Climate-mediated habitat selection in an arboreal folivore

Early View (EV): 1-EV

Climate-mediated habitat selection in an arboreal folivore

Mathew S. Crowther, Daniel Lunney, John Lemon, Eleanor Stalenberg, Robert Wheeler, George Madani, Karen A. Ross and Murray Ellis

M. S. Crowther ([email protected]) and G. Madani, School of Biological Sciences, Univ. of Sydney, Sydney, NSW 2006, Australia. – D. Lunney, E. Stalenberg, R. Wheeler, K. A. Ross and M. Ellis, Office of Environment and Heritage NSW, PO Box 1967, Hurstville, NSW 2220, Australia. DL also at: School of Veterinary and Life Sciences, Murdoch Univ., Murdoch, WA 6150, Australia. ES also at: Evolution, Ecology and Genetics, Research School of Biology, Australian National Univ., Canberra, ACT 0200, Australia. – J. Lemon, Office of the Environment and Heritage NSW, PO Box 20, Gunnedah, NSW 2380, Australia.

The decisions that animals must make to achieve a balance between quantity and quality of resources become more difficult when their habitats are patchy and differ greatly in quality across space and time. Koalas are a prime subject to study this problem because they have a specialised diet of eucalypt leaves and need to balance nutrient and water intake against toxins in the leaves, all of which can change with soil type and climate. Koalas are nocturnal and spend most of the day resting and therefore choose trees for reasons other than feeding, particularly for thermoregulation. We GPS-tracked 40 koalas over 3 yr to determine their shift in tree selection between day and night, and in relation to daily maximum temperature, in a patchy rural landscape in north-western NSW, Australia. The species, degree of shelter, diameter, height and elevation of each visited tree were recorded. We used generalised linear mixed effects models to compare tree use between day and night and maximum daily temperature. Koalas used more feed-trees during the night, and more shelter-trees during the day. They also selected taller trees with more shelter in the day compared with night. As daytime temperatures rose, koalas increasingly selected taller trees at lower elevations. Our results demonstrate that koalas need taller trees, and non-feed species with shadier/denser foliage, to provide shelter from heat. This highlights the need both for the retention of taller, mature trees, such as remnant paddock trees, and the planting of both food and shelter trees to increase habitat area and connectivity across the landscape for arboreal species. Retaining and planting trees that provide optimum habitat will help arboreal folivores cope with the more frequent droughts and heatwaves expected with climate change.

Habitat selection, as defined as the non-random preference of individuals for some habitats within a range of available habitats (Morris 2003), drives population survival and repro-ductive success (Levins 1968, Degabriel et al. 2009). Animals select their habitats on the basis of such attributes as quantity and quality of food, water availability, predation risk, and thermoregulation (Brown 1988, Cresswell et al. 2010, Tuft et al. 2011). These decisions are more difficult when the quality and quantity of these resources differ in space and time (Orians and Wittenberger 1991), particularly when affected by extreme weather (Godvik et al. 2009).

Habitat selection has been traditionally inferred from population densities in different habitats (Rosenzweig 1981, Morris 2003), but a finer method to determine habitat selection, particularly when resources vary spatially and temporally, is to utilise the movements of individuals in the population (Rhodes et al. 2005). These data have hitherto been difficult to obtain, but with advances in Global Positioning System (GPS) tracking are now increasingly common (Cagnacci et al. 2010), but are still beset with difficulties (Matthews et al. 2013). GPS-tracking data have

the advantage that they are unbiased by the researcher, but the disadvantage of non-independent data affecting resource selection functions (Manly et al. 2002). Hence the increased use of generalised linear mixed-effects models (GLMM) in resource-function analysis to account for this non-independence (Gillies et al. 2006).

Climate change is predicted to cause an increasing num-ber of impacts on wildlife such as increased frequency and intensity of extreme weather and shifts in plant phenol-ogy leading to increasing extinction risk (Hughes 2003, Thomas et al. 2004, Steffen et al. 2009, Hovenden and Williams 2010, Duval et al. 2012, Lunney and Hutchings 2012). One of the lesser-explored effects are heatwaves on the movement and habitat selection of individuals.

Koalas are a prime subject to study the effects of spatial and temporal variability in resources. They have a special-ised diet of leaves from trees of the genus Eucalyptus, and need to balance nutrient and water intake against plant secondary metabolites such as toxins and tannins (Moore and Foley 2005, Marsh et al. 2007). Plant nutritional quality can vary with soil type, climate, topography and

Ecography 36: 001–008, 2013 doi: 10.1111/j.1600-0587.2013.00413.x

© 2013 The Authors. Ecography © 2013 Nordic Society Oikos Subject Editor: Douglas A. Kelt. Accepted 23 July 2013

2-EV

plant genetics (Chaves et al. 2003, Loney et al. 2006, Andrew et al. 2010, Duval et al. 2012). Koalas select par-ticular feed trees in order to meet their nutritional require-ments (Moore and Foley 2005, Marsh et al. 2007), and these influence their movements across the landscape (Moore et al. 2010). Koalas feed for up to 5 h d–1, primarily during dusk and night time (Nagy and Martin 1985, Marsh et al. 2007) and spend most of the remaining hours resting and sleeping and therefore choosing trees for thermoregula-tion and protection from predators (Ellis et al. 2002, 2009, Marsh et al. 2007). Non-feed trees are yet to be considered equally with feed trees when defining the suitability of habitat for koalas. This gap is particularly concerning given the rising importance of climate change on folivore populations (Adams-Hosking et al. 2011a, 2012, Seabrook et al. 2011, Lunney et al. 2012a).

