Si budgets for the South Pacific
In the following section, values from previous studies are compared (Table 4)
with the results obtained across this under-studied region of the Pacific
Ocean, which is characterized by the most oligotrophic and Chl a-depleted
waters worldwide . We obtained size-fractionated biomass and
export fluxes during the OUTPACE program, and size-fractionated production
and biomass budgets during the BIOSOPE program. Regarding values obtained at
both ends of the BIOSOPE transects, i.e., in the Peru–Chile upwelling system
and in the HNLC system surrounding the Marquesas Islands, ΣρSi
rates compare well with previous studies from other similar regions (Table 4). Integrated Si production rates at the UPW stations are in the middle
range (42–52 mmol Si m-2 d-1) of what was previously found in
coastal upwellings. Values are, however, almost double what was previously
observed in the Peru upwelling by , although they are lower
than those in the Monterey Bay and Baja Californian upwelling systems
. For oceanic HNLC areas, values obtained
(0.8 to 5.6 mmol Si m-2 d-1) cover the range of rates measured
in HNLC to mesotrophic systems of the North Atlantic, central equatorial
Pacific and Mediterranean Sea. However, integrated rates obtained for the
oligotrophic area of the southeastern Pacific gyre are to our knowledge
among the lowest ever measured, even taking into account the error associated
with budget estimates this close to analytical detection limits. Values range
from 0.04 to 0.20 mmol Si m-2 d-1; they are thus lower than
average values previously measured at BATS and ALOHA stations (0.42 and 0.19 mmol Si m-2 d-1 respectively)
. However, they are similar to measurements
performed in autumn (0.04–0.08 mmol Si m-2 d-1) in a severely
Si-limited regime of the North Atlantic . Previous
studies have documented limitation of diatom Si production by Si
, but more recently evidence of co-limitation by both Si
and Fe was found in the central equatorial Pacific .
This would more than likely be the scenario for the SPG, given the very low
silicic acid (Figs. 2 and 3) and Fe concentrations (0.1 nM and ferricline below
350 m depth, ) measured during both cruises.
Summary of ΣBSi stocks (in mmol Si m-2) for the
OUTPACE and BIOSOPE and other oceanic and oligotrophic systems.
Region
Average integrated Si biomass
References
ΣρSi (mmol m-2)
Coastal upwellings
BIOSOPE: Peru–Chile upwelling
65.7±53.8
This study
Southern California Current coastal waters
53.2±39.3
Krause et al. (2015)
Oceanic area
Southern California Current oceanic waters
1.6±0.3
Krause et al. (2015)
BIOSOPE: southeastern Pacific (HNLC)
11.9±10.9
This study
Oligotrophic area
Mediterranean Sea (BOUM)
1.1–28.2
Crombet et al. (2011)
Sargasso Sea (BATS)
4.0±6.8
Nelson et al. (1995)
Sargasso Sea
0.9–6.1
Krause et al. (2017)
North Pacific subtropical gyre
1.6–12.8
Krause et al. (2013)
North Pacific subtropical gyre (ALOHA)
3.0±1.1
Brzezinski et al. (2011)
Central North Pacific
7.1±3.0
Brzezinski et al. (1998)
Eastern equatorial Pacific
3.8–18.0
Krause et al. (2011)
BIOSOPE: southeastern Pacific gyre
1.1±1.1
This study
OUTPACE: southwestern Pacific gyre
1.0±0.2
This study
OUTPACE: Melanesian Archipelago
2.4±1.0
This study
Summary of Si export fluxes in sediment traps at various depths (in
µmol Si m-2 d-1) for the OUTPACE cruise
compared to other studies.
