2.1 General setup

The general setup of the experiments is fully detailed in Gazeau et al. (2021). Briefly, three experiments were performed at the long-duration stations TYR, ION and FAST during the PEACETIME cruise on board R/V Pourquoi Pas? (Fig. 1). During these experiments (3 to 4 d each), seawater was incubated in 300 L tanks (Fig. S1 in the Supplement) installed in a temperature-controlled container, in which the irradiance spectrum and intensity can be finely controlled and in which future ocean acidification and warming conditions can be fully reproduced. The tanks were made of high-density polyethylene (HDPE) and were trace-metal-free in order to avoid contaminations, with a height of 1.09 m, a diameter of 0.68 m, a surface area of 0.36 m² and a volume of 0.28 m³. The conical base of the tanks was equipped with a sediment trap that was left open during the duration of the experiments and removed at the end. The experimental protocol comprised two unmodified control tanks (C1 and C2), two tanks enriched with Saharan dust (D1 and D2), and two tanks enriched with Saharan dust and maintained simultaneously under warmer (+3 ^∘C) and acidified (−0.3 pH unit) conditions (G1 and G2). At the three stations, tanks were always filled at the end of the day before the start of the experiments: TYR (17 May 2017), ION (25 May 2017) and FAST (2 June 2017). The tanks were filled by means of a large peristaltic pump (Verder© VF40 with EPDM hose, flow of 1200 L h⁻¹) collecting seawater below the base of the boat (depth of ∼5 m), used to continuously supply surface seawater to a series of instruments during the entire campaign. While filling the tanks, seawater was sampled for the measurements of selected parameters (sampling $\text{time}=t-\mathrm{12}$ h). After filling the tanks, seawater was slowly warmed overnight using 500 W heaters, controlled by temperature-regulation units (COREMA©), in G1 and G2 to reach an offset of +3 ^∘C. ¹³C bicarbonate was added to all tanks at 04:00 LT (all times in local time), and G1 and G2 were acidified by addition of CO₂-saturated filtered (0.2 µm) seawater (∼1.5 L in 300 L; collected when filling the tanks at each station) at 04:30 LT to reach a pH offset of −0.3. Sampling for many parameters took place prior to dust seeding (sampling time=t0). Dust seeding was performed between 07:00 and 09:00 LT in tanks D1, D2, G1 and G2. The same dust analog was used and the same dust flux was simulated as for the DUNE 2009 experiments described in Desboeufs et al. (2014). Briefly, the fine fraction (<20 µm) of Saharan soils collected in southern Tunisia, which is a major source of dust deposition over the northwestern Mediterranean basin, was used in the seeding experiments. The particle size distribution showed that 99 % of particles had a size smaller than 0.1 µm, and that particles were mostly made of quartz (40 %), calcite (30 %) and clay (25 %; Desboeufs et al., 2014). This collected dust underwent an artificial chemical aging process by addition of nitric and sulfuric acid (HNO₃ and H₂SO₄, respectively) to mimic cloud processes during atmospheric transport of aerosol with anthropogenic acid gases (Guieu et al., 2010, and references therein). To mimic a wet flux event of 10 g m⁻², 3.6 g of this analog dust was quickly diluted into 2 L of ultrahigh-purity water (UHP water; 18.2 MΩ cm⁻¹ resistivity) and sprayed at the surface of the tanks using an all-plastic garden sprayer (duration=30 min). The intensity of this simulated wet deposition event (i.e., 10 g m⁻²) represents a high but realistic scenario, as several studies reported even higher short wet deposition events in this area of the Mediterranean Sea (Bonnet and Guieu, 2006; Loÿe-Pilot and Martin, 1996; Ternon et al., 2010).

Depending on the considered parameter or process, seawater sampling was conducted 1 (t1 h), 6 (t6 h), 12 (t12 h), 24 (t24 h), 48 (t48 h) and 72 h (t72 h) after dust additions in all three experiments with an additional sample after 96 h (t96 h) at FAST). Acid-washed silicone tubes were used for transferring the water collected from the tanks to the different vials or containers.

2.2 Stocks

2.3 Processes

2.4 Data processing

All metabolic rates were integrated over the duration of the experiments using trapezoidal integrations, and the relative changes (in %) in tanks D and G as compared to the controls (average between C1 and C2) were computed as follows:

where Rate_Treatment is the integrated rate measured in treatments D and G (D1, D2, G1 or G2) and Rate_Controls is the averaged integrated rates between the duplicate controls (treatment C). Daily rates of ¹⁴C-based production were computed from hourly rates assuming a 14 h daylight period. As incubations performed from samples taken at t0 (before dust addition) do not represent what happened in the tanks between t0 and t24 h, as a first assumption, we considered a linear evolution between these rates and those measured from samples at t24 h and recomputed an average value for the time interval t0−t24 h. At FAST, no incubations were performed for ¹⁴C incorporation and oxygen metabolism between t72 h and t96 h; again an average rate between rates measured from samples taken at t48 h and t96 h was used for this time interval. Since bacterial respiration rates were not measured, bacterial growth efficiency (BGE, expressed as a percentage) was estimated based on BP (carbon units) and community respiration (CR, oxygen units). As BP was determined more often than CR during the first 48 h, hourly BP rates were integrated using trapezoidal integrations during the time period when CR was measured. We assumed that heterotrophic prokaryotes were responsible for 70 % of CR (BR$/$CR ratio; Lemée et al., 2002) and used a respiratory quotient (RQ) of 0.8 (del Giorgio and Williams, 2005), following this equation:

When BP varied following an exponential growth, we calculated growth rates (µ_BP) from linear least square regression of the natural logarithm of BP rates versus time.

