The Marine Isotope Stage (MIS) 11 (424-374 ka) was characterized by a protracted deglaciation and an unusually long climatic optimum. It remains unclear to what degree the climate development during this interglacial reflects the unusually weak orbital forcing or greenhouse gas trends. Previously, arguments about the duration and timing of the MIS11 climatic optimum and about the pace of the deglacial warming were based on a small number of key records, which appear to show regional differences. In order to obtain a global signal of climate evolution during MIS11, we compiled a database of 78 sea surface temperature (SST) records from 57 sites spanning MIS11, aligned these individually on the basis of benthic (N = 28) or planktonic (N = 31) stable oxygen isotope curves to a common time frame and subjected 48 of them to an empirical orthogonal function (EOF) analysis. The analysis revealed a high commonality among all records, with the principal SST trend explaining almost 49% of the variability. This trend indicates that on the global scale, the surface ocean underwent rapid deglacial warming during Termination V, in pace with carbon dioxide rise, followed by a broad SST optimum centered at ~410 kyr. The second EOF, which explained ~18% of the variability, revealed the existence of a different SST trend, characterized by a delayed onset of the temperature optimum during MIS11 at ~398 kyr, followed by a prolonged warm period lasting beyond 380 kyr. This trend is most consistently manifested in the mid-latitude North Atlantic and Mediterranean Sea and is here attributed to the strength of the Atlantic meridional overturning circulation. A sensitivity analysis indicates that these results are robust to record selection and to age-model uncertainties of up to 3-6 kyr, but more sensitive to SST seasonal attribution and SST uncertainties >1 °C. In order to validate the CCSM3 (Community Climate System Model, version 3) predictive potential, the annual and seasonal SST anomalies recorded in a total of 74 proxy records were compared with runs for three time slices representing orbital configuration extremes during the peak interglacial of MIS11. The modeled SST anomalies are characterized by a significantly lower variance compared to the reconstructions. Nevertheless, significant correlations between proxy and model data are found in comparisons on the seasonal basis, indicating that the model captures part of the long-term variability induced by astronomical forcing, which appears to have left a detectable signature in SST trends.