We used GPS-tracking to determine tree-selection by koalas in relation to time of day and to daily maximum tem-perature. For specific questions on habitat selection, GPS-tracking has advantages over other commonly used methods, such as scat searches (McAlpine et al. 2006a, b, Rhodes et al. 2006), because it can give data on an animal’s location at a precise time, whereas scats take a long time to decay (Rhodes et al. 2011). GPS-tracking has the advantage over hand-held VHF radio-tracking in that many more readings can be gained, most importantly at night when koalas are feeding. Most previous studies of koala habitat have had either no night time records of tree selection, or been restricted in the number of night readings (Moore and Foley 2005, Matthews et al. 2007), and hence missed a critical part of a koala’s tree choice. The use of GPS track-ing allows for a more thorough study of koala ecology in

relation to habitat use, and it has enabled us to test the fol-lowing hypotheses: 1) koalas select different trees between night and day, reflecting food and shelter needs respectively. 2) Koalas select different trees in relation to temperature, particularly for daytime shelter at high temperatures.

Material and methods

Study area

The study area was the Liverpool Plains surrounding the town of Gunnedah, north-western New South Wales (NSW), Australia (Fig. 1, 30°59′S, 150°16′E). The area is predomi-nantly cleared, productive agricultural land, with either remnant open woodland or restoration plantings (Martin et al. 2004). The woodland patches are dominated by Eucalyptus species that grow on alluvial soils, including poplar box E. populnea, white box E. albens, yellow box E. melliodora, Blakely’s red gum E. blakelyi, river red gum E. camaldulensis and narrow-leafed ironbark E. crebra, and white cypress Callitris glaucophylla. A NSW-wide com-munity survey of koalas in 2006 revealed that this area was unique because residents reported an increase in koala numbers, whereas residents from most other areas reported that local koala populations had stabilised or declined (Crowther et al. 2009, Lunney et al. 2009, 2012a).

Koala capture and GPS tracking

Forty koalas were caught between October 2008 and April 2011, marked with a numbered ear-tag (Leader Products,

Figure 1. Location of koalas used in study, as represented by minimum convex polygons (shaded areas), on the Liverpool Plains, New South Wales, Australia.

3-EV

Craigieburn, VIC) and fitted with Sirtrack GPS collars (Sirtrack, Hawkes Bay, New Zealand). The dataset used in the analysis of tree choice comprised the locations from the GPS collars. For the first 2 seasons, in 2008/2009 and 2009/2010, collars took 1 diurnal reading (17:00) followed by 4-h readings through the night (21:00, 01:00 and 05:00). In the third season (2010/2011), the diurnal readings were increased to 3 (09:00, 13:00, 17:00; i.e. every 4 h).

We determined the accuracy of the GPS collars to account for any positional error around each koala collar reading. In a separate trial, we placed 7 collars on a trigono-metric survey mark in Gunnedah (231°01′47.05717″S, 150°16′04.32014″E) and left them to record 4-h for 7 consecutive days (6 readings d21). The positional error (distance between each GPS-collar reading and the true position of the survey mark) was calculated using the Hawthstools ‘distance between points’ calculator in ArcMap 9.3 (ESRI, Redlands, CA). It has been recommended that readings with higher Horizontal Displacement of Precision (HDOP) values, or fewer satellites, should be dis-carded as they are more likely to be inaccurate (Moen et al. 1997, D’Eon and Delparte 2005, Cargnelutti et al. 2007, Lewis et al. 2007). Examination of the relationships between positional error, HDOP and number of satellites in a collar trial indicated that it would be sensible to include readings irrespective of the number of satellites, and use HDOP alone to predict spatial accuracy for the 2-dimensional fixes required in this koala study.

As there were far more koala collar readings (n 23 000) than we had resources to sample their trees, we concen-trated on those with HDOP 4 as more likely to be accurate. Using the data from the stationary collars, dis-tance root mean square error (dRMS) was calculated from the standard deviations of the eastings and northings of the positional fixes as 9.05 m. As a measure of uncertainty around each collar reading, we used 1 dRMS which defines an area within which there is a 68% probability that an observation will fall. We also identified those days when a collared koala had remained in a single location when there were 3 diurnal readings (triplet readings) with daily deviations of 10 m (and HDOP 4). We then calcu-lated the centroid of each triplet and used these as more accurate diurnal positions, instead of the triplet readings individually. This allowed us to load relatively more accu-rate koala collar points and diurnal triplet centroids, each buffered with a circular polygon representing the GPS position and its associated error, into ArcGIS 9.3 and onto a Personal Digital Assistant (PDA) (Hewlett Packard, Palo Alto, CA) for use in the field. We loaded 639 diurnal triplet centroids, and 15 849 individual collar readings (both day and night) from the 40 koalas.

