Opposing forces of aerosol cooling and El Niño drive coral bleaching on Caribbean reefs
- Centre for Ecology, Evolution, and Conservation, Schools of *Biological Sciences and
- §Environmental Sciences, and
- †Tyndall Centre for Climate Change Research, University of East Anglia, Norwich NR4 7TJ, United Kingdom
-
Edited by Hans Joachim Schellnhuber, Potsdam Institute for Climate Impact Research, Potsdam, Germany, and approved October 25, 2006 (received for review September 26, 2006)
Abstract
Bleaching of corals as a result of elevated sea surface temperatures (SST) is rapidly becoming a primary source of stress for reefs globally; the scale and extent of this threat will depend on how the drivers of SST interact to influence bleaching patterns. We demonstrate how the opposing forces of the El Niño–Southern Oscillation (ENSO) and levels of atmospheric aerosols drive regional-scale patterns of coral bleaching across the Caribbean. When aerosol levels are low, bleaching is largely determined by El Niño strength, but high aerosol levels mitigate the effects of a severe El Niño. High aerosol levels, resulting principally from recent volcanic activity, have thus protected Caribbean reefs from more frequent widespread bleaching events but cannot be relied on to provide similar protection in the future.
Many of the world's coral reefs are severely degraded as a result of regional pollution, overfishing, and land-based development activities (1), and the incidence of widespread coral bleaching is rapidly increasing (2–5). The evidence linking anomalously high sea surface temperatures (SST) to coral bleaching is clear (3, 6). Unusually high temperatures during the summer months result in the expulsion of symbiotic dinoflagellate algae from the coral host (resulting in a characteristic whitening), which may or may not result in the coral acquiring other more stress-tolerant strains of algae (7, 8). This mechanism is risky, however, and does not guarantee the coral's survival. For example, an estimated 16% of the world's coral reefs were severely damaged or killed as a result of exceptionally high SST during the strong 1997–1998 El Niño event (2).
The El Niño–Southern Oscillation (ENSO) is a natural irregular oscillation of the tropical ocean-atmosphere system, between a warm (El Niño, positive ENSO phase) and cold (La Niña, negative ENSO phase) state of the tropical Pacific Ocean, that can influence the global climate (9). Over the past 25 years, there have been several strong El Niño events that have been linked to “mass” coral-bleaching events, particularly in the Pacific (4, 5, 10). In the tropical North Atlantic, weakened northeast trade winds associated with strong El Niño conditions lead to unusually warm SST by reducing the rate of evaporation (11). The thermal inertia of the ocean means that SST variability in the tropical North Atlantic lags that in the Pacific by 4–6 months (11–13). However, although El Niño-related mass bleaching was reported across the Caribbean in 1987 and 1998 (14, 15), very little bleaching was reported in the region during the El Niño episodes of 1982–1983 and 1991–1994, although bleaching was reported in the Pacific Ocean (16, 17).
Whereas El Niño events have been linked with increased thermal stress on coral reefs over the last 3 decades, high levels of atmospheric aerosols have been linked to reduced SST (18, 19). High atmospheric aerosol concentrations can result from events such as major volcanic eruptions, which inject vast quantities of ash and sulfur dioxide into the lower stratosphere, where the latter reacts with water vapor to form sulfuric acid aerosol vapor that may persist for several years. The resultant haze scatters incoming solar radiation, and there is evidence documenting significant regional and global cooling after the eruptions of El Chichón (Mexico) in 1982 and Mount Pinatubo (Philippines) in 1991 (19–21). Even in the absence of volcanic eruptions, high concentrations of tropospheric aerosols such as dust carried from African deserts have been shown to cause cooling of subtropical Atlantic SST (6) and have other climatic impacts (22). Although the regional cooling effect of such aerosols is generally only on the order of 1–2°C (17), the sensitivity of corals to even slight increases in temperature (3, 4) may mean that ENSO-related warming (and therefore the potential for coral bleaching) in the Caribbean could be significantly moderated by the presence of dust and volcanic aerosols in the atmosphere, especially because the atmospheric dust loading of the Northern Hemisphere is approximately twice that of the Southern Hemisphere (23). Atmospheric aerosols could reduce bleaching both directly, through reductions in irradiance levels, and indirectly, through reductions in SST as a consequence of reduced irradiance levels. Consideration of the impacts of ENSO events and atmospheric dust on coral bleaching is further complicated by the observation that major Saharan dust peaks appear to be associated with major El Niño events (22).
Here, we take advantage of recent high variability in atmospheric aerosol levels (largely through volcanic activity) and ENSO frequency and intensity to assess the impact of the opposing forces of atmospheric dust and ENSO events on coral bleaching at a regional scale. We use the multivariate ENSO index (MEI) (24), which is a weighted average of the six major ENSO features (sea-level pressure, zonal and meridional components of surface wind, SST, surface air temperature, and total cloudiness fraction of the sky), and aerosol optical depth (AOD) (25, 26) to quantify the total amount of aerosols and relate these to the geographic extent of coral bleaching in the Caribbean from 1983 to 2000.
Results
Temporal Trends in Climate, Aerosols, and Coral Bleaching.
