The stakes are considerable. The GBMWHA, inscribed by UNESCO in 2000, protects over one million hectares of vegetation celebrated for its Outstanding Universal Values, including ancient Gondwanan plant lineages found nowhere else on Earth. Yet rainforest represents only about one percent of the currently mapped vegetation in the reserve. When vegetation mapping fails to match what grows on the ground, management decisions such as prescribed burning, conservation prioritisation and threat mitigation are built on a distorted picture of where significant biodiversity actually resides. For fire-sensitive rainforest pockets embedded within flammable eucalypt forest, that distortion can be lethal.

The research team set out to validate the regulatory vegetation map units, known as Plant Community Types (PCTs), across the Hawkesbury subsection of the GBMWHA, an area bounded by the Colo, Nepean-Hawkesbury and Grose rivers and the township of Bilpin. The PCT classification is a major numerical undertaking: it was built from more than 90,000 floristic plots spanning 23.5 million hectares of eastern New South Wales, standardised to a modified six-point Braun-Blanquet cover-abundance scale, and embedded within the NSW State Vegetation Type Map (SVTM) that now underpins environmental regulation. Preliminary field survey, however, had hinted at large-scale mapping errors in the region’s remote gullies, where deep sheltered terrain and the moisture-rich Burralow Formation sediments create conditions favourable to rainforest.

To test whether the vegetation in these largely unsurveyed areas could be reliably matched to existing PCTs, the researchers deployed 55 plots across two complementary survey campaigns. Forty-six plots followed the standard NSW government technique for assessing cover and abundance within 20 by 20 metre quadrats, while nine larger circular plots of roughly 700 square metres quantified structural elements such as size class distributions and growth stages, with allometric equations scaling projected foliage area against stem diameter. Sites were stratified along an aridity gradient and split between two topographic positions: ‘Gully’ sites in deep sheltered locations near creeks, and ‘Bench’ sites at more elevated positions at the base of escarpments or on upper slopes. Crucially, the team also recorded bryophytes, the mosses, liverworts and hornworts typically overlooked in vegetation classification, through 15-minute active searches at 26 plots.

The statistical verdict was unambiguous. Using non-metric multidimensional scaling ordinations built on Bray-Curtis dissimilarity, the metric emulating the NSW PCT classification approach, the team found that their field plots occupied vacant ordination space defined by the established PCTs. Permutational multivariate analysis of variance (PERMANOVA) confirmed that the two putative communities differed significantly in floristic composition from all locally mapped PCTs and from all plausible regional rainforest PCTs, with all p-values at or below 0.038 after false discovery rate correction. The results held under both Bray-Curtis and Gower dissimilarity indices, which treat Braun-Blanquet classes differently, demonstrating that the findings were not an artefact of metric choice. Permutational analysis of multivariate dispersion showed the new communities ranked among the tightest groups in ordination space, meaning their distinctiveness was not inflated by internal spread.

The team then ran their data through the NSW government’s Eastern NSW Plot to PCT Assignment Tool, the formal instrument used to meet regulatory obligations. The Tool returned best matches scattered across as many as eight different PCTs for the Gully plots and five for the Bench plots, with no coherent geographic or environmental story. The authors argue that such inconsistency, even when matches fall within the Tool’s statistical threshold of 0.695 distance-to-centroid, indicates a habitat the existing classification does not adequately capture. The 0.695 threshold, they note, represents the 95th percentile of distances for well-sampled PCTs and is designed to test statistical stability, not ecological distinctiveness. Floristically, the new communities sat closest to coastal rainforest types from the Illawarra and Central Coast regions, hundreds of kilometres away, rather than to any locally mapped rainforest.

Bryophyte data added an unexpected layer of discrimination. The 34 bryophyte species recorded separated the Gully and Bench communities in ordination space, with non-overlapping 95 percent confidence intervals around group centroids and a marginally significant PERMANOVA result. The moss Papillaria flavolimbata, which festoons tree branches in Bench rainforest, was never observed doing so in Gully rainforest, while several liverworts and hornworts such as Symphyogyna podophylla and Bazzania adnexa were restricted to creekside Gully sites. The authors suggest that integrating non-vascular plant data into community analysis could become a useful tool for Australian vegetation classification, though they caution that bryophytes varied more within communities than between them, likely reflecting fine-scale responses to moisture, substrate and light.

