I have individual collars that are repeatedly sampled over time (and, very importantly, I don’t have same collars being sampled for all microforms), so collars that have a higher flux one week are more likely to have higher fluxes the next week (this is called auto-correlation). It also means that the same difference in flux magnitude between one week and the next will not be the same relatively. For example, suppose I have two collars (A and B) that are both measuring 91 umol this week but last week collar A measured 90 umol but collar B measured only 10 umol. I can see that the relative increase in flux was much bigger in collar B than in collar A. These relative changes in individual collar measurements are important factors in my analysis so I have to tell R about them. So Microform is my fixed effects but I have collars as random effects. And need to consider that these collars are repeatedly sampled.
Datasets EXAMPLES:
Collars Microform
1 HP
2 HP
3 HP
1 PB
2 PB
3 PB
1 FP
2 FP
3 FP