Content uploaded by Nicholas Gellie
Author content
All content in this area was uploaded by Nicholas Gellie on May 04, 2016
Content may be subject to copyright.
Native Vegetation of the Southern Forests: South-east Highlands,
Australian Alps, South-west Slopes, and SE Corner bioregions
Nicholas James Holman Gellie
School of Resources, Environment and Society, Building No 48, Forestry, The Australian National University
Canberra ACT 0200 AUSTRALIA. Email: nicholas.gellie@anu.edu.au
Abstract: The Southern Forests study area covers an area of about six million hectares of south-eastern New South
Wales, south of Oberon and Kiama and east of Albury and Boorowa (latitude 33° 02’–37 ° 06’ S; longitude 146° 56’
– 147° 06’ E). The total area of existing vegetation mapped was three million hectares (3 120 400 hectares) or about
50% of the study area. Terrestrial, wetland and estuarine vegetation of the Southern Forests region were classied
into 206 vegetation groups and mapped at a scale between 1: 25 000 and 1: 100 000. The classication was based on
a cluster analysis of detailed eld surveys of vascular plants, as well as eld knowledge in the absence of eld survey
data. The primary classication was based on 3740 vegetation samples with full oristics cover abundance data.
Additional classications of full oristics presence-absence and tree canopy data were carried out to guide mapping in
areas with few full oristic samples. The mapping of extant vegetation was carried out by tagging vegetation polygons
with vegetation codes, guided by expert knowledge, using eld survey data classied into vegetation groups, remote
sensing, and other environmental spatial data. The mapping of pre-1750 vegetation involved tagging of soils mapping
with vegetation codes at 1: 100 000 scale, guided by spatial modelling of vegetation groups using generalised additive
statistical models (GAMS), and expert knowledge. Proles of each of the vegetation groups on the CD-ROM* provide
key indicator species, descriptions, statistics and lists of informative plant species.
The 206 vegetation groups cover the full range of natural vegetation, including rainforests, moist eucalypt forests,
dry shrub forests, grassy forests, mallee low forests, heathlands, shrublands, grasslands and wetlands. There are 138
groups of Eucalyptus forests or woodlands, 12 rainforest groups, and 46 non-forest groups. Of the 206 groups, 193
were classied and mapped in the study area. Thirteen vegetation groups were not mapped because of their small size
and lack of samples, or because they fell outside the study area.
Updated regional extant and pre-1750 vegetation maps of southern New South Wales have been produced in 2005,
based on those originally prepared in 2000 for the southern Regional Forest Agreement (RFA). Further validation and
remapping of extant vegetation over 10% of the study area has subsequently improved the quality of the vegetation
map, and removed some of the errors in the original version. The revised map provides a reasonable representation of
native vegetation at a scale between 1: 25 000 and 1: 100 000 across the study area.
In 2005 native vegetation covers 50% of the study area. Environmental pressures on the remaining vegetation include
clearing, habitat degradation from weeds and nutrication, severe droughts, changing re regimes, and urbanisation.
Grassy woodlands and forests, temperate grasslands, and coastal and riparian vegetation have been the most reduced
in areal extent. Over 90% of the grassy woodlands and temperate grasslands have been lost. Conservation of the
remaining vegetation in these formations is problematic because of the small, discontinuous, and degraded nature of
the remaining patches of vegetation.
Cunninghamia (2005) 9(2): 219–254
Introduction
The Comprehensive Regional Assessments (CRAs) were
designed to provide scientic information to create a
Comprehensive, Adequate, and Representative (CAR)
conservation reserve system on public lands, and at the
same time set up systems of Ecologically Sustainable Forest
Management (ESFM) within designated forest regions
in New South Wales, Victoria, Queensland, Tasmania
and Western Australia. One of the key pieces of scientic
information needed was a map of forests and non-forests
within the forest regions. The JANIS report (JANIS 1996)
placed considerable emphasis on using ‘forest ecosystems’ as
general surrogates for the range of biodiversity. The JANIS
report denes a forest ecosystem as:
The aggregate of plants, animals and other organisms, and
the non-living parts of the environment with which these
organisms interact.
........an indigenous ecosystem with an overstorey of trees that
have a greater than 20% canopy cover. These ecosystems
219
*CD-ROM is inside back cover
220 Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests
should normally be discriminated at a resolution requiring
a map-standard scale of 1: 100 000. Preferably these
units should be defined in terms of floristic composition
in combination with substrate and position within the
landscape.’
The conservation representativeness of forest ecosystems in a
given region was a key factor in the conservation assessment
of a forest region, along with other criteria including ora and
fauna species, wilderness, cultural values and old growth.
Keith & Bedward (1999) have detailed a comprehensive
summary of the validity of forest ecosystems as surrogates
for biodiversity. Section 6.1.1 of the JANIS report also
species additional requirements relating to the derivation
and use of forest ecosystems in the CRA/RFA process:
These broad specications provided the framework for
developing a regional scale vegetation map for the Southern
Forests of New South Wales. The approach adopted was
designed to be consistent with the vegetation classication in
the Eden CRA Region (Keith & Bedward 1999); to incorporate
the range of eld survey and mapping data already available;
and to use an overstorey mapping layer to help delineate the
extent of each forest and non-forest ecosystem. In this paper
forest and non-forest ecosystems are synonymous with the
more generic term vegetation groups.
This study presents up-to-date versions of the extant and pre-
1750 vegetation maps of the Southern Forests, based on a
thorough revision of the original CRA work completed in
2000. The specic objectives of this study are to document
the data and methods used to produce these maps, to illustrate
the more recent changes and revisions, to describe a newly
developed hierarchical vegetation classication scheme and
to indicate the current status of vegetation groups across the
study area.
The Southern Forests study area
The Southern Forests study area covers six million hectares
(6 174 400 hectares) of south-eastern New South Wales
(latitude 33° 02’–37° 06’ S; longitude 146° 56’–147° 06’ E)
(Fig. 1). The northern boundary extends from the Crookwell
and Oberon Plateaus, across to the southern part of the
Illawarra near Kiama. The eastern boundary follows the
South Coast down to the boundary of the Eden Forests region.
The south eastern boundary excludes the South-East Forests
region and then follows the NSW state border to Albury. The
western boundary more or less follows the Hume Highway
to Bookham and then heads due north to the Abercrombie
River. The study area includes hardwood forest regions,
based on NSW State Forest’s administration regions at the
time of the CRA, and areas covered by the CRA wilderness
assessment including:
• the Jamberoo escarpment, up to the Southern Highlands
highway between Moss Vale and Kiama;
• small areas of southern Blue Mountains National Park,
for wilderness assessment;
• hardwood forests of Jenolan State Forest adjoining
Kanangra-Boyd NP;
•Pulletop and Livingstone State Forests, south of Wagga.
Landscapes
The landscapes of the study area fall within ve IBRA
bioregions (Thackway & Creswell 1995): Australian Alps,
South Eastern Highlands, South-East Corner, Sydney Basin,
and the NSW South-Western Slopes (Fig. 1).
The elevated plateau of the Australian Alps bioregion extends
from the Victorian border to the Australian Capital Territory.
The highest point is Mount Kosciuszko, 2228m above sea
level. Altitude decreases to 1350 metres in the northern part
of the Alps, within the Australian Capital Territory. The
Australian Alps were formed primarily from large intrusive
granite batholiths, interspersed with metamorphic schists,
gneists, and acid volcanics.
The Murray River drains the high western escarpments of
the Australian Alps. Landscapes on the western footslopes
have formed on granite and granodiorites, and Ordovician
metasediments, including siltstones, mudstones and shales.
Quaternary riverine alluvium lls the valleys of the larger
streams and rivers.
On the south-eastern side of the Australian Alps, the Snowy
River dissects the dry rainshadow areas of the Byadbo and
Delegate catchments of mainly Silurian siltstones and shales.
This dry landscape forms part of the South-East Corner
bioregion and shares afnities with the drier rainshadow
areas of the south coast.
The Murrumbidgee River and its tributaries drain the central
eastern and northern sides of the Kosciuszko and Brindabella
Ranges. On the eastern side of these mountain ranges, the
landscape is mainly undulating high plateaus 600 to 900 m
in elevation with intervening high ranges up to 1400 m, such
as the Yaouk, Cooma, Tinderry, Numeralla, and Gourock
Ranges. The geology of the Numeralla and Gourock ranges
consists of a complex series of Cambrian-Ordovician
schists. The Yaouk and Cooma Ranges consist of Ordovician
and Silurian sediments and shales, interbedded with acid
volcanics in the ranges further to the East. The Cooma-
Monaro plains comprise Tertiary basalt. Further to the south
and east of these plains, Ordovician and Silurian sandstones
are interbedded with Silurian and Devonian granites and
Ordovician metasediments. Cainozoic alluvial plains can be
found on the broad plains to the west of Goulburn in the
upper parts of the Lachlan River catchment. The landscapes
that have formed on these geologies fall largely within the
South Eastern Highlands bioregion.
The abrupt fall on the eastern escarpment coincides with the
catchments of the eastern owing river systems dissecting
the broad Morton plateau in the north and the Deua and
Wadbilliga ranges in the central and southern parts of the
South Coast escarpment. These ranges comprise Ordovician
Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests 221
conglomerates, sandstone, siltstone, and shale, overlying
a thin belt of Comerong acid volcanics, which produce
some of the rugged high outcrops in the Deua ranges. The
foothills of the South Coast are undulating low hills made
up of Ordovician shales, conglomerates, and siltstones,
interspersed with scattered basalt outcrops at Mount
Dromedary, south-west of Narooma, and at Mount Durras,
north of Batemans Bay. These landscape types extend from
the South-East Forests region in Eden up to the edge of the
Morton and Budawang ranges in the lower and middle parts
of the Clyde valley to underlie the coastal foothills extending
southwards through Yadboro and Dampier State Forests.
Alluvial landscapes are found in the lower Shoalhaven, Deua
and Tuross River catchments. The coastal area between
Bermagui and the Illawarra has numerous estuarine lake
systems. Granite batholiths intrude along the coast and are
found west of Nelligen, around Moruya and Bodalla, and in
the central section of the Tuross catchment. These landscapes
extending up from Victoria through Eden and Bega form the
northern part of the South-East Corner bioregion.
Permian siltstones, shales, and sandstones dominate the
geology of the Morton and Budderoo plateaus on the northern
part of the South Coast escarpment and extend in a wedge
down to Mount Durras, north of Batemans Bay. These are part
of the Sydney Basin bioregion. The north-eastern boundary
of the study area straddles the Illawarra escarpment, a high
rainfall part of the Sydney Basin bioregion.
Fig 1. The Southern Forests study area showing bioregions
222 Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests
The northern part of the study area comprises undulating
high plateaus of granite and basalt, surrounded by Devonian
and Ordovician sandstone, quartz sandstone, siltstones and
shales. These plateaus include Oberon, Kanangra-Boyd,
and Crookwell-Taralga, and, with elevations from 900 to
1300 m, dene the Great Divide. To the east of the northern
high plateaus, highly dissected hills 300 to 400 m in elevation,
rapidly descend to the Kowmung and Wollondilly Rivers.
In the dissected valleys, Devonian quartzites, sandstones,
and shales predominate.
Climate
Mean annual rainfall ranges from 465 mm at Cooma on the
Monaro plains to over 2300 mm at Charlottes Pass in the
Snowy Mountains. Rainfall is lowest on the Monaro plains,
increasing east towards the coast and west to the mountains.
Deep valleys or lower plateaus on the leeward side of higher
mountain ranges result in areas of lower rainfall, examples
of which are found in the rain-shadow valleys of the
Araluen Valley, the Monaro plains and the middle Snowy
and Murrumbidgee catchments. By contrast, average annual
rainfall above 1200 mm is found on the Kanangra-Boyd
plateau, and along the South Coast escarpment extending
from Kiama down to the Wadbilliga ranges.
During winter snowfalls are common above 1000 m on the
Australian Alps and the Central and Southern Tablelands.
Snow covers the higher parts of the Australian Alps, above
1600 m, for periods up to four months. Heavy frosts are a
frequent occurrence over most of the South-East Highlands
and Australian Alps during the winter months, becoming
more severe in broader deep valleys fringing the mountain
ranges, as a result of cold air drainage.
Mean annual temperature ranges from 2° C near Mount
Kosciuszko, to 16° C along the coastal plains north of
Ulladulla. Temperature increases as the elevation falls in
both easterly and westerly directions away from the higher
mountain ranges and the Great Dividing Range, which splits
the region in half from north to south (NPWS 1998). These
extremes greatly inuence the distribution and pattern of
vegetation. Seasonality of rainfall also varies across the study
area. The northern coastal escarpment around Kiama shows
a strong summer peak in rainfall, while the western side of
the Kosciusko ranges shows a strong winter rainfall peak.
Land Use History
European land use across the study area has been mainly
forestry, agriculture, mining and human settlement. On the
Central Tablelands and Western Slopes, large areas were opened
up to pastoral development from the 1830s onwards. Much of
this land was covered in grassy woodlands. The steeper, more
infertile and more remote forested land on the Tablelands
was either occasionally cleared or left for rough grazing and
selective logging. Evidence of these activities is still present in
the forest structure on private or public land today.
Mills and Jakeman (1995) provide a detailed account of
the land use changes in the Illawarra and adjacent Southern
Highlands. They document the remorseless cutting and
clearing of the Kiama and Shoalhaven coastal foothills, as
well as the forests on basalt. Harvesting of Toona australis
(Red Cedar) commenced in 1810, following the rst
allocation of land grants to settlers. By 1850 the settlers
had cut out most of the cedar trees. Expansion of farming
caused the clearance of coastal rainforests and forests in
the southern Illawarra, Gerringong, and lower Shoalhaven
causing major losses to the Illawarra Brush. Clearing of
sub-tropical rainforest on the coastal lowland rainforests
began after the 1860s. The Yarrawa Brush and the tall
moist forests on basalt near Robertson were cleared for
agriculture. Kangaroo Valley was not settled till the 1870s
because of poor access, but the open grassy forests in the
valley were gradually cleared for farming thereafter.
From about 1880 until 1950, signicant areas of Crown Land
in the central and northern parts of the study area were made
available for free selection and selective timber cutting.
State forests were established during this period to protect
the forest resource, in response to excessive forest clearing
for agriculture, grazing, and mining. Bago, Buccleuch and
Meragle State Forests in the western highlands had a long
history of forest harvesting since the late 1800s. These sub-
alpine forests were also grazed from time to time under
occupational grazing permits. From 1900 to the early 1950s
state forests were the principal means of conserving forests,
although more intensive harvesting became more widespread
as bullock teams, axe, and crosscut saws were replaced with
mechanised machinery, chainsaws, and a more extensive
road network to gain access to more of the timber resource
after the Second World War.
A succession of major wildres burnt through the Snowy
Mountains and Brindabella Ranges in 1926, and again
in 1939, causing signicant soil loss, and changes in the
forest structure (Costin 1954). These res, coupled with the
intense grazing of the alpine areas, brought about the need
to protect water catchments for hydro-electricity and water
conservation schemes. In 1944 Kosciuszko State Park was
gazetted over the alpine and sub-alpine areas of the Snowy
Mountains plateau.
With the formation of the National Parks and Wildlife
Service in 1967, further Crown Land was dedicated as
National Park including Kanangra-Boyd and southern Blue
Mountains National Parks in 1975, and Morton National
Park in 1976. Although run by a conservation trust from the
1930s, Deua and Budawang National Parks were formally
gazetted in 1977, and Abercrombie and Brindabella
National Parks in 1996.
Extensive pine plantations were established on State Forests
on the eastern slopes of the Tumbarumba and Tumut districts,
on the Oberon Plateau and in Tallaganda State Forest during
the 1960s and 1970s. These pine plantations were established
on areas cleared of native Eucalyptus forest. Clearing of native
Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests 223
forest for pine plantation ceased in the late 1970s as public
concern grew about the impact of large scale conversion of
publicly owned native forest to pine plantation.
During the 1960s, private plantation schemes tried to convert
signicant areas of forest into pine plantation. These schemes
concentrated on clearing of woodlands and heathland in the
Oalen Ford-Braidwood area. Because the schemes overlooked
the infertility and shallowness of the soils, patches of failed
pine plantation remain amongst partially disturbed areas of
native forest in this area. Pine plantations since then have
been established on privately owned pastoral land.
On the South Coast the more rugged and steep areas remained
in public tenure, mainly as State Forest and Crown Land.
Since the 1880s the more fertile ats and valleys have been
cleared for small dairy farms and residential development
close to towns. Since the 1970s, much of the Crown Land
along the South Coast escarpment has been progressively
dedicated as national park. National parks were established
to protect the coastline environment during the 1980s
and 1990s and since 1990 tourism has become a major
contributor to regional economies of the South Coast, ACT,
and the Snowy Mountains. The introduction of intensive
tourism or nature-based tourism poses new challenges for
nature conservation.
Methods
Auditing Vegetation Samples
A thorough compilation of eld survey data was made from
major surveys carried out by CSIRO (Gunn et al. 1969, Austin
1978, Austin and Cawsey 1996, Austin et al. 1996, Doherty
1996, 1997, 1998a, 1998b), State Forests (Binns 1997a and
1997b, Jurskis et al. 1995, State Forests 1999a, 1999b, and
1999c) and individual consultants such as Gilmour (1982,
1983, 1985, 1987), Helman (1983, 1988), Ingwersen (1972,
1983, 1988), Keith and Bedward (1999), Mills (1996a, 1996b
and 1999), and Taws (1997). These eld surveys datasets were
then sorted into four categories of data and classied according
to their major roles in the project (Table 1).
Category I data, comprising full oristics with cover
abundance samples, were used in the analysis and derivation
of vegetation ecosystem groups. The primary classication
method was based on a hierarchical classication method
described in Keith and Bedward (1999) in the Eden RFA
Region, using procedures developed by Belbin (1991).
The selection of a classication dataset followed Keith
and Bedward (1999). Flora data selected for classication
fullled the following criteria:
• Samples with a location on the Australian Map Grid,
with a precision of at least 100 metres;
• The area of plot samples conned to 0.04 or 0.1 hectares;
• Complete list of vascular plant species within the plot;
and
• Species abundances recorded on the six point Braun-
Blanquet cover abundance scale or on a three point scale
used by Gilmour (1982, 1983, 1985, 1987), and Helman
(1983, 1988).
The Gilmour and Helman full oristic survey data, met
three out of the four inclusion criteria for classication of
vegetation groups and their slightly different scaling of cover
was considered similar in its assessment of cover abundance
on most survey plots, and comparable with survey samples
assessed using the Braun Blanquet scaling method.
Category II data were used to derive a presence-absence
classication. The classication method used was identical
to that used with Category I data, except presence-absence
measures was used instead of the Braun Blanquet scale. For
comparison between the classications developed for both
category I and II data, survey data with Braun-Blanquet
cover abundance data (Category I) were converted into
presence-absence scale.
Categories III and IV were primarily intended for classication
of canopy trees to check the quality of the mapping of canopy
vegetation and support the mapping of vegetation groups.
The classication method used with a combined dataset of
categories III and IV was identical to that used with category
II data.
Field Survey Methods
Survey work conducted during the CRA project was designed
to ll environmental and geographical gaps across the study
area to produce an equal spread of sites across the range of
predened environmental strata. In the western and northern
sub-regions, the absence and/or scarcity of substantial
patches of remnant vegetation with an undisturbed ground-
cover, was a major limitation in sampling ora sites evenly
across all environmental strata.
