ARTICLES PUBLISHED ONLINE: 9 OCTOBER 2017 | DOI: 10.1038/NGEO3042

Biological and physical influences on marine snowfall at the equator R. Kiko1*, A. Biastoch1, P. Brandt1,2, S. Cravatte3, H. Hauss1, R. Hummels1, I. Kriest1, F. Marin3, A. M. P. McDonnell4, A. Oschlies1,2, M. Picheral5, F. U. Schwarzkopf1, A. M. Thurnherr6 and L. Stemmann7 High primary productivity in the equatorial Atlantic and Pacific oceans is one of the key features of tropical ocean biogeochemistry and fuels a substantial flux of particulate matter towards the abyssal ocean. How biological processes and equatorial current dynamics shape the particle size distribution and flux, however, is poorly understood. Here we use highresolution size-resolved particle imaging and Acoustic Doppler Current Profiler data to assess these influences in equatorial oceans. We find an increase in particle abundance and flux at depths of 300 to 600 m at the Atlantic and Pacific equator, a depth range to which zooplankton and nekton migrate vertically in a daily cycle. We attribute this particle maximum to faecal pellet production by these organisms. At depths of 1,000 to 4,000 m, we find that the particulate organic carbon flux is up to three times greater in the equatorial belt (1◦ S–1◦ N) than in off-equatorial regions. At 3,000 m, the flux is dominated by small particles less than 0.53 mm in diameter. The dominance of small particles seems to be caused by enhanced active and passive particle export in this region, as well as by the focusing of particles by deep eastward jets found at 2◦ N and 2◦ S. We thus suggest that zooplankton movements and ocean currents modulate the transfer of particulate carbon from the surface to the deep ocean.

T

he distribution and fate of particulate organic matter in the ocean is a result of the interaction of primary production near the surface and advection, sinking, grazing, and remineralization at various depths1–3 . Primary production leads to the uptake of carbon dioxide, whereas remineralization— the breakdown of complex organic molecules back to inorganic forms—leads to carbon dioxide release. At which depth in the ocean remineralization occurs is crucial, as oceanic carbon sequestration—the export of particulate, but also dissolved carbon out of the upper ocean to below the maximum depth of near-surface seasonal mixing4 —is an important aspect of climate regulation5,6 . In the global ocean the export path of particulate matter was found to be controlled by particle properties such as size, density and porosity3,7 , interactions with microbes and metazoans8 , and mesoscale hydrodynamics9,10 . Passive sinking of particulate matter such as detritus, faecal pellets and other particles—which may also combine to form macroscopic aggregates (>0.5 mm) termed marine snow11 —plays a major role in the export of particulate organic matter to depth12 . Mortality and defecation of zooplankton and nekton organisms that migrate daily between the surface layer, where they feed at night-time, and mesopelagic depth, where they hide from visual predation during daytime, also causes a substantial active flux of particulate organic matter to depth13–16 . Marine snow aggregates, sinking at tens to hundreds of metres per day7,17 , have been linked to intense flux events in the deep sea18 . However, recent observations indicate that smaller, slowly sinking particles also contribute significantly to particulate matter flux in the upper 500 m of the ocean19,20 . Their source seems to be the disaggregation

of larger particles and faecal pellets at depth, for example through fragmentation by zooplankton21–23 . Because their slow settling speed exposes them to bacterial remineralization, the contribution of small particles to total particle flux below 1,000 m depth was thought to be negligible19 . Seasonally and regionally increased net primary production in the equatorial Atlantic and Pacific oceans24,25 is linked to enhanced nutrient supply via Ekman-driven equatorial upwelling26 , diapycnal mixing27 and horizontal advection24 (see also Fig. 1, Supplementary Figs 1–6). Sediment trap measurements12 and independent particle size spectra observations28–30 confirm the link between high surface productivity and carbon sequestration in these regions, but the path that settling particles take through the meso- and bathypelagic towards the abyssal ocean below 4,000 m depth is uncertain. Mean Atlantic and Pacific equatorial zonal current velocities are on the order of 10 km d−1 for the deep and intermediate currents and up to 100 km d−1 for the Equatorial Undercurrent31,32 , with mean meridional current velocities generally an order of magnitude smaller than zonal ones33,34 . How these currents impact vertical flux of particles having sinking velocities of tens to few hundred metres per day17 has until now remained unresolved.

Equatorial currents and particulate matter distribution We conducted several full-depth transects combining Underwater Vision Profiler 5 (UVP5) (ref. 35) and Acoustic Doppler Current Profiler (ADCP) measurements to study the mechanisms of equatorial particle dynamics in more detail. Our Lowered-ADCP (L-ADCP) measurements in the Pacific and Atlantic support earlier

1

GEOMAR Helmholtz Centre for Ocean Research Kiel, Düsternbrooker Weg 20, 24105 Kiel, Germany. 2 Faculty of Mathematics and Natural Sciences, University of Kiel, 24148 Kiel, Germany. 3 LEGOS, University of Toulouse, IRD, CNES, CNRS, UPS, 31400 Toulouse, France. 4 University of Alaska Fairbanks, College of Fisheries and Ocean Sciences, Fairbanks, Alaska 99775-7220, USA. 5 CNRS, UMR 7093, Laboratoire d’Océanographie de Villefranche-sur-Mer (LOV), Observatoire Océanologique, F-06230 Villefranche-sur-Mer, France. 6 Lamont-Doherty Earth Observatory, Palisades, New York 10964-8000, USA. 7 Sorbonne Universités, UPMC Univ Paris 06, UMR 7093, LOV, Observatoire Océanologique, F-06230 Villefranche-sur-Mer, France. *e-mail: [email protected] NATURE GEOSCIENCE | ADVANCE ONLINE PUBLICATION | www.nature.com/naturegeoscience © 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

