The habitat requirements of the rufous treecreeper (Climacteris rufa). 1. Preferential habitat use demonstrated at multiple spatial scales

Biological Conservation 105 (2002) 383–394 www.elsevier.com/locate/biocon

The habitat requirements of the rufous treecreeper (Climacteris rufa). 1. Preferential habitat use demonstrated at multiple spatial scales Gary W. Luck1 Centre for Ecosystem Management, School of Natural Sciences, Edith Cowan University, 100 Joondalup Drive, Joondalup, Western Australia 6027, Australia Received 30 May 2001; received in revised form 10 September 2001; accepted 23 September 2001

Abstract Determining the habitat requirements of a species is fundamental to effective conservation, particularly if the species is declining in areas where its habitat is being modified. Multi-scaled investigations of habitat use are essential because different selection processes may operate at different scales. I examined the habitat use of a declining woodland passerine, the rufous treecreeper (Climacteris rufa), at three spatial scales (landscape, woodland and territory) in the wheatbelt of Western Australia. Preferential habitat use was exhibited at all scales. At the landscape scale, wandoo (Eucalyptus wandoo) woodland was used at a significantly greater rate than three other common vegetation types. Territory use within woodlands was positively related to the density of hollowbearing logs, the density of nest sites, and tree age. Within an individual territory, nest sites (hollows) were favoured if they had a spout angle of 550
to the horizontal and an entrance size of between 5 and 10 cm. The rufous treecreeper preferentially used habitat with traits characteristic of old-growth wandoo woodland. Degradation of wandoo through habitat modification (e.g. grazing, logging, fire and removal of deadwood) represents a significant threat to the persistence of treecreepers. # 2002 Elsevier Science Ltd. All rights reserved. Keywords: Habitat use; Spatial scale; Hierarchical analysis; Habitat models; Australia; Climacteris rufa

1. Introduction Where an animal lives is influenced by a number of factors including habitat structure, floristics, food availability, conspecifics, interspecific competition, predation risk and phylogenetic constraints (Hilde´n, 1965; Southwood, 1977; Butler, 1980; Hutto, 1985; Rotenberry, 1985; Muller et al., 1997). The relationship between the distribution, abundance and fitness of a species and the characteristics of its habitat is an important problem in ecology (Morris, 1987; Orians and Wittenberger, 1991; Rosenzweig, 1991; Pribil and Picman, 1997; Clark and Shutler, 1999). The first step in resolving this problem is to determine if species exhibit preferential habitat use. The purpose of my study was to examine correlative relationships between the location E-mail address: [email protected] (G.W. Luck). Present address: Centre for Conservation Biology, Department of Biological Sciences, Stanford University, 371 Serra Mall, Stanford, CA, 94305-5020, USA. 1

of the rufous treecreeper (Climacteris rufa) and structural characteristics of its habitat to establish if habitat use is non-random at multiple spatial scales. Relationships between preferential habitat use and fitness are assessed in a related study (Luck, 2002). Studies of habitat use by birds have often demonstrated the importance of vegetation structure and floristics in determining distribution and abundance (Moen and Gutie´rrez, 1997; Shackelford and Conner, 1997; Micheals and Cully, 1998; Tibbetts and Pruett-Jones, 1999). The structural characteristics of a habitat provide a bird with nest and roost sites, perches, foraging substrates and protection from predators (Cody, 1985; Ford, 1989; Recher, 1991). Birds may be associated with particular habitat types or show a close affinity with a certain plant species (Rice et al., 1984; Chan, 1990; Adams and Morrison, 1993; Storch, 1993; McShea et al., 1995). Single-scale studies of bird habitat use are common, with characteristic spatial scales of investigation being habitat (or vegetation) type (Baines, 1994; Hunt, 1996), individual territories (McShea et al.,

0006-3207/02/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved. PII: S0006-3207(01)00222-1

384

G.W. Luck / Biological Conservation 105 (2002) 383–394

1995; Sodhi et al., 1999) or nest sites (Shields and Kelly, 1997). Studies conducted at only one spatial scale are limited because different factors can influence habitat use at different scales (Wiens et al., 1987; Pribil and Picman, 1997). A more useful approach is to investigate habitat use at multiple spatial scales, preferably within a nested hierarchy (Maurer, 1985; Wiens et al., 1987; Kotliar and Wiens, 1990; Bergin, 1992; Moen and Gutie´rrez, 1997; Hall and Mannan, 1999; Miller et al., 1999). This multiscale approach acknowledges the influence of spatial variation on species behaviour and recognises that there is no single correct scale at which to conduct investigations (Morris, 1987; Levin, 1992; Otis, 1997). Decisions about the appropriate scales of investigation should be based on the relevant ecological traits of the species of interest (e.g. home range size and dispersal ability) to reduce human bias in the selection process (Morris, 1987; Orians and Wittenberger, 1991). I investigated changes in the habitat use of rufous treecreepers across three spatial scales: landscape (woodland selection), woodland (territory selection) and territory (nest-site selection). The scales of investigation are relevant to specific ecological characteristics of the treecreeper. The species is primarily a woodland user, occupies all-purpose territories year round and nests in tree hollows (Luck, 2002). At the woodland and territory scale, I developed predictive habitat models, which are validated with new, independent data by Luck (this volume). Explicit information on habitat requirements contributes to our understanding of how habitat change may affect the persistence of populations. This is particularly important for the rufous treecreeper because the species has declined in abundance and appears to be vulnerable to habitat alteration (Storr, 1991; Saunders and Ingram, 1995).

2. Methods 2.1. Study area This study was conducted in a 12,283 ha remnant of native vegetation in Dryandra Woodland (centred on 32
450 S, 116
550 E), located approximately 160 km south-east of Perth, in the wheatbelt of Western Australia. The region experiences a Mediterranean climate with a mean annual rainfall of 500 mm. The landscape is gently undulating with occasional breakaway slopes and granite outcrops. Vegetation communities in the study area are characterised by open (< 30% projected foliage cover) woodlands of wandoo (Eucalyptus wandoo) and marri (Corymbia calophylla) occurring on mid–lower valley slopes, and powderbark wandoo (Eucalyptus accedens) and brown mallet (Eucalyptus astringens) occurring on upper valley slopes. Brown mallet also occurs in