Table 1. Field Survey Data Categories
Data Flora Survey Project Role
Category Data Type
I Full floristics, Full Floristic classification
cover abundance – derivation of vegetation groups
II Full floristics, Full Floristic classification
presence/absence – support vegetation mapping in
critical geographic gaps of
Category I survey data
III Canopy only, Canopy classification – guide
presence/absence mapping of vegetation groups and
check quality of API mapping, and
validation
IV Partial floristics, Similar role in project to category
cover abundance III survey data
224 Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests
Within identied gaps, sites were located across a variety of
topographic positions to sample as much of the range of local
environmental gradients as possible. Survey samples were
located in areas with adequate access to maximise survey
efciency. Where possible, four locations were sought: one
on a sheltered midslope, an exposed midslope, an exposed
ridge and in a sheltered gully. Within restricted environmental
strata, generally less than 5000 hectares, and where access
was restricted, fewer sites were surveyed. Sites were
positioned as close as possible to the targeted eld location
while avoiding ecotonal boundaries between different forest
and non-forest vegetation; or heavily disturbed sites.
To reect ‘pre-1750’ vegetation patterns as much as possible,
less disturbed areas based on local knowledge and the expert
opinion of the botanists involved, were surveyed. One of the
greatest limitations in extrapolating to pre-1750 vegetation
patterns, especially those mostly cleared, is unknown past
disturbance.
Field survey methods followed plot-based sampling used in
previous CRAs in NSW (DUAP 1999). A standard site data
sheet was based on the one developed for the North-East
CRA ora surveys. The survey approach included collection
of oristic, structural, and physical data in a plot 20 metres
by 20 metres in size, nested within a larger plot of 20 m
by 50 m. This was to ensure compatibility with larger scale
regional surveys of canopy species that had used quadrats
20 m by 50 m in size, while at the same time maintaining
compatibility with other full oristic data collected using
the smaller quadrat size. Floristic data recorded in the
20 m by 20 m plot included a list of all vascular plant species
present and a visual estimation of the cover abundance of
each species according to a modied Braun-Blanquet system
of cover abundance classes (Poore 1955).
Details of each site were recorded, including locality
description, map name, Australian Map Grid easting and
northing, area name, stratication and site number. Within
a 50 m by 20 m plot, physical data such as altitude, slope,
aspect, horizon visibility and azimuth, soil depth and type,
geology, land morphology and element and the presence
of outcropping were also recorded. After completion of
the survey of the vascular plants present, community
structure was estimated over an area of 50 m by 20 m in
size. The structure of the vegetation in terms of the dominant
vegetation layers was described, and could include an
overstorey layer, up to three mid-stratum layers, and up
to three lower or ground layers. The predominant growth
form in each stratum, the upper and lower heights and the
percent cover, as well as three most dominant species, were
recorded. Growth stage of senescent, mature or regrowth
categories was also estimated within the plot.
Disturbance history of re, logging and grazing was also
noted, estimating severity and time since disturbance.
The accuracy of these estimates varied, depending on
the apparent eld evidence available, and the subjective
assessment of eld botanists. Finally, an estimate of the
overall condition of the site was made, incorporating factors
such as the degree of weed infestation as well as the species
and structural diversity.
Preparation of Survey Data
Survey data (sites, oristics, and a master species list) were
stored in an electronic database to allow error checking,
manipulation, and retrieval for analysis. Site records were
checked for geographical accuracy by comparing the relevant
1: 25 000 map sheet against site locality descriptions and
grid coordinates as they appeared on the survey forms.
This information was compared to the location of the site
in ArcView GIS. Accuracy to within 100 m or better was
generally achieved. All grid references were checked in
the GIS to ensure that they fell within the Southern CRA
boundary. Throughout the classication process, there
were random checks of relationships between species
associations and physical site characteristics such as
morphology and geographic position, and corrected any
errors.
Duplicate record checks were conducted in all the database
tables especially the oristics table. Discrepancies found in
oristics records were identied and corrected where possible
or quarantined from analysis. Checks were also conducted to
ensure that all species in the oristics table had either a cover
abundance value or a letter P to indicate species presence.
The database was used to compile a species inventory
to check nomenclatural standards and errors in eld
identication. Synonyms were identied and standardised
according to Harden (1990, 1991, 1992 & 1993) or more
recent publications. Species occurrences were compared
to published geographical ranges to identify errors in
identication or data entry. Where species names had
simply been changed, names were updated, but in many
cases a previous plant taxon may have been split into two
or more new plant taxa. In these cases the geographic
distribution and ecological niches of the new plant taxa
were determined and site data information updated
accordingly. Plant taxa were aggregated to the species
level unless there was some clear geographic or ecological
distinction to enable them to be consistently distinguished in
the data set.
Some of the eld survey data, such as Ingwersen (1992), had
their cover values converted to Braun-Blanquet scores so
that they could be used in the full oristics cover-abundance
classication dataset. Where possible the data were referred
to the original surveyor for advice on conversion to the
Braun-Blanquet scoring method. Unknown species, as well
as exotic species and non-vascular plant species, which
had been recorded in some data sets, were masked out of
the data set to be used for classication. The masking
of these categories of plant species helped to remove any
confounding inuences of anomalous records of native plant
species and the inuence of non-native species data on the
PATN classication.
Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests 225
Environmental spatial data
A variety of spatial data layers describing climate, terrain,
and lithology were compiled from spatial data prepared at
a mapping scale of either 1: 25 000 or 1: 100 000 (Table
2). Climate and terrain variables were derived from digital
elevation data provided by the NSW Land Information
Centre, using algorithms described by Hutchinson (1989).
Most of the data layers were prepared at a resolution of
1: 100 000, which equates to a 100 metre grid cell size.
The climate, terrain, and geology layers were collated
and prepared for use in modelling of the spatial extent of
vegetation groups, whereas the CRAFTI API layer was used
as the primary data layer for mapping extant vegetation.
LANDSAT Thematic Mapper imagery of the whole region
was acquired to facilitate the planning of eld work, checking
of survey data, and to assist mapping of vegetation. Two
complete Landsat scenes were prepared as TIFF images,
using bands 3, 4, and 5. The data projection of the remotely
sensed data was set to Zone 55, AGD66, which followed the
standard set in the CRA. Two LANDSAT scenes were then
digitally enhanced and cut into 8 overlapping sub-scenes that
covered the full extent of the study area, with the exception
of the eastern end of the Jervis and the Beecroft Peninsula in
the Shoalhaven area.
Air Photo Interpretation (API) and Mapping
A classication of canopy presence-absence forest data,
collated and analysed by Austin and Cawsey (1996) was
adapted to enable a more direct comparison between eld
survey data and tree canopy associations. A region-wide
classication of tree canopy species formed a better basis on
which to associate eld survey samples with API mapping
data than forest types based on Research Note 17 (Baur
1989). A correspondence table between canopy oristic
codes and Austin’s canopy associations was developed to
guide the coding of polygons of canopy oristic types drawn
on each air photographic overlay. The detail of these eucalypt
forest codes can be found in a separate report (DUAP
2000a). The coding of rainforest and non-forest component
of the vegetation was based on a collaborative expert agreed
classication. The non-forest API codes covered the full
range of non-eucalypt and non forest types in the study area:
rainforest, tall shrubland, heath, swamp, native grassland,
and coastal estuarine vegetation.
Structural and oristic data layers for all forested lands were
derived from 1997-98 air photos at a scale of 1: 25 000 (except
in the Crookwell area where 1: 50 000 scale photos were used).
Contract air photo interpreters used stereoscopic equipment
to delineate canopy oristics and structural information onto
overlays mounted on 1: 25 000 scale air photographs of each
1: 25 000 map sheet. Contextual information available to
interpreters included other existing vegetation mapping and
survey data. Familiarisation and eldwork was limited to
two to three days per 1: 25 000 mapsheet, depending on the
coverage of native vegetation, the difculty of access, and the
extent of previous API work. Contractual constraints conned
the API mapping to forested areas outside of forested land in
urban areas, intensively developed agricultural land and rural-
urban subdivisions, as well as some large pine plantations
within eucalypt forest remnants. Within the fragmented
woodlands and forests in the western part of the study
area, additional remnant vegetation patches were mapped
subsequently by a small team contracted to validate and rene
the API mapping (Maguire & Hunter 2000). A comprehensive
account of the API methods is contained in a separate report
(DUAP 2000a).
Following scanning and rectication, the layer was merged
with previous canopy oristics mapping undertaken by State
Forests using Research Note 17 (Bauer 1989) and existing
mapping within selected national parks and nature reserves
(Doherty 1996, 1998a). The canopy oristics mapping show
patches of extant forest down to 10 hectares in size and patches
of non-forest vegetation down to 2 hectares. It covers most of
the study area with the exception of an area north of Holbrook.
Mapping of native grasslands was not undertaken in the
canopy oristics mapping, being outside the scope of the
CRAFTI project. A mapping layer of pre-1750 grasslands
through the South-Eastern Highlands bioregion was
produced, based on earlier mapping in the southern part of
the study area (Costin 1954), with additional mapping in the
ACT, Gunning, Crookwell, Mulwaree, Oberon, Tallaganda,
and Yarrowlumla Shires (Rhewinkel & Gellie unpubl.). Fig.
2 shows the nal compiled temperate pre-1750 grasslands
map. In 2005 the larger patches of temperate grasslands still
surviving on Crown reserves and travelling stock routes
were mapped.
Historical Parish Portion Data
State Forests of NSW (1999a) collected historical information
on vegetation cover from a sample of original portion plans
in parishes in the study area. Information that was collected
for each portion plan included the map sheet name, parish
name, portion number, plan number, date of record, notation
on vegetation type and topography, point data on corner
trees used to x survey points, comments or notes and grid
references of corner survey points. SFNSW (1999a) cited a
study by Ryan and Stubbs (1996), which showed that where
the destruction of the vegetation has been complete, the
historical record, and in particular the conditional purchase
plans, was indispensable in reconstructing the pre-settlement
pattern of vegetation.
Where relevant, the historical data layer was used in a
contextual sense by expert botanists when assigning forest
types to cleared areas, but limitations of the data set included
the overly broad nature of many of the early surveyor
descriptions, such as gum, peppermint, or stringybark,
which made it difcult to correlate with individual eucalypt
species.
226 Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests
Classification
Classication of the data was undertaken using hierarchical
classication techniques (Belbin 1991). The approach to
classication was based on methods described by Keith
and Bedward (1999). The analyses were carried out using
the PATN computer program (Belbin 1994). Dissimilarity
between samples was calculated using the Kulzynski
coefcient applied to un-standardized cover-abundance data.
Clustering of samples into a hierarchical classication used
the unweighted pair-group arithmetic averaging technique
(UPGMA). The initial number of vegetation classes within
the classication dataset used in this project was determined
using the homogeneity program (Bedward 1999). This
program measures the extent to which further splitting of
groups yields improved homogeneity of groups.
A dendrogram of the classication analysis was used to view
the results and to rene the classication. Appendix 1a on the
CD-ROM provides a summary view of the nal classication,
based on a grouping of the vegetation groups into vegetation
mapping classes. The suggested vegetation groups in the
classication were evaluated according to three criteria: the
groupings of samples made sense oristically and spatially;
the groups were readily identiable in the eld; and the groups
related to the vegetation patterns apparent in the landscapes
of the study area. The rst step was to check that each group
was oristically distinctive. The second step involved a spatial
assessment of each vegetation group. Further splits were
Table 2. Spatial data layers used in modelling and mapping
(adapted from Thomas et al. 2000)
GIS Layers Resolution of
Data
Climate
Evaporation Average 1: 100 000
Precipitation Annual Average 1: 100 000
Precipitation of the Coldest Quarter 1: 100 000
Precipitation of the Driest Period 1: 100 000
Precipitation of the Driest Quarter 1: 100 000
Precipitation of the Warmest Quarter 1: 100 000
Precipitation of the Wettest Period 1: 100 000
Precipitation of the Wettest Quarter 1: 100 000
Precipitation Seasonality 1: 100 000
Temperature (Maximum) of the Warmest Period 1: 100 000
Temperature (Mean) of the Coldest Quarter 1: 100 000
Temperature (Mean) of the Driest Quarter 1: 100 000
Evaporation Average 1: 100 000
Precipitation Annual Average 1: 100 000
Precipitation of the Coldest Quarter 1: 100 000
Precipitation of the Driest Period 1: 100 000
Precipitation of the Driest Quarter 1: 100 000
Precipitation of the Warmest Quarter 1: 100 000
Precipitation of the Wettest Period 1: 100 000
Precipitation of the Wettest Quarter 1: 100 000
Precipitation Seasonality 1: 100 000
Temperature (Maximum) of the Warmest Period 1: 100 000
Temperature (Mean) of the Coldest Quarter 1: 100 000
Temperature (Mean) of the Driest Quarter 1: 100 000
Temperature (Mean) of the Warmest Quarter 1: 100 000
Temperature (Mean) of the Wettest Quarter 1: 100 000
Temperature (Minimum) of the Coldest Period 1: 100 000
Temperature Annual Mean 1: 100 000
Temperature Annual Range 1: 100 000
Temperature Average 1: 100 000
Temperature Average Maximum 1: 100 000
Temperature Average Minimum 1: 100 000
Context
Rivers and Streams 1: 100 000
Roads and Easements 1: 25 000
Insolation
Solar Radiation Annual Total (11 month) 1: 100 000
Solar Radiation of the Coldest Quarter 1: 100 000
Solar Radiation of the Driest Quarter 1: 100 000
Solar Radiation of the Warmest Quarter 1: 100 000
Solar Radiation of the Wettest Quarter 1: 100 000
Solar Radiation Seasonality 1: 100 000
Lithology & Soils
Lithology divided into 5 broad groups 1: 100 000
Lithology divided into 13 groups based on
mineralogy and geomorphology 1: 100 000
Soils Landscapes Mapping 1: 100 000
Acid Sulphate Soils Mapping 1: 25 00
Terrain
Digital Elevation Model 1: 25 000
Digital Elevation Model 1: 100 000
Aspect 1: 100 000
Distance from the Coast 1: 25 000
GIS Layers Resolution of
Data
Ruggedness (3) - Standard Deviation of
Elevation Values within a 1: 100 0000m
window around the cell 1: 25 000
Skidmore Index (position on a slope from ridge
to gully) 1: 25 000
Slope (degrees) 1: 25 000
Topographic Position of a cell in relation to the
mean of a 250m window around the cell 1: 25 000
Topographic Position of a cell in relation to the
mean of a 500m window around the cell 1: 25 000
Topographic Position of a cell in relation to the
mean of a 1000m window around the cell 1: 25 000
Wetness Index (volume of water draining into a cell,
adjusted for ability to retain water due to slope) 1: 100 000
Vegetation
Eastern Bushlands Vegetation Database (EBDB)
(NPWS 1995) 1: 100 000
API Mapping of Canopy Floristics – State Forests
Data (SFNSW 1989) 1: 25 000
API Mapping of Canopy Floristics – CRAFTI
Data (DUAP 2000a) 1: 25 000
API Mapping of Canopy Floristics – CSIRO Data
(Doherty 1996, 1997, 1998a, 1998b); 1: 25 000
API Review of CRAFTI Mapping Nicholas Graham-
Higgs Pty Ltd (2002a, 2002b, 2002c, 2004, 2005)
and EcoGIS (2002, 2004a, 2004b, 2004c) 1: 25 000
Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests 227
made if sub-groups were oristically distinct and spatially
or ecologically distinct. The distribution of samples of each
vegetation group was also checked at the same time within
Arcview GIS to reveal the location of the samples in relation
to geology, terrain and exposure. The oristic composition of
geographic outliers of the vegetation samples were checked
within the site oristics database to determine if these
samples belonged to other vegetation groups. A vegetation
group was further split if it contained a clearly dened sub-
group at a slightly lower level in the dendrogram tree. This
was only undertaken where the identied sub-group, and the
remainder of the original broader vegetation group could
be readily related to environmental variables and mapped,
or had previously been identied and mapped, and as such,
met the JANIS requirements for a forest ecosystem. In some
instances some closely related groups were lumped if they
could not be related to geology or terrain.
Patterns of elevation, lithology, exposure, and other
topographic variables were related to the spatial distribution
of samples of each vegetation group on the GIS. For example,
if the samples of a vegetation group, such as Yellow Box
Grassy Woodlands, were found within the expected range
of that vegetation group, such as on valley oors with richer
Fig. 2. Map of Pre-1750 Grasslands
228 Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests
soils, then those relationships were used to map that vegetation
group. This process was repeated for each vegetation group
until all groups had been checked on the GIS. Where there was
no recognisable relationship in the distribution of the sites of a
vegetation group on the GIS, further investigations of possible
splitting or lumping of that group were undertaken. Checks on
the oristics of that sample were further undertaken to identify
any eld identication or nomenclature errors. The rst two
iterations in the classication and checking of vegetation of
groups on the GIS, revealed some vegetation groups with no
obvious spatial pattern. This necessitated some revision of
species nomenclature and exclusion of eld survey samples
with errors in eld identication or with low numbers of
native species present.
An iterative classication and validation process was
followed by which classication was checked and errors in
eld identication and nomenclature removed, leading to a
usable and stable classication of eld survey data. Small
errors in the identication of common species appeared
to affect the vegetation classication. Where errors were
suspected, the plant species and their cover abundances
were investigated. If errors were found, any suspect plant
identications and distributions were checked with Flora
of NSW (Volumes 1–4), and with other eld botanists.
Amendments were made to the original database le. The
PATN analysis was re-run and the positions of sample sites
were checked on the dendrogram, to see if the changes made
any differences to the results. There were a few instances
where anomalous groups that either made no geographic
sense or had no obvious indicator species were formed.
The sites in these anomalous vegetation groups were either
reallocated to other vegetation groups or were not used in the
vegetation mapping.
After satisfactory relationships between each of the vegetation
groups and geology and terrain were found on the GIS, the
iterative classication process was ended. Every site within
the PATN Dendrogram was assigned its nal vegetation
number and the FIDEL program was run to provide a nal
prole of species found in each vegetation group. A le
containing all the oristic samples allocated to vegetation
groups in the nal classication was then imported into
the GIS for use in the vegetation mapping across the study
area. The groups derived from this nal classication step
then were dened as vegetation groups, for the purposes of
mapping, description, and reporting. In the original version
of the vegetation mapping, these were termed forest or non-
forest ecosystems, depending on whether they were forest or
non-forest vegetation (Thomas et al. 2000).
A further classication of the vegetation groups into
vegetation classes was undertaken using the same method
to derive the primary classication (Appendix 1a on the CD-
ROM). Instead of the eld samples being treated as objects,
a cluster analysis was performed on the vegetation groups
to lump the vegetation groups into broader vegetation types.
This further analysis produced vegetation classes that became
the next level in the vegetation hierarchy. The vegetation
classication of the study area could now be analysed or
viewed at two levels instead of one created from the original
cluster analysis. To complete the hierarchy, a third level
vegetation formation was added to conform to the state-wide
classication of Keith (2004).
Description of Vegetation Groups
The FIDEL software program (Bedward 1999) was used
to prole indicator species of groups at any level within a
vegetation classication. This program calculates the relative
median cover-abundance and frequency of occurrence of
plant species within and outside a vegetation group. The
default cut-offs in the FIDEL were used to dene the FIDEL
output values. The default cut-offs were attribute scores set
at the 50th percentile; group score cut-offs set to a cover
abundance of 2 and group frequency of 50%, non-group
scores set to a cover abundance of 2 and group frequency of
50%. Table 3 summarises the denitions of indicator species
within vegetation groups.
