Comparison of equatorial Pacific sea surface temperature variability and trends with Sr/Ca records from multiple corals

Author Posting. © American Geophysical Union, 2016. This article is posted here by permission of American Geophysical Union for personal use, not for redistribution. The definitive version was published in Paleoceanography 31 (2016): 252–265, doi:10.1002/2015PA002897.Coral Sr/Ca is widely used to reconstruct past ocean temperatures. However, some studies report different Sr/Ca-temperature relationships for conspecifics on the same reef, with profound implications for interpretation of reconstructed temperatures. We assess whether these differences are attributable to small-scale oceanographic variability or “vital effects” associated with coral calcification and quantify the effect of intercolony differences on temperature estimates and uncertainties. Sr/Ca records from four massive Porites colonies growing on the east and west sides of Jarvis Island, central equatorial Pacific, were compared with in situ logger temperatures spanning 2002–2012. In general, Sr/Ca captured the occurrence of interannual sea surface temperature events but their amplitude was not consistently recorded by any of the corals. No long-term trend was identified in the instrumental data, yet Sr/Ca of one coral implied a statistically significant cooling trend while that of its neighbor implied a warming trend. Slopes of Sr/Ca-temperature regressions from the four different colonies were within error, but offsets in mean Sr/Ca rendered the regressions statistically distinct. Assuming that these relationships represent the full range of Sr/Ca-temperature calibrations in Jarvis Porites, we assessed how well Sr/Ca of a nonliving coral with an unknown Sr/Ca-temperature relationship can constrain past temperatures. Our results indicate that standard error of prediction methods underestimate the actual error as we could not reliably reconstruct the amplitude or frequency of El Niño–Southern Oscillation events as large as ± 2°C. Our results underscore the importance of characterizing the full range of temperature-Sr/Ca relationships at each study site to estimate true error.This study was supported by an NSF Graduate Research Fellowship to A.A. and by NSF-OCE-0926986 and NSF-OCE-1031971.2016-08-0