Tree use

The data on tree characteristics were collected in the field in October 2010, April and September 2011. Daily maximum temperature (Bureau of Meteorology 2012, www.bom.gov.au ) and a day or night category was allocated to each koala collar reading in the PDA. To include both day and

night readings, and the range of temperatures experienced by the koalas, we visited selected sites, based on the GPS readings from different koalas, on those rural properties where records had been made on a number of different times, days, months and years during the tracking period. GPS col-lar readings and centroids were located using the hand-held PDA with GPS. At each selected reading or centroid, we measured the characteristics of trees located within the circu-lar buffer of uncertainty around the central GPS point shown on the PDA.

For each tree, we recorded the GPS location, species, diameter at breast height (DBH) at 1.3 m, tree height, and visually scored the shelter available to koalas using a 3-level ordinal scale (low, medium and high, coded as 1, 2 and 3). For collar readings located in denser vegetation, a differential GPS (Trimble Navigation, Sunnyvale CA) was used in place of the PDA to record tree data and obtain more accurate GPS locations of the trees within the buffer. We extracted elevation for each tree from a Digital Elevation Model (DEM) with a resolution of 25 m. There were 913 measured trees used in the analyses.

Statistical analysis

We assigned the attributes of trees (location, species, eleva-tion, DBH, height and shelter) to koala GPS readings using the Spatial Join tool in ArcMap 9.3 with a ‘join-one-to-many’ operation. Where a number of trees were located within a single collar buffer or centroid buffer, these trees were all matched to the single collar reading (mean 1.62, range 1–8 trees within a single collar buffer). However, the majority of collar readings (2614 or 64%) had only 1 tree in the buffer, and only 3 collar readings had the maximum of 8 trees. A further 13% of collar readings (n 560) had multiple trees of the same species group. Thus only 23% of collar readings were affected by uncertainty associated with multiple trees, which would have had only a minor impact on subsequent analyses.

Some trees were located within a number of overlapping collar buffers, and these trees were duplicated to assign tree attributes to each unique collar reading (average 9.24, range 1–152 collar readings for a single tree). Duplicated trees therefore represent trees that were used for longer time periods, or on a number of different occasions, or by differ-ent koalas. Consequently, we analysed the data correspond-ing to 821 unique trees and 4089 unique koala collar or centroid readings.

To compare differences between the trees used by day and those used by night, we used a generalised linear mixed model (McCulloch and Searle 2001) in the R package ‘glmmML’ (Broström and R Development Team 2008) to account for the non-independence of the measurements for each koala (Gillies et al. 2006, Fieberg et al. 2010). A binary distribution (day versus night) was the response variable, and the factors used were elevation (continuous), DBH (continuous), tree height (continuous), whether the tree was a eucalypt or not (categorical), and the amount of shelter (ordinal). An individual koala was treated as the ran-dom factor. All combinations of variables were used, after they were tested for co-linearity at |r| 0.7 (Dormann et al.

4-EV

Tree selection: temperature

There were 32 models evaluating the effect of high tempera-ture on koala tree use, with 3 in the 95% confidence set (Σwi 0.95); only 1 model had a Dc 2 (Table 3). As the maximum daily temperature increased, koalas selected taller trees at lower elevations, and selected non-eucalypt over eucalypt species (Table 3). DBH was not in any of the top models. There was a significant difference in tempera-tures between used trees for each koala (MStree species

(koala) 113.093, F195,6422 6.850, p 0.001), with koalas using relatively more angophoras (Angophora floribunda) and casuarina when the temperatures were warmer compared to eucalypt species (Fig. 2).

Discussion

Both hypotheses were supported; tree selection by koalas was influenced by time of day (daytime or night time), and tree selection was influenced by high temperatures. The selection of eucalypts at night is consistent with the knowledge that koalas feed at night, in this case the selec-tion of known feed-tree species (red gums and box euca-lypts). In contrast, the daytime tree selection reflects their need to seek shelter during the day rather than feed trees. As temperatures rise to critical and lethal heatwave levels (i.e. above 35°C, Lunney et al. 2012a), koalas give highest priority to thermoregulation by selecting taller trees with higher shelter and at lower elevations (e.g. in sheltered gul-lies). Hence management actions in arboreal folivore con-servation programs must include this broader dimension of habitat and climate refuge.

Holyoak et al. (2008) identified that most published movement studies merely measure and describe the move-ment of organisms, without reference to the ecological or internal factors that drive the movement. By comparing the koalas’ movements to external factors (day/night and temperature), we have applied part of the conceptual framework of the movement ecology paradigm as recom-mended by Nathan et al. (2008). In doing so, studies such as this one add to the general understanding of the causes, mechanisms and spatiotemporal patterns of movement and their role in ecological processes (Nathan et al. 2008).

Tree selection for shelter

Many species vary their habitat requirements between day and night, although most do it for protection from predators (Moreno et al. 1996). It had been expected that

2013), and the best models were selected using the corrected Akaike’s information criterion (AICc) (Burnham and Anderson 2002) using the R package ‘MuMIn’ (Barton and R Development Team 2009). A model was considered useful if the difference in AIC of that model from the best model (Dc) was 2 (Burnham and Anderson 2002).

