The marine reservoir effect in the southern ocean: An evaluation of extant and new Δr values and their application to archaeological chronologies

The last phase of human colonisation of the south‐west Pacific occurred around the 12th Century AD amongst the islands of the subtropical and temperate zones of the Southern Ocean (i.e., Norfolk Island, the Kermadec Islands, New Zealand and the Chatham Islands) (Fig. 1). Archaeological evidence indicates that initial colonisation of the region was rapid, possibly taking less than 100 years. Unfortunately, the chronology of these events has been largely reliant on charcoal radiocarbon determinations that have not been successful in delineating these changes. This paper investigates the use of marine shell for radiocarbon dating as an alternative to charcoal, with specific focus on variation in the marine¹⁴C reservoir (specifically the ?R) of Raoul Island (Kermadec Islands), Norfolk Island and Chatham Island (Fig. 2). Results from known‐age, pre‐AD 1950 shellfish indicate that ?R values south of the South Pacific Convergence Zone are low compared to those recorded for islands within the South Pacific generally (Fig. 1) (see Petchey et al. in press). An average ?R value of ‐19 ± 13¹⁴C yr is recorded for Raoul Island and ‐49 ± 10¹⁴C yr for Norfolk Island. These values are attributed to heightened absorption of atmospheric CO₂in this region. Extant published ?R information from New Zealand (Fig. 2A) also suggests a low average ?R value for these southern waters, but the range of values indicates that considerable variability is possible because of the complex interplay of currents around the New Zealand coastline. In contrast, results from Chatham Island are more variable and much higher (average ?R = 134 ± 83¹⁴C yr). Stable oxygen and carbon isotope data support the hypothesis that these ?R values are caused by upwelling and mixing of¹⁴C‐depleted water along the Chatham Rise. Comparison of marine and charcoal¹⁴C determinations from archaeological sites on Raoul Island support, within the limits of the available data, the ?R values obtained, but further analysis is required to establish the stability of this value over time.