Floristic and structural habitat preferences of yellow-bellied gliders ( Petaurus australis) and selective logging impacts in southeast Queensland, Australia

Pores~ijology

Management ELSEVIER Forest Ecology and Management 98 (1997) 281-295

Floristic and structural habitat preferences of yellow-bellied gliders (Petaurus australis! and selective logging impacts in

southeast Queensland, Australia

T.J. Eyre a3 * , A.P. Smith b a Deparrmenr of Ecosystem Management, Urkersi@ of New England, Armidule, NSW 2351, Australia

’ Austeco Environmental Consulrants, 84 Browsn Street, Armidale. NSW 2350, Australia

Received 19 March 1997; accepted 4 April 1997

Abstract

The floristic and structural habitat requirements of the yellow-bellied glider (Petaurus australis) and the influence of selective logging on habitat quality in the Maryborough District of southeast Queensland, Australia were determined. Yellow-bellied gliders showed a definite preference for forest associations which contained gum-barked and winter flowering species. Within these associations, abundance was correlated with microhabitat variables and a structural variable representing forest age and degree of disturbance using a Poisson regression analysis. The significant explanatory variables included the structural variable, site productivity and the number of dead hollow-bearing trees. These variables relate to the foraging and denning requirements of the yellow-bellied glider. Implications for forest management in southeast Queensland include the need to retain both mature gum-barked eucalypt species and live hollow-bearing trees during harvesting operations. 0 1997 Elsevier Science B.V.

Kewordsr Habitat suitability; Pe~aaurtts australis; Eucal.vprus forest; Forest management: Forest structure

1. Introduction

The yellow-bellied glider (Petaurus australis) has an extensive distribution throughout the forests of eastern Australia where it occurs at low densities in large, exclusive home ranges occupied by small fam- ily groups (Goldingay and Kavanagh, 1991). The yellow-bellied glider is a hollow-dependent species with particular dietary requirements. It feeds pre-

dominantly on plant and insect exudates such as eucalypt sap, nectar, honeydew and manna which satisfy energy requirements, while protein is obtained from arthropods and pollen (Smith and Rus- sell, 1982; Craig, 1985; Goldingay, 1986, 1990; Kavanagh, 1984, 1987). The specialised habitat requirements and low population density and breeding potential of the yellow-bellied glider make it poten- tially vulnerable to habitat disturbance (Craig, 1985; Goldingay and Kavanagh, 1990). In Queensland the yellow-bellied glider has been recognised as a species of management concern (Boorsboom, 1996).

Logging is considered the principal threat throughout the range of the yellow-bellied glider

* Corresponding author. Faculty of Resource Management, Southern Cross University, P.O. Box 157, Lismore N.S.W. 2480, Australia.

0378-l 127/97/$17.00 0 1997 Elsevier Science B.V. All rights reserved. PU SO378-I 127(97)001 1 l-4

(Goldingay and Kavanagh, 1991). This species de- pends on mature and old growth forest resources. including large trees with hollows for nesting, abundant nectar and large areas of shedding bark providing a niche for arthropods, which may be reduced during logging events (Scotts, 1994). Milledge et al. (1991) and Kutt (1995) found significantly higher densities of yellow-bellied gliders in old-growth forests within the intensively managed (cleatfelled) mountain ash and lowland forests of Victoria. An- drews et al. ( 1994) and Smith et al. (1995) found that yellow-bellied gliders preferred mature and old- growth forests in the less intensively (selectively) logged forests of northeast New South Wales. In contrast, broad scale surveys in southeast (Kavanagh and Bamkin. 1995) and northeast (Kavanagh et al., 1995) New South Wales reported that, in preliminary assessments, yellow-bellied gliders showed no association with logging.

In southeast Queensland yellow-bellied gliders are distributed predominantly throughout open eucalypt forests within State Forests (Eyre, 1993). Under State Forest tenure these forests are selectively logged, involving the removal of trees within mer- chantable size classes. and occasional removal of over-mature trees. The intensity of logging varies with the demand for timber products. such as poles or sawlogs. and the productivity of the forest. Selec- tive logging alters the structure of a forest to create an uneven-aged stand dominated by younger stems.

This study had two aims: (i) to identify the habitat requirements of the yellow-bellied glider in a coastal forest of southeastern Queensland. and (ii) to exam- ine the influence of selective logging on habitat use and selection by yellow-bellied gliders within their preferred habitat.

2. Methods

2.1, Study areu

The study area was located in St Marys State Forest (25”39’, l52”26’), 30 km south-west of Mary- borough in southeast Queensland. St Marys State Forest is an expanse of contiguous open forest of 17400 ha and contains two broad forest types which are dominated either by white mahogany Eucalyptus

ucrnenoides with red bloodwood Crqmbiu~ trachyphloia and brown bloodwood C. truchyphloiu or spotted gum C. citriadoru with grey ircmbark E. sider~~phioiu, broad-leaved ironbark E. jihrostr and narrow-leaved ironbark E. L,rebrtr. Other eucalypl species which are scattered throughout the forest in smaller densities include Queensland peppermint k. crsertu. pink bloodwood C. gumm~~>rci-, forest red gum E. tereticornis, grey gum E. longimstrc~tcc. gum-topped box E. moluccnnrr and scribbly gum E. racemosa. and Angophorn costutu. St Marys State Forest has been logged on a regular basis since its gazettal in the 1950s. The most commercially \Talued species include C. citriodorcl, E. fibrtrsct anct k’. sider~phloitr.

The topography is generally undulating with broad flat ridges separated by shallow gullies. The soils arc derived from sedimentary bedrock, and include yellow clayey subsoiIs. leached sandy soils and yellow earths.

2.2. Selectio7l c$‘ .surz~ey sites

Sites were stratified by broad forest association and time since logging. The Queensland Department of Primary Industries Forest Service Area Informa- tion System was used to provide the necessary information on logging history and forest types. The information was digitised into a Geographical lnfor- mation System (Environmental Resource Manage- ment System. E-RMS). which was then used to identify stratified units based on five classes of forest type ( E. ucmenoides/C. truchyphloia. C. citri- odwa/E. sidercjphloia, E. ,fibrosa/C. citrio&Fa, C:. c,itriodwu/E. ct-ebra and C. interrnedin/E. w- rnenoides) and four classes of time since logging (> 40 yrs. 25-40 yrs. IO-25 yrs and < 10 yrs). therefore giving a possible total of 20 strata units. Due to the lack of topographic variation in St Marys. further stratification based on topographic position was not considered necessary for the final selection of field sites.

To avoid bias by over-sampling strata units which covered relatively large areas of forest, the number of sites to be sampled within each strata unit was based on a (log? + 1) transformation of the percent of total area each strata unit encompassed (Table 1).

