Identifying Drivers and Modelling Variability of Blue Carbon Stocks

Tidal marshes, mangroves, and seagrasses serve as dense ‘blue carbon’ (C) sinks. However, these ecosystems are rapidly declining with little understanding of the variability in organic C stocks associated with them or the drivers behind it, presenting challenges for strategic management actions. Our research aims were three-fold: (1) identify ecological, geomorphological, and anthropogenic variables driving variability in C stocks, (2) create a predictive model, and (3) map regional C stock variability at a scale relevant to resource management. First, we undertook a large-scale sampling campaign of sediment C across 96 blue carbon ecosystems in a data-deficient region of southeast Australia. We then combined these C data with spatially explicit environmental data to identify drivers and map variability. We used an information theoretic approach utilizing model selection, dredging, averaging, and validation to generate the best linear mixed effect model for predicting carbon stocks in the region. Mean sediment C_org stock (±SE) to a depth of 30 cm was not significantly different between saltmarsh (87.1 ± 4.90 Mg C_org ha^-1) and mangroves (65.6 ± 4.17 Mg C_org ha^-1), but was significantly lower in seagrasses (24.3 ± 1.82 Mg C_org ha^-1). We identified vegetation type (e.g. salt marsh) as the most important driver of C stock variability, followed by geomorphological variables (e.g. slope), with anthropogenic variables (e.g. land use) being of least importance. Based on our 10-m spatial resolution model, combined with available data on ecosystem extent, we estimated a total of 2.31 million Mg C stored in the top 30 cm of blue C sediments, of which 88% was contained within four major coastal areas due to ecosystem extent. These data can support targeted management of blue C hotspots regionally and these methods can be applied to develop predictive C stock maps at resolutions relevant to resource management in other regions of the globe.