Hey guys, I've been working on a project with my research mentor where I have been delineating watersheds at biological collection sites along the Cahaba River (near Birmingham, AL, an urban area) using ArcGIS, and using these watersheds as "cookie cutters" to cut out a shape of US Land Use data from the USGS. We've been doing this in order to quantify how land is used within the watershed of each site. When I process these data with the biodiversity data derived from each collection event in R, the AIC indicates that Urban Cover within the watershed (as opposed to Agriculture, Grassland, and Forest) best predicts biodiversity, even well away from any urban center. While my project mentor and I think this makes sense (sites further down the river seem to have highest biodiversity, as well as higher Urban Cover), we can't see any other trends. We've tried using river miles from the mouth of the Cahaba and catchment size to see if we can keep Urban Cover constant and find other trends, but AIC still favors Urban Cover as a predictor. Is there a way that we can hold Urban Cover as a constant, so that we can search for trends elsewhere within the data? Would really appreciate your input.
Dear Patrick Lewis
I think you can insering Urban Cover data as additional variables by applying Hidden Markov Model (HMM).
Good luck,
No. You should not make cover constant, because then it ceases to be a variable and you remove its influence on the other factors. You must not examine one factor at a time. You need to examine all of the data together in one large data file. You can do ordination to discover patterns of association, then PERMANOVA and post hoc tests to determine the actual relationships. AIC is really for testing which competing model is best, not how factors interrelate :)
https://www.cbd.int/doc/publications/cbd-ts-78-en.pdf
check this out.
Hi. The higher biodiversity in urbanised areas may be related to the dynamics of the "r" ssp. These are the ones that colonise degraded surfaces, altered by anthropic activity. They are herbaceous plants that very quickly colonise the most disturbed environments (or empty spaces, such as abandoned agricultural areas). In the spaces separating cities from natural habitats, what is known in ecology as the "edge effect" is generated. Such areas are particularly favourable for the germination and establishment of herbaceous plants of this strategy "r". Their absolute and relative frequency is very high (they die when K-strategists germinate and establish). Maybe that is why you find more biodiversity in those areas. By installing simultaneous sampling plots (at time t1), following a gradient of distance to the urban core: d1...dn ), and collecting field data at t2 ...tn , it might to estimate how distance and time influence the increase or decrease of biodiversity. Another aspect is that urban cover increases or decreases over time. Wouldn't it be important to take this into account?In terms of trends...for example, if urban cover increases, is there also an increase or decrease in biodiversity?
Translated with www.DeepL.com/Translator (free version)
- Badges
- Science topic
- Similar topics
- Statistics
- Survey