|
4.1 Sampling Intensity
4.2 New Survey Work
4.3 Map Unit Profiles and Field Identification
4.4 Endangered Ecological Communities
4.5 Defining Vegetation Map Units
4.6 Accuracy of the Vegetation Model
4. DISCUSSION
4.1 Sampling Intensity
The study area now supports a comprehensive coverage of full floristic
site data across all land tenures and environments. The LHCCREMS
project has increased site numbers by around 30%. Most importantly
these sites have sampled private land tenures, which are often the
source of different environments to those present in public tenures
(Pressey et al, in press).
Within the study area some areas have received intensive sampling
in the past (eg. small conservation reserves such as Tomaree National
Park; Glenrock and Awabakal Nature Reserves and areas of major development
pressure eg. Pinney Beach). In contrast, some areas contain sparse
sampling, particularly broader areas of Sandstone environments present
in National Parks. The scale to which communities are defined is
wholly dependent on the data available for use in the analysis
Councils, planners and other land management agencies now have
a foundation of regional data that can be augmented by new survey
work. By using the survey methods used in this report local area
studies can relate floristic values to broader regional assemblages
described in this report. Over time new analyses can be completed,
reviewing and refining community classification and improving map
accuracy.
4.2 New Survey Work
The development of the vegetation model has highlighted several
areas in the region that would benefit from further sampling. In
addition some Councils may require finer scale splits in some of
the Eucalypt dominated Map Units. A review of these issues is presented
below:
- Further sampling on the Wyong Coastal Plain to resolve floristic
values associated with Wetland Communities. A high priority is
to resolve the floristic relationship of Map Unit (QA13) (Payne
1999) to other communities in the region as there are currently
no samples within this unit.
- Further sampling on the Somersby Plateau in Gosford and Wyong
LGAs as the plateau has received minimal sampling effort.
- The eastern slopes of the Wollombi Valley between Bucketty and
Wollombi are not sampled and may reveal greater rain shadow effects
on the vegetation than is currently predicted.
- Sampling in the Congewai Valley south of Cessnock is required
to validate the occurrence of alluvial valley floor communities.
- Additional sampling in riparian environments (Casuarinna
glauca) and lowland depressions supporting E.tereticornis
forest on the Hunter Valley Floor.
- Further splitting of moist eucalypt forest communities delineating
relative dominance of Eucalyptus acmenoides, E pilularis
and Corymbia maculata in the Gosford and Wyong area.
- Some areas could not be accessed near Medowie in the Port Stephens
Shire. Clarification of the floristic values present in these
areas would be valuable.
- Coastal headlands particularly from Norah Head to Catherine
Hill Bay. Sampling intensity in this area was low, further understanding
of the distribution of Coastal Sand Wallum Heath is needed.
- Data on alluvial sediments of the coastal plain would contribute
to the understanding of canopy species distribution especially
between Eucalyptus saligna, Eucalyptus deanii, Eucalyptus
robusta, Eucalyptus pilularis, Eucalyptus tereticornis,
Eucalyptus grandis and Eucalyptus microcorys. Additional
data may also highlight the variables determining the presence
of Melaleuca spp. in the understorey.
- Swamp Mahogany-Paperbark Forest could be further sampled to
potentially model the sub forms outlined in Appendix D. Sufficient
data was not available to reliably model their distribution.
- Map Unit 1: Coastal Wet Gully Forest. Additional data may contribute
to a reliable grouping of these forms in order to allow for modelling.
Further understanding of the distribution of subtropical forms
in the region is needed.
- Map Unit 60: Hunter Valley Tall Open Forest. The high disturbance
of this community means data from these sites has not grouped
very tightly and further refinement could be possible. Sites from
Twelve Mile and Nine Mile Creek in Wallaroo and Medowie State
Forests aligned with this group however more refinement could
be done in this area.
- Map Unit 38: Redgum-Roughbarked Apple Forest. This community
has only been described by 2 sites. Its floristic characteristics
are closely related to Map Unit 37. Additional sites would assist
in improving estimations of its former extent as well as supporting
its definition as a single community.
- Coastal Bangalay communities
near Gosford. Several sites were collected with Eucalptusbotryoides
however this group of sites did not group tightly as a unit. Sites
were aligned to either Littoral Rainforest or Alluvial Tall Moist
Forest Groups. Additional sampling is required to refine potential
communities containing this species in coastal scrub communities.
