Post by Joe Botting on Mar 16, 2011 8:09:11 GMT -5
Hi All,
One of the things we're always trying to get a grip on is a good estimate of the true diversity of a fauna. There are all sorts of approaches ranging from crude graphical techniques (species-abundance curves) to complex statistical approaches with all manner of requirements.
We've just been revisiting an old idea, and discovered that someone else has got there first, and actually managed to make it work. There's a paper here, if you can get at it,
www.esajournals.org/doi/pdf/10.1890/07-2147.1
but if not, this is what you do...
You need to spend a good while systematically making a large sample from a site, and reocrding at least which species are present (you don't have to name them as long as you can recognise them). If you only have the list of species and no abundance data, you need at least two samples to duplicate the list. Count up Sobs (the number of species in total), t (the number of samples), Q1 (the number of species occurring in exactly one sample) and Q2 (the number of species occurring in exactly two samples). An estimate of the total diversity is then given by:
Chao2:
Sest = Sobs + (1-1/t)x(Q12/2Q2)
If you've also got abundance data for the rare things, you can do an equivalent calculation for one sample (which can be the total of many collections, or whatever you can sensibly apply it to):
Chao1:
Sest = Sobs + (f12/2f2)
where f1 is the number of species of which only one specimen is known, and f2 the number for which two are known.
Simple, eh? Well, there are problems. It's an ecological technique but it seems to be applicable also to fossils, albeit with some caveats. For example, you're only estimating the number of the species of preservable things for whatever deposit you're looking for, and it's not clever enough to guess at soft-bodied beasties, for example (unless you're studying the Burgess Shale). It will have problems if you have only a small sample to begin with, or if you have few species with only two specimens (small changes in that figure will cause enormous changes in the estimate). Then, of course, you've got the issue of how to deal with fragments of multi-part skeletons, or moults...
Caveats aside, we think it should be done as a matter of course if you're interested in palaeoecology. Diversity is a fundamental aspect of a fauna that is so far poorly understood, and although these estimates are simple and somewhat crude, they give us something to work on... and you'd be amazed at how little of this has been done up to now. We've already tried it on the fauna of one site we've been sampling for years, and got an answer that we have over 80% of the total preservable fauna, and on another that we only just started on and found that we have indeed barely started, and are missing two-thirds of a ridiculously diverse sponge fauna.
So, anyone fancy having a go at these? We'd love to discuss any results!
Example to follow...
One of the things we're always trying to get a grip on is a good estimate of the true diversity of a fauna. There are all sorts of approaches ranging from crude graphical techniques (species-abundance curves) to complex statistical approaches with all manner of requirements.
We've just been revisiting an old idea, and discovered that someone else has got there first, and actually managed to make it work. There's a paper here, if you can get at it,
www.esajournals.org/doi/pdf/10.1890/07-2147.1
but if not, this is what you do...
You need to spend a good while systematically making a large sample from a site, and reocrding at least which species are present (you don't have to name them as long as you can recognise them). If you only have the list of species and no abundance data, you need at least two samples to duplicate the list. Count up Sobs (the number of species in total), t (the number of samples), Q1 (the number of species occurring in exactly one sample) and Q2 (the number of species occurring in exactly two samples). An estimate of the total diversity is then given by:
Chao2:
Sest = Sobs + (1-1/t)x(Q12/2Q2)
If you've also got abundance data for the rare things, you can do an equivalent calculation for one sample (which can be the total of many collections, or whatever you can sensibly apply it to):
Chao1:
Sest = Sobs + (f12/2f2)
where f1 is the number of species of which only one specimen is known, and f2 the number for which two are known.
Simple, eh? Well, there are problems. It's an ecological technique but it seems to be applicable also to fossils, albeit with some caveats. For example, you're only estimating the number of the species of preservable things for whatever deposit you're looking for, and it's not clever enough to guess at soft-bodied beasties, for example (unless you're studying the Burgess Shale). It will have problems if you have only a small sample to begin with, or if you have few species with only two specimens (small changes in that figure will cause enormous changes in the estimate). Then, of course, you've got the issue of how to deal with fragments of multi-part skeletons, or moults...
Caveats aside, we think it should be done as a matter of course if you're interested in palaeoecology. Diversity is a fundamental aspect of a fauna that is so far poorly understood, and although these estimates are simple and somewhat crude, they give us something to work on... and you'd be amazed at how little of this has been done up to now. We've already tried it on the fauna of one site we've been sampling for years, and got an answer that we have over 80% of the total preservable fauna, and on another that we only just started on and found that we have indeed barely started, and are missing two-thirds of a ridiculously diverse sponge fauna.
So, anyone fancy having a go at these? We'd love to discuss any results!
Example to follow...