Eutrophication -- 14.3 The case of the silicate pump in the Bay of Brest

14.4 Ecological and biogeochemical implications

14.2 Benthic-pelagic coupling and eutrophication

14.3 The case of the silicate pump in the Bay of Brest

14.3 The case of the silicate pump in the Bay of Brest

14.3.1 Si and coastal food webs

Evidence is growing that the nutrient silicic acid (DSi) is playing a major role in the functioning of coastal ecosystems in many regions of the world [104]. The reason is linked to the importance of diatoms in marine food webs [109], and to anthropogenic influences on watersheds and rivers. When DSi is missing, diatoms become replaced by other phytoplankton groups that do not have any requirement for this nutrient, such as dinoflagellates [347]. A wide variety of coastal ecosystems has been documented, where increasing frequency and magnitude of harmful algal blooms have been associated to decreasing Si:N and Si:P ratios, with important consequences for pelagic and benthic food webs [430]. Decreasing Si:P and Si:N ratios first find their origin in eutrophication. Urbanization, agricultural and industrial activities have led to large increases in the delivery of N and P along the land-ocean continuum. On a global basis, the fluxes of these elements to the oceans have increased by a factor two; at the same time, in rivers unaffected by human activities, DSi fluxes have remained constant, as the major source of DSi to rivers comes from natural silicate rock weathering [311]. The second source of decreasing Si:N and Si:P ratios is river manipulation, especially the build up of dams [236]. In the reservoirs behind the dams, growth and sedimentation of diatoms remove biogenic silica (BSiO₂) from the water column, leading to decreased DSi concentrations [104]. Whatever the type of perturbation, decreasing Si:N and Si:P ratios in rivers imply potential DSi limitation for diatoms [137], which becomes true limitation when DSi concentrations decrease below the half saturation constants in the receiving coastal water bodies (e.g. [337]).

14.3.2 The Bay of Brest example

The Bay of Brest is an ecosystem where Si:N ratios in riverine inputs have decreased by a factor of 3 in the past 30 years, mostly due to excessive N inputs from agricultural practices [283]. Indirect evidence of DSi limitation has been provided, on the basis of declines in diatom populations coinciding with DSi concentrations becoming lower than 1 µM by early spring [385]. DSi limitation has then been directly demonstrated from kinetic uptake experiments using the ³²Si radioactive isotope [126]; see also Figure 2. Despite DSi limitation during spring, diatoms typically continue to dominate the phytoplankton during the entire productive period [127]. Several factors have been hypothesized to account for this (apparent lack of) response of the Bay to excessive N inputs. They include the export out of the bay of most of this N during winter [283], the well-mixed nature of the water column which does not favor the development of flagellates [386], and the intensity of Si recycling both in the water column and at the sediment-water interface [385][127].

Although Si recycling has recently been shown to be accelerated under high bacterial activity [44], it remains slower than the recycling of N and P, which are biologically mediated [347]. In the open ocean, this differential recycling rate is at the basis of the so-called silicate pump [140], which removes DSi from surface waters for a long time period. In coastal waters, especially in the semi-enclosed Bay of Brest, the effects of the silicate pump may well be reversed, because of the tight temporal and spatial coupling between sediment and surface waters: following the first diatom blooms and the sedimentation of diatom cells, Si can be retained within the Bay, at the sediment-water interface, instead of being exported to the adjacent coastal ocean; it then becomes directly available for regenerated diatom production, because of the shallow depths of the well-mixed waters [127].

14.3.3 The working hypothesis

Synthesizing 20 years of studies of the Bay of Brest ecosystem, both from a pelagic and benthic point of view, Chauvaud et al. [84] suggested that the functioning of this coastal silicate pump is under the control of benthic suspension-feeders. Suspension feeders dominate the benthic megafauna in the Bay of Brest [467]. Introduced in 1950, the gastropod Crepidula fornicata is now the main benthic suspension feeder in the Bay [82]. Chauvaud et al. [84] have suggested that increased suspension-feeding activity during early spring (filtration and subsequent production of enormous quantities of biodeposits) could lead to an increase in the temporary retention of BSiO₂ in the sediments of the bay, thereby limiting the export of Si out of the bay. Subsequent BSiO₂ dissolution during late spring and summer, enhanced by increasing temperature and bacterial activity, would provide the necessary DSi required by diatoms to maintain their dominance throughout the productive period. It is essential to note that the enormous amount of biodeposits produced by C. fornicata has no equivalent in the ecosystem.

14.3.4 Testing the working hypothesis

ragFig3

Figure: Physical, chemical and biological parameters measured at the SOMLIT site in the Bay of Brest during the year 2000. (a) temperature and salinity. (b) phosphate and chlorophyll a. (c) silicic acid and nitrate. From Ragueneau et al. [382].

