In attempting to explain why I don't find this reasoning circular, I've managed to convince myself that it is. There is an implicit assumption that the control sites in a study are representative of the entire landscape prior to fragmentation. Any difference between the control sites and the fragments is attributed to anthropomorphic effects in the fragments. However, the control site may never have had the same community as the landscape that has become fragmented.
I can't speak on rainforests, but in southern Michigan, I found that forest fragments and preserves were mostly clustered around swamps, ponds, streams, and hills--sites where the topography is too rugged for the plow. Large preserves tend to be on stabilized dune-lands and around wetlands, and I'm certain that, before habitat fragmentation, the plant communities in such sites were different from those in what are now farm fields. I don't think this circularity completely invalidates the approach you describe. Basically, people are going out and seeing what species they find in different habitats and categorizing the species based on what habitats they find them in. They also try to figure out what biological characteristics unite the species within each category and separate them from species in other categories. Hopefully, they have a priori hypotheses, but it is also valid to propose new hypotheses based on your observations, or to note that what you observed is consistent with hypotheses others have proposed. The danger is in not recognizing and acknowledging that your control sites may be unlike your fragments for reasons other than their relative lack of disturbance. Unfortunately, the fundamental problem is that we don't often have good records of what communities were like before disturbance. Using control sites assumes that the fragmented landscape once had a community like that of the control sites. It seems to me that using indicator scores assumes that the fragmented landscape once had a community like those used to produce the indicator scores. On Thu, Jan 29, 2009 at 3:18 PM, Brian D. Campbell < [email protected]> wrote: > Dear listserv members: > > I've been reading a lot of literature recently on the effects of > fragmentation and land-use conversion from forests to agroforests and have > been really troubled by what seems a pervasive issue (at least in my mind); > defining the biodiversity value of a human-modified habitat type (e.g. > either fragment or agroforest). Almost all studies I've reviewed partition > bird communities into categories of "forest species", "rainforest > specialists", "agricultural generalists", among others, and proceed to > compare these among different land-uses. This is not my issue per se, but > rather, I find it very circular if one uses the data/observations they > collected in a study to define these groups; e.g. all species encountered > in > a "control" site of extensive forest were defined as forest species. These > make useful and note-worthy observations but if one then proceed to include > control sites in a statistical comparison then I think there is major issue > with circularity. This also seems to me a very different approach than > having defined a priori (e.g. from distribution lists or other literature) > a > set of forest-candidate species which may or may not be present in any > given > site surveyed. Have others here found similar issues when reviewing papers > dealing with the biodiversity values of secondary forest and agricultural > habitats? > > Brian Campbell > -- James Crants, PhD Scientist, University of Minnesota Agronomy and Plant Genetics Cell: (734) 474-7478
