Environmental, habitat and hydrological variables of floodplain wetlands in the Kiewa, Ovens and Murray Riverina basin

Full description

Floodplain wetlands are among the world's most threatened ecosystems, yet the mechanisms driving their ecological collapse remain poorly understood—hampering effective conservation. Our study addresses this critical gap through an integrated analysis of floodplain wetlands in Australia's Murray-Darling Basin, combining three decades of historical records (1986–2022) with comprehensive contemporary surveys of 73 wetlands using fyke nets, seine nets, and eDNA. Among the key to our research we can highlight:

• Shifted baselines and transformed communities – Contemporary fish assemblages (22 species) represent a depauperate subset of historical communities (28 species), with invasive species now dominating occurrence patterns (common carp present in 93% of wetlands versus 47% historically). Notably, several historically common wetland specialists were undetected, while large-bodied riverine natives now appear in wetlands—a fundamental reconfiguration of community structure.

• Bimodal species strategies – Eight native species showed strong environmental specialization, making them inherently vulnerable to habitat degradation, while six native and four invasive species exhibited stronger spatial signatures, suggesting dispersal-mediated distributions that may facilitate range expansion under changing conditions.

• Persistent native networks – Despite pervasive invasion, residual co-occurrence analysis revealed strong positive associations among native species (e.g., Australian smelt with carp gudgeons, r = 0.926), indicating core native modules that persist within invaded wetlands and could form recovery nuclei for restoration.

• The rare species paradox – Model validation exposed a critical ethical asymmetry: predictive accuracy was high for ubiquitous invasives but failed for 45% of species—predominantly threatened, data-poor natives. This reveals that current analytical tools are optimized for diagnosing problems (invasive spread) but inadequate for guiding solutions (rare native recovery), challenging conventional model validation approaches.

• Decision-focused conservation – We argue for a paradigm shift from purely statistical validation toward decision-focused metrics that prioritize actionable insights—such as identifying key refuge wetlands—even when statistical scores are modest. Our three-model approach (habitat, water quality, hydrology) provides a transferable template for linking specific management actions to biodiversity outcomes in floodplain wetlands globally.