Region
Sediment
Average
References
trap (m)
Si export fluxes
(µmol m-2 d-1)
Coastal upwellings
Southern California Current coastal waters
100
8000±5760
Krause et al. (2015)
Oceanic area
North Atlantic (NABE)
400
10–145
Honjo and Manganini (1993)
North Atlantic (POMME)
400
2–316
Mosseri et al. (2005); Leblanc et al. (2005b)
North Pacific subtropical gyre (ALOHA)
150
14–300
Brzezinski et al. (2011)
Oligotrophic area
Sargasso Sea (BATS)
150
17–700
Nelson et al. (1995)
Sargasso Sea (BATS)
150
130
Brzezinski and Nelson (1995)
200
113
300
85
OUTPACE: southwestern Pacific gyre
153
1.8
This study
328
0.5
OUTPACE: Melanesian Archipelago
153
1.6
This study
328
1.6
519
2.5
The approximate surface area of midocean gyres was estimated to be 1.3×108 km2 (representing approximately one-third of the global ocean), yielding a
global contribution of only 26 Tmol Si yr-1 gross silica production,
i.e., approximately 9–13 % of the budget calculated for the global ocean of
240 Tmol Si yr-1 according to Nelson et al. (1995). This budget has
been recently revised down to 13 Tmol Si yr-1 when considering budgets
from the North Pacific reducing the contribution of
subtropical gyres to 5 %–7 % of global marine silica production
. However, the range provided in
in the calculation of their global Si production fluxes for
midocean gyres was of 0.2–1.6 mmol m-2 d-1. Our values
would, once again, lower the contribution of these vast oceanic regions to
global Si production, although the present data are only based on two
production station measurements and warrant further measurements for this
region. Nevertheless, it can be expected that the most ultra-oligotrophic
region of the world ocean would contribute even less to total Si production
than the other oligotrophic systems listed in Table 4 and that, in particular,
the Si production in the ultra-oligotrophic southern tropical gyre would be
lower than the northern tropical gyre.
Integrated Si biomass also reflects the very low contribution of diatoms in
this system, which was more than 2 times lower in the South Pacific gyre
than in the Melanesian Archipelago (Table 5). In the SPG, the lowest Si
stocks were measured (∼1 mmol Si m-2) and were similar to
lower-end values found in the ultra-oligotrophic eastern Mediterranean Basin
in autumn and in other oligotrophic areas of the North Pacific subtropical
gyre and of the Sargasso Sea (Table 5 and references therein). It is probable
that ΣρSi production and BSi stocks could have been slightly
higher less than a month earlier in the season in the western part of the
OUTPACE transect in the MA. Indeed, the satellite-based temporal evolution of
Chl a at stations LD-A and LD-B showed decreasing concentrations at the time
of sampling , while the situation did not show any
temporal evolution for the SPG, thus suggesting that the biogenic silica
budget for this area is quite conservative under a close to steady-state
situation.
Lastly, our Si export flux measurements by drifting sediment traps are the
lowest ever measured and are about 2 orders of magnitude lower than those
from other oligotrophic sites such as BATS in the Atlantic or ALOHA in the
Pacific Ocean (Table 6). They represent a strongly negligible fraction of
surface Si stocks, implying no sedimentation at the time of sampling, and
that active recycling and grazing occurred in the surface layer. Indeed,
surface temperatures higher than 29 ∘C at all long-duration sites may favor
intense dissolution in the upper layer, while active zooplankton grazing was
also documented, removing between 3 % and 21 % of phytoplankton stocks daily
. The virtual absence of silica export from the surface
layer agrees well with the conclusion of that no siliceous
sediment is accumulating beneath the central ocean gyres.
Siliceous plankton community structure in the tropical South Pacific
The main feature observed during OUTPACE was a bi-modal distribution of
diatom communities, either at the surface and/or at the DCM level depending
on stations, which deepened towards the east, following the increasing
oligotrophy gradient, similarly to what was previously described in the
Mediterranean Sea (Crombet et al., 2011). A similar feature, showing a
particularly deep DCM, up to 190 m in the SPG at 1.2 times the euphotic depth
, was observed during BIOSOPE, revealing a known strategy for
autotrophic plankton cells in nutrient-depleted waters to stay at the depth
where the best light vs. nutrient ratio is obtained .
While DCMs are common in midocean gyres and are known to be often
dominated by pico-sized phytoplankton , studies documenting
phytoplankton community structure in the tropical South Pacific Ocean, an
area formerly called a “biological desert”, are still very scarce. In the
review of planktonic diatom distribution by , which
references biocenoses for all main oceanic water bodies and for which thousands of
articles were processed, the diatom composition for the southern tropical region
was referred to as “No species given (flora too poor)”. Since then only a
few studies mentioning phytoplankton community structure, mostly located
along the equator were published, such as , , , and . In some
biogeographical distribution of phytoplankton, including diatoms, is given for
the entire Pacific region, yet the southern tropical region is limited to
more historical Russian data and relies on very few stations. The only diatom
distribution for the southern tropical gyre was published for the present data
set by in the BIOSOPE special issue. Hence the present data
contribute to documenting a severely understudied yet vast area of the
world ocean. The oceanic regions covered during both cruises may be clustered
into three main ecological systems with relatively similar diatom community
structures: the nutrient-rich coastal upwelling system near the Peru–Chile
coast, where diatom concentrations exceeded 100 000 cells L-1; the
Fe-fertilized areas of the Melanesian Archipelago and west of the Marquesas
Islands, where concentrations could locally exceed 10 000 cells L-1;
and all the other ultra-oligotrophic regions (mainly the South Pacific gyre
system) characterized by extremely low diatom abundances, usually <200 cells L-1.
