I went to quite an interesting talk this afternoon by Dr Julian Caley from the Australian Institute of Marine Science. Julian’s visiting us at QUT for a few weeks and presented some interesting work he and his colleagues have done on estimating marine biodiversity (they estimate somewhere between 500,000 and 2,000,000 species) and the statistical and ecological issues faced.
At least one QUT Bayesian Statistics PhD graduate has ended up at AIMS, Dr Rebecca O’Leary. Some of the work members of BRAG have done with AIMS involve expert elicitation around the issues of fish population but Julian said today that fish and coral populations, commonly used as a proxy for biodiversity, only account for about 2% of the total species in marine environments. Some of the expert elicitation involves not only quantifying the named species and the discovered but unnamed species but quantifying the prevalence of undiscovered species. Apparently the place to be working isn’t in fish and coral species (which are already well known and quite easy to study) but in nematodes and isopods which are expected to have many undiscovered species.
Julian talked about some very interesting ways of estimating the prevalence of unobserved species by looking at which Orders are present (higher level taxa) that cover the species you’re interested in, which other species in the same taxa are prevalent (cross-taxa) or which species are present in the genus you’re interested in (subset taxa). Not having a huge amount of experience in classification and regression trees I wasn’t entirely sure about how one would go about working through these various methods but I could see it being quite an interesting way of estimating biodiversity and quantifying the uncertainty.
After Julian’s talk and some afternoon tea a few members of BRAG stayed back to do a brief workshop on the statistical issues in ecology and to try to match up which methods we’re good at with what the problems in marine ecology are, in an effort to stimulate some work that could be started while Julian visits us.
I’m sure there’s also a lot of spatial modelling and NP Bayes to be done with the data that AIMS has available to them, so I’ll have to keep an eye out on positions they have going.