We used Sr/Ca and stable isotope data from well dated and preserved corals from the northeastern Caribbean to determine the seasonal environmental conditions for four continuous years during the Eemian, the last time the Earth was in a prolonged warm phase. We determined that the seasonal range in SST during the Eemian was 25°–30° C. This is ∼1–2° larger than at present and caused primarily by winter cooling and, only to a small degree, by summer warming. As climate modeling studies indicate, the bias towards colder winters can be explained by changes in low latitude insolation induced by altered orbital parameters, modulated by atmospheric CO2 levels that were lower than today. Milankovitch forcing at higher latitudes was probably less important.
 The astronomical theory of paleoclimates [Berger, 1988], best known in its Milankovitch version, states that changes in the seasonal distribution of insolation brought about by changes in the Earth's orbital parameters exert a dominant influence on climate. One approach to corroborate this theory would be to study seasonal data, preferably from a period in the past in which the Earth's orbital parameters were markedly different from the present values. The last interglacial, which is believed to have occurred sometime between 140,000–130,000 to 117,000 a BP [Adams et al., 1999], depending on location, is one such period [Muhs et al., 2002]. It is commonly known as the Eemian on land, approximately equivalent to Marine Isotope Stage 5e. It is generally believed that the Eemian was the last time the Earth experienced warm conditions reminiscent of today. Sea level was higher than today [Bard et al., 1990]. The mean atmospheric CO2 concentration was at pre-industrial levels (∼270 ppm from ice core measurements [Petit et al., 1999]). There are indications that it was interrupted for a short period by a cold interval [Stirling et al., 1998]. Coral reef building, which requires relatively stable climate conditions, took place for a period of at least 5,000 a, and concluded most likely before the beginning of the cold interval at ca. 122,000 a BP [Adams et al., 1999].
 During the peak of the last interglacial, eccentricity was much greater than today (0.0394 as compared to 0.0167) and, in fact, greater than at any other time during the past 150,000 a, and perihelion (longitude of perihelion with respect to vernal equinox: 95.41° as compared to 282.04°) took place in northern summer (June) instead of northern winter (January). Obliquity, the tilt of the Earth's axis (24.04° as compared to 23.45°) was also slightly greater than today (all orbital element data are for 127,000 a BP and AD 1950, respectively [Berger, 1978]). The near-minimum precession during the Eemian (perihelion was quite close to the summer solstice, 90°) enhanced the seasonality in the northern hemisphere and weakened it in the other (Figure 1; Tuenter et al., ). The difference in obliquity added another small contribution to these seasonal insolation changes, by strengthening the contrast between the hemispheres during summer and winter, anti-symmetrical to the equator. A stronger seasonal contrast might therefore be expected for Eemian climate records from the northern hemisphere.
 Here we present two different SST proxy records derived from a Caribbean coral living at the height of the Eemian to test the theory that the seasonal distribution of insolation during Eemian times should result in a greater seasonality in low northern latitudes. We compare these seasonal proxy data with insolation and results from climate modeling.
 Reef building corals have thrived during only short time intervals in Earth's history. For example, in Western Australia Eemian reef building was confined to just a few thousand years between 127,000 to 122,000 a BP [Stirling et al., 1998]. This is also true for most other areas of the tropics because sea level was constant only at the height of the Eemian allowing enough time for extensive coral building to take place. Abundant coral growth during the Eemian is evident on 4 to 6 m high reef terraces throughout the Caribbean.
 Our Eemian record was obtained by drilling through a colony of Montastraea faveolata occurring on a raised reef terrace [Taggart, 1993] 4.1 m above sea level on Isla de Mona (18°N, 67°W) between Hispaniola and Puerto Rico in the northeastern Caribbean. The core was radiometrically dated by U-series alpha spectrophotometry and U/Th thermal ionization mass spectrometry (TIMS) [Cheng et al., 1998]. Two TIMS U-Th dates gave ages of 127.5 ± 1.4 ka and 126.3 ± 1.9 ka, confirming that the corals grew during Substage 5e. Scanning electron microscope and X-ray diffraction studies of the aragonite show that the aragonite has undergone little or no diagenesis. The 238U concentration of 2.5 ppm of the fossil coral is typical for the modern species and the similarity between the initial 234U/238U ratios of the Isla de Mona 5e corals, at 1.152 ± 0.003, to that of modern corals at 1.144 ± 0.004 [Cheng et al., 1998] is further evidence that the samples are pristine.