1

NATURE GEOSCIENCE DOI: 10.1038/NGEO3042

ARTICLES Depth (m)

a

b

P1: Sep. 2011 200 400 600 800 1,000

−4 −2 0 2 4 Latitude (°)

I1: Apr. 2010

−4 −2 0 2 4 Latitude (°)

1,350 1,200

d

1,050

Latitude

20° N

900 750

0°

600 20° S

450

100° E

140° E

180° E

140° W

100° W

60° W

20° W

20° E

60° E

100° E

Longitude

e

P3: Aug. 2015

f P4: Aug. 2015

g

h

P5: Apr. 2015

A1: May 2014

i A2: Sep. 2015

j A3: Nov. 2012

300 150 0

Net primary production (mgC m−2 d−1)

−4 −2 0 2 4 Latitude (°)

c

P2: May 2011

22.5

Depth (m)

17.5 15.0

2,000

12.5 10.0

3,000

7.5 5.0

4,000

Particles 0.14 to 16.88 mm (# L−1)

20.0

1,000

2.5 5,000 −4 −2 0 2 4 Latitude (°)

−4 −2 0 2 4 Latitude (°)

−4 −2 0 2 4 Latitude (°)

−4 −2 0 2 4 Latitude (°)

−4 −2 0 2 4 Latitude (°)

−4 −2 0 2 4 Latitude (°)

0.0

Figure 1 | Primary production is enhanced at the Atlantic and Pacific equator and fuels marine equatorial snowfall. a–j, Satellite-derived mean annual net primary productivity for 2010 to 2015 (d) and particle abundanceprofiles (a–c and e–j) across the equator obtained between April 2010 and September 2015. Particle size range is 0.14 to 16.88 mm diameter. Net primary productivity values exceeding the colour scale are represented in white.

studies31,32 , showing that the intermediate and deep equatorial current systems of the Atlantic and Pacific consist of latitudinally alternating zonal jets with nearly depth-independent eastward flow at 2◦ N and 2◦ S (Figs 2 and 3). Furthermore, vertically alternating eastward and westward jets with a vertical extent of 300 to 700 m were again observed between about 1◦ N and 1◦ S (Figs 2 and 3). Our UVP5 observations reveal enhanced abundances of particles near the equator even at bathypelagic depth (Figs 1–3, Supplementary Table 1, Transects P5 and A1 to A3). Small, micrometric particles (MiPs; particles with 0.14 to 0.53 mm diameter) are almost exclusively confined to a 2◦ N to 2◦ S latitudinal band at the Atlantic and the central Pacific equator (Figs 2 and 3, Supplementary Table 2, Transects P5 and A1 to A3) consistent with a narrow latitudinal surface productivity (Fig. 1 and Supplementary Figs 1 and 3). A broader distribution of both satellite-derived primary productivity and MiP abundance at depth can be found in the less productive western part of the Pacific equator (Figs 1–3, Transects P3 and P4), outside of the major equatorial upwelling. The abundance of large macroscopic particles (MaPs; particles with 0.53 mm to 16.88 mm diameter) varies more strongly, but is also enhanced at the equator (Figs 2 and 3 and Supplementary Table 3). In contrast, particle abundance in the equatorial region of the Indian Ocean and surface productivity was found to be low (Fig. 1—Transect I1), which can be explained by the lack of steady equatorial easterlies, associated upwelling and the concomitant localized nutrient input into the euphotic zone at the equator. Size–sinking speed relationships36 suggest relatively high (20 to >100 m d−1 ) sinking speeds and a locally constrained export route for MaPs. Our particle backtracking experiments show 2

that this also holds true for the equatorial system. Even the strong equatorial currents do not affect the export localization substantially when sinking speeds >50 m d−1 are assumed (Supplementary Fig. 7). That upwelling precedes or coincides with the occurrence of MaPs even at bathypelagic depth indicates that a fast and deep penetrating flux is mainly generated in bloom situations, consistent with earlier observations of flux events18 . Our observations at 23◦ W in September 2015 and November 2012 were preceded by upwelling (Supplementary Fig. 1, Transects A2 and A3) and MaP abundance at bathypelagic depth was about twofold higher in November 2012 (Fig. 2, Transect A3) compared to the non-upwelling situation in May 2014 (Fig. 2, Transect A1; see also Supplementary Table 3). Comparatively low MaP abundance at depth in September 2015 (Fig. 3, Transect A2) directly after the upwelling may be related to the rather episodic nature of flux events which, depending on the particle properties, can last only a few days6,17 . Almost no MaPs were observed in May 2014, when sampling at 23◦ W (Fig. 2, Transect A1) was preceded by several months of increased surface temperatures and relatively low primary productivity (Supplementary Fig. 1). MaP abundance at depth was also low in the western Pacific (Fig. 3, Transects P3 and P4), where net primary productivity is low (Fig. 1 and Supplementary Figs 1 and 2), whereas it was high in the central Pacific (Fig. 3, Transect P5), where continuous upwelling was observed prior to sampling (Supplementary Fig. 3).

