Graeme Lorimer details the workings of NaturePrint and lists some concerning and unexplained features.
AMONG the many types of information available to you from the Victorian Department of Environment, Land, Water and Planning’s website is a map of Victoria called ‘NaturePrint – Strategic Natural Values’, which I’ll contract to NaturePrint. It uses colour-coding to show the importance of flora and fauna at any location, with a resolution of 75 m. It is used by land managers, policy makers and granting bodies to help them decide where to direct their efforts. One example is given in the article on page 3. You might use it to search for the highest value examples of nature in your district, or to see just how well your favourite spots rate.
To access NaturePrint, just Google ‘Biodiversity Interactive Map’ and you’ll soon have a rather bare map of Victoria on your screen. Drag your mouse diagonally across an area of interest to zoom in. In the menu of ‘map layers’ on the right of the map, click the little box next to the words, ‘NaturePrint v2.0 Strategic Natural Values’ (circled on the screenshot next page) and you’ll see the NaturePrint map. Click the words themselves to see a legend for the colour-coding. Use the tools at the top of the map to zoom and pan around Victoria.
You might be surprised at what you see; e.g. that most of the open space and golf courses along the Yarra River from Kew to Heidelberg (see the screenshot) have higher conservation value than half of the Grampians (Gariwerd) National Park, as do many residential areas in outer Melbourne and even part of the Calder Park racetrack.
But as with so much of the information you get online, you have to be careful about what you believe and how much you rely on it.
The methodology for NaturePrint is explained in an undated ‘Facts Sheet’ titled ‘NaturePrint v2.0: Elements and integration’ (DSE, c. 2013), which I will now summarise. An initial phase involves condensing the flora and fauna of Victoria, excluding aquatic fauna, into 98 ‘assemblages’ of species that tend to occur together. A computer model was devised to predict which of the assemblages would most likely be a good fit to the non-aquatic flora and fauna of any location in Victoria. This was done for a 75m×75m grid covering the whole of Victoria. Of course, in reality, there are vastly more than 98 distinct combinations of non-aquatic flora and fauna across the 40,000,000+ grid squares, but reducing them to 98 assemblages makes computations manageable. Thus, detailed biological data and expert interpretation are sacrificed in order to obtain complete coverage of the whole state with a uniform methodology.
The same process was applied to 17 assemblages of river fish and 53 surrogates for wetland assemblages.
For each grid square, NaturePrint combines the assemblage data with the following information:
- A measure of ‘landscape connectivity’ obtained by computer analysis of vegetation mapping across Victoria;
- An estimate of the ecological condition of vegetation, from another computer model; and
- Which (if any) of a selected range of rare or threatened species have been reported in each grid square since 1970.
All the foregoing information goes into a software package called ‘Zonation’. (Search online for ‘Zonation Helsinki’.) The model ranks every grid square across Victoria for how important it is for conserving the chosen assemblages and threatened species, mediated by certain policy preferences. The policy preferences are explained on the basis that biodiversity management is ‘cheaper, easier and more likely to succeed’ if it focuses on areas that rate highly in vegetation cover, ecological condition, landscape connectivity and ‘compactness’. This policy preference favours larger tracts of habitat.
Zonation starts by identifying the least important grid square in Victoria and then it works iteratively, with each iteration determining the next most important grid square and placing it on a progressive list. At each iteration, all the grid squares already on the list are ‘eliminated’, i.e. treated as if they cease to have any conservation value. Zonation re-calculates the importance that each remaining grid square would have in this artificially reduced landscape. The least important grid square that abuts an eliminated square is placed on the progressive list and a new iteration begins.
The ‘metadata’ description that DELWP issues with NaturePrint says it ‘identifies areas that contribute most to protecting the full range of biodiversity values, and identifies the relative contribution of all areas to protecting the full range of biodiversity values. The analysis behind the map incorporates the best information available on the species distribution for all Victorian plants and animals including fish and freshwater crayfish, combined with habitat connectivity and recoverability layers. It explicitly considers rare and threatened species and communities.’
