Spatial thinning distance

[makani...@gmail.com]

I'm looking to run Maxent in R and handle the sampling bias in the dataset by spatially thinning/filtering the occurrence data. I see there are packages like spThin to optimize the amount of points retained, but how does one decide the desired distance between points? Anyone have a good method for selecting this distance?

Cheers,

Makani

[Joao Frederico Berner]

Hello Makani,

I'm in favour of retaining 1 occurrence record per grid cell, instead of using distance. Of course, it all depends on your taxon and resolution of predictor variables.

Best regards,

João

[sdidel...@gmail.com]

Hello Dear Makani,

If you are using the new version of Maxent which uses the Inhomogeneous Poisson Process models (See here: https://onlinelibrary.wiley.com/doi/10.1111/ecog.03049)  to estimate the relative abundance (number of observed records per unit area) for the target species, to my knowledge, I don't see the relevance to thin your occurrence data. Of course, you can remove duplicates records at the same spatial location.

Best wishes !
Idelphonse

[calde...@gmail.com]

Hi Makani,

 

I have a dataset of white-tailed prairie dog burrow occurrences. I thinned to 30 meters to match the resolution of my environmental variables. This gives me roughly one occurrence per pixel. It’s been a while, but I believe this method came from Kramer-Schadt et al (2013).

 

Thanks,

Cal

--
You received this message because you are subscribed to the Google Groups "Maxent" group.
To unsubscribe from this group and stop receiving emails from it, send an email to maxent+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/maxent/acc7c1f0-5b05-4db8-a87f-2dc6804cb39an%40googlegroups.com.