Compound poisson gamma model and biomass data

I have some biomass data (total count -discrete- of individuals and total dry weight -continuous-) that I would like to model with environmental parameters recorded at different spatial scales (temperature -continuous-). My biomass data is zero inflated mainly due to scale inaccuracy and as a results I have many sampling points that present a heigh number of individuals but no biomass reading. I do however have some true zeros and I don't to ignore them in the analysis.

The sampling effort between blocks presents some discrepancies (lost samples or logistical problems) and I have been reading that CPG was more robust than a delta approach in this case.

CPG is completely new to me and I am still struggling a bit to understand all the math behind although I do get the general picture and philosophy (I think...).

To put it into context, all the sampled blocks are part of a fragmentation experiment in the same landscape and before running a CPG I was gonna test for spatial autocorrelation between sampling units for each block.

Am I going down the right road?


To clarify my question, is this the right approach to deal with this data set? I want to be able to see if there is a trend between biomass and temperature data at different scales but I also want to take into account all my biomass readings that have a zero output but still have a count of individual recorded.