3.1 Initial conditions

Initial conditions in terms of the chemical and biological standing stocks measured while filling the tanks at the three stations are fully described in Gazeau et al. (2021). Briefly, the three experiments were conducted with surface seawater collected during stratified oligotrophic conditions typical of the open Mediterranean Sea at this period of the year (Table 1). Nevertheless, a dust event took place nine days before sampling at station TYR as evidenced from the particulate inventory of lithogenic proxies (Al, Fe) in the water column (Bressac et al., 2021), likely stimulating phytoplankton growth before the start of the experiment. Nitrate and nitrite (NO_x) concentrations were maximal at station FAST with a NO_x to dissolved inorganic phosphate (DIP) molar ratio of ∼4.6. Very low NO_x concentrations were observed at stations TYR and ION (14 and 18 nmol L⁻¹, respectively). DIP concentrations were the highest at station TYR (17 nmol L⁻¹) and the lowest at the most eastern station (ION, 7 nmol L⁻¹). Consequently, the lowest NO_x:DIP ratio was measured at TYR (0.8), compared to ION and FAST (2.8 and 4.6, respectively). Silicate (Si(OH₄)) concentrations were similar at TYR and ION (∼1 µmol L⁻¹) and the lowest at FAST (∼0.6 µmol L⁻¹). Both POC and DOC concentrations were the highest at station TYR (12.9 and 72.2 µmol L⁻¹, respectively) and the lowest at FAST (6.0 and 69.6 µmol L⁻¹, respectively). Very low and similar concentrations of chlorophyll a were measured at the three stations (0.063–0.072 µg L⁻¹). Phytoplankton communities at stations TYR and ION were dominated by prymnesiophytes followed by cyanobacteria, while, at station FAST, the phytoplanktonic community was clearly dominated by photosynthetic prokaryotes. At all three stations, the proportion of pigments representative of larger species was very small (<5 %; Gazeau et al., 2021). Heterotrophic prokaryotes were the most abundant at station FAST (6.15×10⁵ cells mL⁻¹) and the least abundant at station ION (2.14×10⁵ cells mL⁻¹).

Table 1Initial chemical and biological stocks as measured while filling the tanks (initial conditions in pumped surface water; sampling time: t−12 h). NO_x: nitrate and nitrite, DIP: dissolved inorganic phosphorus, Si(OH)₄: silicate, POC: particulate organic carbon, DOC: dissolved organic carbon, TEPs: transparent exopolymer particles, TChl a: total chlorophyll a. Values shown for ¹⁴C incorporation rates and percentages of extracellular release (%PER) as well as for net community production (NCP), community respiration (CR) and gross primary production (GPP) were estimated from samples taken at t0 in the control tanks. For heterotrophic bacterial production (BP), rates were estimated from seawater sampled at t−12 h.

Download Print Version | Download XLSX

Relatively similar ¹⁴C-based particulate production rates were measured at the start of the experiments (t0) in the control tanks (C1 and C2) at station ION and FAST (ca. 0.014–0.015 $\mathrm{\mu }\mathrm{g}\mathrm{C}{\mathrm{L}}^{-\mathrm{1}}{\mathrm{h}}^{-\mathrm{1}}$). At both stations, ca. 80 % of the production was attributed to larger (>2 µm) cells and the percentage of extracellular release (%PER) did not exceed 45 %. Lower rates were estimated at station TYR (total particulate production of 0.08 $\mathrm{\mu }\mathrm{g}\mathrm{C}{\mathrm{L}}^{-\mathrm{1}}{\mathrm{h}}^{-\mathrm{1}}$) from which 87.5 % was due to large cells >2 µm. A larger amount of ¹⁴C incorporation was released as dissolved organic matter at station TYR compared to the two other stations (PER ca. 60 %). Metabolic balance derived from oxygen measurements showed that, at all three stations, the community was net heterotrophic with a higher degree of heterotrophy at station TYR (NCPs were −1.9, −0.2 and −0.8 $\mathrm{\mu }\mathrm{mol}{\mathrm{O}}_{\mathrm{2}}{\mathrm{L}}^{-\mathrm{1}}{\mathrm{d}}^{-\mathrm{1}}$ at TYR, ION and FAST, respectively, as measured in the controls from seawater sampled at t0). CR and GPP rates were respectively the highest and the lowest at station TYR compared to the other two stations. Finally, BP rates were the highest at station FAST (35.8 $\mathrm{ng}\mathrm{C}{\mathrm{L}}^{-\mathrm{1}}{\mathrm{h}}^{-\mathrm{1}}$), intermediate at ION (26.1 $\mathrm{ng}\mathrm{C}{\mathrm{L}}^{-\mathrm{1}}{\mathrm{h}}^{-\mathrm{1}}$) and the lowest at TYR (21.3 $\mathrm{ng}\mathrm{C}{\mathrm{L}}^{-\mathrm{1}}{\mathrm{h}}^{-\mathrm{1}}$).

3.2 Changes in biological stocks

DOC concentrations showed a general positive trend during the three experiments and a large variability between duplicates (Fig. 2). This variability appeared as soon as 1 h after dust seeding (t1 h) while the range of variation at t0 (before dust seeding) was rather moderate (difference between minimal and maximal values in all tanks of 1.3, 6.2 and 4.3 µmol C L⁻¹ at station TYR, ION and FAST, respectively). As a consequence of this variability, no clear impact of dust seeding (D) could be highlighted at station TYR and FAST. Indeed, DOC concentrations in the two duplicates (D1 and D2) were higher than values in the controls (C1 and C2) in only 33 % of the samples along the experiments (after dust seeding). In contrast, at station ION, DOC concentrations appeared impacted by dust seeding as higher concentrations were almost systematically (83 % of the time after dust seeding) measured for this treatment as compared to control tanks at the same time. At all stations, this impact was somewhat exacerbated under conditions of temperature and pH projected for 2100 (G1 and G2) as DOC concentrations were higher almost all the time in these tanks than in control tanks (83 %–100 % of the samples after dust seeding, depending on the station).

Figure 2Dissolved organic carbon (DOC) concentrations and ratio between total hydrolysable amino acids (TAA) and DOC concentrations measured in the six tanks (controls: C1, C2; dust addition under present conditions of temperature and pH: D1, D2; dust addition under future conditions of temperature and pH: G1 and G2) during the three experiments (TYR, ION and FAST). The dashed vertical line indicates the time of seeding (after t0).

Download

Total hydrolysable carbohydrates and amino acid concentrations along the three experiments are shown in Fig. S2 in the Supplement. TCHO concentrations were quite variable between tanks before dust seeding (t0; 649–954, 569–660 and 600–744 nmol L⁻¹ at station TYR, ION and FAST, respectively), and no visible impact of the treatments was visible at station TYR (TCHO tended to decrease everywhere). In contrast, at station ION and FAST, values in dust-amended tanks increased and appeared higher than in control tanks towards the end of the experiments, although the large variability between duplicates tended to mask this potential effect. An impact of dust seeding was much clearer for TAA concentrations that showed larger increases throughout the three experiments in tanks D1 and D2 as compared to control tanks, this effect being exacerbated for warmer and acidified tanks (G1 and G2). The ratio between TAA and DOC concentrations (Fig. 2) showed positive trends in tanks D and G during all three experiments with a clear distinction between treatments at the end of the experiments ($\mathrm{G}>\mathrm{D}>\mathrm{C}$). The strongest increase was observed at station FAST in tanks G where final TAA$/$DOC ratios were above 3 %.

