A new botany?

John Ladasky jladasky at pmgm.Stanford.EDU
Mon Nov 23 18:23:17 EST 1998

In article <gPS1GDAmedW2EwED at wharfe.demon.co.uk>,
David Brear  <dbrear at wharfe.demon.co.uk> wrote:
>In The Independent today there is an article which suggests that
>our previously understood relationships between plants are all

	Well, maybe not *all* of it... but perhaps a large part of
our understanding was wrong.

>Dr Mark Chase of Kew, Kare Bremer of the University of
>Uppsala in Sweden and Peter Stevens of Harvard have
>investigated the genetic links of plants. The work 'involved the
>detailed comparison of three genes for each of 565
>representatives of all the families of flowering
>Maybe this is old news to many readers but what do you think?

	It's not old news to me.  However, the method of analysis is
somewhat familiar.  I'm an immunologist, but my recent research has
led me to ask some questions about evolution -- not of plants, but of
monkeys.  The issues involved in analyzing DNA sequences are believed
to be the same between all organisms.

>The original articles (two in one day's paper - well done The
>Independent!) are at 

	I just had a look at this.  Good article, but I think that
there's a little hype.

>All opinions welcome!

	I will be interested to see the original study when it comes
out, and I should probably reserve my comments until then.  But, since
you asked, and since science is such a self-critical profession (es-
pecially when you're criticising someone else work!), and since this
is Usenet...

	How technical do you want me to get?

	In order to draw a good genetic tree from DNA data, you need
two things: you need the data from your sequences to be similar enough
so that it's obvious how to align them, but you also need differences
because that's how you tell them apart.  As a rule of thumb, the more
different two sequences are, the less related they are.  Some portions
of a DNA sequence can change and have no effect on the function of the
protein that it encodes, while other mutations do affect function.
The latter category of mutations will be subject to natural selection.
The former category are considered to be genetically "neutral."  

	Neutral DNA sequences are considered to be the best place to
look for evolutionary data, since the mutations that occur there can
safely be considered to be random.  Only trouble is, a long stretch
of neutral DNA is also dispensable.  As an organism evolves, it can
delete these chunks of neutral DNA.

	I believe that three genes will not provide enough data to
draw a truly robust genetic tree for flowering plants.  There are two
main reasons that I think this is true.   The first is that the 
study is actually looking at *genes*, like RUBISCO.  This enzyme is a 
highly-tuned machine, and a lot of the mutations are probably not neu-
tral.  Any mutation that makes RUBISCO work poorly will be quickly 
eliminated from the genetic record.  Second, the genetic history of
flowering plants is *very* long, hundreds of millions of years.  This
means that there has been a lot of time for evolution to occur.  It
is almost certain that any mutations observed have occurred multiple
times.  This means that bases in the DNA sequence may have changed,
and then changed back!

	A conclusive paper establishing once and for all that chimp-
anzees are closer living relatives of humans than gorillas was just
published last year.  In order to establish this with a probability
of >99%, fourteen different stretches of DNA were examined in all the
primates, most of them neutral.

	I will guess that at least a similar number of genes will be
needed in order to establish the relationships among flowering plant
families.  Just finding that many sequences common to all flowering
plants will not be easy!  And then, within families?

>David Brear
>The Wild Flower Page
>Wild Flower News

Rainforest laid low.
"Wake up and smell the ozone,"
Says man with chainsaw.					- John Ladasky

More information about the Plantbio mailing list