relatively extensive plantations that were established from 1925 to 1962 (Department of Conservation and Land Management, 1995). Lateritic uplands support dense shrublands of Dryandra and Petrophile spp., sometimes with a jarrah (Eucalyptus marginata) overstorey. Mixed shrublands of Banksia, Grevillea, Hakea, Gastrolobium and other genera occur in isolated areas. 2.2. Woodland selection 2.2.1. Field methods At the landscape scale (the 12,283 ha remnant), I examined treecreeper use of four major woodland types, classified by the predominant tree species: wandoo, powderbark wandoo, brown mallet (including the plantations) and mixed woodland (powderbark wandoojarrah-marri). These vegetation types were the ones most likely to be used in the region based on reports in the literature (Kitchener et al., 1982; Rose, 1996; Recher and Davis, 1998), and they covered approximately 92% of the study area. In 1997, I conducted a pilot study to determine the best methods for detecting rufous treecreepers at 30 locations that were known to contain the species. Results from the study showed that initial detection of the species was primarily aural (87%), with 90% of detections occurring within 5 min. Most (85%) aural detections were of birds 4100 m from the observer with an apparent detection limit of approximately 150 m. These results were used as a guide when conducting the main study. To locate sample sites for the main study, I randomly selected sections of dirt roads from a topographic map of the study area and used a vehicle to travel a distance of 500 m from the beginning of each section of road. At this point, I classified the site into one of the four main woodland types (or other vegetation type) based on the predominant overstorey plant species occurring within a 100 m radius of the vehicle. Sites in the main woodland types were marked on a map and in the field with flagging tape to facilitate re-location. I then travelled at least another 500 m before locating the next sample site. This was repeated for each section of road. I located 200 sites (50 per woodland type), which were surveyed for the presence of treecreepers on five occasions (once per season): mid-breeding season (November 1997), summer (January 1998), autumn (April 1998), winter (July 1998) and early breeding season (September 1998). Surveys were conducted in fine weather conditions between 06:00 and 12:00 h. The order of roads surveyed was randomised for each survey period. At each sample site, I spent a maximum of 5 min listening for treecreeper calls and scanning the woodland with binoculars. If a treecreeper was detected, I recorded

G.W. Luck / Biological Conservation 105 (2002) 383–394

the time to detection and the approximate location of the bird. Each location was marked with flagging tape and its distance from the vehicle was measured by pacing. These marked locations were used to identify approximate areas of use of the species at any given site so more detailed habitat data could be collected (Section 2.3). Detection times and distances were used to examine detectability differences between woodland types. This is important because differences in detectability may affect assessments of habitat selection (Thomas and Taylor, 1990). 2.2.2. Data handling and analysis I examined seasonal differences in the number of detections recorded overall and in each woodland type using chi-square. Differences in detection time and distance for each season and woodland type were analysed using repeated-measures analysis of variance (ANOVA) after data were log10 transformed. I considered the species to be present at sites where detection frequency was  three (out of five surveys), and absent from sites with nil detections. These values were used to increase the probability that sites where the species was recorded as present were occupied by resident birds rather than transients or dispersers temporarily occupying a habitat type, and sites where it was recorded as absent reflected true absence rather than an inability to locate birds. I compared the total number of sites where the species was recorded as present among the four woodland types using chi-square. 2.3. Territory selection 2.3.1. Field methods The next (finer) scale of investigation was that of territory selection within a woodland. Based on the results of the presence/absence surveys, I randomly selected 50 sites with treecreepers and 50 sites without from which to collect detailed habitat data. I estimated the species’ area of use (referred to as territory from here on) at each site containing treecreepers based on the three or more flagged locations identified during the presence/ absence surveys. For sites where treecreepers were absent, a ‘‘pseudo-territory’’ was established centred on a point located 100 m perpendicular to the road. The boundary of a pseudo-territory extended from the centre point in a radius of 80 m. This covered an area of approximately 2.5 ha, comparable to the average size of a treecreeper territory in the study area (Luck, 2001). In a standard hierarchical analysis, territories would be nested within a single woodland type (e.g. Wiens et al., 1987). This was not possible in my study because a given woodland type did not contain suitable numbers of the two territory categories. In each territory and pseudo-territory, I collected data on potentially important habitat attributes during 1998

385

and 1999 (Table 1). The selection of attributes was based on prior knowledge of rufous treecreeper ecology, data from other studies on hollow-nesting birds, and observations of the species’ behaviour in my study area. I randomly located up to 10, 2020 m quadrats within the boundaries of each territory and pseudo-territory, and all data were collected within these quadrats. The appropriate sample size for each woodland type was approximated by plotting the mean of the most variable habitat characteristic measured (tree diameter at breast height; DBH) against sample size until an asymptote was obtained. The number of quadrats differed for each woodland: wandoo, 10; powderbark wandoo, eight; brown mallet, five; and mixed woodland, eight. To calculate percent ground and shrub cover, each quadrat was dissected with four evenly spaced 20 m transects, and five sampling points per transect (20 per quadrat) were located at 5 m intervals. At each sampling point, a 10 mm diameter, 2 m high levy pole divided into 10 cm height classes was placed vertically and a substrate was recorded if it came in contact with the pole. Only one hit per substrate type was recorded (i.e. presence or absence). The substrates ground vegetation, litter and bare ground were considered as mutually exclusive. Percent cover of woody-stemmed shrubs was not sufficient to classify shrubs into height classes; so total shrub cover was calculated. Shrub cover was not mutually exclusive from ground substrates (e.g. shrub and litter could be recorded at the same sampling point). 2.3.2. Data handling and analysis Habitat variables that did not meet assumptions of normality were transformed (see Table 5) after being assessed using frequency distributions, normal probability plots and the Shapiro-Wilks test. I examined correlative relationships between the presence of rufous treecreepers and characteristics of their habitat using logistic regression. Multicollinearity between the independent habitat variables was analysed using the Pearson correlation coefficient, and r50.70 was considered a suitable criterion for either omitting a variable or creating a composite variable using principal components analysis (Adler and Wilson, 1985; Tabachnick and Fidell, 1996). Variable retention was based on statistical and biological considerations (Section 3). I used ‘‘interactive’’ (sensu Henderson and Velleman, 1981) regression modelling, whereby all possible subsets of habitat variables were analysed and the best combination was selected based on improvements in the fit and predictive power of the model (Henderson and Velleman, 1981; James and McCulloch, 1990). Model fit was assessed using significant changes in 2loglikelihood with the addition or deletion of variables based on the goodness-of-fit statistic (Z2—distributed as w2), the Hosmer-Lemeshow goodness-of-fit test and R2 variance

386

G.W. Luck / Biological Conservation 105 (2002) 383–394

Table 1 The habitat characteristics measured in each territory and pseudo-territory Habitat characteristic

Method of measurement

Tree density (ha1)

