Finding Field Sites and Comparing Two Development Patterns

Hello again everyone! I suppose it is time to discuss more about what I am studying. My current work is to determine the long-term (20+ years) effect of exurban neighborhood development on stream morphology, stream health, and salamander occupancy/diversity. I'm selecting small streams where salamanders are the dominant predator of the ecosystem (1st/2nd order, no fish). The goals of this project are to see if in the long-term a stream that is disturbed for exurban development returns to pre-disturbance conditions, or remains permanently degraded.

It has been found that streams and their associated salamanders recover after timber harvest. Exurban neighborhoods also involve large vegetation loss for development, and are also allowed to grow said vegetation back, like in a timber plantation. There are obviously many difference between forests and low density neighborhoods though, so we need studies to show us if these areas recover. If you have questions about the specifics of this work leave me a question in the comments, I don't want to bore my readers with all the technical jargon and heavy background information.

This week I'll be traveling to Gatlinburg and Pigeon Forge, TN to locate my field sites on the Tennessee side of the mountains. While looking at the mapping software I've come across an interesting observation. It appears that we are in fact seeing two distinct patterns of the development. A typical Gatlinburg area community looks like this (note the condo/house density and the amount of vegetation removed).

Our typical NC neighborhood looks more like the following image (far less dense, but much larger houses and often associated golf courses):

I am currently trying to decide how to incorporate this variation into the study. I'm wondering whether I should try to find TN sites similar to the NC ones for the sake of consistency, even if they are not representative of the majority of the neighborhoods I've seen on the TN side, or if I should select this dense neighborhoods in TN. It might be cool to capture the two ranges of development, but I also worry that adding more variables will reduce the statistical significance of whatever findings I may have.