observation covariates changing over occasions in pcount

[Walter Di Nicola]

Hello, 

I am working on a n-mixture model to estimate deer abundance and I would like to include a variable that gives information on the period of the year for the abundance part of the model. 

For that, I made a new object ("Period") where there are all my sites and where I have a a new value for each occasion (e.g. for occasion 1 I have period 1 and so on), however, I do not know how to include it in the unmarkedFramePCount. As it is for the abundance part of the model, I should include it within the siteCovs, however, since period of the year changes at each occasion, I cannot format it as a single column. 

Does anyone know how can i account for this variable?

Thank you!

[Juergen Niedballa]

Hello,

this is likely not possible out of the box because, as you said, only site covariates can be used on abundance. So, alternatives are:

• include period as a covariate on detection
• consider a multi-season model
• if there is sufficient variation in period across stations, and deployments are relatively short, you can maybe use the deployment start or midpoint as a covariate on abundance. Depends on the data if that is feasible.It would only be an approximation though and not precise down to the occasion.
  
• or you could split the data between periods, and still analyse it in a single-season framework. So for example Site1 would be several rows instead of one row, with site covariates organized by e.g. " Site1_Period1", "Site1_Period2" (likewise for detection histories and observation covariates)
  

In any case, this is not part of camtrapR or directly supported in the package workflow, so you'd likely need a custom solution here. The Hierarchical Modelling in Ecology Google group may be a good place to discuss potential approaches in more detail.

https://groups.google.com/g/hmecology/

Best regards,

Jürgen