Reef-building scleractinian (stony) corals obligately host dinoflagellate algal symbionts (genus Symbiodinium), whose loss during mass coral-reef ‘bleaching’ as a result of increased seawater temperatures is a significant threat to coral reef ecosystems worldwide1. Although acclimatization and/or adaptation of reef coral hosts and/or their algal symbionts are recognized as potentially mitigating the frequency and severity of these bleaching events1,2,3, no studies have been undertaken to test whether such responses actually occur on affected reefs.

The genus Symbiodinium is diverse, and many corals are relatively flexible in the type(s) of algal symbiont they contain, although one type is usually dominant in any given species and environment4,5,6,7. Because corals containing different symbionts can vary in their sensitivity to bleaching4,8 and can modify their symbiont communities in response to environmental change4,7,9,10, we investigated whether severe bleaching and mortality can select for stable host–symbiont combinations that are more thermally tolerant, raising the overall bleaching resistance of the reef as a result.

We undertook molecular surveys of Symbiodinium in shallow (less than 7 m depth) scleractinian corals from five locations in the Indo-Pacific that had been differently affected by the 1997–98 El Niño–Southern Oscillation (ENSO) bleaching event (Fig. 1; for sample details, see supplementary information). For these large-scale comparisons, restriction-fragment length polymorphisms in the symbionts' large-subunit ribosomal DNA were used to distinguish Symbiodinium in clades A, C or D5,9.

Figure 1: Distribution of Symbiodinium algae in shallow-water (less than 7 m depth) scleractinian corals from Kenya, Mauritius, Saudi Arabia and Pacific Panama.
figure 1

Pie charts show distribution of symbionts by site, except those for Panama, which show the distribution of symbionts at a single site before (1995: all colonies healthy), during (1997: some colonies healthy, others severely bleached) and after (2001: all colonies healthy) the 1997–98 El Niño. Size of pie charts is scaled to the square-root of sample size to reflect an equal area for each sample, as indicated by the inset scale. See supplementary information for sampling details for each country.

In Panama, we surveyed ecologically dominant corals in the genus Pocillopora before, during and after ENSO bleaching. Colonies containing Symbiodinium in clade D were already common (43%) in 1995 and were unaffected by bleaching in 1997, while colonies containing clade C bleached severely8. By 2001, colonies containing clade D had become dominant (63%) on these reefs.

We also surveyed corals from the Persian (Arabian) Gulf that experience extremely high seasonal temperatures (>33 °C). Even these reefs were severely bleached in 1998 (ref. 11), when temperatures in some locations exceeded 38 °C. We compared these with corals from the Red Sea that do not experience such seasonal temperature highs (typically 29 °C) and which were relatively unaffected in 1998 (ref. 11). We found that, in 2000–01, Gulf reefs were dominated by the same clade-D symbionts as were found in Panama (62% of scleractinian colonies), whereas Red Sea reefs only rarely contained these symbionts (1.5% of colonies).

This comparison was repeated in the western Indian Ocean in 2000–02: corals from reefs in Kenya that were severely bleached by the combined effects of ENSO and the Indian Ocean Dipole in 1998 were compared with corals from Mauritius that had fortuitously escaped significant bleaching during the same event11. The findings were similar: in Kenya, 15–65% of colonies contained clade D, depending on site, compared with only 3% of colonies in Mauritius.

Taken together, these results indicate that corals containing thermally tolerant Symbiodinium in clade D are more abundant on reefs after episodes of severe bleaching and mortality, and that surviving coral symbioses on these reefs more closely resemble those found in high-temperature environments.

We have yet to determine the processes by which these symbiont shifts occur in coral populations (for example, there could be differential mortality of coral hosts and/or bleaching-induced symbiont change in individual colonies12). It is possible that affected reefs may revert to their original symbiont communities over time if they do not experience repeat warming episodes13. We also recognize that some corals (most notably the genus Porites) remain common on devastated reefs, even though they do not often contain clade-D symbionts. Nevertheless, we propose that the symbiont changes described here are a common feature of severe bleaching and mortality events, and predict that these adaptive shifts will increase the resistance of these recovering reefs to future bleaching.