Estimating species richness using Royle Biometrics

[mferreira]

Hi

I'm interested to estimate amphibian richness across several habitats. My hypothesis is that a particularly habitat, that is being destroyed, has higher species richness that other habitats. I have 122 sites surveyed in 4 periods with 3 to 6 sweeps, with dip-net, in each site for period across five distinct habitats.

I don't know how to use WinBugs and someone from my Lab suggested to use Royle Biometrics with the counts of species for each sweep, for example the data looks like this: (2, 3, 3, 2, 0, 1 ) - 2 species in the first sweep, 3 in the second, and so on... I've done this for each period and the results make sense, and they are almost to good to be true.

Can this be done? I don't find any papers using this approach. Do you see any problem using this method?

Merry Christmas to all
Mário

[Alan Dextrase]

Hello Mario,

I believe the inherent problem with your proposed approach is that it ignores species-specific detection probabilities. You can try psi(species)p(species) models in presence [there are a few posts on the forum about this], but probably the best solution is to go the WinBUGS route. Elise Zipkin and Andy Royle have established a web site that includes software code and relevant references (http://137.227.242.23/pubanalysis/communitymodeling/).

Alan

[mferreira]

Thanks a lot Alan!!!

I was already starting to go WinBUGS. This site is goingo to be very useful!!

Cheers
Mário