Climate change is now widely recognized as the major environmental problem facing the world (IPCC, 2013) and is at the heart of human, environmental and economical issues. On a global scale, the oceanic biological carbon pump (BCP) influences climate trends: it consists of the photosynthetic fixation of carbon dioxide (CO2) by oceanic algae (phytoplankton) in the upper illuminated ocean, followed by the downward flux of some of this material mainly due to gravitational settling. The BCP transfers approximately 5–15 GT of carbon (C) from the surface ocean to the oceans' interior every year (Henson et al., 2011).

The efficiency of our oceans to take up excess CO2 largely depends on the availability of fixed nitrogen (N) (Falkowski, 1997) in the surface ocean. In the vast nitrate (NO3-)-limited oligotrophic gyres, which cover ∼ 60 % of the global ocean surface, fixed N is principally provided through the biological fixation of atmospheric dinitrogen (N2) by N2-fixing (or diazotrophic) organisms (Karl et al., 2002). Diazotrophs fix N2 gas dissolved in seawater (the largest reservoir of N on Earth), turning it into ammonium and organic N compounds. On the global scale, they provide the major external source of N for the ocean, surpassing atmospheric and riverine inputs (Gruber, 2004), and thus act as “natural fertilizers”, contributing to sustaining life and the BCP through the so called “N2-primed prokaryotic C pump” (Karl et al., 2003, 2012).

Important progress on the magnitude and the ecological role of marine N2 fixation in biogeochemical cycles has been made by the international oceanographic community over the last 2 decades. This includes the landmark discovery of unicellular diazotrophic organisms of pico- and nanoplanktonic size termed unicellular diazotrophic cyanobacteria (UCYN; see, e.g., Zehr et al., 2001) and new and unexpected ecological niches where diazotrophs are active, such as N-rich oxygen minimum zones (see, e.g., Dekaezemacker et al., 2013; Fernandez et al., 2011). Thus, we have gained a much better understanding of this process. However, a critical question that remains poorly studied is the fate of N newly fixed by diazotrophs (or diazotroph-derived N, hereafter referred to as DDN) in oceanic food webs and its impact on CO2 uptake and export (BCP) (Mulholland, 2007). The VAriability of vertical and tropHIc transfer of diazotroph derived N in the south wEst Pacific (VAHINE) project proposes a scientific contribution to answer these questions, based on a combination of experimentation and modelling involving recently developed innovative techniques. The acronym VAHINE was chosen in order to refer to the Pacific culture where this experiment has been performed with the help of local people. The main scientific research priorities of the project were (i)

to quantify the DDN which enters the planktonic food web,

Below, we focus on the technical challenges and the methods developed to answer the scientific questions of the project. Studying the fate of DDN in the ocean is technically complex. First, it requires appropriate methodologies to trace the passage of DDN through the different components of the planktonic food web. During the VAHINE project, we made intensive use of high-resolution nanometre-scale secondary ion mass spectrometry (nanoSIMS) in combination with flow cytometry cell sorting and 15N2 labelling to trace the passage of 15N-labelled DDN into several groups of non-diazotrophic phytoplankton and bacteria. This technique and results are presented in detail in Bonnet et al. (2016b) and in this special issue (Berthelot et al., 2016; Bonnet et al., 2015) and will not be detailed in this introduction.

Second, carrying out this research requires the monitoring of the chemical, biological and biogeochemical characteristics of a water body affected by a diazotroph bloom for a long period of time (15–30 days) to be able to follow plankton community changes, track the N transfer in the different compartments of the ecosystem (dissolved or particulate phases, small or large plankton, export material) and elaborate biogeochemical budgets. Small-scale laboratory microcosm experiments have been frequently used in ocean biogeochemical studies, but their limited realism can make extrapolations to natural systems difficult to justify. They limit the duration of experiments to a few days (usually 24 to 72 h); the small volumes used (a few litres maximum) limit the number of parameters measured and they do not include export terms. To overcome these difficulties, we decided to use the technology of large-volume mesocosms. Mesocosms are now widely used in ecological studies (Riebesell et al., 2013; Stewart et al., 2013) and enable the isolation of water masses of several cubic metres from physical dispersion for several weeks, without disturbing temperature and light conditions, taking into account the biological complexity of the planktonic ecosystem; they thus provide a powerful approach to maintain natural planktonic communities under close-to-natural self-sustaining conditions for several weeks. Moreover, the responses obtained from mesocosm studies (isolated from hydrodynamics) provide useful parameterizations for ecosystem and biogeochemical models.