Number of trees (52 cm DBH) per quadrat converted to density ha1. Multi-stemmed trees were considered a single tree if stems diverged above the ground. As above for all trees estimated to be 510 m in height. As above for all trees 5<10 m in height. As above for all trees <5 m in height. As above for all wandoo trees. As above for all wandoo trees 510 m in height. As above for all trees with at least one hollow with an entrance size estimated to be large enough to allow access by a treecreeper. Trees were scanned for hollows from the ground using binoculars. As above for all hollows with an entrance size estimated to be large enough to allow access by a treecreeper. As above for all logs (downed wood) with a hollow deemed suitable for treecreeper use (also see log biomass). Calculated for each tree as size class (sapling—1, subcanopy—2, canopy—3) DBH. A mean value was assigned to each territory. DBH measurements were taken on the thickest stem to the nearest cm using a diameter tape. Percent amount of standing deadwood in each tree was subjectively estimated to the nearest 10% and a biomass figure was calculated as percent deadwoodtree size. A mean biomass figure was then calculated for each territory. The thickest stem of each tree was sighted at eye level through binoculars at a distance of 25 m and percent amount of decorticating bark was estimated to the nearest 10%. Bark biomass was calculated as percent barktree size, and a mean value was calculated for each territory. Downed wood was considered a log if 510 cm in diameter at the widest point. Only logs where >50% of total log length fell inside the quadrat boundaries were measured; fallen trees were considered a single log. A size value was calculated for each log as total log lengthlength of log 510 cm in diameter. These values were summed for each quadrat and the total assigned to each territory. Calculated for each territory as proportion of sampling points with ground vegetation (e.g. herbs and annuals). As for ground vegetation. Litter classified as leaves, bark and woody debris <10 cm in diameter. As for ground vegetation. As for ground vegetation. Percent cover was calculated for all woody-stemmed shrubs combined. As for ground vegetation. Measured at each sampling point by sighting vertically through a 4 cm diameter monocular tube and recording the presence or absence of leaves. A Shannon-Wiener diversity index (Zar, 1996) was calculated for all ground cover comprised of ground vegetation, litter and bare ground. As above for ground vegetation, shrub, sapling, subcanopy and canopy cover.

Canopy tree density (ha1) Subcanopy tree density (ha1) Sapling density (ha1) Wandoo density (ha1) Wandoo canopy density (ha1) Density of hollow-bearing trees (ha1) Density of hollows (ha1) Density of hollow-bearing logs (ha1) Tree size

Deadwood biomass

Bark biomass

Log biomass

Percent ground vegetation Percent litter Percent bare ground Percent shrub cover Percent canopy cover S-W diversity index of ground cover S-W diversity index of vegetation structure

explained for each model (Tabachnick and Fidell, 1996). The predictive capability of the retained model is examined and validated by Luck (this volume). Modelling was conducted using SPSS 8.0 (Norusis, 1998).

measured once. The methods used for data analysis followed those described in Section 2.3.

3. Results 2.4. Nest-site selection 3.1. Detectability The finest scale of investigation was of nest-site selection within individual territories. Nest-site selection was examined in 30 territories that were extensively monitored from 1997 to 2000, as part of a broader study on the ecology of the treecreeper (Luck, 2000). Nest tree and hollow measurements were collected from used nest sites (n=48) and randomly selected unused sites (n=48) that occurred within the boundary of a given territory (Table 2). Multiple nesting hollows used by the same female were not considered as independent and only one of these hollows (chosen randomly) was included in the analysis. Hollows used on multiple occasions were only

There was no significant seasonal difference in the number of detections recorded in each woodland type or overall (Table 3), but there was a significant seasonal difference in the time to detection (Table 4). Detectability was lowest in summer, but occurred more readily during the breeding season probably as a result of the constant calling of nestlings and fledglings. There were no significant differences in detectability between woodland types and no woodlandseason interactions (Table 4). Therefore, the data on woodland selection should not be affected by detectability differences between woodlands.

387

G.W. Luck / Biological Conservation 105 (2002) 383–394 Table 2 Nest-site characteristics measured at each used and unused site Nest-site characteristics

Methods of measurement

Tree DBH (cm) Percent deadwood Tree height (m) Number of hollows Hollow height (m)

Measured per Table 1. Percent amount of standing deadwood in the nest tree subjectively estimated to the nearest 10%. Highest point of the nest tree measured using an inclinometer and calculated via trigonometry. Measured per Table 1. Height of hollow entrance from the ground. Measured using extendable poles to a height of 8 m, or with an inclinometer. Hollow height divided by tree height. Measured as angle to horizontal of branch or trunk which nest was placed in, estimated to the nearest 10
. Branch angle may not coincide with entrance angle (e.g. a front opening hollow in a trunk). Horizontal diameter of widest section of entrance hole measured externally using a 30 cm ruler fixed to the end of extendable poles and read through binoculars. For nests higher than 8 m, entrance size was estimated relative to the size of adult treecreepers by observing birds entering and leaving the nest (or just estimated for unused hollows). For hollows with more than one opening, I considered the entrance to the hollow to be the one that was used most frequently by the birds. Compass direction to which entrance hole opened divided into nine aspect classes: north (337.5 <22.5
); northeast (22.5< 67.5
); east (67.5<112.5
); southeast (112.5<157.5
); south (157.5<202.5
); southwest (202.5<247.5
); west (247.5< 292.5
); northwest (292.5<337.5
); vertical aspect (facing upwards). Measured by standing directly below the hollow, sighting vertically through a 4 cm diameter monocular tube and estimating percent field of view covered by leaves.

Relative height of hollow (m) Spout angle (
) Entrance size (cm)

Aspect (
)

Percent canopy cover

Table 3 The number of detections in each season and for each woodland type (n=200, 50 for each woodland) Woodland type

Wandoo Powderbark Brown mallet Mixed woodland Overall a

w2a 4

Number of detections Mid-breeding

Summer

Autumn

Winter

Early breeding

43 7 6 5 61

46 5 7 4 62

46 10 13 7 76

44 9 10 7 70

43 7 10 5 65

0.20 0.98 2.90 1.28 2.31

Chi-square values are not significant (P>0.10).

