Hi Alex - have you had a look at "Ants: Standard Methods for Measuring and Monitoring Biodiversity" by Agosti et al. (Smithsonian, 2000)? It has sections on the diversity of ants, and their use in monitoring environmental change, plus sampling effort and choice of methods (among other topics),

Dave

Dear Alex,

I am fish ecologist. My PhD aims at assessing the impact of benthic marine habitat shifts (from marine forest to barren grounds) on fish assemblage structure. This looks quiet similar to your problematic. Hence, I gave you some suggestions, which I hope won't be too much naive.

I think you need to refine your initial questions and the expected (socio-economic?) outcome, since the 2 experiments you suggest are really different:

With the local study including new habitats, you could adress new questions related to landscape ecology. For instance, you could compare for a same type of habitat, "large" patch versus "small" patch (I thought to this because you spoke about forest ecotone). Results might be particularly usefull for improving management practices (land use, how do we harvest forest?).

The regionnal experiment for sure will help to confirm/infirm your local observations. And it might be particularly usefull in answering to your third question: what are the other environmental features affecting ant assemblage strucutres (if I well understood you question).

Please, I am interested in how do you estimate ant participation to ecosystem functions. Do you go beyound discussing the detected functionnal diversity patterns?

Just a remark, I think your questions 1 and 2 are similar, is not it? I would suggest to present them in another way:

The aim is to assess the impact of human-driven habitat shifts on ant assemblage structures, by comparing among habitats both biological and functionnal diversity of ant assemblages.

I hope this helps,

ciao

Pierre

Hi and thanks Dave and Pierre,

Yes Dave, I have read the book about ant sampling methods. But the question I address is not linked to methodology, but rather a choice of experiment to answer some questions.

About the questions, Pierre, certainly there is a bit of redundancy, probably because I haven't deepened enough each of them. I have modified the question, hoping it is clear now. Thanks for the remark. About ant association to processes I use a functional group approach based on feeding ecology of ants obtained on the basis of empirical data. This allows me to estimate intensity of ant participation on several processes, which change depending on the habitat. I’d like to understand how species are filtered, and the consequences of the number of species decline and substitution onto processes.

So certainly adding habitats will help providing more insight about shifts in a gradient, but I’m afraid such conclusions will be limited due to the “local” nature of the study.

Alex

Hi Alex,

I know your questions are not linked with ant sampling methods, but sampling is your first and in in many times your only connection with your model organisms. So, using only baits maybe is not a better alternative for your purpose. Baits is very selective method and samples only a small part of ant diversity (and consequently, ant functional diversity). This bias is more strong in tropical areas, and you will miss fungus-growers, kinetic trap-jaw species, several specialist predators, honeydew feeders, and so on (specially, if you are working at French Guyane). Maybe using Winkler samples or even pitfall traps will give a better picture of the ant functional diversity in your study areas.

Hi Fabricio,

About the totality of diversity I agree with you that only a subset of species is captured with baiting. But for the pruposes of the study, which include comparing the communities attracted to different baits, other sampling devices are just a control for the percentage of diversity obtained, and yet, this will only include a subset of the species present at an ecosystem. For example if I'll use pitfall, I'll get a community quite similar to that on baits, which captures essentially a subset of "pitfall species" (ground foraging + bait attracted). But winkler samples will provide a quite different group of species inhabiting the litter. Also some arboreal ants can go down and forage on items, so to get the arboreal control I should be fogging too. Several studies have already provided information about communities composition using pitfall and winkler (also in French Guiana), I guess I can use literature to justify groups of species that weren't captured but are very likely to be present (anyway I got pitfall samples of 1 control plot of each habitat type, + winkler + beating, just to provide a fast image).

Ok, Alex I understood (I missed the baiting comparison in the first question).

So, going back to your original question. My suggestion is to go for the second option. Functional traits are (in principle) designed for species level analysis. Therefore increasing your sample scale will be more appropriated in my point of view. In addition, you will have more replicates of your treatments, which will give more power for your analysis. If there is such a filter to functional traits between treatments, the identity of species will change between areas, but the trait convergence will show up.

By the way, what functional traits are you using?

I guess by this far we can agree that to understand trait convergence and in order to obtain robust data, regional replication is far more interesting.

About the traits, I am thinking on combining morphological, nesting, behavioural, thermic and density/abundance traits. Morphologic measures like weber lengths, eye position, femur length, head width and length, and mandible length. Ant’s resistance to desiccation, and the resources they use for nesting as well as resource acquisition strategies. Altogether should provide me insight of functional declines and correlations between traits that can be helpful to understand cascading effects on functional properties of ant communities linked to habitat shifts. Considering this last point, I must precise that species composition between highly contrasted habitats results in extreme variations of diversity and the characteristics of species. Therefore a smoother gradient (with intermediate and transition areas) may also help to understand gradients in dominance and trade-offs. This will certainly depend on the stressors between habitats. I guess that the main ones that we can associate to these habitats shifts are things like “cover/shade”, “litter depth” and “perturbation regime”. What do you think?

Nice Alex. You have a complete functional traits data in hands, and I agreed with you about the main stressors. Besides those 3 stressors, maybe temperature range, which will be strong correlated with cover/shade will also being interesting to look.