Is it possible to add nested random effects?

[sanne-verdonck]

Hi everyone,

I have a question regarding nested random effects and whether it is possible to add them to a sdmTMB model or not.

My data was collected across multiple plots which are distributed across two forest complexes themselves. This means I have repeated measurements per plot and the plots are nested into a specific forest complex. With glmmTMB you can add this nested random effect as follows: (1|site/plot). Using the same syntax does not work for sdmTMB. Hence, my question: is it possible to add nested random effects to a sdmTMB model? And if so, how do you implement it?

Cheers,
Sanne

[seananderson]

You should be able to do the same thing by coding your plot IDs in your data as unique across all sites not just within sites and then having a random intercept for site and a random intercept for plot. As a warning, if your sites are spatially arranged and you’re also including a spatial random field, you might struggle to estimate the model. One is assuming the sites are IID normally distributed and the other is assuming they are random fields (MVN with some covariance function). Theoretically you could have random site noise on top of the random field, but unless you have a lot of data that may be hard to estimate and you might want to pick one or the other.

[sanne-verdonck]

Thank you for your fast reply!