The El Niño-Southern Oscillation (ENSO) drives large changes in global climate patterns from year to year, yet its sensitivity to continued anthropogenic greenhouse forcing is uncertain. We analyzed fossil coral reconstructions of ENSO spanning the past 7000 years from the Northern Line Islands, located in the center of action for ENSO. The corals document highly variable ENSO activity, with no evidence for a systematic trend in ENSO variance, which is contrary to some models that exhibit a response to insolation forcing over this same period. Twentieth-century ENSO variance is significantly higher than average fossil coral ENSO variance but is not unprecedented. Our results suggest that forced changes in ENSO, whether natural or anthropogenic, may be difficult to detect against a background of large internal variability.

The El Niño-Southern Oscillation (ENSO) represents the largest source of year-to-year global climate variability. While Earth system models suggest a range of possible shifts in ENSO properties under continued greenhouse gas forcing, many centuries of preindustrial climate data are required to detect a potential shift in the properties of recent ENSO extremes. Here we reconstruct the strength of ENSO variations over the last 7,000 years with a new ensemble of fossil coral oxygen isotope records from the Line Islands, located in the central equatorial Pacific. The corals document a significant decrease in ENSO variance of~20% from 3,000 to 5,000 years ago, coinciding with changes in spring/fall precessional insolation. We find that ENSO variability over the last five decades is~25% stronger than during the preindustrial. Our results provide empirical support for recent climate model projections showing an intensification of ENSO extremes under greenhouse forcing. Plain Language SummaryRecent modeling studies suggest that El Niño will intensify due to greenhouse warming. Here new coral reconstructions of the El Niño-Southern Oscillation (ENSO) record sustained, significant changes in ENSO variability over the last 7,000 years and imply that ENSO extremes of the last 50 years are significantly stronger than those of the preindustrial era in the central tropical Pacific. These records suggest that El Niño events already may be intensifying due to anthropogenic climate change.

The integrity of coral-based reconstructions of past climate variability depends on a comprehensive knowledge of the effects of post-depositional alteration on coral skeletal geochemistry. Here we combine millimeter-scale and micro-scale coral Sr/Ca data, scanning electron microscopy (SEM) images, and X-ray diffraction with previously published d 18 O records to investigate the effects of submarine and subaerial diagenesis on paleoclimate reconstructions in modern and young sub-fossil corals from the central tropical Pacific. In a 40-year-old modern coral, we find secondary aragonite is associated with relatively high coral d 18 O and Sr/Ca, equivalent to sea-surface temperature (SST) artifacts as large as À3 and À5°C, respectively. Secondary aragonite observed in a 350-year-old fossil coral is associated with relatively high d 18 O and Sr/Ca, resulting in apparent paleo-SST offsets of up to À2 and À4°C, respectively. Secondary Ion Mass Spectrometry (SIMS) analyses of secondary aragonite yield Sr/Ca ratios ranging from 10.78 to 12.39 mmol/mol, significantly higher compared to 9.15 ± 0.37 mmol/mol measured in more pristine sections of the same fossil coral. Widespread dissolution and secondary calcite observed in a 750-year-old fossil coral is associated with relatively low d 18 O and Sr/Ca. SIMS Sr/Ca measurements of the secondary calcite (1.96-9.74 mmol/mol) are significantly lower and more variable than Sr/Ca values from more pristine portions of the same fossil coral (8.22 ± 0.13 mmol/mol). Our results indicate that while diagenesis has a much larger impact on Sr/Ca-based paleoclimate reconstructions than d 18 O-based reconstructions at our site, SIMS analyses of relatively pristine skeletal elements in an altered coral may provide robust estimates of Sr/Ca which can be used to derive paleo-SSTs.

