Estimating Carbon Flux From Optically Recording Total Particle Volume at Depths Below the Primary Pycnocline

Alexander B Bochdansky,Dennis A Hansell,Robert B Dunbar,Gerhard J Herndl

doi:10.3389/fmars.2019.00778

Abstract

Optical instruments can rapidly determine numbers and characteristics of water column particles with high sensitivity. Here we show the usefulness of optically assessed total particle volume below the main pycnocline to estimate carbon export in two systems: the open subarctic North Atlantic and the Ross Sea, Antarctica. Both regions exhibit seasonally high phytoplankton production and efficient export (i.e., a strong biological pump). Total particle volumes in the mesopelagic (200-300 m) were significantly correlated with those in the overlying surface mixed layer (50 - 60 m), indicating that most particles at depth reflect export from the surface. This connectivity, however, is modulated by the physical structure of the water column and by particle type (e.g., the presence of colonies of the haptophyte Phaeocystis antarctica versus diatoms). Evidence from both regions show that a strong pycnocline can delay or may even prevent particles from settling to deeper layers, which then succumb to disintegration, or microbial and/or zooplankton consumption. Strong katabatic winds in the Ross Sea may deepen the mixed layer, causing a rapid transfer of particles to mesopelagic depths through the mixed-layer pump. Independent estimates of seasonally integrated export production in the Ross Sea, based on upper water column carbon mass balance, were significantly correlated (in the order of shared variance) with 1) total particle volumes from images, 2) particulate organic carbon, and 3) chlorophyll fluorescence, all recorded at a depth range of 200 – 300 m. Carbon export was not significantly correlated with particle abundance measured by a Coulter Counter at the same depth range. Measuring total particle volume below the primary pycnocline is therefore a useful approach to estimate carbon export at least in regions characterized by seasonally high particle export.

Highlights

Imaging systems have been used to characterize plankton and particles in the sea for many decades
We focused on three areas in the western Ross Sea: north of Franklin Island (“south site”), south of Coulman Island (“north site”), and Terra Nova Bay (TNB), each revisited several times during the expedition to record temporal changes (DeJong et al, 2017)
The density structure was very different in the Arctic basin with considerably higher sigma-t values overall than those observed in the subarctic North Atlantic (Figure 2)

Summary

Introduction

Imaging systems have been used to characterize plankton and particles in the sea for many decades. In contrast to other optical systems such as transmissometry, optical backscattering, forward scattering laser diffraction, and fluorometry, particle imaging systems allow a much more detailed analysis of particle morphology such as size, aspect ratio, roughness, and porosity using relatively simple imageanalytical procedures. Image surveys can further be expanded to the classification of particles. We are at a threshold where machine learning will greatly improve throughput in the classification of particles once trained on specific targets by expert human operators (e.g., Zheng et al, 2017; Luo et al, 2018). Imaging systems can accurately survey relatively rare macroscopic particles in areas such as the deep sea where other optical systems such as optical backscatter or transmissometry reach their lower detection thresholds (Bochdansky et al, 2016)

Methods

Results

Discussion

Conclusion