I am comparing different sites based on their floristic composition in R. Therefore, I have created a huge community datamatrix (presence/absence data) from 53 sites including over 1000 species. To Perform PCoA on the data I first transformed the raw community datamatrix into a distance matrix using Bray-Curtis distance and took the square root of the resulting values to avoid negative eigenvalues. Than I performed PCoA using cmdscale() and calculated the explained variance of the first Coordinates as followed:


As result I get explained variance of the first coordinate 9.3 % and second coordinate 7.3 % which appears to be extremely low. However, when I plot the results along the two axes it looks fairly nice and the plot shows exactly what I have expected. So, does anyone know why I get such low values of explained variance and if this really matters for me? Has this something to do with the huge amount of species which are, please correct me if I am wrong, in that case my variables?

Interestingly enough I found no PCoA plots on the internet which actually indicate the explained variance of the axes....

Thank you for your help!!!

  • $\begingroup$ With 53 variables the average proportion of variance will be less than 2%. With respect to this, 9.3% and 7.3% are fairly high. From the limited information given, it's impossible to ascertain whether this ought to matter to you. $\endgroup$
    – whuber
    Jun 24 at 20:24

Maybe a silly answer but for me worked well. I got from 18% to 38% on the first axis (sum of the 3 main axes >50%) simply cleaning my data from non statistically significant variables and setting each variable in the correct form since R sometimes do not categorize properly the data. Check str() of you dataset and try to clean up something


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.