Figure 3Particulate organic carbon (POC) concentrations and transparent exopolymer particle carbon content (TEP-C) measured in the six tanks (controls: C1, C2; dust addition under present conditions of temperature and pH: D1, D2; dust addition under future conditions of temperature and pH: G1 and G2) during the three experiments (TYR, ION and FAST). The dashed vertical line indicates the time of seeding (after t0).

Download

Particulate organic carbon (POC) concentrations strongly decreased at all stations between t−12 h and t0, this decrease being the largest at station TYR where concentrations dropped from 25.7 to 9.6–13.2 µmol C L⁻¹ (Fig. 3). After dust seeding, POC concentrations did not show clear temporal trends for the three experiments although a slight general increase could be observed at station FAST. Furthermore, no impact of dust seeding and warming/acidification could be observed on POC dynamics. While concentrations of TEP-C were rather constant through time in control tanks at the three stations, a large increase was observed in dust-amended tanks (D and G) with TEP-C reaching values up to ∼2 µmol C L⁻¹ in tank G1 at station TYR after 24 h (i.e., ∼17 % of POC concentration, Fig. 3). In all cases except for tank G2 at station ION, TEP-C further decreased towards the end of the experiments, although concentrations remained well above those observed in the controls. As the variability between duplicated tanks G was rather high, no impact of warming/acidification on TEP dynamics could be highlighted at the three stations.

3.3 Changes in metabolic rates

¹⁴C-based particulate production rates as measured during the different time intervals at the three stations were low in control tanks (maximal total particulate production of 0.34 $\mathrm{\mu }\mathrm{g}{\mathrm{L}}^{-\mathrm{1}}{\mathrm{h}}^{-\mathrm{1}}$ at station FAST) and did not show any particular temporal dynamics (Fig. 4). In these tanks, the vast majority of particulate production was attributed to cells above 2 µm (65 %–89 %). The percentage of extracellular release (%PER) was overall maximal at station TYR and minimal at station FAST with a tendency to decrease with time at the three stations, although large variations were observed between duplicates.

Figure 4¹⁴C-based production rates ((a) <2 µm and (b) >2 µm size fractions, (c) total particulate) estimated from 8 h incubations on samples taken in the six tanks (controls: C1, C2; dust addition under present conditions of temperature and pH: D1, D2; dust addition under future conditions of temperature and pH: G1 and G2) during the three experiments (TYR, ION and FAST). The percentage of extracellular release (d, %PER) is also shown.

Download

Dust addition alone did not have any clear positive impact on all ¹⁴C-based rates at station TYR, with even an observable decrease in production rates from larger cells (>2 µm) compared to the controls. In contrast, at this station, dust seeding under warmer and acidified conditions (tanks G) had a positive effect on particulate production rates, this effect being particularly visible for cells <2 µm and to a lesser extent on dissolved production with a general decrease in %PER. An important discrepancy between the duplicates of treatment G was observable at the end of the experiment with much larger rates measured in tank G2.

In contrast to station TYR, an enhancement effect of dust addition was clearly visible at station ION where all rates increased towards the end of this experiment reaching a maximal total particulate production of 0.6–0.7 $\mathrm{\mu }\mathrm{g}{\mathrm{L}}^{-\mathrm{1}}{\mathrm{h}}^{-\mathrm{1}}$ in tanks D1 and D2. Since this positive effect was similar between small and larger cells, dust addition alone had no effect on the partitioning of production at this station, with cells >2 µm representing ∼80 % of total production. Although it was also positively impacted and increasing with time, dissolved production appeared less sensitive than particulate production leading to an overall decrease in %PER at station ION following dust addition. These positive impacts of dust seeding on ¹⁴C-based particulate production rates were even more visible at this station under warmer and acidified conditions (tanks G), with maximal rates more than doubled compared to those measured under present conditions of temperature and pH (1.5–1.6 $\mathrm{\mu }\mathrm{g}{\mathrm{L}}^{-\mathrm{1}}{\mathrm{h}}^{-\mathrm{1}}$). Dust seeding under warmer and acidified conditions had a slight impact on the partitioning of particulate production at station ION, with smaller cells benefiting the most from these conditions. %PER remained between 20 % and 30 %.

At station FAST, similarly to station ION, total particulate production rates were clearly enhanced by dust addition (tanks D), reaching maximal values during the incubation time interval t48–56 h. No clear increase was observed for total particulate production on the next incubation (t96–120 h) while production rates of cells larger than 2 µm increased and rates of smaller cells decreased. However, in contrast to station ION, there was much less impact of warming/acidification on all measured rates at station FAST, although rates measured on smaller cells (<2 µm) did not decrease at the end of the experiment as observed under present environmental conditions. %PER under both present conditions of temperature and pH (tanks D) decreased during this experiment, reaching values lower than in the controls and in tanks G.

The initial enrichment of the tanks in ¹³C bicarbonate led to an increase in the ¹³C signature of dissolved inorganic carbon (δ¹³C C_T) of above 300 ‰, with generally lower values measured in warmer and acidified tanks (G; Fig. S3 in the Supplement). After this initial enrichment, δ¹³C C_T levels decreased linearly in all tanks. At stations TYR and ION, the isotopic signature of dissolved organic carbon (δ¹³C DOC; Fig. S3) increased with time, although these increases were rather low and limited to ∼4 ‰ over the course of the experiments. In contrast to station TYR, at ION, an enhanced incorporation of ¹³C into DOC was visible after 24 h in tanks D and G in comparison to control tanks. A similar observation was done at station FAST, especially at the end of the experiment, although much more variability was observed at this station.

Figure 5Incorporation of ¹³C into particulate organic carbon (δ¹³C POC) in the six tanks (controls: C1, C2; dust addition under present conditions of temperature and pH: D1, D2; dust addition under future conditions of temperature and pH: G1 and G2) during the three experiments (TYR, ION and FAST). The dashed vertical line indicates the time of seeding (after t0).

