what stats to use?

Dear All,

As I am not experienced in stats I would like to ask for some stats suggestions for a study.
study one: control impact with one impacted site and one reference site. Two species and biometrics taken. I need to see if the the impacted site increases biometrics.
study two: control-impact: two impacted sites with 3 reference sites and 3 species. Same biometrics taken and need to see if the impacted sites increase the biometrics.
The design of the second one is unbalanced on one of the impacted sites (very low numbers collected).

What sort of test would be most appropriate?
When to use PCA for ordination and when nMDS?
when to use PERMANOVA?
(I am using R for all stats).

Help is greatly appreciated.
Thank you.