T.J. Eyre, A.P. Smith/ Forest Ecology and Management 98 (1997) 281-295 283

Table I Sampling rule used to determine the number of sites to be sampled in each strata unit

Percent of total area Number of sites to be sampled

encompassed by strata unit (log, + 1) transformation

< 0.3 Not sampled 0.3-2 2 2.1-4 3 4.1-8 4 8.1-16 5 > 16 6

For example, if a strata unit made up 4% of the total area, then three sites were established within that strata unit. This led to 74 sites being established.

Because yellow-bellied gliders occupy large home ranges (30-65 ha; Goldingay, 1992; Goldingay and Kavanagh, 19931, special precautions were required during site selection. Sites were located within strata units > 50 ha in area and positioned at least 200 m from boundaries with adjacent strata units to reduce the influence of patch dynamics. Sites were also located at least 1 km apart to ensure independence between the sites (i.e. to avoid oversampling gliders within a single home range). At each site a 400 X 50 m transect was marked out in 20 m intervals.

2.3. Census of yellow-bellied gliders

Three observers were positioned 150 m apart along each transect at the 50 m, 200 m, and 350 m marks. Each census began with a five minute period listening for characteristic vocalisations of the yellow-bellied glider. Pre-recorded vocalisations of the yellow-bellied glider and the powerful owl Ninox strenua were then broadcast for three minutes each from the 200 m mark to encourage a response (Kavanagh and Rohan-Jones, 1982; Goldingay, 1994). The recordings were played at a volume which was audible up to 200 m from the broadcast point. Finally, 15 min was spent by each observer searching with 100 W or 50 W spotlights. The time, estimated distance and direction of each yellow-bellied glider heard or seen during the census by an observer was recorded to allow mapping of approxi- mate glider locations at the time of each census.

Gliders mapped further than 200 m from any point along the transect were not included in the analyses.

The census was conducted during spring, autumn, winter and summer at each site between July 1992 and September 1994. However, since logging operations were undertaken at eight of the survey sites before the fourth census period, only the data obtained during the first three census periods were used. The number of gliders detected during the first three census periods were combined to provide a single abundance value for each site.

A number of variables pertaining to the prevailing weather and survey methods (covariates, Table 2) were recorded at the time of census so that any effects upon the detectability of yellow-bellied gliders could be recognised during data analysis.

2.4. Habitat measurement

Habitat variables were measured and recorded between July 1992 and November 1992. At each site variables were recorded either along the transect (Table 3) or from seven 10 X 10 m subplots (Table 4). Transects (400 m long) of varying widths (25 or 50 m) were used to measure habitat variables. The 400 X 50 m transect was used so that hollow-bearing trees, a widely dispersed variable, could be ade- quately sampled, thus giving a value per 2 ha. Data from the seven subplots were pooled and a mean value calculated for each site.

2.5. Statistical analyses

2.5.1. Covariates and detectability of yellow-bellied gliders

Data obtained during the first three census periods were combined to produce a data matrix of 222 rows (74 sites X 3) and nine columns. The relationship between counts of yellow-bellied gliders, covariates and habitat variables recorded for each site were analysed using a Generalised Linear Model assuming Poisson error (McCullagh and Nelder, 1989) with a stepwise forward regression process to produce an equation of the form:

N=eR

where N is the number of yellow-bellied gliders detected at each site and R is a function of the

284 T.J. Eye. A.!? Smith/Forest Ecolqg und Managernrr~t 5% I IO971 X3-241.5

Table 2 Covariates recorded at the time of yellow-bellied glider census

Description of covariates

Moonphase - 96 moon out Moonlight

----__-.-.̂ .~ Categories of covariate

- .____---- o- 100% 0 No moonlight I Low 2 Medium 3 High

Cloud cover percent of sky covered Air temperature Estimate of wind velocity in canopy

Precipitation

Time after dusk Time of year season

o- 100% o-30°C 0 Still no movement m canopy I Gentle upper branches in motion 7 Moderate entire canopy in motion 3 Strong majority of tree in motion 0 No rain I Fog or periodic drizrle 2 Light rain 3 Medium rain 1 Heavy rain O-6 h I -Winter 1 Spring 3 Summer 4 Autumn

Total Eucul~ptus flower cover (m’ ha - ’ )

significant variables in the equation. Poisson regression analyses were used because counts of gliders on each site were discrete and assumed to conform to a Poisson distribution.

As no significant relationship was found between the covariates and the number of gliders detected during each census, the effects of covariates upon glider activity or detectability was not considered to have influenced the reliability of the results describing relationships between glider density and habitat variables.

2.5.2. Overstorey Joristic classification An agglomerative, hierarchical clustering algo-

rithm was used to classify the sites into groups with similar overstorey floristics. The Bray-Curtis index was used to determine the dissimilarity between the mean % basal area of each overstorey tree species at each site. The Unweighted Pair Group Method using Arithmetic Means (UPGMA) was then applied to perform a cluster analysis on the above similarity matrix. The UPGMA dendogram, showing the dissimilarity between the sites, was then examined for groups of floristic similarity between the sites.

Analysis of Variance (ANOVA) was used to in- vestigate the relationship between the overstorey groups identified during the classification process and yellow-bellied glider abundance.. Counts of gliders obtained during the first three census periods were pooled for each site and then normalised using the transformation ,j( x + 0.5)

ANOVA was also used to explore relationships between the overstorey groups and al1 habitat variables other than those pertaining to the basal area of overstorey species. Tukey tests were used to determine significant differences between means. and to identify homogenous groups.

2.5.3. Ouerstorey structure classiJi:catiotl Two categorical habitat variables representing

structure of the overstorey were generated and added to the database. Sites were classified into a simu&ed structural gradient using an agglomerative, hierarehi- cal clustering algorithm. The Bray-Curtis measure was used to determine the dissimilarity between the number of stems in each of five increasing~ diameter classes (S-30 cm, 3 l-50 cm, 5 I-70 cm, 7 1-W cm and > 90 cm). The first structural variable

T.J. Ewe. A.P. Smith/ Forest Ecology and Management 98 (1997) 281-295 785

Table 3 Description of habitat variables recorded at each survey site (400 X 50 m)

Code Description of variable

Time since logging (years) Soil type

Topographic position Elevation (ml

Variubles recorded along 400 X 25 m transect

Time since last logging event in years, used as a surrogate for stand age Three categories: yellow clayey subsoils, leached sandy soils and yellow earths Three categories: ridge/top slope, midslope and gully/lower slope Obtained from 1:25 000 topographic map

Site index Site quality at each site was determined by using the site index table provided by Grimes and Pegg (1979) for St Marys State Forest. The mean height of 12 trees with diameters 30 i I cm was calculated to give a value of height at 30 cm diameter for each site. The site index. using the mean height at 30 cm diameter. was then derived from the site index table.