Download

The incorporation of ¹³C into particulate organic carbon (δ¹³C POC) is shown in Fig. 5. At all stations, δ¹³C POC increased with time but reached lower enrichment levels at station TYR as compared to ION and FAST. At station TYR, incorporation rates appeared smaller in dust-amended tanks under present environmental conditions (tanks D). As for ¹⁴C-based production rates, an important discrepancy was observed between duplicates under future conditions of temperature and pH (tanks G) with much higher final δ¹³C POC at the end of the experiment in tank G2. At station ION, enrichment levels obtained at the end of the experiment were more important in dust-amended tanks reaching maximal levels of 73 ‰ in tank G2 at t72 h. This enhancement effect was even more visible at station FAST with maximal enrichment levels of 146 ‰ (tank D2 at t96 h). Since no sampling occurred at t72 h, these enrichment levels cannot be directly compared to what was measured at station TYR and ION. However, by interpolating values at t72 h assuming a linear increase between these time intervals, enrichment levels appeared similar although slightly higher for tanks D between station ION and FAST.

Figure 6(a) Net community production (NCP), (b) community respiration (CR) and (c) gross primary production (GPP) rates estimated using the oxygen light–dark method (24 h incubations) on samples taken in the six tanks (C1, C2, D1, D2, G1 and G2) during the three experiments (TYR, ION and FAST).

Download

NCP rates as measured using the O₂ light–dark method showed that, under control conditions, the communities remained in a net heterotrophic state (NCP<0; Fig. 6) the vast majority of the time throughout the three experiments. This was especially true at station TYR where the lowest NCP rates were measured. At this station, dust addition whether under present or future conditions of temperature and pH did not switch the community towards net autotrophy but even drove the community towards a stronger heterotrophy. This was related to the fact that while gross primary production rates were not positively impacted, community respiration increased in tanks D and G. At station ION, dust addition alone (tanks D) led to a switch from net heterotrophy to net autotrophy after two days of incubation due to a stronger positive effect of dust on GPP than on CR. Under future environmental conditions (tanks G), the same observation was made with higher NCP and GPP rates than in tanks D. CR rates reacted quickly to these forcing factors in tanks G and initially (first incubation) drove the community towards a much stronger heterotrophy as compared to the other tanks. Finally, at station FAST, similarly to what was observed at ION, the community became autotrophic after two days of incubation in dust-amended tanks as, although both GPP and CR were positively impacted by dust addition, this impact was less important for CR. Warming and acidification had a limiting impact on this enhancement, with a lower final NCP in tanks G compared to tanks D, a difference that can be related to an absence of effects of these environmental stressors on GPP while CR clearly increased at higher temperature and lower pH.

Figure 7Heterotrophic bacterial production rates (BP) and cell-specific maximum hydrolysis velocity (V_m) of the alkaline phosphatase (both over 1–2 h incubations) on samples taken in the six tanks (C1, C2, D1, D2, G1 and G2) during the three experiments (TYR, ION and FAST).

Download

While BP remained constant or gradually increased in control tanks depending on the station, a clear and quick fertilization effect was observable following dust addition (treatment D and G) at all stations (Fig. 7). At station TYR, BP rates sharply increased to reach maximal values at t24 h, with an even stronger increase observed under warmer and acidified conditions (tanks G). After this initial increase, rates slightly decreased towards the end of the experiment. This fertilization effect appeared less important at station ION where lower maximal rates were obtained after 24 h as compared to station TYR. Nevertheless, the same observations can be made, namely, (1) higher rates were measured under future temperature and pH levels and (2) after this initial sharp increase, rates gradually decreased towards the end of the experiment especially in tanks G. At station FAST, a much stronger effect of warming/acidification was observed with an important increase in BP in tanks G until 24 or 48 h post-seeding, depending on the duplicate. A sharp decline was observed for this treatment until the end of the experiment although rates remained higher than those measured in tanks C and D. The impact of dust addition under present environmental conditions (tanks D) was somehow more limited than at the other stations with a gradual increase until t72 h with maximal rates ∼40–100 % higher than rates measured in the controls. However, BP increased exponentially between t0 and t12 h in all tanks including controls, and in all experiments (Table 2). The growth rate of BP (µ_BP) in control tanks was the highest at TYR, intermediate at ION and the lowest at FAST. µ_BP increased significantly in all dust-amended tanks compared to controls. Under future environmental scenarios, µ_BP tended to increase compared to treatment D but with a variable relative change (Table 2).

Table 2Heterotrophic bacterial production (BP) growth rates (μ_BP in h⁻¹) estimated from the exponential phase of BP growth, observable from at least four sampling points, between t0 and t12 h, during the three experiments (TYR, ION and FAST) in the six tanks (controls: C1, C2; dust addition under present conditions of temperature and pH: D1, D2; dust addition under future conditions of temperature and pH: G1 and G2). Values±SE are shown.

Download Print Version | Download XLSX

Table 3Estimated bacterial growth efficiency (BGE in %) during the course of the three experiments (TYR, ION and FAST) in the six tanks (controls: C1, C2; dust addition under present conditions of temperature and pH: D1, D2; dust addition under future conditions of temperature and pH: G1 and G2). BGE was calculated based on integrated heterotrophic bacterial production (BP) and community respiration (CR) rates by applying a bacterial respiration to a CR ratio of 0.7 and a respiratory quotient of 0.8 (see Material and methods).

Download Print Version | Download XLSX

BGE increased in dust-amended tanks under present environmental conditions (treatment D) at TYR and ION, while no changes were detectable at station FAST due to a strong discrepancy between control duplicates and overall higher BGE at this station in the controls (Table 3). In contrast, warming and acidification exerted the strongest effect at station FAST with a doubling of BGE between treatment G and D. Although an increase in BGE was also observed at the two other stations in treatment G as compared to present environmental conditions (treatment D), this increase was more limited (ca. 1- to 1.4-fold increase).

The alkaline phosphatase V_m (AP V_m) increased in all experiments after dust seeding, with amplified effects in G treatments (Fig. S4 in the Supplement). Note that AP V_m also increased in the controls at TYR and FAST. In contrast, leucine aminopeptidase V_m (LAP V_m) showed a succession of peaks instead of a continuously increase (Fig. S4). It was higher in dust-only treatment (D) as compared to the controls at TYR and FAST. A larger variability between duplicates at ION prevents such an observation. At all stations, maximum velocities were measured under future environmental conditions (G). V_m being possibly influenced by enzyme synthesis but also by the number of cells inducing such enzymes, we also computed specific AP V_m per heterotrophic bacterial cell (Fig. 7). Specific AP V_m slightly increased during all experiments in controls and dust-amended tanks (D), with no visible differences between these treatments, and a clear over-expression of this enzyme was observed under warmer and more acidified conditions (treatment G) especially at station FAST where velocities were enhanced by a ∼8-fold at t96 h.

