I've completed an analysis of biodiversity in forestry and found it to be highly acceptable method. Since it is an hypotheses testing technique it can be used to test for causality.I tested a model with 12 variables that tested for hierarchy and colinearity. I hope to submit the manuscript in November/December. Also search for publications by Eric Lamb, University of Saskatchewan. 

Of course the approach started in biology with Sewall Wright. This book shows the usefulness of SEM in biology and conservation

Cause and correlation in Biology.

Shipley B. 2000.

Cambridge: Cambridge University Press.

£45.00 (hardback). 317 pp.

now in a 2nd edition - see review in http://aob.oxfordjournals.org/content/90/6/777.full

I would start by drawing up your causal model and see if you really do need mediators and moderators etc on the causal paths and the extent to which interventions and randomization can reduce the need for complex statistical modeling.

One of the principles you need to bear in mind while trying to establish causality is temporal order (like classic mediation analysis). 

There is some information you may find useful at

http://www.nwrc.usgs.gov/SEM/index.html

This is also a useful ref on SEMs: Eisenhauer N, Bowker MA, Grace JB, Powell JR (2015) From patterns to causal understanding: Structural equation modeling (SEM) in soil ecology. Pedobiologia

I have used SEMs to evaluate the indirect effects of land-use intensity on diversity. It is a very powerful tool but requires careful consideration of the model structure. I recommend that you also think about composite and latent variables because they can be very useful in ecological studies.

Here are some links with more detailed information:

http://www.structuralequations.org/

http://www.r-bloggers.com/ecological-sems-and-composite-variables-what-why-and-how/

http://www.billshipley.recherche.usherbrooke.ca/