Baker replies

Because corals that were transplanted downwards did not bleach in response to reduced irradiance, they failed to exchange their 'high-light' algal symbionts for the more suitable 'low-light' algae that were already found in the deep-water colonies at this site (and/or at other sites nearby). As a result, they contained inappropriate algae for their new environment, which led to chronic stress and eventual mortality.

In contrast, corals that were transplanted upwards experienced severe bleaching as a result of increased irradiance. Consequently, suboptimal low-light algae were removed, allowing high-light algae to become dominant in the newly vacant hosts. Such corals survived well as a result, despite their initial bleaching. This explanation is particularly powerful because it unifies coral bleaching, symbiont change and host mortality.

Hoegh-Guldberg et al. suggest that my findings fail to support the ABH because they do not provide evidence of 'new' symbionts in transplanted corals. The ABH is not limited to this constraint. Regardless of the origin of replacement symbionts (which, as I pointed out, may "colonize" and/or "proliferate inside" hosts) or the proximate environmental causes of bleaching (for example, light or temperature), if bleached reef corals change the composition of their symbiont communities faster than unbleached corals, and if more rapid symbiont change proves beneficial, then bleaching has adaptive value. Even if adult colonies are unable to form symbioses with unusual or new algae (which is unlikely, given the recent discovery of some scleractinian coral colonies containing symbionts that are usually found in foraminifera1), cryptic populations of diverse symbionts may still occur at low abundance in many coral hosts2.

There is no field evidence that symbiont genotypes change after bleaching events because the necessary molecular investigations have not yet been undertaken. Despite this, one of the best available long-term data sets on mass coral bleaching and mortality reveals that far fewer corals in the far-eastern Pacific Ocean died after the 1997–98 El Niño event (0–26%) than after the 1982–83 El Niño event (52–97%; ref. 3), even though the magnitude and duration of sea-surface temperature anomalies in the region in 1997–98 exceeded those of 1982–83 (ref. 4). These observations indicate that surviving reef corals may be more resistant to recurrent thermal stress through having experienced earlier episodes of severe bleaching and mortality, as predicted by models of symbiont change5.

Furthermore, field experiments with bleached corals6 and laboratory studies of model invertebrate–algal symbioses7 support some of the assumptions of the ABH. We should not mistake an absence of evidence for evidence of absence, and instead need to document worldwide patterns of coral–algal associations and their response to mass-bleaching events. The real question is not whether coral–algal associations can adapt by recombining, but rather how, and over what timescales, they do so.

Although episodes of mass coral bleaching and mortality will occur in the future, my findings suggest that they may not recur with the frequency and severity predicted by some studies8. This should stimulate efforts to protect the remaining three- quarters of the world's coral-reef ecosystems9 by reducing the compounding effects of anthropogenic factors that are still under our influence.