Mesopelagic particle distribution and flux An increase especially in MiP, but also in MaP abundance (Figs 2 and 3) and calculated particulate organic carbon (POC) flux36 (Fig. 4) at about 200 to 600 m depth is a dominant feature

NATURE GEOSCIENCE | ADVANCE ONLINE PUBLICATION | www.nature.com/naturegeoscience © 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

NATURE GEOSCIENCE DOI: 10.1038/NGEO3042 b

A1: May 2014

a

c

1,000

1,000

2,000

2,000

Depth (m)

Depth (m)

a

ARTICLES

3,000

4,000

e

A2: Sep 2015

f

e

P4: Aug. 2015

f

h

P5: Apr. 2015

i

3,000

d

1,000

1,000

2,000

2,000

Depth (m)

Depth (m)

c

5,000

5,000

3,000

4,000

3,000

4,000

5,000

5,000

g

h

A3: Nov. 2012

i

g

1,000

1,000

2,000

2,000

Depth (m)

Depth (m)

P3: Aug. 2015

4,000

d

3,000

3,000

4,000

4,000

5,000

b

−4 −2 0 2 4 Latitude (°)

−0.8 −0.4

0.0 0.4

Zonal current velocity (m s−1)

−4 −2 0 2 4 Latitude (°)

0.00 0.15 0.30 0.45 POC (mg m−3)

5,000

−4 −2 0 2 4 Latitude (°)

−0.8 −0.4 0.0 0.4

0.00 0.05 0.10 0.15

Zonal current velocity (m s−1)

POC (mg m−3)

Figure 2 | Zonal current velocity and calculated POC content transects across the equatorial Atlantic at 23◦ W. a–i, Data were obtained in May 2014 (a–c; Transect A1); September 2015 (d–f; Transect A2); and November 2012 (g–i; Transect A3). From left to right: Zonal current velocity, MiP POC content and MaP POC content. Please note different scaling for MiP and MaP POC content. Eastward current velocity values exceeding the upper limit of the colour scale are represented in dark red.

at mesopelagic depth and seems to be an important step in particulate matter flux to depth. We observed this intermediate particle maximum (IPM) within the westward flowing Equatorial Intermediate Current (EIC; Fig. 2 transects A1 and A3, Fig. 3 transect P5), but also within extra-equatorial eastward currents (Fig. 2 transects A2 and A3, Fig. 3 transects P3 and P4). An IPM

−4 −2 0 2 4 Latitude (°)

−4 −2 0 2 4 Latitude (°)

0.00 0.15 0.30 0.45 POC (mg m−3)

−4 −2 0 2 4 Latitude (°)

0.00 0.05 0.10 0.15 POC (mg m−3)

Figure 3 | Zonal current velocity and calculated POC content transects across the equatorial Pacific at different longitudes. a–i, Data was obtained at 157.5◦ E (a–c; Transect P3) and 165◦ E (d–f; Transect P4) in August 2015, and at 151◦ W (g–i; Transect P5) in April 2015. From left to right: Zonal current velocity, MiP POC content and MaP POC content. Please note different scaling for MiP and MaP POC content. Eastward current velocity values exceeding the colour scale are represented in dark red.

within the EIC was also reported earlier28–30 and was suggested to be caused by meridional funnelling of slowly sinking particles towards the equator via a convergent equatorward flow at 100 to 300 m depth28 . More recent observations in the Atlantic33 and Pacific34 indicate that the equatorward meridional flow is relatively strong at

NATURE GEOSCIENCE | ADVANCE ONLINE PUBLICATION | www.nature.com/naturegeoscience © 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

3

NATURE GEOSCIENCE DOI: 10.1038/NGEO3042

ARTICLES 0

P3: Aug. 2015

P4: Aug. 2015

P5: Apr. 2015

Depth (m)

1,000 2,000 3,000 4,000 5,000 −5 0

0

5

10

15 −5

A1: May 2014

0

5

10

15 −5

A2: Sept. 2015

0

5

10

15

A3: Nov. 2012

Depth (m)

1,000 2,000 3,000 4,000 5,000 −5

0

5

10 15 −5

0

5

10

15 −5

0

5 −2

Sv day−night difference (dB) or POC flux (mgC m

10

15

−1

d )

Figure 4 | POC flux increase coincides with diel vertical migration depth at the equatorial Atlantic and Pacific. Day–night difference in 38 kHz (blue) or 75 kHz (cyan) ADCP volumetric backscatter Sv (dB) observed between 1◦ S and 1◦ N. Shading indicates statistical difference of the signal to zero (one-sided Students t-test, p-value < 0.01). Red lines and shading indicate the calculated mean POC flux and standard deviation, respectively, averaged for the 1◦ S to 1◦ N section. Dotted black lines indicate the POC flux extrapolated from the POC flux observed at 100 m depth using a Martin-curve approach38 with an exponent b = −0.86.

about 50 to 100 m depth, but weak below this depth. Particle tracking (Supplementary Videos 1–5 and Supplementary Fig. 8) also shows that small, slowly sinking particles observed in the EIC IPM in the Atlantic cannot be linked to such an export route from the surface. Very slowly sinking particles (0 to 2 m d−1 sinking speed) are advected eastwards and off-equatorial, but fail to return to the equator, whereas faster sinking particles (>10 m d−1 sinking speed) are not affected by the meridional flow. As we assume sinking speeds of 10 to 20 m d−1 for the MiP size class and >20 m d−1 for the MaP size class, it seems unlikely that the IPM signal in these size classes is caused by meridional funnelling. The rapid eastward transport of MiPs by the Equatorial Undercurrent below the surface down to 200 m depth would furthermore require a productive source region west of the 23◦ W section to sustain an increase in particle abundance at mesopelagic depth. Primary productivity west of the 23◦ W section, however, is comparatively low (Fig. 1). The fact that IPMs can also be observed outside the EIC in eastward and westward currents furthermore supports the interpretation that the IPM generation mechanism is not directly related to the equatorial current field. Part of the flux that fuels the IPM could be brought to IPM depth through gut flux associated with the diel vertical migration of zooplankton and nekton. Diel migrators feed at the surface during night-time and migrate to mesopelagic depth during daytime where released faecal pellets constitute the active gut flux13–16,37 . At the Atlantic and Pacific equator the migration generally reaches 300–500 m depth37 . To sustain a POC increase of 0.08–0.11 mgC m−3 over a 400 m depth range, as observed in the strongest IPMs at Transects P5 and A3 would require POC 4