3.2. Woodland selection Rufous treecreepers were recorded on three or more occasions at a total of 55 sites. Woodland selection differed significantly among the four woodland types (w23=62.1, P < 0.001). Wandoo was the most commonly used woodland type with the treecreeper being recorded at 39 (78%) of 50 possible sites. Six sites (12%) were occupied in powderbark wandoo and brown mallet, respectively, and four sites (8%) were occupied in mixed woodland. 3.3. Territory selection Table 5 summarises the values of each habitat variable measured. SDEN, SCDEN, WDEN, DHBT, PGV and PLIT (see Table 5 for full variable names) were correlated with other variables and were discarded because they were nested within other habitat measures or retained variables provided more detailed information. The highly correlated (r> 0.75) variables WCDEN, DHOL, TSIZ and DWBM were included in a

principal components analysis. Two principal components were derived from this analysis that had an eigenvalue > 1.0 and explained 90.8% of cumulative variance in the data. WCDEN and DHOL had high factor loadings with the first principal component (0.88 and 0.87, respectively). This component was interpreted as the number of potential nest sites (as treecreepers primarily nest in hollows in wandoo canopy trees) and formed the composite variable NSITE. The variables TSIZ and DWBM had high factor loadings with the second principal component (0.75 and 0.74, respectively). This component was interpreted as a measure of tree age (older trees are generally larger and have a greater biomass of standing deadwood) and formed the composite variable TAGE. Because the territory selection analysis was not structured as a standard nested hierarchy, I included the dummy variable ‘‘woodland type’’ (i.e. wandoo, powderbark wandoo, brown mallet or mixed) in the regression analysis to determine if this was a significant predictor of treecreeper territory use. A total of 13 variables were analysed using interactive logistic regression to determine the most parsimonious

388

G.W. Luck / Biological Conservation 105 (2002) 383–394

model. The final model (Table 6) was highly significant (w23=94.16, P < 0.001), explained 81.3% of variance in the data (Nagelkerke R2), and was not significantly different from the statistically perfect model (HosmerLemeshow goodness-of-fit C8=4.12, P=0.846). Territory selection by rufous treecreepers was positively related to the density of hollow-bearing logs ha1 (DHLOG), the density of NSITE ha1 (wandoo canopy

Table 4 The time and distance to detection in each season and woodland type (meanS.E.) Season

Time (s) to detection

Distance (m) to detection

Mid-breeding Summer Autumn Winter Early-breeding ANOVA

34.84.58 80.16.62 61.46.90 53.58.11 42.64.41 F4, 244=8.81, P <0.001

84.85.83 77.46.02 68.65.28 73.35.23 84.85.83 F4,244=1.46, P >0.10

Woodland type Wandoo Powderbark Brown mallet Mixed woodland ANOVA

trees and hollows), and TAGE (tree size and standing deadwood biomass). Associations between the probability of occurrence of the rufous treecreeper and the composite variables NSITE and TAGE are difficult to interpret as principal component scores. Therefore, I plotted probability of occurrence against the original habitat measures DHLOG, WCDEN, DHOL, TSIZ and DWBM (Fig. 1). The probability of treecreeper occurrence dropped below 0.5 (50%) when DHLOG was < 15 ha1, WCDEN was < 25 ha1 and DHOL was < 50 ha1. The relationships plotted in Fig. 1 are simplifications of the actual situation, as these habitat variables interact to influence probability of occurrence. Structural differences between territories and pseudoterritories are further illustrated when comparing the means of wandoo sites only (Table 7). Each of the significant habitat variables identified by the model (i.e. not factor scores) had higher mean values in sites containing treecreepers. 3.4. Nest-site selection

49.53.82 57.811.92 53.27.97 62.66.37 F3,72=1.81, P >0.10

75.73.38 84.49.57 80.38.28 83.54.66 F3,72=0.83, P>0.10

There was no significant interaction between woodland type and season for time to detection (F12,317=1.44, P=0.15) or distance to detection (F12,317=0.82, P=0.63).

The values, transformations and full variable names of each nest-site characteristic are included in Table 8. None of the variables were correlated and all were included in the logistic regression analysis. The final model identified SPNG and SIZE as significantly related to nest-site selection in treecreepers (Table 9). This model was significantly different from the constant-only

Table 5 The values (meanS.E.) of each of the habitat variables measured in territories and pseudo-territories (refer to Table 1 for calculations) Habitat characteristic

1

Tree density (ha ) Canopy tree density (ha1) Subcanopy tree density (ha1) Sapling density (ha1) Wandoo density (ha1) Wandoo canopy density (ha1) Density of hollow-bearing trees (ha1) Density of hollows (ha1) Density of hollow-bearing logs (ha1) Tree size Deadwood biomass Bark biomass Log biomass Percent ground vegetation Percent litter Percent bare ground Percent shrub cover Percent canopy cover S-W diversity of ground cover S-W diversity of vegetation structure a b

Sitesa

Code

TDEN CDEN SCDEN SDEN WDEN WCDEN DHBT DHOL DHLOG TSIZ DWBM BBM LBM PGV PLIT PBG PSC PCC SWG SWV

Territories (50)

Pseudo-territories (50)

208.88.98 89.13.94 70.73.72 49.44.83 140.412.59 53.94.46 31.62.03 91.16.70 20.31.37 66.52.48 18.21.12 15.80.93 437.536.47 16.20.99 67.41.31 16.40.93 8.00.66 50.41.37 0.80.02 0.90.01

285.021.05 103.211.80 85.0 12.47 96.4 12.04 52.9 10.62 6.31.53 10.2 1.30 23.5 3.23 10.4 0.98 47.5 3.22 10.4 0.69 8.30.83 325.348.24 8.21.33 74.1  1.84 17.8  1.09 15.9 1.54 50.2 1.97 0.60.03 0.80.02

Numbers in brackets are sample sizes. Transformations conducted prior to principal components analysis and logistic regression.

Transformationb

Square root Square root Square root Square root

Log10 Square root Arcsine Arcsine Arcsine Arcsine Arcsine

389

G.W. Luck / Biological Conservation 105 (2002) 383–394

model (w22=11.4, P < 0.01) and not significantly different from the perfect model (Hosmer-Lemeshow goodnessof-fit C7=10.7, P=0.151), but it only explained 24.9% of variance in the data (Nagelkerke R2). Treecreepers tended to use hollows as nest sites if the spout angle was 550
to the horizontal (82% of hollows, n=48) and the horizontal diameter of the entrance hole was between 5 and 10 cm (72% of hollows). Table 6 The habitat variables included in the final territory model showing values of the Wald statistic, levels of significance (Sig.) and proportion of variance explained (R) Variables

Coefficients

S.E.

Walddf

Sig.

R

Constant NSITEa DHLOG TAGEb

2.1071 3.1780 0.1944 0.9340

0.8954 0.8694 0.0601 0.4812

5.071 13.361 10.481 3.761

0.0243 0.0003 0.0012 0.0523

0.2863 0.2473 0.1129

a b

Composite variable of WCDEN and DHOL. Composite variable of TSIZ and DWBM.

4. Discussion The rufous treecreeper exhibited non-random habitat use at all spatial scales examined. At the landscape scale, wandoo woodland was favoured over other available woodland types. Within wandoo, habitat with an abundance of hollow logs and potential nest sites Table 7 Each significant habitat variable (meanS.E.) included in the logistic regression model in territories and pseudo-territories of wandoo sites only Wandoo sites onlya

Habitat characteristic

TSIZ DHOL WCDEN DHLOG DWBM a

Territories (36)

Pseudo-territories (12)

67.4 2.94 106.4 6.79 66.4 3.98 21.1 1.53 19.3 1.37

32.0 2.13 35.5 2.00 25.0 2.24 7.0 1.43 8.8 0.54

Numbers in brackets are sample sizes.

