Georges et al. 1986). Therefore, wetlands
with high heavy metal contamination may limit the long-term
persistence of turtles in urban environments.
Although model support was not strong, there was some
evidence that C. longicollis was found in higher relative
abundance at wetlands more distant from rivers. Rivers are
11 15 potential corridors for freshwater turtles in urban environments
(Chelazzi et al. 2007), particularly where urban barriers may
reduce over-land movement between wetlands. However, Rees
et al. (2009) observed that urban barriers did not inhibit the
movements of C. longicollis in Canberra. Furthermore, there
was no support for other measures of wetland isolation (road
density and inter-wetland distance) to explain abundance, or
occupancy, in our study. However, C. longicollis often
inhabits shallow, ephemeral waterbodies remote from
permanent rivers (Chessman 1988), where it can exploit food
resources in the absence of other turtle species (Georges et al.
1986). Therefore, the pattern that we found may reflect natural
habitat preferences. Other facets of land-use composition can
also influence the occurrence of turtles (Quesnelle et al. 2013).
The type of land-use surrounding a wetland (e.g. forest,
industrial) will influence the availability of suitable terrestrial
habitat for nesting and aestivation, or influence wetland
Factors influencing occurrence of a freshwater turtle
productivity and prey availability. Therefore, the persistence of
turtles at these sites may be influenced by other measures of
habitat quality.
Human-mediated dispersal of wildlife is another consequence
of urbanisation. There are numerous accounts of C. longicollis
being released by members of the public into urban environments,
either from the pet trade or from illegal collection elsewhere in
the species natural range (D. Stokeld, pers. obs.; Nick Clemann,
Department of Environment and Primary Industries, pers. comm.;
Jon Birkett, Melbourne Zoo, pers. comm.). How frequent or
widespread this practice has been is unknown; however, it
may have significantly influenced the distribution of this
species in the greater Melbourne area. If preferential release of
turtles by people occurs at urban wetlands, with greater
accessibility or a perception of suitability for turtles, this may
have contributed to the higher relative abundance at wetlands
more distant from rivers.
Our findings, and those of Rees et al. (2009) and Roe et al.
(2011), suggest that C. longicollis is likely to persist in urban
landscapes across its range, and there is little reason to believe
this species currently requires conservation management in
these environments. However, long-term studies focussed on
demographic parameters, or survivorship, may elucidate
undetected effects of urbanisation. Elsewhere within this
species range, there is evidence it may be in decline because
of the loss of important natural wetlands and climatic changes
(Chessman 2011). Predation by introduced foxes has also been
implicated as a threat to the long-term survival of populations
(Kennett et al. 2009).
The present study is the first landscape-scale study of urban
impacts on a freshwater turtle in Australia. The findings
supported our hypothesis that C. longicollis is influenced by
factors associated with urbanisation. However, in contrast to
many studies on freshwater turtles in North America, it may
be relatively resilient to habitat degradation and habitat
fragmentation resulting from urbanisation. Reasons for this
may include (1) its opportunistic feeding behaviour (Chessman
1984), (2) its ability to occupy a wide range of lotic and lentic
water bodies (Chessman 1988; Kennett and Georges 1990)
and (3) its highly plastic behavioural response to changing
environmental conditions (Roe and Georges 2008a). Species
with a wide niche breadth, in terms of diet and habitat
preferences, such as C. longicollis, are often associated with a
greater tolerance of anthropogenic habitat alteration (Swihart
et al. 2006).
The effects of urbanisation on turtle species are likely to be
species-specific (Eskew et al. 2010). Whereas some species
may be more sensitive to habitat fragmentation and decreased
habitat quality, others may maintain high abundance even in
low-quality or polluted wetlands (DeCatanzaro and Chow-
Fraser 2010). Whereas C. longicollis may appear resilient to
urbanisation, other Australian species, such as the critically
endangered western swamp turtle (Pseudemydura umbrina)
have specialised habitat requirements (Burbidge 1981), and
different life-history traits, which will likely make them more
vulnerable to urbanisation. Currently, there are relatively few
species of turtles inhabiting Australia’s cities; however, as
urban development expands, a wider range of species may
become impacted. Therefore, further studies are required to
Wildlife Research 169
ascertain the impacts of urbanisation on this wider group of
species.
Acknowledgements
Tim Jessop assisted with statistical analyses. Chhavi Bhandari and Stefano
Canessa helped with ArcGIS. Katelyn Power assisted with fieldwork. This
study was undertaken under Permit #10005308, issued by Department of
Environment and Primary Industries, University of Melbourne Animal
Ethics Committee #0811058.1, and Melbourne Water #850/101/0208–5.