Initial hydrological and biogeochemical conditions (i.e. conditions in ambient waters the day of mesocosm deployment – 13 January, day 0) are summarized in Table 2. Seawater temperature was 25.30 ∘C, which is slightly lower than the temperature reported in this season at the Amédée lighthouse station, while salinity (35.15) was typical for the season (Le Borgne et al., 2010). NO3- and DIP concentrations were both reported to be 0.04 ± 0.01 µmol L-1, and Chl a concentrations from fluorescence data (0.11 µg L-1) were typical of oligotrophic systems and in the range reported in the literature for this location (Fichez et al., 2010). DON and dissolved organic P (DOP) concentrations were 4.65 ± 0.46 and 0.100 ± 0.002, and ambient N2 fixation rates were 8.70 ± 1.70 nmol N L-1 d-1 before the mesocosm deployment.

Seawater temperature measured daily by vertical CTD profiles inside the mesocosms and in the lagoon waters (Fig. 4a–d) gradually increased over the 23 days of the experiment from 25.50 ∘C the day of the mesocosm closure (day 2) to 26.24 ∘C on day 23. This warming is the typical trend observed in New Caledonia in the course of the summer season (Le Borgne et al., 2010). The water column was vertically homogeneous over the course of the experiment, except on the two first days, which were characterized by a slight stratification inside and outside the mesocosms. Data indicate therefore a good reproducibility between the three mesocosms and between the mesocosms and the Noumea lagoon waters.

Salinity data (Fig. 4e–h) indicate a small and gradual increase in the three mesocosms during the 23-day experiment (35.2 to 35.4), suggesting a probably higher level of evaporation in the mesocosms compared to the Noumea lagoon. Moreover, lagoon waters constantly receive some low-salinity waters from the coast due to rainfall advected by tide currents, which may also explain the slightly lower salinity values measured in the Noumea lagoon (35.40) compared to inside (35.47) at the end of the experiment.

NO3- concentrations (Fig. 5a–d) remained below 0.1 µmol L-1 during the whole experiment in all mesocosms and in the lagoon waters. Average concentrations over the 23-day experiment and the three depths samples were close to detection limits of the method (0.01 µmol L-1) and are thus difficult to quantify accurately: they were 0.04 ± 0.02, 0.02 ± 0.01, 0.02 ± 0.02 and 0.06 ± 0.04 µmol L-1 in M1, M2, M3 and in the lagoon waters, respectively. DIP concentrations (Fig. 5e–h) were also close to detection limits (0.005 µmol L-1) and on average 0.04 ± 0.01, 0.03 ± 0.01 and 0.03 ± 0.02 µmol L-1 before the DIP fertilization (days 2 to 4, hereafter called P0) in M1, M2 and M3 (average over the three depths). They increased after the fertilization on day 5 to 0.73 ± 0.07, 0.98 ± 0.01 and 0.77 ± 0.03 µmol L-1 in M1, M2 and M3. The intensity of the DIP fertilization differed slightly among the mesocosms, likely reflecting the different volume of the mesocosms (see above). Subsequently DIP concentrations decreased steadily towards initial concentrations by the end of the experiment: 0.03 ± 0.01, 0.03 ± 0.01 and 0.05 ± 0.02 µmol L-1 in M1, M2 and M3, respectively (average of 23 days over the three depths). However, the DIP pool was first exhausted in M1 (day 14), then M2 (day 19) and finally M3 (day 23). A more detailed description of the evolution of stocks and fluxes of biogenic elements during the experiment can be found in Berthelot et al. (2015).