To further test the species of tree used by the koalas, we used a log-linear mixed effects model with day/night and tree species as fixed factors, and an individual koala as the random factor. Due to the large number of eucalypt species, we grouped them into 5 categories for analysis: red gums (E. blakelyi and E. camaldulensis), box eucalypts (E. populnea, E. albens and E. melliodora), ironbark (E. crebra), eucalypts introduced to the area.

To investigate the use of trees in relation to daily maxi-mum temperature, we again used a generalised linear mixed model with temperature as the response variable with a Gaussian distribution in the R package, ‘nlme’ (Pinheiro et al. 2008). All combinations of variables were used, and the best models were selected using AICc (Burnham and Anderson 2002) using the R package ‘MuMIn’ (Barton and R Development Team 2009). Again, a model was considered useful if Dc 2 (Burnham and Anderson 2002). We then tested, using a nested ANOVA, the differences in maximum temperature with the use of individual tree species, with tree species nested within individual koalas in SPSS ver. 20 (SPSS, IBM, Armonk, NY, USA).

Results

Tree selection: day versus night

There were 32 models comparing day versus night tree use by koalas, with 6 in the 95% confidence set (Σwi 0.95). Only 3 models had a Dc 2 (Table 1). Koalas selected more non-eucalypt species during the day and more eucalypt species at night. They also selected taller trees with more shelter during the day (Table 1). DBH was considered in only 1 of the top 3 models, although koalas may be using trees with a larger DBH during the day (Table 1).

Koalas used relatively more box eucalypts and red gums during the night compared to day, and relatively more Brachychiton and casuarina Casuarina cristata during the day compared to night, as indicated by the significant inter-action between day and night and these species (Table 2). Koalas used cypress (Callitris glaucophylla), but there was no difference between day and night. This was the same for introduced eucalypts and ironbarks.

Table 1. Coefficients and standard errors of the 3 best generalised mixed models, with AICc values, change in AICc values (Dc) and Akaike weight (wi) comparing koala use of day versus night trees. Positive coefficients indicated that used trees tended to have higher values of that factor during the day; (2) indicates factor was not in model.

(Intercept) DBH ElevationNon-eucalypt

versus eucalypt Height Shelter Log-likelihood AICc Dc wi

21.686 – – 0.863 0.085 0.029 0.008 0.093 0.039 23852.941 7715.891 0 0.34021.682 0.002 0.001 – 0.873 0.085 0.024 0.009 0.084 0.040 23852.400 7716.813 0.922 0.21422.141 – 0.001 0.003 0.863 0.084 0.030 0.008 0.095 0.039 23852.838 7717.688 1.797 0.13822.103 0.002 0.001 0.001 0.003 0.869 0.084 0.024 0.009 0.084 0.040 23852.300 7718.624 2.733 0.087

5-EV

models developed from principles of biophysics (Nash 1998, Porter et al. 2008, Kearney and Porter 2009) to pre-dict the distribution of another folivorous marsupial, the greater glider Petauroides volans, in response to climate change. They predicted that the species will undergo con-siderable range contractions, arising largely from increased temperatures causing water loss from evaporative cooling. Unlike the greater glider and other arboreal marsupials, with rare exceptions, koalas do not shelter in tree hollows by day and are thus more exposed to high temperatures and other extreme weather conditions. Habitat selection at the scale of individual trees is the only recourse for a heat-stressed koala. Our findings show the importance of shelter trees in tree selection by koalas and identify that these shelter trees critically affect their capacity to with-stand climate change, particularly in hot years. This has major implications for the proposed restoration of koala habitat, such as after the extensive coal mining proposed for the Liverpool Plains (Lunney et al. 2012b).

Tree selection for feeding

Koalas selected known feed-tree species at night, favouring box eucalypts and red gums, consistent with the view that koalas feed primarily at night. Why koalas choose 1 tree over others within a species at night requires more detailed observations of feeding and analysis of leaf chemistry of individual trees (Moore and Foley 2005, Moore et al. 2005). Koalas face particular challenges when selecting suitable food, and their choices depend on their nutritional needs at the time. Eucalypt leaves have variable concentrations of essential nutrients, and contain toxic plant secondary metabolites which can vary with soil, climate, topography and plant genetics (Chaves et al. 2003, Loney et al. 2006, Andrew et al. 2010, Duval et al. 2012). Some plant compounds, such as plant secondary metabolites, act as feeding deterrents by reducing the palatability of plants or reducing the availability of nutrients (Robbins et al. 1987, DeGabriel et al. 2008). Koalas favour particular species of eucalypt over others, but trees of the same species can have very different leaf chemistry, and thus palatability, even within a small cluster of trees (Moore and Foley 2005, Andrew et al. 2010, Moore et al. 2010).