gut fluxes of 6.6–8.5 mgC m−2 d−1 , respectively (Supplementary Fig. 9). Maximum mesozooplankton gut fluxes of 4.5 mgC m−2 d−1 to below 200 m depth were observed at the oligotrophic Bermuda Atlantic Time Series station13 and of 21.0 mgC m−2 d−1 for the North Atlantic14 . Assuming that active POC gut flux can be approximated as 1%–4% of the migrating biomass (calculated from Table 4 in ref. 13 and Table 4 in ref. 14, respectively), results in maximum gut fluxes of 1.5–6.2 mgC m−2 d−1 associated with the migration of only mesozooplankton to below 120 m depth for the equatorial Pacific at 140◦ W (ref. 15). A study that also included micronekton revealed gut fluxes of 2.4–19.4 mgC m−2 d−1 associated with the migration of mesozooplankton and micronekton to below 160 m depth for the equatorial Pacific at 145◦ E (ref. 16). Comparison of day- and nighttime biomass distribution based on 75 and 38 kHz shipboard ADCP backscatter data (Fig. 4 and Supplementary Fig. 10) shows that an increase in acoustic scatterer biomass (size range of scatterers >5 and >10 mm, respectively) can be observed at the upper border of the IPM at 1◦ S to 1◦ N in the Atlantic and Pacific. In May 2014 (Fig. 4, transect A1) the increase in POC flux at depth is minor, coinciding with a relatively small increase in ADCP backscatter at depth (see also Supplementary Fig. 10, transect A1). Our analysis further shows that at this time of low migration activity the extrapolation of the POC flux from surface values to depth using the often-cited Martincurve approach38 seems reasonable. Importantly, the Martin-curve approach that assumes a decline of particulate matter flux according to a power law with an exponent of −0.86 fails to explain the POC flux increase at IPM depth when strong migration activity was observed. Zooplankton and nekton gut fluxes reported in or calculated from the literature13–16 would therefore largely suffice to explain the IPM, and our analysis suggests a consistency between migration activity and POC flux increase at IPM depth.

Bathypelagic particle distribution and flux Bathypelagic POC flux averaged for the 1,100–1,300 m and the 2,900–3,100 m depth layers inside the equatorial belt at 1◦ S to 1◦ N is generally elevated in comparison to the off-equatorial regions and enhanced by up to a factor of three (Fig. 5; see also Supplementary Table 4). These observations are in line with earlier sediment trap studies that show enhanced POC flux in the equatorial upwelling regions of the Atlantic and Pacific12 . Enhanced MaP flux during or after times of high upwelling activity at the equator seems to be one factor in POC flux increase at the equator. MiP POC content and flux are almost constant throughout the entire bathypelagic depth range and shape the baseline flux with elevated values at the Atlantic and western Pacific equator (Figs 2, 3 and 5 and Supplementary Table 2). Importantly, MiPs fuel on average 65 to 86% of total POC flux at the Pacific and 78–80% at the Atlantic equator at 2,900–3,100 m depth. These observations are contrary to the assumption that small particles contribute little to particulate matter flux below 1,000 m depth19 . Many global biogeochemical models parameterize a downward increasing sinking speed of marine particles39,40 , attributing particle flux at bathypelagic depth to MaPs41 . Also the often-applied powerlaw flux function42 approximates to vertically increasing sinking speed and/or vertically decreasing remineralization rates43 . These assumptions are made because slow sinking particles formed at the surface should be remineralized at shallower depth, precluding their penetration to depth. However, we observe high concentrations of MiPs at bathypelagic depth, an observation contrary to these model assumptions. Assuming sinking speeds between 10 and 20 m d−1 (calculated from the size limits of the MiP class and published size-sinking speed relationships36 ) and remineralization rates between 0.03–0.12 d−1 (ref. 44), MiP transfer to 3,000 m depth of at least 150 days would exceed their lifetime. Likewise, if we assume a maximum sinking speed of 15 m d−1 and a remineralization rate of 0.05 d−1 , only 0.005% of the export flux

NATURE GEOSCIENCE | ADVANCE ONLINE PUBLICATION | www.nature.com/naturegeoscience © 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

P3: Aug. 2015

a

14 12 10 8 6 4 2 0

b

P4: Aug. 2015

c

ARTICLES

P5: Apr. 2015

d

A1: May 2014

e

A2: Sept. 2015

f

A3: Nov. 2012

100

g

h

i

j

k

l

80

10 8

60

6 4

40

2

% MiP flux

0 POC flux (mgC m−2 d−1)

POC flux (mgC m−2 d−1)

POC flux (mgC m−2 d−1)

NATURE GEOSCIENCE DOI: 10.1038/NGEO3042

m

n

o

p

q

20

6 5

0

4 3 2 1 0

−4 −2

0

2

Latitude (°)

4

−4 −2

0

2

4

Latitude (°)

−4 −2

0

2

4

Latitude (°)

−4 −2

0

2

Latitude (°)

4

−4 −2

0

2

4

Latitude (°)

Figure 5 | POC flux at bathypelagic depth is dominated by micrometric particles and enhanced at the equator. a–q, Calculated MiP (dark grey) and MaP POC flux (light grey) averaged for 650–750 m (a–f), 1,100–1,300 m (g–l) and 2,900–3,100 m depth bins (m–q). Total POC flux is indicated by the entire shaded area. Percentage of MiP flux to total flux for each sampled profile is indicated by coloured dots at the bottom of each panel. Diamonds refer to published12 annual averages (white) or weekly measurements from the observation season (grey) of POC flux from sediment trap data.

