A description of two NSW state-wide vegetation maps and comments on their relevance to biodiversity assessment
Dr John Benson, Ecologist and member of National Parks Association of NSW
The recent comment on biodiversity decline in New South Wales canvassed that unreliable vegetation mapping could compound deficiencies in recently enacted biodiversity regulations. After the change of government in New South Wales in 2011, it appears a decision was made for NSW to adopt maps in biodiversity regulations replacing previous on-ground site assessment and property planning. This transition to maps raises challenges because unlike a google street map, vegetation maps contain errors leading to significant uncertainty in site-based land-use decisions.
Context
In 2016 the Biodiversity Conservation Act (Biodiversity Act) and Local Land Services Amendment Act (LLS Act) were passed by the New South Wales’ Parliament signalling significant changes to the way agricultural land use is managed in NSW. This followed an expert panel review of existing laws. Some would suggest an error was made by the panel when it agreed to the removal of the “maintain or improve test” enshrined in previous laws. This opened the floodgates for the compilation of loose regulations that commenced operation on 25 August 2017. There have been numerous news articles and submissions pointing out the deficiencies in these regulations, particularly the self-assessable, loosely-framed farm and equity codes in the LLS Act. These appear to facilitate land clearing even in over-cleared landscapes with threatened habitats and species. The codes allow clearing of up to 625 hectares per property over three years. Furthermore, although clearing is supposed to be restricted in areas with less than 10% of native vegetation remaining, exemptions provide ways around this.
The LLS Act self-assessment and Biodiversity Act development application decisions will either be based on or heavily rely on vegetation maps. Two different forms of vegetation maps are being produced across New South Wales and this can be confusing for both expert and lay persons. The two maps are described in the following.
NSW Native Vegetation Regulatory Map (NVRM)
Under previous native vegetation laws property land use change was assessed via a Property Vegetation Plan (PVP) developed in conjunction with Catchment Management Authority staff. The PVP highlighted important vegetation to retain and trade-offs for minor clearing were allowed so long as there was no net loss.
In contrast, the new regulations will depend largely on the use of a NSW Native Vegetation Regulatory Map (NVRM) produced over the last few years. Landholders are expected to interpret this map when they self-assess their property. The map reliability is therefore critical. Essentially, the NVRM divides NSW into two main categories: native vegetation and non-native vegetation. Certain sensitive lands are included to expand the “native vegetation” category. These include RAMSAR internationally recognised wetlands, riparian vegetation, and listed critically endangered ecological communities. Endangered and vulnerable ecological communities are not included, and such exclusion could move these communities closer to collapse. Sensitive land also includes steep and erodible areas, high conservation value native grasslands and certain areas of koala habitat (based on local government koala plans). A State-wide Koala Critical Habitat Map is under development but like any map it will suffer from inaccuracies.
Land not mapped as native vegetation or of sensitive character is not subject to biodiversity assessment. This includes isolated farm trees that are important for species of birds, arboreal mammals and invertebrates – some of which assist agriculture. The only hope for these often-old growth farm trees is that landholders voluntarily maintain them.
As with any map, the NVRM will contain significant errors. Data on some of the mapped sensitivity features are poor and there are fuzzy boundaries between native and non-native vegetation, especially defining high conservation grasslands and naturally sparse woody tree cover such as open woodlands on inland floodplains. Thankfully, however, in an age where auto-computer modelling is all-pervasive, even over areas where data are lacking, a sensible decision was made to produce the NVRM through human expert visual interpretation of satellite and airborne (flown) imagery. There is, therefore, hope that it will contain reasonable reliability, especially if it is regularly updated with better data. Due to the current map-based regulations, the NVRM is vital to biodiversity assessments and compliance but because the map is currently (as of January 2018) in “transition phase” there is effectively little regulation at present.
