I can recommend Past 3.25 that performs all these multivariable analyzes and I enclose the manual is friendly to understand when to use each Test. I recommend that in these types of analysis before using them you must analyze the collinearity of the values.

Unfortunately I will not be working with R or MINITAB.

I recommend CCA using vegan package in R.

In vegan, function cca performs correspondence analysis, or optionally constrained correspondence analysis.

But be aware of assumptions of this method.

Please read the following links and see the fallowing paper:

https://www.researchgate.net/post/What_are_the_assumptions_of_Canonical_Correspondence_Analysis

http://ordination.okstate.edu/robust.htm

https://stats.stackexchange.com/questions/249569/assumptions-for-canonical-correspondence-analysis

From a more philosophical perspective I prefer to keep the analyses separate rather than to conduct one (e.g. CCA) that attempts to combine very different types of variable into one ordination. I am also wary of the idea that many of these relationships are linear. You could do some analyses based on the underlying resemblance matrices (Bray-Curtis for fish, normalised euclidean distance for environmental variables) such as BIO-ENV, which could show which habitat variables most closely match patterns in fish community structure, and then visualise those relationships using something like bubble-plots in MDS. In Primer you can also produce NMDS plots on which you can overlay vectors showing the direction of increase in the environmental variables.

I suggest you to try spatial autocorrelation analysis. Bellow there are two links about that. You can do it in R, python, ArcGIS or any other GIS software.

Good luck.

Article Spatial autocorrelation of ecological phenomena

https://gisgeography.com/spatial-autocorrelation-moran-i-gis/