Spatial mapping revealed the true scale of the discrepancy. Using high-resolution aerial imagery, canopy height modelling and terrain analysis to isolate gullies and benches, then ground-truthing the output across two field campaigns, the team estimated a conservative minimum extent of 2,086 hectares of rainforest in the study area, against just 645 hectares documented in the SVTM. The overlap between the two mapping products was a mere 12 percent, implying the SVTM may carry a false positive rate exceeding 40 percent, which would push the actual increase in rainforest extent closer to seven or eight times the official figure. The team mapped over 40 distinct patches of Bench rainforest ranging from 1 to more than 30 hectares, and proposed two new vegetation types for inclusion in the PCT classification: ‘Hawkesbury Warm Sheltered Gully Rainforest’ and ‘Hawkesbury Subtropical-Dry Bench Rainforest’, intermediate forms bridging warm temperate, subtropical and dry rainforest.

The botanical significance of these communities is striking. Bench rainforest ranked highest in mean species richness among all regional rainforest PCTs with more than one reference site, and Gully rainforest ranked third. The communities harbour regionally rare species and notable range extensions, including the white cedar relative Dendrocnide photinophylla, apparently restricted to a single location with fewer than fifty living trees more than 100 kilometres from its nearest known population, and several hundred individuals of the Critically Endangered tree Rhodamnia rubescens across at least five populations. The team also recorded species such as Gmelina leichhardtii, Ficus henneana, Diospyros pentamera and Calanthe triplicata, many documented decades ago in botanical surveys of the Wheeny Creek area by ecologist Rosemary Floyd, whose historical records the authors argue were overlooked during the development of contemporary mapping.

The refugial story behind these rainforests is one of the study’s most compelling threads. As Australia drifted north and aridified from the Late Eocene, rainforest contracted under fire regimes favouring sclerophyllous species, persisting in pockets where conditions remained suitable. The Hawkesbury landscape, with mountain peaks receiving 1,200 to 1,300 millimetres of annual precipitation, deeply incised gorges and the semi-impermeable clay layers of the Burralow Formation acting as aquitards, appears to have buffered these assemblages from long-term climate drying and fire. The authors speculate that under past climates, more extensive rainforest links once existed between the Illawarra, Hawkesbury and Central Coast regions, dispersed along the hydrological conduit of the Hawkesbury-Nepean River and the movements of fruit-eating fauna.

That refugial buffering is now under threat. By 2090, the study region is predicted to warm by up to 3.4 degrees Celsius, lose up to 10 percent of annual rainfall and suffer more severe fire weather days. The Gully and Bench rainforests cover less than 1,000 hectares across four small subcatchments, making them acutely vulnerable to wildfires like those of 2019-2020, which affected hundreds of thousands of hectares. The researchers observed evidence that some rainforest patches within historical fire perimeters had been transformed into perpetual vine thickets with few large trees. Myrtle rust, the introduced pathogen Austropuccinia psidii, was consistently observed on Rhodamnia rubescens and other species, while invasive plants including Ochna serrulata and Lantana camara compete with threatened populations. Illegal four-wheel drive and mountain bike activity in the ridgetops of the Burralow, Wheeny and Gospers Creek catchments risks igniting unplanned fires and transporting weeds and pathogens into undisturbed areas.

The authors urge rapid revision of the PCT mapping and targeted measures to protect the remaining patches, warning that fires and urban development pressures will likely threaten the integrity of these rainforests in future. Prescribed burning is implemented across the region to reduce wildfire risk to lives and property, and rapid urban expansion is converting former rural lands to housing at the bushland interface, raising the probability of unplanned ignitions. Because 65 percent of the surveyed rainforest sites are officially mapped as sclerophyll forest, there is a material risk of off-target impacts from fire management activities within the World Heritage Area. The new community descriptions, diagnostic species lists and revised mapping provide land managers with the tools to plan burns around rainforest boundaries and to conserve what the authors describe as remnants that hold the stories of plant evolution over geological time, contributing more to the Outstanding Universal Values of the World Heritage Area than has previously been ascribed.

Subject of Research: Identification of distinctive new rainforest communities in the Greater Blue Mountains World Heritage Area, Australia

Article Title: Identification of distinctive new rainforest communities in the Greater Blue Mountains World Heritage Area, Australia

Article References: Bendall, E. R., Wong, V., Bell, S. A. J., & Gallagher, R. V. (2026). Identification of distinctive new rainforest communities in the Greater Blue Mountains World Heritage Area, Australia. Vegetation Classification and Survey, 7, 103-125. https://doi.org/10.3897/vcs.186732

Image Credits: AI Generated

DOI: 10.3897/VCS.186732

Keywords: Identification, distinctive, rainforest, communities, Greater, Blue, Mountains, World, Heritage, Area, Australia, scientific research