An indicator or diagnostic species has high relative frequency
within the Target Vegetation Group and a relatively low
frequency in other vegetation groups. These plants species
are usually described as Positive Species. Plant species
which are described as Negative or Constant are considered
unreliable indicator species. In some instances, species that
occurred at a frequency between 30 and 50% were still
regarded as useful indicator species, even though FIDEL
labelled them as uninformative. These are described as
‘Less Informative’ species in Table 3. When combined with
‘Positive Diagnostic’ indicator species they contribute to the
description and prole of a vegetation group.
Based on the information summarised in the FIDEL
le, a description of each vegetation group was derived.
The vegetation groups were assigned vernacular names
following the conventions of Keith and Bedward (1999).
The last part of the name describes the vegetation structure,
such as rainforest, forest, woodland, scrub, heath, swamp,
or grassland. The name may include a place name if it is
found in a distinct part of the study area. A description of the
understorey is generally included to give a more complete
description of the vegetation group.
Descriptive proles of vegetation groups, the rst level in the
hierarchy, were produced to enable identication of these in
the eld. These vegetation proles are available on the CD-
ROM accompanying this paper (Appendix 5). Each prole
includes information about the oristic composition, structure,
habitat, and occurrence of each ecosystem. Floristic data is
presented in a table showing vegetation group frequency
and cover-abundance of each species within the described
unit, together with its frequency and abundance in all the
other vegetation groups. Fidelity classes in the table describe
possible indicator species in each vegetation group.
Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests 229
The naming convention of the second level in the hierarchy,
the vegetation class, is based on regional descriptors and
levels of moistness in the understorey, where appropriate
to give more detailed vegetation descriptions that reect
the environmental and physiographic conditions of that
vegetation class. A regional descriptor was based on
botanical regions, such as South Coast, Southern Tablelands,
and South-West Slopes. If a vegetation class fell within a
particular part of the study area, it was given a sub-regional
name descriptor, such as Lower Snowy or Morton Plateau
to help with locating that vegetation class within the study
area. Finally a vegetation structural descriptor completed the
description of the vegetation classes, using commonly used
terms for vegetation formation or sub-formations, such as
rainforest, moist eucalypt forests, ash type eucalypt forests,
montane eucalypt forests, dry shrub eucalypt forests, grassy
eucalypt woodlands or forests, grasslands, heathlands, and
shrublands, swamps and swamp heaths, sub-alpine and
alpine herbelds and ferns, and coastal vegetation. The
term moist shrub or layered shrub forest replaces the term
wet sclerophyll, as some forests in this formation can have
a variety of understorey types. For example Vegetation
class 2c, Moist Layered Shrub Forest, has a mixed layered
understorey, comprising rainforest, broad and ne leaved
shrub plant species. A second example is vegetation class 2d,
Tableland Grass/Fern Forest, has or a mixed grassy or ferny
ground layer, with a layer of scattered shrubs, depending on
the site and past disturbance.
The third and highest level in the hierarchy of vegetation
classication is the vegetation formation. The naming
convention for a vegetation formation follows of Keith
(2004), with some additional formations to cover Grassy
Forests, and Montane Tablelands Forests, Wet Heaths/
Swamps, Vegetation on Rock Outcrops, Sub-alpine Grassy
Forests, and Riparian Forests. The Montane Tableland
Forests are neither grassy nor shrubby and fall into a
category between Wet and Dry Sclerophyll Forests. Coastal
and Tableland Grassy Forests are termed Grassy Woodlands
in Keith (2004). In the study area the Coastal and Tablelands
Grassy Forests have a forested rather than a woodland cover.
In places with higher levels of clearing and disturbance, the
current forested cover may have been reduced to a woodland
state. These Grassy Forests are distinctively different from
the Grassy Woodlands found further west on the tablelands
and south-west slopes.
Mapping of extant vegetation
To facilitate mapping, the study area was divided into three
subregions based on operational regions of State Forests
of NSW and modied somewhat to conform to bioregions.
The Commonwealth and State Governments adopted these
sub-regions as negotiation regions for the RFA. The sub-
regions were:
• South Coast sub-region, encompassing State Forests
of NSW’ hardwoods region, extending west to the Monaro
Highway and including southern part of Wingecarribee
and Kiama local government areas;
• Western sub-region, encompassing the State Forests of
NSW’ hardwood region in the Riverina and including the
ACT;
• Northern sub-region, covering the southern Central
Tablelands, and the northern half of the Southern
Tablelands.
The mapping method is based upon an expert knowledge
system of vegetation mapping that correlates the vegetation
groups, dened in the cluster analysis, with environmental
spatial data. The spatial distribution of each vegetation group
was related primarily to the CRA forest type layer (CRAFTI)
which was used as the primary mapping layer for mapping
existing vegetation. Other key environmental data to guide
the mapping were geology, elevation, aspect, and topographic
position. The Landsat Thematic Mapper (TM) layer was used
as a photographic image on which all the key environmental
layers and classied survey data were overlain. The patterns
of dry and wet sclerophyll forest were clearly visible and in
most cases matched the lines drawn by air photo-interpreters
to distinguish the broad vegetation types, such as rainforest,
wet sclerophyll, dry sclerophyll, heath forest, grassy forests,
heaths and swamps. It also became an important tool in
investigating spatial patterns of each vegetation group, as it
helped to differentiate between wet and dry sclerophyll sites
Table 3. Definitions of Diagnostic Species (after Keith & Bedward 1999) using frequency and cover abundance C/A
Other Vegetation Groups
Frequency ≥ 0.5 and Frequency < 0.5 or C/A < 2 Frequency = 0
cover abundance (C/A) > 2
Target Vegetation
Group Frequency ≥ 0.5 and Frequent Positive Diagnostic Positive Diagnostic
C/A > 2
Frequency < 0.5 or Negative diagnostic Less Informative Positive Diagnostic
C/A ≤ 2
Frequency = 0 Negative diagnostic Uninformative -
230 Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests
on opposing aspects, the more productive forests on valley
oors or along riparian zones, and any sharp transitions in
vegetation, usually associated with geology or soil type.
The mapping method depended on an accurate and reliable
air photo interpretation to produce a high quality vegetation
map. The CRA forest type layer was rst checked to ensure
that it was a satisfactory base layer for vegetation mapping
purposes using a two step process:
a) a preliminary check of air photo-interpretation of the
CRAFTI layer in the ofce. All available mapped and site
data was used to check polygon boundaries and API codes
with the CRAFTI layer, using expert local and regional
knowledge of the distribution of tree species;
b) eld checks of API polygons, using a hard copy map
version of the CRAFTI layer, as well as expert knowledge
of the distribution of tree species during travel between ora
surveys in the eld.
An Arcview project within Arcview GIS 8.3 was created to
enable mapping of vegetation against a backdrop of Landsat
TM imagery and other contextual spatial data. Requisite
spatial data was loaded into the view window. Data loaded
into this view included:
• classied site data taken from the PATN analysis;
• remotely sensed imagery;
• environmental spatial data; and
• other contextual layers such as roads and utility easements
and reserve boundaries.
The mapping sub-regions, South Coast, Western, and
Northern, were then progressively mapped in that order.
Small sections of a sub-region were mapped at a time to
make the mapping of vegetation groups consistent within
and between different sections of the sub-region. Each
vegetation group in that section of the map was mapped using
the sample sites allocated to that vegetation group in PATN,
and by applying extensive expert knowledge of the likely
spatial pattern of that vegetation group in the landscape. If
there was a clear relationship between the expected pattern
and the boundaries of CRAFTI polygons, then a numeric
vegetation code corresponding to the vegetation group was
placed in the vegetation code eld of the CRAFTI layer.
Generally there was a clear relationship between the expected
vegetation pattern and the CRAFTI mapping, and vegetation
codes were directly assigned to the relevant API polygons.
Some polygons required rening to adjust boundaries or to
split into ner polygons to match the expected vegetation
pattern. These changes were made in Arcview GIS and the
vegetation group codes reassigned as appropriate.
Using this highly repetitive process across 3 120 400 hectares,
three botanists, Phil Gilmour, Nic Gellie, and Michael
Doherty completed the vegetation map of the three mapping
sub-regions in three months (Table 4). The mapping of extant
vegetation by this method proved to be the only efcient,
consistent, and repeatable method within the time constraints
imposed by the Regional Forest Agreement (RFA) process.
The three maps were amalgamated to create a uniform and
consistent vegetation map across the whole Region. The
most recent version of the extant vegetation map has recent
national parks and nature reserves mapping overlain on the
region-wide vegetation maps. Recent mapping includes the
vegetation maps of new RFA reserves in the South-West
Slopes, South Coast, and Far South Coast Regions of the
Department of Environment and Conservation (EcoGIS,
2002, 2004a, 2004b, 2004c, Graham-Higgs Pty Ltd (2002a,
2002b, 2002c, 2004, and 2005).
Models of vegetation
Modelling of the distribution of vegetation groups was
undertaken simultaneously with the mapping of extant
vegetation. The modelling was seen as a supportive process
to the expert vegetation mapping of extant and pre-1750
vegetation and was used to test some of the assumptions
held by the mapping experts about the distributions of
vegetation groups within a regional landscape setting.
The use of Generalised Additive Modelling (GAMS) was
consistent with methods devised by Austin & Belbin (1982)
and Yee & Mitchell (1991). A modelling software program,
based on S-PLUS and Arcview GIS, was adapted for this
purpose (Ferrier 2002). The modelling of vegetation groups’
distributions were based on the classied site samples
arising out of the classication phase of the project, as well
as the detailed environmental layers (Table 2). GAM models
were prepared for those vegetation groups that had been
signicantly cleared and that had sufcient plot data (at least
10 vegetation samples) to produce a valid model. Table 5
summarises which vegetation groups had GAM models run
in each sub-region of the study area. The number refers to the
map legend number of the vegetation groups.
Within this subset of vegetation groups, all the sites within
a given vegetation group were assigned to presence sites,
while the remaining sites in the other groups were assigned
to absence sites. With up to 3400 absence sites available,
the GAM models produced reasonable models when there
were at least 20–30 samples assigned to presence data, with a
reasonable distribution of samples assigned as absence data.
Some of the models were also useful in identifying possible
sites to reallocate to other vegetation groups.
The next stage of model production involved setting realistic
minimum thresholds for the probability distribution for
each model that would correspond to a realistic map of the
pre-1750 distribution of each vegetation group. The spatial
extent of a GAM model was usually constrained to the spatial
extent of the sample sites within each vegetation group.
Where there was an over-prediction in the spatial extent
of a model, the model’s minimum probability thresholds
were adjusted higher to ensure that the model produced a
Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests 231
tighter t. In some cases the models were constrained with a
geographic area mask, set by an expert botanist, to eliminate
spurious predictions outside the known or expected range of
vegetation groups.
Map validation
Occasional short trips were undertaken during mapping of a
sub-region to conrm the vegetation patterns in recognisable
problem areas. These eld checks helped to compare the
vegetation to the vegetation classication and evaluate
the accuracy of the CRAFTI or mapping of state forests.
Field trips also helped to resolve problematic areas such
as geographic limits of similar vegetation groups that are
separated by understorey components rather than canopy
species, and are not well differentiated in the API layer.
Additional full oristic surveys were also undertaken in
areas identied as problematic for vegetation mapping.
To assist the ofce validation of the mapping, outside experts
reviewed the extant vegetation map of the South Coast and
Western sub-regions, and feedback and recommendations
were incorporated into subsequent versions. Representatives
of State Forests and the National Parks and Wildlife Service
were involved in these reviews.
In the 2005 version of the extant vegetation map, more
recent vegetation surveys and mapping have been used to
validate the classication and the mapping. The mapping of
RFA reserves on the South Coast and in the western parts
of the study area have been used to validate sections of the
vegetation map and to conrm the validity of vegetation
groups in the vegetation classication. The Department of
Environment and Conservation provided reports prepared
by Nicholas Graham-Higgs Pty Ltd (2002a, 2002b, 2002c,
2004, 2005) and EcoGIS (2002, 2004a, 2004b, 2004c), which
were incorporated into the current version of the vegetation
map. Additional eld survey information was kindly made
available for the alpine areas (McDougall & Walsh in prep.)
and for the temperate and grassy woodlands (Rhewinkel
unpubl.).
Derivation of a Pre-1750 Vegetation Map
Two maps in GIS format were used to model the previous
extent of vegetation on cleared land: maps of acid sulphate
soils along the coastal strip, and soil landscapes maps
throughout the rest of the study area. The acid sulphate soils
layer (Murphy et al. (1998), based on detailed soil survey
and mapping of acid sulphate soils of the lowland coastal
areas of the South Coast, was essential for modelling the
extent of previously cleared vegetation groups.
A separate CRA project, Lithology and Soil Landscape
Mapping, produced a soil landscapes maps for the entire
study area (DLWC 2000). This soil landscapes layer was
a multi-attributed polygon coverage that used the CSIRO
(1997) classication of lithology as a basis for the mapping
of soil landscapes. Within cleared areas of each CRA sub-
region, the soil layers were assigned to vegetation groups,
based on the relationship between the broad soil landscapes
and assigned API polygons on the adjoining forested
area, and the overlay of GAMS. These relationships were
explored visually within ARCVIEW GIS to help determine
the possible distribution and extent of vegetation groups
within the soil landscapes. As soil landscapes were mapped
at a scale of 1: 100 000, some mapped landscapes were larger
than the corresponding API polygons, mapped more nely at
a scale of 1: 25 000. In these instances the detailed pattern
on the extant vegetation map could not be transferred, and
the most dominant vegetation group was mapped on the
adjoining cleared land.
The pre-1750 vegetation map was prepared using the
following GIS layers:
1. Extant Vegetation of the South Coast, Western, and
Northern sub-regions
2. Pre-1750 Temperate Grasslands
3. Acid Sulphate Soils of the South Coast (DLWC 1998)
4. Soil Landscapes (DLWC 2000)
Table 4. Mapping of extant vegetation across mapping regions. The South Coast Region includes Commonwealth lands adjoining
Jervis Bay and the Western Region includes the ACT.
Mapping Region Area of Sub-Region Area of Extant Mapping Experts Completion Time
hectares (ha) Vegetation (ha) (months)
South Coast 1 842 600 1 255 200 Phil Gilmour and Nic Gellie 2
Western 2 912 600 1 446 700 Phil Gilmour, Michael Doherty,
and Nic Gellie 2
Northern 1 436 800 418 400 Nic Gellie 1
Total Area 6 192 000 3 120 400
232 Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests
A sequential merging of the above vegetation layers,
converted to 25 m grids, was undertaken. The pre-1750
vegetation map was compiled in the order and presence
of grid cell values from each of these layers. The grid cell
values were taken from the extant vegetation layer rst, and
then from the next vegetation layer, the pre-1750 grasslands
layer, and then from the pre-1750 soils layers, until all the
grid cells in the study area had a value of a vegetation group.
The nal merged product was checked to see if there were
any missing grids or values in the nal map. The pre-1750
vegetation map was recompiled after removal of unassigned
polygons in the extant vegetation or pre-1750 soils maps.
Results
Audit of existing survey data
The data audit produced 9656 samples that could be potentially
used in classication and mapping in the study area. Of
these 2395 samples met the audit criteria for vegetation
classication using full oristics cover-abundance data, plus
1415 additional samples from the South-East Forests region
(Table 6). Tables 7, 8, 9, and 10 summarise the datasets.
Categories II, III, and IV were datasets from survey work
carried out in the last 25 years that failed to meet criteria for the
initial classication but were used for supplementary analyses
to test for additional groups and to support the preparation
and validation of the extant and pre-1750 vegetation maps.
Previous ora surveys have centred on selected state forests
and conservation reserves and catchment areas within the
Brindabellas, Tinderry, Tallaganda, Deua and Budawang
Ranges (Fig. 3). Smaller ora surveys have been carried out
in the ACT and Monaro grasslands, Eurobodalla coastline,
Jervis Bay and Beecroft Peninsular, in Kosciusko National
Park, and in the Crookwell and Gunning Shires.
Field surveys
Between 1997 and 1999, CRA ora surveys added 1216 full
oristic cover abundance survey samples, as well as another
359 canopy only presence-absence samples. The latter
dataset was used mainly in the validation of the API layer.
The surveys focussed principally on major gaps in survey
effort in state forests and national parks (Fig. 3). There were
also some eld surveys done on Crown Land to supplement
the other surveys. Private land was only surveyed with
agreement from private landholders
Between 2001 and 2005, 418 additional eld survey samples
have been added to the survey database (EcoGIS, 2002,
2004a, 2004b, 2004c). This additional data has been used to
validate the rst version of vegetation classication and maps
produced in 2000. Survey data from reports by Graham-
Higgs Pty Ltd (2002a, 2002b, 2002c, 2004, 2005) and the
P5MA mapping (Tindall et al., in prep.) are not included in
the current database.
The current survey database contains 5444 full oristic survey
samples with cover abundance data, of which about 4180
samples have been used in the more vegetation classication
in 2005. An additional 6675 samples were also catalogued
and checked that included either partial oristics data with
cover-abundance scores; or canopy oristics data; or full
oristics data with presence absence scoring values.
The eld surveys since the audit have added another
33% of full oristic cover abundance samples across
critical environmental gaps in the study area. While the
extra sampling has not covered all the geographical and
environmental gaps, it has provided sufcient samples for
a region-wide classication to be undertaken with some
Table 6. Audited sample sites available for classification and mapping.
Data Category Flora Survey Data Type Southern Forests South-East Forests Total
No of Samples No of Samples
I Full floristics cover abundance 2395 1415 3810
II Full floristics presence absence 612 589 1201
III Canopy only presence absence 4442 0 4442
IV Partial floristics cover abundance 0 203 203
Total 7449 2207 9656
Table 5. List of GAMS models (Thomas et al. 2000)
Southern Forests Vegetation Groups
CRA Sub-Region
South Coast 2, 5, 9, 15, 16, 21, 48, 57, 58, 59, 61, 62, 68,
73, 74, 75, 90, 109, 112, 113, 114, 146, 160,
161, 171, 174,
Western 76, 80, 82, 85, 97, 98, 103, 106, 107, 110, 111,
116, 117, 120, 146, 147, 153, 154, 157, 158,
160, 161, 162
Northern 15, 17, 73, 89, 90, 100, 109, 121, 113, 115, 121,
122, 125, 133, 134, 135, 116, 153, 161, 162
Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests 233
Table 7. Full Floristic Cover-abundance Datasets (Thomas et al. 2000)
Reference Location No of Samples Plot size
(ha)
Southern Region
Benson 1994 Monaro Grasslands 62 0.01
Binns 1997a Bago & Meragle SF 102 0.1
Binns 1997b Carabost & Woomargama SF 7 0.1
CSIRO 1999 Clyde Mountain and Escarpment 47 0.1
Doherty 1996 Tinderry NR 50 0.04
Doherty 1997 Mundoonen NR 17 0.04
Doherty 1998a Brindabella NP 130 0.04
Doherty 1998b Burrinjuck NR 33 0.04
Eurobodalla Shire Council & NPWS 1998 Eurobodalla Shire 30 0.04
Forward et. al. 1997 Kosciuszko NP - fire monitoring plots 49 0.04
Gilmour 1985 Deua National Park 127 0.1
Gilmour et al. 1987 Williamsdale and Mt Tennant locality in the Namadgi NP 113 0.04
Helman 1983 Clyde River and Mt Dromedary Rainforest Surveys 231 0.1
Helman 1983 South Coast Random Rainforest Survey 40 0.1
Helman et al. 1988 Upper Cotter catchment, ACT 135 0.04
Hibberd and Taws 1993 Gunning, Crookwell, Boorowa Travelling Stock Route Surveys 321 0.04
Ingwersen 1992 Namadgi NP 201 0.04
Jurskis et al. 1995 Queanbeyan/Badja EIS 112 0.1
Lockwood et al. 1997 Eurobodalla NP 55 0.04
NPA 1998 Benanderah SF 30 0.04
Sharpe 1991 ACT grasslands 30 0.01
Skelton and Adam 1994 Beecroft Peninsula 168 0.04
Steenbecke 1990 Middle Kowmung River, Kanangra-Boyd NP 150 0.04
Taws 1997 Booderee National Park 101 0.04
Togher 1996 Abercrombie National Park 32 0.04
Eden Region
SFNSW 1999a SFNSW - miscellaneous Eden 81 0.04
SFNSW 1999b SFNSW Mt Pericoe Flora Reserve 24 0.04
SFNSW 1999c SFNSW Mt Waalimma Flora Reserve 22 0.1
Dodson et al. 1988 Glenbog SF 30 Unknown
Fanning & Rice 1989 Bondi SF 50 0.1
Keith 1999 Wadbilliga - mallee heath survey 14 0.1
Keith & Bedward 1999 NPWS Eden CRA Survey 269 0.1
Keith & Bedward 1999 South-East Forests Combined Surveys since 1991 698 0.04
Keith & Bedward 1999 Eden CRA Validation Surveys 136 0.1
Williams 1997 NPWS Bermagui NR, Biamanga NP, Goura & Wallaga Lake NRs 91 0.04
condence. The addition of the new samples has helped to
validate some of the earlier mapping decisions made in the
absence of good coverage of eld survey samples, as well as
expert eld knowledge.