Fig. 1. Relationship between the predicted probability of occurrence of the rufous treecreeper and: (a) hollow log density; (b) hollow density; (c) wandoo canopy tree density; (d) tree size index; and (e) deadwood biomass. Dashed lines are 95% confidence intervals.

390

G.W. Luck / Biological Conservation 105 (2002) 383–394

Table 8 The values of each nest-site characteristic (mean S.E.; see Table 2 for calculations) and a summary of the transformations conducted prior to logistic regression analysis Nest-site characteristic

Tree DBH (cm) Percent deadwood Tree height (m) Number of hollows Hollow height (m) Relative height of hollow (m) Spout angle (
) Entrance size (cm) Percent canopy cover a

Statusa

Code

DBH DWD TRHE NHOL HOHE REHE SPNG SIZE CANC

Transformation

Used (48)

Unused (48)

44.32.30 37.24.38 15.20.63 5.90.61 8.30.48 0.60.03 63.13.56 7.40.47 35.64.50

45.5 3.03 45.5 5.48 15.8 0.81 6.0 0.79 9.1 0.46 0.6 0.03 44.4 4.53 9.4 1.00 42.1 5.06

Arcsine Square root

Log10

Numbers in brackets are sample sizes. Aspect is not included in the table.

Table 9 The variables included in the final nest-site model showing values of the Wald statistic, levels of significance (Sig.) and proportion of variance explained (R) Variables

Coefficients

S.E.

Walddf

Sig.

R

Constant SPNG SIZE

2.6087 1.8440 0.0496

1.1346 0.6691 0.0425

5.291 7.601 4.151

0.0215 0.0058 0.0423

0.2051 0.1246

(wandoo trees with hollows), and large, older trees, was preferentially selected to establish territories. Within a given territory, hollows with a spout angle of 550
and an entrance size of 5–10 cm were favoured as nest sites. In a study of the foraging behaviour of the species, Luck et al. (2001) found that treecreepers also preferentially used large wandoo trees as foraging substrates. Nonrandom habitat use at multiple spatial scales means that investigations confined to a single scale are misleading and a hierarchical approach should be adopted (Kotliar and Wiens, 1990). Hierarchical habitat selection by the rufous treecreeper is illustrated in Fig. 2. Potential scales of habitat use probably represent a continuum, but partitioning into discrete units facilitates interpretation (Wiens et al., 1987). For the rufous treecreeper, interpretations of habitat use are scale dependent and different selection processes operate at different scales, as has been found for other bird species (Bergin, 1992). Orians and Wittenberger (1991) suggested that nest-site selection drives habitatuse decisions at larger spatial scales because individuals are committed to a nest site for the duration of the nesting attempt. The availability of nest sites is often recognised as one of the most important limiting factors in the habitat use of birds (Sedgewick and Knopf, 1990; Bergin, 1992; Matsuoka et al., 1997). However, for sedentary species that occupy all-purpose territories, which must provide suitable nesting and foraging habitat, territory choice is probably the most important

consideration. This is particularly the case for rufous treecreepers in Dryandra because breeding birds occupy territories for extended periods (e.g. 3 years; Luck, 2001), territory quality is positively correlated with reproductive success (Luck, 2001), and potential nest sites do not appear to be limited (see below). Rufous treecreepers have been recorded using wandoo woodland in other regions of the Western Australian wheatbelt (Kitchener et al., 1982; Rose, 1996) suggesting a close affinity with this woodland type. Wandoo is an important habitat for many bird species in the wheatbelt, and a number of these species have declined in abundance since European colonisation (Saunders and Ingram, 1995). The preferential clearance of this woodland for agriculture is undoubtedly one of the main reasons why many of the bird species that use this habitat type are now uncommon in the wheatbelt. The selection of territories by rufous treecreepers in wandoo woodland suggests a close association between the species and particular structural characteristics that are common to older woodland habitat. Undisturbed, old-growth wandoo has an abundance of large trees, dead limbs or dead trees, hollows and logs. These habitat traits are identified as being important in the habitat selection of numerous hollow-nesting birds in North America (Sedgwick and Knopf, 1990; Shackelford and Conner, 1997; Hershey et al., 1998; Steeger and Hitchcock, 1998; Hooge et al., 1999; Lahaye and Gutie´rrez, 1999; Savignac et al., 2000) and Australia (Saunders et al., 1982; Traill, 1991; Bennett et al., 1994; Pell and Tidemann, 1997). Large trees in particular provide substantial foraging and nesting resources for many bird species (Kavanagh et al., 1985; Braithwaite et al., 1989; Sedgwick and Knopf, 1990; Ford and Barrett, 1995; Steeger and Hitchcock, 1998; Flemming et al., 1999; Weikel and Hayes, 1999). This highlights the importance of conserving old-growth forests and woodlands for the protection of a considerable proportion of avian diversity.

G.W. Luck / Biological Conservation 105 (2002) 383–394

Fig. 2. A diagrammatic representation of hierarchical habitat use in the rufous treecreeper. Non-random use was exhibited at each spatial scale.

Research on other Climacteris species indicates that tree species, type of bark, logs, ground cover, standing deadwood and hollow abundance probably influence habitat use (Noske, 1982, 1986; Recher et al., 1985; Ford et al., 1986; Brooker et al., 1990; Recher and Davis, 1997). Type of ground cover may be important for rufous treecreepers because the species spends a substantial amount of time foraging on the ground during winter and early spring (Recher and Davis, 1998; Luck et al., 2001). A high density of hollow logs is important because these provide refuges for recently fledged young, which are weak fliers (Rose, 1996; Luck, 2000). The statistical relationship between the structural characteristics of nest hollows and hollow use by treecreepers was weak, probably reflecting the fact that potential nest hollows are abundant in Dryandra. An abundance of suitable hollows would weaken statistical power in identifying important characteristics influencing hollow selectivity (Pribil, 1998), because a higher proportion of random unused sites that are compared with used sites may indeed be suitable for nesting.