3.4 Inorganic and organic material export

Both total mass and organic matter fluxes, as measured from analyses of the sediment traps at the end of each experiment, were extremely low under control conditions (Fig. 8). Only less than 30 % of the dust introduced at the surface of the tanks were recovered at the end of the experiment (3 or 4 d after) in the sediment traps with $\text{TYR}>\text{ION}>\text{FAST}$. The composition of the exported material was quite similar for each experiment with no significant difference between D and G treatments with 3 %–5 % opal, 4 % organic matter, 35 %–36 % CaCO₃ and 48 %–54 % lithogenic matter (Fig. S5 in the Supplement). Additions of dust in tanks D and G led to a strong increase in both fluxes with a large variability between the duplicates of treatment D at ION. No clear changes between tanks maintained under present and future conditions of temperature and pH could be highlighted.

Figure 8Total mass and organic matter fluxes measured in the sediment traps at the end of the three experiments (TYR, ION and FAST) in the six tanks (C1, C2, D1, D2, G1 and G2).

Download

4.1 Initial conditions of the tested waters and evolution in controls

As discussed in detail in the companion paper from Gazeau et al. (2021), the three sampling stations were typical of stratified (mixed layer depth of 10–20 m) oligotrophic conditions encountered in the open Mediterranean Sea in late spring–early summer. Briefly, the low NO_x:DIP ratio nutrient concentrations suggest that communities found at the three stations experienced N and P co-limitation at the start of the experiments. The composition of the smaller-size phytoplankton communities differed substantially, with autotrophic nano-eukaryotes dominating at stations TYR and ION and a larger contribution from autotrophic pico-eukaryotes and cyanobacteria at station FAST. The observed low total chlorophyll a concentrations and the small contribution of large phytoplankton cells at the start of the three experiments are characteristic of LNLC areas in general, and of surface Mediterranean waters in late spring and summer (Siokou-Frangou et al., 2010). DOC concentrations at the start of the experiments were in the same range (60–75 µmol C L⁻¹) as those measured from samples collected in surface waters using clean sampling procedures (Van Wambeke et al., 2021), revealing no contamination issues from our sampling device. TAA concentrations as measured in the tanks at t0 were also consistent with measurements from surface water samples (Van Wambeke et al., 2021) with an average across stations and treatments of 254±36 nmol L⁻¹ (Fig. S2 in the Supplement). In contrast, TCHO appeared higher at t0 (average across stations and treatments of 681±98 nmol L⁻¹) than concentrations based on clean in situ sampling (average of 595±43 nmol L⁻¹; Van Wambeke et al., 2021). The decrease in POC concentrations between pumping (t−12 h) and t0 for the three experiments, especially at station TYR (likely linked to higher initial concentrations), was likely a consequence of sedimentation of senescent cells and/or fecal pellets in our experimental systems, which are designed to evaluate the export of matter thanks to their conical shape. TEP concentrations were not quantified at t−12 h and therefore there is no possibility to evaluate if sedimentation of these particles occurred before t0 in our tanks. At t0, larger and more abundant TEPs were measured at station TYR compared to the two other stations (data not shown).

As a consequence of a very low availability of inorganic nutrients, TChl a and ¹⁴C-based production rates were very low, all typical of oligotrophic conditions. Nano- and micro-phytoplanktonic cells (>2 µm) contributed most of the ¹⁴C-based particulate production (∼80 %), as found also on several on-deck incubations at the three stations (on average 73±6 %; Marañón et al., 2021). %PER values were also very similar to those measured during these on-deck incubations (∼40 %–45 %; see Marañón et al., 2021). This suggests there was no significant impact of our experimental protocol on rates and partitioning of ¹⁴C-based production rates (i.e., sampling from the continuous seawater supply, delay of 12 h before initial measurements, artificial light etc.). The low values of chlorophyll stocks as well as of ¹⁴C-based production rates are consistent with previous estimates based on direct measurements, satellite observations and modeling approaches in the same areas in late spring–early summer (e.g., Bosc et al., 2004; Lazzari et al., 2016; Moutin and Raimbault, 2002).

The metabolic balance was in favor of net heterotrophy at all stations at the start of the experiments (NCP<0). Net heterotrophy in the open Mediterranean Sea at this period of the year has been reported by Regaudie-de-Gioux et al. (2009) and Christaki et al. (2011) in agreement with our measurements at t0 in control tanks (Table 1). The lowest NCP and the highest CR rates were measured at station TYR, suggesting that the autotrophic plankton community was not very active at this station. This was confirmed by the ¹⁴C-based particulate production rates, which were about half those measured at the other two stations. The community at TYR was most likely relying on regenerated nutrients, as shown by the highest levels of ammonium (${\mathrm{NH}}_{\mathrm{4}}^{+}$) measured at the start of this experiment (Gazeau et al., 2021). As discussed in Guieu et al. (2020b), a dust deposition event took place several days before the arrival of the vessel in this area, likely on 10–12 May. This dust event was confirmed by inventory of particulate aluminum in the water column at several stations of the Tyrrhenian Sea including TYR, 6 to 9 d after the event (Bressac et al., 2021). This dust deposition likely stimulated phytoplankton growth and POC accumulation shortly after the deposition and consequently increased the abundance of herbivorous grazers (copepods) and attracted carnivorous species (Feliú et al., 2020), subsequently driving the community towards a net heterotrophic state that characterized the initial condition of the experiment at this station. The favorable conditions for BP growth at TYR were also confirmed by the highest μ_BP growth rates obtained among the three experiments (Table 2; 0.06–0.07 h⁻¹) in controls tanks.

The two other stations, although both also showing a slight net heterotrophic state, were clearly different from each other in terms of initial biological stocks and metabolic rates. Indeed, whereas TChl a and abundances of pico- and nano-autotrophic cells (flow cytometry counts; Gazeau et al., 2021) were higher at FAST compared to ION, the autotrophic community was not more efficient at fixing carbon at FAST, as shown by similar initial ¹⁴C-based production rates. In contrast, both heterotrophic prokaryotic abundances and BP were much higher at station FAST as compared to ION, leading to initial higher CR and lower NCP. At ION, the initial NCP closer to metabolic balance further suggests a tight coupling between heterotrophic prokaryotes and phytoplankton at this station, as discussed by Dinasquet et al. (2021).