To be more precise, NaturePrint:
- Provides just one of a wide range of possible maps for this purpose, depending on such things as the models and model options chosen, the embedded policy preferences, the selection of flora and fauna species and the degree to which the biological information is simplified;
- The only sense in which it covers ‘all Victorian plants and animals’ is that the assemblages are intended to be surrogates for the vastly larger range of species in Victoria;
- It only incorporates a small part of ‘the best information available’; e.g. it considers only a subset of rare or threatened species, it ignores information about the abundance of any species and it takes no account of any evidence that a species no longer occurs at a particular location where it was once recorded; and
- The claim that rare or threatened communities are explicitly considered is hard to reconcile with the description of the methodology given by DSE (c. 2013).
- There are other limitations that should also be borne in mind, such as:
- Zonation was created to determine things like minimum conservation reserve systems or where best to add revegetation to the landscape. In that context, it makes sense for each iteration to treat all the less important grid squares as eliminated. However, such a treatment is less appropriate in the case of NaturePrint;
- NaturePrint only provides a ranking of grid squares, and just like in a league ladder, the rankings say nothing about how much better each one is from the one below;
- A computer model of ‘landscape connectivity’ provides a less reliable guide to the movements of wildlife (and the pollen and seeds they carry) than direct observational evidence and expert interpretation;
- NaturePrint takes no account of differences between the amount of effort and commitment put into nature conservation in each grid square or the trajectory toward improvement or deterioration, which clearly affects the value for conserving flora and fauna;
- There is considerable tension between the bias toward large tracts of habitat and the need to provide all Victorians with opportunities for contact with nature in their daily lives, as promoted by IFFA and recognised in the state government’s new Biodiversity Plan;
- The many assumptions, simplifications and limitations that go into the various levels of modelling have only been disclosed in general terms; and
- No indication has been given of how sensitive the results are to any of the matters above.
There is also an inherent drawback in any system like NaturePrint that substitutes computation for expert interpretation. I’ll use the example of threatened species records to illustrate. When I see a record of a Red-tailed Black- Cockatoo in Croydon South or a Swift Parrot in Carnegie, I interpret it as an example of an animal moving through, not as an indication that the particular location of the observation is special. A quick check with the observer us ually confirms this. On the other hand, a recent record of a Tree Goanna (or Lace Monitor) at a location always counts much more, because I know that this threatened species must be resident there (or nearby) and that the importance of the observation applies not just to the particular location of the observational record but to a substantial area of habitat that is required for such an animal to live. I will usually investigate the extent of that habitat and treat it as all important. By contrast, NaturePrint is quite undiscriminating about threatened species reco rds. It assigns a standard level of importance to the particular grid cell corresponding to each observational record, not to the habitat represented by the record or the importance of that habitat to the plant or animal.
NaturePrint’s various limitations are the costs of obtaining a consistent, objective picture of biodiversity conservation values across the whole of Victoria. It is therefore quite appropriate that the metadata for NaturePrint says, ‘The dataset is to be used for large scale and strategic planning only. It is not suitable for site appraisal and statutory planning.’
Nevertheless, NaturePrint is used for statutory planning. The planning controls over removal of native vegetation rely, to a significant degree, on each grid square’s ‘Strategic Biodiversity Value Score’ (SBS). DEPI (2013) says that SBS has been calculated from the NaturePrint Zonation output by squaring each grid square’s ranking and converting the values to a range from zero to one – effectively, just a re-scaling. However, that claim conflicts with the observation that if you look at NaturePrint and SBS on ‘Biodiversity Information Map’, they have peaks and troughs at different locations from each other.
Whatever the explanation for such discrepancies, it is apparent that significantly different maps can be produced by NaturePrint for essentially the same purpose. This observation highlights why it is so important that no information is available about the reliability of the mapping.
In summary, NaturePrint has some very appealing features. For a decision-maker, it’s very enticing to reduce the complex factors that make up the conservation value of a site into a single score or ranking. Consequently, NaturePrint has attracted a lot of usage for many purposes. However, its shadowy basis hides some concerning, unexplained features that leave me wondering which purposes it can reliably fulfil.
A comment was sought in vain from DELWP.
DEPI (2013), Biodiversity information tools used in Victoria’s native vegetation permitted clearing regulations – Fact sheet. Department of Environment and Primary Industries, Melbourne.
DSE (c. 2013), NaturePrint v2.0: Elements and integration – A technical explanation of the NaturePrint v2.0 analysis. Department of Sustainability and Environment, Melbourne.