Friday, April 17, 2009

Worse Than Alligators in the Sewers

Writing in the latest New Yorker, Burkhard Bulger suggests that Florida is like Club Med for exotic tropical species, "an exlusive seaside getaway, far from the fang and claw of the usual tropical crowd." Exotic species have been checking in for decades, but the potential gravity of the problem didn't enter public consciousness until a group of Everglades tourists captured footage of an exotic Burmese python's epic battle with an alligator in 2003. Shortly thereafter, the hero of Bulger's piece - Everglades biologist Skip Snow - started to discover hatchling pythons, prompting state wildlife managers to quickly switch from telling him "no problem at all" when he raised concerns about Everglades pythons to telling him "you might as well give up". Although it may not yet be time to give up, I was surprised by the overly simplistic strategies supported some professional wildlife managers (e.g., "It's time to stop studying these things and start killing them"). I would hope that if we've learned one thing about invasive species it is that simple brute force extermination does not work, particularly in an area as large as the Everglades and when the strategy being employed is as simple as the intentional road-killing or "rapid-acceleration removal method" practiced by some Florida biologists. One source of scientific information with management significance comes in the form of recent climate envelope and niche modeling studies conducted by Rodda et al. (1) and Pyron et al. (2). Although Rodda et al.'s climate envelope models suggest probable expansion of Burmese Pythons throughout the southern United States, Pyron et al.'s niche modeling analyses suggest a much smaller potential range, and that concern about such expansion should not be a the top of the list of challenges facing managers of python populations. (Is this the first time niche modeling studies have recieved a nod on the pages of the New Yorker?) Of course, a range of phylogenetic and phylogeographic studies are also providing insight on the history, biology, and management of invasive species.

Although most Floridians seem more likely to be concerned about the safety of their house pets than the loss of an ecosystem, Bulger does a beautiful job driving home the profound significance of the latter when he suggests that some invasive species can "...change the way we see a place. A parrot in Miami is like a McDonald's in Kathmandu: a sign that you are everwhere and nowhere at once."


Dan Warren said...

I think it's worthwhile to point out that there's growing concern that the default behavior of Maxent may cause it to produce overly restrictive estimates of species' environmental tolerances due to overfitting of the data. This has the effect of producing excellent AUC scores in the regions from which the training data is drawn, but produces overly restrictive distribution estimates when the model is transferred to other places or times. You can fix this behavior to a large extent by tweaking the regularization parameter, which causes the marginal suitability functions to be a lot less complex.

I haven't read the Pyron et al paper in detail, but I did skim it and don't see any suggestion that they experimented with different regularization parameters. I don't want to suggest that their prediction of a narrow future range IS driven by model overfitting, but it's 100% consistent with that scenario.

Glor said...

Good point Dan. I suspect that most previous work on range projections into the past and into unoccupied areas in the present day is going to need to be scrapped due to problems like this.