null models for a single species
On Tue, Mar 8, 2011 at 2:55 PM, Penner, Johannes
<Johannes.Penner at mfn-berlin.de> wrote:
Dear Prof. Oksanen, thank you very much for the quick reply! I will try to give more details on the problem. The data is from a colleague and she is investigating what are the responsible factors (environmental and genetic) for selection of ponds by reproducing frogs (one species only). So far all the usual ordination (nmds, pca, cluster analysis, etc.) has provided no result. No environmental factor tested seems to be important. Therefore, one of the hypotheses is that the frogs do not select at all and take ponds at random.
I don't think you can - you can only test if they are random *in relation to the factors you measured*. These frogs could have a preference for the ponds with a certain characteristic X, but which is not reflected in any of your measured factors - so *in regards of the factors measured*, the distribution is random, but still highly dependent on the factor X. Bottomline: I think you have to refine your question before you can answer it. Cheers, Rainer
We would now like to test that statistically but we do not know how. Do you mean that by "structured model" and "structured hypothesis" or the structure of the ponds in the landscape? Thank you very much for your help! With kind regards Johannes -----Urspr?ngliche Nachricht----- Von: Jari Oksanen [mailto:jari.oksanen at oulu.fi] Gesendet: Dienstag, 8. M?rz 2011 13:42 An: Penner, Johannes; r-sig-ecology at r-project.org Betreff: Re: [R-sig-eco] null models for a single species On 8/03/11 10:54 AM, "Penner, Johannes" <Johannes.Penner at mfn-berlin.de> wrote:
Dear List members, I would like to test whether an observed occupancy of lakes in a landscape has occurred randomly (by chance) or not. How can I do that? The problem is that it concerns only a single species and I would like to use binary data only. At first I thought of generating null models and test the observed occupancy against the randomly generated one. However, this needs more than one species... Any hints are highly appreciated!
Johanne, Actually many of the null models as defined in vegan would work here: you only provide a one-column matrix. Although they work, they would not make much sense: null models of type "r00" and "c0" would only give you random permutation of your data (and "c0" would give you the data). Naturally, this is one way to go: just permute your observations. For simple permutations you can use sample() function of base R, and for constrained permutation you can download Gavin Simpson's 'permute' package from http://www.r-forge.r-project.org/. However, if you have a structured model and a structured hypothesis you can do much better than have a simple permutation. I have no idea of your hypothesis, though, and I can't help here. Cheers, Jari Oksanen
_______________________________________________ R-sig-ecology mailing list R-sig-ecology at r-project.org https://stat.ethz.ch/mailman/listinfo/r-sig-ecology
NEW GERMAN FAX NUMBER!!! Rainer M. Krug, PhD (Conservation Ecology, SUN), MSc (Conservation Biology, UCT), Dipl. Phys. (Germany) Centre of Excellence for Invasion Biology Natural Sciences Building Office Suite 2039 Stellenbosch University Main Campus, Merriman Avenue Stellenbosch South Africa Cell:? ? ? ? ?? +27 - (0)83 9479 042 Fax:? ? ? ? ? ? +27 - (0)86 516 2782 Fax:? ? ? ? ? ? +49 - (0)321 2125 2244 email:? ? ? ? ? Rainer at krugs.de Skype:? ? ? ? ? RMkrug Google:? ? ? ?? R.M.Krug at gmail.com