Since 1983, SST over the Caribbean region have increased significantly (r = 0.58, n = 18, P < 0.02), whereas the strength of ENSO events and values of AOD have varied annually but with no clear temporal trend (Fig. 1 a–c). Reports of coral bleaching from throughout the Caribbean between 1983 and 2000, collated for all 1° latitude/longitude cells containing coral reefs (Fig. 1 d), also show no clear trend but five distinct peaks in 1987, 1990, 1995, 1998, and 1999.
Temporal variation in SST, associated warming and cooling forces, and coral bleaching across the Caribbean between 1983 and 2000. (a) Mean August–October SST anomaly from the 1961–1990 base period [MOHSST6; U.K. Met Office Historical Sea Surface Temperature (27)]. (b) ENSO strength (www.cdc.noaa.gov/people/klaus.wolter/MEI/mei.html). (c) Mean August–October AOD (http://gacp.giss.nasa.gov/data_sets) over the Caribbean region (except for 1994 where data were insufficient). (d) Percentage of Caribbean 1° longitude/1° latitude cells containing coral reefs (total n = 92) in which coral bleaching was reported at least once, during August–October.
Drivers of SST Anomalies and Coral Bleaching.
Regression analysis was first used to relate annual variation in SST anomalies across the Caribbean to the bimonthly ENSO strength (MEI) of Jan/Feb and mean August–October AOD data over the Caribbean region. SST anomalies were not significantly related to the strength of ENSO events (r 2 = 0.11, n = 17, P = 0.18), but were weakly related to AOD levels (b = −0.14, r 2 = 0.3, n = 16, P = 0.03). However, AOD and ENSO strength together were highly significant predictors of SST anomalies (r 2 = 0.62, P < 0.001; AOD: b = −1.96, P < 0.001; MEI: b = 0.15, P < 0.005), indicating that SST values are highest when the ENSO is positive and AOD levels are low, but across the spectrum of ENSO values, SST values are low when AOD levels are high. Because positive SST anomalies are linked to coral-bleaching events (3) (Fig. 1), we then used regression analysis to relate variation in the geographic extent of coral bleaching over the Caribbean region, measured as the percentage of coral reef cells in which bleaching was reported (C) to the same MEI and mean AOD data as above. The extent of bleaching was significantly, negatively related to AOD levels (Fig. 2) but not to ENSO strength (r 2 = 0.12, n = 18, P = 0.16). However, when both predictors were included in a multiple regression model, both significantly influenced bleaching, indicating that once the effect of AOD was controlled, high values of MEI (indicating strong El Niño conditions) did result in significant bleaching (log10 C = [log10 AOD × −3.00] + [MEI × 0.23] − 1.47; r 2 = 0.70, n = 17, P < 0.001). The partial correlation between bleaching and MEI is illustrated in Fig. 3. These analyses can be used to predict the level of bleaching that would be expected to have occurred in the absence of high AOD levels; in the 5 years in which volcanic activity resulted in high AOD levels (1983–1984 and 1991–1993), the spatial extent of bleaching would have been ≈5-fold higher in the absence of these high aerosol loadings.
Effect of atmospheric dust on the extent of coral bleaching in the Caribbean. The extent of coral bleaching during August–October in the Caribbean declines with increasing concentrations of atmospheric aerosols (measured as AOD) over the region (log10 C = [log10 AOD × −2.10] − 0.74; r 2 = 0.33, n = 17, P = 0.016).
Partial effect of ENSO strength on coral bleaching in the Caribbean. The residual variance of the relationship between bleaching extent and AOD (Fig. 2) is positively related to the MEI (partial correlation r = 0.61, n = 17, P = 0.001).
Discussion
This work quantifies how aerosols and ENSO events combine to drive regional-scale bleaching of corals in the Caribbean, which has not been described previously. Over the last 2 decades, atmospheric dust has had a greater modifying impact on coral bleaching in the Caribbean than positive ENSO events. Variation in the MEI from 1983 to 2000 has produced an 8-fold variation in spatial extent of coral bleaching. In contrast, variation in atmospheric dust concentrations during the same period has produced a 19-fold variation in bleaching extent (Fig. 2). This variation is influenced greatly by the volcanic activity that occurred during this period; when we excluded those years where major volcanic eruptions affected the amount of dust in the atmosphere (1983–1984 and 1991–1993), variation in atmospheric dust produced only a 4-fold variation in extent of coral bleaching (although the direction and statistical significance of the effects of AOD and MEI on the extent of bleaching were unchanged). The eruptions of El Chichón (Mexico, 1982) and Mount Pinatubo (Philippines, 1991) were thus fortuitous events that caused a substantial reduction in the thermal stress affecting coral reefs in the Caribbean during strong ENSO conditions. The timing of these eruptions during a period of major Caribbean coral reef decline (28) was particularly remarkable because there have been only three other similar eruptions since the late 19th century (20).