will reach 3,000 m depth. To sustain the observed MiP POC flux of ∼2 mgC m−2 d−1 at 3,000 m depth exclusively via MiPs would therefore require an export flux of ∼32,000 mgC m−2 d−1 , which by far exceeds the observed production in this region (Fig. 1). A suggested very low45 , or vertically decreasing46 , degradability could explain the transport of these particles to depth: to sustain the observed MiP POC flux of ∼2 mgC m−2 d−1 at 3,000 m depth exclusively via MiPs, and assuming complete export of surface production of ∼800 mgC m−2 d−1 as MiPs (Fig. 1), would therefore require a remineralization rate of 0.03 d−1 , which is at the lower end of observed estimates for marine aggregates, measured at 4 ◦ C (ref. 44). However, recent observations suggest that remineralization rates of small, slowly sinking particles are enhanced in comparison to larger, faster sinking particles23 . As an alternative hypothesis we suggest that MiPs are generated at depth through shedding from or disintegration of larger, fast sinking MaPs, following a period of several weeks of microbial colonization and attack47 during the MaPs relatively fast (>20 m d−1 and up to >100 m d−1 ; ref. 17) descent to depth. Zooplankton-mediated fragmentation3,21–23 might also play a role at bathypelagic depth, although encounter rates should be low due to relatively low zooplankton and MaP abundances. As noted, MiPs in the bathypelagic zone are especially abundant between 2◦ S and 2◦ N and form a veil of small particles that can be observed down to the sea floor (Figs 2 and 3; Transects A1 to A3 and P5). Enhanced MiP abundance is likely related to the enhanced productivity at the surface and resultant stronger active and passive export of particulate matter in general. However, the deep equatorial current system also seems to favour the maintenance of this veil of small particles, as suggested by spatially coherent current and particle patterns. By enhancing the meridional gradient of potential vorticity48 , the largely depth-independent flanking eastward jets prevent the poleward dispersion of particles away from the

equatorial belt49 . The intermediate and deep equatorial jet system therefore constrains the latitudinal and favours the longitudinal extent of enhanced MiP abundance, and also contributes to a zonal advection of particulate matter from the more productive eastern areas.

Conclusions Our results suggest that gut flux via diel vertical migration creates a strong deviation of particle flux from the often-assumed powerlaw function for particle flux38 , shuttling particulate matter to the mesopelagic layer and thereby enhancing particulate matter flux to the bathypelagic. Contrary to model assumptions that assign bathypelagic particle flux to large, fast sinking particles39–42 , we observe that bathypelagic particle flux is mainly maintained by an almost constant flux of rather small particles down to >4,000 m depth. The poleward dispersion of this veil of particles is prevented by a high gradient of potential vorticity of the deep equatorial jets. Zooplankton diel vertical migrations, particulate matter disintegration, and the intermediate and deep equatorial currents seem key in the generation and maintenance of marine equatorial snowfall and for carbon sequestration in the deep ocean.

Methods Methods, including statements of data availability and any associated accession codes and references, are available in the online version of this paper. Received 2 July 2017; accepted 8 September 2017; published online 9 October 2017

References 1. Behrenfeld, M. J. & Falkowski, P. G. Photosynthetic rates derived from satellite-based chlorophyll concentration. Limnol. Oceanogr. 42, 1–20 (1997).

NATURE GEOSCIENCE | ADVANCE ONLINE PUBLICATION | www.nature.com/naturegeoscience © 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