Vegetation classification
2273 samples from the audit were combined with the 1216
samples from the eld surveys, together with a sub-set
of 251 samples from the Eden dataset, selected within an
adjoining buffer zone within the South-East Forests Region.
It was decided to use a smaller number of samples from the
South-East Forests to remove any bias in the vegetation
classication from the much higher density of sampling in
that region. The combined number of Southern and South-
East Forest survey samples amounted to 3740 using samples
with all vascular plants and cover abundance measures. This
nal dataset included some datasets such as Helman (1983),
Gilmour et al. (1987), and Helman et al. (1988) which had
comparable methods of assessing vegetation cover to that
of the modied Braun-Blanquet method (Poore 1955). The
authors of the original surveys were consulted about making
consistent and repeatable changes to the cover-abundance
scores, to make the datasets consistent with other surveys
with Braun-Blanquet scoring systems.
Preliminary testing of levels of vegetation classication
used classication levels between 100 and 200 groups.
Homogeneity analysis (Bedward (1999) of 3740 oristic
samples grouped into ascending number of groups (Fig. 4)
was used to select a potential starting point in the vegetation
234 Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests
classication. The upper inection point in the curve (Fig. 4)
occurred between 150 and 200 groups. After reviewing the
classication between 150 and 200 groups, an intermediate
starting point of 170 groups was selected, with the intention of
adding further groups if eld knowledge and data supported
further splitting of some of the 170 groups.
A classication of 170 vegetation groups provided sufcient
detail to show the pattern of vegetation at the scale of forest
compartment, between 100 and 300 hectares in size, or
equivalent to a small water catchment of similar size. This
level of classication seemed to correspond to vegetation units
that could be mapped at a scale of 1: 25 000 and provided
sufcient detail to distinguish forest types on the basis of
overstorey and understorey species that could be readily
identiable in the eld. A 170 group classication appeared
to fulll the classication and mapping requirements outlined
in the Introduction section. The dendrogram of the 170 group
classication showed that samples from the adjoining South-
East Forests region fell into similar groups classied by Keith
and Bedward (1999). This meant that the vegetation groups
in the study area had equivalents in the South-East Forests
Region and provided inter-regional comparability in vegetation
classication.
In addition to the vegetation groups classied using vegetation
samples, a number of vegetation groups were derived from
expert knowledge of gaps in the sampling effort. These were
often in unsurveyed areas, in remote terrain, or could not be
sampled adequately because of extensive clearing of native
vegetation. New vegetation groups were created to correspond
with those derived in previous vegetation mapping work or
which corresponded with anomalies identied in either the
CRAFTI layer or on the remotely sensed images (Appendix
1b on the CD-ROM)
Table 8. Presence-Absence Survey Datasets (Thomas et al. 2000)
Reference Location No of Samples Plot Size
(ha)
Southern
Clarke 1989 NSW Coastal Vegetation Survey 72 20m
length
transects
Gunn et al. 1969 Queanbeyan Shoalhaven area 261 0.04
Ingwersen 1972 Black Mountain Nature Park 59 0.04
Ingwersen 1983; Ward & Ingwersen 1988 Tidbinbilla Nature Reserve 81 Presumed
0.04
Ingwersen et al. 1974 Mount Ainslie and Majura Nature Park 139 Presumed
0.04
Eden Region
Breckwoldt and Breckwoldt 1979 Eden/Southern Coastal Vegetation Survey in Goura,
Bermagui, & Bournda NRs, as well as Mimosa Rocks and
Wallaga Lake NPs 374 0.04
Outhred 1986 NPWS Eden Survey of Wadbilliga NP 215 Nested
Quadrats
Table 9. Canopy Floristics Survey Datasets (Thomas et al. 2000)
Reference Survey location No of Samples Plot Size
(ha)
Southern
CSIRO 1999 Canopy Surveys - South Eastern NSW 3921 0.1
NPWS 1996 Tumut Interim Assessment Project 521 0.1
Table 10. Partial Floristics Cover Abundance Survey Datasets (Thomas et al. 2000)
Reference Location No of Samples Plot Size
(ha)
Southern
Mills, 1996a, 1996b and 1999 Conjola, Cudmirrah, Illawarra, Jervis Bay, Kangaroo Valley,
Nowra, Red Rock NR, and Ulladulla localities 182 Presumed
0.04
Mills, 1999 Coastal Shoalhaven 21 Presumed
0.04
Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests 235
A lower point of inection can be found at the level of 30
groups (Figure 4) and seemed to correspond to a much
broader ecological classication, equivalent to a vegetation
class (Keith 2004). This intermediate classication fell
somewhere between a vegetation formation and a vegetation
group. The underlying heterogeneity in some of the original
30 vegetation classes required further differentiation into
separate vegetation groups to produce a classication of
48 vegetation classes. A hierarchical classication of 48
vegetation classes was created, based on a hierarchical
grouping of the 170 vegetation groups derived from the
PATN analysis (Appendix 1a on the CD-ROM). The rest of
the vegetation groups, with group numbers greater than 170,
appear in a separate table after the dendrogram (Appendix 1b
on the CD-ROM). A further 6 vegetation classes were created
from the expert derived vegetation groups, after matching
where possible to the existing 48 vegetation classes in the
classication dataset (Appendix 2 on the CD-ROM).
A complete list of the vegetation groups (Appendix 3 on the
CD-ROM) includes the vegetation group numeric code; the
vegetation group name; whether it was forest or non-forest;
its mode of derivation in terms of classication, such as
whether it was derived directly from the cluster analysis or
instead was expert derived.
As well as a full oristic classication, a canopy only dataset
of 9139 samples was created by merging the full oristics
cover abundance and presence-absence datasets with the
Fig. 3. Distribution of full floristic cover abundance survey samples
236 Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests
canopy only dataset (Table 11). This is less than the nal
number of 10238 samples available in 2005. This dataset,
once classied, was a signicant aid in mapping vegetation
where there were few full oristic survey samples available
in the more remote parts of the study area. Areas where these
proved valuable included the central parts of Kosciuzko
National Park, the lower footslopes of the South-East
Highlands bioregion, and in the middle Shoalhaven River
catchment.
Mapping
The map of extant vegetation covers 3 120 400 hectares of
forests and non-forests. 193 vegetation groups were mapped
on the extant vegetation map, and 199 on the pre-1750
vegetation map. 12 vegetation groups were not mapped
because they occurred in small patches or as narrow lineal
features and had insufcient oristic samples to map them
accurately (11, 39, 72, 84, 85, 142, 155, 182, 199, 202) or
they fell outside the study area (6, 52). A digital version
of the extant vegetation map can be found in the mapping
directory on the CD-ROM. It is produced in both Arcview
grid and shapele formats.
To map each of the ecosystems separately was considered
cumbersome. Vegetation classes, which reveal some of the
broader patterns and relationships of the vegetation class
classes to physiography, climate and soils at regional and
sub-regional scales, were mapped. Maps of vegetation
classes were grouped by formation according to Keith’s
(2004) statewide classication. Maps of each of the
vegetation formations have been prepared (Appendix 4 on
the CD-ROM), in accordance with the hierarchy laid out in
Appendices 1a and 1b.
Table 12 shows estimates of pre-1750 and extant areas of
vegetation formations, together with other statistics on the
vulnerability classes, the extent of remaining vegetation, as
well as areas of each vegetation groups within conservation
reserves. The most widespread and dominant vegetation
formation is Dry Grass/Shrub Forest, with 1 026 800 hectares
(~33% of the total extant area of vegetation). The next most
common formation is Dry Shrub Forests mainly situated
on the eastern side of the study area, within the South-
East Corner and Sydney Basin bioregions. It covers 481
400 hectares (~16%). This is closely followed by Montane
Tableland Forests with an area of 474 500 hectares (~15%),
which occupies the central part of the study area, at higher
elevations than the Dry Grass/Shrub Forest formation. The
Moist Eucalypt Forest formation occupies a slightly lesser
area of 445 800 hectares (~14%).
The vegetation formation of Heath Forests/Heathlands/
Mallee Low Forests has an area of 150 300 hectares (~5%).
The Sub-Alpine Forests/Woodlands formation covers an
area of 112 800 hectares (4%) similar in area to that of Ash
Eucalypt Forests, which covers an area of 110 700 hectares
(4%). In the Alpine Area bioregion, the Alpine/Sub-alpine
Complexes formation has a somewhat smaller area of nearly
80 400 hectares (~3%). Because of extensive clearing and
modication of the both woodland and grassland ground
cover, the once extensive vegetation formation of Grassy
Woodlands and Grasslands now occupy only 89 600 hectares
(~3%). It once occupied a possible area of 1 314 800 hectares
previously. Vegetation on rock outcrops amounts to 48 300
hectares (2%), scattered in small pockets through out its
range. Much smaller areas of the remaining formations,
including Wet Heaths & Rainforests, Sedgelands, Riparian
Forests, Freshwater Wetlands and Coastal Complexes, make
up the remainder of the area. Their areas vary between 10
000 and 30 000 hectares in specialised habitats of the study
area, ranging in areal extent between 0.3 and 1%.
Discussion
Classification and mapping
The vegetation mapping, with some extensive subsequent
revisions between 2000 and 2005, has produced a map with
193 native vegetation groups. Some of the smaller and ner
vegetation groups, such as linear riparian vegetation, have not
been mapped. In 2000 the primary focus was to map forested
ecosystems. In 2003 the scope of the study was broadened to
map to cover the full range of vegetation in the study area.
418 samples were added to the original 3740 ora survey
samples. The more recent classication in 2003 conrmed
some of the vegetation groups suggested by botanists in the
rst version of the vegetation map and conrmed most of the
original vegetation groups created in the rst stable version
of the vegetation classication in 1999. Detailed checking of
vegetation in the new RFA reserves has largely conrmed the
stability of the vegetation classication (Graham-Higgs Pty
Ltd (2002a, 2002b, 2002c, 2004, 2005) and EcoGIS (2002,
2004a, 2004b, 2004c)).
Fig. 4. Graph of Homogeneity Values, showing possible starting
points in the vegetation classification
Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests 237
184 mapped vegetation groups have remained stable in the
classication, while the remaining 8 vegetation groups,
together with some individual samples have remained
inherently unstable in the hierarchical classication
(e.g. 31, 59, 64, 66, 89, 97, 99, 100). These elastic and unstable
samples and vegetation groups either form new groups or mix
in with other samples of more stable vegetation groups. This
complexity and instability in a hierarchical classication is
an ever-present and complicating factor in a classication of
this size.
In the South-East Forests region, Keith & Bedward (2000)
identied 72 vegetation groups in an area of 550 000
hectares. By comparison the Southern Forests study area
here has 193 vegetation groups for an area of vegetation of
three million hectares (3 120 400 hectares). Gilmour (pers
comm) contended that the classication of vegetation groups
in the Southern Forests study area should have been ner than
that used in the nal vegetation map. A ner classication
however, might make the task of identifying ecosystems in
the eld more difcult, because there could be a smaller
range of indicator species to separate similar vegetation
groups.
The larger size of vegetation groups in the Southern Forests
study area may be related to the broader environmental
gradients operating over the atter topography and less
diverse soils, particularly on the tablelands and western
slopes. There is a higher number of vegetation groups found
on the more rugged and diverse topography of the Kosciuzko
and Namadgi Ranges. It is possible that the higher number
of groups in the Brindabella Ranges may be due to the
heterogeneity created by different ora surveys, using
slightly different survey methods. Doherty (pers comm) has
indicated that disturbance patterns and observer bias may
have contributed to some of the extra variation and detailed
ecosystems in the Brindabella area.
26 vegetation groups in the South-East Forests region (Keith
& Bedward 1999) were found to be directly comparable with
those in the Southern Forests study area ( 1, 2, 7, 13, 18, 22,
23, 28, 32, 33, 35, 47, 48, 49, 53, 54, 55, 61, 64, 134, 135,
138, 157, 164, 165, 166) (Appendix 3 on the CD-ROM).
Map accuracy
The API mapping layer was only checked during the Southern
CRA in small areas (Hunter 2000). The API mapping layer
often had inconsistent polygon API codes which reected the
varying interpretations made by air photo interpreters. These
inconsistencies meant that assigning vegetation group codes
to API polygons became quite complex and involved a great
deal of checking using all available site and mapping data. In
these instances the mapping expert had to use the classied
oristic data and local knowledge to guide the assignment
of API polygons to vegetation. Difculties in vegetation
mapping were also experienced in places where there was
an inadequate coverage of site data. In these situations
the mapping expert would refer to previous mapping of
vegetation and reports to infer possible relationships between
the API polygons and the classied site data. The canopy-
only site data proved to be invaluable in places where there
was no full oristic data.
The 2005 version of the vegetation map may still contain
errors in places where there has been inconsistent or erroneous
air photo interpretation, and where there is a scarcity of
eld survey data, and lack of local knowledge. One of the
main limiting factors was having well stratied oristic
samples across the full range of environmental gradients.
Though many gaps in the geographical coverage of survey
samples have been lled by ongoing survey, some areas still
remain undersampled. However the average sample density
across the study area has reached 0.06 samples per square
kilometre, with a range from 0.01 up 0.33 samples per square
kilometre (Fig. 5). The highest density of samples is found
in Namadgi National Park, Jervis Bay, Bago and Meragle
State Forests, and in the middle Kowmung catchment in
the Blue Mountains. The sampling density is just under one
third the average density of 0.18 in the South-East Forests
region (Keith and Bedward 1999), and 0.2 in the survey of
the Cumberland Plain of western Sydney (Tozer 2003). A
sampling density between 0.1 and 0.15 samples per square
kilometre in this study area would need a further 500 to 750
oristic samples. As most of the study area has less variable
landscape than those in the South-East Forests region, a
lower density of between 0.1 and 0.15 samples per square
kilometre may be acceptable.
Since the production of the vegetation map in 2000, there has
been ongoing survey, and verication of the API. Most of
the new data has come from validation surveys of new RFA
reserves in the Western and South Coast sub-regions. The
vegetation map has also been updated in problematic areas.
There has been a major update of the mapping of Grassy-
Shrubby Forests, Ash Forests, Heathlands, and Alpine &
Sub-alpine Complexes, indicated by the lighter grey areas
in Fig. 6. In total 330 000 hectares, representing about 10%
of the area has been remapped. Despite this more remapping
work, the extant vegetation map requires further checking
within some of the larger national parks, such as southern
Blue Mountains, Kosciuszko and Morton National Parks, as
well as on private and Crown lands through the northern part
of the study area.
Recommended areas for further checking and revision of the
map include:
• the northern sub-region in grasslands, grassy woodlands,
and dry forests in the central and upper parts of the
Abercrombie catchment and on the Crookwell and
Goulburn plateaus;
• the western sub-region in grassy woodlands and montane
grass/shrub forests, the northern part of the Cotter
catchment in the ACT, a vast tract of central and northern
Kosciuszko National Park, the western and eastern
238 Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests
escarpments of Kosciuszko National Park, the Snowy
River catchment and Delegate areas, and all through the
sub-alpine and alpine complex, particularly in the non-
forest areas, with particular focus on the grasslands, bogs,
alpine herbelds, and feldmarks; and nally,
• the eastern escarpment sub-region in central Morton
and Budawang National Parks, the grassy woodlands
and grasslands through the broad valleys in Numeralla
and middle Shoalhaven river catchments, and through the
Numeralla and Wadbilliga ranges. Some of the additional
sampling has already been completed in the northern part
of the study area under the auspices of the P5MA project
(Tindall et al., in prep).
Biogeographical and environmental relationships
The study area spans ve bioregions, the South-east
Highlands (SEH), Australian Alps (A), South-Western
Slopes (NSS), South-East Corner (SEC), and Sydney Basin
(SB). The bioregions reect the varying dominant inuences
of geological substrate and climate that apply at a continental
scale. At a regional scale the vegetation in the study area
largely conforms to these bioregions with a strong gradient
between warm and moist climates in the Kiama-Illawarra
escarpment area, to the cooler and drier tableland plateau, and
nally onto the moist and cool alpine areas. The vegetation
responds to these general environmental gradients, as well as
the underlying soil fertility.
Fig. 5. Map reliability - the mean number of samples per square kilometre, within circular radii of 20 kilometres, across the Southern
Forests study area
Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests 239
1. Rainforests
The rainforest formation is found mostly in the South-East
Corner bioregion comprises sub-tropical, warm-temperate,
and cool temperate elements (Map 1 on the CD-Rom). Along
the Southern Illawarra, Deua, and Budawang escarpments,
vegetation class 1a forms mist forests in narrow sheltered
gullies, on moderately fertile soils derived from Ordovician
and Permian siltstones, in moderately cool temperatures and
high rainfall. Gondwanan elements include plant species
Eucryphia moorei, Fieldia australis, Dicksonia antarctica,
and Pyrrosia rupestris. An outlier of cool temperate forest is
found in moist gullies in the Geehi catchment of Kosciuzko
NP, where the dominant tree is Atherosperma moschatum. In
the milder and warmer climes below the coastal escarpment
a warm-temperate/subtropical rainforest (vegetation class
1b) predominates on sites with moderately fertile soils, high
rainfall, and nds full expression around the Kiama and
Kangaroo Valley escarpments. Warm temperate rainforest
species, such as Acmena smithii, Ficus coronata, and
Claoxylon australe, are common throughout vegetation class
2b. Doryphora sassafras and Ceratopetalum apetalum, are
more common dominant tree species in the warm temperate
rainforests. In the sub-tropical/warm temperate rainforests,
more northern rainforest species, such as Dendrocnide
excelsa, Baloghia inophylla, and Livistona australis are
represented. In the northern part of SEC bioregion, warm
temperate rainforests tend to be found on the cooler sheltered
slopes and aspects, whereas the warm temperate sub-tropical
rainforests are found on the warmer and more exposed
slopes.