Chl a fluorescence was homogeneous throughout the water column during the course of the experiment (Fig. 4i–l). Chl a slightly increased (by 0.1 to 0.2 µg L-1) in the three mesocosms after DIP fertilization on days 5 and 6. After day 6, it consistently declined back to the initial (before fertilization) concentrations of 0.12–0.15 µg L-1. On days 12, 13 and 14, Chl a concentrations re-increased dramatically to reach 0.61, 0.65 and 1.02 µg L-1 in M1, M2 and M3 at day 23, respectively, indicating that the three mesocosms were relatively synchronized but the intensity of the phytoplankton bloom differed between the mesocosms, with a greater increase observed in M3 compared to M2 and M1. In the lagoon waters, Chl a concentrations also gradually increased over the experiment (concentrations reached 0.35 µg L-1 at day 23) but to a lower extent compared to that of the mesocosms.

The goal of this special issue is to present the knowledge gained regarding the fate of DDN in an LNLC ecosystem based on the large data set acquired during the VAHINE mesocosm experiment. VAHINE was a multidisciplinary project involving close collaborations between biogeochemists, molecular ecologist, chemists, marine opticians and modellers. Most of the contributions to this special issue have benefited from this collective and collaborative effort. The philosophies and summarized results of the different papers composing the special issue are presented briefly hereafter, and a synthesis paper of all the multidisciplinary approaches used to answer the main scientific questions of the VAHINE project is given in the synthesis paper Bonnet et al. (2016a).

First, thanks to the high-frequency (daily) sampling of the same water body for 23 days, this project provided a unique opportunity to characterize the diversity of the planktonic assemblage using several complementary approaches and to investigate species successions in relation to hydrological parameters, biogeochemical stocks and fluxes during a diazotroph bloom in an LNLC ecosystem. By using polymerase chain reaction (PCR), which targeted a component of the nitrogenase gene (nifH), sequencing and qPCR assays, Turk-Kubo et al. (2015) fully characterized the diazotroph community composition within the mesocosms and the New Caledonian (Noumea) lagoon and calculated in situ growth and mortality rates for natural populations of diazotrophs, which is rarely accomplished. They revealed that the diazotroph community was dominated by diatom–diazotroph associations (DDAs) during the first period of the experiment after the DIP fertilization (days 5 to 14, hereafter called P1) and that a bloom of UCYN-C occurred during the second half (days 15 to 23, hereafter called P2), providing an unique opportunity to compare the DDN transfer and export efficiency associated with different diazotrophs. Complementary to this approach, Pfreundt et al. (2016b) used 16S tag sequencing to examine the temporal dynamics of the prokaryotic community and observed clear successions of prokaryotes during the experiment in relation to biogeochemical parameters. In a second study, Pfreundt et al. (2016a) also used metatranscriptomics to investigate the microbial gene expression dynamics from diazotrophic and non-diazotrophic taxa and highlighted specific patterns in the expression of genes involved in N, DIP, iron and light utilization along the different phases of the experiment. Van Wambeke et al. (2015) revealed that heterotrophic bacterioplankton production and alkaline phosphatase activity were statistically higher during P2, concomitant with the UCYN-C bloom. Their results suggest that most of the DDN reached the heterotrophic bacterial community through indirect processes, like mortality, lysis and grazing. In parallel, Leblanc et al. (2016) focused on the phytoplankton assemblages and dynamics from pigment signatures, flow cytometry and taxonomy analyses and revealed a monospecific bloom of the diatom Cylindrotheca closterium and an 2-fold increase in Synechococcus and nanophytoeukaryotes during P2.

Tedetti et al. (2015) used bio-optical techniques to describe the spectral characteristics and the variability of dissolved and particulate chromophoric materials according to the phytoplankton community composition and revealed a coupling between the dynamics of the N2 fixation and that of chromophoric material in the south-west Pacific. Berman-Frank et al. (2016) analysed the spatial and temporal dynamics of transparent exopolymeric particles (TEPs), which are sticky carbon-rich compounds that are formed, degraded and utilized in both biotic and abiotic processes, and measured a relatively stable TEP pool available as both a carbon source for plankton communities and facilitating aggregation and flux throughout the experiment.

