What dissimilarity index would you use (e.g.: Bray-Curtis, Euclidean, Manhattan, etc...) to analyze data from a community with 10 - 12 species, over several years.
I would go for Bray-Curtis but I'm not completely sure that double absences should be disregarded
You can use beta diversity as a proxy for dissimilarity, which uses presence-absence data and is independent of sample size. But Bray-Curtis is good too because it is based on counts and can include absences because it is bound between 0 and 1and utilises the sum of the lesser values for only those species in common between both sites
Hi!
I would suggest first to define if you can interpret unambiguously the double zeros, since this has a great influence on the dissimilarity measures you can choose from (and a great ecological importance).
If not, Bray-Curtis is certainly an option; otherwise I would choose the Euclidean distance for clear double zeros interpretation.
I would suggest you to use simpson's dissimilarity, because it is not affected by species richness. You can find the function in "betapart" package in R. Also, take a look at Anderson et al. 2011 (Ecology Letters) for a better explanation on the choice of a dissimilarity index.
- Badges
- Science topic