Re: Nature of the fossil record

From

Date
Reply-To

Both Norm and Doug have addressed what I feel to be the major issue with the
S-L effect. The S-L effect is driven by the absence of data. Why these data
are absence results from two sources - one of which can be "fixed", the other
cannot - and as students of history we must learn to live with. (1) Some
portion of the 0's in the taxon/time matrix result from sampling error - we
simply have not mined all the Cretaceous rocks on the earth and therefore do
not really "know" where the last "knowable" data point lies. (2) Bias produces
other 0's in the taxon/time matrix - the taxon was differently removed by
predators, wave action, geo-cemical factors; it wasn't in the right habitat,
it was really extinct,... The important point with (2) is that it is bias,
and in a statistical sense unknowable. Even if we reduced sampling error to 0,
we would still have bias in our data and a S-L effect. In the real world we willprobably never be able to reduce sampling error to 0 and we cannot know what
the bias component is. It is not pretty. Charles Marshall has a proceedure to
estimate confidence limits for the last occurences of a taxon approaching an
event (GSA, 94), but it requires continuous rather than interval data to be
valid. And again, these produce null models which can be falsified.

D.R. Lindberg
davidl@ucmp1.Berkeley.Edu

I really don't want to belabor this point (and I certainly don't want Doug
to feel like he's being picked on) but Doug's comment shows just how
insidious the S&L Effect is and how easy it is to see whatever we want to
see in it's image. Is the Signor-Lipps Effect PRIMARILY a preservation
phenomenon? Any particular Signor-Lipps pattern might be due to
differential preservation. But then again it might be due to sample
size/relative abundance interactions, or ecological changes. That's the
problem with negative evidence. You simply can't know precisely why
something is not there. Signor and Lipps (1982) list three sources for the
effect without indicating which one is dominant. Within it's domain the
S&L Effect is simply a statement of uncertainty. But there must be a
domain outside of which the S&L Effect can be ignored. If this isn't true
then I guess I don't understand how biostratigraphy is possible.

Norman MacLeod
Senior Research Fellow
N.MacLeod@nhm.ac.uk (Internet)
N.MacLeod@uk.ac.nhm (Janet)

Address: Dept. of Palaeontology, The Natural History Museum,
Cromwell Road, London, SW7 5BD

Office Phone: 071-938-9006
Dept. FAX: 071-938-9277