Photo of a sugar glider, from 2008 Bush Biodiversity Calendar, courtesy Elizabeth Kakoschke, Bingara. The calendar features native animals and lists threats to their survival.

Efforts to restore wildlife habitat outside national parks need to be supported by more accurate ways of measuring biodiversity, according to a major study by NSW Department of Primary Industries (DPI) researchers.

The study, funded by the NSW Environmental Trust, sought to ‘ground-truth’ four ranking systems under development for assessing the conservation value of vegetation in rural and agricultural landscapes.

The ranking systems are predictive models which aim to rapidly assess such factors as the benefits and losses to biodiversity of changes in land use.

The researchers’ aim was to investigate the accuracy of the ways of predicting vertebrate species richness within forest remnants and eucalypt plantations established as habitat for wildlife on agricultural land.

After comparing the scoring system used in the models with data on the vegetation and fauna recorded from 120 sites in NSW and northern Victoria, the research team concluded the models were poor predictors of biodiversity in agricultural and other altered landscapes.

NSW DPI forest biodiversity research leader, Dr Rod Kavanagh, said while vertebrate species are not the only component of biodiversity, models could be expected to do a better job in predicting occurrences of more familiar species, including birds, mammals, reptiles and frogs, than was the case.

The extent to which ranking systems could predict the richness of all species, including invertebrates and fungi is unknown.

"Accordingly, they may not provide a reliable method of identifying high quality sites for biodiversity and thus poorer quality sites could end up being chosen for protection over higher quality ones."

Dr Kavanagh said the models were slightly better at predicting vertebrate species richness within forest and woodland remnants than within eucalypt plantations, but the differences were not all that great.

"More effort needs to be made to tweak them to better represent vertebrate species richness in plantations and other restored habitats, as well as in remnant vegetation."

The study tested the attributes and rankings from the models against data on vegetation and vertebrate species richness gathered from 120 sites regarded as representative of the vegetation types, patch size and planting ages typically found in agricultural landscapes.

Dr Kavanagh said two important factors - the presence of water and extent of surrounding vegetation cover - were ignored in some of the ranking systems.

"Proximity to water, whether in dams or rivers, is an important habitat requirement for many species," he added.

The study found while there is no need to develop new models, existing ones need work.