I really do not understand what q1 and q2 should be, can you please elaborate on that?

I am a bit confused by your model structure. For one I am unsure why you would want to include random slopes instead of random effects (i.e., 1|plot). For another I wondered if you wanted to include observation time point at least in the random part of your equation.

Hi Sven.

q0, q1 e q2 (Hill Numbers) are indices to calculate traditional metrics of taxonomic diversity (species richness, Shannon’s diversity, and Simpson’s diversity); where q = 0 refers to the species richness of the communities, q = 1 refers to Shannon’s diversity and q = 2 refers to Simpson’s diversity, which attributes greater weight to the most common species.

Ultimanently talking to colleagues, I realized that my model really had problems or at least my interpretation of how I should do it was confusing.

I thought about adding slopes, because I believed that this way I would be doing a paired test. However, I see that performing a mixed model with 'treatment' as a fixed effect and 'parcel' as a random factor would already be a paired test.

Note that the 'treatment' variable refers to my samples before and after the fire. So to use it as a random effect I would have to nest the samples (this to control the temporal pseudo-replication).

Since I used the data from the plot and not the subplot, I don't have to worry about spatial pseudo-replication. So, in this case, the mixed model is not necessary and I can perform more reliable tests with just a paired T-test.

Anyway, thanks for the availability.

My best regards,

Jakelyne.