Number of stems (ha-’ ) and basal area (m2 ha- ‘) of each Each stem was identified to species and recorded within 10 cm diameter overstorey species classes. The basal area of each species for each site was determined by

multiplying the number of stems recorded within each diameter class by the basal area of trees at the midpoint of each diameter class. The basal area of each diameter class for each species was summed to give a total basal area for each site

Basal area of trees with decorticating bark (ma ha- ’ 1

Number of old cut stumps (ha- ’ 1

Number of recent cut stumps (ha-’ )

Total basal area of gum-barked trees which shed bark in strips (E: longirostrata, E. moluccana and E. tereticornis) Total number of old cut stumps (estimated > 5 yrs old) from past logging operations Total number of recently cut stumps (estimated < 5 yrs old) from past logging operation

Variables recorded along 400 X 50 m transect

Number of dead hollow-bearing trees (per 2 ha) Number of live hollow-bearing trees (per 2 ha)

Total number of dead stems > 20 cm in diameter with observable hollows Total number of living stems > 20 cm in diameter with observable hollows

(STRUCTA) was based on the total number of stems in each of the five diameter classes. The second structural variable (STRUCTB) was based on the

number of stems in each of the five diameter classes within four broad floristic groups, including gum- barked (C. citriodoru, E. tereticornis. E. moluccana,

Table 4 Variables recorded for seven subplots (10 X IO ml located at each 50 m mark along transect. Data from each subplot was combined and a mean value calculated for each site

Code Description of variable

Slope Rock Aspect Canopy height (ml Height of tallest tree (m) Height of shrub understorey (ml Canopy cover Shrub cover Number of shrubs (ha- ’ ) Number of Acacia stems (ha-’ 1

Measured in degrees 8 cover Measured in degrees Height of overstorey canopy Height of tallest living or dead tree Height of shrubs > 0.5 m Projective foliage cover of overstorey Projective foliage cover of shrub understorey ( > 0.5 ml Total number of shrub stems, excluding eucalypts Total number of Acacia stems

286 T.J. Eyre, A.P. Smith/ Forest Ecology and Management 98 C IYY7) 281-295

A. costatu and E. longirostruta). ironbarks (E. such as C. citriodoru. A. costuta, E. molucccma, E. siderophloia. E. jibrosa, and E. crebru), blood- racemosu and E. tereticomis were uncommon. and woods (C. intermedia and C. truchyphloia) and E. longirostrutn was not present at any of the sites in stringybarks (E. acmenoides). this class.

ANOVA was used to determine whether there were significant differences between the mean number of stems in each of the five diameter classes in each of the categories of STRUCTA. Similarly. ANOVA was used to identify significant differences in the mean number of gum, ironbark, bloodwood, and stringybark stems in the five diameter classes in each of the categories of STRUCTB. Relationships between the groups of both STRUCTA and STRUCTB and logging variables (years since logging and cut stumps) was also explored using ANOVA. Tukey tests were used to distinguish significant differences between the means.

Class 1 had the highest mean tree basal area of the seven floristic groups, however mean canopy height was relatively low when compared to the remaining classes.

3.1.2. Cluss 2 This class was dominated by C. intermed& and

E. ucmenoides, and occurred at 10 sites. C. irac$- phloicr was also common at most sites in this class. The three ironbark species and C. c+triodoru were present at a number of these sites, but only in small densities. Mean site index was the lowest in this class.

2.5.4. Analysis of habitat data 3.1.3. Class 3 A lack of independence between the predictor

variables violates an important assumption of regression analysis. A Pearson correlation matrix was generated so that intercorrelation between the predictor variables could be identified. Canopy height was correlated with site index and was removed from the data set. Stepwise Poisson regression analyses were then used to produce a model that described the relationship between the pooled number of yellow- bellied gliders and the recorded habitat variables recorded at each site.

Class 3 consisted of eight sites. where the dominant species were E. crebru and C. citriodorcr. E. moluccnnu, E. longirostrata and E. ncmenoides were also present at the majority of sites in this class. This class had the lowest mean number of stems.

3. I .4. c1u.s.s 4 E. jibrosa and C. citriudoru were the dominant

overstorey species of the nine sites belonging to class 3. E. acmenoides was also common species in this class. The mean tree basal area for this class was relatively high, though the mean number oft stems was the highest of all the classes.

3. Results 3.1.5. ClLlS.~ 5

3.1. Overstorey floristic ussociations

Six overstorey classes were identified at St Marys, using a dissimilarity threshold of approximately 0.5. A summary of the mean % basal area of each tree species, mean number of stems and mean total basal area within each class is shown in Table 5.

A total of 12 sites formed class 5. where C. citriodotw and E. .siderophloia were the dominant overstorey species. E. moluccuna and E. tereticornis were also relatively common in this class. E. UP menoides and C. intermedia were present at a number of the sites in this class, though in small densities.

3.1.1. Class 1 3.1.6. Class 6 This class occurred at nine sites, where the over-

storey was dominated by C. trachyphloia. Sub-dominant species included E. acmenoides and C. intermedia. E. exserta was present at most sites in this class, though in small numbers. Smooth-barked species

This class comprised 26 sites where C. citriodora and E. siderophloia were the dominant species. E. acmenoides and C. intermedia were present at a number of the sites in this class, whereas E. tereticomis and E. moluccana were occasionally present.

Tabl

e 5

Sum

mar

ised

desc

riptio

n of

th

e se

ven

tree

asso

ciatio

n cla

sses

Tree

as

socia

tion

clas

s I

2 3

4 5

6 Do

mina

nt

spec

ies

C.

trach

yphlo

ia E.

acm

enoi

des

C. c

itrio

dom

E.

jibr

osa

C.

citrio

dora

C.

cit

riodo

ra

.?I

C.

inter

med

ia C.

int

erm

edia

E. c

rebr

a C.

citr

iodor

a E.

sid

erop

hloi

a E.

Sid

erop

hloi

a fi

/E.

acm

enoi

des

/ E.

m

oluc

cana

,,”

a M

ean

% b

asal

ar

ea

P 3 E.

acm

enoi

des

14.7

2*3.

4 35

.48k

5.3

7.8Ok

2.3

14.7

6 f

2.2

6.34

* 3.

5 9.

96

k 1.

7 A.

co

stata

” 2.

39

k 0.

8 2.1

7kO.

9 0

0.78

+

0.4

1.26

k

0.9

0.55

f

0.3

2 E.

cre

bra

0 0.

98

f 0.

6 26

.6k5

.6

I .52

+

1.5

1.15

+1.1

0.1

6kO.

2 s \

E. e

xser

ta

4.12

+ 1.

6 2.