Koalas changed their use of trees between night and day and, with increasing temperature, away from food trees towards non-food trees, and at lower elevations. Hence koalas have to sacrifice dietary intake for shelter in response to high temperatures and to ensure that they meet their dietary intakes at night when temperatures are cooler. As temperatures continue to rise, this window of feeding will shorten as koalas seek shelter rather than food and reduce

trees selected by koalas by day would differ from those selected by night (Hindell et al. 1985, Matthews et al. 2007) and we have confirmed that view. Koalas are nocturnal and sleep by day among the foliage in the trees. Their dense fur provides some protection to temperature extremes (Degabriele and Dawson 1979) and they are also able to regulate their exposure by changing positions in the tree (Matthews et al. 2007), but there is a limit to their capacity to cope with extreme weather conditions (Gordon et al. 1988). Exposure to prolonged high temperatures can result in heat stress, dehydration, and eventually death of individuals. Approximately a quarter of the Gunnedah koala population perished as a result of a series of heatwaves in late November and early December 2009, just before the end a prolonged drought (Lunney et al. 2012a). The impact of these heatwaves showed that shelter is critical dur-ing heatwaves if death from dehydration is to be avoided.

Koala distribution has long been affected by climate; since the Quaternary the range of koalas contracted to areas with climate refugia during times of glacial maxima (Adams-Hosking et al. 2011a, b). Future anthropogenic climate change is predicted to result in higher tempera-tures, lower rainfall, and increased frequency and severity of extreme events, such as heatwaves, throughout eastern Australia (CSIRO and Bureau of Meteorology 2007, Pittock 2009). The ability of arboreal animals to find refuge from extreme conditions will impact the long- term survival of local populations and will limit their dis-tribution across a landscape.

Understanding the physiological responses of animals to particular environmental conditions is important to make robust predictions about how they will cope with cli-mate change. Kearney et al. (2010) used mechanistic niche

Table 2. Choice of tree species by koalas, using a mixed-effects log-linear model. A positive estimate indicates a relatively higher use of that tree species during the night, while a negative estimate indicates a relatively higher use by day.

Estimate SE Z value p

Angophora day versus night 20.073 0.313 20.232 0.817Box day versus night 0.426 0.203 2.093 0.036Brachychiton day versus night 20.797 0.272 22.929 0.003Callitris day versus night 0.095 0.219 0.434 0.665Casuarina day versus night 20.390 0.261 21.493 0.135Eucalypt (introduced)

day versus night20.185 0.232 20.798 0.425

Eucalypt (unknown) day versus night

20.118 0.254 20.463 0.643

Geijera day versus night 0.071 0.242 0.291 0.771Ironbark day versus night 20.265 0.231 21.146 0.252Other day versus night 20.102 0.334 20.306 0.759Red gum day versus night 0.689 0.208 3.312 0.001

Table 3. Coefficients and standard errors of the 2 best generalised mixed models, with AICc values, change in AICc values (Dc) and Akaike weight (wi) comparing tree use to temperature. Positive coefficients indicated that used trees tended to have higher values of that factor at higher temperatures; (–) indicates factor was not in model.

(Intercept) DBH ElevationNon-eucalypt vs

eucalypt Height Shelter Log-likelihood AICc Dc wi

37.741 – 20.031 0.008 0.520 0.154 0.117 0.014 – 218931.230 37874.473 0 0.81637.312 – 20.030 0.008 0.490 0.158 0.116 0.0142 0.062 0.074 218932.567 37879.151 4.678 0.079

6-EV

south-western Queensland have shown the impact of drought on habitat selection, including tree choice (Seabrook et al. 2011, Smith et al. 2013a, b).

GPS methods have the advantage over radio-tracking in the number of simultaneous positions that can be obtained, but have the disadvantage of the accuracy of the readings when being used for fine-scale habitat selection, such as the use of individual trees. Our study was successful because the landscape configuration was largely well-spaced trees on agricultural land, with most areas of limited tree species and size diversity. Habitat-selection studies of arboreal fauna in heavily forested areas, particularly areas of high floral diversity (Campbell and Sussman 1994, Wilson et al. 2007), would need to consider radio-tracking on foot.

Conclusions

Koalas use feed trees by night and shelter trees by day, and select taller trees with more shelter, at lower elevations, and more often non-eucalypt species, as daytime temperatures increase to heatwave levels. Thus, tree choice by koalas goes beyond feed trees and crucially includes trees that provide shelter, particularly under extreme weather conditions. The management implication is that the long-standing emphasis on retaining or planting feed trees for koalas needs to be expanded to include shelter trees, and these trees need to include the mix of trees within the home range of each koala. There is a particular need for trees in the sheltered gullies for protection from heat and possibly to provide higher leaf water content. Climate change will cause major impacts on many tree-dependent fauna, par-ticularly in fragmented rural landscapes. Our study high-lights the need for both the retention of taller, mature trees such as remnant paddock trees (as koalas selected taller trees

their activity to reduce the risk of heat stress. Ellis et al. (2010) found that non-food trees were 2°C cooler than food trees for koalas. In addition, plant nutritional quality is predicted to decline due to elevated atmospheric CO2 concentrations (Kanowski 2001, Williams et al. 2003, Duval et al. 2012, Lunney et al. 2012a), compelling folivores to travel further to seek suitable foods, or to eat more and thus consume more toxins, to meet their dietary needs (Wiggins et al. 2006). These aspects of climate change will result in escalated threats for folivorous mammals, such as the koala (Adams-Hosking et al. 2011a, 2012, Lunney and Hutchings 2012).

