Earth is heading towards a climate that was last experienced more than 3 Myr during the “mid-Pliocene warm period”1. Atmospheric carbon dioxide (pCO2) concentrations were ~400 ppm, global sea level oscillated in response to orbital forcing2,3 and peak global mean sea level (GMSL) may have reached ~20 m above present4,5. For sea-level rise of this magnitude extensive retreat or collapse of the Greenland, West Antarctic and marine based sectors of the East Antarctic ice sheets are required. Yet the relative amplitude of sea-level variations within glacial-interglacial cycles remains poorly-constrained. Here, we show sea-level varied on average by 13 ± 5 m over glacial-interglacial cycles during the mid- to late Pliocene, ~3.3 - 2.5 Myrs. We calibrated a theoretical relationship between modern sediment transport by waves and water depth and then applied the technique to Pliocene grain size in shallow-marine sediments from Whanganui Basin, New Zealand, thereby estimating past sea level variation. The resulting PlioSeaNZ record is independent of the deep ocean oxygen isotope (δ18O) record for global ice volume3, and in phase with ~20 kyr duration cycles of insolation over Antarctica, paced by eccentricity-modulated orbital precession between 3.3 and 2.7 Ma6. Thereafter, sea-level fluctuations are paced by ~41 kyr cycles in Earth's axial tilt as ice sheets stabilise on Antarctica and intensify in the northern hemisphere3,6. Sensu stricto, we provide the amplitude of relative sea-level (RSL) change, rather than absolute GMSL change. However, glacio-isostatic adjustment (GIA) simulations of RSL change, show that the PlioSeaNZ record approximates eustatic sea level (ESL), defined here as GMSL unregistered to the centre of the Earth. Nonetheless, under conservative assumptions, our estimates limit maximum Pliocene sea level to less than +25 m and provide new constraints on polar ice-volume variability under climate conditions Earth is on track to experience this century.

Tables below contain relative amplitudes for glacial/deglacial phases and the continuous relative sea level record.Grant et al PlioSeaNZ: Paleobathymetry and relative sea level for the PlioSeaNZ record. Samples from Siberia-1 core (39-350 m depth) and Rangitikei River Section (425 – 818 m stratigraphic height) are presented as a composite stratigraphic height, with ∑V>63 (%; Sand), paleobathymetry and error from Eqn. 7 and relative sea-level (RSL) with age following backstripping. Grant et al PlioSeaNZ: Relative sea-level (RSL) amplitudes for deglacial (G-IG) and glacial (IG-G) transitions. RSL and uncertainty are calculated from Eqn 10 and ages are reported for each event as the mid-range, with error defined as ±5 kyr resolution of magnetostratigraphy.The relative sea-level (RSL) with age following backstripping. Note that backstripping is undertaken for the Siberia-1 core and Rangitikei outcrop separately, the different relative values reflect this and are normalised for display in Fig. 2. RSL and uncertainty are calculated from Eqn 10 and ages are reported for each event as the mid-range, with error defined as ± 5 kyr resolution of magnetostratigraphy.

Supplement to: Grant, Georgia Rose; Naish, Timothy R; Dunbar, Gavin B; Stocchi, Paolo; Kominz, Michelle A; Kamp, Peter J; Tapia, C A; McKay, Robert M; Levy, Richard H; Patterson, Molly O (2019): The amplitude and origin of sea-level variability during the Pliocene epoch. Nature