Climate change is now the leading cause of coral-reef degradation and is altering the adaptive landscape of coral populations 1,2 . Increasing sea temperatures and declining carbonate saturation states are inhibiting short-term rates of coral calcification, carbonate precipitation and submarine cementation [3][4][5] . A critical challenge to coral-reef conservation is understanding the mechanisms by which environmental perturbations scale up to influence long-term rates of reef-framework construction and ecosystem function 6,7 . Here we reconstruct climatic and oceanographic variability using corals sampled from a 6,750-year core from Pacific Panamá. Simultaneous reconstructions of coral palaeophysiology and reef accretion allowed us to identify the climatic and biotic thresholds associated with a 2,500-year hiatus in vertical accretion beginning ∼4,100 years ago 8 . Stronger upwelling, cooler sea temperatures and greater precipitation-indicators of La Niña-like conditions-were closely associated with abrupt reef shutdown. The physiological condition of the corals deteriorated at the onset of the hiatus, corroborating theoretical predictions that the tipping points of radical ecosystem transitions should be manifested sublethally in the biotic constituents 9 . Future climate change could cause similar threshold behaviours, leading to another shutdown in reef development in the tropical eastern Pacific.Climatic and oceanographic variability have played a dominant role in the development of reefs throughout the Phanerozoic eon 10 , and the recent past is no exception. In Panamá and several other locations in the Pacific, coral reefs stopped accreting vertically for 2,500 years, beginning ∼4,100 cal yr BP (ref. 8; calibrated 14 C calendar years before 1950; Fig. 1a). Correlations with regional palaeoclimate proxies suggest that enhanced variability of the El Niño/Southern Oscillation (ENSO) was the ultimate cause of reef shutdown in the tropical eastern Pacific 8 (TEP). Climatic shifts at that time led to environmental and cultural impacts on a global scale 11,12 .In this study we investigated the long-term impacts of environmental variability on coral physiology and reef development in the TEP to ascertain the climatic, oceanographic and biotic controls on ecosystem state in the past. We quantified the range of environmental conditions that corals experienced during the past ∼6,750 years to determine whether significant changes in climate or oceanography were associated with changes in coral physiology or reef accretion. We then evaluated the environmental and physiological thresholds that characterized the catastrophic phase shift to the hiatus.In Pacific Panamá, El Niño-like periods are characterized by a warm, dry climate and a reduction in seasonal upwelling. Those conditions are reversed during La Niña-like periods (Supplementary Discussion and Supplementary Fig. 1). Contemporary environmental variability is high at Contadora Island, a site in Pacific Panamá that is exposed to intense seasonal upwelling and ...

Quantitative estimates of natural climate variability are required to detect anthropogenic climate trends in the tropical Pacific; however, instrumental records from this region are too short and scarce. Coral oxygen isotopic (δ18O) and strontium to calcium (Sr/Ca) records are often used to extend instrumental observations; however, differences in the mean Sr/Ca and δ18O values of Porites spp. colonies from the same reef can introduce large uncertainties in coral‐based climate reconstructions. To quantify intercolony variability at Palmyra Atoll, we generate monthly resolved Sr/Ca and δ18O time series from five Porites spp. colonies that grew between 1980 and 2010. Monthly to interannual variability in Sr/Ca and δ18O is well‐reproduced among different colonies; however, we document intercolony offsets in mean Sr/Ca of ±0.09 mmol/mol (1σ) or ~1 °C, and in mean δ18O of ±0.12‰ (1σ) or ~0.1 °C. The sensitivity of each proxy to climate also varies across colonies, with Sr/Ca‐SST slopes ranging from −0.06 to −0.1 mmol mol−1 °C−1 and δ18O‐SST slopes ranging from −0.25 to −0.35‰ °C−1. Intercolony variability in both coral Sr/Ca and δ18O reduces the reproducibility of coral‐based δ18Osw reconstructions across overlapping colonies. Accounting for both intercolony variability and slope error suggests that SST reconstructions using Sr/Ca from a single Palmyra coral have an uncertainty of ±1.3 °C (1σ); however, replicating Sr/Ca records across multiple colonies can greatly reduce this uncertainty. A composite Sr/Ca record built using five modern cores, for example, offers a reduced error of ±0.6 °C (1σ) in mean SST reconstructions, ~2.5 times smaller than errors associated with reconstructions from single corals.