Strat. Confidence Intervals

From
Date
Reply-To
I also agree that we've had enough discussion about a problem that, thus
far at least, hasn't been a problem.  PaleoNet was created to enhance
communication among paleontologists (which I define in the broad sense as
any serious student of the science regardless of how he or she makes a
living) about paleontology (which I also define in the broadest possible
sense).  Let's not worry about the cranks until they get here (then we'll
send in the clowns).

Now, about that Paleobiology issue, I agree it's very interesting.  I'd
like to briefly underscore something that Marshall says in his paper on
strat. confidence intervals about interpretations of these mathematical
constructs having to take stratigraphic reality into consideration. I
couldn't agree more.  However, there is a tendency in the strat. confidence
interval literature to make the greatly simplifying, but stratigraphically
unlikely, assumption that lithostratigraphic distance (e.g., between fossil
occurrences) is equivalent to relative time throughout the known range of
the taxon as well as beyond it's last (or first) occurrence.  In order to
accurately estimate the confidence interval using either the Sadler and
Strauss, or the Marshall methods one needs an accurate method of inferring
the gap size distribution.  Since sediment accumulation rates vary,
lithostratigraphic thickness cannot be used for this purpose.  It may be
that in particular sections or cores sediment accumulation rates are indeed
constant enough to allow distance to be used as a proxy for time, but this
must be demonstrated, not assumed.  Moreover, constancy of lithology is not
a valid test for either the continuity of the local stratigraphic record or
the constancy of sediment accumulation rates - as those of us who deal with
deep-sea cores know all too well.  In short, in order to apply the
important insights to be gained through the application of stratigraphic
confidence intervals, we are going to have to become even more concerned
with straightening out the detailed chronostratigraphy of our sections and
cores.  As a stratigrapher, I like the implications of that.

Also, as someone who routinely deals with deep-sea cores I'm more
optimistic that Charles Marshall about the applicability of his methods to
deep-sea microfossil data.  While it is true that the more abundant species
do tend to show up in virtually every sample (implying very tight
confidence limits about the first and last appearances of these taxa) such
species usually make up only a small proportion of the overall taxic
assemblage.  Occurrence data for microfossils tends to follow either a
log-normal or broken stick distribution with most microfossil taxa
exhibiting a distinctly 'gappy' pattern.  Unfortunately, the existence of
these gaps is often obscured in the micropaleo. literature because of the
tradition of drawing lines on our biostratigraphic and relative abundance
charts over the samples contain no data for a particular species.  Now that
we understand that the gap distribution contains information that can be
important we can only hope that this practice will be discontinued.

One other thing, if these gaps are telling us anything about the
spatio-temporal distribution of populations in the past then we should also
be able to test a variety of interesting meta-population models using
fossil data.  That, however, is another story.

Norm MacLeod


----------------------------------------------------------------------------
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
----------------------------------------------------------------------------