2. Moist Eucalypt Forests
Moist Eucalypt Forests, sometimes adjoining rainforests,
are found in four distinct bands, corresponding to areas with
high annual average rainfall between 1100 and 1700mm
and fertile soils, such as kraznozems and red earths. The
vegetation classes, 2a, 2b, and 2c, are mainly found below
the escarpments of the South-East Corner and the southern
Sydney Basin bioregions. Outlying pockets of vegetation
class 2a, Brown Barrel Moist Shrub Forest, are found east of
Canberra in the Brindabella ranges, and around Blackjack
Mountain in southern Kosciuszko, and along the Alpine
Way in the Upper Murray valley catchment, where high
rainfall and fertile deep soils coincide. Vegetation class 2a
is found along the high parts of the South Coast escarpment
between Deua and the northern Budawangs, with another
area around the Robertson and Macquarie Pass areas (Map
2 on the CD-ROM).
Vegetation class 2b, South Coast and Hinterland Layered
Shrub Forests, is more restricted to the sheltered slopes
and gullies of the South Coast hinterland in lower rainfall
areas on moderately fertile soils. Common tree species
in this vegetation class are Eucalyptus muelleriana,
E. cypellocarpa, and Angophora oribunda. The understorey
can be a diverse mixture of shrubs, ferns, grasses, and
forbs and may include in the shrub layer Acacia mabellae,
Synoum glandulosum, Acacia falciformis, Notelaea venosa,
Pittosporum undulatum, and Hibbertia dentata, as well as
vines Cissus hypoglauca, Tylophora barbata, Pandorea
pandorana, and Eustrephus latifolius. In the ground layer
ferns Pteridium esculentum and Doodia aspera may be
dominant, along with a scattered layer of grasses and herbs
such as Poa meionectes and Schelhammera undulata.
Vegetation class 2c, South Coast/Central Coast Hinterland
Moist Shrub/Fern Forests, occurs on deep soils on sheltered
slopes, derived from Permian mudstones below the escarpment
and along deep gullies closer to the Coast. It is probably
the most productive forest type in the study area, has as
dominant eucalypt species, Corymbia maculata, Eucalyptus
saligna/botryioides, and E. pilularis, over an understorey
of rainforest species, including Synoum glandulosum,
Elaeocarpus reticularis, Livistona australis, as well as ferns,
Calochlaena dubia and Blechnum cartilagineum.
Vegetation class 2d, Tableland Moist Fern/Herb-Grass
Forests, occurs on moderately fertile soils on elevated
mountain plateaus with rainfall between 1100 and
1300 mm rainfall through the South-East Highlands
bioregion, This tableland forest type does not quite t either
the typical wet or dry sclerophyll categories of Keith (2004),
having a Poa spp. tussock grasses, intermixed with moist
herbs, and scattered shrubs. The dominant trees are usually
Eucalyptus dalrympleana, E. robertsonii subsp. robertsonii,
with sub-dominant trees of E. viminalis.
3. Ash Eucalypt Forests
The Ash Eucalypt Forests occupy specic environmental
niches on the exposed eastern and western escarpments
between 900 and 1400 metres, usually on the lee side of
main ridges in better developed soils, such as kraznozems,
and deep red earths (Map 3 on the CD-ROM).
Vegetation class 3a, South Coast Escarpment White Ash
Shrub Forests, tends to be an ecotonal vegetation type
between vegetation class 2a Brown Barrel moist shrub
forest and vegetation class 7e, South Coast Escarpment
Peppermint-Silvertop Ash Forests. It has as its dominant
trees, Eucalyptus fraxinoides, Eucalyptus cypellocarpa,
Eucalyptus fastigata, and Eucalyptus sieberi. The understorey
is intermediate between wet and dry sclerophyll forest, with
elements of moist shrub and heath elements intermingling
in the understorey. Vegetation class 3b, Sub-Alpine Ash
Shrub Forest, has an overstorey usually dominated by
Eucalyptus delegatensis, with co-dominant tree species
Eucalyptus pauciora, and Eucalyptus dalrympleana.
It has a drier understorey, comprising sub-alpine shrubs,
such as Polyscias sambucifolia subsp. B, Daviesia latifolia,
Coprosma hirtella, Tasmannia xerophila, and Tasmannia
lanceolata. A more exposed, higher altitude variant of this
vegetation class is vegetation group 86, whereas vegetation
240 Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests
group 87 is found in more montane, sheltered slopes,
mainly in the central western parts of Kosciuszko National
Park. Vegetation classes 3a and 3b represent the typical
re sensitive montane tall forests, which can succumb to
stand replacing res at intervals as short as 40 and 60 years.
Because of the local site conditions, Ash Eucalypt Forests
are fast growing and highly productive in their respective
montane and sub-alpine environments.
4. Montane Tableland Forests
A widespread formation through the higher parts of the South-
East Highlands bioregion is the Montane Tableland Forests,
which comprise 3 main vegetation classes, 4a, 4b, and 4c
(Map 4 on the CD-ROM). Its vegetation classes tend to be
found on less well developed soils in areas of slightly less
rainfall than either vegetation formations 2 or 3. These forests
are commonly associated with yellow podzolics and red
earths on granites and metamorphic rocks. Vegetation class 4a
corresponds to the more productive vegetation class Montane
Narrow Leaved Peppermint Forests, which occupies sheltered
slopes between 700 and 900 metres on the atter plateau
around Tumut, Black Andrew, and Woomargama to the west
and north of the Kosciuszko ranges, and Blackjack Mountain
in southern Kosciuszko National Park. Typical forest tree
dominants include Eucalyptus dalrympleana, Eucalyptus
robertsonii subsp. robertsonii, Eucalyptus macrorryncha,
and Eucalyptus bridgesiana. The understorey comprises
Platylobium formosum subsp. formosum, Daviesia latifolia,
Poa sieberiana var. sieberiana, and Pteridium esculentum.
In the northern part of the South-East Highlands bioregion,
vegetation class 4b, Southern/Central Tablelands Montane
Shrub/Grass Forests, is a forest dominated by Eucalyptus
pauciora, Eucalyptus dalrympleana, and Eucalyptus dives.
It occupies the more productive red earths in the Crookwell
and Oberon plateau with moderate to high rainfall. The
understorey is not that dissimilar to vegetation class 4c.
Fig. 6. Extent of mapping and validation work since 2000.
Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests 241
Vegetation class 4c can be found in the higher parts of the range
of this formation, between 900 and 1300 metres. A complex
suite of vegetation groups occupy a range of sites from
exposed and rather infertile soils to deeper and moderately
fertile soils. The dominant eucalypts are Eucalyptus pauciora
and Eucalyptus dalrympleana subsp. dalrympleana. In drier
and lower elevations of this formation, Eucalyptus dives and
Eucalyptus rubida becomes more dominant tree species,
with Daviesia mimosoides, Persoonia chamaepeuce, and
Poa sieberiana subsp. sieberiana becoming more dominant
in the understorey. On moister sites the understorey changes
to a moderate cover of Poa spp. and Daviesia latifolia, with
a greater range of herbs, such as Stellaria pungens, Asperula
scoparia, Viola betonicifolia subsp. betonicifolia, and Acaena
novae-zelandiae.
5. Grass/Shrub Forests
The Grass/Shrub Forests occupy considerable areas of the
South-East Highlands bioregion on moderate to steep slopes
with red and yellow podzolic soils on sedimentary and acid
volcanic rocks. They occur as far south as the lower Snowy
near Byadbo, and skirt the eastern edge of Kosciuszko
and the Namadgi ranges. They occupy a large part of the
Numeralla ranges, as well large areas of the Crookwell and
Oberon plateaus. They are also found on the western and
north-western parts of the study area on steep or rolling hills
above valley oors. Refer to Map 5 on the CD-ROM.
Vegetation class 5a comprises dry grassy forests found on
granite batholiths and in rocky gorges of the South-East
Corner and Sydney Basin bioregions. Vegetation group 54
is found on undulating slopes on clay loams derived from
granite in the coastal valleys from Wallaga Lake to the
Illawarra district. Tree dominants are Eucalyptus globoidea,
Eucalyptus tereticornis, and Angophora oribunda, with a
sparse shrub understorey of Bursaria spinosa and Leucopogon
juniperinus, and a grassy understorey of Themeda australis,
Echinopogon ovatus, and Eragrostis leptostachya. The gorge
forests comprise vegetation groups 51 and 174, which are
found in the Araluen and Bungonia gorges. In the Araluen
Valley an open grassy forest of Eucalyptus melliodora and
Eucalyptus maidenii predominates while in the Bungonia
gorges, Eucalyptus tereticornis is found in the deeper
parts of the lower gorges while Eucalyptus moluccana and
Eucalyptus bosistoana are common eucalypts in the open
grassy forests at the top of the gorge.
A mosaic of vegetation classes 5b, 5c, 5d, and 5e found within
the South-Eastern Highlands bioregion. Vegetation class 5b
is found on red podzolic clay soils where average rainfall
ranges between 600 and 800 mm. Trees dominants include
Eucalyptus mannifera, Eucalyptus dives, Eucalyptus rossii,
and Eucalyptus macrorryncha, with a common ground cover
of Daviesia leptophylla, Joycea pallida, Poa sieberiana
subsp. sieberiana, and Gonocarpus tetragynus. Vegetation
class 5c is a western and northern variant of vegetation class
5b, occupying the transition between the western Southern
Tablelands and the South-West Slopes. Dominant tree species
in unit 5c are Eucalyptus macrorryncha, Eucalyptus nortonii,
and Eucalyptus polyanthemos, over an understorey of mixed
grasses and herbs, such as Danthonia spp, Elymus scaber,
Hydrocotyle laxiora, Daucus glochidiatus, Gonocarpus
tetragynus and Lomandra liformis subsp. coriacea.
Vegetation class 5d, Central Southern Tableland Dry Grass
Forests, contains a diverse range of vegetation groups which
occupy the valley oors and slopes on heavier clay podzolics
and yellow earths. The dominant eucalypts of this class are
Eucalyptus dives, Eucalyptus rubida, Eucalyptus bridgesiana,
Eucalyptus viminalis, and Eucalyptus pauciora. Common
species in the understorey include grasses Themeda australis,
Poa sieberiana, and Dichelachne spp, as well as low shrubs,
such as Bossiaea buxifolia and Pultenaea procumbens.
6. Grassy Woodlands and Grasslands
The Grassy Woodlands and Grasslands are found principally
in the atter lower valleys of the South-East Corner and
South-West Slopes bioregions in areas of rainfall between
600 and 800 mm (Map 6 on the CD-ROM).
Vegetation class 6a, Lower Snowy White Box Woodland, is
a distinct unit found in the lower Snowy gorges on shallow
soils on granitic and metamorphic rocks in rainshadow areas
below 650 mm annual average rainfall. The forest canopy
comprises Eucalyptus albens and Callitris glaucophylla,
with a sparse understorey of grasses Themeda australis
and Austrostipa spp., along with scattered shrubs of Acacia
deanei subsp. parvipinnula, Chrysocephalum spp. and
Lissanthe strigosa. Vegetation class 6b is found on the
western side of the Kosciusko ranges, on the extreme edge
of the study area in the South-West Slopes bioregion, on
moderately fertile clays on undulating terrain from Holbrook
up to Boorowa Shires. An open tree canopy of Eucalyptus
albens, Eucalyptus blakelyi, and Eucalyptus polyanthemos
subsp. polyanthemos, overtops a grassy ground layer of
Microlaena stipoides, Danthonia racemosa var. racemosa,
and Austrostipa scabra subsp. scabra.
Vegetation class 6c, Southern Tablelands Yellow Box/Apple
Box Grassy Woodland, is found in the central and central
northern parts of the South-East Highlands bioregion on red
and yellow podzolics on sedimentary rocks. A woodland
cover of trees Eucalyptus blakelyi, Eucalyptus melliodora,
and Eucalyptus bridgesiana overtop a grassy understorey
of Themeda australis, Danthonia racemosa var. racemosa,
Microlaena stipoides var. stipoides, and Austrostipa scabra
subsp. falcata.
Vegetation class 6d, Temperate Grasslands, have been
included in this vegetation class as they show close similarity
to vegetation class 6c in the vegetation classication. Within
this class, there are two distinct sub-regional groups, the
Monaro Grasslands comprising vegetation group 158 in the
Cooma-Monaro district, and vegetation groups 152, 153, and
157 in the Canberra and Yass localities. Common species
242 Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests
found in both sub-regional groups include Themeda australis,
Poa sieberiana subsp. sieberiana, Danthonia caespitosa,
Chrysocephalum apiculatum, Geranium solanderi var.
solanderi, and Calocephalus citreus. Vegetation class 6d
is commonly found on the Monaro plains and through the
Canberra region, on red podzolic soils derived from basalt or
sedimentary siltstones, where frosts are reasonably common,
and the average annual rainfall varies between 500 and
700mm.
7. Dry Shrub Forests
Dry Shrub Forests dominate in the eastern part of the study
area, principally in the South-East Corner and Sydney Basin
bioregions (Maps 7.1 and 7.2 on the CD-ROM). Despite the
relatively high rainfall these areas receive, the low fertility
and water holding capacity of the shallow soils limit
growth of more lush and taller plants. The Dry Sclerophyll
Forests formation is made of 9 different heterogeneous
vegetation classes that reect the diversity of overstorey
and understorey species and life forms, mainly reecting
the drier and more infertile soil types.
Vegetation class 7a, South Coast/Hinterland Dry Shrub
Forests, is primarily found on shallow yellow/orange infertile
podzolics on dry bare ridges at elevations between sea level
and 700 metres. The average rainfall varies between 850 and
1100 mm. Common tree species in this vegetation class are
Syncarpia glomulifera, Corymbia gummifera, Eucalyptus
agglomerata and Eucalyptus sieberi, overtopping a sparse
cover of shrubs Acacia obtusifolia, Oxylobium ilicifolium,
Leucopogon lanceolatus, and in the ground layer Platysace
lanceolatus, Entolasia stricta, Joycea pallida, and
Dianella caerulea. On the more undulating slopes and
ridges, closer to the South Coast, vegetation class 7b can
be found on yellow podzolics derived from metamorphic
sediments and Permian siltstones. Common tree species in
this vegetation class are Corymbia maculata, Eucalyptus
longifolia, Eucalyptus muelleriana, and Eucalyptus
paniculata subsp. paniculata Typical understorey species
include Allocasuarina littoralis, Oxylobium ilicifolium,
Platysace lanceolatus, Hardenbergia violacea, and grasses
Entolasia stricta, Poa meionectes, and Imperata cylindrica
var. major. Towards Jervis Bay and the lower reaches of
the Shoalhaven River, an interesting and diverse vegetation
class 7c, South Coast Mixed Species Dry Shrub Forests,
can be found on yellow earths and podzolics on clay
rich sedimentary rocks. The overstorey commonly has
Eucalyptus punctata, Eucalyptus eugenioides, Eucalyptus
pilularis, and Corymbia maculata, while the understorey
has the tall shrub Allocasuarina littoralis as a common
feature, and common small herbs and grasses, including
Brunoniella pumilio, Entolasia stricta, Lagenifera gracilis,
and Gahnia radula.
Vegetation class 7d, Coastal Bangalay /Blackbutt Dry
Shrub Forests, is found on pockets on coastal dune deposits
around Jervis Bay and Moruya on sandy podzolic soils.
The forest overstorey comprises Eucalyptus botryoides,
Eucalyptus pilularis, and occasionally Corymbia gummifera
in the Jervis Bay area. The understorey has tall heath shrubs
Banksia serrata, Banksia integrifolia, Monotoca elliptica,
and Acacia longifolia subsp. sophorae closer to the beach.
The ground layer is usually a bracken/grass understorey,
comprising Pteridium esculentum, Lomandra longifolia,
and Imperata cylindrica var. major, as well as Lepidosperma
laterale and Gonocarpus teucrioides.
Along the boundary between the coastal and the South-Eastern
Highlands bioregions, vegetation class 7e, South Coast
Escarpment Peppermint-Silvertop Ash Forests, occurs on
high ridges and plateaus, on red podzolic soils on Ordovician
ne sediments, usually in areas of moderately high average
rainfall between 900 and 1100m. The dominant trees are
Eucalyptus dalrympleana subsp. dalrympleana, Eucalyptus
radiata subsp. radiata, and Eucalyptus sieberi, usually with
a bracken understorey comprising Pteridium esculentum,
and other heath and grass-like species such as Oxylobium
ellipticum, Leucopogon lanceolatus subsp. lanceolatus,
Dianella tasmanica, and Lomandra obliqua. Within the
Kanangra and Coxs River Gorge country, vegetation class
7f occupies steep ridges and slopes on Devonian sediments
below vegetation class 7e, in areas of average annual
rainfall between 800 and 950 mm. Common tree species
are Eucalyptus punctata, Eucalyptus blaxlandii, Eucalyptus
agglomerata, Eucalyptus tereticornis, and Eucalyptus
melliodora. This dry forest type has scattered shrubs, such
as Olearia viscidula, Persoonia linearis, Stypandra glauca,
and herbs Pratia purpurascens and Gonocarpus tetragynus.
Further to the south, vegetation class 7g represents a
complex mixture of forests dominated by Eucalyptus
sieberi, Eucalyptus agglomerata, and Eucalyptus dives.
This type occurs on yellow podzolic soils on undulating
topography on low fertility sediments in areas with average
annual rainfall between 700 and 900 mm. The understorey
is usually an open heath understorey with common species
such as Allocasuarina littoralis, Lomatia silaifolia, Dianella
revoluta subsp. revoluta, and Entolasia stricta.
Outliers of the dry shrubby forest formation occur further
to the west on the western escarpment of Kosciuzko. These
include vegetation class 7h, Western Southern Tablelands
Dry Shrub Forests, which is found on the steep rocky slopes
above the Tumut River. Vegetation class 7i, South-West
Slopes Red Ironbark Dry Shrub/Grass Forests, is found
further to the west on raised foothills of the South-Western
Slopes bioregion. These two vegetation classes occur in
small pockets on shallow red podzolic soils in very specic
environmental niches. Descriptions of these types can be
found on the enclosed CD-ROM. These two vegetation
classes represent dry sclerophyll forests that could be
classied as heath forests, as the understorey can be quite
dense in places.
Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests 243
8. Heath Forests, Mallee Low Forests, and Heathlands
Several structural formations are included on the one
vegetation class as they are share a dense heath understorey
either in the open or underneath a tree canopy. Heathy
forests, heathlands and mallee low forests are located on
infertile sediments derived from Permian sandstones or
granites, mainly in the South-East Corner and Sydney Basin
bioregions (Map 8 on the CD-ROM).
Vegetation class 8a, Sandstone Plateau Heath Forests,
contains ve closely related vegetation groups that are
found on relatively poor yellow or red podzols on Permian
sandstones on the Morton plateau or on Ordovician
sediments in the Clyde River catchment in areas of annual
average rainfall between 1000 and 1200 metres. Overstorey
dominants include Eucalyptus sclerophylla and Corymbia
gummifera, with Eucalyptus sieberi becoming dominant
along the escarpment and western Morton plateau. Common
shrubs include Banksia ericifolia, Banksia paludosa, Hakea
teretifolia, Leptospermum trinervium, Petrophile sessilis,
Aotus ericoides, and Allocasuarina distyla, along with
smaller forbs and grasses, such as Epacris microphylla subsp.
microphylla, Actinotis minor, Entolasia stricta, Leptocarpus
tenax and Lomandra glauca.
The next vegetation class 8b, South Coast/Hinterland
Heath/Shrublands, occurs either primarily along the
coast on infertile wet shallow clay podzolic soils within
reasonably close proximity to the coast or higher up on the
Cambewarra and Kiama escarpments on organosols with
average annual rainfalls between 1400 and 1500 mm. This
class of wet heaths has Banksia ericifolia, Hakea teretifolia,
Leptospermum squarrosum, Leptospermum juniperinum,
Epacris obusifolia, Epacris microphylla subsp. microphylla,
Darwinia leptantha, and Dillwynia oribunda as common
heath plant species, with a diverse array of sedges and ferns
present in the ground cover layer. The ground cover layer
includes sedges adapted to wet impeded soils, such as Restio
fastigiatus, Lepidosperma liformis, Leptocarpus tenax,
Xanthosia resinosa, Gleichenia dicarpa, Lepidosperma
forsythii, and Gymnoschoenus sphaerocephalus.