Implications for future research: methods

The contrast between trees used by day and night draws attention to the limitation of studies of daytime movements of koalas. Studies restricted solely or largely to daytime movements, or those based on the collection of scats from under trees, are not able to determine all the critical elements of koala habitat preferences. Koala scats can last for months, depending on location and humidity (Rhodes et al. 2011), so koala scats under a tree do not allow analyses to be made of the time, date, or use of the tree by a koala. Habitat analyses based on scats will not pick up the critical resource of the shelter trees, and especially why a non-eucalypt was selected. This matter will be of rising importance as climate change progresses, forcing further changes in daily habitat selection by koalas. The optimum approach lies in combin-ing the 2 field techniques – the labour-intensive GPS collar research with the economical, scat-based surveys that can cover wide areas rapidly and enable landscape ecological analyses to detect broad patterns of habitat selection over entire regions. The Liverpool Plains is at the western, drier edge of the koala’s range, and scat-based studies in semi-arid

Figure 2. Mean temperature standard errors compared to relative use by koalas of different tree species.

7-EV

Campbell, A. F. and Sussman, R. W. 1994. The value of radio tracking in the study of Neotropical rain forest monkeys. – Am. J. Primatol. 32: 291–301.

Cargnelutti, B. et al. 2007. Testing Global Positioning System performance for wildlife monitoring using mobile collars and known reference points. – J. Wildl. Manage. 71: 1380–1387.

Chaves, M. M. et al. 2003. Understanding plant responses to drought – from genes to the whole plant. – Funct. Plant Biol. 30: 239–264.

Cresswell, W. et al. 2010. Predator-hunting success and prey vulnerability: quantifying the spatial scale over which lethal and non-lethal effects of predation occur. – J. Anim. Ecol. 79: 556–562.

Crowther, M. S. et al. 2009. Using broad-scale, community survey data to compare species conservation strategies across regions: a case study of the koala in a set of adjacent ‘catchments’. – Ecol. Manage. Restor. 10: s88–s96.

CSIRO and Bureau of Meteorology 2007. Climate change in Australia: technical report 2007. – CSIRO.

DeGabriel, J. L. et al. 2008. A simple, integrative assay to quantify nutritional quality of browses for herbivores. – Oecologia 156: 107–116.

Degabriel, J. L. et al. 2009. The effects of plant defensive chemistry on nutrient availability predict reproductive success in a mammal. – Ecology 90: 711–719.

Degabriele, R. and Dawson, T. J. 1979. Metabolism and heat-balance in an arboreal marsupial, the koala (Phascolarctos cinereus). – J. Comp. Physiol. 134: 293–301.

D’Eon, R. G. and Delparte, D. 2005. Effects of radio-collar position and orientation on GPS radio-collar performance, and the implications of PDOP in data screening. – J. Appl. Ecol. 42: 383–388.

Dormann, C. F. et al. 2013. Collinearity: a review of methods to deal with it and a simulation study evaluating their performance. – Ecography 36: 27–46.

Duval, B. D. et al. 2012. CO2 effects on plant nutrient concentration depend on plant functional group and available nitrogen: a meta-analysis. – Plant Ecol. 213: 505–521.

Ellis, W. et al. 2010. Climate change and the koala Phascolarctos cinereus: water and energy. – Aust. Zool. 35: 369–377.

Ellis, W. A. H. et al. 2002. Tree use, diet and home range of the koala (Phascolarctos cinereus) at Blair Athol, central Queensland. – Wildl. Res. 29: 303–311.

Ellis, W. A. H. et al. 2009. Spatiotemporal dynamics of habitat use by koalas: the checkerboard model. – Behav. Ecol. Sociobiol. 63: 1181–1188.

Fieberg, J. et al. 2010. Correlation and studies of habitat selection: problem, red herring or opportunity? – Phil. Trans. R. Soc. B 365: 2233–2244.

Gillies, C. S. et al. 2006. Application of random effects to the study of resource selection by animals. – J. Anim. Ecol. 75: 887–898.

Godvik, I. M. R. et al. 2009. Temporal scales, trade-offs, and functional responses in red deer habitat selection. – Ecology 90: 699–710.

Gordon, G. et al. 1988. A koala (Phascolarctos cinereus Goldfuss) population crash during drought and heatwave conditions in south-western Queensland. – Aust. J. Ecol. 13: 451–461.

Hindell, M. A. et al. 1985. Tree species selection by free- ranging koala populations in Victoria. – Aust. Wildl. Res. 12: 137–144.

Holyoak, M. et al. 2008. Trends and missing parts in the study of movement ecology. – Proc. Natl Acad. Sci. USA 105: 19060–19065.

Hovenden, M. J. and Williams, A. L. 2010. The impacts of rising CO2 concentrations on Australian terrestrial species and ecosystems. – Austral Ecol. 35: 665–684.

with more shelter during higher temperatures), as well as the planting of more trees to increase habitat area and connectivity across these fragmented landscapes. The con-servation implications of this case study are considerable, especially for restoration projects that focus solely on food trees. The ecological significance of this finding is that there is a large suite of obligate arboreal fauna that will be influ-enced by climate-mediated selection. Further, given that a day/night difference was demonstrated in this study, it points to the importance of ensuring that comprehensive analyses of habitat include day/night studies. Since climate change is occurring it is necessary to include weather vari-ables in interpreting habitat selection by arboreal fauna.

