Hello all, sorry to forgo introduction and bombard you with a query as my 1st post, but I'm running short of time and it only just occurred to me to pose my question to a stats forum.

I'm using PAST software to perform PERMANOVA and nMDS + SIMPER analyses on a multivariate dataset, (species abundance data, with a large amount of zero values), to investigate differences in species composition/abundance at six sites of varying depth and habitat.

My problem is, I'm having a hard time coming up with a reliable means to determine statistical power, (so far as I understand it: is my sample-size large enough to robustly detect species differences among sites?) Note: rather than a designed experiment, this was an investigative study 'piggybacking' on a commercial fishing operation; sites were assigned post-data collection, with unequal number of samples at each.

Any suggestions or comment on my methodology would be most welcome. Thanks!