How can I test for the correlation between a spatial distance matrix and a species co-occurrence distance matrix?

Question

I want to test whether there is a correlation between the spatial distances of 18 plots and the dissimilarity in their species composition. I also have information about whether they are urban (as a binary distance matrix).

I thought I could use a mantel's test but after doing some readings I think this would be inappropriate due to the spatial autocorrelation of my distance matrix. I was also hoping there might be a way to run a linear model with matrices and include whether or not the two plots are urban as a factor in the model.

Could someone please point me towards a method (preferably R based) that I could look into? I've never worked with matrices before so recommendations for basic intro books with R code would also be appreciated.

Thank you.

Answer 1

I think that dbRDA (distance based Redundance Analysis) can solve your problem. You need a Y-variable (dissimilarity matrix of your species composition) and two X-variables (Lat and Long, in this case you don't need a spatial distance matrix, you use your coordinates). If you want, you can use more X-variables too, maybe ecological variables.

R use capscale() function of VEGAN package to do dbRDA. Then you can use anova() and RsquareAdj() or plot() to explore your results (dbRDA object) and the importance of your variables to explain your species differences between your plots.