Correlation between RSF covariates

[Rebecca Leloudis]

Hi Chris,

I have what might be a bit of a broad question. I'm running rsf's using different land cover classes, and I'm concerned about possible spatial correlation between classes and its impact on the reliability of my rsf results. My understanding is that in traditional rsf's that use logistic regression, high spatial correlation between land cover classes (for example, classes of wetland) would make the final selection coefficients somewhat unreliable. I'm having difficulty understanding, however, if this is the case in ctmm's rsf functions, given that they use a point process model. My rsf results suggest strong selection for forested wetland, but strong selection against emergent and scrub/shrub wetland. Yet when I used Ripley's K to test for clustering among wetland classes, all 3 appeared to be spatially correlated. Would this clustering of wetlands invalidate the results of my rsf?

Apologies if this question is extremely basic or confusing, I struggle with the underlying statistics behind some parts of the ctmm package.

Thanks so much for your time!

[Christen Fleming]

Hi Rebecca,

I don't think the spatial autocorrelation matters. This isn't a process that's being sampled across the landscape but over time. It would be different if you were going out and taking soil measurements.

Best,

Chris