Lifemapper

[Robert Mesibov]

I will be very interested to see if TAXACOMers have more success with
predictive modelling than we've had in Tasmania. A "GARP-like" algorithm
was the basis of CORTEX, a biogeographic modelling system at 1 km x 1 km
scale which was based partly on climate modelling (BIOCLIM) and whose
inputs were historical records:

Peters, D. & Thackway, R. 1998. A new biogeographic regionalisation for
Tasmania. Hobart: Commonwealth of Australia & Parks & Wildlife Service,
Tasmania,

CORTEX didn't do badly with trees and birds but failed dismally with "lower
life forms" (not my phrase),  including frogs. In 1997 I  evaluated the
species models for 76 terrestrial invertebrates and found that 32 of the
species (42%) filled only a small portion of their predicted ranges. The
failure had nothing to do with data gaps; most of the species had very well
known ranges.

In my unpublished evaluation report I suggested 2 reasons for "unfilled range":

1. The species hasn't yet equilibrated with current environmental
conditions, or is blocked from doing so by  dispersal barriers. (Sounds
good, but hard to test.)
2. The 'species' is better understood as 2 or more ecologically convergent
species. Between them, these "equivalent" species divide up the predicted
range in an allopatric/parapatric mosaic. (Also sounds good, but only
seemed to apply to a few of the 32 failures.)

CORTEX was next tried on a rare beetle, this time using site data as well
as 1 km square generalisations:

Meggs, J.M. & Taylor, R.J. 1999. Distribution and conservation status of
the Mt Mangana Stag Beetle, Lissotes menalcas (Coleoptera: Lucanidae). Pap.
Proc. Roy. Soc. Tasmania 133: 23-28.

"The CORTEX model (fig. 2) showed suitable environment for the beetle
occurring in many areas well to the north and south of the known
locations." (Meggs & Taylor, p. 25) Although absence of evidence is not
evidence of absence, L. menalcas has yet to be found outside those known
areas, even after deliberate searching.

Programs like CORTEX and LIFEMAPPER might be expected to give better and
better results as more and more locality records are fed in. The Tasmanian
experience is that the better the real range is known, the more obvious the
program's failure becomes, at least for invertebrates. At some point, one
already reached for many Tasmanian creatures, you can pencil in a range
boundary on a map of localities-as-dots and get a much more credible result
than you can by environmental envelope modelling.

While I respect the efforts and intentions of range modellers, I can't help
thinking that modelling is no substitute for field work, especially for
mapping those "lower life forms" that don't disperse very easily or
quickly, and which are therefore of particular conservation concern.

Dr Robert Mesibov
Honorary Research Associate
Queen Victoria Museum and Art Gallery
Home contact: PO Box 101, Penguin, Tasmania, Australia 7316
(03) 64371195; 61 03 64371195