For most of the chemical and biological stocks (e.g., nutrients, pigments) presented in Gazeau et al. (2021), no major changes took place during the three experiments under control conditions. Here, we further show that DOC, POC and TEP concentrations did not exhibit strong changes during the experiments. For DOC, large variability between the duplicates (C1 and C2) potentially masked an increase towards the end of the experiments. The same holds true for autotrophic metabolic rates, as ¹⁴C-based particulate production rates showed no marked variations during the three experiments, although a slight increase was visible at FAST until t48 h. The communities at the three stations remained heterotrophic under the nutrient-limited conditions in the controls. However, heterotrophic prokaryotes probably benefited from initial inputs of available organic matter issued from other stressed eukaryotic organisms and/or POC decay between t−12 h and t0, which could be due to both sedimentation and degradation. This was reflected in the progressive increase in BP, their variable initial growth rates (μ_BP ranged from 0.02 to 0.06 h⁻¹ in control tanks according to the experiment) as well as increasing TAA$/$DOC ratios at the three stations. Finally, an initial increase in BP during incubations is generally described and classically attributed to a bottle effect, which favors large, fast-growing bacteria and often induces mortality of some phytoplankton cells (Calvo-Díaz et al., 2011; Ferguson et al., 1984; Zobell and Anderson, 1936).

4.2 Impact of dust addition under present environmental conditions

The addition of nitrogen and phosphorus in the experimental tanks through dust seeding (+11 to +11.6 µmol L⁻¹ and +22 to +30.8 nmol L⁻¹ for NO_x and DIP, respectively, in dust enriched, i.e., D1 and D2, versus controls; Gazeau et al., 2021) had very contrasting impacts on the metabolism of the communities, depending on the station. At TYR, surprisingly, the relieving of nutrient (N, P) limitation had a negative impact on ¹³C incorporation as well as on both particulate and dissolved ¹⁴C-based production rates (as seen by the relative changes compared to the control presented in Fig. 9). These observations are fully corroborated by the observed relative decrease in GPP in these tanks (D1 and D2) relative to controls and by the negative impact of dust-addition on TChl a concentrations as discussed by Gazeau et al. (2021). Integrated ¹⁴C-incorporation rates converted to P (using a C:P molar ratio of 245:1 determined in the particulate organic matter in surface waters of the northwestern Mediterranean Sea during stratification; Tanaka et al., 2011) showed that phytoplankton P requirements in treatment D (∼2 nmol P L⁻¹) were much lower than the release of DIP through dust addition at station TYR (+20.4 to +24.6 nmol P L⁻¹; Gazeau et al., 2021). This suggests that the observed strong decrease in DIP at TYR following dust addition was due to a utilization by the heterotrophic compartment. Indeed, in contrast to the autotrophic compartment, both heterotrophic prokaryotic abundances (Gazeau et al., 2021) and BP (this study, Fig. 9) showed that heterotrophic prokaryotes reacted quickly and strongly to the increase in DIP availability. Integrated BP increased by almost 400 % in tanks D1 and D2 as compared to controls (Fig. 9). Such relative increases in BP, surpassing by far the observed relative increases in CR, suggest a much more efficient utilization of resources by heterotrophic prokaryotes in this treatment (i.e., BGE increased by 200 % as compared to the controls; Fig. 9). As such, at TYR, the addition of dust drove the community to an even more heterotrophic state. Such an absence of response of the autotrophic community despite the input of new N and P from simulated wet deposition was never observed in dust enrichment experiments performed in the Mediterranean Sea (Guieu and Ridame, 2020). To the best of our knowledge, it is the first time that a negative effect of dust addition is experimentally demonstrated on the metabolic balance. The apparent utilization of nutrients, especially DIP (Gazeau et al., 2021), by heterotrophic prokaryotes was extremely fast, starting right after dust addition and driving DIP concentrations back to control levels at the end of the experiment (t72 h). While heterotrophic prokaryotic abundances increased until the end of the experiment, BP rates increased exponentially during the first 24 h, and then BP reached a plateau. Heterotrophic prokaryotes appeared limited by nutritive resources although DIP concentrations were not yet back to their initial level and no relative increase in the AP V_m per cell compared to the control was observed in these tanks. Independent nutrient experiments showed a direct stimulation of BP in the dark after addition of DIP (Van Wambeke et al., 2021), suggesting a great competition with phytoplankton for DIP utilization at TYR. After 24 h, abundances of heterotrophic prokaryotes continued to increase while BP stabilized, suggesting a lesser extent of lysis and viral control than in the other experiments (abundances of heterotrophic nanoflagellates decreased; Dinasquet et al., 2021). This limitation of BP was potentially a consequence of relatively less available access to labile DOC sources, as ¹⁴C-based production rates decreased relative to the controls at t24 h and t48 h although BP increased by 200 %–800 %. The very tight coupling between phytoplankton and bacteria at all stations investigated was further confirmed by the absence of an important ¹³C incorporation into DOC (Fig. S3).

Figure 9Relative difference (%) between integrated rates measured in tanks D (D1, D2; dust addition under present conditions of temperature and pH) and G (G1, G2; dust addition under future conditions of temperature and pH) as compared to the controls (C1, C2) during the three experiments (TYR, ION and FAST). Vertical boxes represent the range observed between the two replicates per treatment.

Download

At stations ION and FAST, in contrast to TYR, both the autotrophic and heterotrophic community benefited from dust addition relative to the controls (Fig. 9). Interestingly, while the relative increase in integrated autotrophic processes (GPP and all ¹⁴C-based production rates) was more important at FAST than at ION, the opposite was observed for BP. Estimated BGE values even suggest an absence of response to dust addition at station FAST compared to the controls. The different (relative) responses of BP at the two stations could be partly explained by the dynamics of BP in the control tanks as no clear pattern could be observed at ION while a continuous increase was observed at FAST. As shown by Gazeau et al. (2021), at FAST, abundances of heterotrophic prokaryotes were much higher at the start of the experiment, further increased until t48 h and then declined until the end of the experiment.