The Baker Foundation supported Andrew Hamer and Rodney van der Ree
and, together with The School of Botany at The University of Melbourne,
supported Danielle Stokeld. We thank the many land owners and managers
who allowed us to survey turtles in their wetlands.
References
Alberti, M. (2005). The effects of urban patterns on ecosystem function.
International Regional Science Review 28, 168–192. doi:10.1177/01600
17605275160
ANZECC and ARMCANZ (2000). ‘Australian and New Zealand Guidelines
for Fresh and Marine Water Quality. Vol. 1: The Guidelines. National
Water Quality Management Strategy No. 4.’ (Australian and New Zealand
Environment and Conservation Council & Agriculture and Resource
Management Council of Australia and New Zealand: Canberra.)
Bishop, C. A., Struger, J., Shirose, L. J., Dunn, L., and Campbell, G. D. (2000).
Contamination and wildlife communities in stormwater detention ponds
in Guelph and the greater Toronto area, Ontario, 1997 and 1998. Part II –
Contamination and biological effects of contamination. Water Quality
Research Journal of Canada 25, 437–474.
Bishop, B. E., Savitzky, B. A., and Abdel-Fattah, T. (2010). Lead
bioaccumulation in emydid turtles of an urban lake and its relationship
to shell disease. Ecotoxicology and Environmental Safety 73, 565–571.
doi:10.1016/j.ecoenv.2009.12.027
Bower, D. S. (2011). Conservation biology of freshwater turtles in the lower
Murray River of Australia. Ph.D. Thesis, University of Canberra.
Bower, D. S., Hutchinson, M., and Georges, A. (2012). Movement and habitat
use of Australia’s largest snake-necked turtle: Implications for water
management. Journal of Zoology 287, 76–80. doi:10.1111/j.1469-
7998.2011.00891.x
Brooks, S. P., and Gelman, A. (1998). General methods for monitoring
convergence of iterative simulations. Journal of Computational and
Graphical Statistics 7, 434–455.
Burbidge, A. A. (1981). The ecology of the western swamp tortoise
Pseudemydura umbrina (Testudines: Chelidae). Wildlife Research 8,
203–223. doi:10.1071/WR9810203
Burger, J., Carruth-Hinchey, C., Ondroff, J., McMahon, M., Gibbons, J. W.,
and Gochfeld, M. (1998). Effects of lead on behavior, growth, and
survival of hatchling slider turtles. Journal of Toxicology and
Environmental Health. Part A. 55, 495–502. doi:10.1080/00984109
8158296
Burgin, S. (2008). Comparison of sympatric freshwater turtle populations
from an urbanized Sydney catchment. Aquatic Conservation-Marine
and Freshwater Ecosystems 18, 1277–1284. doi:10.1002/aqc.945
Burnham, K. P., and Anderson, D. R. (2002). ‘Model Selection and
Multimodel Inference: a Practical Information-theoretic Approach.’
2nd edn. (Springer Science: New York.)
Cagle, F. R. (1939). A system of marking turtles for future identification.
Copeia 170–173. doi:10.2307/1436818
Chelazzi, G., Naziridis, T., Benvenuti, S., Ugolini, A., and Crivelli, A. J.
(2007). Use of river-wetland habitats in a declining population of
the terrapin (Mauremys rivulata) along the Strymon River, northern
Greece. Journal of Zoology 271, 154–161. doi:10.1111/j.1469-7998.
2006.00193.x
170 Wildlife Research
Chessman, B. C. (1984). Food of the snake-necked turtle, Chelodina
longicollis (Shaw) (Testudines, Chelidae) in the Murray Valley,
Victoria and New South Wales. Australian Wildlife Research 11,
573–578. doi:10.1071/WR9840573
Chessman, B. C. (1988). Habitat preferences of fresh-water turtles in the
Murray Valley, Victoria and New South Wales. Australian Wildlife
Research 15, 485–491. doi:10.1071/WR9880485
Chessman, B. C. (2011). Declines of freshwater turtles associated with
climatic drying in Australia’s Murray–Darling Basin. Wildlife
Research 38, 664–671. doi:10.1071/WR11108
Clevenger, A. P., and Wierzchowski, J. (2006). Maintaining and restoring
connectivity in landscapes fragmented by roads. In ‘Connectivity
Conservation’. (Eds K. R. Crooks and M. Sanjayan.) pp. 502–535.
(Cambridge University Press: New York.)
Closs, G., Downes, B. J., and Boulton, A. J. (2004). ‘Freshwater Ecology:
a Scientific Introduction.’ (Blackwell Publishing Company: Malden,
MA.)
