MioTEX: A comprehensive GDGT-based ocean temperature compilation for the Miocene

Abstract. The Miocene epoch (~23 to 5.3 million years ago) was a period when global mean temperatures were warmer than today, with estimated atmospheric CO2 levels similar to those expected in coming centuries. Climate models generally underestimate reconstructed Miocene global mean and high-latitude ocean temperatures, a persistent discrepancy that raises questions about both the reliability of proxy reconstructions and the ability of climate models to represent warm climate states. Resolving this discrepancy requires a critical, unified reevaluation of the existing proxy record. Here we compile and reevaluate existing ocean temperature records reconstructed using the TEX86 paleothermometer. The TEX86 paleothermometer, which derives upper ocean temperatures from the distribution of the membrane lipids of marine archaea, has been widely used to reconstruct past sea surface and shallow subsurface (0–200 m) temperatures. Here, we introduce MioTEX, a global compilation of Miocene ocean temperature records based on the TEX86 proxy. This effort, developed within the MioOcean initiative, applies uniform screening criteria and calibration methods to published records from 46 sites spanning multiple ocean basins. We find distinct temperature variations among different latitudinal bands and ocean basins, as well as absolute temperature differences arising from the choice of calibration. Compared with previous SST reconstructions suggesting an Early to Middle Miocene collapse and Late Miocene establishment of the modern meridional SST gradient, our standardised, multi‑site synthesis quantifies when and where such gradient changes occurred across ocean basins, with the strongest cooling recorded at high-latitude sites in the South Atlantic and Southwest Pacific. The MioTEX dataset is publicly available at Zenodo (https://doi.org/10.5281/zenodo.22258386, Sosdian et al., 2026) and presented on an interactive Streamlit-based web platform, allowing users to explore data, apply different calibrations, and test screening criteria.