The second point to be illustrated is that the bloom of diazotrophs (UCYN-C) obtained in the closed water body of the mesocosms following DIP fertilization offered the opportunity to track the fate of DDN in the ecosystem: Berthelot et al. (2015) describe the evolution of C, N and P pools and fluxes during the course of the experiment and report a 3-fold increase in Chl a concentrations and N2 fixation rates and a 5-fold increase in C export during the second half of the experiment (UCYN-C bloom). They also reveal that the e ratio that quantifies the efficiency of a system to export particulate organic C was significantly higher (p<0.05) during P2 than during P1, indicating that the production sustained by UCYN-C was more efficient at promoting C export than the production sustained by DDAs. Complementary to this approach, Knapp et al. (2015) reported the results of δ15N measurements on DON, particulate organic N (PON) and particles from sediment traps and further substantiated these results with a significantly (p<0.05) higher contribution of N2 fixation to export production during P2 (56 ± 24 % and up to 80 % at the end of the experiment) compared to P1 (47 ± 6 %). Bonnet et al. (2015) explored the fate of DDN on shorter timescales and revealed that ∼ 10 % of UCYN-C from the water column were exported daily to the traps, representing as much as 22.4 ± 5.5 % of the total particulate organic C (POC) exported at the height of the UCYN-C bloom. This export was mainly due to the aggregation of small (5.7 ± 0.8 µm) UCYN-C cells into large (100–500 µm) aggregates. Using a nanoSIMS approach, they also showed that 21 ± 4 % of the DDN was transferred to non-diazotrophic plankton, mainly picoplankton (18 ± 4 %) followed by diatoms (3 ± 2 %) during P2. The same nanoSIMS approach was used by Berthelot et al. (2016) in a parallel experimental study to compare the DDN transfer efficiency into non-diazotrophic plankton, whether it comes from UCYN-C, UCYN-B or Trichodesmium. They showed that the transfer was twice as high during a Trichodesmium bloom than during a UCYN-B or UCYN-C bloom, arguing that filamentous diazotroph blooms are more efficient at promoting non-diazotrophic production in N-depleted areas. In parallel, Hunt et al. (2016) estimated a mean ∼ 30 % contribution of DDN to zooplankton biomass in the mesocosms based on natural 15N isotope measurements on zooplankton. They also provided evidence for direct ingestion and assimilation of UCYN-C-derived N by the zooplankton, results that were complemented by qPCR assays on several diazotroph phylotypes in zooplankton guts. Spungin et al. (2016) took advantage of the Trichodesmium bloom occurring outside the mesocosms to specifically investigate its decline and understand changes in genetic underpinning and features that could elucidate varying stressors or causes of mortality of Trichodesmium in the natural environment.

The third point to be highlighted is that modelling was used at every stage of the project. Simulations performed with the 1D vertical biogeochemical mechanistic Eco3M-MED model were used prior to the VAHINE experiment to help in the scientific implementation of the project (timing and quantification of the DIP fertilization). Gimenez et al. (2016) validated the model using the in situ data measured during the whole experiment and provided additional information such as stoichiometry of planktonic organisms that could not be inferred from in situ measurements and offered the opportunity to deconvolute the different interlinked biogeochemical processes occurring in the ecosystem to help understand the fate of DDN in oligotrophic ecosystems and the impact of N2 fixation on carbon export.

Finally, a synthesis study by Bonnet et al. (2016a) attempts to summarize our knowledge and the unresolved questions regarding the fate of DDN in the ocean to synthetize and link the major experimental and modelling results obtained during the project and described in the VAHINE special issue. It reconciles the diverse and complementary methodological approaches used in this study to answer the scientific questions of the VAHINE project. After putting the different experimental findings in perspective, the modelling approach has also been used in the synthesis article as a tool to investigate the impact of N2 fixation on marine productivity, export and food web composition by artificially removing N2 fixation in the model.