Crawley, M. J. (2007). ‘The R Book.’ (John Wiley & Sons Ltd: West
Sussex, UK.)
Day, R. D., Segars, A. L., Arendt, M. D., Lee, A. M., and Peden-Adams, M. M.
(2007). Relationship of blood mercury levels to health parameters in
the loggerhead sea turtle (Caretta caretta). Environmental Health
Perspectives 115, 1421–1428.
De Lathouder, R., Jones, D. N., and Balcombe, S. R. (2009). Assessing
the abundance of freshwater turtles in an Australian urban landscape.
Urban Ecosystems 12, 215–231. doi:10.1007/s11252-008-0077-4
DeCatanzaro, R., and Chow-Fraser, P. (2010). Relationship of road
density and marsh condition to turtle assemblage characteristics in the
Laurentian Great Lakes. Journal of Great Lakes Research 36, 357–365.
doi:10.1016/j.jglr.2010.02.003
Dunson, W., and Mazzotti, F. J. (1989). Salinity as a limiting factor in
the distribution of reptiles in Florida Bay: a theory for the estuarine
origin of marine snakes and turtles. Bulletin of Marine Science 44,
229–244.
Eskew, E., Price, S., and Dorcas, M. (2010). Survival and recruitment of
semi-aquatic turtles in an urbanized region. Urban Ecosystems 13,
365–374. doi:10.1007/s11252-010-0125-8
ESRI (2010). ArcGIS 9.3. Environmental Systems Research Institute Inc.,
Redlands, CA, USA.
Fahrig, L. (2003). Effects of habitat fragmentation on biodiversity. Annual
Review of Ecology Evolution and Systematics 34, 487–515. doi:10.1146/
annurev.ecolsys.34.011802.132419
French, S. S., Fokidis, H. B., and Moore, M. C. (2008). Variation in stress
and innate immunity in the tree lizard (Urosaurus ornatus) across an
urban–rural gradient. Journal of Comparative Physiology. B,
Biochemical, Systemic, and Environmental Physiology 178, 997–1005.
doi:10.1007/s00360-008-0290-8
Georges, A., Norris, R. H., and Wensing, L. (1986). Diet of the freshwater
turtle Chelodina longicollis (Testudines, Chelidae) from the coastal dune
lakes of the Jervis Bay Territory. Australian Wildlife Research 13,
301–308. doi:10.1071/WR9860301
Gibbs, J. P., and Shriver, W. G. (2002). Estimating the effects of road
mortality on turtle populations. Conservation Biology 16, 1647–1652.
doi:10.1046/j.1523-1739.2002.01215.x
Giles, J. C., Kuchling, G., and Davis, J. (2008). Populations of the snake-
necked turtle Chelodina oblonga in three surburban lakes of Perth,
Western Australia. In ‘Urban Herpetology. Vol. 3’. (Eds J. Mitchell,
R. Brown and B. Bartholomew.) pp. 275–283. (Society for the Study of
Amphibians and Reptiles: Salt Lake City, UT.)
Google Inc. (2011). Google Earth (version 6). Mountain View, CA, USA.
Grgurovic, M., and Sievert, P. (2005). Movement patterns of Blanding’s
turtles (Emydoidea blandingii) in the suburban landscape of eastern
Massachusetts. Urban Ecosystems 8, 203–213. doi:10.1007/s11252-
005-4380-z
D. Stokeld et al.
Hamer, A. J., and McDonnell, M. J. (2010). The response of herpetofauna
to urbanization: inferring patterns of persistence from wildlife
databases. Austral Ecology 35, 568–580. doi:10.1111/j.1442-
9993.2009.02068.x
Harden, L. A., Price, S. J., and Dorcas, M. E. (2009). Terrestrial activity
and habitat selection of eastern mud turtles (Kinosternon subrubrum) in
a fragmented landscape: implications for habitat management of golf
courses and other suburban environments. Copeia 2009, 78–84.
doi:10.1643/CH-08-037
Hilbe, J. M. (2011). ‘Negative Binomial Regression.’ (Cambridge University
Press: New York.)
Holliday, D. K., Elskus, A. A., and Roosenburg, W. M. (2009). Impacts
of multiple stressors on growth and metabolic rate of Malaclemys
terrapin. Environmental Toxicology and Chemistry 28, 338–345.
doi:10.1897/08-145.1
Kennett, R. M., and Georges, A. (1990). Habitat utilization and its
relationship to growth and reproduction of the eastern long-necked
turtle, Chelodina longicollis (Testudinata: Chelidae), from Australia.
Herpetologica 46, 22–33.
