Data for improving SDM transferability of global freshwater invaders with non-conservative niches

Species distribution models (SDMs) are powerful tools for addressing
global ecological challenges. Many species, particularly invaders, exhibit
dynamic niches in novel environments, which reduces SDM transferability,
raising concerns about prediction accuracy. Improving model
transferability and validating predictions with robust methods and
independent occurrence data are essential to address these issues.
However, the extent of transferability improvement required to achieve
reliable predictions remains unclear. This work aims to quantitatively
assess how niche dynamics influence model transferability and prediction
accuracy, using freshwater invasive golden, zebra, and quagga mussels as a
case study. We categorized mussel occurrence data based on invasion
histories and quantified niche dynamics using bioclimatic variables. We
analyzed the improvements in SDM transferability by optimizing occurrence
data and various environmental variables. Finally, we established the
relationship between niche dynamics and transferability-related indices,
identifying their threshold ranges necessary for accurate distribution
predictions. Zebra mussels maintained high niche stability during
invasions, while golden and quagga mussels exhibited niche expansion and
unfilling. Incorporating invasive-range data and diverse environmental
variables improved model transferability and prediction accuracy for zebra
mussels but was less effective for the other two species. We found
positive correlations between SDM transferability and niche stability and
negative correlations with niche expansion and unfilling. We also
identified potential minimum thresholds for accurate mussel distribution
predictions: a Continuous Boyce Index of 0.900, a Linear Correlation of
Suitable Habitat Area of 0.800, and a Schoener’s D of 0.200. For
freshwater invaders with non-conservative niches, accurate predictions can
be achieved by improving transferability when niche dynamics meet specific
thresholds. Our findings therefore offer strategies for refining SDMs to
improve prediction accuracy for species with dynamic niches and provide
valuable insights for managing freshwater ecosystems in the context of
global change.