An extensive area of this vegetation class is found on
shallow yellow podzolic soils, in areas of average yearly
rainfall between 950 and 1200mm on Permian mudstone
through Morton and Budderoo National Parks. It bears close
oristics resemblance to 8a, Sandstone Plateau Heath Forests.
Overstorey species may include Eucalyptus sclerophylla,
Eucalyptus tenella, Eucalyptus sturgessiana, Eucalyptus
langleyii, and Coyrmbia gummifera. The heathy understorey
may include Allocasuarina distyla, Leptospermum lanigerum,
Kunzea capitata, Hakea teretifolia, and Banksia ericifolia.
The ground cover may be a mixture of forbs, sedges and
grasses, such as Lepidosperma urophorum, Entolasia stricta,
Lepyrodia scariosa, and Schoenus ericatorum.
Vegetation class 8c, High Plateau Mallee Low Open Forests,
is found on shallow infertile poorly drained organosols
on the Kanangra Boyd Plateau and Murrun stock route in
well dened mallee patches on Permian sandstones, with
an average annual rainfall between 800 and 1000 metres.
There is usually a moderately dense layer of Allocasuarina
nana, alongside mallee clumps of Eucalyptus stricta, with
a dense cover of Hakea dactyloides, Banksia ericifolia,
Banksia marginata, Leptospermum trinervium, Isopogon
anemonifolius, and Playsace linearifolius. The ground cover
comprises Carex appressa, Ptilothrix deusta, Patersonia
fragilis, and Lindsaea linearis.
A closely related vegetation class 8d, Eastern Montane
Heath/Tall Shrubland, occurs along the western spine of the
South Coast escarpment from Nerriga down to Wadbilliga
National Park. The soils are usually lithosols, derived from
hard quartzite or shale rocks, in areas with average annual
rainfall between 700 and 950 mm. The heaths are usually
found on open windswept ridges, in moderately cool climatic
conditions. This vegetation class contains two vegetation
groups, 134 and 135, which have a number of plant species
in common. In the heath shrub layer Allocasuarina nana is
dominant, with Brachyloma daphnoides, Banksia canescens,
Hakea dactyloides, Hibbertia pedunculata, and Kunzea sp
C, while the ground cover layer may have Lomandra glauca,
Stylidium graminfolium, Gonocarpus tetragynus, Austrostipa
pubinodes, Lepidosperma gunnii, and Amperea xiphoclada
var. xiphoclada.
9. Swamp Forests and Sedgelands
This vegetation class comprises the swamp and swamp forests
found on soils with impeded drainage, usually associated with
valley bottoms and creek lines. These vegetation formations
are conned to narrow wet areas usually in moderate to high
rainfall areas at high elevations, on organosols which collect
water from adjoining areas (Map 9 on the CD-ROM).
Vegetation class 9a, South Coast/Hinterland Swamp Forests,
is found on infertile organosols overlying colluvial sandy
substrates. The dominant trees are usually Syncarpia
gummifera, Eucalyptus robusta, Eucalyptus tereticornis,
Eucalyptus longifolia or Corymbia gummifera, with a tall
shrub layer of Melaleuca linearifolia or Melaleuca biconvexa,
Melaleuca squarrosa or Leptospermum polygalifolium subsp.
polygalifolium. The ground layer may have either Gahnia
spp. or Gleichenia dicarpa as dominant ground cover, as well
as a range of sedges and forbs, depending on site conditions.
This vegetation class contains a heterogeneous group of
vegetation that has adapted to particular site conditions in the
Jervis Bay area, usually related to different type of colluvium
that has lled the creeklines or valley oors.
Fringing the eastern Southern and Central Tablelands
escarpments is another heterogeneous swamp heath/low
244 Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests
forest group, vegetation class 9b, which occurs on sandy
organosols in shallow river ats in the middle and upper
Shoalhaven catchment. This is called Eastern Tablelands
Swamp Heath/Low Forests. The overstorey tree dominants
are usually sparse and may comprise Eucalyptus pauciora,
Eucalyptus viminalis, Eucalyptus rubida subsp. rubida or
Eucalyptus aggregata, with Leptospermum myrtifolium,
Leptospermum juniperinum, Epacris microphyllus subsp.
microphyllus, as possible heath shrubs. The ground layer
usually has a range of grasses and herbs, such as Poa
sieberiana subsp. sieberiana, Microlaena stipoides var.
stipoides, and Gonocarpus micranthus subsp. micranthus,
and sedges Schoenus apogon, Lepyrodia anarthria, and
Selaginella uliginosa subsp. uliginosa.
A widespread vegetation class in the South-East Highlands
bioregion is vegetation class 9c, Southern Tableland Montane
Wet Heaths / Swamps. This vegetation class is found in the
Upper Shoalhaven and Queanbeyan River catchments on
either side of the Tallaganda ranges and further to the west
in the higher montane climates of northern Kosciuzko and
the ACT. The vegetation groups 123, 124, 125, and 126,
are conned to creek or river ats, with deep organosols
in moderate to high rainfall areas between 750 and
1200 mm, occupying a diverse range of montane and lower
sub-alpine environments. Typical heath shrubs include
Baeckea utilis, Epacris paludosa, Epacris brevifolius,
Leptospermum myrtifolium and Hakea micrantha. The
ground cover layer may have either a dense cover of Poa
labillardieri or Carex gaudichaudiana, with smaller sedges
such as Restio australis and Empodismus minus. The inter-
tussock space may be lled with forbs, such as Euchiton
gymnocephalus, Hydrocotyle peduncularis, Hypericum
japonicum, or Oreomyrris eriopoda. A closely related unit,
vegetation class 9d, Southern Tablelands Swamp Grasslands,
is found on soils in similar climatic zones of the South-East
Highlands bioregion in the ACT and further to the north in
the Oberon area. It tends to have a more dominant grassy/
sedge layer of Poa labillardierii and Carex appressa, with
herbs such as Acaena novazelandiae, as well as small sedges
and rushes Carex inversa and Juncus licaulis.
The nal vegetation class in this heterogeneous formation
is vegetation class 9e, Southern Tablelands Swamps /Open
Woodlands, which is found mainly in central and northern
Kosciuszko ranges and along narrow streams through
the Tumut, Tumbarumba and Woomargama districts. The
soils are freer draining and are typically found in frost
hollows in areas with average annual rainfall between 800
and 1200mm. The overstorey typically has Eucalyptus
pauciora, Eucalyptus stellulata and Eucalyptus rubida,
in some of the low lying ats in the higher parts of the
Kosciuzko, ACT, and Tallaganda ranges. The grass and
sedge ground cover usually contains Carex appressa and
a range of small grasses and forbs, including Hydrocotyle
laxiora, Hydrocotyle peduncularis, Dichondra repens and
Mentha diemenica. In some cases this vegetation class is
likely to have a more diverse heath and grassy ground layer,
which includes heath species Leptospermum myrtifolium,
Hakea micrantha, Bossiaea foliosa, Acrotriche serrulata,
Mirbelia oxyloboides, Grevillea lanigera, as well as
Themeda australis and Poa spp., and small forbs Acaena
novazelandiae, Viola betonicifolia, Dichondra repens, and
Asperula scoparia.
10. Vegetation on Rock Outcrops/Screes
This vegetation formation covers shrublands, woodlands,
and forests on rocky outcrops and scree slopes across a
wide range of climates and soil substrates in the South-East
Corner, Sydney Basin, and South-West Slopes bioregions
(Map 10 on the CD-ROM).
Vegetation class 10a, Sub-Alpine/Montane Rocky Heath
Complex, is conned to the rocky tors and western escarpment
of the Alps and has a patchy occurrence through the Scabby
and Bimberi ranges of the ACT. Its dominants usually include
Kunzea muelleri, Leptospermum micromyrtus, Leptospermum
namadgiensis, Kunzea ericoides, Phebalium squamulosum
subsp. ozothamnoides, and Oxylobium alpestris.
The second vegetation class 10b, South Coast/Southern
Tableland Hinterland Rocky Shrublands/Forests, covers a
discrete set of vegetation groups in the classication that
correspond to shrublands and rocky forests on Silurian
granites in the upper Snowy River to steep rocky screes
on Ordovician sediments on the mid slopes of the South
Coast escarpment. In the Snowy River valley average
annual rainfall varies between 600 and 700 mm per annum
while along the South Coast it can be between 900 and
1000 mm per annum. There is a range of tree species adapted
to these rocky exposed conditions, including Eucalyptus
smithii, Eucalyptus olsenii, Eucalyptus bauerlenii,
Eucalyptus blaxlandi, and Eucalyptus fraxinoides. Either
under canopy under moderately sheltered sites or on more
exposed slopes Acacia sylvestris, Eriostemon trachyphyllus,
Phebalium coxii, may be present. An unusual group within
vegetation class 10b, can be found on rhyolite outcrops in
Deau and southern Budawang National Park. Eucalyptus
stenostoma dominates the tree canopy while the shrub layer
usually comprises Allocasuarina littoralis, Leptospermum
trinervium, Phebalium coxii, Boronia ledifolia, Leucopogon
setiger with a ground cover of Patersonia spp. The ground
cover can be quite sparse on these rocky screes.
Further to the south in the Snowy River Valley, Acacia
sylvestris shrubland form may overlap with vegetation class
10c, Lower Snowy Acacia Shrubland / Cypress Pine Forest,
which has overstorey dominants of Acacia doratoxylon,
Callitris endlicheri, with a similar shrub layer of Eriostemon
trachyphyllus.
On the extreme south-western edge of the study area in the
South-West Slopes bioregion, vegetation class 10d, South-
West Slopes Acacia Shrublands, can be found on tops of
sandstone ridges, in areas of average annual rainfall between
600 and 700 mm per annum. The shrubland usually has a
Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests 245
moderately dense tall shrub layer of Acacia doratoxylon,
along with Callitris endlicheri, Allocasuarina verticillata,
Eucalyptus blakelyi x Eucalyptus dwyeri, with a simplied
ground cover of Stypandra glauca, Gonocarpus elatius,
Crassula sieberiana subsp. sieberiana, and Triptilodiscus
pygmaeus.
Vegetation class 10e, South-West Slopes Black Cypress Pine
Forests. It is found on granitic outcrops along sections of
the Murray valley between Corryong and Holbrook, and up
into the lower Abercrombie river valley, north of Boorowa.
Annual average rainfall varies between 600 and 700 mm
per annum, which is quickly shed on the bare surfaces of
these rocky outcrops. Soils are usually skeletal. Common
overstorey dominants are Eucalyptus goniocalyx, Eucalyptus
macrorryncha, and Callitris endlicheri. The understorey is
usually made up of Xanthorrhoea glauca var. angustissima,
Acacia verniciua, Brachyloma daphnoides, Dodonaea
viscosa subsp. angustissima, Stypandra glauca, Gonocarpus
elatius, and Lomandra liformis spp coriacea.
The last class in this vegetation formation is 10f, Ribbon Gum
Forests on Limestone, which is found on limestone outcrops
in the river catchments of the Jenolan, Kowmung, lower
Goodradigbee, Deau, and Yarragobilly Rivers. Unfortunately
it has not been sampled. Typically it has an overstorey of
Eucalyptus viminalis, with a shrub layer of Bursaria spinosa
ssp. spinosa and Senecio spp., and a matt of herbs and
forbs, comprising Dichondra repens, Hydrocotyle laxiora,
Pellaea falcata var. falcata, in amongst moist grasses, such
as Microlaena stipoides var stipoides.
11. Riparian Forests
The riparian forests mapped possibly represent a fraction of
the actual and previous extent of the riparian forests in the
study area. There are two principal vegetation classs shown
in Map 11 on the CD-ROM, which are associated with wide
rivers or streams in the study area. The rst vegetation class
11a, River Oak Riparian Grass/Herb Forest, is associated
with fast moving perennial streams (Keith 2004), and usually
occurs on third or fourth order streams throughout the study
area. The dominant tree is Casuarina cunninghamiana. The
understorey and ground cover is typical of other moist riparian
or valley bottom moist forests/woodlands (vegetation groups
48, 92, or 162) in the study area.
The other vegetation class 11b, South-West Slopes Grass/
Sedge Forests, is associated with streams in the South-West
Slopes bioregion and comprises two vegetation groups 43
and 162. The soils in these two vegetation groups are usually
derived from colluvium or alluvium and are relatively deep
and moist for most of the year. Typical overstorey dominants
include Eucalyptus camaldulensis on major rivers and
streams and Eucalyptus blakelyi further upstream on the
minor streams and atter valley oors with perennially moist
soils. The understorey usually features Carex appressa, Carex
inversa, Eleocharis spp, Pratia peduncularis, and a range of
forbs, grasses, and rushes adapted to temporary inundation
on oodplains and on shallow riparian creeklines.
12. Sub-alpine Low Forests/ Woodlands
Sub-alpine low forests/woodlands are conned to the higher
elevations of the New South Wales Alps between 1500 and
1650 metres (Map 12 on the CD-ROM). The formation of
sub-alpine low forests comprises two vegetation classes.
A northern rocky heathy type found on the Scabby and
Gudgenby Ranges on granitic substrates is conned to a
narrow environmental niche in the Southern ACT ranges.
This is categorised as vegetation class 12a, Southern
ACT Sub-alpine Snow Gum Heath Forest. The southern
vegetation class 12b, Kosciuzko Sub-alpine Low Forest, is
more widespread from south of Thredbo up into the ranges
surrounding Nungar Plain. Vegetation class 12b is associated
with organosols derived from a diverse range of parent
materials, such as sandstones, cherts, granites, granodiorites,
and acid volcanics. In the range of both vegetation classes
average annual rainfall varies between 1300 and 2000 mm
per annum.
These low forests are usually dominated by Eucalyptus
debeuzevillei in vegetation class 12a and Eucalyptus
niphophila and Eucalyptus pauciora, in vegetation class
12b. The understorey in vegetation class 12a tends to have
a denser cover of heath shrubs, including Kunzea muelleri,
Leptospermum micranthus, Leptospermum namadgiensis,
Oxylobium alpestre, Kunzea ericoides, and Westringia
lucida. Vegetation class 12b on the other hand has a greater
mixture of alpine heaths and grasses. The shrub layer is
somewhat different to that in vegetation class 12a and
usually comprises Bossiaea foliosa, Olearia phlogopappa,
Ozothamnus secundiorus, Leucopogon montanus,
Leucopogon hookeri, Hovea montana and Tasmannia
xerophila. The dominant grass and herb plants species in the
ground layer may comprise Poa hiemata, Poa ensiformis,
Oreomyrrhis eriopoda and Stellaria pungens.
13. Alpine and Sub-alpine Complexes
Alpine and sub-alpine complexes largely fall within the
bounds of the Alpine Area bioregion (Map 13 on the CD-
ROM). High rainfall, low fertility soils, and cold to cool
temperatures, with a high frequency of frosts in winter lead to
a relatively short summer growth period. There are currently
an inadequate number of vegetation samples to describe
adequately this highly heterogeneous sub-alpine and alpine
vegetation. The work of McDougall and Walsh (in prep.)
may help to unravel the complexity of the vegetation in this
vegetation formation.
Four separate vegetation classes have been identied
and mapped as complexes. Vegetation class 13a, Alpine
Feldmarks, is conned to the bare snow patches and bare
246 Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests
rocky ground at elevations above 1700 mm. Below the
Alpine Feldmarks two vegetation classes 13b, Alpine/Sub-
alpine Herbelds, and 13c, Alpine Bogs/Fens, dominate the
more open and less well drained slopes of the Alps. In the
wetter less well drained areas, vegetation class 13c, Alpine
Bogs/Fens predominate. Vegetation class 13d, Sub-alpine
Grasslands, is more common in the northern and eastern
parts of the Alps at slightly lower elevations on grassy plains,
where it can be subjected to intense frosts in winter.
14. Coastal Complexes
The coastal complex vegetation is conned to the immediate
surrounds of estuaries, sand-dunes, and low lying areas
adjoining estuarine lakes along the South Coast. Because
of the patch size and linear shape of the vegetation groups
present, it is difcult to present the information at an adequate
scale to differentiate the separate vegetation classes (Map 14
on the CD-ROM).
Map unit 14a, Coastal Dune Complex comprises ve separate
vegetation groups, which occupy distinct zones in the fore,
mid, and hind-dunes. The zonation in the dunes reects the
transition from an undeveloped sandy soil in a foredune
through to a well developed sandy podzol in the hind-dune.
The type of plant species present in each of the distinct
dunal zones reect the soil differentiation and salt loading
coming in off the sea. Details of the oristics composition of
each of these vegetation groups can be found in the detailed
description of each vegetation group on the accompanying
CD-ROM.
In the low lying areas above the tidal zone, vegetation class
14b, Coastal Swamp Heath/Forest Complex, is found on acid
sulphate soils in semi-inundated areas. Usually these areas
adjoin wetlands, dominated by species such as Phragmites
communis, and areas of open water in brackish areas of
freshwater estuaries. Around the edges of saline estuaries, a
complex mosaic of salt marshes and mangroves interweave
in intricate patterns on mudats, reecting the varying impact
of waterlogging and salt concentration in the soils (Keith
2004). This complex of inter-tidal estuarine vegetation is
mapped as vegetation class 14c.
15. FreshwaterWetlands
There are three vegetation classes of freshwater wetlands in
the study area. Vegetation class 15a comprises the coastal
freshwater lagoons and sand dune wetlands. Vegetation
Class 15b comprises the montane lakes, such as found in
the lakebeds of Lake George and Lake Bathurst. Vegetation
class 15c is found in and adjoining billabongs in amongst
vegetation group 43 on along the Murrumbidgee and Murray
Rivers. Since none of these vegetation classes were sampled,
there are no detailed vegetation descriptions. However the
vegetation classes described here correspond to those in
Keith (2004) where details of the characteristic species can
be found.
Disturbance Regimes
Typical disturbance agents include re, clearing, grazing,
logging, drought, and predation by insects or diseases. These
disturbance agents vary in their frequency, season, timing in
relation to drought or rain, spatial extent, intensity, and likely
combinations within each vegetation formation. These are
referred to here as disturbance regimes and are summarised
in Table 11 which is intended to be a guide to the general
trends in disturbance regimes since European settlement.
The introduction of feral and domestic animals and exotic
plants, combined with changed re regimes or timber
getting, have signicantly altered the historical pattern of
natural disturbance regimes of the grassy forested and non-
forested ecosystems, with consequent changes in species
diversity and structure of these grassy forests. Some of
the Dry Shrub Forest and Moist Forests formations have
probably had much less change to their historical pattern of
disturbance processes. In the late 1800s, clearing associated
with free selection on Crown Land, and timber getting in the
early part of the 20th century converted extentive areas of old
growth forest to pasture, potato farms, or regrowth forest.
Early accounts referred to eucalypts of gigantic proportions
in the Yarrawa Brush area near Robertson (Jervis 1962).
Dendrochronological studies (Banks 1982, Pulsford 1989)
have suggested an increase in re frequency in the sub-alpine
forests and lower Snowy forests, associated with European
use from 1870 to 1950. Frequent spring or autumn burning,
associated with maintenance of ‘green pick’ for introduced
stock, became an annual practice. This was in marked
contrast to infrequent res prior to European settlement.