Kennett, R., Roe, J., Hodges, K., and Georges, A. (2009). Chelodina
longicollis (Shaw 1784) – eastern long-necked turtle, common long-
necked turtle, common snake-necked turtle. In ‘Conservation Biology
of Freshwater Turtles and Tortoises: a Compilation Project of the IUCN/
SSC Tortoise and Freshwater Turtle Specialist Group, Chelonian
Research Monographs No. 5. Vol. 5’. (Eds A. Rhodin, P. Pritchard,
P. van Dijk, R. A. Saumure, K. Buhlmann, J. Iverson and R. Mittermeier.)
pp. 031.1–031.8. Available at http://www.iucn-tftsg.org/cbftt/ [verified
22 July 2014].
Lee, S. Y., Dunn, R. J. K., Young, R. A., Connolly, R. M., Dale, P. E. R.,
Dehayr, R., Lemckert, C. J., McKinnon, S., Powell, B., Teasdale, P. R.,
and Welsh, D. T. (2006). Impact of urbanization on coastal wetland
structure and function. Austral Ecology 31, 149–163. doi:10.1111/j.1442-
9993.2006.01581.x
Luiselli, L. (2004). Dietary shifts of sympatric freshwater turtles in pristine
and oil-polluted habitats of the Niger Delta, southern Nigeria. The
Herpetological Journal 14, 57–64.
MacKenzie, DI, and Bailey, LL (2004). Assessing the fit of site-occupancy
models. Journal of Agricultural Biological & Environmental Statistics 9,
300–318. doi:10.1198/108571104X3361
MacKenzie, D. I. (2006). ‘Occupancy Estimation and Modeling: Inferring
Patterns and Dynamics of Species.’ (Elsevier: Boston, MA.)
MacKenzie, D. I., Nichols, J. D., Lachman, G. B., Droege, S., Andrew Royle,
J., and Langtimm, C. A. (2002). Estimating site occupancy rates when
detection probabilities are less than one. Ecology 83, 2248–2255.
doi:10.1890/0012-9658(2002)083[2248:ESORWD]2.0.CO;2
Mason, C. F. (1996). ‘Biology of Freshwater Pollution.’ 3rd edn. (Longman
Group Limited: Harlow, Essex, England.)
McCarthy, M. A. (2007). ‘Bayesian Methods for Ecology.’ (Cambridge
University Press: Cambridge, UK.)
McDonnell, M. J., and Pickett, S. T. A. (1990). Ecosystem structure and
function along urban-rural gradients: An unexploited opportunity for
ecology. Ecology 71, 1232–1237. doi:10.2307/1938259
McKinney, M. L. (2002). Urbanization, biodiversity, and conservation.
Bioscience 52, 883–890. doi:10.1641/0006-3568(2002)052[0883:
UBAC]2.0.CO;2
McKinney, M. L. (2008). Effects of urbanization on species richness: a review
of plants and animals. Urban Ecosystems 11, 161–176. doi:10.1007/
s11252-007-0045-4
Melbourne Water (2012). ‘History of Our Rivers and Creeks.’ Available at
http://www.melbournewater.com.au/content/rivers_and_creeks/
[verified 20 April 2012].
Mitchell, J. C., and Brown, R. E. J. (2008). Urban herpetology: global
overview, synthesis, and future directions. In ‘Herpetological
Conservation. Vol. 3’. (Eds J. C. Mitchell, R. E. J. Brown and
 Factors influencing occurrence of a freshwater turtle
B. Bartholomew.) pp. 1–30. (Society for the Study of Amphibians and
Reptiles: Salt Lake City, UT.)
Parris, K. M. (2006). Urban amphibian assemblages as metacommunities.
Journal of Animal Ecology 75, 757–764. doi:10.1111/j.1365-2656.
2006.01096.x
Patrick, D. A., and Gibbs, J. P. (2010). Population structure and movements
of freshwater turtles across a road-density gradient. Landscape Ecology
25, 791–801. doi:10.1007/s10980-010-9459-0
Quesnelle, P. E., Fahrig, L., and Lindsay, K. E. (2013). Effects of habitat
loss, habitat configuration and matrix composition on declining wetland
species. Biological Conservation 160, 200–208. doi:10.1016/j.biocon.
2013.01.020
Rees, M., Roe, J. H., and Georges, A. (2009). Life in the suburbs: behavior
and survival of a freshwater turtle in response to drought and urbanization.
Biological Conservation 142, 3172–3181. doi:10.1016/j.biocon.2009.
