Post date: Jul 10, 2020 6:48:40 PM
Uncertainty in classifying endangered species
In most jurisdictions around the world, endangered species are classified with reference to quantitative or semi-quantitative criteria developed by the International Union for Conservation of Nature (IUCN). The database used for the IUCN Red List (called SIS) includes a module that replicates the algorithm from Akçakaya et al. (2000; 2001). So far, its use for species added to the IUCN Red List has been optional, but Resit Akçakaya has been trying to—and he hopes getting close to—making it required. This would mean that all assessments would have to use the module that incorporates uncertainty of every data point that goes into the assessment. The main issue will be making sure the assessors enter uncertainty in comparable and consistent ways. There is a set of guidelines, developed over ~20 years, that are very well used. But they need a lot more specific information about determining uncertainty bounds for each parameter. There are very different attitudes and preferences over this. Many academics would focus on things like whether to use 90% or 95% confidence intervals for the upper and lower bounds that the algorithm requires and, if so, what about credible intervals, etc. Of course, almost no species has data to calculate such statistics, and most assessors are a lot more practical; some even would rather not think about uncertainty at all. So, there is a lot to do in finding practical yet defensible ways to help people decide what interval to use.
There are two other developments that have uncertainty components that you might be interested in.
1. Inferring extinctions. The algorithms in RAMAS Red List (Akçakaya et al. 2001) are all about the probability that a species will go extinct sometime in the future. A parallel question is the probability that the species is already extinct, which is addressed by Andrew Solow's equations, etc. With Mark Burgman's lead, we developed and implemented a method to calculate this probability (see "When is a species extinct?"). This method is now officially part of the Red List guidelines: https://www.iucnredlist.org/resources/ex-probability. It uses a very simple interval approach. What has been interesting is that so far assessors appear to be more confident (specifying narrower bounds) than what seems prudent. But, unfortunately, as with the other algorithm, it's hard to know for sure.
2. The Green List. Over the last several years, most of my focus has been on the IUCN Green List of Species, now called the IUCN Green Status of Species. Incorporating uncertainty into this system has been challenging. What we came up with (described in this Appendix this paper) tries to balance practical considerations and robustness, but I suspect it will need to be improved over the coming years.
So, uncertainties are not at all neglected; they are a major aspect of these three different but related methods. The difficulties in all of these are fundamentally the same. We want a very simple yet defensible method, and simple and practical ways of deciding on the uncertainty bounds in each case where there is often very little or no data at all. And, then we need to convince thousands of assessors to use them!
Resit would be happy to hear any thoughts and suggestions.
Additional references
Akçakaya, H.R., S. Ferson, M.A. Burgman, D.A. Keith, G.M. Mace, and C.R. Todd. 2012. Commentary: IUCN classifications under uncertainty. Environmental Modelling & Software 38: 119-121
Regan, H.M., H.R. Akçakaya, S. Ferson, K.V. Root, S. Carroll, and L.R. Ginzburg. 2003. Treatments of uncertainty and variability in ecological risk assessment of single-species populations. Human and Ecological Risk Assessment 9: 889-906.
Akçakaya, H.R., W.T. Root and S. Ferson. 2001. RAMAS Red List: Threatened Species Classifications Under Uncertainty. Applied Biomathematics, Setauket, New York
Akçakaya, H.R., S. Ferson, M. Burgman, D. Keith, G. Mace and C. Todd. 2000. Making consistent IUCN classifications under uncertainty. Conservation Biology 14: 1001-1013
Colyvan, M., M. Burgman, C. Todd, H.R. Akçakaya and C. Boek. 1999. The treatment of uncertainty and the structure of the IUCN categories. Biological Conservation 89:245-249.
Solow, A.R. 1993. Inferring extinction from sighting data. Ecology 74: 962-964
Solow, A., and T. Helser. 2000. Detecting extinction in sighting data. Quantitative Methods for Conservation Biology, edited by S. Ferson and M. Burgman, Springer.
Solow, A.R. 2016. On Bayesian inference about extinction. Proceedings of the National Academy of Science (US) 113 (9) E1132.