69+

1.0

0.68

+

0.5

0.61

+O

.l 1.

83*0

.8

1.88

,0.5

2

E. f

ibros

a 0

4.88

+

3.2

5.88

k3.2

36

.27

+ 4.

0 0

0.69

+

0.4

2 2 C.

int

erm

edia

16.3

5 +4

.3

32.4

1 +

7.9

0.60

+

0.3

5.59

+

3.2

5.49

+

1.9

8.71

-f

1.4

h C.

citr

iodor

aa

2.92

+2.9

6.

54

f 2.

4 35

.27

f 3.

4 31

.34k

4.4

34.0

7 *

3.1

48.0

k

2.3

a

E. m

oluc

cana

a F

0.93

f

0.9

0 17

.5

k 8.

5 1.

13*0

.7

16.2

2k3.

5 1.

67 k

0.

7 00

‘1

E.

lon

giros

trata

* 0

0 4.

93

k 1.

48

0.81

kO

.8 0.

26

f 0.

2 0.

79

-10.

4 f:

E. s

ider

ophl

oia

0.7

1+

0.5

4.33

f

1.8

0.44

+

0.4

3.91

+ 1.

5 21

.74j1

3.2

21.1

6&

1.4

E. r

acem

osaa

0.

74+0

.7

0 0

0 0

0.23

+0

.2

i

E. t

eret

icom

iP

1.42

f0.8

0.

55

+ 0.

5 2.1

8kO.

6 3.

41*2

.1

11.6

4&-3

.3

3.91

+0

.6

8 C.

tra

chyp

hloia

55.8

6+4.

7 9.

92k3

.4

0 0.

31

to.3

0

2.31

+

1.1

2 ra

Mea

n ba

sal

area

19

69.7

+ 14

4.1

1402

.9

+ 14

9.8

1478

.0+9

4.1

1595

.7

+ 15

2.2

1471

.5+-

13

5.0

1495

.9

i 55

.2

F Cm

2 ha

-‘)

s M

ean

num

ber

of

308.

22

+ 26

.2

354.

12*6

9.12

29

9.8

k 26

.1

493.

63

f 89

.8

423.

92

f 48

.6

376.

39

f 22

.1

z

stem

s (h

a-’

) zz

2

Mea

n ca

nopy

26

.3f

1.1

26.6

+

0.6

29.0

+

0.6

28.3

rt

0.7

29.3

+

0.9

29.3

+

0.5

heigh

t (m

) ?

Mea

n sit

e ind

ex

28.6

& 1.

1 27

.3

+ 0.

5 29

.3

k 0.

6 28

.6

f 0.

8 29

.4+0

.7

29.6k

O.4

No.

sites

9

10

8 9

12

26

s

Mea

ns

are

f sta

ndar

d er

ror.

“Sig

nifie

s gu

m-b

arke

d sp

ecie

s.

288 T.J. Eyre, A.P. Smith/Forest Ecology and Management 98 (1997) 281-295

3.2. Overstorey jloristic associations preferred by yellow-bellied gliders

Yellow-bellied gliders were detected at 46 of the 74 sites surveyed in St Marys State Forest over the three census periods.

Yellow-bellied gliders were detected in each of the six overstorey classes (Fig. la>. There was a significant difference in mean glider abundance between the overstorey classes (Fs,68 = 3.421, P = 0.008). Mean glider abundance was highest in the C. citriodora/ironbark dominated classes (classes 3, 4, 5, and 6) which formed one homogenous group, and

:

L

Fig. I. The mean ( + se.) (a) number of yellow-bellied gliders. (b) basal area of trees with decorticating bark and (c) number of live trees with hollows within each overstorey floristic class. The overstorey floristic classes are as described in the text and Table 5.

lowest in the bloodwood/stringybark dominated classes (1 and 2) which formed a second homogenous group.

3.3. Relationships between or~erstorey,floristic c&~se.s and habitut variables

The means of two habitat variables. number of’ live hollow-bearing trees and basal area of trees with decorticating bark were significantly different between the overstorey classes (F5,6h = 6.515, P = 0.000 and F5,6x = 13.645, P = OBOO respectivety). Tukey tests showed that the number of live hollow- bearing trees was greater in overstorey classes 1 and 2. which formed a homogenous group, relative to classes 3-6 which formed another homogenous group (Fig. lb).

The basal area of trees with decorticating bark was lowest in classes 1 and 2, which again formed a homogenous group, relative to classes 3. 4. 5 and 6 (Fig. 1~). This result reflects the composition of tree species within the overstorey classes, where classes 1 and 2 contain relatively low numbers of bark-shedding species (see Table 5).

The stringybark/bloodwood overstorey associations avoided by yellow-bellied gliders contained significantly more live hollow-bearing trees, a variable expected to be important in yellow-bellied glider habitat selection. To avoid confounding between tree hollow abundance and floristic association, sites in the stringybark/bloodwood classes were excluded from analysis of structural habitat analyses. The reduced dataset contained a total of 55 sites.

3.4. Ollerstorey structural chss$cation

3.4.1. STRUCTA The classification of stems in the five diameter

classes (STRUCTA) produced three groups; SAA with 10 sites, SAB with 22 sites and SAC with 23 sites. SAA had the lowest mean number of stems in the 5 I-70, 7 I-90 and > 90 diameter classes whereas SAC had the highest mean number of stems in these diameter classes (Fig. 2). However the difference was significant only for the 71-90 diameter group using a single factor ANBVA (F2,s- = 6.515, P = 0.012).

T.J. Eyre, A.P. Smith/ Forest Ecology and Management 98 (1997) 281-295 289

SAB

STRUCTA overstorey structure group

SAC

Fig. 2. Proportion of stems in each of the five diameter classes within each of the STRUCTA structural groups. Stem numbers in each diameter cl&s were transformed using a (log + 1) transformation.

The mean number of years since logging was significantly different between SAA and SAC (F2,52 = 4.480, P = 0.0161, where SAA had the lowest and SAC the highest mean (Fig. 3) but there was no significant difference between the mean number of cut stumps (an index of logging intensity) among the groups.

3.4.2. STRUCTB Classification of the number of stems in the five

diameter classes within broad floristic groups (STRUCTB) also produced three groups; SBA with 15 sites, SBB with 10 sites and SBC with 30 sites. The significant differences among the groups of

35,

T

SAC

Fig. 3. The mean number (+ s.e.) of years since logging in the STRUCTA groups SAA. SAB and SAC.

STRUCTB using ANOVA and Tukey tests are sum- marised in Table 6.

