An assessment of the use of sediment traps for estimating upper ocean particle fluxes

This review provides an assessment of sediment trap accuracy issues by gathering data to address trap hydrodynamics, the problem of zooplankton "swimmers," and the solubilization of material after collection. For each topic, the problem is identified, its magnitude and causes reviewed using selected examples, and an update on methods to correct for the potential bias or minimize the problem using new technologies is presented. To minimize hydrodynamic biases due to flow over the trap mouth, the use of neutrally buoyant sediment traps is encouraged. The influence of swimmers is best minimized using traps that limit zooplankton access to the sample collection chamber. New data on the impact of different swimmer removal protocols at the US time-series sites HOT and BATS are compared and shown to be important. Recent data on solubilization are compiled and assessed suggesting selective losses from sinking particles to the trap supernatant after collection, which may alter both fluxes and ratios of elements in long term and typically deeper trap deployments. Different methods are needed to assess shallow and short- term trap solubilization effects, but thus far new incubation experiments suggest these impacts to be small for most elements. A discussion of trap calibration methods reviews independent assessments of flux, including elemental budgets, particle abundance and flux modeling, and emphasizes the utility of U-Th radionuclide calibration methods. In a synthesis of four annual time-senes 234Th trap comparison studies in the upper ocean, a factor of two trap under-collection of 234Th-carrying particles was found. At present it is unclear whether this under-collection reflects missed episodic flux events, issues with the comparison between radionuclide models and trap flux, or a more general pattern that should be applied to other elemental fluxes. A summary of recommendations for best practices, ways to compensate for biases, and documentation of methods used is also provided.