O'Brien et al. reply —

Brierley et al.1 question our findings of elevated temperatures in the tropical warm pools during the Pliocene2. Focusing specifically on the mid-Pliocene warm period (about 3.3 to 3 million years ago), as framed by Brierley et al., we continue to find evidence for warmer than Holocene temperatures in the western Pacific warm pool in good agreement with PlioMIP simulations, especially in light of new p co 2 reconstructions3.

First, seawater Mg/Ca values were almost certainly lower than modern during the Pliocene (for example, ref. 4), although the exact magnitude and implications for the Mg/Ca ratio of foraminiferal calcite remain uncertain5. Nevertheless, if the western Pacific warm pool was the same temperature as pre-industrial times, as Brierley et al. contest, we would expect the Mg/Ca values of the foraminifera to be lower than those of the Holocene. Therefore, the observation of similar Holocene and mid-Pliocene Mg/Ca values indicates that sea surface temperatures were warmer than during the pre-industrial Holocene in the western Pacific warm pool, regardless of the correction used.

Second, Brierley et al. suggest that because sea surface temperature (SST) estimates based on uncorrected Mg/Ca and the alkenone proxy agree at site ODP 847 in the east tropical Pacific, the Mg/Ca seawater correction may be site-specific, and therefore unnecessary at site ODP 806 in the heart of the warm pool. However, they provide no mechanistic rationale for this. Moreover, as we discussed in the Supplementary Information of our Article, five out of six low-latitude sites for which Mg/Ca and palaeotemperature estimates exist show better agreement when the seawater Mg/Ca correction is applied. We also argue that faunal-based temperatures that appear to confirm little warming are biased by the upper limit of the modern calibration dataset, which is about 30 °C (ref. 6).

Third, we can also rule out a simple expansion of the western Pacific warm pool based on and Mg/Ca temperature reconstructions from the South China Sea (ODP 1143) and Mg/Ca temperature estimates from the warm pool centre (ODP 806). Specifically, the Holocene Mg/Ca temperature difference between the sites7 persists throughout our 5-million-year records2,8, implying that both the western Pacific and South China Sea were 1 to 2 °C warmer than Holocene during the Mid-Pliocene.

Fourth, a comparison between Plio–Pleistocene ; and Mg/Ca temperature estimates (with an Mg/Ca correction applied) and corresponding Holocene core-tops also indicates that the western Pacific warm pool, South China Sea and western Atlantic warm pool were warmer than Holocene estimates (Supplementary Fig. 1). Similarly, applying a regional, Bayesian-based TEX86 calibration9 yields Pliocene SSTs that are ∼1–2 °C higher than modern, measured SSTs for both ODP 806 and ODP 1143. We also point out that for all of these proxy approaches, the Pliocene data points represent an average of several thousand years, and thus do not necessarily represent maximum Pliocene warmth.

Finally, the long-term trends in the data from the heart of the western Pacific warm pool10 and southern South China Sea2,10 clearly demonstrate that the warm pool temperatures have cooled from the Late Miocene to the Pleistocene (Supplementary Fig. 1a,b). In summary, the data we presented in ref. 2 and provide here strongly suggest that the warm pools of the Atlantic and Pacific were warmer than the Holocene during the Mid-Pliocene warm period and Pliocene as a whole.