Research by Costin (1966) produced evidence of the negative
impacts of frequent grazing and burning on water yield and
sediment production. When grazing was excluded from the
sub-alpine and alpine zones, there was a sudden reversal in
the re frequency on the alpine and sub-alpine vegetation in
the Snowy Mountains. However from 1960–1980 the drier
Montane Tableland Forests and the Dry Grass/Shrub Forests
on the western footslopes of the ranges became the focus of
more frequent intensive burning as a re mitigation strategy.
Since then, the NPWS have adopted a more conservative re
management in these forests, limiting burning to strategic
property protection. The wildres in a severe drought year in
the 2002/2003 re season were widespread, given the large
numbers of dry lightning storm ignitions, and a dry continuous
fuel cover over a wide area. Much of the forest regrowth
in the sub-alpine forests is regrowing from lignotuberous
resprouts and seedlings in heavily burnt areas.
Rabbit plagues between 1880 and 1950s were a regular
feature until the introduction of Myxomatosis rabbit virus
in the 1950s. These rabbit plagues also had a major impact
on vegetation cover and fuel build-up, especially during
droughts, and led to widespread sheet and gully erosion
(Costin 1954).
Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests 247
Gilmour and Plumwood (1982) collated anecdotal evidence
from neighbours of Budawang National Park that conrmed
the frequent, chaotic, and uncontrolled pattern of burning
prior to the introduction of re permits under the 1949
Rural Fires Act of NSW. Occasional extensive hot res were
known during drought seasons such as 1913/14, 1929/30 and
1939/40. Characteristic uncontrolled spring burning on the
western side of Morton, Budawang, and Deua National Parks
has become much less frequent since the 1950s, with only
occasional summer wildres of smaller size than in earlier
decades of the twentieth century. National parks’ management
from the mid 1970s has become more conservative, with
burning associated with protection of life and property
along national park boundaries. Occasional larger prescribed
burning has occurred within parts of Deau and Wadbilliga
National Parks but the gross area burnt within a ten to fteen
year period is suggesting a longer re frequency between 25
to 40 years. Extensive and intense wildres during El Nino
drought periods could explain the lack of smaller, more
numerous intervening re events during non-drought years.
Since the introduction of exotic perennial grasses in the
1960s and 1970s, invasive perennial grasses such as
Phalaris (Phalaris aquatica) have invaded grassy forests
and woodlands adjoining improved pastures in agricultural
landscapes of the South-West Slopes and South-Eastern
Highlands bioregions. These grassy woodlands have lost
many of their original native understorey species, and very
few grassy forests in the valleys or lower slopes can be found
in their original native condition on private land, on travelling
stock routes, or on larger Crown Land blocks. In western
lower elevations, invasive weeds, such as St Johns Wort
(Hypericum perforatum) have spread into larger patches of
grassy forests on State Forests, National Parks, and Crown
Lands. Attempts to control these invasive species have been
sporadic and poorly-coordinated. Fire and drought can
exacerbate invasion of aggressive weeds and limited natural
control agents, grazing of exotic stock, and benign neglect on
public and private land, may threaten and possibly eliminate
some of the more threatened natural ecosystems, particularly
in the grasslands, grassy forests and woodlands.
Conservation significance and threatening processes
It is estimated that nearly 4.5 million hectares (4 447 900
hectares) of forests or woodlands covered 72% of the Southern
Forests study area at the time of European settlement.
The remaining 28% of vegetation was made of non-forest
vegetation formations, such as tall shrublands, heaths,
swamps, or grasslands (Table 12). In the last 200 years, the pre-
1750 area of forests has been reduced to about three million
hectares (3 120 400 hectares), about 44% of the study area,
while the area of non-forest vegetation has been reduced to
428 000 hectares, about 7% of the estimated pre-1750 area.
Despite these reductions however, there are still considerable
areas of native vegetation, covering about 50% of the
Southern Forests study area.
However the clearing and modication of native vegetation
has not been uniform across the area. The vegetation
formations most affected by change have been the Grassy
Woodlands and Temperate Grasslands, and the Riparian
and Grass/Shrub Forests. These formations are found on
the more fertile and less rugged land, that provided the
most suitable sites for agricultural settlement in the 19th and
early 20th century. In the Shoalhaven and Kiama precincts,
rainforests were cut for their timber in the 19th century, and
then cleared. Considerable areas of the montane tablelands
and the less productive dry grass/shrub forests were cleared
for pine plantations in the Oberon and Tumut-Tumbarumba
areas in the 1960s as part of Australia’s commitment to
self-sufciency in wood products. The shift in population
to the coast in the last 20 years has led to further clearing
of signicant coastal complex vegetation and the swamp
forests/wet heaths, and sedgelands.
The forests in the more rugged parts have been less changed
since European settlement. The least affected formations
have been the high elevations Alpine/Sub-alpine Complex
and the Sub-alpine Low Forests. The extent of clearing has
been as low as 1–3%. Nevertheless extensive summer grazing
and occasional intense res have affected the ground-cover
and soils. Severe res at intervals between 40 and 60 years
have modied the structure of the Sub-alpine Low Forests
and Ash Eucalypt Forests towards a regrowth rather than an
old growth form.
The Heath forests/Heathlands, as well as Vegetation on Rock
Outcrops/Screes, have not been signicantly cleared because
of their unsuitability for agriculture or forestry, but are
currently under pressures from more frequent re regimes,
particularly along the South Coast escarpment, which seems
to be experiencing more frequent drought periods at the
height of summer.
The production of the rst vegetation map in 2000 focussed
on the mapping of forest ecosystems, and did not include
the mapping of extant native grasslands. The mapping of
pre-1750 grasslands was undertaken to limit errors in pre-
1750 forest ecosystem estimates. The 2005 version of the
vegetation map now contains extant grasslands areas. Pre-
1750 grassland area is estimated at 450 000 hectares, with
about 12 500 hectares or about 3% remaining. The estimate of
12 500 hectares is a preliminary one, and with further eld
investigations, might rise to 20 000 hectares, including
partially degraded natural temperate grasslands. This still
represents a small fraction of pre-1750 area.
As part of the assessment of the criteria of comprehensiveness
and representativeness in 1999 (JANIS 1996), experts
assessed and ranked the vegetation groups. Based on the
estimates of pre-1750 and extant areas, and the extent of
clearing since 1750, these experts reached consensus on the
rarity and vulnerability rankings of each vegetation group.
The latter estimates have been since updated to reect the
2005 version of the pre-1750 and extant vegetation map, as
well as accounting for the current reserve system.
248 Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests
Rarity of each vegetation group was considered separately
to that of vulnerability. The criterion of area less than
1000 hectares was used to identify rare vegetation groups
(JANIS 1996). Because of possible errors in the estimation
of pre-1750 and extant areas, this latter criterion was adjusted
where local knowledge assessed some vegetation groups
as being either rare or not rare. The rare vegetation groups
preliminarily identied by vegetation formation: Rainforest
(164, 172, 168, 197); Ash Forest (62); Grass/Shrub Forests
(183); Heaths/ Heath Forests (136, 184, 200, 216); Swamp
Heaths/Forests (65, 125, 126, 212); Vegetation on Rocky
Outcrops (192); Alpine Complex (204, 205); Coastal
Complex (195, 196); and Wetlands (188, 189, 199).
Vulnerability ranking was based on an estimated area of
percentage cleared since pre-1750 (1 was considered very
high vulnerability, 5 was very low). The type and signicance
of environmental pressures and the vulnerability of a
vegetation group to those pressures were then considered.
The environmental pressures considered were grazing,
clearing, logging, weeds, recreational pressure, re, urban
development and hobby farming. The vegetation formation
with the highest number of vulnerability class 1 or 2 scores
is the Grassy Woodlands/Temperate Grasslands, with over
Table 11. Inferred historical and current disturbance regimes in the major forest formations of the Southern Forest study area
Vegetation formation Historical disturbance Settlement period (1820–1950s) Current regimes
regimes (pre 1820s) (1950s to present)
2. Wet Sclerophyll Forests Very infrequent intense Infrequent fires at 15–30 year intervals Infrequent intense fires at 30–50 year
fires at 50–100 year during droughts. Selective timber cutting intervals. Intensive timber harvesting.
intervals. and some clearing. Sporadic insect Sporadic insect defoliations.
defoliations.
4. Montane Tableland Infrequent moderate to Frequent low to moderate intensity fires Infrequent moderate to high intensity
Forests high intensity fires at at 5–10 year intervals. Timber harvesting, fires at infrequent intervals, more than
20–50 intervals. Low moderate grazing and extensive clearing 20 years. Moderate to high grazing
levels of grazing. for pasture. pressures and timber harvesting.
5. Dry Grass/Shrub Forests Infrequent moderate to Frequent low to moderate intensity fires Infrequent moderate to high intensity
high intensity fires at 12–30 at 5–10 year intervals. Timber harvesting, fires at infrequent intervals, more than
year intervals. Low levels moderate grazing and extensive clearing 20 years. More frequently burnt as part
of grazing. for pasture. of prescribed burning programmes in
the 1950s to early 1970s.
7. Dry Shrub Forests Infrequent moderate to low Frequent low to moderate intensity fires, More frequent intense summer fires at
intensity fires, at 10–20 year at 5–10 year intervals, mainly in early 10–20 year intervals. Timber harvesting
intervals, from late spring Spring. Occasional intense summer fires and low levels of grazing.
to late summer, low levels at 21–40 year intervals. Selective timber
of grazing. Occasional high cutting.
intensity fires
6. Grassy Woodlands/ Light grazing, occasional Frequent fires of varying intensity. Heavy grazing, infrequent fires,
Grasslands fires of moderate intensity. Heavy grazing, fencing and weed incursion. Replacement by
fragmentation. introduced perennial grasses.
12. Sub-alpine Low Occasional medium sized Frequent fires of varying intensity, Occasional intense fires during
Forests / Woodlands fires of moderate to high with occasional high intensity fires. drought years. Very light grazing
intensity. Very light grazing. Heavy grazing. by native animals.
Fig.7. Vegetation formations showing vegetation groups within
each vulnerability classes by vegetation formation
Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests 249
11 vegetation groups in vulnerability class 1 (Fig. 7). The
next most threatened vegetation formation comprises the
Riparian Forests, with four vegetation group s spread between
vulnerability classes 1, 2, and 3. Both these formations are
found on atter more fertile country in the study area. The
much lower representation of high vulnerability classes in
the remaining vegetation formations reect the less fertile
and hillier terrain of the study area, which has been less
cleared and modied in the last 200 years.
The current reserve system covers 1.5 million hectares
(1 565 200 hectares) and occupies 25% of the pre-1750 area.
It caters well for the least vulnerable vegetation formations
in the steeper, less fertile terrain, but is still missing some
of the more vulnerable vegetation formations and vegetation
groups (Table 12). After applying the JANIS criteria, the
total target area, comprising all vegetation groups, comes to
771 100 hectares. After deducting the area of vegetation
in the current reserves, there still remains about 220 600
hectares to be conserved on private land in some form.
The priority of the recent Southern Regional Forest
Agreement was to transfer land from Crown Land and State
Forests to formal conservation areas such as National Parks
and Nature Reserves. This process largely overlooked the
more vulnerable vegetation groups in the Grassy Woodlands/
Temperate Grasslands and Grassy Forests formation, which
are principally on private land on the Tablelands and
Western Slopes, and reects the current bias in the reserve
system towards the less vulnerable vegetation formations.
The remaining conservation target of 220 600 hectares
includes 196 300 hectares in the most vulnerable vegetation
formations. Given the extent of past clearing and ongoing
modication of native vegetation on private lands, it may be
difcult to attain the goal of a truly comprehensive, adequate,
and representative (CAR) reserve system. Without some form
of protection through conservation agreements and incentive
schemes, vulnerable vegetation formations will most likely
continue to degrade from the combined impacts of habitat
fragmentation, weed infestation, and loss of native species.
Conclusion
The approach used to map vegetation in the Southern
Forests relied heavily on air photo interpretation. This is in
marked contrast to the decision-tree mapping approach of
Keith & Bedward (1999) in the South-East Forests region,
which relied heavily on environmental variables to predict
the occurrence of vegetation. Keith and Bedward could not
nd relationships between classied survey data and the API
mapping, and so discarded the API mapping as a modelling
layer. While there is a scale difference in the size of eld
survey samples, relative to the size of the API polygons,
Table 12 Conservation significance of vegetation formations
Broad Formation Pre-1750 Extant % of total % Cleared Area in % of pre- JANIS
area (ha) area (ha) extant area reserves (ha) 1750 area in target not in
reserves reserves (ha)
01 Rainforests 40 100 29 700 1% 26% 15 100 38% 1 300
02 Moist Eucalypt Forests 577 600 445 800 14% 23% 240 900 42% 1 400
03 Ash Eucalypt Forests 110 900 110 700 4% 0% 90 700 82% 100
04 Montane Tableland Forests 654 900 474 500 15% 31% 310 900 47% 9 700
05 Dry Grass/Shrub Forests 2 296 600 1 026 800 33% 56% 282 900 12% 143 300
06 Grassy Woodlands/Grasslands 1 314 800 89 600 3% 93% 39 700 3% 45 400
07 Dry Shrub Forests 620 700 481 400 16% 22% 244 100 39% 5 600
08 Heath Forests, Mallee Low
Forests, & Heathlands 155 900 150 300 5% 3% 109 100 70% 100
09 Swamp Forests, Wet Heaths,
& Sedgelands 51 000 30 100 1% 42% 12 100 24% 4 300
10 Vegetation on Rock Outcrops 49 600 48 300 2% 3% 29 300 59% 1 800
11 Riparian Forests 53 100 11 400 0.4% 79% 3 300 6% 2 200
12 Sub-alpine Low Forests 112 800 112 200 4% 1% 107 100 95% 0
13 Alpine-Subalpine Complex 83 300 80 400 3% 3% 76 600 92% 0
14 Coastal Complex 33 000 10 200 0.3% 68% 3 200 10% 1 600
15 Freshwater Wetlands 20 100 19 000 1% 5% 200 1% 3 800
Totals 6 174 400 3 120 400 50% 1 565 200 25% 220 600
250 Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests
the use of eld knowledge and eld sampling has helped to
overcome some of the potential correlation issues between
the two sources of mapping data. Ferrier et al. (2002) argues
the case for integrating traditional vegetation mapping and
environmental layers to map vegetation communities in
the North-East Forests of New South Wales. This could be
considered if the API mapping in the study area could be
thoroughly checked for consistency in coding and modied
to match more closely the environmental gradients and
natural vegetation patterns.
This method of vegetation mapping has proved to be
versatile, exible, and adaptable in a consistent and more
or less repeatable manner. The methods employed have
some considerable advantages over mathematical modelling
techniques, such as GAMS (Austin et al. 1996) or a decision
tree expert system (Keith & Bedward 1999). Its advantages
in mapping vegetation include:
• production using the best available GIS and eld
knowledge, using intuitive and interactive GIS methods;
and
• rapid revision using a wide variety of site and mapping
data, as well as local and regional knowledge of vegetation.
Using their combined knowledge, a botanist and an air photo
interpreter can produce a draft vegetation map simply and
quickly. Using the methods described in this paper, data from
further validation work can be quickly incorporated into the
map, using adaptive management methods to update it.
Although the method of vegetation mapping does not adhere
to the rules of strict objectivity and transparency, it does
incorporate a range of eld survey data, eld knowledge,
survey and mapping data from other research and survey
reports, and iterative eld validation. The method allows
for ready update of both the pre-1750 and extant vegetation
maps, as more ora survey and improved API mapping
data become available. Since the completion of the project
in 2000, the original vegetation map has been extensively
checked and updated for use in re management and
management plans in new RFA and existing reserves.
The continued use and adaptation of the original work
demonstrates a continuing use and acceptance of this work
in planning and management. With further eld validation
on private land, this regional vegetation map could be used
in State of Environment reports of vegetation condition at
levels of state, regional or local government.
Acknowledgements
I wish to acknowedge the support and direct sponsorship of
the original vegetation mapping project of the southern CRA
by the Department of Urban Affairs and Planning of NSW
and the Commonwealth Department of Environment and
Heritage. The Department of Environment and Conservation
also contributed funds to the preparation of this paper for
inclusion in Cunninghamia. I also acknowledge the support
given to me by the School of Resources, Environment, and
Society, at the Australian National University who enabled
me to complete this paper while working concurrently on
my Masters Thesis.
I am deeply indebted to the staff of the Southern CRA Unit who
untiringly and professionally undertook project management,
eld planning, logistical and technical support. This paper
is a tribute to their work. They include the following: Peter
Beukers, Virginia Thomas, Jess Szigethy-Gyula, Helen
Achurch, Allison Trewick, Anita Zubovic, Nicole Walkey,
Mark Robey, Tanya Harrison, Naomi Robbie, Will Inveen,
Malcolm Stephens, David Keith, Genevieve Wright, Peter
Hesp, Rob Mezzatesta, Steve House, Brenton Marchant,
Bianca Redden, Chris Beer, and Mary Grgic. I also wish
to acknowledge the contribution of personnel within State
Forests of NSW, including Doug Binns, Bob Bridges, and
Andrew Stirling.
The study would not have been possible without external
help and contractors. They include: Michael Doherty, Phil
Gilmour, Mike Austin, Michael Bedward, Jackie Miles,
Heather Stone, Isobel Crawford, Anne Duncan, Roger
Lembit, Mark Robertson, Owen Maguire, Simon Hunter,
Rob Streeter, Mark Fisher, Naomi Robbie, Will Inveen, and
Matt Brookes. I wish to thank especially Phil Gilmour and
Michael Doherty, who undertook a signicant part of the
original vegetation mapping work in the western and eastern
part of the study area.
I acknowledge the provision of survey data contributed by
David Keith, Doug Binns, John Benson, Frank Ingwerson,
Mike Austin, Bob Outhred, Jacky Miles, Sarah Sharpe,
Kevin Mills, Nicki Taws, Phil Gilmour, and Carol Helman.
Finally I wish to thank Phil Gilmour, Michael Doherty and
Susan Jackson for their editorial advice and proof-reading of
the manuscript. Finally the referees of this paper provided
invaluable assistance and advice on the style and content.
Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests 251
References
Austin, M.P. (1978) Chapter 5 Vegetation in: M.P. Austin & K.D.
Cocks (eds). Land use on the South Coast of New South Wales
(CSIRO: Canberra).
Austin, M.P. & Belbin, L. (1982) A new approach to the species
classication problem in oristic analysis. Australian Journal
of Ecology 7: 75–89.
Austin M.P. and Cawsey M. (1996a) Preliminary forest community
classication for south-eastern NSW. Unpublished report
prepared for NSW National Parks & Wildlife Service (CSIRO
Division of Wildlife & Ecology: Canberra).
Austin M.P, Cawsey, M.; Coops, N. & Barnett, G. (1996) Pre-1750
mapping of vegetation for the South Coast forests area. Final
Report Volumes 1 & 2. Unpublished report prepared for NSW
National Parks & Wildlife Service (CSIRO Division of Wildlife
& Ecology: Canberra).
Banks J. (1982) The Use of dendrochronology in the interpretation
of the dynamics of snowgum forests. PhD Thesis.Dept of
Forestry, ANU, Canberra.
Baur, G.N. (1989) Forest types in New South Wales. Research Note
17 (Forestry Commission of NSW: Sydney).
Bedward, M. (1998) Albero decision tree modelling system — A
very incomplete user’s guide. Unpublished report. 13pp.
Bedward, M. (1999) Fidel: A utility to prole classication groups
in terms of attribute delity (Version 2.1 for Windows 9x/NT).
