Yes, if you are expecting excess zero-count data you can use Poisson :)

I have the feeling (i.e., I don't know) that the answer is more complicated than what Andrew Paul McKenzie Pegman is suggesting here. As an alternative to a Zero-inflated Poisson model (ZIP) you may want to consider a zero-inflated negative bionomial GLM (ZINB), so have a look at these ressources:

Zuur AF, Ieno EN, Walker NJ, Saveliev AA, Smithm GM. 2009. Mixed effects models and extensions in ecology with R [electronic resource], New York ; London : Springer, 2009.

However, I think that your data are not large enough to get much out of these models!

If you have excess zeros, you might just have one or two observations for an order, hard to say then if these observations are due to size or distance of the cavity. By looking at more orders than you have sites, you will get lots of 'results' just by chance. Likewise, once a taxon is found at more than 9 or 10 sites, you can't get much information out of these too.

You have 12 trees/observations, barely enought to cover all nine possible combinations of your two continuous variables if you would group them into three classes each (small, medium, large cavities; close to, middle-ish, and far away from the ground), more problematic so if there is an interaction term between size and distance of these cavities (e.g., small cavities are more likely to be colonised if they are closer to the ground).

Ecologically, tree cavities size is probably a good proxy for age of the cavity, which in turn has an effect of the presence/absence of taxa if the colonisation and extinction rates are not the same (and they probably differ for various taxa anyway). By setting all populations to zero during your trial, you in fact remove the factor cavity age from your system.

Furthermore, you model assumes that distance to the ground is the same as distance to the nearest population when it comes to colonisation. This may or may not be the case for each taxa you are looking at. Likewise, is one meter on a smooth beech tree bark the same as one meter on a rough oak tree bark, from a worm point of view? Are there any other cavities on the tree that may act as a source population?

Also, comparing models using abundance might not be useful if your taxa of interest differ widely in reproduction rates. 250 beetles is not the same as say 250 mites. Not to mention that when you count individuals at the order level, you might blurr ecological differences between species (e.g., larger cavities can support more predatory species within a given animal size class, but at lower abundances).

Does that makes any sense? I would probably have a look at this paper if you haven't explored your data yet:

Zuur AF, Ieno EN, Elphick CS. 2010. A protocol for data exploration to avoid common statistical problems. Methods in Ecology and Evolution 1(1): 3-14.

(I know that I haven't really answered your questions, but some of my comments might be useful when dealing with reviewers at some stage),

Cheers,

Claas

Hello, I see that the answer the hypothesis you want to answer is if there are differences in the abundance of your samples according to the different types of trees and cavities, if so, you could test Poison distributions, Sorensen's ordering maps, or multivariate hypothesis tests like PERMANOVA or MANOVA.