Acknowledgements – We remain indebted to Liverpool Plains Land Management and its executive officer David Walker, the NSW Environmental Trust (Restoration), and Angus Robinson and the Foundation for National Parks and Wildlife for their sustained support for this project. The assistance and co-operation of many landholders in the Gunnedah Shire who helped us with this project is very much appreciated. Allan Peatman provided expert assistance with the use of a differential GPS to capture tree position data. Anni Blaxland-Fuad assisted with design of the PDA form and use of ArcPad, and Chris Togher provided expert GIS advice. We are also indebted to Gunnedah veterinarian David Amos for his sustained interest in this research. Will Dorrington provided access to geo-referenced aerial photographs. We are also most appreciative of many co-workers and volunteers for their assistance during the project, particularly Mark Krockenberger, Jo Griffith, Mel Retamales, Corinna Orscheg, Harry Parnaby, Jessica Bryant, Libby Dwyer, Becky Usmar and Sarita Guy. Ben Moore provided critical comments on the draft. The Animal Ethics Committee approved protocol was from the Office of Environment and Heritage NSW (080211/02).

References

Adams-Hosking, C. et al. 2011a. Modelling climate- change-induced shifts in the distribution of the koala. – Wildl. Res. 38: 122–130.

Adams-Hosking, C. et al. 2011b. Modelling the potential range of the koala at the last glacial maximum: future conservation implications. – Aust. Zool. 35: 983–990.

Adams-Hosking, C. et al. 2012. Modelling changes in the distribution of the critical food resources of a specialist folivore in response to climate change. – Divers. Distrib. 18: 847–860.

Andrew, R. L. et al. 2010. Genetic and environmental contributions to variation and population divergence in a broad- spectrum foliar defence of Eucalyptus tricarpa. – Ann. Bot. 105: 707–717.

Barton, K. and R Development Team 2009. MuMIn: multi-model inference. – R package ver. 0.12.0.

Broström, G. and R Development Team 2008. glmmML: generalized linear models with clustering. – R package ver. 0.81-2.

Brown, J. S. 1988. Patch use as an indicator of habitat preference, predation risk, and competition. – Behav. Ecol. Sociobiol. 22: 37–47.

Burnham, K. P. and Anderson, D. R. 2002. Model selection and multimodel inference: a practical information-theoretic approach. – Springer.

Cagnacci, F. et al. 2010. Animal ecology meets GPS- based radiotelemetry: a perfect storm of opportunities and challenges. – Phil. Trans. R. Soc. B 365: 2157–2162.

8-EV

Moreno, S. et al. 1996. Cover is safe during the day but dangerous at night: the use of vegetation by European wild rabbits. – Can. J. Zool. 74: 1656–1660.

Morris, D. W. 2003. Toward an ecological synthesis: a case for habitat selection. – Oecologia 136: 1–13.

Nagy, K. A. and Martin, R. W. 1985. Field metabolic rate, water flux, food consumption and time budget of koalas, Phascolarctos cinereus (Marsupialia, Phascolarctidae) in Victoria. – Aust. J. Zool. 33: 655–665.

Nash, L. T. 1998. Vertical clingers and sleepers: seasonal influences on the activities and substrate use of Lepilemur leucopus at Beza Mahafaly Special Reserve, Madagascar. – Folia Primatol. 69: 204–217.

Nathan, R. et al. 2008. A movement ecology paradigm for unifying organismal movement research. – Proc. Natl Acad. Sci. USA 105: 19052–19059.

Orians, G. H. and Wittenberger, J. F. 1991. Spatial and temporal scales in habitat selection. – Am. Nat. 137: S29–S49.

Pinheiro, J. et al. 2008. nlme: linear and nonlinear mixed effects models. – R package ver. 3.1-88.

Pittock, A. B. 2009. Climate change: turning up the heat. – CSIRO Publishing.

Porter, W. P. et al. 2008. Modeling large mammals on real landscapes. – Molecules to Migration, Fourth International Conference in Africa for Comparative Physiology and Biochemistry.

Rhodes, J. R. et al. 2005. A spatially explicit habitat selection model incorporating home range behavior. – Ecology 86: 1199–1205.

Rhodes, J. R. et al. 2006. Modeling species’ distributions to improve conservation in semiurban landscapes: koala case study. – Conserv. Biol. 20: 449–459.

Rhodes, J. R. et al. 2011. The consequences of using indirect signs that decay to determine species’ occupancy. – Ecography 34: 141–150.

Robbins, C. T. et al. 1987. Role of tannins in defending plants against ruminants: reduction in protein availability. – Ecology 68: 98–107.

Rosenzweig, M. L. 1981. A theory of habitat selection. – Ecology 62: 327–335.

Seabrook, L. et al. 2011. Drought-driven change in wildlife distribution and numbers: a case study of koalas in south west Queensland. – Wildl. Res. 38: 509–524.