4.3 Map Unit Profiles and Field Identification
Each profile contains a list of species that can help identify
one Map Unit from another. The species data has been drawn from
sites used in the analyses. Consequently, the species listed do
not represent the total floristic list within any given Community.
However, they provide indications of the relative probabilities
of encountering species using a randomly stratified sampling strategy.
To simplify matters, there are two types of species used to describe
communities. These are identified in the column headed fidelity
class. Positive species will be those that only occur in a single
Map Unit or at an abundance and frequency greater than all other
communities. The also table provides figures on average cover abundance
scores achieved within the Map Unit as compared to all others. Summary
structural data describing height and cover abundances for each
vertical strata of the Map Unit may guide recognition.
During the classification process floristic similarity between
sites overruled structural similarity. Variations in structural
form have been identified in the profiles where captured data describes
such variations. Often the difference between 2 forms may be the
addition of one or two canopy species. While this difference may
make a significant visual impact in the field, the floristic composition
is relatively consistent. For example sites that grouped as Map
Unit 1: Coastal Wet Gully Forest showed no significant grouping
where emergent trees were present. As a result such assemblages
have been left as one unit. Where API permits, the structural sub-forms
have been delineated on the map. These variations have been tabulated
in Appendix D for field reference based on structural form.
Uninformative species are given to assist with the description
of the Map Unit. These have been chosen by selecting the most frequently
occurring uninformative species at the authorÕs discretion. All
tree species recorded at sites within a Map Unit have been given
as well as some conspicuous though uninformative species. These
are useful for readers who may wish to relate such species to canopy
based classification systems. It is also explicitly describes the
variability in tree species which will make up any vegetation classification
system. Other species have been chosen if they occur at high frequencies
amongst sites at the group. These aid the picture of some prominent
structural characteristics such as shrubbiness, which may not be
clear from the positive diagnostic species.
Canopy Labels describe key canopy species likely to be found within
the Map unit. It is important to consider that classifications based
on a full floristic analyses do not always describe communities
identified by distinct canopy species therefore they are not always
good indicators. They have been presented for the purpose of easing
field recognition of map units however they do not contest to always
be reliable indicators of the community. The likelihood of species
occurring depends largely on the abundance and frequency with which
it occurs within the map unit. The greater the frequency the greater
the chance of it occurring.
The species presented remain exploratory. On going review and refinement
will be possible as new data is collected with ongoing council survey
efforts. Producing a definitive list of positive diagnostic species
for each Map Unit may only be resolved, if at all, after many years
of field-testing and refinement. Further, identification may also
be thwarted by previously unsampled variation within and between
communities (Keith & Bedward, 1999).
4.4 Endangered Ecological Communities
There are two communities found in the study area which are listed
under the Threatened Species Conservation Act (1995) as Endangered
Ecological Communities.
4.4.1 Norah Head Low Heath Woodland
The community ÒLow Woodland with heathland on indurated (hardsetting)
sandÓ (Community Qhd7 of Payne, 1997) is found at Norah Head. Analyses
of site data taken from within its mapped extent grouped this community
with a number of other sites describing Coastal Clay Heaths. It has
been mapped and recognised as a component of this group as Map Unit
48a.
4.4.2 Sydney Coastal River Flat Forest
"Sydney Coastal
River-Flat Forest:" is stated to occur in or near Dharug National
Park( Gazettal 12/02/99). The determination provides for a wide variety
of species to be included in the definition of this community. The
most recent review of the floristic composition of this community
is provided by NPWS (2000). This work divides the listed community
into two groups Alluvial Woodland and Riparian Forest. Alluvial Woodland
describes a woodland dominated by a wide variety of Eucalypts but
most commonly Eucalyptus tereticornis, E.amplifolia and Angophora
floribunda with occassional Casuarina cunninghamia subsp. Cunninghamia.
The community is supported by a dense ground cover dominated bygrasses
such as Oplismenus aemulus, Microlaena stipoides var stipoides,
Entolasia marginata and Echinopogon ovatus. A stratum of small
trees is sometimes present including Acacia parramattensis and
Melaleuca linariifolia. A sparse shrub cover of Bursaria
spinulosa also occurs.
