Description: We present a revised and extended high Arctic air temperature reconstruction from a single proxy that spans the past ∼12,000 y (up to 2009 CE). Our reconstruction from the Agassiz ice cap (Ellesmere Island, Canada) indicates an earlier and warmer Holocene thermal maximum with early Holocene temperatures that are 4-5 °C warmer compared with a previous reconstruction, and regularly exceed contemporary values for a period of ∼3,000 y. Our results show that air temperatures in this region are now at their warmest in the past 6,800-7,800 y, and that the recent rate of temperature change is unprecedented over the entire Holocene. The warmer early Holocene inferred from the Agassiz ice core leads to an estimated ∼1 km of ice thinning in northwest Greenland during the early Holocene using the Camp Century ice core. Ice modeling results show that this large thinning is consistent with our air temperature reconstruction. The modeling results also demonstrate the broader significance of the enhanced warming, with a retreat of the northern ice margin behind its present position in the mid Holocene and a ∼25% increase in total Greenland ice sheet mass loss (∼1.4 m sea-level equivalent) during the last deglaciation, both of which have implications for interpreting geodetic measurements of land uplift and gravity changes in northern Greenland.

Description: The topographic survey of the studied outcrops is based on several thousands of measurements per study site and the measurement of the sample elevation with reference to sea level using a real-time kinematic GPS Trimble R8. The maximum vertical (Z) and horizontal (X and Y) elevation errors are of ± 2.0 cm and a few millimetres, respectively. During the measurement, the surveys were related to the French Polynesian Geodetic Network (Réseau Géodésique de Polynésie Française; RGPF), to operating tide gauges or tide gauge data sets, to probes that were deployed during the field work, to the instantaneous sea level or to modern adjacent microatolls growing in a similar environment than their fossil counterparts. In the absence of geodetic datum or tide gauges, probes were deployed for four to five days in order to measure the sea-level position and to compare the data to the elevation of modern microatolls. The relative sea-level curve, which is presented in this paper, is based on data acquired on islands for which longer tidal records and geodetic data are available. After acquisition, the raw data were processed with the aims: 1) to estimate the elevation of individual dated fossil microatolls based on local tide gauge parameters, and 2) to compare the elevation of all dated fossil microatolls according to the same vertical reference. The link between tide gauge data and the position of the living and fossil microatolls can be established using RGPF. However, a topographic reference at the scale of French Polynesia (4,167 km^2), which is mandatory to achieve the second objective, does not exist, as tide gauge observations are incomplete and the NGPF (Nivellement Général de Polynésie Française) vertical datum that is associated to the RGPF is not homogeneous at this regional scale. The official geodetic system in French Polynesia is the RGPF, which is associated with the NGPF vertical datum. The French Polynesian Geodetic Network is a semi-dynamic system with different levels established by the Naval Hydrographic and Oceanographic Service (Service Hydrographique et Océanographique de la Marine; SHOM) in cooperation with the National Geographic Institute (Institut Géographique National; IGN). The selection of microatolls for dating has been based on the lack of erosion features, the absence of local moating effects and their mineralogical preservation, demonstrating that our database is robust. The chemical preparation, mass-spectrometer measurements and age dating were performed in the years 2014 to 2016 mostly directly after field collection. The data are presented in Supplementary Table 2 following recommendations from Dutton et al. (2017). The best-preserved samples, as indicated by X-ray Powder Diffraction (XRD) measurements, comprise 97.5% aragonite on average (n = 281). Additionally, no secondary aragonite or calcite crystals were revealed by thin section and Scanning Electron Microscope (SEM) observations.

Description: Characteristics, elevations and ages of 165 samples from various mid-late Holocene features, such as in situ Porites microatolls, in situ reef flats, conglomerates and reworked reef blocks, collected from twelve islands in French Polynesia. This table complements the database on which the mid-late Holocene sea-level curve has been initially reconstructed (Hallmann et al., 2018). Δ modern-fossil represents the difference in elevation between modern and Holocene microatolls at the same study site and in a similar environment (maximum vertical error is of ± 2 cm). NGPF = altimetric reference of French Polynesia. The elevations have been corrected for subsidence based on rates of 0.14 mm/yr for Moorea, 0.05 mm/yr for Bora Bora and Tahaa as well as 0.03 mm/yr for Maupiti, the Gambier Islands and Raivavae. Error values for ages and elevations are 2-sigma. Uncertainties for measured elevations related to NGPF are ± 14 cm for samples from Bora Bora, Maupiti, Raivavae, Tikehau and the Gambier Islands; ± 22 cm for samples from Fakarava, Hao, Makemo, Manihi, Moorea, Rangiroa and Tahaa. Uncertainties for estimated elevations are ± 10 cm for Δ modern-fossil and ± 22 cm for NGPF elevations.

Description: Uranium/Thorium isotopic composition of 165 samples from various mid-late Holocene features, such as in situ Porites microatolls, in situ reef flats, conglomerates and reworked reef blocks, collected from twelve islands in French Polynesia. Recommendations of Dutton et al. (2017) were followed for the presentation of U/Th age data. For each parameter, the first column contains the value and the second column the statistical error. All statistical errors are two standard deviations of the mean (2σ mean). All samples have been corrected for initial 230Th by using a 230Th/232Th activity ratio of 0.66 ± 0.2 (Fietzke et al., 2005). Non-reported data consist of 230Th/232Th ratios which became negative due to background corrections. 238U Concentrations are not corrected for the background.