08.019
Riley, S. P. D., Busteed, G. T., Kats, L. B., Vandergon, T. L., Lee, L. F. S.,
Dagit, R. G., Kerby, J. L., Fisher, R. N., and Sauvajot, R. M. (2005).
Effects of urbanization on the distribution and abundance of amphibians
and invasive species in southern California streams. Conservation Biology
19, 1894–1907. doi:10.1111/j.1523-1739.2005.00295.x
Roe, J. H., and Georges, A. (2007). Heterogeneous wetland complexes,
buffer zones, and travel corridors: landscape management for
freshwater reptiles. Biological Conservation 135, 67–76. doi:10.1016/
j.biocon.2006.09.019
Roe, J. H., and Georges, A. (2008a). Maintenance of variable responses for
coping with wetland drying in freshwater turtles. Ecology 89, 485–494.
doi:10.1890/07-0093.1
Roe, J. H., and Georges, A. (2008b). Terrestrial activity, movements and
spatial ecology of an Australian freshwater turtle, Chelodina longicollis,
in a temporally dynamic wetland system. Austral Ecology 33, 1045–1056.
doi:10.1111/j.1442-9993.2008.01877.x
Roe, J., Rees, M., and Georges, A. (2011). Suburbs: dangers or drought
refugia for freshwater turtle populations? The Journal of Wildlife
Management 75, 1544–1552. doi:10.1002/jwmg.219
Semlitsch, R. D., and Bodie, J. R. (2003). Biological criteria for buffer zones
around wetlands and riparian habitats for amphibians and reptiles.
Conservation Biology 17, 1219–1228. doi:10.1046/j.1523-1739.2003.
02177.x
Snodgrass, J. W., Casey, R. E., Simon, J. A., and Gangapura, K. (2008).
Ecotoxicology of amphibians and reptiles in urban environments: an
www.publish.csiro.au/journals/wr
View publication stats
Wildlife Research 171
overview of potential exposure routes and bioaccumulation. In ‘Urban
Herpetology. Vol. 3’. (Eds J. Mitchell, R. Brown and B. Bartholomew.)
pp. 177–196. (Society for the Study of Amphibians and Reptiles: Salt
Lake City, UT.)
Spiegelhalter, D., Best, N. G., Carlin, B. P., and van der Lin, A. (2002).
Bayesian measures of model complexity and fit. Journal of the Royal
Statistical Society. Series B. Methodological 64, 583–639. doi:10.1111/
1467-9868.00353
Spiegelhalter, D., Thomas, A., Best, N., and Lunn, D. (2003). ‘WinBUGS
User Manual.’ (MRC Biostatistics Unit: Cambridge, UK.)
Swihart, R. K., Lusk, J. J., Duchamp, J. E., Rizkalla, C. E., and Moore, J. E.
(2006). The roles of landscape context, niche breadth, and range
boundaries in predicting species responses to habitat alteration.
Diversity & Distributions 12, 277–287. doi:10.1111/j.1366-9516.2006.
00242.x
Turtle Taxonomy Working Group (2010). ‘Turtles of the World, 2010
Update: Annotated Checklist of Taxonomy and Synonymy,
Distribution, and Conservation Status. In “Conservation Biology of
Freshwater Turtles and Tortoises: a Compilation Project of the IUCN/
SSC Tortoise and Freshwater Turtle Specialist Group, Chelonian
Research Monographs No. 5’. (Eds A. Rhodin, P. Pritchard, P. van
Dijk, R. A. Saumure, K. Buhlmann, J. Iverson and R. Mittermeier.)
pp. 000.85–000.164. Available at http://www.iucn-tftsg.org/cbftt/
[verified 22 July 2014].
UNFPA (2007). ‘State of the World Population 2007: Unleashing
the Potential Urban Growth.’ (United Nations Population Fund:
New York.)
Walsh, C. J., Sharpe, A. K., Breen, P. F., and Sonneman, J. A. (2001). Effects
of urbanization on streams of the Melbourne region, Victoria, Australia.
I. Benthic macroinvertebrate communities. Freshwater Biology 46,
535–551. doi:10.1046/j.1365-2427.2001.00690.x
Walsh, C. J., Roy, A. H., Feminella, J. W., Cottingham, P. D., Groffman, P. M.,
and Morgan Ii, R. P. (2005). The urban stream syndrome: current
knowledge and the search for a cure. Journal of the North American
Benthological Society 24, 706–723. doi:10.1899/04-028.1
Wintle, B. A., and Bardos, D. C. (2006). Modeling species-habitat
relationships with spatially autocorrelated observation data. Ecological
Applications 16, 1945–1958. doi:10.1890/1051-0761(2006)016[1945:
MSRWSA]2.0.CO;2