NSW State Vegetation Type Map (SVTM)
The second form of mapping in New South Wales is a State Vegetation Type Map (SVTM). This aims to map greater than 1,000 classified plant community types (PCTs) which represent a fine thematic level of vegetation classification. The science literature on vegetation classification, which forms a basis for type mapping, is too large to summarise here. PCTs can be classified subjectively or quantitatively depending on the quality and coverage of flora plot data. These plot data are samples of plant species often assigned abundance ratings suitable for statistical analysis that yield groupings of similar flora plots. Few of Australia’s 89 bioregions (18 in NSW) contain sufficient quantitative plot data to allow for a fully objective approach to vegetation type classification and mapping. Only a few coastal bioregions in NSW are sampled well enough to be subjected to exploratory numerical classification or model vegetation mapping, but gaps in data still demand expert supervision and amendment.
There are more than 800 existing vegetation type maps across New South Wales produced over 50 years for different purposes, such as documenting vegetation types as biodiversity surrogates, or producing fire plans in national parks, state forests or in local government areas. Some are highly accurate and map State vegetation type classifications, others are of poor reliability and superseded. Up to the end of the 20th Century, most vegetation maps were produced through stereoscopic interpretation (API) of wet film, black and white or colour, aerial photographs. Recently, with the capture of high resolution airborne digital imagery over most of New South Wales, it is possible to apply digital API in 3-dimensional space at fine scales to produce reliable vegetation type maps, along with information on vegetation condition. Importantly, 3-D digital API can map poorly-sampled vegetation types, including numerous plant communities west of the Great Dividing Range. In contrast, the use of computer modelling to produce vegetation maps leaves out poorly plot-sampled but often distinct plant communities. This contravenes the precautionary principle in nature conservation and is a reductionist approach to biodiversity documentation and assessment. It completely ignores the role of expert knowledge and qualitative information.
A key concern about the SVTM Map is its unreliability, as revealed in independent peer-reviewed published and other field checked validations. The SVTM map appears to be less than 20% accurate in depicting the correct PCT at the site or property scale. Why? Because to develop this map quickly an experimental method was selected that involves using pattern recognition software to automatically map any patterns, supplemented by modelling from often limited flora plot data, to label the mapped areas (map polygons). While these methods are sold as “repeatable evidence-based science”, for modelling to be effective it probably requires a great deal more observational data than are available. And there are major limitations in using pattern recognition software for mapping vegetation types because it confuses disturbance patterns (differential grazing, logging, floods, fire, shadow etc) with species assemblages. It also struggles to separate different plant species assemblages that contain a similar canopy structure.
If the SVTM is advanced as a keystone data layer for environmental planning, site assessment, use in prioritisation of properties for allocating NSW Biodiversity Funds, or for calculating offsets for developments, there could be confusion and disappointment for map-users such as local councils, consultants, government agencies and landholders. The map’s unreliability could be taken advantage of by those favouring development – including at locations containing threatened ecological communities that are not mapped correctly. The expert 3-D digital API method mentioned above is a viable alternative means for producing more reliable State or regional vegetation maps – including for regions currently lacking quantitative flora plot data.
Conclusion
The application of NSW biodiversity regulatory codes, compliance or environmental planning could be compounded by inaccurate vegetation maps, whether these are inscribed in regulations such as the NVRM or used in supplementary ways such as the SVTM. Any maps used in regulations or advanced for multi-purpose use should be independently, third-party validated for reliability and should meet exacting standards applicable to land-use decisions.
Concerning the SVTM, major gaps in flora sample sites across New South Wales limits over-reliance on quantitative vegetation type classification and associated auto-modelled vegetation mapping.
The NVRM should be subject to regular revision aimed at eliminating errors to the maximum practical degree. Future reviews of the New South Wales biodiversity regulations should reduce dependence on vegetation maps per se and re-emphasise on-ground property planning checks of vegetation classification and condition. This should be done under a “maintain or improve” scenario. Only then can we expect to stop and reverse the decline of native vegetation and fauna habitat across the largest tenure in the state, privately managed lands.
You must log in to post a comment.