Global patterns in seagrass leaf and sediment carbon isotope fractionation have implications for carbon provenance calculations in blue carbon accreditation
Emma A. Ward, Marianna Cerasuolo, Federica Ragazzola, Sarah Reynolds, Joanne Preston
Abstract
Seagrass carbon sequestration is known to be an accumulative process of both autochthonous and allochthonous carbon sequestration, however, carbon accreditation focuses on increasing autochthonous organic carbon only. In seagrass carbon accreditation methodologies peer-reviewed published data may be utilised as evidence for the deduction of a percentage of allochthonous carbon from the total seagrass sediment organic carbon. These literature-based proxies are often derived from stable isotope mixing models, which utilise seagrass and sediment δ 13 C values. This study looks at global seagrass sediment and leaf δ 13 C analyses, and demonstrates that climatic bioregion, geomorphology and seagrass morphological traits explain global patterns in seagrass leaf and sediment isotope δ 13 C ratios. Multi-factor analysis of mixed data shows a separation between seagrass bioregions and different leaf-size populations, specifically; north temperate regions from tropical and south temperate regions; medium leaf-size individuals to all others. Analysis of variance confirmed a significant difference (p < 0.001) in the Δδ 13 C seagrass‐sediment between bioregions and species sizes classifications. KMeans clustering of the seagrass and sediment δ 13 C and sediment depth data suggests that three main clusters can be identified (1) small deltas, (2) tidal systems coastlines, (3) an aggregation of lagoons, arheic and fjords coastlines. If proxies are used for blue carbon accreditation, this paper presents an informed criterion to improve the selection of allochthonous sediment organic carbon proxies based on their derivative sediment and seagrass δ 13 C values. However, proxy values from the literature are not a direct substitute for site specific δ 13 C seagrass leaf and sediment data, and their use in context dependent mixing models.