I had 16S rRNA sequences from 16 marine sites; I did not have duplicates and seldom people collect these samples in duplicates or triplicates. I calculated diversity indexes using Mothur “summary.single” command. Now reviewer of the journal to which I have submitted my manuscript has asked me to include error bars; I am not sure, probably he meant to include standard deviation which this command generates in output file. And he/she has also asked me to indicate which samples are significantly different terms of these values (e.g. shannon)?
I was suggested in the last post that differences between alpha indices can be tested by T test or chi-squared test. I am confused, because I do not have samples in replicates. In this case how can I do a T test or chi squared test, when i have >2 sample (16 samples in total)?
If you don’t have any replicates then you can’t do statistics. Can you group your samples into treatment groups? For example, you may have collected 8 samples from shallow waters and 8 from deep. You could then do statistics comparing shallow and deep using a T-test (not a X2 test). If you have more than three treatment groups then you would need to do an ANOVA. If you can’t group your samples meaningfully, then you can’t do statistics and are really limited to being able to do a descriptive comparison of the samples. In general, you want to minimize the number of treatment groups and increase the number of replicates within each treatment group.
