I collected five samples of animal tissue from each of 3 different sampling sites (15 samples total). For each sample, I determined bacterial community composition, so I have a standard multivariate table of species/abundance data. At each sampling site, I also sampled various environmental variables. A simplified version of my environmental data looks like this (I have 8 variables total):
Site Temp pH Nitrogen A 20 7 24 B 22 6.1 60 C 19 7.4 41
The main purpose of my analysis was to demonstrate similarities or differences between animals in the different sites, which I did using an NMDS plot showing that one site is similar to another, and those two sites are different from the third:
A reviewer has asked that I demonstrate how my environmental variables may influence the trends seen in this plot. I understand the principles of vector-fitting and know how I could do it using, for example, constrained ordination or
envfit in vegan, but my concern is that I only have one set of environmental variables for all five animals at each site. Assigning each animal the environmental data from its site would be artificially replicating my data (and create issues of no variance between samples).
Thoughts on how I could go about fitting my unreplicated environmental variables to this ordination?