Fig. 2: Deconstructing POC flux attenuation.

a–c, Microbially mediated component of the POC flux attenuation, MR (a), the residual POC flux (b) and the cumulative POC flux (c) as a function of depth at the six study sites investigated (same colour code as for Fig. 1c). Note the different flux scales for SG and BEN and for all sites in a. Owing to ship logistics, incubation times varied by threefold between sites (Extended Data Table 2). The coloured curves correspond to the best fit of a power-law model (not fitted for MED and SPSG where only two depths below the PPZ were sampled), and error bars were derived using uncertainty-propagation laws (Methods). These Martin curves are presented only to demonstrate that the cumulative POC flux approach reproduces a power-law attenuation trend. A fixed RQ of 0.70 (that is, for plankton-based particles such as marine snow⁴⁵) was used to calculate MR. The coloured-shaded rectangles in a represent the range of values of MR obtained based on the upper (that is, 0.89 for phytoplankton cells⁴⁴) and lower (that is, 0.50 for methane oxidation within particles, the theoretical minimum for RQ⁴⁶) bounds of the range of RQ values used in the sensitivity analysis (Methods). The beige-shaded areas represent the PPZ where new particles are produced.