Hello
I am investigating the shared pollinator communities of a native wildflower, and a related invasive hybrid. The hybrid is replacing the native plant. I have collected data which lists all insect visitors to both plants. I want to see if the bee communities between plants are significantly different. I have tried using nmds in vegan and mvabund in R, but it doesn't seem to work when only comparing two things (for example it seems normally multiple sites are species are compared). How might I go about comparing the assemblages? Secondly there have been some interesting results, for example bumble bees don't seem to visit one plant. How might I test if that is a significant result or more likely just a feature of my sampling.
Any help is very appreciated.
Hello Cameron; Your description of the study sounds like the data a comparison of frequencies. That sounds like a Chi Square test is an appropriate place to begin. It is simple and quick...you can do it by hand. Best regards, Jim Des Lauriers
Thank you Andrew Paul McKenzie Pegman and James Des Lauriers . So I have abundance of species visiting the two flowers I wish to compare. I want to know two things: whether the assemblages are significantly different, and whether I can confidently say that some species are specific to one plant and not the other.
For the first aim what would I perform the chi squared test upon? (I have attached my species observation table and the distribution of bees between the two plants (the network web)). For example, I observed 15 bee species, but only seven were shared between the two plant species. How would I perform the chi squared test on this data. The second aim would be to ask, is it significant that I only found bumble bees (excluding the carder bee) upon one plant.
thank you again for your help.
James Des Lauriers , Andrew Paul McKenzie Pegman i should mention that my observations were conducted in 8 different sites across a region that I am investigating.
Hello Cameron; Your table is arranged in a way that makes a Chi Squared analysis easy. Before you start, it would be good for you to read a formal description of this test so that you know that your data conform to the restrictions of the test. Particularly note that Chi Square calculation uses frequencies only; it never uses percentages or proportions.
My impression of the logic is that your null is that the insects are NOT making any choice and approach the two species randomly. If that is the case, with what frequency would each species visit the plants? Those frequencies are the null. Then the actual frequencies are compared to that.
I’d consider each site separately for a beginning. There may be something inherent in some sites that you aren’t yet aware of. The test might reveal that.
Best regards, Jim Des Lauriers
You calculate the mean abundance of bees across all plant species and this is your expected value. Then you calculate observed minus expected values and use these to calculate the Chi square value. If p is < 0.05, you then can do pair-wise comparisons to see where the differences are :)
Andrew Paul McKenzie Pegman James Des Lauriers thank you both very much for your help, I think i've cracked it now.