391

In the Northern Hemisphere, the population density of secondary cavity-nesting (=hollow-nesting) species (i.e. those that do not excavate their own cavity) is often considered to be limited by the availability of cavities (Newton, 1994; Pribil, 1998), although some studies indicate that this is not always true (Waters et al., 1990; Welsh and Capen, 1992). Research on hollow-nesting cockatoos and parrots in Western Australia found that hollow abundance was probably not a factor limiting population density (Saunders, 1979; Saunders et al., 1982). This appears to be the case for rufous treecreepers in my study area. Hollow density was relatively high (91  6.70 ha1) and the average treecreeper territory (2.6 ha) probably contained many potential nest hollows. This relationship does not consider competition for hollows with other taxa, which has been identified as an important limiting factor for some hollow-nesting birds (e.g. Garnett et al., 1999). I only measured the external characteristics of hollows, and certain internal characteristics (e.g. depth of hollow) may influence hollow use. However, the statistical correlations between treecreeper hollow selection and spout angle and entrance size probably reflect important biological relationships. An angle of 550
to the horizontal ensures that the nest cup is close to parallel with the ground, thereby providing a relatively stable platform on which to lay the eggs. An entrance size of 5–10 cm allows easy hollow access by adult birds, reduces predation risk by larger nest predators, and ensures greater protection of the nest from adverse climatic conditions (e.g. rain) than hollows with larger entrance sizes. Rufous treecreepers did not use wandoo woodland exclusively and were recorded in all other major woodland types. The species showed a level of habitat use flexibility, although powderbark, mallet and mixed woodland sites favoured by treecreepers had some structural similarities to wandoo (e.g. open understorey, hollow logs and hollow-bearing trees). Changes in the extent of use of non-preferred woodland types may coincide with fluctuations in population density and the level of habitat saturation in wandoo woodland. In Dryandra, the survival rate of treecreepers during my study was relatively high and natality far outweighed adult mortality (Luck, 2001). If these results are representative, the population density in Dryandra is likely to be high possibly resulting in reduced habitat selectivity. Although some of the structural habitat characteristics I measured had strong correlations with the presence of rufous treecreepers, these may not be the actual variables influencing the habitat use of the species (i.e. they may be surrogates for other important factors such as food availability). Also, I did not consider intra and interspecific interactions (e.g. competition and predation), which may affect habitat use (Mac Nally, 1990).

392

G.W. Luck / Biological Conservation 105 (2002) 383–394

Identifying the actual variables influencing a species’ distribution requires comprehensive data collection and may be difficult if these variables are consistently correlated with other habitat characteristics. Given these limitations, it appears that old-growth woodland (specifically wandoo) is extremely important habitat for rufous treecreepers, consistent with other studies of hollow-nesting birds (Saunders et al., 1982; Sedgwick and Knopf, 1990; Pell and Tidemann, 1997; Shackelford and Conner, 1997). Determining the relationship between woodland age, tree size, and the formation of hollows is a contentious issue (Mawson and Long, 1994, 1997; Stoneman et al., 1997) and hollow formation for a particular tree species may vary throughout its range owing to different edaphic and climatic conditions (Saunders et al., 1982; Bennett et al., 1994). In Dryandra, the average DBH of a wandoo tree that provided a nesting hollow for the rufous treecreeper was 46 ( 1.89) cm. Acknowledging potential limitations, Rose (1993) estimated that wandoo trees of this size, in Dryandra, would be 150 years of age. Therefore, younger stands of wandoo may not be suitable breeding habitat for the treecreeper and other hollow-nesting species. This has significant implications for habitat restoration in degraded regions like the Western Australian wheatbelt, where recovery of particular woodland types is likely to be a long-term process. Habitat clearance and fragmentation in the agricultural regions of southern Australia has led to a decline in the abundance of many woodland birds (Ford et al., 2001) and severe modification of remaining woodland habitat is likely to augment this process. Threats to the population viability of species like the rufous treecreeper include removal of dead wood for fuel or to ‘‘tidy up’’ woodland remnants, logging of large trees, and lack of replacement of older trees owing to poor seedling recruitment. Conserving woodland birds in agricultural landscapes requires careful management and restoration of remnant native vegetation. Future directions for landscape restoration have been outlined in detail by other researchers (e.g. Recher, 1993; Barrett et al., 1994) and the results of my study support their conclusions. In summary, management actions must involve removing disturbance (e.g. grazing) from remnant vegetation, ensuring regeneration of endemic species and maintaining important habitat characteristics (e.g. large trees). Determining the traits that may influence habitat choice in species is fundamental to developing strategies aimed at habitat protection and species conservation.

Acknowledgements This work was supported by grants from the Australian Bird Study Association, Birds Australia (Stuart

Leslie Bird Research Fund), the Centre for Ecosystem Management (Edith Cowan University), CSIRO Sustainable Ecosystems and the Ecological Society of Australia. The Department of Conservation and Land Management granted permission to conduct the fieldwork in Dryandra. Peter Cale, Robert Lambeck, Harry Recher and two anonymous reviewers provided constructive comments on drafts of the manuscript.

References Adams, E.M., Morrison, M.L., 1993. Effects of forest stand structure and composition on red-breasted nuthatches and brown creepers. Journal of Wildlife Management 57, 616–629. Adler, G.H., Wilson, M.L., 1985. Small mammals on Massachusetts islands: the use of probability functions in clarifying biogeographic relationships. Oecologia 66, 178–186. Baines, D., 1994. Seasonal differences in habitat selection by black grouse (Tetrao tetrix) in the northern Pennines, England. Ibis 136, 39–41. Barrett, G.W., Ford, H.A., Recher, H.F., 1994. Conservation of woodland birds in a fragmented rural landscape. Pacific Conservation Biology 1, 245–256. Bennett, A.F., Lumsden, L.F., Nicholls, A.O., 1994. Tree hollows as a resource for wildlife in remnant woodlands: spatial and temporal patterns across the northern plains of Victoria, Australia. Pacific Conservation Biology 1, 222–235. Bergin, T.M., 1992. Habitat selection by the western kingbird in western Nebraska: a hierarchical analysis. Condor 94, 903–911. Braithwaite, L.W., Austin, M.P., Clayton, M., Turner, J., Nicholls, A.O., 1989. On predicting the presence of birds in Eucalyptus forest types. Biological Conservation 50, 33–50. Brooker, M.G., Braithwaite, R.W., Estbergs, J.A., 1990. Foraging ecology of some insectivores and nectarivorous species of birds in forests and woodlands of the wet-dry tropics of Australia. Emu 90, 215–230. Butler, R.G., 1980. Population size, social behavior, and dispersal in house mice: a quantitative investigation. Animal Behaviour 28, 78– 85. Chan, K., 1990. Habitat selection in the white-plumed and the fuscous honeyeater at an area of sympatry. Australian Journal of Ecology 15, 207–217. Clark, R.G., Shutler, D., 1999. Avian habitat selection: pattern from process in nest-site use by ducks? Ecology 80, 272–287. Cody, M.L., 1985. Habitat Selection in Birds. Academic Press, London. Department of Conservation and Land Management, 1995. Dryandra Woodland Management Plan 1995–2005. Lands and Forest Commission, Perth. Flemming, S.P., Holloway, G.L., Watts, E.J., Lawrance, P.S., 1999. Characteristics of foraging trees selected by pileated woodpeckers in New Brunswick. Journal of Wildlife Management 63, 461–469. Ford, H.A., 1989. Ecology of Birds—An Australian Perspective. Surrey Beatty and Sons, Chipping Norton. Ford, H.A., Barrett, G., 1995. The role of birds and their conservation in agricultural systems. In: Bennett, A., Backhouse, G., Clark, T. (Eds.), People and Nature Conservation: Perspectives on Private Land Use and Endangered Species Recovery. Surrey Beatty and Sons, Chipping Norton, pp. 128–134. Ford, H.A., Noske, S., Bridges, L., 1986. Foraging of birds in eucalypt woodland in north-eastern New South Wales. Emu 86, 168–179. Ford, H.A., Barrett, G.W., Saunders, D.A., Recher, H.F., 2001. Why have birds in the woodlands of southern Australia declined? Biological Conservation 97, 71–88.

