133845.pdf

180 PAUL J. MARKWICK

(habitat 'islands'), although few habitat 'islands'

are as absolute as ocean islands (Rosenzweig

1995).

The association of areas of high species diver-

sity with low latitudes, high temperatures and

high incident solar energy fluxes, has given rise

to the widely promoted species-energy theory

(Pianka 1966; Currie 1991; Rohde 1992; Stevens

1992; Currie & Fritz 1993; Latham & Ricklefs

1993; Wright et al. 1993; Brown 1995; Rosen-

zweig 1995). Currie (1991), comparing North

American terrestrial species diversity against 21

environmental variables, found that species

diversity correlated best with potential evapo-

transpiration (PET), considered to be an appro-

priate measure of ambient energy. However,

Kerr & Packer (1997) found that this was only

true for mammals in North America where PET

< 1000 mm a-

1

(Canada and Alaska). Stevens

(1989) has suggested that climate (energy) vari-

ability is the most important determinant of

species numbers, with few taxa being able to

survive large seasonal variations. Such taxa

should therefore have the largest latitudinal

ranges (Rapoport's rule), but this pattern has

not been found in all groups (Rohde et al. 1993;

Roy et al. 1994). For many this theory remains

equivocal (Pianka 1966; Rohde 1992; Roy et al.

1994; Brown 1995; Rosenzweig 1995; Kerr &

Packer 1997) because not all groups show clear

diversity gradients and the mechanism by which

energy can dictate the number of species is

uncertain.

The problem is that individual species

respond to different environmental factors (and

combinations of factors) and to different extents

depending on their physiology and ecology

(Root & Schneider 1993). Measuring only the

number of species (taxonomic diversity) rather

than the distribution of differences between

organisms (functional and ecological diversity)

may obfuscate the processes dictating diversity

patterns (Gaston 2000). It is therefore essential

to be able to examine the macroscale spatial

structure of diversity in the context of physi-

ology and behaviour (what an organism does),

as well as taxonomy (what an organism is

called), although for traditional classification

methods there is often considerable overlap

between the two. What is more, species richness

does not vary only with latitude (Brown 1995),

nor is it independent of history (evolution and

palaeobiogeography). To understand global

patterns therefore requires large, intercontin-

ental datasets. Unfortunately, there have been

few studies at this scale, and these are restricted

to analyses at relatively coarse taxonomic levels

and resolutions/grain (e.g. Gaston et al. 1995).

This paper presents a new, digital, geographic

information system-based dataset with which

the relationships between present-day terres-

trial biodiversity, biogeography and climate are

examined. Examples of the observed macro-

scale modern patterns are illustrated (both as

maps and bivariate plots) in order to facilitate

comparison with previous studies of North

America, especially that of Currie (1991). The

consequences of derived relationships for

interpretations of palaeoclimate and palaeo-

ecology are discussed. As a test of Ostrom's

(1970) suggestion of using diversity gradients

to retrodict palaeoclimate, an experiment is

presented in which the Middle Eocene

palaeotemperature of Messel, Germany, is

reconstructed using modern-day regressions

between observed taxonomic diversity and

temperature, and the results compared with

values obtained from other methods.

Methods

The dataset used in this study is part of a large

computer-based ecological database of fossil

and modern faunal and floral localities compiled

by Markwick (1996). The database is designed to

facilitate analysis at any specified taxonomic

level, such that differences between the

response of families, genera and species can be

systematically analysed. Data can also be

queried for any combination of parameters

included in the database. Since this study was

begun in 1990, the database has been integrated

into a geographic information system (ArcView

GIS and Arclnfo). The fossil data include

detailed specimen, environmental and strati-

graphic information on about 6000 Cretaceous

and Cenozoic fossil vertebrate localities, with

taxonomic and ecological data for 22 000 extant

and fossil vertebrate and floral taxa (including

habitat, size and diet). The modern data

analysed here draw on the climate information

from 1060 climate stations (Fig. 1) taken from

Muller's (1982) compilation for vegetation

studies. Each station contains monthly data for

14 climate variables including mean daily tem-

perature, mean precipitation, radiation and

potential evapotranspiration (PET). A large

number of additional parameters, including

annual metrics and combinations of variables,

have been calculated using these data. A list of

variables mentioned in this paper, their abbrevi-

ations and explanations are given in Table 1. The

propensity for stations to occur in lowland sites

reflects Muller's (1982) original requirements:

acceptable stations must contain data for a large

array of climate parameters representing time