Do you have to publish your whole analysis?

Hi Justine,

I have had similar results to this and wanted to know what your thoughts are re the implications of a publication. I am aware most report both alpha and beta diversity. However, If one's aim is to ultimately measure dissimilarities between two different ecosystems, would it be okay to just report beta-diversity here and not alpha?

For example in the plots below (please correct if req):
Obs features shows greater abundance in ecosystem X
Pielou's evenness shows site Y has a more even distribution of observed features
Shannon (richness + evenness) shows mean diversity is higher in Y
Faith_PD based on phylogenetic tree distances shows mean diversity is higher in X

However none of these is significant statistically. Would it be worth mentioning them at all if beta measures (on phylogenetic tree) are significant and say, show site X as having > diversity.

This choice is up to you. I would recommend reporting results from all tests performed, and focusing the discussion on the results that are most surprising.

If the alpha diversity is expected to be different between two groups and you find that it's not, this results is absolutely worth reporting! The biological context matters more than the p values.

You can also highlight differences in findings:

So X has more taxa but Y is more even. Interesting!

One last thing:

This is a very small sample size and is likely underpowered. I'm not supprised that these are not passing your alpha threshold with n=5. If you get more samples, your power (and p-values :stuck_out_tongue_winking_eye: ) will improve.

Go get more samples!

1 Like

Apologies, and thank you Justine for splitting to a new topic. I thought it was related due to similar results.

1 Like

Hi @skose82,

Sorry, I should have made a note. I split the topics because its almost a year later.


1 Like

Ah yes, the sample power is low as it was a quick preliminary test to see if beta diversity differed between the sites. Hence the wonderment of reporting the alpha - but I guess it can't hurt?
Thank you for your response!

1 Like

It also is worth considering where you are intending to submit for publication. Different publications might have more or less desire to incorporate more or less figures. Your reviewers may also ask to see certain plots like this alpha diversity. There is also the consideration of if the publisher charges by the figure or not, in which case there may be incentive to not include figures of alpha in the main text.
I personally, based on how you've described it here, would start out with putting these alpha figures in the supplemental as it doesn't seem to be your main interest but is still worth referencing briefly at least. If your reviewers think it should be in the main text and adjusted from there, then they will say so. Of course, as the above have noted, it really depends ultimately on how you are talking about everything and putting everything together.