This layer reflects the probability of finding a tree species that a koala is known to use for food or shelter. The statewide Koala Habitat Information Base (KHIB) has been developed as part of the NSW Koala Strategy. It delivers the best available state-wide spatial data on koala habitat, likelihood, koala preferred trees and koala sightings for NSW. It will be an important resource to assist government agencies, local councils and private land holders with koala conservation decisions. The Koala Tree Index v1.1 is one dataset under the KHIB. Tree indices were developed for each of the nine Koala Modelling Regions and mosaiced together into a statewide raster. Boosted Regression Tree (BRT) individual tree species distribution models were developed for important food and shelter species, with each KMR possessing a characteristic set of species whose combined distributions added to create index. The choice which species to include in each region was determined by a process of expert elicitation. The statewide product is available for download below as a zipped 5m tif image readable in any spatial software package. An ArcGIS mxd is also supplied for suggested symbology. All Koala Habitat Information Base datasets are available for download at the links below under 'Dataset relationship'. For further information on the data layers and their development, please see the Koala Habitat Information Base Technical Guide.

The statewide koala baseline survey is a standalone snapshot of the occurrence, distribution and relative abundance of koalas in NSW in 2025. The baseline survey covers the plausible range of koalas in NSW and over 1,000 sites were sampled across land tenures using heat-sensing drones and passive acoustic recorders. The survey also provides a baseline population estimate from which to evaluate koala population trends over time. This dataset comprises state-wide occupancy and abundance models derived from acoustic and drone data and environmental covariates as described in the NSW koala baseline technical report, 2025

The Koala Likelihood Map (KLM) predicts the likelihood of finding a koala relative to other arboreal mammals across a 10-km grid covering NSW. It is built using existing arboreal mammal records from the past 20 years (currently 1999 to 2019) and represents the likelihood of koalas as the proportion of all records within a grid cell that are koalas. The records of other arboreal mammals provide a measure of survey effort independent of koalas and allow identification of areas where other arboreal mammals have been recorded, but not koalas. The map also includes a measure of the confidence in the koala likelihood estimate. This enables deficiencies in the data to be highlighted, and recommendations to be made for areas requiring further survey. The KLM is a useful tool that can be used to inform a range of koala conservation and management issues, however it is not static and should be updated regularly as new data become available. The KLM was first developed in 2014 for use in private native forestry regulation, on behalf of the NSW Environment Protection Authority. An updated and refined version of the map (NSW Koala Baseline Likelihood Map 2016) was produced in 2016 and has been used to inform provisions for koala protection under the Coastal Integrated Forestry Operations Approvals and is planned to inform the future review of the Private Native Forestry Code of Practice. This latest version of the KLM (v2.0 August 2019) includes new data from BioNet and Spot Assessment Technique (SAT) survey databases, as well as SAT data from a targeted state-wide field survey program. The KLM v2.0 (August 2019) is delivered under the NSW Koala Strategy's Koala Habitat Information Base. This comprises several layers of spatial information, including: Koala Habitat Suitability Model (KHSM) – the probability of finding koala habitat at any location; Koala Tree Suitability Index (KTSI) – the probability of finding a tree species that koalas are known to use for food or shelter; Koala Likelihood Map (KLM) including a confidence layer – predicts the likelihood of finding a koala at a location; Areas of Regional Koala Significance (ARKS) – identifies key koala populations and management areas with potential for long-term viability as well as priority threats to key koala populations; Native vegetation of NSW – this is a high-resolution map of native tree cover and water bodies; and all koala sightings recorded in NSW Bionet. All Koala Habitat Information Base (KHIB) datasets are available for download below under 'Dataset Relationship'.

The statewide koala baseline survey is a standalone snapshot of the occurrence, distribution and relative abundance of koalas in NSW in 2025. The baseline survey covers the plausible range of koalas in NSW and over 1,000 sites were sampled across land tenures using heat-sensing drones and passive acoustic recorders. The survey also provides a baseline population estimate from which to evaluate koala population trends over time. This dataset comprises point data of koala presence and absence derived from raw acoustic data, captured from passive acoustic recorders in koala breeding season.

This dataset contains map data relevant to the Site Investigation Area for Koala Plans of Management Map, as referenced in Chapter 4 of State Environmental Planning Policy (Biodiversity and Conservation) 2021 (Biodiversity and Conservation SEPP). The map is referenced in Chapter 4 of the Biodiversity and Conservation SEPP as the ‘State Environmental Planning Policy (Koala Habitat Protection) 2021—Site Investigation Area for Koala Plans of Management Map’. However, the 2021 SEPP has been repealed and replaced by Chapter 4 of the Biodiversity and Conservation SEPP. The description and method for this layer is detailed below. Site Investigation Area for Koala Plans of Management Map This map identifies areas that are likely to have koala use trees, as well as environmental features such as soil type, topography and climate suitable for sustaining koalas. This map was developed from the Koala Habitat Information Base. The map does not show core koala habitat, and is only relevant as an investigation area, when councils resolve to prepare a Koala Plan of Management. Outside of this process, the map is not relevant and is not used. The map only captures land in the LGAs listed in schedule 21 of the SEPP. The map also excludes the lands that the SEPP does not apply to, such as national parks and state forests. The Site Investigation Area Map is not publicly available. To access this map, please contact your local council or the Department of Planning, Housing and Infrastructure. Contact data.broker@environment.nsw.gov.au for more information.