In addition to these global influences on aerosols, concentrations of atmospheric aerosols may be generally greater over Caribbean reefs because of the transport of millions of tonnes of dust from the African Sahara and Sahel annually (29, 30), in contrast to the isolated oceanic coral reefs of the Pacific and Indian oceans. We would consequently expect the relationship between positive ENSO and coral bleaching in the Southern Hemisphere to be more direct. Note that dust from the Sahel has been implicated in the transport of pathogens of coral reefs across the Atlantic (29, 30); thus, atmospheric aerosols may not always provide benefits to corals.
Although SST will be influenced by many local and global processes, the inclusion of just two predictors (ENSO strength and AOD) explained 70% of the variation in spatial extent of coral bleaching occurring across the Caribbean region between 1983 and 2000. In fact, our model also predicted very closely the extent of bleaching observed in the widespread bleaching event of 2005 in the Caribbean, during a moderately warm El Niño. AOD data for August–October 2005 are not currently available, but between 1983 and 2000, June AOD levels were significantly correlated with August–October AOD levels (r = 0.71, P < 0.002). Using the AOD values for June 2005 and ENSO strength in January/February 2005 (11–13), our model predicted that 38% of Caribbean coral reef cells should have experienced bleaching; the observed extent in 2005 was ≈33% of cells. Coral bleaching is thus largely under the dual control of El Niño events raising SST and atmospheric aerosols lowering it.
The nonvolcanic aerosol loading of the atmosphere is predicted to decline in the future, as cleaner technologies are developed (31), and grassland expansion into deserts such as the Sahara (32) could also reduce levels of dust transported across the Atlantic. Our analyses show that a 10% decline in AOD results in a 36% increase in coral bleaching extent. Should the projected increases in SST as a result of climate warming (33) occur in combination with any increase in the frequency of El Niño events (11) and reductions in levels of atmospheric aerosols (31), the resulting increases in coral bleaching will severely threaten these already highly degraded ecosystems. The protection afforded to Caribbean reefs by volcanic eruptions in the last 2 decades clearly cannot be relied on to mitigate the impact of rising SST in the future.
Methods
Spatial Extent of Coral Bleaching in the Caribbean.
Reports of coral bleaching during August–October were collated across the Caribbean from 1983 to 2000. Reports were allocated to 1° latitude/longitude cells depending on their location, and each cell was allowed a maximum of one value per year (see ref. 3 for further details).
AOD Values.
Monthly AOD data from radiances of the AVHRR (Advanced Very High Resolution Radiometer) are available in 1° latitude/longitude cells (25, 26) (http://gacp.giss.nasa.gov/data_sets). Data for August–October were averaged across the Caribbean to calculate a single AOD value per year, except for 1994, when insufficient data were available.
Identifying the Time Lag Between ENSO and Caribbean Coral Bleaching.
Previous studies indicated that SST variability in the North Atlantic lagged ENSO variability by ≈4–6 months (11–13). We expanded this range to explore lag times of 2–8 months, calculating a series of multiple regressions of the effect on coral bleaching (August–October, log scale) of AOD (August–October, log scale) and bimonthly MEI values (www.cdc.noaa.gov/people/klaus.wolter/MEI/mei.html) from the preceding December/January to June/July (n = 7). Both AOD and MEI were statistically significant predictors of coral bleaching extent in all months from December/January to April/May, and the best fit was obtained with the January/February MEI.
Predicting Bleaching Extent in the Absence of Volcanoes.
Average AOD levels for the Caribbean region during August–October were calculated from all years except the volcano-affected years (1983–1984 and 1991–1993). AOD values for each year were then calculated as anomalies from this mean, and these values were used to adjust SST for each year to that which would have been expected without the volcanic aerosols (18). Adjusted SST values were then used to predict the number of cells that would have been expected to bleach in each year, in the absence of major volcanic eruptions, by using the linear regression relating SST anomalies to extent of bleaching (3).
Acknowledgments
We thank N. Rayner, L. Bohn, I. Geogdzhayev, J. Merrill, J. Palutikof, and K. Briffa for assistance; A. Perry for discussions and 2005 bleaching data; and the U.K. Meteorological Office–Hadley Centre, Centers for Disease Control–National Oceanic and Atmospheric Administration, and Global Aerosol Climatology Project–National Aeronautics and Space Administration for access to data. This work was supported by the Tyndall Centre for Climate Change Research, the Joint Nature Conservation Committee, and Natural Environment Research Council (London).
Footnotes
- ‡To whom correspondence should be addressed. E-mail: j.gill{at}uea.ac.uk
-
↵ ¶Present address: Department of Biological Sciences, Simon Fraser University, Burnaby, BC, Canada V5A 1S6.
-
Author contributions: J.A.G., A.R.W., J.P.M., and I.M.C. designed research; J.A.G., A.R.W., and J.P.M. analyzed data; J.A.G., A.R.W., J.P.M., and I.M.C. wrote the paper; J.P.M. performed research.
-
The authors declare no conflict of interest.
-
This article is a PNAS direct submission.
- Abbreviations:
- AOD,
- aerosol optical depth;
- ENSO,
- El Niño–Southern Oscillation;
- MEI,
- multivariate ENSO index;
- SST,
- sea surface temperatures.
- © 2006 by The National Academy of Sciences of the USA