G.W. Luck / Biological Conservation 105 (2002) 383–394 Garnett, S.T., Pedler, L.P., Crowley, G.M., 1999. The nesting biology of the glossy black-cockatoo (Calyptorhynchus lathami) on Kangaroo Island. Emu 99, 262–279. Hall, L.S., Mannan, R.W., 1999. Multiscaled habitat selection by elegant trogons in southeastern Arizona. Journal of Wildlife Management 63, 451–461. Henderson, H.V., Velleman, P.F., 1981. Building multiple regression models interactively. Biometrics 37, 391–411. Hershey, K.T., Meslow, E.C., Ramsey, F.L., 1998. Characteristics of forests at spotted owl nest sites in the Pacific Northwest. Journal of Wildlife Management 62, 1398–1410. Hilde´n, O., 1965. Habitat selection in birds. Annales Zoologici Fennici 2, 53–75. Hooge, P.N., Stanback, M.T., Koenig, W.D., 1999. Nest-site selection in the acorn woodpecker. Auk 116, 45–54. Hunt, P.D., 1996. Habitat selection by American redstarts along a successional gradient in northern hardwoods forest: evaluation of habitat quality. Auk 113, 875–888. Hutto, R., 1985. Habitat selection by nonbreeding, migratory land birds. In: Cody, M. (Ed.), Habitat Selection in Birds. Academic Press, London, pp. 455–476. James, F.C., McCulloch, C.E., 1990. Multivariate analysis in ecology and systematics: panacea or pandora’s box? Annual Review of Ecology and Systematics 21, 129–166. Kavanagh, R.P., Shields, J.M., Recher, H.F., Rohan-Jones, W.G., 1985. Bird populations of a logged and unlogged forest mosaic at Eden, New South Wales. In: Keast, A., Recher, H.F., Ford, H., Saunders, D. (Eds.), Birds of Eucalypt Forests and Woodlands: Ecology, Conservation and Management. Royal Australasian Ornithologists Union and Surrey Beatty and Sons, Chipping Norton, pp. 272–281. Kitchener, D.J., Dell, J., Muir, B.G., Palmer, M., 1982. Birds in Western Australian wheatbelt reserves—implications for conservation. Biological Conservation 22, 127–163. Kotliar, N.B., Wiens, J.A., 1990. Multiple scales of patchiness and patch structure: a hierarchical framework for the study of heterogeneity. Oikos 59, 253–260. Lahaye, W.S., Gutie´rrez, R.J., 1999. Nest sites and nesting habitat of the northern spotted owl in northwestern California. Condor 101, 324–330. Levin, S.A., 1992. The problem of pattern and scale in ecology. Ecology 73, 1943–1967. Luck, G.W., 2000. Landscape Differences in the Ecology of the Rufous Treecreeper (Climacteris rufa). PhD dissertation, Edith Cowan University, Perth. Luck, G.W., 2001. The demography and cooperative breeding behaviour of the rufous treecreeper (Climacteris rufa). Australian Journal of Zoology 49 (in press). Luck, G.W., 2002. Determining habitat quality for the cooperatively breeding rufous treecreeper (Climacteris rufa). Austral Ecology (in press). Luck, G.W., Charmantier, A., Ezanno, P., 2001. Seasonal and landscape differences in the foraging behaviour of the rufous treecreeper (Climacteris rufa). Pacific Conservation Biology 7, 9–20. Mac Nally, R., 1990. Modelling distributional patterns of woodland birds along a continental gradient. Ecology 71, 360–374. Matsuoka, S.M., Handel, C.M., Roby, D.D., Thomas, D.L., 1997. The relative importance of nesting and foraging sites in selection of breeding territories by Townsend’s warblers. Auk 114, 657–667. Maurer, B.A., 1985. Avian community dynamics in desert grasslands: observational scale and hierarchical structure. Ecological Monographs 55, 295–312. Mawson, P.R., Long, J.L., 1994. Size and age parameters of nest trees used by four species of parrot and one species of cockatoo in southwest Australia. Emu 94, 149–155. Mawson, P.R., Long, J.L., 1997. Size and age of nest trees: a reply to Stoneman, Rayner and Bradshaw (1997). Emu 97, 181–182.

