Wed, Aug 17, 2022: 10:15 AM-10:30 AM
514B
Background/Question/MethodsThe Global Biodiversity Framework (GBF) of the Convention on Biological Diversity has proposed ambitious goals to stop or reverse biodiversity declines, with explicit milestones for 2030. Yet, few people have explicitly evaluated whether we would even be able to assess whether policies result in improved biodiversity trends (i.e., a biodiversity power-analysis). We ask three questions: Would we be able to detect improvements in biodiversity trends, if policies were effective? Would an alternative framework for assessment be beneficial? How many populations need to be sampled to reach a given level of power? We use the Living Planet Database, and use the outcome of Bayesian Hierarchical models, to derive estimates of power and the number of populations needed to be sampled. We focus on 10 geographic/taxonomic systems across the world, where the mean estimates from the posterior distribution suggest catastrophic declines.
Results/ConclusionsOur results suggest that even where mean estimates suggest catastrophic declines, we will often fail to detect improvements with high (95%) certainty, even if new policies reduce net declines to zero, and even if thousands of populations were sampled post-policy. This is because the ability to detect change is limited by the current uncertainty, prior to policy action. We suggest an alternative risk framework for understanding biodiversity change, and we 1) we quantify the sampling (number of populations) needed to improve our certainty of current trends to 95% (e.g., at 200 populations, power begins to level off, except for freshwater Neotropical fish, where power remained < 50% even with 500 populations sampled), 2) we estimate power at different levels of risk (e.g., accepting 70% certainty, we could obtain 75% power for 7 of the 10 systems, with 200 populations sampled post-policy), and also 3) we consider sampling required to detect improvements against a fixed threshold. Our results highlight challenges and potential solutions to biodiversity goals and their assessment.
Results/ConclusionsOur results suggest that even where mean estimates suggest catastrophic declines, we will often fail to detect improvements with high (95%) certainty, even if new policies reduce net declines to zero, and even if thousands of populations were sampled post-policy. This is because the ability to detect change is limited by the current uncertainty, prior to policy action. We suggest an alternative risk framework for understanding biodiversity change, and we 1) we quantify the sampling (number of populations) needed to improve our certainty of current trends to 95% (e.g., at 200 populations, power begins to level off, except for freshwater Neotropical fish, where power remained < 50% even with 500 populations sampled), 2) we estimate power at different levels of risk (e.g., accepting 70% certainty, we could obtain 75% power for 7 of the 10 systems, with 200 populations sampled post-policy), and also 3) we consider sampling required to detect improvements against a fixed threshold. Our results highlight challenges and potential solutions to biodiversity goals and their assessment.