Centres of Endemic?

[Don McAllister]

Norbert Hahn wrote:

> Dear All,
>
> I am sitting with a perplexing problem of trying to define the term "centre
> of endemism"!

This is a fundamental problem, whether one is trying to identify centres of
endemism, species (or other levels/clades of taxa richness - biodiversity), or
areas rich in crop origins (Vasilov centres).

One wants to count the number of taxa in different areas and find which areas
are richer - 'centres.'  The problem lies in choosing the working area units for
which one does the counts.  Most, like countries, states, counties, islands,
lakes and ecozones of various sorts, are unequal in size.

One can make a correction, species per unit area, for example, or even plot
species-area curves (or straighten the curve with a log-log graph) and see
whether points lie above or below the line of best fit.  But a bigger geographic
unit, like the United States, may conceal a centre within its boundaries and its
large size may dilute the average number per unit area, so that a centre will be
undetected.  Freshwater mussels and fishes, for example, are very rich in the
Cumberland-Tennessee drainages, while most of western U.S. is species poor. So
the use of big units can conceal species-rich areas.

The second problem with using geopolitical units is that rich areas may straddle
the boundaries.  To what degree do species-rich areas in the Mekong River basin,
for example, straddle the boundaries of countries.

One solution to identifying species hotspots, centres of endemism, etc., is to
use relatively small equal-area grids on which to map distributions of species
and subsequently count how many occur in each grid cell.  If the cells are small
enough, then they will probably be smaller than the 'hotspot' or centre.  And if
the data is good enough - both detailed and evently spaced, then one should be
able to detect various centres using a consistent tool.

So the excellence of your data may impose limits on how finely you will be able
to resolve patterns - unless you are willing or able to go out and fill in gaps
in your sampling.  Most of us do have gaps imposed by road networks, ship
tracks, expeditions, difficulty of terrain, lack of budgets to hire helicopters,
etc.  The ideal is to carry out biological surveys themselves which are based on
fine grid patterns.  Then your data will show better where the species are - an
not the highways!

Fred Schueler, myself and various co-authors tried this approach first in a
paper on freshwater fishes of U.S. and southern Canada, using one degree
latitude-longitude 'squares.'  It nicely located species-rich hotspots, though
it suffered from the fact that lat-long squares diminish in size towards the
poles.

Our second paper used truly equal areas, in a grid of our own devising.  In that
study we sought various centres for a sizable world coral reef fish data base
developed by Callum Roberts and Julie Hawkins.  Moreover, we were able to define
'endemic' in terms of the number of grid cells occupied - thus removing the
second variable from the traditional approach. You can find this second paper in
a book, Mapping the Diversity of Nature, edited by R.I. Miller (Chapman and
Hall, 1994).  So that study let us look for hotspots in richness of all species
(or by family) as well as endemic species.

D. Rathbert et al (1999), Environmental correlates of species richness for
native freshwater fish in Oregon, U.S.A. J. Biogeogr. 24: 257-273, used an
equal-area grid of 375 hexagons that covered Oregon, to analyze data on
freshwater fish distribution and variables like temperature, presence of exotic
species, stream density, and basin connectivity.

don
Don McAllister