Ensemble distribution models in conservation prioritization: from consensus predictions to consensus reserve...

Area of Science:

• Ecology
• Conservation Biology
• Biodiversity Informatics

Background:

• Conservation planning increasingly uses species distribution models (SDMs) and ensemble forecasting.
• The impact of different ensemble prediction summarization methods on conservation outcomes remains understudied.
• Current methods often summarize ensemble predictions before conservation prioritization.

Purpose of the Study:

• To evaluate how different methods of summarizing ensemble SDM predictions affect conservation planning outcomes.
• To compare conventional consensus methods (applied before prioritization) with a novel post-selection consensus method.

Main Methods:

• Utilized an ensemble of predicted distributions for 146 Western Palaearctic bird species.
• Applied four consensus methods and variability-discounted distributions as inputs for spatial conservation prioritization.
• Developed and tested a novel method involving 100 resampled datasets from the ensemble, followed by prioritization and averaging of ranks.

Main Results:

• Networks derived from predicted distributions were more representative of rare species than random networks.
• Different ensemble summarization methods yielded varying levels of representativeness in reserve networks.
• The novel post-selection consensus method demonstrated superior representation of rare species compared to pre-selection methods.

Conclusions:

• Retaining information on predicted distribution variability throughout prioritization yields better conservation outcomes.
• Summarizing ensemble predictions after reserve selection is more effective than pre-selection summarization.
• Emphasizes the importance of considering model-based uncertainty in conservation planning using predicted distribution data.