Holocene relative sea-level histories of far-field islands in the mid-Pacific

Holocene relative sea-level (RSL) data from far-field islands in the mid-Pacific have been used to validate the ice-melting histories of glacial isostatic adjustment (GIA) models. However, a lack of quality control in the reconstruction of RSL hinders the understanding of regional variability that can constrain ice-equivalent sea-level changes. Here, we present a standardised database of Holocene RSL data from five regions in the mid-Pacific (Cook Islands, Tuamotu Islands, Christmas (Kiritimati) Island, Gilbert Islands and Fiji). We categorised the data as high or low quality based on the susceptibility of samples to age and/or elevation errors. Of the 614 data points that were reviewed, 25% were rejected and 100 sea-level index points (SLIPs) were reinterpreted as limiting data. The new database consists of 141 SLIPs and 262 marine and 56 terrestrial limiting data points reconstructed from a variety of sea-level indicators (e.g., coral microatolls, mangrove peat, beachrock, and beach ridges), of which 71% provide high-quality constraints on RSL. The early to mid Holocene RSL evolution in the Cook Islands, Gilbert Islands and Fiji are poorly constrained due to a lack of high-quality SLIPs and limiting data during this period. The Tuamotu Islands provided the only record of early to mid Holocene evolution of RSL, indicating rapid RSL rise from between −22.9 m and −15.2 m at ∼9.0 ka to between −0.2 m and 0.5 m by ∼6.5 ka, at rates as high as 9.8 ± 5.1 mm/a, with a slowdown in the rate of RSL rise sometime between ∼8.2 ka and ∼6 ka. The Christmas (Kiritimati) Island record indicates stable RSL within ∼1.5 m of present-day levels over the past ∼6.6 ka. In the late Holocene, the Cook Islands record suggests a gradual fall in RSL over the past ∼2.9 ka at rates below 0.1 ± 4.3 mm/a. SLIPs at Fiji also indicate a slight fall in RSL at rates of less than 0.5 ± ∼4.4 mm/a at ∼4 ka, following which RSL fell from above 0.9 m at ∼3 ka to between −0.3 m and 0.6 m by ∼2.5– ∼2.1 ka. We highlight the importance of standardisation and quality control to critically evaluate the processes controlling RSL and validate GIA models. Indeed, the new standardised database has implications for the timing of the mid- Holocene highstand, which has been used to support the ICE-4G and ICE-7G_NA models. Due to the poor constraints of data in the mid-Pacific islands, particularly in the early Holocene, there remains no unique solution for a global ice-melting history.