I have species abundances with associations between environmental variables. I realize the RDA will only be able to tell you the strength of the relationship of all the species abundances with the environmental variables but not the nature of the relationship.
Here is an example where the species (red) and site in (black)
oxy and dbo are negatively related to each other. Is there a way to infer the nature of their relationship?
Let's assume that oxy and dbo have a quadratic relationship with a hump up at intermediate levels of oxy. Say, I am intersted only in the effect of oxy and dbo on species abundance (red). I could construct a mixed-effects model with poisson distribution, with species.abundance ~ oxy + dbo Then, how do I account for oxy and dbo having a negative relatiosnhip in this equation? How can I add altitude in?
In summary:
- What is the advantage of a PCA over mixed-effect models
- Can we identify the nature of relationships from a PCA?
- If not, how can we write a mixed-effects model that'll incorporate those relationships?
