How to further investigate an increase in diversity

Hi all,

I have an experiment where I’ve compared the gut and stomach microbiomes of a territorial reef fish receiving enriched nutrients to those without. I found that both stomach and gut microbiomes increase in diversity when nutrients is added but interestingly enough DeSeq and ANCOM and PICRUST all highlight taxa that are at higher relative abundances in the controls.

So from what I can tell, the increase in diversity in the enriched samples is due to an influx of rare taxa that do not have high enough abundance to statistically mean anything. The other possibility is that adding nutrients increases the rate of unassigned taxa which would also explain the results. How might one move forward to better explain the increase in diversity?

I realize I am omitting a lot of important information so feel free to ask for more info if needed.
Thanks,
Sam

Without seeing the data, it sounds rather like the nutrient addition is causing blooms of different clades in a more or less chaotic fashion, not a consistent change. Are you seeing greater beta dispersion in the nutrient addition? (check out the permdisp method in qiime diversity beta-group-significance action).

1 Like

Hi Nicholas,

That makes sense. This sounds like the test I’ve been looking for. Thank you!

1 Like

So I actually get this result with the permdisp method where it looks like my controls have higher dispersion that my experimentals (assuming I am reading it correctly)....this is strange no? I am trying to wrap my head around what higher diversity but lower dispersion could mean. And I get this result for jaccard and bray-curtis and then it seems to diminish in both unweighted and weighted unifrac matrices, p~0.1, but trend still there.

I guess this means there is an increase in diversity but the increase is uniform and decreases the variance.
Attached is my plot.
Thanks again for the feedback.

Could you share your PCoA plots?

Yes,
Attached is a plot with just the gut samples (colored by treatment), just the stomach samples, and then a plot of the gut, stomach, and algae (their food) microbiomes and the same plot also with shapes corresponding to treatment. Hope this helps. Sidenote: not sure if the component percentages are accurate they seem very low to me despite solid clustering in some of them.
Thank you,
Sam

Gut%20OTU%20PCoA%20Plot Stomach%20OTU%20PCoA%20bray%20curtis

So it looks like control is more dispersed (note: this complicates interpretation of the permanova result — a significant permdisp result could indicate that the control is just more dispersed but not necessarily significantly different from EXP)