 M. faveolata is well suited to the task of reconstructing past SSTs in the tropical Atlantic Ocean [Leder et al., 1996]. Oxygen isotopic ratios and trace elements in the aragonitic framework of scleractinian coral species are the primary agents for SST reconstruction [Gagan et al., 2000]. For stable isotope and trace elements analyses 24 samples were taken between annually produced high-density bands using the method of Watanabe and Oba . The δ18O standard deviations (2σ; 95% confidence) for fifteen duplicate measurements of powdered carbonate samples measured by a Finnigan MAT 251 mass spectrometer were 0.04‰. To convert oxygen isotopes to SSTs we used the equation of Leder et al. . Using this equation to convert modern stable isotopes derived from M. faveolata to SST in the northeast Caribbean has consistently provided reliable results [Watanabe et al., 2002].
 Stable oxygen isotopes record SST and the stable isotopic composition of the water in which the coral lived. Isla de Mona, situated between Hispaniola and Puerto Rico in the Mona Passage, a throughway for water exchange between the Atlantic Ocean and Caribbean Sea, lacks river discharge, and ocean salinity remains constant throughout the year suggesting that changes in coral δ18O in Isla de Mona reflect only local SST variability. To convert Sr/Ca to SST we simply used the modern local temperature range for calibration. This yielded a slope of 0.55 similar to the 0.6 from the Bahamas for M. faveolata [Guilderson, 1997]. Trace element ratios are primarily a function of SST. For Sr/Ca, the primary assumption is that oceanic Sr was relatively constant [de Villiers, 1999]; that the increased growth rate of the coral did not cause appreciable kinetic effects [de Villiers et al., 1995]; and that the analyzed aragonitic material was pristine so that possible distortion of the SST signal due to diagenesis was negligible [Enmar et al., 2000]. Precision for Sr/Ca is better than 0.15% (1 σ).
 To relate the change in SST to the solar forcing at the time, we compute the anomaly of daily insolation as a function of season and latitude as well as for a fixed latitude of 18°N for 127,000 a BP (Figures 1 and 2; the value of the solar constant is set at 1367 W m−2). We define Eemian northern summer and winter as those periods that correspond to the same interval in true longitude as June–August (JJA) and December–February (DJF) today [Joussaume and Braconnot, 1997; Montoya et al., 2000]. For present day, we employ the usual definitions for the meteorological seasons.
 Compared to the present, the change in global annual-mean insolation during the Eemian was negligible (0.21 W m−2). At the latitude of Isla de Mona (18°N), annual-mean insolation was 393.37 as compared to 393.87 W m−2, a change of just 0.50 W m−2 or 0.13%. However, seasonality in the Eemian at low northern latitudes was greater than today because of altered orbital parameters of the Earth [Berger, 1978]. During northern summer, local insolation was greater than today (489.53 as compared to 442.87 W m−2, an increase of 46.66 W m−2 or 10.54%), and during northern winter it was smaller than today (296.77 as compared to 328.60 W m−2, a decrease of 31.84 W m−2 or 9.69%; Figure 2). Overall, the seasonal range in insolation was amplified in the Northern Hemisphere, but attenuated in the Southern Hemisphere [Montoya et al., 2000].