5

NATURE GEOSCIENCE DOI: 10.1038/NGEO3042

ARTICLES 2. Burd, A. B. et al. Assessing the apparent imbalance between geochemical and biochemical indicators of meso-and bathypelagic biological activity: What the @$#! is wrong with present calculations of carbon budgets? Deep-Sea Res. II 57, 1557–1571 (2010). 3. Giering, S. L. et al. Reconciliation of the carbon budget in the ocean’s twilight zone. Nature 507, 480–483 (2014). 4. Lampitt, R. S. et al. Ocean fertilization: a potential means of geoengineering? Phil. Trans. R. Soc. A 366, 3919–3945 (2008). 5. Kwon, E. Y., Primeau, F. & Sarmiento, J. L. The impact of remineralization depth on the air-sea carbon balance. Nat. Geosci. 2, 630–635 (2009). 6. Falkowski, P. G., Barber, R. T. & Smetacek, V. Biogeochemical controls and feedbacks on ocean primary production. Science 281, 200–206 (1998). 7. Alldredge, A. L. & Gotschalk, C. In situ settling behavior of marine snow. Limnol. Oceanogr. 33, 339–351 (1988). 8. Steinberg, D. K. et al. Bacterial vs. zooplankton control of sinking particle flux in the ocean’s twilight zone. Limnol. Oceanogr. 53, 1327–1338 (2008). 9. Siegel, D. A. & Deuser, W. G. Trajectories of sinking particles in the Sargasso Sea: modeling of statistical funnels above deep-ocean sediment traps. Deep-Sea Res. I 44, 1519–1541 (1997). 10. Waite, A. M. et al. The wineglass effect shapes particle export to the deep ocean in mesoscale eddies. Geophys. Res. Lett. 43, 9791–9800 (2016). 11. Alldredge, A. L. & Silver, M. W. Characteristics, dynamics and significance of marine snow. Prog. Oceanogr. 20, 41–82 (1988). 12. Honjo, S., Manganini, S. J., Krishfield, R. A. & Francois, R. Particulate organic carbon fluxes to the ocean interior and factors controlling the biological pump: a synthesis of global sediment trap programs since 1983. Prog. Oceanogr. 76, 217–285 (2008). 13. Schnetzer, A. & Steinberg, D. K. Active transport of particulate organic carbon and nitrogen by vertically migrating zooplankton in the Sargasso Sea. Mar. Ecol. Prog. Ser. 234, 71–84 (2002). 14. Zhang, X. & Dam, H. G. Downward export of carbon by diel migrant mesozooplankton in the central equatorial Pacific. Deep-Sea Res. II 44, 2191–2202 (1997). 15. Morales, C. E. Carbon and nitrogen fluxes in the oceans: the contribution by zooplankton migrants to active transport in the North Atlantic during the Joint Global Ocean Flux Study. J. Plankton. Res. 21, 1799–1808 (1999). 16. Hidaka, K., Kawaguchi, K., Murakami, M. & Takahashi, M. Downward transport of organic carbon by diel migratory micronekton in the western equatorial Pacific: its quantitative and qualitative importance. Deep-Sea Res. I 48, 1923–1939 (2001). 17. Turner, J. T. Zooplankton fecal pellets, marine snow, phytodetritus and the ocean’s biological carbon pump. Prog. Oceanogr. 130, 205–248 (2015). 18. Beaulieu, S. E. Accumulation and fate of phytodetritus on the sea floor. Oceanogr. Mar. Biol. 40, 171–232 (2002). 19. Alonso-González, I. J. et al. Role of slowly settling particles in the ocean carbon cycle. Geophys. Res. Lett. 37, 2921–2927 (2010). 20. Durkin, C. A., Estapa, M. L. & Buesseler, K. O. Observations of carbon export by small sinking particles in the upper mesopelagic. Mar. Chem. 175, 72–81 (2015). 21. Dilling, L. & Alldredge, A. L. Fragmentation of marine snow by swimming macrozooplankton: a new process impacting carbon cycling in the sea. Deep-Sea Res. I 47, 1227–1245 (2000). 22. Iversen, M. H. & Poulsen, L. K. Coprorhexy, coprophagy, and coprochaly in the copepods Calanus helgolandicus, Pseudocalanus elongatus and Oithona similis. Mar. Ecol. Prog. Ser. 350, 79–89 (2007). 23. Cavan, E. L., Trimmer, M., Shelley, F. & Sanders, R. Remineralization of particulate organic carbon in an ocean oxygen minimum zone. Nat. Commun. 8, 14847 (2017). 24. Pérez, V., Fernández, E., Marañón, E., Serret, P. & García-Soto, C. Seasonal and interannual variability of chlorophyll a and primary production in the equatorial Atlantic: in situ and remote sensing observations. J. Plankton Res. 27, 189–197 (2005). 25. Pennington, J. T. et al. Primary production in the eastern tropical Pacific: a review. Prog. Oceanogr. 69, 285–317 (2006). 26. Subramaniam, A., Mahaffey, C., Johns, W. & Mahowald, N. Equatorial upwelling enhances nitrogen fixation in the Atlantic Ocean. Geophys. Res. Lett. 40, 1766–1771 (2013). 27. Moum, J. N., Perlin, A., Nash, J. D. & McPhaden, M. J. Seasonal sea surface cooling in the equatorial Pacific cold tongue controlled by ocean mixing. Nature 500, 64–67 (2013). 28. Walsh, I. D. et al. Particle dynamics as controlled by the flow field of the eastern equatorial Pacific. Deep-Sea Res. II 44, 2025–2047 (1997). 29. Gorsky, G., Le Borgne, R., Picheral, M. & Stemmann, L. Marine snow latitudinal distribution in the equatorial Pacific along 180◦ . J. Geophys. Res. 108, 8146 (2003). 30. Bochdansky, A. B., van Aken, H. M. & Herndl, G. J. Role of macroscopic particles in deep-sea oxygen consumption. Proc. Natl Acad. Sci. USA 107, 8287–8291 (2010). 6

