Re: NANFA-L-- Diversity Indicies

Bruce Stallsmith (fundulus at
Thu, 29 Sep 2005 13:16:20 -0400

Since I'm the kinda geek who likes haggling over statistics, allow me to
suggest that you use rarefaction to compare species composition between
communities if you have unequal sample sizes. I won't try to descibe it
here. But look for the original description of this approach in: Hurlbert,
S. H. (1971) The non-concept of species diversity: A critique and
alternative parameters. Ecology 52: 577-586. Hurlbert has pissed off lots of
people with his bluntness, but since he's pretty much right and can
demonstrate why, his criticisms are important and useful.

--Bruce Stallsmith
along the Tennessee
Huntsville, AL, US of A

>From: "Todd D. Crail" <tcrail at UTNet.UToledo.Edu>
>Reply-To: nanfa-l at
>To: <nanfa-l at>
>Subject: Re: NANFA-L-- Diversity Indicies
>Date: Thu, 29 Sep 2005 12:44:56 -0400
>Yeah IBI isn't going to work. I didn't describe my situation very well,
>sorry about the argument there.
>So if I'm understanding what I'm reading about Hill's, it will account for
>the discrepantcy in the biomass? I'm just looking for a simple way to
>compare multiple yet similar sites within a watershed, using the fish
>community sampled as the response variable.
>The problem I'm running into is something like this:
>One of my sites consisted of:
>green sunfish(1)
>least darter(95)
>If I caluculate Simpsons on that site alone, I get 1.15, in part, because
>least darter are so proportionately more represented than other species
>you compare it to another replicate site:
>fathead (1)
>...which comes out 1.28.
>The second site is more "diverse"? It's definately more even, but c'mmon!
>And we also loose the fact the least darter is a large proportion of the
>individuals found at the first site, and more aggrivatingly, is a SSC fish.
>The whole diversity issue irons out when all the replicates are taken in
>consideration, but I'd rather drive my point home with a hammer, rather
>some feather-like p value that doesn't make any sense to anyone besides the
>0.05% I work with lol.
>What I'd really like to do is find a way to compare each site against the
>whole of the sample, where each replicate receives a score against the
>whole. I would think there'd be some way to compare stream segments along
>gradient (perhaps by stream order?) but I guess everyone has been too busy
>counting mayflies and snails they can't speciate to look at something
>to identify, like fish.
>I dunno. I dunno if this works for on-list traffic either. I guess I
>rationalize it that if there isn't an easy way to compare parts of streams
>by a response variable such as fish community, then the NANFA list is
>something useful and doesn't require understanding how greek symbols relate
>to each other to figure out if part of a stream fares better than another
>part of the same system. It sure seems can't voice my opinion here at
>school... Apparently masters students are to only be seen and not heard.
>bondage... {obscure x-ray specs reference}
>Thanks for your help and continued tolerance :)
/ This is the discussion list of the North American Native Fishes
/ Association (NANFA). Comments made on this list do not necessarily
/ reflect the beliefs or goals of NANFA. For more information about NANFA,
/ visit Please make sure all posts to nanfa-l are
/ consistent with the guidelines as per
/ To subscribe, unsubscribe, or get
/ help, visit the NANFA email list home page and archive at