 Results of δ18O and Sr/Ca analyses from a modern coral from Puerto Rico and the Eemian coral from Isla de Mona are presented in Figure 3. When converted to Eemian SSTs both proxies yield similar results. The fact that two different proxies yield similar temperature ranges gives us confidence in our results and therefore we believe that the Eemian range given here is reliable. Two major differences between the modern coral and the Eemian coral are apparent. The first is that the seasonal SST range of ∼5°C in the Eemian sample (approximately 25°–30°C) is one to two degrees larger than the present range of ∼3–4 °C (1965–1999; Figure 3 and Table 1). The second is that, in contrast to local insolation, the greater seasonality in Eemian SST was caused primarily by winter cooling and only to a small degree by summer warming. Thus, at our site, while the seasonal range of SST was larger during the Eemian than at present, its annual mean value was possibly slightly lower.
|Ave. of Max.||29.7||29.7||30.0||29.6||30.0|
|Ave. of Min.||26.0||26.8||26.0||24.6||25.0|
- a Values in first column are the observations (1966–1998) taken at La Parguera (LP), Puerto Rico. Second and third column show calculated SST values for the Present derived from LP coral Sr/Ca and δ18O. Fourth and fifth columns show SST values for Eemian corals taken at Isla de Mona (50 km from LP).
 We realize that the Eemian samples are basically random snap shots of peak Eemian SST conditions. However, we believe that they are indicative of Eemian SST conditions for at least three thousand years (roughly from 127,000–124,000 a BP) for three reasons: First, sea level was nearly constant between 127,000 and 122,000 a BP. Second, although this interval was possibly punctuated by a cold event, the occurrence of coral reef building by itself provides evidence for otherwise relatively stable climatic conditions. Third, the orbital elements vary only very slowly through time, the fastest being the longitude of the perihelion with a period of about 20,000 a. Between 128,000–124,000 a BP, perihelion fell into northern summer (JJA), and insolation conditions were similar to 127,000 a BP (Figures 1 and 2).
 Furthermore, detailed information on Eemian climate is rare; few records actually resolve the seasonal cycle and few exist from outside the mid-to-high northern latitudes. Pacific records can be complicated by large salinity/precipitation changes associated with ENSO/monsoon variability. Atlantic corals should provide a more robust global SST signal. Indeed, work by Diaz (pers. com.) has shown that SST changes in the northeastern Caribbean are coherent with changes in global SST during the last 100 years. Finally, any data from low latitudes are particularly well suited because the local response of the climate system is less distorted by high-latitude albedo effects and ocean circulation changes and, thus, allows an assessment of the sensitivity of the climate system to radiative perturbations only.
 Using Eemian insolation and CO2 forcings in their CLIMBER-2 model, Kubatzki et al.  showed that the Caribbean experienced summer and winter near-surface air temperature anomalies of <1°C and ∼−2°C respectively in comparison to present-day (or modern) climate with present CO2 levels and of <2°C and ∼−1°C in comparison to present-day climate with pre-industrial CO2 levels (the CO2 effect deduced from the CLIMBER-2 equilibrium simulations maybe higher than in reality because the present-day climate has not yet equilibrated with the high industrial CO2 levels). Also, Tuenter et al.  found in the ECBilt model that the difference in surface air temperature in the Caribbean for minimum and maximum precession insolation amounts to >1°C during summer and <−1°C during winter, not taking into account any CO2 change. The Eemian SST average and annual range we derived from the coral proxies is therefore consistent with what one would expect from altered insolation at the coral site of 18°N (Figure 2) and lower CO2 levels. Ocean and atmospheric circulation changes probably also played a role, but it is uncertain whether the Atlantic Ocean meridional overturning circulation strengthened or weakened relative to pre-industrial times [Kubatzki et al., 2000].
 Geochemical paleo proxies from a M. faveolata coral colony on Isla de Mona (18°N 67°W) in the northeastern Caribbean, growing at the height of the Eemian, indicate a greater seasonal range in SSTs relative to today, caused primarily by winter cooling and, to a smaller degree, by warmer summers. A greater seasonal range is what would be expected from the seasonal distribution of insolation at that time. The apparent bias towards the winter season is due to greater seasonality combined with the lower CO2 concentration in the atmosphere (about 270 ppm as compared to more than 360 ppm today). Thus, our findings demonstrate the orbital control of seasonality at low latitudes. At the same time they show the importance of changes in atmospheric CO2 levels, because it appears that, in the annual mean, the present-day climate is already warmer than the last interglacial. This may be contrary to common belief, but is in accordance with climate model results [Montoya et al., 1998; Montoya et al., 2000].