393

McShea, W.J., McDonald, M.V., Morton, E.S., Meier, R., Rappole, J.H., 1995. Long-term trends in habitat selection by Kentucky warblers. Auk 112, 375–381. Michaels, H.L., Cully, J.F., 1998. Landscape and fine scale habitat associations of the loggerhead shrike. Wilson Bulletin 110, 474–482. Miller, D.A., Hurst, G.A., Leopold, B.D., 1999. Habitat use of eastern wild turkeys in central Mississippi. Journal of Wildlife Management 63, 210–222. Moen, C.A., Gutie´rrez, R.J., 1997. California spotted owl habitat selection in the central Sierra Nevada. Journal of Wildlife Management 61, 1281–1287. Morris, D.W., 1987. Ecological scale and habitat use. Ecology 68, 362–369. Muller, K.L., Stamps, J.A., Krishnan, V.V., Willits, N.H., 1997. The effects of conspecific attraction and habitat quality on habitat selection in territorial birds (Troglodytes aedon). American Naturalist 150, 650–661. Newton, I., 1994. The role of nest sites in limiting the numbers of holenesting birds: a review. Biological Conservation 70, 265–276. Norusis, M.J., 1998. SPSS 8.0 Guide to Data Analysis. Prentice Hall, New Jersey. Noske, R.A., 1982. Comparative Behaviour and Ecology of Some Australian Bark-Foraging Birds. PhD dissertation, University of New England, Armidale. Noske, R.A., 1986. Intersexual niche segregation among three barkforaging birds of eucalypt forests. Australian Journal of Ecology 11, 255–267. Orians, G.H., Wittenberger, J.F., 1991. Spatial and temporal scales in habitat selection. American Naturalist (supplement) 137, S29–S49. Otis, D.L., 1997. Analysis of habitat selection studies with multiple patches within cover types. Journal of Wildlife Management 61, 1016–1022. Pell, A.S., Tidemann, C.R., 1997. The impact of two exotic hollownesting birds on two native parrots in savannah and woodland in eastern Australia. Biological Conservation 79, 145–153. Pribil, S., 1998. Reproductive success is a misleading indicator of nestsite preferences in the red-winged blackbird. Canadian Journal of Zoology 76, 2227–2234. Pribil, S., Picman, J., 1997. The importance of using the proper methodology and spatial scale in the study of habitat selection by birds. Canadian Journal of Zoology 75, 1835–1844. Recher, H.F., 1991. The conservation and management of eucalypt forest birds: resource requirements for nesting and foraging. In: Lunney, D. (Ed.), Conservation of Australia’s Forest Fauna. Royal Zoological Society, Mosman, pp. 25–34. Recher, H.F., 1993. The loss of biodiversity and landscape restoration: conservation, management and survival. An Australian perspective. In: Saunders, D.A., Hobbs, R.J., Ehrlich, P.R. (Eds.), Nature Conservation 3: Reconstruction of Fragmented Ecosystems. Surrey Beatty and Sons, Chipping Norton, pp. 141–151. Recher, H.F., Davis, W.E., 1997. Foraging ecology of a mulga bird community. Wildlife Research 24, 27–43. Recher, H.F., Davis, W.E., 1998. The foraging profile of a wandoo woodland avifauna in early spring. Australian Journal of Ecology 23, 514–527. Recher, H.F., Holmes, R.T., Schulz, M., Shields, J., Kavanagh, R., 1985. Foraging patterns of breeding birds in eucalypt forest and woodland of southeastern Australia. Australian Journal of Ecology 10, 399–419. Rice, J., Anderson, B.W., Ohmart, R.D., 1984. Comparison of the importance of different habitat attributes to avian community organization. Journal of Wildlife Management 48, 895–911. Rose, A.M., 1996. Territorial and breeding behaviour of the rufous treecreeper (Climacteris rufa) in the Stirling Ranges, Western Australia. Corella 20, 55–61.

394

G.W. Luck / Biological Conservation 105 (2002) 383–394

Rose, P.W., 1993. Production of Habitat Hollows by Wheatbelt Eucalypts. Final Report—Project R053. Department of Conservation and Land Management, Perth. Rosenzweig, M.L., 1991. Habitat selection and population interactions: the search for mechanism. American Naturalist (supplement) 137, S5–S28. Rotenberry, J.T., 1985. The role of habitat in avian community composition: physiognomy or floristics? Oecologia 67, 213–217. Saunders, D.A., 1979. The availability of tree hollows for use as nest sites by white-tailed black cockatoos. Australian Wildlife Research 6, 205–216. Saunders, D.A., Ingram, J.A., 1995. Birds of Southwestern Australia. An Atlas of Changes in Distribution and Abundance of the Wheatbelt Fauna. Surrey Beatty and Sons, Chipping Norton. Saunders, D.A., Smith, G.T., Rowley, I., 1982. The availability and dimensions of tree hollows that provide nest sites for cockatoos (Psittaciformes) in Western Australia. Australian Wildlife Research 9, 541–556. Savignac, C., Desrochers, A., Huot, J., 2000. Habitat use by pileated woodpeckers at two spatial scales in eastern Canada. Canadian Journal of Zoology 78, 219–225. Sedgwick, J.A., Knopf, F.L., 1990. Habitat relationships and nest site characteristics of cavity-nesting birds in cottonwood floodplains. Journal of Wildlife Management 54, 112–124. Shackelford, C.E., Conner, R.N., 1997. Woodpecker abundance and habitat use in three forest types in eastern Texas. Wilson Bulletin 109, 614–629. Shields, S.J., Kelly, J.F., 1997. Nest-site selection by belted kingfishers (Ceryle alcyon) in Colorado. American Midland Naturalist 137, 401–403. Sodhi, N.S., Paszkowski, C.A., Keehn, S., 1999. Scale-dependent habitat selection by American redstarts in aspen-dominated forest fragments. Wilson Bulletin 111, 70–75. Southwood, T.R.E., 1977. Habitat, the templet for ecological strategies? Journal of Animal Ecology 46, 337–365.

Steeger, C., Hitchcock, C.L., 1998. Influence of forest structure and diseases on nest-site selection by red-breasted nuthatches. Journal of Wildlife Management 62, 1349–1358. Stoneman, G.L., Rayner, M.E., Bradshaw, F.J., 1997. Size and age parameters of nest trees used by four species of parrot and one species of cockatoo in south-western Australia: critique. Emu 97, 94–96. Storch, I., 1993. Habitat selection by capercaillie in summer and autumn: is bilberry important? Oecologia 95, 257–265. Storr, G.M., 1991. Birds of the south-west division of Western Australia. Records of the Western Australian Museum. Supplement No. 35. Tabachnick, B.G., Fidell, L.S., 1996. Using Multivariate Statistics, 3rd Edition. HarperCollins, New York. Thomas, D.L., Taylor, E.J., 1990. Study designs and tests for comparing resource use and availability. Journal of Wildlife Management 54, 322–330. Tibbetts, E., Pruett-Jones, S., 1999. Habitat and nest-site partitioning in splendid and variegated fairy-wrens (Aves: Maluridae). Australian Journal of Zoology 47, 317–326. Traill, B.J., 1991. Box-ironbark forests: tree hollows, wildlife and management. In: Lunney, D. (Ed.), Conservation of Australia’s Forest Fauna. Royal Zoological Society of New South Wales, Mosman, pp. 119–123. Waters, J.R., Noon, B.R., Verner, J., 1990. Lack of nest site limitation in a cavity-nesting bird community. Journal of Wildlife Management 54, 239–245. Weikel, J.M., Hayes, J.P., 1999. The foraging ecology of cavity-nesting birds in young forests of the northern coast range of Oregon. Condor 101, 58–66. Welsh, C.J.E., Capen, D.E., 1992. Availability of nesting sites as a limit to woodpecker populations. Journal of Wildlife Management 48, 31–41. Wiens, J.A., Rotenberry, J.T., Van Horne, B., 1987. Habitat occupancy patterns of North American shrubsteppe birds: the effects of spatial scale. Oikos 48, 132–147. Zar, J.H., 1996. Biostatistical Analysis, 3rd Edition. Prentice-Hall International Incorporated, New Jersey.