During the year 2000, the hypothesis of a "biologically active silicate pump" was tested. Figure 3 shows the physical, chemical and biomass parameters recorded weekly at the monitoring SOMLIT station located near the bay entrance. These parameters characterize the productive season which begins in late March in the bay with the increase in chlorophyll a corresponding to the decrease in nutrient concentrations. A succession of phytoplankton blooms occurred throughout the spring and summer. By late July, DSi and DIP start to accumulate again in the water column, followed by DIN two months later, when the productive period ends.

Direct evidence of DSi limitation in the Bay of Brest.

Direct evidence of DSi limitation during spring has been obtained through two kinetic experiments performed when the diatoms Rhizosolenia sp. and Chaetoceros sp. where dominating the phytoplankton population (Figure 4). Having similar K_m values close to 1.3 µM, these diatoms were both limited to only 20% of their maximal uptake velocity by late spring/early summer. Note that Chaetoceros sp. exhibited a V_max/K_m ratio ten times higher than that of Rhizosolenia sp., suggesting a higher affinity and thus, a higher ability to take up DSi at low concentrations. Thus, diatoms were clearly limited by ambient DSi concentrations and were dependent upon Si recycling during early summer. By late summer, DSi was accumulating again in the water column (Figure 3), suggesting that DSi inputs exceeded the diatom demand by that time.

ragFig4

Figure: Two kinetic experiments performed on 17 May 2000 when the diatom Rhizosolenia was dominating and on 21 June 2000 when the diatom Chaetoceros was dominating. Experiments have been performed using the ³²Si radioactive isotope [469]. A 3L water sample has been distributed into eight 250 ml polycarbonate incubation bottles. These bottles have been enriched with silicic acid up to 20 µM, spiked with ³²Si and incubated for 24 hours at light saturation. Following liquid scintillation counting [281], the specific uptake rate (V) is plotted against the Si(OH)₄ concentration of the flasks at the beginning of the incubation. These Michaelis-Menten type of curves have been fitted using the non-linear regression method of Wilkinson [497], allowing the determination of the maximal uptake velocity (V_max) and the half saturation constant (K_m). From Ragueneau et al. [382].

Direct effect of C. fornicata on DSi benthic fluxes.

To study the possible effects of benthic suspension feeders on DSi recycling at the sediment-water interface, benthic fluxes were measured seasonally at two contrasting sites, displaying respectively low (ca. 30 ind. m^-2) and high (ca. 1700 ind. m^-2) concentrations of C. fornicata. Sediment cores were retrieved manually at 20 m depth using scuba diving; 3 replicates were taken at each site. Following a time zero sampling at sea, cores were then rapidly (within 1 - 2 hours) incubated in the laboratory at the temperature of the bay waters (from 8°C during winter to 16°C during summer, (Figure 3). The DSi concentration in the water overlying the sediment was monitored every hour during the first 6 hours, and then two to three times between 20 and 24 hours following the core collection. Homogenisation of the overlying water was ensured by pumping water, using a peristaltic pump, 2 - 3 cm above the sediment water interface and redistributing it near the water surface. The flow rate was adjusted so that one water volume was renewed every hour. The slopes of the DSi increase measured in the cores during their incubation show a 20-fold difference, during late summer, depending on whether C. fornicata was abundant or rare (Figure 5). The corresponding DSi fluxes are typical of those encountered in coastal environments [161, and references therein][502, and references therein]. Mean values at these two contrasting sites were measured every two months, throughout the productive period (Figure 6). Two important observations can be made. First, whatever the season considered, fluxes are always higher at the site with C. fornicata compared to the site where C. fornicata is absent. Being only of a factor of two by late spring, the difference between the fluxes measured at the two sites becomes more than one order of magnitude by mid-summer and fall. Secondly, maximum DSi fluxes were measured in late spring at the site without C. fornicata and in late summer at the site with C. fornicata. Thus, both the amplitude of the DSi benthic flux in the presence of C. fornicata and the delay in the timing of the maximum of these fluxes provide strong evidence that the BSiO₂ produced during spring is indeed being retained by the activity of the suspension feeders and then gradually released to the overlying waters, following dissolution. Owing to the well-mixed nature of the water column in this macrotidal ecosystem, DSi is then immediately available for diatoms production [386].

ragFig5

Figure: Sediment core incubation experiments conducted during late summer in the Bay of Brest. Sediment cores were collected at Rozegat using diving, at two sites located within 300 m but exhibiting contrasted densities of the suspension feeder Crepidula fornicata. Filled symbols: site Rozegat with high densities (1243 ind. m^-2) of C. fornicata, [467]. Open symbols: site Rozegat without any C. fornicata. Note the factor of 20 between the mean flux measured at the site with C. fornicata (triplicates, mean: 6.3 mmol Si m^-2 d^-1) and the mean flux measured at the site without C. fornicata (duplicates, mean: 0.3 mmol Si m^-2 d^-1). From Ragueneau et al. [382].

Validation at the bay scale.

