I'm working with ecological community data collected from different sites and want to make comparisons between the sites using descriptors like richness, Shannon Weiner's diversity, and Simpson's evenness. There are some sites with multiple temporal replicates and others with just one sample. What is the best way to summarize the community data for sites with multiple replicates so they can be compared with the sites with only one sample? For example, the dataset looks something like this, where the values are counts of each species:

          SiteA_Rep1  SiteA_Rep2  SiteA_Rep3  SiteB  SiteC  SiteD_Rep1  SiteD_Rep2
SpeciesA      1           4           2         0      2        0           6
SpeciesB      0           2           0         7      4        1           2
SpeciesC      7           6           3         6      2        3           0

I am trying to summarize the data so that I have one count for each species at each site. I have tried taking the mean or median count for each species at a site with multiple replicates, but this inflates the richness because in effect I am increasing the sampling intensity at that site and increasing the chance that a rare species is encountered.

I have also looked at rarefaction methods for estimating richness, but I need the species counts for my other metrics.

Is there an ideal way to turn these replicates into one "sample" while preserving counts for each of the species?


You could calculate the metrics you want to compare (i.e. Shannon-Weiner and Simpsons indices) for each site, and then take an average of those metrics for the sites that were sampled on multiple occasions, before carrying out your comparisons to other sites.

Of course this does lose some of your information, by summarising the multiple samples into a single metric - but that is what you asked to do. If I were you I would consider presenting the results from all your sites separately - showing the metrics you obtain from multiple re-samples of the same site could be interesting. For instance, you might get different results because you sampled at different times of year. It would be interesting to know, for example, if Site B was more diverse than Site A in summer and Site B in winter.

  • $\begingroup$ Right, thanks for the suggestion! Can't believe I overlooked this option, but it makes the most sense $\endgroup$
    – Fragilaria
    May 18 at 13:16

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.