Assessing the source and delivery processes of organic carbon within a mixed land use catchment using a combined n-alkane and carbon loss modelling approach

Abstract

Abstract Purpose Understanding fluxes of soil organic carbon (OC) from the terrestrial to aquatic environments is crucial to evaluate their importance within the global carbon cycle. Sediment fingerprinting (SF) is increasingly used to identify land use-specific sources of OC, and, while this approach estimates the relative contribution of different sources to OC load in waterways, the high degree of spatial heterogeneity in many river catchments makes it challenging to precisely align the source apportionment results to the landscape. In this study, we integrate OC SF source apportionment with a carbon loss model (CLM) with the aim of: (i) reducing ambiguity in apportioning OC fluxes when the same land use exists in multiple locations within a catchment; and (ii) identifying factors affecting OC delivery to streams, e.g., buffer zones. Methods Two main approaches were used in this study: (i) identification of the sources of freshwater bed sediment OC using n-alkane biomarkers and a Bayesian-based unmixing model; and (ii) modelling and analysis of spatial data to construct a CLM using a combination of soil OC content modelling, RUSLE soil erosion modelling and a connectivity index. The study was carried out using existing OC and n-alkane biomarker data from a mixed land use UK catchment. Results Sediment fingerprinting revealed that woodland was the dominant source of the OC found in the streambed fine sediment, contributing between 81 and 85% at each streambed site. In contrast, CLM predicted that arable land was likely the dominant source of OC, with negligible inputs from woodland. The areas of the greatest OC loss in the CLM were predicted to be from arable land on steeper slopes surrounding the stream channels. Results suggest extensive riparian woodland disconnected upslope eroded soil OC and, concomitantly, provided an input of woodland-derived OC to the streams. It is likely the woodland contribution to streambed OC is derived from litter and leaves rather than soil erosion. Conclusion This study demonstrates how location-specific OC sources and delivery processes can be better determined using sediment fingerprinting in combination with CLM, rather than using sediment fingerprinting alone. It highlights that, although wooded riparian buffer strips may reduce the impact of upslope, eroded soil OC on waterways, they could themselves be a source of OC to stream sediments through more direct input (e.g., organic litter or leaf debris). Characterising this direct woodland OC as a separate source within future fingerprinting studies would allow the contributions from any eroded woodland soil OC to be better estimated.