Before budgeting Si fluxes within the bay of Brest ecosystem, a simple, albeit important, calculation provides strong support that the working hypothesis has significance at the scale of the whole system. 11×10³ t of dry organic matter are produced by suspension feeders [84]. By analogy with zooplankton fecal pellets in surface waters [457], we can make the hypothesis that these biodeposits are enriched in biogenic silica because Si is not retained by these organisms for their metabolism. Typically, in the open ocean, the Si:C ratio increases by a factor of ca. 5 between food and feces, between production and export in surface waters, or between the deposition at the sediment-water interface and the accumulation below the bioturbated layer [384]. Using this factor and a typical Si:C ratio of 0.04 - 0.09 for the bay phytoplankton [385][126] yields a Si:C ratio in the biodeposits close to 0.2 - 0.5. Thus, the biodeposits can lead to the temporary deposition of 167 - 417 × 10⁶ mol Si. Distributed spatially over the 180 km² of the Bay and temporally over the whole year, this leads to a potential DSi benthic flux of 2 - 6 mmol Si m^-2 d^-1. Interestingly enough, this range is of the same magnitude as the fluxes measured using core incubations (Figure 6).

ragFig6

Figure 6. Synthesis of DSi benthic fluxes measured at the two constrasting sites during the productive period in the Bay of Brest. Black bars: site Rozegat with C. fornicata; white bars: site Rozegat without C. fornicata. These fluxes represent the mean values of the fluxes measured in triplicates (see Figure 3.

Preliminary budgets.

Having provided direct evidence for (i) DSi limitation and (ii) a direct influence of suspension feeders on BSiO₂ retention and DSi availability, one needs to compare the magnitude of the DSi benthic fluxes with both river inputs and the diatom demand (silica production). Nutrient river inputs were measured on a weekly basis in the Aulne and the Elorn rivers rivers, which bring most of the freshwater to the Bay of Brest, by members of the ECOFLUX network [379]. For comparison purposes, benthic DSi fluxes have been extrapolated to the whole bay (180 km²) by applying the fluxes measured at the site with C. fornicata to the area of the bay covered by C. fornicata (90 km²) and the fluxes measured at the site without C. fornicata to the area still unaffected (90 km²). Reasonable estimates of integrated silica production can be obtained on the basis of primary production measurements and the use of appropriate Si:C ratios [372]. Under nutrient replete conditions, diatoms grow with a Si:C ratio close to 0.13 [60]. In the Bay of Brest, this ratio is typically twice as low, due to the coupled influence of DSi limitation and the presence of non-siliceous algae [385][127]. A ratio of 0.06 has therefore been chosen as a reasonable mean of converting C primary production into a BSi production that can then be compared to river and benthic fluxes (Figure 7).

ragFig7

Figure 7: Seasonal budgets of DSi fluxes in the Bay of Brest. All data in mol Si d^-1. White bars: river fluxes. Grey bars: sum of river and benthic fluxes. Black bars: estimates of silica production. See text for explanations on budget calculations. (a) Benthic DSi fluxes have been extrapolated to the whole Bay by applying the fluxes measured at each site (Figure 4) to half of the Bay, i.e. the present extension of the invasive C. fornicata. The grey bars do not represent only the benthic fluxes shown on Figure 4 and extrapolated to the whole Bay; they represent the sum of the river and benthic DSi inputs, which can be directly compared to the diatom demand (black bars). (b) Same description, only C. fornicata has been artificially removed from the system by applying the benthic flux measured at the site without C. fornicata to the whole Bay and not only to half of it. From Ragueneau et al. [382].

Four budgets were made for the productive season (Figure 7a), neglecting DSi inputs from the adjacent ocean, as they represent less than 5% of the diatom demand during the productive period [385]. These budgets clearly demonstrate the importance of suspension-feeder activity on the Si cycle allowing for DSi to be available for diatom production during late spring and summer. By early spring, river Si inputs can sustain nearly 100% of the diatom demand; diatoms do not depend upon recycling at the sediment-water interface, especially if we add the winter stock of DSi that can account for one third of the initial diatom demand [385]. By late spring, river inputs have decreased by a factor of three and can sustain only 30% of the diatom demand. The rest must be met by recycling at the sediment-water interface. By mid-summer, river inputs are even smaller and DSi benthic fluxes alone can sustain diatom demand. Because recycling also occurs in the water column, DSi is probably available in excess and starts to accumulate in the water column (Figure 3). Note that in September, DSi inputs exceed the diatom demand by about 140×10⁴ mol Si d^-1(Figure 7a). The Bay volume is close to 2×10⁹ m³ on average, which means that DSi should accumulate at a rate of roughly 0.07 µmol L^-1 d^-1. From late July onwards, DSi increases linearly from 0 to 12 µM in 5 months (Figure 3). This corresponds to a rate of 0.08 µmol L^-1 d^-1, which is very consistent with the above budget calculation. During fall, diatom demand decreases sharply whereas benthic fluxes are still high and river inputs have increased again due to rainfall. DSi continues to accumulate in the water column at the mean rate calculated above and will soon reach its winter maximum concentrations. The budgets presented demonstrate unambiguously the importance of recycling at the sediment-water interface in sustaining diatom demand throughout the productive period.

14.3 The case of the silicate pump in the Bay of Brest