Table 6 shows that this classification method effectively separated the structural groups on a floristic and structural level. Group SBA was characterised by significantly high numbers of gum and ironbark stems in the smallest diameter class, and significantly low numbers of gum stems in the three largest diameter classes compared with that of groups SBB and SBC. Group SBB was characterised by significantly large numbers of bloodwood and stringybark stems across all diameter classes. Com- pared with groups SBA and SBB, group SBC had significantly large numbers of gum stems in the three largest diameter classes, and on average larger numbers of ironbark stems in the two largest diameter classes, though this was not significant.

There was a significant difference in the mean number of years since logging between the three groups of STRUCTB (F2.52 = 8.248, P = O.OOl), being much greater in Group SBC than groups SBA and SBB (Fig. 4a). The mean number of cut stumps was also significantly different among the structural groups ( F2,52 = 3.287, P = 0.04), being greatest in SBA and lowest in SBC (Fig. 4b).

These results indicate that group SBA was characterised by sites with small numbers of bloodwood or stringybark stems, were logged recently and logged at a greater intensity relative to sites belonging to groups SBB and SBC. Group SBB was characterised

290 T. J. E.vre. A.P. Smith / Forest Ecology and Management 98 t 1997) 2X1 -29.~

Table 6 Summary of the ANOVA and Tukey tests of the significant difference of mean number of stems per diameter class within floristic groups between STRUCTB groups SBA, SBB and SBC

Variable F ?.<I P _-_.---.--.--

Groups showing significant differences between means from Tukey test (P c 0.05)

Bloodwoods (5-20 diameter) Bloodwoods (2 l-40 diameter) Bloodwoods (41-60 diameter) Stringybarks (21-40 diameter) Stringybarks ( > 80 diameter) Gums (41-60 diameter) Gums (61-80 diameter1 Gums (> 80 diameter1 Ironbarks (5-20 diameter) Ironbarks (2 l-40 diameter)

12.297 0.000

27.60') 0.000

Y.897 0.000

3.628 0.033

5.90') 0.005

11.567 0.000 3.113 0.05

5.151 0.008

3.667 0.032

3.713 0.031

SBA and SBB, SBB and SBC SAA and SBB. SBB and SBC SBA and SBB. SBB and SBC SBB and SBC SBA and SBB. SBB and SBC SBA and SBC. SBB and SBC SBA and SBC SBA and SBC, SBB and SBC SBA and SBC, SBB and SK SBB and SBC

--

by sites with high numbers of stringybark and bloodwood stems in the overstorey, and logged most recently but at a reduced intensity. Group SBC repre- sents the sites with low numbers of bloodwood and stringybark stems, with older forests that had been the least intensively logged.

It would seem that STRUCTB reflect togging intensity and time since logging gradients mure effectively than STRUCTA. This can be attributed to differences in the commercial value of the various eucalypt species in St Marys, where C’. citriodora and the ironbark species are logged more regutarly than the less valued stringybark and bloodwoods species (Queensland Forest Service. 1994). STRUCTB separates the bloodwood and stringybark

T species from C. citriodoru and the ironbark species,

-L- r-

Fig. 4. The mean number t + se.1 of (al years since logging and (bl cut stumps in the STRUCTB groups SBA, SBB and SBC.

thus revealing a clearer forest age and logging intensity gradient between the groups.

3.5. Relationship between habitat mrinhles and yellow-bellied glider abundance

The structural variables STRUCTA and STRUCTB were added to the habitat dataset prior to the Poisson regression analyses. Due to the significant statistical interaction with STRUCTA and/or STRUCTB. the variables time since logging and number of cut stumps were removed.

Three variables. including forest structure within broad floristic groups (STRUCTB), site index, and the number of dead hollow-bearing trees. best ex- plained the number of yellow-bellied gliders in the C. citriodotz/ironbark forests of St Marys State Forest (Table 7). There were fewer yellow-bellied gliders in forests with SBA and SBB~structure/floristic types than in forests with SIX structure/floris-

T.J. Eyre, A.P. Smith/Forest Ecology and Management 98 f 1997) 281-295 291

Table 7 4. Discussion Significant variables from Poisson regression model describing the abundance of yellow-bellied gliders in the C. citn’odora/ironbark forests of St Marys State Forest, south-east Queensland

4.1. Overstorey jloristic associations and yellow-bellied glider distribution

Variables Coefficients s.e.

Constant - 1.9979 1.259 SBA 0 -

SBB - 0.3929 0.25 1 SBC 0.4779 0.092 Site index 0.0688 0.041 Number of dead trees with hollows 0.0438 0.024

tic types (Fig. 5a). Numbers of yellow-bellied gliders increased as site index and number of dead hollow- bearing trees increased (Fig. 5b and c, respectively).

@’ /

Fig. 5. Relationships between the predicted number of yellow-bellied gliders and, as described by the Poisson regression model (Table 7). the significant variables (a) STRUCTB, (b) site index and (c) dead trees with hollows. For each prediction the two remaining variables were held at their mean. Vertical bars refer to the 95% confidence interval.

Preference by yellow-bellied gliders for particular overstorey floristic associations as shown in St Marys State Forest (Fig. la>, has also been reported in the mixed forests of Waratah Creek in southeast New South Wales where only two of seven forest types were favoured by gliders (Kavanagh, 1984; Goldin- gay and Kavanagh, 1993). The forest types preferred by yellow-bellied gliders at Waratah Creek were dominated by eucalypt species providing a winter nectar resource, sap trees and foraging resources associated with decorticating bark, such as invertebrates and honeydew (Kavanagh, 1984, 1987; Goldingay, 1986). The forest types least preferred by gliders at Waratah Creek were dominated by fibrous barked and summer-flowering species (Kavanagh, 1987).

In St Marys State Forest, the clear preference by yellow-bellied gliders for the C. citriodora/ironbark floristic associations (classes 3, 4, 5 and 6) over the stringybark/bloodwood associations (classes 1 and 2) follows the pattern described for Waratah Creek. Classes 1 and 2 are dominated by E. acmenoides and C. intermedia and/or C. trachyphloia which are all fibrous barked and which, in southeast Queensland, tend to flower concurrently from late summer through spring (Blake and Roff, 1988). Conversely, classes 3, 4, 5, and 6 were dominated by C. citriodora, a winter-flowering species, and either E. crebra, E. fibrosa or E. siderophloia, which each provide nectar from late winter to mid summer (Blake and Roff, 1988). The C. citriodora/ironbark forests in St Marys, due to the staggered flowering periods of C. citriodora and the ironbark species, provides a con- tinuous nectar resource over the winter months- when glider energy requirements are relatively high and alternative food sources are low-through to mid summer. In the Maryborough district C. citriodora, E. fibrosa and E. crebra are known by apiarists to produce large quantities of pollen, and E. siderophloia is one of the most valuable of the coastal nectar-producing trees (Blake and Roff, 1988).

