[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Subject Index][Author Index]

Re: advice for the undergrad

At 8:34 PM +0200 7/5/08, Michiel Pillet wrote:
>I was also looking at some computer science courses, but... Why is it so
>important to take a computer language?

One reason is that paleontology is a very data-rich science, and with data
often comes the need for statistical analysis. If your data are from some
routine situation for which there is a standard statistical test, then you
probably won't need to do any programming; you'll probably be able to do
the analysis in any statistical software package.

But if you have data that don't follow common situations, or if you are
asking unusual questions, there may not be any existing analysis that is
appropriate. In that case, you'll have to write your own analysis from
scratch, and programming skill will certainly come in handy.

Just off the top of my head, here are some examples of novel analyses that
required substantial programming:

1) Alroy et al, 2008 in Science: This paper, which just came out two days
ago, is the culmination of over a decade of work. How many species were
there on earth during the Cambrian? What about the Permian, or the Eocene?
You can just count the number of taxa known from the fossil record, but
that is subject to numerous biases (e.g., the amount of exposed rock from
each interval, the number of paleontologists working on each interval,
etc.). Alroy's analysis attempts to account for these factors in order to
estimate the true trajectory of life over the Phanerozoic.


2) Similarly, we may want to estimate the extinction and origination rate
at any time during earth history (e.g., what percent of species went
extinct in the Permian extinction? How does that compare to the Cretaceous
extinction?) These questions are subject to similar biases as in (1), plus
the Signor-Lipps effect (incorrect estimation of originations and
extinctions due to missing data). Mike Foote used the statistical
programming language R to run his novel analyses.

3) How do extinctions at one level of a food web affect the other levels of
the web? If you kill off, say, 40% of primary producers, what will happen
to the herbivores and carnivores? Peter Roopnarine used C++ to create a
food web model with which you can manipulate selected guilds in a food web
and see how extinctions cascade through the web.


All of these analyses would not have been possible without extensive
programming to implement the authors' novel ideas.



Steve C. Wang
Associate Professor of Statistics
Swarthmore College