Does anyone have a good source for correcting for spatial auto correlation when comparing a species assemblage (site X species matrix) to a geographic distance matrix? I know a mantel test will tell me how correlated the variables are but how do you correct for this effect in subsequent analyses?
Hi Israel,
I found numerical ecology (P. Legendre and Loic F J Legendre) and Numerical ecology with R (Daniel Borcard, François Gillet and Pierre Legendre) really helpful for learning how best to deal with Community Data in spatial frameworks; they go through a couple of nice examples testing for spatial autocorrelation in multivariate datasets and then discuss different options for correcting for this. The attached paper outlines a method of partialling out the effect of spatial structure in the data in a partial canonical functions analysis using spatial eigenfunctions. These analysis can be implemented readily in R and are discussed in the books above. I hope this helps and best of luck,
Jan
Article Estimating and Controlling for Spatial Structure in the Stud...
There are some regression techniques that are specifically designed for this type of data (spatial autoregressive models). You can implement these relatively easily in R. For example, check out the spatial error model and/or conditional autoregressive models available in pacakge 'spdep'. Also googling 'conditional autoregressive models R' comes up with some useful PDF and powerpoint tutorial materials. Good luck!
Hi Alice,
Do you know if those models exist for use with multivariate response variables (i.e., community matrices); if so would definitely be a neat approach, but I haven't found much yet. Thanks and all the best,
Jan
Hi Israel,
To control for the effects of spatial autocorrelation, you can use spatial eigenvector mapping (SEVM). This approach removes all significant autocorrelation from the residuals of your model. Have a look at Griffith & Peres-Neto (2006) Spatial modeling in ecology: the flexibility of eigenfunction spatial analyses. Ecology 87, 2603–2613.
Good luck
Hi Jan, sorry I didn't appreciate the multivariate response aspect of the question! I have not ever dabbled in multivariate response data, so I can't really offer any help with that. I recently had a spatial analysis workshop with Fangliang He from the University of Alberta and his research group seem to work on community dynamics and assemblages alot - with a focus on spatial autocorrelation and what it tells us/how to deal with it. Perhaps it is worth having a look at his publications or contacting him (he is very approachable)? https://www.ualberta.ca/~fhe/
Maybe this is useful http://onlinelibrary.wiley.com/doi/10.1890/10-0602.1/abstract
Spatial Efficiency metric (SPAEF) is proven to be robust when comparing two raster maps. Python and Matlab codes are available at: http://space.geus.dk/tools_products/index.html
Two papers on the metric:
- https://www.hydrol-earth-syst-sci.net/22/1299/2018/hess-22-1299-2018-assets.html
- https://www.geosci-model-dev-discuss.net/gmd-2017-238/
- Badges
- Science topic
- Similar topics
- Mathematics
- Algebra
- Matrix