31. Firing, E., Wijffels, S. E. & Hacker, P. Equatorial subthermocline currents across the Pacific. J. Geophys. Res. 103, 21413–21423 (1998). 32. Gouriou, Y. et al. Deep circulation in the Equatorial Atlantic Ocean. Geophys. Res. Lett. 28, 819–822 (2001). 33. Perez, R. C. et al. Mean meridional currents in the central and eastern equatorial Atlantic. Clim. Dynam. 43, 2943–2962 (2014). 34. Perez, R. C. & Kessler, W. S. The three-dimensional structure of tropical cells in the central equatorial Pacific Ocean. J. Phys. Oceanogr. 39, 27–49 (2009). 35. Picheral, M. et al. The Underwater Vision Profiler 5: an advanced instrument for high spatial resolution studies of particle size spectra and zooplankton. Limnol. Oceanogr.-Meth. 8, 462–473 (2010). 36. Kriest, I. Different parameterizations of marine snow in a 1D-model and their influence on representation of marine snow, nitrogen budget and sedimentation. Deep-Sea Res. I 49, 2133–2162 (2002). 37. Bianchi, D., Galbraith, E. D., Carozza, D. A., Mislan, K. A. S. & Stock, C. A. Intensification of open-ocean oxygen depletion by vertically migrating animals. Nat. Geosci. 6, 545–548 (2013). 38. Martin, J. H., Knauer, G. A., Karl, D. M. & Broenkow, W. W. VERTEX: carbon cycling in the northeast Pacific. Deep-Sea Res. I 34, 267–285 (1987). 39. Schmittner, A., Oschlies, A., Matthews, H. D. & Galbraith, E. D. Future changes in climate, ocean circulation, ecosystems, and biogeochemical cycling simulated for a business-as-usual CO2 emission scenario until year 4000 AD. Glob. Biogeochem. Cycles 22, GB1013 (2008). 40. Aumont, O., Ethe, O., Tagliabue, A., Bopp, L. & Gehlen, M. PISCES-v2: an ocean biogeochemical model for carbon and ecosystem studies. Geosci. Model Dev. 8, 2465–2513 (2015). 41. Yool, A., Popova, E. E. & Anderson, T. R. MEDUSA-2.0: an intermediate complexity biogeochemical model of the marine carbon cycle for climate change and ocean acidification. Geosci. Model Dev. 6, 1767–1811 (2013). 42. Maier-Reimer, E. Geochemical cycles in an ocean general circulatiol model. Preindustrial tracer distributions. Glob. Biogeochem. Cycles 7, 645–677 (1993). 43. Kriest, I. & Oschlies, A. On the treatment of particulate organic matter sinking in large-scale models of marine biogeochemical cycles. Biogeosciences 5, 55–72 (2008). 44. Iversen, M. H. & Ploug, H. Temperature effects on carbon-specific respiration rate and sinking velocity of diatom aggregates—potential implications for deep ocean export processes. Biogeosciences 10, 4073–4085 (2013). 45. Aumont, O. et al. Variable reactivity of particulate organic matter in a global ocean biogeochemical model. Biogeosciences 14, 2321–2341 (2017). 46. McDonnell, A. M., Boyd, P. W. & Buesseler, K. O. Effects of sinking velocities and microbial respiration rates on the attenuation of particulate carbon fluxes through the mesopelagic zone. Glob. Biogeochem. Cycles 29, 175–193 (2015). 47. Biddanda, B. A. & Pomeroy, L. R. Microbial aggregation and degradation of phytoplankton-derived detritus in seawater. I Microbial succession. Mar. Ecol. Prog. Ser. 42, 79–88 (1988). 48. Greatbatch, R., Brandt, P., Claus, M., Didwischus, S. & Fu, Y. On the width of the equatorial deep jets. J. Phys. Oceanogr. 42, 1729–1740 (2012). 49. Ménesguen, C., Hua, B. L., Fruman, B. & Schopp, R. Dynamics of the combined extra-equatorial and equatorial deep jets in the Atlantic. J. Mar. Res. 67, 323–346 (2009).

Acknowledgements This study was supported by the German Science Foundation through the Collaborative Research Center 754 ‘Climate-Biogeochemistry Interactions in the Tropical Ocean’ and by the German Federal Ministry of Education and Research through the cooperative project ‘RACE’. The enthusiastic and continued support of UVP5 operations and maintenance by Jerome Coindat and Sylvain Fevre (Hydroptic) is gratefully acknowledged. For the Tara Oceans expedition we thank the commitment of the CNRS (in particular Groupement de Recherche GDR3280), European Molecular Biology Laboratory (EMBL), Genoscope/CEA, VIB, Stazione Zoologica Anton Dohrn, UNIMIB, Fund for Scientific Research—Flanders, Rega Institute, KU Leuven, The French Ministry of Research and the French Government ‘Investissements d’Avenir’ programmes OCEANOMICS (ANR-11-BTBR-0008). We are also grateful for the support and commitment of Agnès b. and Etienne Bourgois, the Veolia Environment Foundation, Région Bretagne, Lorient Agglomération, World Courier, Illumina, the EDF Foundation, FRB and the Prince Albert II de Monaco Foundation. This article is contribution number 59 from Tara Oceans. The contributions of M.P. and L.S. were supported by the Chair VISION of the CNRS and UPMC. A.M. acknowledges support from the National Science Foundation, the US Global Ocean Carbon and Repeat Hydrography Program, and the University of Alaska Fairbanks. US GO_SHIP funding through NSF OCE-1437015 is gratefully acknowledged for the acquisition of the P16N L-ADCP data. S.C. and F.M. acknowledge support by the French national programme LEFE/INSU (ZEBRE) and IRD, which also supported the CASSIOPEE cruise. The backward tracking experiments and the underlying ocean model integrations were performed at the North-German Supercomputing Alliance (HLRN) and the computing centre at Kiel University. We would like to thank captains and crews of RV Meteor, RV Maria S. Merian, RV Ron Brown, RV L’Atalante and the Tara schooner for their support, P. Vandromme for

NATURE GEOSCIENCE | ADVANCE ONLINE PUBLICATION | www.nature.com/naturegeoscience © 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

NATURE GEOSCIENCE DOI: 10.1038/NGEO3042 conducting UVP5 deployments during RV Meteor cruise M119, DT-INSU and US-IMAGO for the acquisition and processing of CTD, L-ADCP and UVP5 measurements and G. Eldin for the shipboard ADCP data from the CASSIOPEE cruise. Furthermore, we would like to thank G. Krahmann for processing of L-ADCP and CTD data from RV Meteor cruises M106 and M119 and RV Maria S Merian cruise MSM22, as well as H. Mehrtens for help with data management. F. Melzner and C. Bowler provided very valuable comments on the article.

Author contributions R.K. and L.S. led the project and designed the study. A.M.P.M., H.H., M.P., L.S., I.K. and R.K. processed and analysed UVP5 data. F.U.S. and A.B. conducted particle backtracking simulations. P.B., R.H., A.M.T., S.C. and F.M. provided CTD and ADCP data. R.K.

ARTICLES compiled all data and led the drafting of the manuscript. All authors contributed to the interpretation of the results and provided substantial input to the manuscript.

Additional information Supplementary information is available in the online version of the paper. Reprints and permissions information is available online at www.nature.com/reprints. Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Correspondence and requests for materials should be addressed to R.K.

Competing financial interests The authors declare no competing financial interests.

