The biomes of Western Australia: a vegetation-based approach using the zonality/ azonality conceptual framework

Vegetation patterns of are the result of environmental drivers operating across a range of ecological and evolutionary scale. Those vegetation patterns at global and continental scales underpin the existence of large-scale functional entities called biomes. Traditionally, terrestrial biomes are recognised and distinguished by way of physiognomic traits thought to reflect the action of selective filters such as climate or disturbance (fire, grazing). This approach is well suited to recognising zonal biomes; it fails, however, to account for the role of other drivers such as soil and hydrology, potentially leading to over simplified representations of the biome patterns in ecologically complex regions such as Western Australia. It is, therefore, of vital importance to consider these patterns and processes and recognise the azonal biomes as well. Using a previously published physiognomic map of the vegetation of Western Australia, we adopted a bottom-up approach to translate the vegetation classification categories into a new biome classification of Western Australia that would more accurately account for the zonal and azonal structures recorded across the landscape. Five continental zonal biomes (Australian Eucalyptus Savanna, Australian Hummock Grassland, Mulga Shrubland, Australian Temperate Woodland, Australian Oceanic Temperate Forest) and one relict biome (Australian Vine Thicket) were identified. We further distinguished 34 types of azonal biomes, including three new ones: argillobiome, duplo-pedobiome and metallobiome. Predictive modelling (using Classification and Regression Trees) was used to define the potential extent of zonal biomes, with the model returning an overall accuracy of 90%. This study shows that a bottom-up approach based on vegetation maps can be used as an effective approximation for defining a robust classification of biome patterns. The predicted zonal biomes correspond well with existing traditional biogeographic classifications, yet some residual climatic discordance warrants further study.

Automated summary

Vegetation patterns are influenced by environmental drivers operating at different scales, forming large functional entities known as biomes. Traditional biome recognition is based on physiognomic traits reflecting selective filters such as climate or disturbance, but it may oversimplify biome patterns by not considering other drivers. Predictive modelling based on vegetation maps can effectively define robust biome patterns, aligning well with existing biogeographic classifications but with some residual climatic discordance.