Data collected on location and tree use of Koalas in Kanangra Body National Park

The spatial data maps koala populations and habitat for three study areas in the Midcoast LGA of New South Wales. The three study areas, Kundle, Khappinghat and Tea Gardens are geographically separate, and each are approximately 16, 000 hectares and square or rectangular in shape. The data maps both ‘Likely koala habitat’ and Occupied koala habitat’ on all lands except NPWS Estate and Forestry Corporation of NSW estate. ‘Likely koala habitat’ occurs where there is greater than 15% dominance of preferred koala food trees. ‘Occupied koala habitat’ is a spatial subset of ‘Likely koala habitat’ where koala populations are currently viable based upon known presence and their generational persistence over time. The maps were produced under the NSW Government's Koala Strategy in collaboration with Midcoast Council to inform Council’s 2024 MidCoast Koala Conservation Strategy.

The Maxent modeling algorithm was used to build the species distribution model at 270 m spatial resolution using species occurrence points and environmental layers as predictors (Phillips et al. 2006). Species occurrence points were primarily obtained from CNDDB (California Natural Diversity Database) and other CDFW sources, GBIF (Global Biodiversity Information Facility), PRBO (Point Blue Conservation Science) and Arctos museum databases. Vegetation, distance to water, elevation, and bioclimatic variables (Franklin et al. 2013) were used as predictor variables. The models were run at 270 m spatial resolution with five replications using cross-validation as a method of sample evaluation. Cross-validation involved the partitioning of the sample data into n subsets, fitting the models to n-1subsets, and testing the model on the one subset not used in fitting the model. Initial model runs showed that our models converged around 2,000 iterations and for this reason we ran all models with 2,500 maximum iterations. Maxent was implemented in R using the ‘dismo''package (Hijmans et al. 2011). Model evaluation was carried out using the ‘PresenceAbsence''package in R (Freeman and Moisen 2008). We used AUC as a metric to evaluate model performance. The package also computes threshold values using several accuracy metrics to translate predicted probability maps into binary suitable and unsuitable habitats. We selected the MeanProb, a threshold set based on the mean predicted probability of species occurrences. The output from Maxent are grid datasets in a multiband ‘tif''format with one band for each replication. We averaged the five replicated maps and created a mean grid for each species. The grid was then symbolized to represent low (threshold-50), medium (50-75) and high (75-100) habitat suitability, with pixel values that are below the threshold excluded. Models were reviewed by CDFW species experts; please review the use limitations.For more information see the project report at [https://nrm.dfg.ca.gov/FileHandler.ashx?DocumentID=85358].

This is a collection of 171 habitat quality models for fauna species that were mapped across forest areas in the Upper North East (UNE) and the Lower North East (LNE) NSW during the Comprehensive Regional Assessment (CRA) in 1999. They are 100m grids stored in MGA Zone 56 projection. 34 models are mapped on public tenure and 137 over all tenure. The ‘public land’ fauna models were those that modelled fairly on public land, some using Presence-Absence modelling, and were restricted to public land because the systematic surveys were carried out there (eg. primarily NEFBS, State Forest EIS, CRA). Some all-tenure (Presence-only) models were most likely have been cut to public land if it was considered that they modelled better there. In this case there would be two versions of the same model, but only one was used in the CRA. It was decided that the flora models would not be published due to their poor quality and their need for updating with better records in the time since. Note that a revised edition of approximately a third of the models were produced in 2008: https://iar.environment.nsw.gov.au/dataset/revised-northern-cra-habitat-models-2008 The original models were produced as part of a Comprehensive Regional Assessment (CRA) for the Regional Forest Assessment (RFA) process. The specific objective of these models was to identify core areas of forest capable of sustaining viable populations of priority species. Habitat quality models were derived using known distributions of species combined with abiotic, biotic, terrain, habitat and geographic layers within a GIS. These known species-habitat relationships were then used to model predicted distributions and thus areas of significant habitat for the species of concern. Flora and fauna experts were used to validate the models and define areas of high-quality habitat for each species. The models are either mapped across All Tenure (at) or Public Tenure (pt). Each species model is named with the Catalogue of Australian Vertebrates (CAVS) code. Fauna models were developed using logistic regression models (generalised additive models) of species presence and absence to mapped environmental features. Where statistical models were judged by the expert panel to be inadequate, qualitative or expert models were derived. Additionally, fauna experts were used to identify habitat quality. Probability levels were used where appropriate to define high (class 1), intermediate (class 2), and marginal (class 3) habitat. Flora models were produced using a combination of GAM inference of species sightings with mapped environmental features and a boolean overlay of selected environmental features along with expert review. Expert judgement was employed to categorise flora habitat into two classes of potential habitat: Occupied habitat (class 1) that shows validated point localities or population areas with a surrounding buffer to account for local seed bank or regeneration. High quality habitat (class 2) which is the rest of the model constructed using the boolean overlay of environmental layers. See Table 3A (pg.33-38) in report for full a breakdown of species models, methods used and assessment of model confidence. The report notes that models were not validated due to time constraints and that results should be viewed as a "minimum estimate of high-quality habitat for the purposes of the CRA." The official report, Modelling areas of habitat significance for vertebrate fauna and vascular flora in north-east NSW 1999, expands on the methodology and outputs. The report is stored for internal access under P:\Corporate\Products\Biodiversity\Habitat\CRA_Northern MODELLING AREAS OF HABITAT SIGNIFICANCE FOR VERTEBRATE FAUNA AND VASCULAR FLORA IN NORTH EAST NSW A project undertaken as part of the NSW Comprehensive Regional Assessments, April 1999 Project number NA 23/EH The fauna species modelled are as follows: • 0021 Rose-crowned Fruit-Dove • 0023 Superb Fruit-Dove • 0035 Brush Bronzewing • 0174 Bush