NATURE GEOSCIENCE | ADVANCE ONLINE PUBLICATION | www.nature.com/naturegeoscience © 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

7

NATURE GEOSCIENCE DOI: 10.1038/NGEO3042

ARTICLES Methods Satellite-derived sea surface temperature and primary production estimates were retrieved from http://www.science.oregonstate.edu/ocean.productivity (MODIS based Eppley-VGPM1 ). High-resolution full-depth particle size spectra (0.14 mm to 26.8 mm equivalent spherical diameter, ESD) were obtained with an Underwater Vision Profiler 5 (UVP5; ref. 30). The UVP5 was mounted on a Rosette equipped with a Seabird SBE 11plus CTD, an upward- and a downward-looking ADCP. Particle data shown in Fig. 1 to Fig. 3 was averaged in 20-m-depth bins and regridded on a 600 × 150 (latitude × depth) grid using the function grid data from the Python module scipy. A 38 kHz and/or a 75 kHz shipboard ADCP were continuously run during transect occupations. To estimate diel vertical migration depth and intensity, ADCP echo intensity was transformed into volumetric backscatter Sv (dB) (ref. 50) and 4.8 h (0.2 days) of daytime or night-time data from each day during which the ship operated between 1◦ S and 1◦ N were selected. To statistically test for differences in ADCP volumetric backscatter, daytime and night-time data were averaged in 10 time intervals of 28.8 min lengths and the daytime values were subtracted from the night-time values to obtain the difference in volumetric backscatter that indicates differences in biomass distribution. A one-sided Student’s t-test was conducted to identify differences of the mean of the ten resulting values from zero (p-value < 0.01) and the mean was calculated to visualize the differences in biomass distribution during day- and night-time. The organism or particle size that contributes to the acoustic backscatter depends on the acoustic frequency or wavelength (λ). The minimum size of particles that influences the sound scattering can be estimated51 to be (0.25–0.5) × λ. With λ ≈ 20 mm for a 75 kHz instrument and λ ≈ 39.5 mm for a 38 kHz instrument, scatterers must be larger than 5–10 mm, and 10–20 mm, respectively. Therefore, large copepods, euphausiids and small pelagic fishes contribute most to the ADCP backscatter signal. The simulation of the relative particle distribution in backward tracking experiments was conducted with a Lagrangian model52 using output from the high-resolution ocean general circulation model TRATL01 (refs 53,54). A few hundred thousand particles were seeded across the equatorial Atlantic at 23◦ W at 300 to 500 m (IPM depth) or at 3,000 m depth. Particles with sinking speeds of 10 or 50 m d−1 were advected backwards using 5-daily varying three-dimensional velocities from the ocean general circulation model. With 1/10◦ horizontal resolution, 46 vertical levels and a realistic atmospheric forcing, TRATL01 realistically simulates the spatio-temporal structure of the equatorial current system54 , however, with still to weak intermediate and deep circulation along the equator53 . POC flux was calculated assuming that particle mass and sinking speed can be calculated using empirically derived relationships for marine aggregates

(see ref. 36, reference 2a of Table 1 and reference 9 of Table 2 for mass and sinking speed of a particle, respectively). Assuming a carbon:nitrogen ratio of 106:16, this yields an expression for the sinking flux of a size class characterized by its diameter ESD (in cm) of 2.8649 ESD2.24 (mgC m d−1 ). Multiplying with the particle number in this size class (in particles m−3 ), and integrating over all size classes between 0.014–1.688 cm, we obtain the total POC flux (mgC m−2 d−1 ) for this size range. Flux calculations based on optical measurements have limitations as they are based on assumptions of the sinking speed and carbon content of the particles observed. The parameterization used was derived from in situ measurements of particulate matter sinking speeds and carbon content, and has been shown to best reproduce profiles of marine snow and particulate organic matter at the same time36 . Data availability. All UVP5 data, L-ADCP data from all cruises except the 2015 occupation of the P16N section by RV Ron Brown, as well as shipboard ADCP volumetric backscatter data that support the findings of this study, are available at https://doi.pangaea.de/10.1594/PANGAEA.874873. L-ADCP data from RV Ron Brown cruise is available at https://currents.soest.hawaii.edu/ go-ship/ladcp/2015_P16N.html. Net primary productivity data used in Fig. 1 were obtained from http://orca.science.oregonstate.edu/ 1080.by.2160.monthly.hdf.eppley.m.chl.m.sst.php.

References 50. Mullison. Backscatter estimation using broadband Acoustic Doppler Current Profilers – Updated (2017); http://go.nature.com/2fsJv1p 51. White, M., Mohn, C. & Kiriakoulakis, K. in Environmental Sampling, in Biological Sampling in the Deep Sea (eds Clark, R., Consalvey, M. & Rowden, A. A.) 57–79 (John Wiley, 2016). 52. Blanke, B., Arhan, M., Madec, G. & Roche, S. Warm water paths in the Equatorial Atlantic as diagnosed with a general circulation model. J. Phys. Oceanogr. 29, 2753–2768 (1999). 53. Duteil, O., Schwarzkopf, F. U., Böning, C. W. & Oschlies, A. Major role of the equatorial current system in setting oxygen levels in the eastern tropical Atlantic Ocean: A high-resolution model study. Geophys. Res. Lett. 41, 2033–2040 (2014). 54. Schwarzkopf, F. U. Ventilation Pathways in the Tropical Atlantic and Pacific Oceans with a Focus on the Oxygen Minimum Zones: Development and Application of a Nested High-Resolution Global Model System PhD thesis, Christian-Albrechts-University Kiel (2016).

NATURE GEOSCIENCE | www.nature.com/naturegeoscience © 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.