Unpublished report (NSW National Parks & Wildlife Service:
Queanbeyan).
Bedward, M., Keith, D.A. & Pressey, R.L. (1992) Homogeneity
analysis: Assessing the utility of classications and maps of
natural resources. Australian Journal of Ecology 17: 133–139.
Belbin, L. (1991) The analysis of pattern in bio-survey data. pp.
176–190. In: Nature conservation: cost effective biological
surveys and data analysis, C. R. Margules & M. P. Austin,
editors, (CSIRO: Canberra).
Belbin L. (1994) PATN. Pattern analysis package (CSIRO:
Canberra).
Benson, J.S. (1994) The native grasslands of the Monaro region:
southern tablelands of New South Wales. Cunninghamia 3:
609–650.
Binns, D.L. & Kavanagh, R.P. (1990a) Flora and fauna survey of
Nalbaugh State Forest (part) Bombala district, Eden region,
south eastern New South Wales. Forest Resources Series No. 9
(Forestry Commission of NSW, Sydney).
Binns, D.L. & Kavanagh, R.P. (1990b) Flora & fauna survey, Nullica
State Forest (part) Eden district, Eden region. Forest Resources
Series No. 10 (Forestry Commission of NSW: Sydney).
Binns, D.L. (1995) Flora survey & vegetation of Tallaganda State
Forest. Environmental Impact Statement of Forest Harvesting
in Tallaganda State Forest (State Forests of NSW: Sydney).
Binns, D.L. (1997a) Unpublished survey data. Flora survey of
Bago-Maragle State Forests (State Forests of NSW: Sydney).
Binns, D.L. (1997b) Unpublished survey data. Flora survey of
Carabost, Woomargama, and Tumut State Forests (State Forests
of NSW: Sydney).
Breckwoldt, R. & Breckwoldt, G. (1979) A preliminary vegetation
survey of Goura, Bermagui and Bournda Nature Reserves,
Mimosa Rocks & Wallaga Lake National Parks. Unpublished
report (NSW National Parks & Wildlife Service.
Clarke, P.J. (1989) Coastal Dune Vegetation of New South Wales.
Technical Report 89/1 (Coastal Studies Unit, University of
Sydney: Sydney and Soil Conservation Service: Sydney.
Commonwealth of Australia (1992) National Forest Policy
Statement. A New Focus for Australian Forests (Commonwealth
of Australia: Canberra).
Costin A.B. (1954) A study of the ecosystems of the Monaro. NSW
Minister for Conservation, Sydney.
Costin A.B. (1966) Management opportunities in Australian high
mountains catchments. pp 565–577. In: W.E. Sopper & H.W.
Hull (eds) Forest Hydrology (Pergamon Press).
CSIRO Division of Wildlife & Ecology (1997) Production of a
lithology and nutrient index layer — Southern and Eden CRA
Regions. Unpublished report prepared for NSW National Parks
& Wildlife Service: Queanbeyan.
CSIRO Division of Wildlife & Ecology (1999) Unpublished survey
data: Clyde Mountain and environs and Kosciuszko National
Park. Unpublished report prepared for NSW National Parks &
Wildlife Service.
Dept of Mineral Resources (2000) Geology of the CRA Region. A
report undertaken for the NSW CRA/RFA Steering Committee
( NSW Dept of Urban Affairs & Planning: Sydney).
Dodson, J.R, Kodela, P.G & Myers, C.A. (1988) Vegetation survey
of the Tantawangalo Research Catchments in the Eden Forestry
Region, New South Wales. Forest Resource Series No. 4 (State
Forests of NSW: Sydney).
Doherty, M.D. (1996) Vegetation survey and mapping of Tinderry
Nature Reserve. Unpublished report prepared for NSW National
Parks & Wildlife Service, Queanbeyan District (CSIRO
Division of Wildlife & Ecology: Canberra).
Doherty, M.D. (1997) Vegetation survey and mapping of
Mundoonen Nature Reserve. Unpublished report prepared for
NSW National Parks & Wildlife Service, Queanbeyan District
(CSIRO Division of Wildlife & Ecology: Canberra).
Doherty, M.D. (1998a) Vegetation survey and mapping of
Brindabella National Park and adjacent Vacant Crown Lands.
Unpublished report prepared for NSW National Parks &
Wildlife Service, Queanbeyan District. (CSIRO Division of
Wildlife & Ecology: Canberra).
Doherty, M.D. (1998b) Vegetation survey and mapping of
Burrinjuck Nature Reserve. Unpublished report prepared for
NSW National Parks & Wildlife Service, Queanbeyan District
(CSIRO Division of Wildlife & Ecology: Canberra).
DLWC (2000) Soil and regolith Attributes for Southern CRA/RFA
Modelling Purposes. A report undertaken for the NSW CRA/
RFA Steering Committee (NSW Dept of Urban Affairs &
Planning: Sydney).
DUAP (1999) Forest ecosystem classications for the upper and
lower North East CRA Regions: A report undertaken for the
NSW CRA/RFA Steering Committee (NSW Dept of Urban
Affairs & Planning: Sydney).
DUAP (2000a) CRAFTI Southern Report: A project undertaken as
part of the NSW Comprehensive Regional Assessments (NSW
Dept of Urban Affairs & Planning: Sydney).
DUAP (2000b) JANIS Conservation Requirements: A project
undertaken as part of the NSW Comprehensive Regional
Assessments (NSW Dept of Urban Affairs & Planning:
Sydney).
EcoGIS (2002) Vegetation, rare plants and weeds mapping in
Monga NP. Unpublished Report (NSW Dept of Environment
& Conservation: Far South Coast Region).
EcoGIS (2004a) Mapping of vegetation ecosystems in new and
existing conservation reserves, SWS Region. Unpublished
Report (NSW Dept of Environment & Conservation: South
West Slopes Region).
252 Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests
EcoGIS (2004b) Vegetation survey & mapping of reserves, Upper
Murray Area Unpublished Report (NSW Dept of Environment
& Conservation: Snowy Mountains Region).
EcoGIS (2004c) Mapping of vegetation ecosystems in Bugong
National Park, and Cambewarra Range and Tapitallee Nature
Reserves. Unpublished Report (NSW Dept of Environment &
Conservation: Far South Coast Region).
Environment Australia (2000) Response to disturbance of ora and
fauna. Report of the NSW Comprehensive Regional Assessments
(NSW Dept of Urban Affairs & Planning: Sydney).
Eurobodalla Shire Council & NPWS (1998) Vegetation surveys
within Eurobodalla Shire. Unpublished survey data.
Unpublished report to the Eurobodalla Shire Council (NPWS
of NSW: Queanbeyan).
Fanning, F.D. & Mills, K. (1989) Natural resource survey of the
southern portion of Rockton section, Bondi State Forest.
Forest Resources Series No. 6 (Forestry Commission of NSW:
Sydney).
Fanning, F.D. & Rice, B. (1989) Natural resource survey of the
northern portion of Rockton section, Bondi State Forest.
Forest Resources Series No. 7 (Forestry Commission of NSW:
Sydney).
Fanning, F.D. & Mills, K. (1990) Flora and fauna survey of the
Myanba Creek catchment, Coolangubra State Forest, Eden
region. Forest Resources Series No. 11 (Forestry Commission
of NSW: Sydney).
Fanning, F.D. & Fatchen, T.J. (1990) The Upper Wog Wog Creek
catchment of Coolangubra and Nalbaugh State Forest, (Mines
Road area), New South Wales. A fauna and ora survey. Forest
Resources Series No. 12 (Forestry Commission of NSW:
Sydney).
Fanning, F.D. & Clarke S.S. (1991) Flora and fauna survey of Jingo
Creek catchment, Nullica State Forest, Eden region. Forest
Resources Series No. 14 (Forestry Commission of NSW:
Sydney).
Fanning, F.D. & Mills, K. (1991) The Stockyard Creek Catchment
of Coolangubra State Forest, New South Wales: A Fauna
and Flora Survey. Forest Resources Series No. 13 (Forestry
Commission of NSW: Sydney).
Ferrier, S, M. Drielsma, G. Manion, & G. Watson (2002) Extended
statistical approaches to modelling spatial pattern in biodiversity
in northeast New SouthWales. Biodiversity and Conservation
11: 2309–2338.
Forward, L.R. and Hall, R.M. (1997) Alps vegetation re response
monitoring system — Final Project Report for Australian Alps
Liaison Committee (NSW National Parks & Wildlife Service:
Sydney).
Gilmour, P.M. & Plumwood, V. (1982) A survey of the vegetation of
Budawang National Park. Unpublished report (NSW National
Parks & Wildlife Service: Nowra District).
Gilmour, P.M. (1983) A survey of the vegetation of Nadgee Nature
Reserve. Unpublished report (NSW National Parks & Wildlife
Service of NSW: Eden District).
Gilmour, P.M. (1985) Vegetation survey of Deua National Park.
Unpublished report (NSW National Parks & Wildlife Service:
Narooma District).
Gilmour, P.M., Helman, C.E. & Osborne, W.S. (1987) An ecological
study of the Mount Tennant-Blue Gum Creek Area, A.C.T.
Unpublished report (Conservation Council of the South-East
Region and Canberra: Canberra).
Gunn, R.H., Story, R., Galloway, R.W., Duffy, P.J.B., Yapp, G.A. &
McAlpine, J.R. (1969) Lands of the Queanbeyan-Shoalhaven
Area, A.C.T. and N.S.W. Land Research Series No. 24 (CSIRO:
Melbourne).
Harden G. (1990–93) Flora of New South Wales. Volumes 1–4
(UNSW Press: Kensington)
Helman, C.E. (1983) Inventory analysis of southern NSW rainforest
vegetation. Master of Science Thesis; University of New
England, Armidale NSW.
Helman, C.E., Gilmour, P.M., Osborne, W.S., & Green, K. (1988)
An ecological survey of the Upper Cotter Catchment Wilderness
Area Namadgi National Park, A.C.T. Unpublished report
(Conservation Council of the South-East Region & Canberra).
Hibberd, J. & Taws, N. (1993) The long paddock revisited 15 Years
on: a comparative study of the condition and use of travelling
stock reserves in the Southern Tablelands 1977–1993.
Unpublished report (Nature Conservation Council of NSW).
Hutchinson M.F. (1989) A new objective method for interpolation
of meteorological variables from irregular networks applies to
the estimation of monthly mean solar radiation, temperature,
precipitation, and windrun. Technical Memorandum 89/5–
104pp (CSIRO Division of Water Resources: Canberra).
Ingwersen, F. (1972) Black Mountain Reserve preliminary
development and management plan (Dept of the Capital
Territory).
Ingwersen, F., Evans, O. & Grifths, B. (1974) Vegetation of
the Ainslie-Majura Reserve: Dept of the Capital Territory
Conservation Series No.2 (Australian Government Publishing
Service: Canberra).
Ingwersen, F. (1983) Numerical analysis of the timbered vegetation
in Tidbinbilla Nature Reserve, ACT Australia. Vegetatio 51:
157–179.
Ingwersen, F. (1992) Vegetation surveys in Gudgenby Nature
Reserve 1980–1992. Unpublished survey data (ACT Parks &
Conservation: Canberra).
Jervis, J. (1962) The Yarrawa Brush, pp. 41-58. In: A history of the
Berrima District 1798–1963 (Berrima County Council).
JANIS (1996) Proposed nationally agreed criteria for the
establishment of a comprehensive, adequate and representative
reserve system for forests in Australia. Report of the Joint
ANZECC / MCFFA National Forest Policy Implementation
Sub-committee (Commonwealth of Australia: Canberra).
Jurskis, V., Shields, R. & Binns, D. (1995) Flora survey, Queanbeyan/
Badja Environmental Impact Statement Area, Southern Region,
New South Wales. Environmental Impact Statement Queanbeyan
and Badja Management Areas Supporting Document 3 (State
Forests of NSW: Sydney).
Keith, D.A (1999) Allocasuarina survey within Wadbilliga National
Park. Unpublished survey data (NSW National Parks & Wildlife
Service: Sydney).
Keith, D.A. (2004) Ocean shores to desert dunes. The native
vegetation of New South Wales and the ACT (NSW Dept of
Environment & Conservation: Sydney).
Keith, D.A. & Bedward, M. (1999) Native vegetation of the South-
East Forests region, Eden, New South Wales. Cunninghamia
6(1): 1–218.
Keith D.A. & Benson D.H. (1988) The natural vegetation of the
Katoomba 1: 100 000 map sheet.Cunninghamia 2(1): 1–146.
Lockwood, M., Wise, P., Lane, M., Williams, J., Godd, L &
Costello, D. (1997) Eurobodalla National Park Flora Survey:
Johnstone Centre of Parks, Recreation and Heritage Report No.
84 (Charles Sturt University: Albury).
Maguire, O. & Hunter, S. (2000) CRAFTI Southern Floristic Field
Validation Report and Eucalypt Remnant Mapping of Tumut
and Tarcutta areas, South-West Slopes NSW. Unpublished
report (NSW National Parks & Wildlife Service:Queanbeyan).
McDougall K.L. & Walsh N.G. (in prep.) Treeless vegetation of the
Australian Alps. Cunninghamia.
Cunninghamia 9(2): 2005 Gellie, Vegetation of the Southern Forests 253
Mills, K. (1993) The natural vegetation of the Jervis Bay region of
New South Wales. Report to the National Estates Grants Scheme
1990/1991 (NSW Heritage Assistance Program: Sydney).
Mills, K. (1999) Vegetation Survey Sheets for the Shoalhaven
Region. Unpublished survey sheets compiled by author (NSW
National Parks & Wildlife Service: Queanbeyan).
Mills, K & Associates Pty Limited (1996a) Flora and fauna
assessment, Milton Ulladulla Structure Plan. (Shoalhaven City
Council: Nowra).
Mills, K & Associates Pty Limited (1996b) The natural vegetation
of the Nowra Area, City of Shoalhaven, New South Wales.
Vegetation Communities and Rare Plant Species. Unpublished
report (Shoalhaven City Council:Nowra).
Mills, K. & Jakeman, J. (1995) Rainforests of the Illawarra District
(Coachwood Publishing: Jamberoo).
National Parks Association (1998) Benandera State Forest
Biodiversity Survey. Unpublished report (National Parks
Association: Sydney).
Nicholas Graham-Higgs Pty Ltd (2002a) Meroo National Park
and Barnunj State Recreation Area Vegetation survey and
mapping. Unpublished Report (NSW Dept of Environment &
Conservation: South Coast Region).
Nicholas Graham-Higgs Pty Ltd (2002b) Murramarang National
Park and Offshore Islands Vegetation survey and mapping.
Unpublished report (NSW Dept of Environment & Conservation:
South Coast Region).
Nicholas Graham-Higgs Pty Ltd (2002c) Kooraban and Gulaga
National Parks Vegetation survey and mapping. Unpublished
report (NSW Dept of Environment & Conservation: Far South
Coast Region).
Nicholas Graham-Higgs Pty Ltd (2004) Conjola and Morton
National Parks Vegetation survey and mapping. Unpublished
report (NSW Dept of Environment & Conservation: South
Coast Region).
Nicholas Graham-Higgs Pty Ltd (2005). Flora survey & vegetation
re-mapping of Bungonia State Conservation Area, Bees
Nest Nature Reserve and North-Western Morton National
Park. Unpublished report (NSW Dept of Environment &
Conservation: South Coast Region).
NPWS (1995) Eastern Bushland Database. GIS layer prepared for
the vegetation of the Eastern Coastal and Tablelands Regions
of NSW. Unpublished report (National Parks and Wildlife
Service: Sydney)
NPWS (1996) Interim Forest Agreement Process (IFA) preparation
of data and databases Volume II: Pre 1750 vegetation and forest
disturbance mapping for the Tumut study area. Unpublished
report for the Resource and Conservation Assessment Council.
(National Parks & Wildlife Service of NSW: Queanbeyan).
NPWS (1998) Vertebrate fauna survey. Unpublished report for
the Joint Commonwealth NSW Regional Forest Agreement
Steering Committee as part of the NSW Comprehensive
Regional Assessments (National Parks & Wildlife Service of
NSW: Queanbeyan).
NPWS (2000). Southern CRA vegetation surveys 1997–2000.
Unpublished survey data (National Parks & Wildlife Service of
NSW: Queanbeyan).
Outhred, R. (1986) Vegetation survey of Wadbilliga National Park.
Unpublished survey data (NSW National Parks & Wildlife
Service: Eden District).
Poore, M.E.D. (1955) The use of phytosociological methods in
ecological investigations. I. The Braun-Blanquet system.
Journal of Ecology 43: 226–244.
Pulsford I.F. (1991) History of disturbance in the white cypress
pine (Callitris glaucophylla) forests of the Lower Snowy River
Valley, Kosciuszko National Park. Masters Thesis. Australian
National University, Canberra.
Ryan, M & Stubbs, B.J. (1996) Pre-1750 vegetation model:
historical study. Consultancy report (National Parks & Wildlife
Service of NSW: Sydney).
SFNSW (1999a) Collection of vegetation data from historical
portion plan surveys of the Southern CRA Region. A project
undertaken for the Joint Commonwealth NSW Regional
Forest Agreement Steering Committee as part of the NSW
Comprehensive Regional Assessments (State Forests of NSW:
Sydney).
SFNSW (1999b) Various vegetation surveys within State Forests
Unpublished survey data. (State Forests of NSW: Sydney).
Skelton, N, J. & Adam, P. (1994) Beecroft Peninsula Vegetation
Survey. Unpublished Report (Australian Nature Conservation
Agency: Canberra).
Steenbecke, G.L. (1990) An Investigation into the ora and
vegetation of the middle Kowmung River valley, eastern New
South Wales. Honours Thesis, University of Sydney.
Taws, N. (1997) Vegetation survey & mapping of Jervis Bay
Territory. Unpublished Report (Environment Australia:
Canberra)
Thackway, R. & Cresswell, I.D. (Eds) (1995) An Interim
Biogeographic Regionalisation for Australia: a framework
for establishing the national system of reserves, Version 4.0
(Australian Nature Conservation Agency: Canberra).
Thomas V., Gellie, N.J.H., & Harrison T. (2000) Forest ecosystem
classication and mapping for the Southern Comprehensive
Regional Assessment (National Parks & Wildlife Service of
NSW: Queanbeyan).
Tindall, D., Pennay, C., Tozer, M.G., Turner, K. & Keith,D.A.(in
prep.). NSW Monitoring and Evaluation Trials. NSW Dept of
Infrastructure, Planning and Natural Resources 29. Native
vegetation map report series. No. 4: Araluen, Batemans Bay,
Braidwood, Burragorang, Goulburn, Jervis Bay, Katoomba,
Kiama, Moss Vale, Penrith, Port Hacking, Sydney, Taralga,
Ulladulla, Wollongong. NSW (NSW Dept of Environment &
Conservation and NSW Dept of Infrastructure, Planning and
Natural Resources: Sydney).
Togher, C. (1996) A Report on the biodiversity and land
management of the Abercrombie River Catchment (National
Parks Association: Sydney)
Ward, J.E. & Ingwersen, F (1988) A checklist of the vascular plants
of the Tidbinbilla Nature Reserve. Conservation Memorandum
No.8 (Parks & Conservation: ACT)
Williams, J. (1997) Vegetation survey of Bermagui Nature Reserve,
Biamanga National Park, Goura Nature Reserve and Wallaga
National Park. Unpublished report (NSW National Parks and
Wildlife Service: Narooma District).
Yee, T.W. & Mitchell, N.D. (1991) Generalised additive models in
plant ecology. Journal of Vegetation Science 2: 587–602.
Manuscript accepted 20 October 2005