The upwelling area was characterized by a distinct community, not
found in the other regions, composed of typical neritic and centric colonial
species such as Skeletonema sp., Bacteriastrum spp.,
Chaetoceros compressus, Thalassiosira subtilis and
T. anguste-lineata. These first three species were already
documented as abundant in the Chile upwelling by , whereas
T. anguste-lineata was reported along the Chilean coast from 20
to 36∘ S and was also documented in the upwelling system
west of the Galapagos Islands . The highest ρSi
production values were measured at the offshore UPW station where
Bacteriastrum spp. and Chaetoceros compressus co-occurred as the two dominant
species, whereas ρSi rates were halved at the closest coastal station UPX, associated with lower abundances of diatoms, with co-occurring dominance
by Skeletonema sp. and Thalassiosira anguste-lineata. The HNLC regions off
the Marquesas Islands and in the eastern gyre (stations 14–20, BIOSOPE)
and the oligotrophic region (N-deprived but Fe-fertilized region of the MA),
with bloom situations at stations 5 and LD-B (OUTPACE), showed strong
similarities in terms of diatom community structure and were all mainly
dominated by the medium-sized pennate diatoms of the Pseudo-nitzschia delicatissima and subpacifica species complex. These pennate species
are commonly reported for the central and equatorial Pacific Ocean
. During BIOSOPE,
Pseudo-nitzschia delicatissima were often seen forming “needle
balls” of ∼100 µm diameter, which suggests an antigrazing
strategy from micrograzers , a strategy already described
by several authors . Predominance of
pennate diatoms over centrics has previously been observed in the N-depleted
environment of the equatorial Pacific (Blain et al., 1997; Kobayashi and
Takahashi, 2002) and could correspond to an ecological response to
diffusion-limited uptake rates, favoring elongated shapes, as suggested by
. Furthermore, net samples from the OUTPACE cruise showed
a numerically dominant contribution of Cylindrotheca closterium over
0–150 m at most stations of the MA (Appendix C1), with a strong dominance at
LD-B, even though their contribution to biomass is minor given their small
size. However, it should be noted that if small fast-growing pennates were
numerically dominant, their relative contribution to C biomass was very small
compared to that of few larger centrics such as Pseudosolenia calcar-avis, which when present dominated in terms of biomass, similarly to
what had already been observed in the South Pacific with large
Rhizosolenia . Pseudo-nitzschia sp. and Cylindrotheca closterium have been shown to bloom upon Fe-addition experiments
and may reflect the
significantly higher dissolved Fe concentrations measured in the MA (average
1.9 nM in the first 100 m) compared to the SPG (0.3 nM) . In
the equatorial Pacific, Fe-amendment experiments evidenced the rapid growth
of Cylindrotheca closterium, with a high doubling rate close to 3 d-1 , which can explain why this species is often
numerically dominant.
Fast-growing colonial centric diatoms such as Chaetoceros spp. were
notably absent from the MA, except at stations 5 and LD-B, where mesoscale
circulation increased fertilization and allowed a
moderate growth (observed in both Niskin samples and net hauls), resulting in
an increased contribution of diatoms to total C biomass of approximately 10 % (Fig. 9c). Other typical bloom species such as Thalassiosira spp.
were completely absent from the species from the Niskin samples but observed
at low abundance in some net haul samples. Nonetheless, very large centrics
typical of oligotrophic waters such as Rhizsolenia calcar-avis
were present in low numbers at all stations and in all
net hauls, and they represented a non-negligible contribution to biomass despite
their low abundance.