We can rule out a potential limitation of BP from DIP availability at station FAST as DIP levels remained much higher in tanks D than in the controls (Gazeau et al., 2021). Furthermore, the amount of maximum DIP reached before its decline compared to TYR and ION showed a less important direct DIP uptake, suggesting that communities were not as much P limited at FAST compared to the other stations at the start of the experiment. Finally, no increase in specific AP V_m was observed in these tanks as compared to the controls (Fig. 7), suggesting no particular additional needs for AP synthesis per unit cell following dust addition. A potential explanation resides in the competition between heterotrophic bacteria and phytoplankton for DIP utilization. At station ION, P requirements of the autotrophic community were low compared to the initial input of DIP following dust seeding (∼9 nmol P L⁻¹ as compared to an input of +22 to +23.3 nmol P L⁻¹; Gazeau et al., 2021). In contrast, at FAST, the autotrophic community consumed a much larger proportion of the initial DIP input (∼25 nmol P L⁻¹ as compared to an input of 30.8–31.3 nmol P L⁻¹) and phytoplankton appeared as a winner for the utilization of DIP towards the end of the experiment at this station. It seems that heterotrophic bacteria and phytoplankton were more in balance and less stressed at the start of the experiment at FAST; i.e., phytoplankton abundances showed no decrease between t−12 h and t0 and BP did not increase as much as during the other two experiments, suggesting a strong predation pressure (μ_BP was the lowest of the three experiments: ca. 0.02 h⁻¹ in the controls).

The explanation for the observed differential responses of the autotrophic community at the two stations (FAST>ION) is not evident and further complicated by the fact that the sampling strategy differed between the two stations (i.e., no sampling at t72 h, replaced by a sampling at t96 h). It is, however, unlikely that this different sampling strategy was responsible for the different changes in computed integrated autotrophic rates at the two stations. As a maximal increase in nano-eukaryote abundance was observed at t72 h at FAST (followed by a drastic reduction at t96 h; Gazeau et al., 2021), excluding this sampling point in the calculation of autotrophic metabolic rates would most likely have led to an underestimation of these rates rather than an overestimation. Furthermore, a similar partitioning of ¹⁴C-based production rates throughout the two experiments did not provide clear insights into which size group benefited the most at station FAST compared to ION. Two non-exclusive explanations could be proposed: (1) as mentioned above, a less important immediate consumption of DIP by heterotrophic bacteria leading to a higher availability of new DIP for phytoplankton growth at FAST (+31 vs. +22 to +23 nmol L⁻¹ at FAST and ION, respectively; Gazeau et al., 2021) along with (2) the presence of a potentially more active community at the start of the experiment at FAST with a much higher contribution from smaller cells (i.e., pico-eukaryotes, Synechococcus; Gazeau et al., 2021) that are well known to be better competitors for new nutrients (e.g., Moutin et al., 2002) and that were less stressed at the start of the experiments.

During both experiments at ION and FAST, communities switched from net heterotrophy to net autotrophy between 48 and 72 h following dust addition (Fig. 6), leading to a positive integrated NCP at both stations (Fig. 9). This is an important observation since, to the best of our knowledge, the present study constitutes the first investigation of the community metabolism response to dust addition. However, it is important to discuss the timing of such a switch in community metabolism. Since heterotrophic prokaryotes reacted faster than autotrophs to the relief of nutrient (N, P) limitation (i.e., BP already increased by 150 %–500 % at t24 h, while ¹⁴C-based production rates increased only after 48–72 h), NCP was first lower (and negative) in the dust-amended tanks as compared to the controls. Marañón et al. (2010) and Pulido-Villena (2008, 2014) have already reported on a much faster response of the heterotrophic prokaryote community to dust enrichment in the central Atlantic Ocean and Mediterranean Sea, respectively. As DIP concentrations at the completion of their 48 h incubations did not differ from those in the controls, it is unlikely that primary production rates and consequently NCP would have further increased. In contrast, during our experiments, DIP concentrations in dust-amended tanks (D) reached initial levels only after 72 h at TYR and ION and remained far above ambient levels at FAST until the end of the experiment (t96 h). During the PEACETIME cruise, high-frequency sampling of CTD (conductivity, temperature, depth) casts allowed the evolution of biogeochemical properties and fluxes to be followed before and after wet dust deposition that took place in the area around FAST on 3–5 June (Van Wambeke et al., 2020). As in our experiment, a rapid increase in BP was responsible for the observed in situ decline in DIP concentrations in the mixed layer following the rain with no detectable changes in primary production (Van Wambeke et al., 2020). The intensity of the wet deposition event that was simulated during our experiments was, by far, higher, but still representative of a realistic scenario (Bonnet and Guieu, 2006; Loÿe-Pilot and Martin, 1996; Ternon et al., 2010).

The most intriguing result concerning the export of inorganic and organic matter is that these fluxes were maximal at the end of the experiment at TYR in the dust-amended tanks despite the fact that ¹⁴C-based production was relatively low and not enhanced by dust addition. Based on previous studies (Bressac et al., 2014; Louis et al., 2017; Ternon et al., 2010), organic matter export was most likely mainly due to the formation of organic-mineral aggregates triggered by the introduced lithogenic particles (referred thereafter to as POC_litho). Indeed, Louis et al. (2017) showed that such an aggregation process occurs within 1 h after dust deposition. These authors further demonstrated the key role of TEPs as the conversion of dissolved organic matter (DOM) to POC was mediated by TEP formation/aggregation activated by the introduction of dust. As TEP concentrations were only measured on two occasions after seeding, with the first measurement occurring at t24 h, the dynamics of these particles cannot be studied in detail. Nevertheless, it is very likely that the sharp decrease in TEP-C (Fig. 3) between t24 h and t72 h was related to POC_litho export. The coefficient linking POC_litho to Litho_flux (i.e., the mass of sedimented particles) measured here (0.02) is consistent with values reported for other experiments conducted in the Mediterranean Sea (Louis et al., 2017).

Even though ¹⁴C-based production rates were enhanced in the dust-amended tanks at stations ION and FAST, the amount of POC exported at the end of these experiments remained lower than at TYR, with fluxes ∼10–20 $\mathrm{mg}\mathrm{C}{\mathrm{m}}^{-\mathrm{2}}{\mathrm{d}}^{-\mathrm{1}}$.