 We thank the crew and diving contingent of the R.V. Pezma, Mark Riegle, Robin and Andy Bruckner and Milton Carlo for their considerable help. This paper was supported by the Royal Swedish Academy of Sciences (Hierta-Retzius Foundation), and the Kungl and Hvitfeldska Foundations. We are also grateful to the Department of Marine Botany, Göteborg University, for the use of their freezing microtome. AP acknowledges the support of the Alexander von Humboldt Foundation through a Feodor Lynen Research Fellowship during a stay at Scripps Institution of Oceanography in La Jolla, California.
- Sudden climate transitions during the Quaternary, Progress in Physical Geography, 23(1), 1–36, 1999.
- U-TH ages obtained by mass spectrometry in corals from Barbados; sea level during the past 130,000 years, Nature, 346(6283), 456–458, 1990.
- Long-term variations of daily insolation and Quaternary climatic changes, Journal of Atmospheric Sciences, 35, 2363–2367, 1978.
- Milankovitch theory and climate, Reviews of Geophysics, 26, 624–657, 1988.
- Uranium-thorium-protactinium dating systematics, Geochimica Et Cosmochimica Acta, 62(21–22), 3437–3452, 1998.
- Seawater strontium and Sr/Ca variability in the Atlantic and Pacific oceans, Earth and Planetary Science Letters, 171(4), 1999.
- Biological controls on coral Sr/Ca and delta-O-18 reconstructions of sea surface temperatures, Science, 269(5228), 1247–1249, 1995.
- Diagenesis in live corals from the Gulf of Aqaba. I. The effect on paleo-oceanography tracers, Geochimica Et Cosmochimica Acta, 64(18), 3123–3132, 2000.
- New views of tropical paleoclimates from corals, Quaternary Science Reviews, 19(1–5), 45–64, 2000.
- Tropical Atlantic SSTs over the last 20,000 years: Implications on the mechanism and synchroniety of the interhemisphere climate change, Columbia University, New York, 1997.
- Sensitivity of paleoclimate simulation results to season definitions, Journal of Geophysical Research, 102(D2), 1943–1956, 1997.
- Comparison of the last interglacial climate simulated by a coupled global model of intermediate complexity and an AOGCM, Climate Dynamics, 16(10–11), 799–814, 2000.
- The origin of variations in the isotopic record of scleractinian corals, 1, Oxygen, Geochimica Et Cosmochimica Acta, 60(15), 2857–2870, 1996.
- Climate simulation for 125 kyr BP with a coupled ocean-atmosphere general circulation model, Journal of Climate, 13(6), 1057–1072, 2000.
- Timing and warmth of the Last Interglacial period: New U-series evidence from Hawaii and Bermuda and a new fossil compilation for North America, Quaternary Science Reviews, 21(12–13), 1355–1383, 2002.
- 420000 years of climate and atmospheric history revealed by the Vostok deep antarctic ice core, Nature, 399, 429–436, 1999.
- Timing and duration of the Last Interglacial: Evidence for a restricted interval of widespread coral reef growth, EPSL, 160, 745–762, 1998.
- Tectonic and eustatic correlations of radiometrically dated marine terraces in northwestern Puerto Rico Rico and Isla de Mona, Puerto Rico, Ph.D. thesis, University of Puerto Rico, Mayagüez, Puerto Rico, 1993.
- The response of the African monsoon to remote and local forcing due to precession and obliquity, submitted to Global and Planetary Change, 2002.
- Daily reconstruction of water temperature from oxygen isotopic ratios of a modern Tridacna shell using a freezing microtome sampling technique, Journal of Geophysical Research-Oceans, 104(C9), 20,667–20,674, 1999.
- Evaluation of the fidelity of isotope records as an environmental proxy in the coral Montastraea, Coral Reefs, 21(2), 169–178, 2002.