Unsure about whether to use merge.groups


I am trying to answer the following problems and am unsure which way would be the best to go about it:

I have three samples taken from six individuals at three time-points. I am wanting to see if samples taken from an individual at the first time-point more closely resemble samples taken from them at the other time-points than they do samples taken from the other individuals.

My first thought was to run an AMOVA with samples grouped in my design file based on which individual they were taken from and which time-point they come from ie. Person1-Timepoint1, Person2-Timepoint2 etc. then compare the pairwise differences between these groups. However, I’m not sure this is the best way to answer these questions.

The other way I though of doing this was using merge.groups. I would input the same design file as above into the merge.groups command along with my subsampled shared file. I would then put a distance file created from this and the following design file into amova:

Person1_Timepoint1 Person1
Person2_Timepoint1 Person2
Person3_Timepoint1 Person3
Person4_Timepoint1 Person4
Person5_Timepoint1 Person5
Person6_Timepoint1 Person6
Person1_Timepoint2 Person1
Person2_Timepoint2 Person2
Person3_Timepoint2 Person3
Person4_Timepoint2 Person4
Person5_Timepoint2 Person5
Person6_Timepoint2 Person6
Person1_Timepoint3 Person1
Person2_Timepoint3 Person2
Person3_Timepoint3 Person3
Person4_Timepoint3 Person4
Person5_Timepoint3 Person5
Person6_Timepoint3 Person6

Can you see any problems with using this to answer my question?

Thanks for your help,

Hi Laura,

I’d keep them separate - I wouldn’t pool these samples since they aren’t technical replicates. You’ll probably want to use the adonis function in the R package vegan to do a more complex two-way or repeated measures ANOVA.