Ecology Letters, (2005) 8: 643–651
doi: 10.1111/j.1461-0248.2005.00763.x
LETTER
Environmental and geometric constraints on Indo-Pacific coral reef biodiversity
D. R. Bellwood,1* T. P. Hughes,1 S. R. Connolly1 and J. Tanner2 1
Centre for Coral Reef
Biodiversity, Marine Biology Department, James Cook University, Townsville, Qld. 4811, Australia 2
South Australia Research and
Development Institute, PO Box 120, Henley Beach, SA 5022, Australia *Correspondence: E-mail:
[email protected]
Abstract The Indo-Australian Archipelago supports the world’s richest coral reef biodiversity hotspot. Traditional hypotheses that account for such exceptional biodiversity have highlighted the importance of environmental variables such as habitat area and energy input. Recently, however, an additional explanation has been proposed based on geometric constraints in the placement of geographical ranges within a bounded domain, which cause a mid-domain peak in species richness; the mid-domain effect (MDE). Here, for the first time, we examine the relative importance of area, energy and MDE jointly on species richness patterns. Model selection indicates that the best model incorporates MDE and reef area, but no energy effect; moreover, this best-fit model captures all major features of reef fish and coral species richness patterns. Habitat area is the major environmental factor influencing species richness. The prevention of further fragmentation and loss of habitat area is of critical importance for the conservation of coral reef biodiversity. Keywords Biodiversity, coral reefs, fishes, hot spots, mid-domain effect, spatial models. Ecology Letters (2005) 8: 643–651
INTRODUCTION
One of the central goals in ecology and biogeography is to explain variation in species richness. This question has a long history and has spawned numerous hypotheses (Whittaker et al. 2001; Hawkins & Diniz-Filho 2002; Pimm & Brown 2004). Global biodiversity patterns and latitudinal gradients, in particular, have received considerable attention, especially in recent years. Detailed reviews of the topic have highlighted the need to increase the scale of studies, to look at global rather than just regional processes, to go beyond single taxon patterns, to consider evolutionary histories and to adopt a more pluralistic approach rather than a traditional division into ecological or historical explanations (Willig et al. 2003; Ricklefs 2004; Wiens & Donoghue 2004). Here, for the first time, we conduct a series of analyses to evaluate the relative importance of environmental factors (relating to habitat area, historical habitat availability, and the extent and variability of energy input), and of geometric constraints in range locations (the mid-domain effect, MDE) as determinants of regional scale biodiversity patterns and species richness on coral reefs. Of the numerous hypotheses examined in previous studies of coral
reefs, two variables have been identified as leading contenders in explaining regional-scale biodiversity: energy and habitat area (Fraser & Currie 1996; Bellwood & Hughes 2001). These two variables reflect two alternative general hypotheses that have emerged from studies of global-scale biodiversity gradients in other ecosystems. The energy hypothesis predicts that high and consistent levels of energy supply support high species richness. This idea has received wide support, particularly at large spatial scales (Roy et al. 1998, 2000; Mittelbach et al. 2001; Allen et al. 2002; Harmelin-Vivien 2002; Hawkins et al. 2003). The area hypothesis highlights the importance of habitat area: across all scales, the greater the area, the higher the number of species that can be supported (Raup 1972; Rosenzweig 1995, 1999). Historical area availability may also be important, by promoting speciation or inhibiting extinction (Rosenzweig 1995; Wiens & Donoghue 2004). In the last decade an additional, and controversial, hypothesis has been proposed based on the observation that randomized distributions of the locations of geographical ranges generate a peak in species richness in the middle of a geographical domain, the MDE (Colwell & Hurtt 1994; Colwell & Lees 2000). This peak arises because the 2005 Blackwell Publishing Ltd/CNRS
644 D. R. Bellwood et al.
30 250
500 250
30 30
c
o 14
18
22
26
30
Figure 1 Map of the Indo-Pacific domain showing the longitudinal and latitudinal mid-domain (white cross, midway between Africa and the
Americas, centred on the equator) and 67 study locations (white circles), which were chosen to provide a wide range of species richness. Two important environmental factors are indicated: reef area [scaled into four categories (>3%, 2–3%, 1–2% and 99.7% support) (Table 1). When MDE, reef area and mean SST were combined in a spatial multiple regression analysis, model selection by AIC indicated that reef area + MDE is the best model, for both corals and fishes (Table 2), and likelihood ratio tests yielded the same conclusions (Supplementary Material). The trend component of the MDE + area models explained 79% of the variation in species richness in fishes and 71% in corals. A two-dimensional representation of the predicted species richness of fishes and corals based on these models (represented by contours fitted to the predictions for each of the 67 sites, Fig. 3) captures all major characteristics of observed fish and coral distribution patterns remarkably well.
Figure 2 Proportion of variation in reef fish and coral species richness at 67 sites, explained by the mid-domain effect (MDE) and eight
environmental variables that relate to energy or habitat area. ND, normalized distance; GA, great arc distance; A1, reef area; A2, area of substratum 0–30 m deep; A3, area of substratum 30–200 m deep; E1, mean annual sea surface temperature; E2, solar irradiance range; E3, sea surface temperature range; E4, mean solar irradiance; E5, productivity. *** P < 0.001; **P < 0.01; *P < 0.05. Dark bars indicate variables selected for additional analyses. 2005 Blackwell Publishing Ltd/CNRS
Global coral reef biodiversity 647
Table 1 Selection of the best energy, area,
and mid-domain variables from regression models incorporating spatial autocorrelation
Fishes
Corals
Model set
Model
AIC
% Support
AIC
% Support
Energy
Mean SST SST range Productivity Solar mean Solar range Reef area Shallow Deep ND GAD
836.4 855.0 857.7 856.8 847.6 842.6 854.3 855.9 834.2 854.0
>99
