Hello all,

I am new to the forum and this is my first post. I would appreacite if somebody could provide advice on the following problem:

I am working on mangrove fish ecology and right now I am dealing with multivariate data of fish abundance data from several species. In the following lines I will try to explain the experimental design for which I would like to run a Permanova test. There are some issues related to the how to define my factors that I would like to be sure, before running the analysis and any advice will be of extreme help.

From December 2009 to November 2010, I sampled mangrove fishes in a semi-diurnal macrotidal area on a monthly basis. I used block nets that were set at the entrance of small creeks (3-8 m wide). Four creeks were repetitively sampled during this time (two in a low salinity area and two in a high salinity area). Each month I sampled the four creeks at day and night. The date of sampling each month was designed to alternate between spring and neap tides. That means that I have six samples (from the four creeks) from spring tides and six samples (from the same four creeks) from neap tides. In total, I have an n of 96 block net catches. Block nets captured fishes that entered with the flood tide; these fishes were taken at the next low tide.

I would like to answer if (1) spring/neap tides, (2) day/night and (3) salinity zone, had an effect on the community structure of the mangrove fish assemblages in my study area. Since I have fixed experimental units sampled over time (creeks), I think I am dealing with a repeated measures design. This is not contemplated in a permanova test (as far as I know). I would appreciate any suggesting to deal with this aspect.

To run the Permanova test I have thought about the factors as follows:

Factor 1. Creek (four levels) ==> random (nested in time of day)
Factor 2. Salinity zone (two levels: low and high) ==> random
Factor 3. Time of day (two levels: day and night) ==> random (nested in tide)
Factor 4. Tide (two levels: spring and neap) ==> random
Factor 5. Month (twelve levels=Dec09-Nov10) ==> fixed

I would appreciate if somebody could comment on the way I have organized and defined my factors for the permanova model, and also if somebody could suggest ways of dealing with the repeated measures nature of my samples.

thanks in advance!