For my study I am comparing diversity of mangrove epibiont communities among estbalished, reforested and deforested mangrove sites.
Asking a question to the theoretical background of the Shannon Index: In which extend can I use the comparison of Shannon-Indeces in-between sites with different environmental settings ? Do habitats need to be similar to a certain degree to apply and compare the Indeces
Hi Verena,
This is a very good question, and one that people don't often think about. The Shannon index can be difficult to interpret, so comparisons between habitats may or may not be useful. While making comparisons, it may be beneficial to also think about (and compare) species dominance, evenness, and a more basic measure like richness.
If you want to do direct comparisons of Shannon diversity, keep in mind that this index is in a log scale. Thus, it may make sense to take the exponent of the index for comparisons. There is some information on this on the internet, as well as in one of my papers (was not species diversity, but was Shannon index) (http://www.int-res.com/abstracts/meps/v499/p1-18/).
Finally, it may make sense to look at similarities, rather than diversity. For this you can use multivariate approaches such as NMDS, ANOSIM, PERMANOVA, etc.
Cheers and good luck,
Dan
Dear Verena,
you should know that this index function actually consisted from two arguments - species amount and equitability (evenness). It is like traffic into the city – can be over one road or by 10 or more, and maximum when all these flows of cars is equal. Really its origin from Bolzman'z formula of entropy from Physics look Wikipedia.
Your question just into the target. What is the main aim of investigation? Important is if you are going to use structural or physiological indices. In first case to have appropriate results should compare ecosystems or communities of the same level (order). But anyway it produce hardly interpretative results. So, I recommend you to use S.Hurlbert equation (The Nonconcept of Species Diversity: A Critique and Alternative Parameters .Ecology, Vol. 52, No. 4 (Jul., 1971), pp. 577-586
Andrey
Dear Andrey, thank you for you help, this is the kind of information I am looking for..
The biggest challenge i see here (If I understand your question correctly) is the error distribution for the diversity index. We can assume that the diversity index has normally distributed error so using classical method of Anova and the like will not be appropriate. For you to test if these habitats have different indices you will need to use a bootstrap method to get the confidence interval for the index of each habitat and then check if these confidence interval overlap.
The biggest challenge i see here (If I understand your question correctly) is the error distribution for the diversity index. We can assume that the diversity index has normally distributed error so using classical method of Anova and the like will not be appropriate. For you to test if these habitats have different indices you will need to use a bootstrap method to get the confidence interval for the index of each habitat and then check if these confidence interval overlap.
In my opinion a difference between several types of diversity might clarify your data somewhat. You have the alpha- diversity: diversity within 1 plot (in your case 1 site); beta-diversity: diversity between different plots (2 sites); gamma diversity: diversity in the whole area (all sites)
(Over-simplification in figure)
Thanks for all the answers, that were really helpful! However, as my online research and the discussions show, habitat difference is a large confounding factort that needs to be considered applying the Shannon Index to any data analysis. I think this is often overlooked in many studies!!!
There are good and bad diversity measurement concepts and the distinction between them is determined by properties, contexts and individual judgements.
I suggest you try to asses the "true" species diversity, using species richness and the effective number of species, sensu Jost (2006).
Jost, L. 2006. Entropy and diversity. - Oikos 113: 363–374
Best regards
- Badges
- Science method