The significantly greater basal area of trees in the C. citriodora/ironbark forests which regularly shed

292 T. .I. Eye, A. P. Smith / Forest Ecology and Management 98 ( 19971281-39.5

bark in strips, as opposed to the stringybark/bloodwood forests (Fig. lb), suggests that these forests provide a greater supply of invertebrates and honeydew for foraging purposes. Since yellow-bellied gliders incise only gum-barked tree species in St Marys State Forest including C. citriodora, E. moluccana, E. longirostrata and E. tereticorqis (Eyre, 19931, the higher proportion of these species in the C. citriodoru/ironbark forests (see Table 5) provides a greater sap tree resource.

4.2. Habitat selection within C. citriodoru / ironbark forest associations

Both floristics and structure of the overstorey influenced habitat suitability for yellow-bellied gliders within the preferred C. citriodoru/ironbark forests. Gliders were more common in forests with relatively high densities of ironbark, and in particular, gum-barked stems > 50 cm diameter. As previ- ously discussed, the ironbark and gum-barked species provide a greater foraging resource than the stringybark and bloodwood species in St Marys. The preference by yellow-bellied gliders for forests with relatively high proportions of gum-barked stems > 50 cm diameter (see Fig. 5a) may reflect the increased food availability or harvesting efficiency in relatively large trees. Kavanagh (1987) found that yellow-bellied gliders predominantly chose to forage for most food resources in the larger trees in his study area. Similarly, Goldingay (1990) found that the species chose to feed on nectar in large trees, since the larger trees had greater flower cover. Large gum-barked trees have a greater surface area for decorticating bark, and may provide concentrated shelter for arthropods.

In the context of timber management, site index can be defined as the productive capacity of a forest site, and conveys the interaction of environmental influences of climate, parent materials and topography (Schreuder et al., 1993; Inions, 1990). A site index table had been formulated for St Marys forests from 50 yrs of data collected from detailed growth plots, and provides a reliable surrogate for site productivity (Grimes and Pegg, 1979). The relationship between yellow-bellied gliders and site index as shown by the model suggests that gliders were more common in the more productive forest areas (Fig.

5b). There is a general correlation between forest productivity and abundance of other arboreal mammals, particularly the folivorous greater glider (Petuuroides volans), in forests of eastern Australia (Smith, in press). Other studies have shown an association between arboreal marsupial densities and components of forest productivity. Braithwaite et al. (1984) and Braithwaite (1984) attributed an association between arboreal marsupial abundance and con- centrations of foliar nutrients in forests of southeast New South Wales forests with a component of forest productivity. namely the nutrient status of the under- lying parent materials.

The significance of hollow-bearing trees in the selection of habitat by yellow-bellied gliders (Fig. 5~) was expected since the dependence of this and other arboreal marsupials upon tree hollows for shei- ter and nesting has been well documented (Mackow- ski. 1984; Craig, 1985; Lindenmayer et al., 1991; Goldingay, 1992). That the hollow-bearing trees were dead was not initially expected since the few data available show that yellow-bellied gliders tend to den in living trees. In Victoria, Craig (1985) recorded 15 yellow-bellied glider den trees, all of which were large living trees with diameters > 90 cm, and in north Queensland 45 dens used by yellow-bellied gliders were located in living trees and only one den was in a dead tree (Environment Science and Ser- vices, 19911. The extensive home range of the yet- low-bellied glider (Goldingay and Kavanagh, 19911 suggests that the species does not require tree hoi- lows in the same densities as the more sedentary arboreal marsupial species. even though it is known to utilise a number of hollows within their home range (Henry and Craig. 1984; Craig, 1985). In north Queensland, yellow-bellied glider groups used a mean of 4.3 & 1.0 den trees within a mean home range of 52.8 + 8.6 ha, or approximately 0.16 hollow-bearing trees per 2 ha (Environment Science and Services, 1991). In Victoria Craig (1985) recorded yellow-bellied glider groups using approximately 0.24 hollow-bearing trees per 2 ha. These studies suggest that the number of live hollow-bearing trees in the C. citriodora/ironbark forests of St Marys. with a mean ( f se.) value of 4.4 It 0.3 per 2 ha, should not be a limiting resource for ~yellow-bellied gliders, though it is expected that the number of hollow-bearing trees recorded are over-estimates due

T.J. Eyre. A.P. Smirh / Forest Ecology and Management 98 (1997) 281-295 293

to difficulties in observing true hollows from the ground.

We attribute the apparent importance of dead hollow-bearing trees to yellow-bellied gliders to competition with other hollow dependent species, and possible scarcity of hollows in living trees with characteristics suitable for denning. Characteristics of the hollow-bearing tree, including the tree species and the number of hollows, can influence whether a hollow-bearing tree would be occupied by arboreal marsupials (Smith and Lindenmayer, 1988; Linden- mayer et al., 1990b; Gibbons, 1994; Gibbons and Lindenmayer, 1996). Hollow characteristics which influence use by arboreal marsupials include en- trance size and orientation, whether the hollow-supporting wood is living or dead, height of the hollow above ground level, hollow volume and the position of the hollow in the tree (Ambrose, 1982; Mack- owski, 1984; Gibbons, 1994; Gibbons and Linden- mayer, 1996).

4.3. Implications for forest management

The yellow-bellied glider has often been described as dependent on old growth forest and sensitive to habitat alteration due to logging (Craig, 1985; Goldingay and Kavanagh, 199 1; Kavanagh, 1991; Scotts, 1994; Andrews et al., 1994; Smith et al., 1995). However studies which aimed to assess the impact of logging on arboreal marsupials, including Kavanagh and Bamkin (1995) in southeast New South Wales, and Kavanagh et al. (1995) in northeast New South Wales, found no significant association between logging and yellow-bellied gliders. These studies compared arboreal marsupial abundance in logged or unlogged forests and did not consider the on-site structural characteristics of the forest. Further, the studies of Kavanagh and Bamkin (1995) and Kavanagh et al. (1995), as pointed out by the authors, were each conducted on a regional scale which restricted due consideration of the interactions between logging intensity and forest type upon glider distribution, which was shown to be important in the present study.

Yellow-bellied gliders in St Marys were found to prefer the more productive areas of the commercially viable C. citriodora/ironbark forests which are most valued for timber harvesting. The importance of the

structural variable STRUCTB in influencing yellow-bellied glider abundance in St Marys suggests that the species is sensitive to the intensity of selective logging practices, since the structural groups reflected a logging intensity gradient. Less than one- third of the sites sampled in the C. citriodora/ironbark forests (n = 15) were classified into the structural group that reflected the most intensively logged gradient (SBA). This suggests that the area of forest that has been logged at an intensity which renders the habitat unsuitable for yellow-bellied gliders is currently not extensive in St Marys. However an increase in the extent of more intensive selective logging, as a response to an increase in demand, may have deleterious impacts upon gliders.