One difference with the N-replete Marquesas HNLC system was that the
hydrological conditions of the MA were highly favorable for the growth of
diazotrophs, with warm waters (>29 ∘C) and depleted N in the surface layer
associated with high Fe levels, while P was likely the ultimate controlling
factor of N input by N2 fixation in this region . N2-fixation rates were among the highest ever measured in
the open ocean during OUTPACE in this region , and the
development of a mixed community composed of filamenteous cyanobacteria such
as Trichodesmium spp. and other spiraled-shaped species, unicellular
diazotrophs such as UCYN, Crocosphaera watsonii, and
DDAs was observed (Appendix C1). The highest
rates were measured at the surface at stations 1, 5, 6 and LD-B
and the major contributor to N2 fixation in MA
waters was by far Trichodesmium . In the Niskin
cell counts, diatoms known to live in association with the diazotroph
Richelia intracellularis (such as Hemiaulus hauckii,
Chaetoceros compressus and several species of Rhizosolenia
such as R. styliformis, R. bergonii, R. imbricata
and the centric Climacodium frauenfeldianum known to harbor a genus
related to Cyanothece sp.; ) were all found in
low abundance in the water sample cell counts, contributing to less than 1 %
of total diatoms. Exceptions were observed at sites 1 and 2, where their
contributions increased to 2.3 % and 8 % respectively. The low contribution of
DDAs to the diazotrophs community was confirmed by direct cell counts and
nifH gene sequencing . Notably, the presence of
Richelia intracellularis was not observed in the Niskin lugol-fixed
water samples, but Rhizosolenia styliformis with Richelia,
and some isolated Richelia cells were observed abundantly in net
hauls. The latter were found to be dominant at stations 1 and LD-B, where the
highest fixation rates were measured. Richelia cells, alone or in association with
R. styliformis, were much less abundant in the South Pacific gyre, where Fe is
prone to be the limiting nutrient for N2-fixation rates despite higher
P availability, pointing to less favorable growth conditions for diazotrophs.
Yet, the overall dominance of Trichodesmium, Crocosphaera
and other filamenteous cyanobacteria (Appendix C1) in the net samples reveals
that DDAs were very minor contributors to N2 fixation during OUTPACE.
This was also evidenced through NanoSIMS analyses .
In order to explain the growth of diatoms in this severely N-depleted region,
one can quote the use of diazotroph-derived nitrogen (DDN), i.e., the
secondary release of N2 fixed by diazotrophs, which was shown to be
efficiently channeled through the entire plankton community during the VAHINE
mesocosm experiment . In this latter study off the shore of New
Caledonia, Cylindrotheca closterium grew extensively after
stimulation with diazotrophy after P addition in large-volume in situ mesocosms
in New Caledonia . As previous studies had already
observed a co-occurrence of elevated C. closterium with several
diazotrophs , this recurrent association tends
to confirm our previous hypothesis of a likely efficient use of DDN released
as NH4 by this fast-growing species . This could be
another factor, besides Fe availability, explaining its success. A similar
hypothesis may be invoked for the presence of Mastogloia woodiana, a pennate
diatom known to be occasionally dominant in the North Pacific subtropical
gyre blooms . It is also a characteristic
species of oligotrophic areas , often observed in
association with other DDAs, which could similarly benefit from secondary
N release .
Lastly, the ultra-oligotrophic region of the SPG investigated during both
OUTPACE and BIOSOPE revealed a common baseline contribution of diatoms with
less than 200 cells L-1 at the DCM and close to zero at the surface.
In addition, a dominance of small and large pennate species was observed,
such as Nitzschia bicapitata, Pseudo-nitzschia delicatissima, Thalassiothrix longissima, Thalassionema elegans and Pseudoeunotia sp., which have already been documented
for the equatorial Pacific by . Occasional occurrences of
some emblematic species of oligotrophic regions were also observed, such as
Chaetoceros dadayi, C. peruvianus, C. tetrastichon
or Planktoniella sol. It can be noted that radiolarians were also
more abundant and more diverse in the ultra-oligotrophic SPG during OUTPACE
than in the MA, while unfortunately no information regarding radiolarians is
available for the BIOSOPE cruise.
Potential siliceous organisms in the picoplanktonic (<2–3 µm)
size fraction. (a) Siliceous scale-bearing Parmale (Tetraparma pelagica in SEM,
photo courtesy of J. Young), (b) centric diatom (Minidiscus trioculatus),
(c) Synechoccocus cell showing Si assimilation in red (28Si) in NanoSIMS
(photo courtesy of Mathieu Caffin).