The recovery of the introduced dust (traced by the lithogenic mass recovered in the traps) was low (27 % at TYR, ∼20 % at ION and 13 %–19 % at FAST) reflecting that a majority of the dust particles (the smaller ones that are the most abundant according to the particle size distribution of the dust) still remained in the tanks after 3 or 4 d following dust addition. This has been already observed in pelagic mesocosms (Bressac et al., 2012) as those small particles can aggregate to organic matter and eventually sink. The higher export efficiency observed ($\text{TYR}>\text{ION}>\text{FAST}$) is likely linked to the higher initial abundance and higher production of TEPs during the experiment (Fig. 3). At TYR, impacted by a strong dust event several days before the experiment started (see above), the likely stimulation of the autotrophs after this in situ event should have been followed by the production of a fresh and abundant DOM, comparable to the “post-bloom situation” in Bressac and Guieu (2013).

4.3 Impact of dust addition under future environmental conditions

Warming and/or acidification had a clear impact on most evaluated stocks and metabolic rates. Gazeau et al. (2021) have already discussed temperature- and pH-mediated changes in nutrient uptake rates and autotrophic community composition in these experiments. Briefly, they showed that warming and acidification did not have any detectable impact on the release of nutrients from atmospheric particles. Furthermore, these external drivers did not drastically modify the composition of the autotrophic assemblage, with all groups benefiting from warmer and acidified conditions. Here, we showed that the difference in the response of plankton community metabolism to dust addition under present and future conditions of temperature and pH was highly dependent on the sampling station (Fig. 9). At all stations, ¹⁴C-based particulate production rates were enhanced under future conditions as compared to those measured under present environmental conditions (treatment D), although this pattern was not observed for ¹³C incorporation into POC at stations ION and FAST. At ION, no differences could be detected and at FAST an even lower ¹³C enrichment was measured at the end of the experiment. These contrasting patterns between ¹⁴C-uptake rates and ¹³C enrichment of POC are likely explained by the fact that the latter covered the whole experimental period (including dark periods) and represents net community carbon production while ¹⁴C-based rates were measured over 8 h incubations in the light, providing an estimate in between gross and net carbon production.

Similarly, the heterotrophic compartment was more stimulated, as BP rates increased strongly at all stations under future conditions compared to treatment D. The relatively smaller increase in CR rates, compared to BP, leading to higher BGE suggests a better utilization of resources by heterotrophic prokaryotes under future environmental conditions. Overall, CR was more impacted than GPP, with the consequence that all integrated NCP rates decreased under future environmental conditions compared to present conditions (treatment D). At station TYR, as discussed previously, dust addition under present conditions did not lead to a switch from net heterotrophy to net autotrophy. This pattern was even more obvious under warmer/acidified conditions, with a larger decrease in integrated NCP at this station. The decrease in integrated NCP at station FAST relative to controls, as well as the smaller increase in all ¹⁴C-based production rates relative to those observed at station ION, must be taken with caution. As already discussed, the fact that for these processes (O₂ metabolism and ¹⁴C incorporation) no samples were taken at FAST at t72 h when maximal cell abundances were recorded for all autotrophic groups (pico- and nano-eukaryotes, autotrophic bacteria) must have artificially led to an underestimation of these integrated metabolic rates. The question of the timing appeared even more preponderant under warmer/acidified conditions, especially at station FAST, where the very important increase in BP led to a full consumption of DIP before t48 h (Gazeau et al., 2021) and drove the community towards a strong heterotrophy. The metabolic balance further switched to a slight autotrophy at t72 h when heterotrophic bacterial activity appeared limited by nutrient availability.

Both elevated partial pressure of CO₂ (pCO₂) and warming are major global change stressors impacting marine communities. Elevated pCO₂ may directly facilitate oceanic primary production through enhanced photosynthesis (Hein and Sand-Jensen, 1997; Riebesell et al., 2007), although the effects appear to be species- and even strain-specific (e.g., Langer et al., 2009). Warming affects organisms by enhancing their metabolic rates (Brown et al., 2004; Gillooly et al., 2001). Although recent studies suggest large differences in temperature sensitivity between phytoplankton taxa (Chen and Laws, 2017) and no significant overall difference between algae and protozoa (Wang et al., 2019), mineralization rates are usually believed to be more impacted by warming than primary production rates, potentially leading to a decline in net oceanic carbon fixation (Boscolo-Galazzo et al., 2018; Garcia-Corral et al., 2017; Lopez-Urrutia and Moran, 2007; Regaudie-de-Gioux and Duarte, 2012) and carbon export efficiency (Cael et al., 2017; Cael and Follows, 2016). Overall, our experimental setup did not allow warming to be discriminated from acidification effects, precluding an evaluation of their potential individual impacts. Nevertheless, we could speculate to which extent a 3 ^∘C warming and a doubling of CO₂ can explain some of the observed differences between D and G (for instance, a 2-fold increase in ¹⁴C-based production rates at ION). For photosynthesis, meta-analysis studies indicate minor effects of pCO₂ on most investigated species (Kroeker et al., 2013; Mackey et al., 2015). Recent studies show a strong, although species-dependent, temperature sensitivity of phytoplankton growth (Chen and Laws, 2017; Wang et al., 2019), suggesting that a 3 ^∘C warming could explain most of the increased carbon fixation in G compared to D. With respect to NCP, our results are in line with the general view and suggest a weakening of the so-called fertilization effect of atmospheric deposition in the coming decades.

In contrast, we did not observe an additional impact of future environmental conditions on the export of organic matter after dust addition as, at each station, this export was of the same order of magnitude for treatments D and G. This result is in agreement with the findings of a similar experiment in coastal Mediterranean waters that considered only pH change (Louis et al., 2017) but stands in contrast with the findings of Müren et al. (2005), who showed a clear decrease in sedimentation following a 5 ^∘C warming in the Baltic Sea. Only a few studies have addressed the combined effect of both temperature and pH changes on aggregation processes and export but none considered dust as the particulate phase. These studies, focused mainly on the formation of TEPs, were inconclusive on the impact of these combined factors (Passow and Carlson, 2012, and references therein). The potential effect of warming and acidification on biogenic carbon export was certainly, over the rather restricted duration of the experiments, insignificant as compared to the large amount of carbon exported through the lithogenic pump. Although a longer experimental period would likely be necessary to clearly support an impact of future conditions on export, those changes occur on a long timescale that cannot be easily mimicked by experimental approaches. Only in situ co-located observations (atmospheric flux and export in sediment traps) over long temporal scales would be necessary to ascertain the interactive effects of these stressors at the decadal timescale.