I would like to know if the following analysis strategy is correct conceptually, if it makes sense. I found difference in fish assemblage structure between two estuarine sectors, through PERMANOVA. I would like to identify what environmental factors are associated with this difference. Then, I used DISTLM (Distance‐based linear modelling) utilising information criteria to find the best model of variables related to the multivariate data cloud. I then run a canonical analysis of principal coordinates (CAP) to find axes maximizing this spatial differences with linear combinations of variables from the best DISTLM model. I'm not so sure if I can use these variables from DISTLM in the CAP. If not, can anyone suggest an alternative analysis strategy?