I'm comparing species counts between areas. My data is very similar to the below.

Species  Location    count
Yellow-frog Africa  3
Blue-frog   Africa  0
Green-frog  Africa  20
Yellow-frog Europe  0
Blue-frog   Europe  100
Green-frog  Europe  0

I'm not sure about what statistical test to use to compare between sites?? Especially so it takes into account that its zero count data.

My desired output would be to see if there is a statistical difference between:

  1. The sites overall
  2. Species (i.e. are there more yellow frogs in Europe or Africa).
  3. The species within each site (i.e. are there more blue frogs compared to yellow frogs)

My stats knowledge is pretty basic. I assume its regression, but I'm not sure about the family. Poisson regression/ binomial/ Zero-inflated.

[P.s. I'm working in R]

  • $\begingroup$ Present the data in 2x3 contingency table. Can perform Fisher's exact test. $\endgroup$ – user158565 Dec 23 '18 at 3:32
  • $\begingroup$ Can you tel us some more about how you obtained this data, and what is the question you want to ask the data? $\endgroup$ – kjetil b halvorsen Dec 23 '18 at 10:07
  • $\begingroup$ Sampling of the species was random, i.e. we would go to a pond and count the number of frogs of each species. The same frog occurs in Europe and Africa. My priority is that I need to know if there are differences in frog species composition between africa and europe. $\endgroup$ – imasnake Dec 23 '18 at 10:28

If your priority goal is to compare species composition between sites, I recommend using multivariate statistics for community analysis. For this you would need to set up your count data as below.

| Continent | Species 1 | Species 2 | Species 3 |
| Africa    |   0       |   3       |      1    |
| Europe    |   1       |   1       |      0    |

Then you can analyze the similarity of the continents using PERMANOVA or ANOSIM in the Vegan Package and visualize the pre-defined groups (continents) using ordination such as NMDS.

| cite | improve this answer | |
  • $\begingroup$ Thanks @bicyclerider. I think this looks like the right way to go. I'm getting the following error when running PERMANOVA/ANOSIM: "'nperm' >= set of all permutations: complete enumeration. Set of permutations < 'minperm'. Generating entire set" I think it's because I'm only comparing species counts between two sites, with no other variables. $\endgroup$ – imasnake Dec 27 '18 at 11:43
  • $\begingroup$ Hi @imasnake, I am not sure what the error is. Did you define the number of permutations you would like to include? You might use 9999 permutations or more. Do you have different sites per continent? If so, I recommend adding a column for Site and including a single row for each site. You can still keep the continent column, just list Africa for each site that is in Africa or Europe for each site that is Europe. In general, each site should be a separate row and each species should be a separate column. Hope that helps. $\endgroup$ – bicyclerider Dec 29 '18 at 1:12

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.