Of some concern is the apparent importance of dead hollow-bearing trees to yellow-bellied gliders. Though the reasons for this remain somewhat un- clear, the ephemeral nature of the hollow resource does pose a threat to the long-term conservation status of the species. Dead hollow-bearing trees provide a short term hollow resource relative to that of live hollow-bearing trees. Lindenmayer et al. (1990a) showed that in montane ash forests of Victoria the rate of decay in hollow-bearing trees was rapid, particularly in senescent or dead trees. They predicted a substantial decline of trees with hollows due to the senescence and eventual collapse of dead trees which were providing an essential hollow resource for leadbeater’s possum Gymnobelideus leadbeateri. Dead and decaying trees are also most susceptible to the effects of fire, the degree of damage being proportional to the intensity of the fire (Inions et al., 1989). Until more live hollow-bearing trees become available through active recruitment, dead trees with hollows will need to be maintained during harvesting operations. Currently in Queensland there is no for- mal prescription for the retention of hollow-bearing trees or recruitment trees, where the number, choice and configuration of hollow-bearing trees is at the discretion of the District Forester (Queensland Forest Service, 1994).

5. Conclusions

The major factor influencing the occurrence of yellow-bellied gliders in St Marys State Forest is

294 T.J. Eyre, A.P. Smith/ Forest Ecology and Management 98 (I9971 281-295

forest association, where forests containing gum- barked and winter flowering species were preferred. Within the preferred forest associations, yeilow-bellied gliders are more abundant in the more productive forests with relatively high densities of ironbark and gum-barked species > 50 cm diameter and dead trees with hollows. This leads to conflict with forest harvesting practices, where high intensity selective logging can reduce the suitability of habitat for yellow-bellied gliders. However long term conservation of the yellow-bellied glider is compatible with low intensity selective logging, where a high proportion of ironbark and, in particular, gum-barked stems > 50 cm diameter are retained during logging events. The recruitment of potential hollow-bearing trees will be essential, and dead hollow-bearing trees will need to be maintained to provide an interim hollow resource.

Acknowledgements

The instigator of this project, John Kehl, is grate- fully acknowledged for his enthusiasm, support and ideas during the early stages of the project. Many thanks to the Forest Wildlife Section of the Depart- ment of Natural Resources for logistic support, and in particular Luke Hogan for his untiring assistance in the field and organisation of equipment. Many warm thanks to Amanda Stuart, Mick Whitehead and Harvi, for their energetic assistance and great com- pany while in the field. Greg Smyth, Paul Lloyd and Tracy Bell are also thanked warmly for field assistance. Ross Goldingay and Geoff Smith are thanked for their valuable comments on an earlier draft of this paper. This study would not have been possible without the financial and logistic support provided by the Queensland Department of Primary Industries Forest Service.

References

Ambrose, G.J., 1982. An ecological and behavioural study of vertebrates using hollows in eucalypt branches. Ph.D. thesis, LaTrobe University, Melbourne, Vie., 447 pp.

Andrews, S.P., Gration, G.. Quin, D., Smith, A.P., 1994. Descrip- tion and assessment of forestry impacts on fauna of the Urbenville Forestry Management Area. A report to State

Forests of New South Wales. Austeco and State Forests oi New South Wales, Armidale, N.S.W.. 129 pp.

Blake, S.T., Roff, C.. 1988. The Honey Flora of Queensland. Queensland Department of Primary Industries, Brisbane. Qld. 218 pp.

Boorsboom, A. (Ed.). 1996. Flora and Fauna Information System Species Management Manual. Departmefit of Natural Re- sources, Brisbane, Qld, in press.

Braithwaite, L.W., 1984. The identification of conservation a&a:. for possums and gliders within the Eden woodchip &cession district. In: Smith, A.P., Hume. I.D. (Eds.), Possums and Gliders. Surrey Beatty & Sons, Sydney, N.S.W., pp. ?Ot --508.

Braithwaite, L.W.. Turner, J., Kelly, J.. 1984. Studies on the arboreal marsupial fauna of eucalypt forests being harvested for woodpulp at Eden, New South Wales. III. Relationships between fauna1 densities. eucalypt occurrence and foliage nutrients, and soil parent materials. Aust. WildI. Res: 1 1, 41-48.

Craig, S.A., 1985. Social organization, reproduction and feeding behaviour of a population of yellow-bellied gliders, Petauru~ australis (Marsipialia: Petauridae). Aust. Wildl. Res. 12. I- 18.

Environment Science and Services. 1991. Tully-Millstream Hy- dro-Electric Scheme: Yellow-bellied Glider Baseline Study. Prepared for Queensland Electricity Commission. Brisbane. Qld.

Eyre. T.J.. 1993. Distribution, habitat and conservation &tams of the yellow-bellied glider Petaurus austrulis in x!othem Queensland. Prepared for Queensland Department of Primaq Industries Forest Service, Brisbane, Qld.. 90 pp.

Gibbons. P., 1994. Sustaining key old-growth characteristic rn native forests used for wood production: Retention of trees with hollows. In: Norton, T.W.. Dovers, S.R. (Eds.), Ecology and Sustainability of Southern Temperate Ecosystems, CSFRG, Melbourne, Vie., pp. 59-84.

Gibbons. P., Lindenmayer, D.B., 1996. Issues associated with the retention of hollow-bearing trees within eucalypt forests managed for wood production. For. Eco. h4anage. 83. 245-2.79.

Goldingay, R.L., 1986. Feeding behaviour of the yellow-bltied glider Petwus australis (Marsupialia: Petauridae) at Bombala. New South Wales. Aust. Mammal. 9, 17-25.

Goldingay, R.L.. 1990. The foraging behaviour af a nectar feeding marsupial. Petaunts australis. Gecologia 85. 19 I- 199.

Goldingay, R.L.. 1992. Socioecology of the yellow-bellied gilder (Petawus austrulisf in a coastal forest. Aust. J. Zool. 40. X7-278.

Goldingay. R.L.. 1994. Loud calls of the yellow-b&lied glider, Petaurus austmlis: territorial behaviour by an at&real ma&- pial?. Aust. J. Zool. 42, 279-293.

Goldingay, R.L., Kavanagh, R.P.. 1990. Socioecology 01. the yellow-bellied glider, Petuurus azistrcrlb. at Waratah Creek. NSW. Aust. J. Zool. 38, 327-341.

Goldingay, R.L.. Kavanagh, R.P., 1991. The yellow-bellied glider: a review of its ecology. and management considerations. In: Lunney, D. (Ed.), Conservation of Australia’s Forest Fauna. Royal Zoological Society of New South Wales, Mosman. N.S.W., pp. 365-375.

