Mantel Test winth environmental data

Is possible to make a Mantel test comparing my environmental data with my sequences? If the answer is yes, how can I input my environmental data?

Have you tried this?

You’ll need to generate a distance matrix for the environmental parameters, we haven’t gotten that far yet.

I’m having the same question as posted here, and since we’re one year further than the original post, I’m giving it another try.

Any recommandations on how to generate a distance matrix on environmental data? I’m not familiar with R, so I’m hoping for another, easier way.

I would have for example pH, concentrations of a lot of salts, TOC/DOC, EC, O2, t°, etc, for 7 samples, and I would love to find correlations between those parameters and the 16S data from these samples (OTUs or just the sequences). So, working with mantel(phylip1=X.dist, phylip2=Y.dist) seems great, IF I would find out how to get a distance matrix on my environmental parameters.



Now, I managed to make a distance matrix for my environmental data, which wasn’t that hard at all. Just inputted it in a straightforward statistical software, made a correlationmatrix and transformed it to a distance matrix (the easiest way is distancematrix = 1 - correlationmatrix). Don’t know for sure of this is totally correct, but it seemed a good way to start.

Then I could finally input these data in mothur to call upon mantel(). But I get the error message that my two matrices are not the same size. Well, of course they are not. I don’t nearly have as many environmental parameters than I have OTUs. Only the amount of samples is the same, but this isn’t reflected in the matrices.

So, am I completely wrong here? It did seem strange to me to compare two matrices of different sizes, so I wasn’t completely surprised by the error message. And indeed I found that Mantel requires equally sized matrices.

I guess this means you can’t use Mantel to compare environmental data with otu’s. Unless you are so fortunate to have 1641 environmental paramaters to go with your 1641 otu’s.


Okay now, don’t bother. After some good night’s rests and a lot of easter eggs, I see now that it didn’t make sense what I was trying to do. Relating overall genetic distances to environmental parameters like TOC and temperature would in my case just result in a lot of really silly data :oops:
If I would like to use Mantel, for example for sampling depth in my case, I believe I would have to construct OTU distance matrices for each depth and compare those.

But well, formulating my problem on this forum hàs helped me find the answer. I think. :wink:


Glad we could help :slight_smile: