Conclusions
Ecosystem spatial pattern modelling – sometime called ecosystem distribution modelling – is in early stages of maturity (Naas et al 2024), and there are few evaluations of suitable approaches. This is expected to change. While species distribution models greatly outnumber models for other levels of biodiversity (Zurell et al. 2022), interest in community and ecosystem spatial modelling is increasing rapidly (Geary et al. 2020, Simpkins et al. 2022). Much of the immediate demand has come from conservation scientists tasked with developing models to identify ecosystems for protection and restoration. This has prompted development of standardized classifications and mapping tools, including those for the red list of ecosystems (Keith et al. 2022, Murray et al. 2022). Yet very few countries presently have suitable models for mapping ecosystems (Xiao et al. 2024) and available models emphasize vegetation proxies. Our framework offers an alternative approach. Our joint emphasis on biotic and abiotic response variables from field surveys contrasts with other approaches. Biotic and abiotic properties are the building blocks of ecosystems (Holling 1992). Predicting regular shifts in groups of these properties is essential for resolving and understanding ecosystem pattern (Levin 1998, Levin et al. 2001). Such patterns express ecosystem assembly outcomes, including the factors shaping lower levels of ecological organization (Keith et al. 2022). Overall, the framework we present is intended to help advance general understanding of ecosystem patterns, which remain poorly understood (Loreau 2020).