Evidence for active Si uptake in the picoplanktonic size fraction in the tropical South Pacific
The pico-sized fraction (<2–3 µm) represented 11 % of BSi stocks on average
during BIOSOPE, and 26 % of BSi stocks during OUTPACE (Fig. 6), which is a
non-negligible contribution. The significant contribution of the pico-sized
fraction to the BSi stocks during both cruises could be explained by the
presence of detrital components; however, its contribution to Si(OH)4
uptake during BIOSOPE was really surprising but can be explained in the light
of new findings. Indeed, recent studies have evidenced that the
picophytoplanktonic cyanobacteria Synechococcus can assimilate Si
, which could
explain why Si stocks were detected in this size fraction. The first
hypothesis was to consider broken fragments of siliceous cells passing
through the filter or interferences by lithogenic silica, but these
hypotheses were invalidated during BIOSOPE when Si uptake measurements using
32Si were also carried out on this pico-sized fraction and revealed a
non-negligible uptake, mainly in the Chilean upwelling systems (Fig. 7). Our
results are thus in line with previous findings, as no other organisms below
2–3 µm are known to assimilate Si, except some small-sized Parmales, a poorly
described siliceous armored planktonic group which span over the 2–10 µm
size class, such as Tetraparma sp. , or small
nanoplanktonic diatoms such as Minidiscus ,
close to the 2 µm limit (Fig. 11a, b). The latter two species could occur in
the 2–3 µm size fraction, but are very easily missed in light microscopy and
require SEM imaging or molecular work for correct identification. The presence of
Parmales or nanoplanktonic diatoms may explain the measurement of BSi in
this 0.4–3 µm size-class for the OUTPACE cruise, but can be excluded as
responsible for the Si uptake measured during BIOSOPE on filters below 2 µm.
Rather, during OUTPACE, NanoSIMS imaging revealed that cytometrically sorted
Synechococcus cells accumulated Si (Fig. 11c), confirming their
potential role in the Si cycle in the southern tropical gyre.
According to , the Si content of Synechococcus, in
some cases, could exceed that of diatoms, but these authors suggested that
they might exert a larger control on the Si cycle in nutrient-poor waters
where these organisms are dominant. In the present study, the largest
contribution of the pico-sized fraction to absolute ΣρSi uptake
rates occurred at both ends of the transect in the Peru–Chile upwelling
region and at the MAR station (Table 1), locations which also corresponded to
the highest concentrations of Synechococcus observed
. However, compared to diatoms, this only represented 1 % to 5 % of total ΣρSi
uptake, which is not likely to drive the
Si drawdown in this environment. This low relative contribution to ΣρSi was similarly found at the other end of the transect at HNL and MAR
station, but where absolute uptake rates were moderate. The largest
contribution of the pico-sized fraction was measured in the SPG (GYR and EGY
sites), where despite very low ρSi values, the relative ΣρSi
uptake between 0.2 and 2 µm reached 16 % to 25 %. Station GYR as well as
stations 13 to 15 are areas that are highly depleted in orthosilicic acid,
with concentrations <1 µM from the surface to as deep as 240 m. Hence, it is
probable that Synechococcus could play a role in depleting the Si of
surface waters in this area, which are devoid of diatoms.
During the OUTPACE cruise, there were no clear correlations between
Synechococcus distributions and the measured 0.4–3 µm BSi
concentrations. This could be explained by the extremely wide range of
individual cellular Si quotas estimated to vary between 1 and 4700 amol Si cell-1 (with an average value of 43) from cells collected in the northwestern Atlantic , where Synechococcus
contributed up to 23.5 % of ΣBSi . In the latter
study, a first-order estimate of the contribution of Synechococcus
to the global annual Si production flux amounted to 0.7 %–3.5 %, which is
certainly low, but comparable to some other important input or output fluxes
of Si . Using the range of measured Si cellular content
per Synechococcus cells given in of 14 to 64 amol Si cell-1 and Synechococcus abundance data from the same
casts obtained in flow cytometry (data courtesy of S. Duhamel, Lamont
Doherty, NY), this yields a potential contribution of 3 % to 14 % of
Synechococcus to the small BSi fraction, which is close to the
previous estimates.