Smith, A. G. et al. 2013a. At what spatial scales does resource selection vary? A case study of koalas in a semi-arid region. – Austral Ecol. 38: 230–240.

Smith, A. G. et al. 2013b. Out on a limb: habitat use of a specialist folivore, the koala, at the edge of its range in a modified semi-arid landscape. – Landscape Ecol. 28: 415–426.

Steffen, W. et al. 2009. Australia’s biodiversity and climate change. – CSIRO Publishing.

Thomas, C. D. et al. 2004. Extinction risk from climate change. – Nature 427: 145–148.

Tuft, K. D. et al. 2011. Predation risk and competitive interactions affect foraging of an endangered refuge-dependent herbivore. – Anim. Conserv. 14: 447–457.

Wiggins, N. L. et al. 2006. Spatial scale of the patchiness of plant poisons: a critical influence on foraging efficiency. – Ecology 87: 2236–2243.

Williams, S. E. et al. 2003. Climate change in Australian tropical rainforests: an impending environmental catastrophe. – Proc. R. Soc. B 270: 1887–1892.

Wilson, R. F. et al. 2007. Importance of canopy connectivity for home range and movements of the rainforest arboreal ringtail possum (Hemibelideus lemuroides). – Wildl. Res. 34: 177–184.

Hughes, L. 2003. Climate change and Australia: trends, projections and impacts. – Austral Ecol. 28: 423–443.

Kanowski, J. 2001. Effects of elevated CO2 on the foliar chemistry of seedlings of two rainforest trees from north-east Australia: implications for folivorous marsupials. – Austral Ecol. 26: 165–172.

Kearney, M. R. and Porter, W. P. 2009. Mechanistic niche modelling: combining physiological and spatial data to predict species’ ranges. – Ecol. Lett. 12: 334–350.

Kearney, M. R. et al. 2010. Correlative and mechanistic models of species distribution provide congruent forecasts under climate change. – Conserv. Lett. 3: 203–213.

Levins, R. 1968. Evolution in changing environments. – Princeton Univ. Press.

Lewis, J. S. et al. 2007. Effects of habitat on GPS collar performance: using data screening to reduce location error. – J. Appl. Ecol. 44: 663–671.

Loney, P. E. et al. 2006. How do soil nutrients affect within- plant patterns of herbivory in seedlings of Eucalyptus nitens? – Oecologia 150: 409–420.

Lunney, D. and Hutchings, P. 2012. Wildlife and climate change: are robust strategies for Australian fauna possible? – In: Lunney, D. and Hutchings, P. (eds), Wildlife and climate change: towards robust conservation strategies for Australian fauna. Royal Zoological Society of New South Wales, pp. 180–201.

Lunney, D. et al. 2009. Combining a map-based public survey with an estimation of site occupancy to determine the recent and changing distribution of the koala in New South Wales. – Wildl. Res. 36: 262–273.

Lunney, D. et al. 2012a. Koalas and climate change: a case study on the Liverpool Plains, north-west NSW. – In: Lunney, D. and Hutchings, P. (eds), Wildlife and climate change: towards robust conservation strategies for Australian fauna. Royal Zoological Society of New South Wales, pp. 150–168.

Lunney, D. et al. 2012b. An ecological approach to koala conservation in a mined landscape. – Proceedings Life-of-Mine 2012. The Australasian Inst. of Mining and Metallurgy, pp. 345–354.

Manly, B. F. J. et al. 2002. Resource selection by animals: statistical design and analysis for field studies, 2nd ed. – Kluwer.

Marsh, K. J. et al. 2007. Behavioural contributions to the regulated intake of plant secondary metabolites in koalas. – Oecologia 154: 283–290.

Martin, W. K. et al. 2004. The value of habitat reconstruction to birds at Gunnedah, New South Wales. – Emu 104: 177–189.

Matthews, A. et al. 2007. Tree use by koalas (Phascolarctos cinereus) after fire in remnant coastal forest. – Wildl. Res. 34: 84–93.

Matthews, A. et al. 2013. The success of GPS collar deployments on mammals in Australia. – Aust. Mamm. 35: 65–83.

McAlpine, C. A. et al. 2006a. Testing alternative models for the conservation of koalas in fragmented rural-urban landscapes. – Austral Ecol. 31: 529–544.

McAlpine, C. A. et al. 2006b. The importance of forest area and configuration relative to local habitat factors for conserving forest mammals: a case study of koalas in Queensland, Australia. – Biol. Conserv. 132: 153–165.

McCulloch, C. E. and Searle, S. R. 2001. Generalized, linear and mixed models. – Wiley.

Moen, R. et al. 1997. Accuracy of GPS telemetry collar locations with differential correction. – J. Wildl. Manage. 61: 530–539.

Moore, B. D. and Foley, W. J. 2005. Tree use by koalas in a chemically complex landscape. – Nature 435: 488–490.

Moore, B. D. et al. 2005. Eucalyptus foliar chemistry explains selective feeding by koalas. – Biol. Lett. 1: 64–67.

Moore, B. D. et al. 2010. Palatability mapping: a koala’s eye view of spatial variation in habitat quality. – Ecology 91: 3165–3176.