T.J. Eyre, A.P. Smith/Forest Ecology and Management 98 (I9971 281-295 295

Goldingay, R.L., Kavanagh, R.P., 1993. Home-range estimates and habitat of the yellow-bellied glider (Petaurus australis) at Waratah Creek, New South Wales. Wildl. Res. 20, 387-404.

Grimes, R.F., Pegg, R.E., 1979. Growth Data for a spotted gum-ironbark forest in South-east Queensland. Technical Pa- per WO. 17. Department of Forestry Queensland, Brisbane, Qld. 30 pp.

Henry, S.R., Craig, S.A., 1984. Diet, ranging behaviour and social organization of the yellow-bellied glider (Petaurus australis Shaw) in Victoria. In: Smith, A.P., Hume, I.D. (Eds.), Pos- sums and Gliders. Surrey Beatty & Sons, Sydney, N.S.W., pp. 331-341.

Inions, G.. 1990. Classification and evaluation of sites in karri (Eucalyptus diuersicolor) regeneration. I. Edaphic and cli- matic attributes. For. Ecol. Manage. 32, 117-134.

Inions, G.B., Tanton, M.T., Davey, S.M., 1989. Effect of fire on the availability of hollows in trees used by the common brushtail possum Trichosurus uulpecula Kerr, 1792, and the ringtail possum, Pseudocheirus peregrinus Boddaerts, 1785. Aust. Wildl. Res. 16, 449-458.

Kavanagh, R.P., 1984. Seasonal changes in habitat use by gliders and possums in southeastern New South Wales. In: Smith, A.P., Hume, I.D. (Eds.), Possums and Gliders. Surrey Beatty & Sons, Sydney, N.S.W., pp. 527-543.

Kavanagh, R.P., 1987. Forest phenology and its effect on foraging behaviour and selection of habitat by the yellow-bellied glider, Petauru.? australis Shaw. Aust. Wildl. Res. 14, 371-384.

Kavanagh, R.P.. 1991. The target species approach to wildlife management: gliders and owls in the forests of southeastern New South Wales. In: Lunney, D. (Ed.), Conservation of Australia’s Forest Fauna. Royal Zoological Society of New South Wales, Mosman, N.S.W., pp. 377-383.

Kavanagh. R.P., Bamkin, K.L., 1995. Distribution of nocturnal forest birds and mammals in relation to the logging mosaic in southeastern New South Wales, Australia. Biol. Conserv. 71, 41-53.

Kavanagh, R.P.. Debus. D., Tweedie, T., Webster, R., 1995. Distribution of nocturnal forest birds and mammals in north- eastern New South Wales: relationships with environmental variables and management history. Wildl. Res. 22, 359-377.

Kavanagh, R.P., Rohan-Jones, W.G., 1982. Calling behaviour of the yellow-bellied glider, Petaurus australis Shaw (Marsupialia: Petauridae). Aust. Mammal. 5, 95- 112.

Kutt, A.S., 1995. Arboreal marsupials and nocturnal birds in thinned regrowth, unthinned regrowth and old lowland forest, East Gippsland, Victoria. Aust. For. 57, 123-130.

Lindenmayer, D.B., Cunningham, R.B., Tanton, M.T., Smith, A.P., 1990a. The conservation of arboreal marsupials in the montane ash forest of the central highlands of Victoria, southeast Australia: II. The loss of trees with hollows and its implications for the conservation of Leadbeater’s Possum

Gymnobelideus leadbeateri McCoy (Marsupialia: Petauridae). Biol. Conserv. 54, 133-145.

Lindenmayer, D.B., Cunningham, R.B., Tanton, M.T., Smith. A.P., Nix, H.A., 1990b. The conservation of arboreal marsupials in the montane ash forests of the central highlands of Victoria, southeast Australia: I. Factors influencing the occu- pancy of trees with hollows. Biol. Conserv. 54, 1 I1 - 131.

Lindenmayer, D.B., Cunningham, R.B.. Tanton. M.T.. Smith, A.P., Nix, H.A., 1991. Characteristics of hollow-bearing trees occupied by arboreal marsupials in the montane ash forests of the central highlands of Victoria, southeast Australia. For. Ecol. Manage. 40, 289-308.

Mackowski, C.M., 1984. The ontogeny of hollows in blackbutt, Eucalyptus pilularis and its relevance to the management of forests for possums, gliders and timber. In: Smith, A.P.. Hume, I.D. (Eds.), Possums and Gliders. Surrey Beatty & Sons, Sydney, N.S.W., pp. 517-525.

McCullagh. P., Nelder, J.A., 1989. Generalized Linear Models, 2nd edn. Chapman & Hall, London, 5 11 pp.

Milledge, D.R., Palmer, C.L., Nelson, J.L., 1991. Barometers of change: the distribution of large owls and gliders in mountain ash forests of the Victorian Central Highlands and their potential as management indicators. In: Lunney, D. (Ed.), Conserva- tion of Australia’s Forest Fauna. Royal Zoological Society of New South Wales. Mosman, N.S.W., pp. 53-65.

Queensland Forest Service, 1994. Harvesting, Marketing and Re- source Management Manual. vol. 2. Queensland Department of Primary Industries, Brisbane, Qld., 7-8.

Schreuder, H.T., Gregoire, T.G., Wood, G.B., 1993. Sampling Methods for Multiresource Forest Inventory. Wiley, New York.

Scotts, D., 1994. Sustaining sensitive wildlife within temperate forest landscapes: regional systems of retained habitat as a planning framework. In: Norton, T.W., Dover& S.R. (Eds.), Ecology and Sustainability of Southern Temperate Ecosys- tems, CSIRO, Melbourne, Vie., pp. 85-106.

Smith, A.P. (in press). Ecosystem management in Australia. In: Copeland, C., Lewis, D. (Eds.), Ecosystem Management: The Legacy of Science. Halstead Press, Sydney, N.S.W.

Smith, A.P., Andrews. S.P., Gration, G., Quin, D., Sullivan, B., 1995. Description and assessment of forestry impacts on fauna of the Urunga-Coffs Harbour Management Areas. Coffs Har- bour-Urunga Management Area Environmental Impact State- ment Supporting Document No. 4. State Forests of New South Wales, Sydney, N.S.W.

Smith, A.. Russell, R., 1982. Diet of the yellow-bellied glider Petaurus australis (Marsupialia:Petauridae) in north Queens- land. Aust. Mammal. 5, 41-45.

Smith, A.P., Lindenmayer, D.B., 1988. Tree hollow requirements of leadbeater’s possum and other possums and gliders in timber production as forests of the Victorian central highlands. Aust. Wildl. Res. 15, 347-362.

Floristic and structural habitat preferences of yellow-bellied gliders ( Petaurus australis) and selective logging impacts in southeast Queensland, Australia

Documents