This combination of species most closely aligns with Map Unit 14
Wollombi Redgum-Roughbarked Apple Woodland. A similar suite of canopy
species and a shared dense cover of grass and herb species are evident
although differences in shrub species occur.
Map Unit 14 extends along alluvial valley floors from the Hawkesbury
River along the Dubbo Gully and into the Wollombi Valley proper.
At the southern end of the distribution the community grades out
of Map Unit 40 Swamp Oak øRushland Forest adjacent to the river
into the broad plains. Eucalyptus deannii and Acacia parramattensis
are present. Further from the river, Redgum species tend to dominate
in the canopy. Deeper into the Wollombi Valley Casuarina cunninghamia
subsp cunninghamia is prevalent on stream banks. Evidence of
dry rainforest species become apparent in closer proximity to the
Hunter Valley floor.
4.5 Defining Vegetation Map Units
It can be observed in the Map Unit Profiles that some assemblages
are defined to a finer scale than others. The varying scale to which
communities are described is wholly dependent on the data available
for use in the analysis. In sum, the more data the greater the ability
to define floristic assemblages at a fine scale.
As a result not all Map Units have been defined to the same level.
Where this occurs attempts have been made in the Map Unit profile
descriptions to describe these variations. It is possible with the
collection of more data that some of these variations will be able
to be defined to a point where modelling these distributions is
possible.
4.6 Accuracy of the Vegetation Model
The accuracy of the vegetation model is difficult to compare with
other mapping in the region, as such testing has not been explicit
in those works. As a guide to modelling performance, Keith and Bedward
(1999) achieved an accuracy of 72% within a 100 metre radius for
the map of the South East Forests. Around 80% was achieved for a
window of 250 metre radius.
The map produced for this project will certainly differ from other
mapping in the region as different communities have been described
and mapped. Caution will need to be applied in its review to ensure
that comparisons are not made on the basis of previous communities
and mapping methods.
Inaccuracies of the model can be caused for several reasons. The
easiest to detect are those problems that derive directly from the
spatial layers used for predictions. These are already visible in
some areas in the Watagan Range where soil landscape mapping has
not edge matched features between the 1:250 000 series needed to
cover the Cessnock area to the more detailed 1:100 000 series along
the coastal area. Similarly, within the extant vegetation map, the
numerous observers involved in air photo interpretation have produced
some uneven boundaries. For the most part these will remain cosmetic,
until a means can be found to clean the various data sets. Others
problems may exist where features are simply not described in the
spatial layers. Shale lenses in sandstone, and basalt outcrops amongst
Carboniferous sediments will mean vegetation patterns relating to
these features will not be displayed. In a similar vein, some discussion
has already been made of the difficulties of low-lying terrain.
Climatic and terrain variables derived from the GIS may also contain
error.
The benefit of the modelling approach is that refinements are possible
and anticipated. The inclusion of new information from systematic
data collection, wider field experience, and the input of finer
and improved predictive layers will continually improve map accuracy.
It should be noted that the accuracy of modelling varies across
map units. Firstly, a greater degree of accuracy is expected within
Map Units that received a higher sampling density, as physical attributes
determining their distribution are better understood. Secondly,
the accuracy of spatial data layers determines model accuracy. Map
units relying on features such as soil types, which are not mapped
accurately, should expect a greater degree of inaccuracy ie. Shale
lenses are less likely to be mapped as accurately as others.
4.6.1 Accuracy of Identifying Extant Vegetation
The aerial photos used to create the extant vegetation layer vary
in age from 1970 to 1998 (refer to Table 2.6), therefore the level
of accuracy in identifying extant vegetation across the map is inconsistent.
This can be observed in Lake Macquarie where photos of 1:4000 scale
were used to define wetlands, however this degree of definition was
not available elsewhere. Where older photos have been used a higher
degree of inaccuracy is expected due largely to a higher probability
of land clearing since the year of the photo.
The study area would benefit from the application of consistent
criteria to describe extant vegetation. This can be achieved at
a broad scale using satellite image interpretation or aerial photograph
interpretation. Recent mapping by NPWS (2000) provides a guide to
the criteria that may be employed for regional API mapping.
Extant vegetation is highly variable in its condition. The condition
of modelled map units has not been described here. As exotic species
were excluded from the dataset no inferences have been drawn on
the relative condition of remnants in the study area. It is anticipated
that this will form an additional component within the REMS biodiversity
program.
|
