Evolution of the riverine nutrient export to the Tropical Atlantic over the last 15 years: is there a link with Sargassum proliferation?

The Tropical Atlantic is facing a massive proliferation of Sargassum since 2011, with severe environmental and socioeconomic impacts. As a contribution to this proliferation, an increase in nutrient inputs from the tropical rivers, in response to climate and land use changes or increasing urbanization, has been often suggested and widely reported in the scientific and public literature. Here we discuss whether changes in river nutrient inputs could contribute to Sargassum proliferation in the recent years or drive its seasonal cycle. Using long-term in situ and satellite measurements of discharge, dissolved and particulate nutrients of the three world largest rivers (Amazon, Orinoco, Congo), we do not find clear evidences that nutrient fluxes may have massively increased over the last 15 years. Moreover, focusing on year 2017, we estimate that along the year only 10% of the Sargassum biomass occurred in regions under river plume influence. While deforestation and pollution are a reality of great concern, our results corroborate recent findings that hydrological changes are not the first order drivers of Sargassum proliferation. Besides, satellite observations suggest that the major Atlantic river plumes suffered a decrease of phytoplankton biomass in the last two decades. Reconciling these observations requires a better understanding of the nutrient sources that sustain Sargassum and phytoplankton growth in the region.


Context
Before 2010, holopelagic Sargassum spp. were preferentially found in the Sargasso Sea and in the Gulf of Mexico. They now develop in large quantities on the southern part of the North Atlantic between 0 • N and 10 • N forming a 'Sargassum belt' stranding in millions of tons on the coasts of the Lesser Antilles, Central America, Brazil and West Africa. (e.g. Smetacek and Zingone 2013, Wang and Hu 2016, Langin 2018, Wang et al 2019. Satellite imagery pointed to the presence of large amounts of Sargassum in areas under seasonal influence of the Amazon plume (Gower et al 2013, Sissini et al 2017, Oviatt et al 2019, Wang et al 2019 raising the hypothesis that river nutrient fluxes might play a role in this proliferation (Langin 2018, Oviatt et al 2019, Wang et al 2019. A possible influence of the Congo has also been invoked in several studies (Djakouré et al 2017, Oviatt et al 2019. A recent study by Johns et al (2020), however, did not find strong evidence to support this hypothesis as there appears to be a spatiotemporal mismatch between Sargassum occurrence and these riverine sources of nutrients. Given the importance of this question and the present discrepancies in the scientific literature we find it important to examine to which extent the riverine source of nutrients may contribute to the proliferation of pelagic Sargassum. Indeed, several elements give support to a possible influence of the riverine sources of nutrients. First, rivers export nitrogen (N) and phosphorus (P), which are key limiting nutrients required for Sargassum growth (Lapointe 1986(Lapointe , 1995. Specifically, the Amazon also contains important concentrations of dissolved organic substrates that could be an important source of nutrient for Sargassum growth as reviewed in Oviatt et al (2019). Second, the Tropical Atlantic receives the fresh and nutrient rich waters of the three largest rivers on the planet-in terms of flow (Amazon, 209 000 m 3 s −1 , Congo, 42 000 m 3 s −1 and Orinoco, 35 000 m 3 s −1 ), which alone represent 21% of the total global riverine flow (Milliman and Farnsworth 2011). Their low-saline and productive plumes extend thousands of kilometers far offshore (Muller-Karger et al 1988, Signorini et al 1999. Third, the watersheds undergo strong climatic and anthropogenic pressures that are thought to have the potential to modify oceanic biogeochemical systems. For instance, Seitzinger et al (2010) estimated that the total river input of N to the coastal seas has approximately doubled since the 70 s, with South America representing ∼20% of the global increase. The Amazon basin already shows some signs of a transition to a disturbance-dominated regime in response to agricultural expansion and climate variability (Davidson et al 2012). The region experiences a strong anthropogenic pressure associated with a rapid urbanization (Richards and Vanwey 2015), intense hydropower dam construction (Latrubesse et al 2017), and increase of mining and oil extraction contamination (e.g. Moquet et al 2014). The overall consequences of these changes in terms of nutrient budget remain uncertain since they can act as a source or a sink of nutrients.
In this context, the long-term evolution of the continental nutrient export to the Tropical Atlantic is investigated on the basis of in situ observations of the major dissolved and particulate nutrients exported by the three main rivers of the basin (Amazon, Orinoco and Congo). Satellite estimates of chlorophyll provide an independent set of observations to monitor the long-term changes of biological activity in the large river plumes. Finally, the large-scale seasonal distribution of Sargassum for year 2017 is confronted to numerical experiments of river plume dispersal. We focused on this year because basin scale Sargassum fractional coverage observations from MODIS were available (Berline et al 2020), with concurrent observations carried out during two cruises in the Tropical

River nutrient fluxes
The productivity of the Sargassum is enhanced by N and P availability (Lapointe 1995). At global scale, the rivers carry N to oceanic coastal zone in dissolved and particulate forms in almost equal proportion (Joo et al 2013) while P is mainly exported as particulate form (90%-95% of the total P flux to the ocean; Ruttenberg 2004). About 25%-45% of the particulate P (Ruttenberg 2004) and a significant proportion of particulate N are reactive in the sea water and bioavailable for marine organisms including the seaweed. The dissolved and particulate N and P fluxes measured or estimated at the seaward-most stations for the Amazon, Orinoco, and Congo basins are shown in figure 1 for the last two decades. These data were collected by the SO-HYBAM observatory and are presented together with riverine flux calculation methods in the supplementary material.
For the three rivers, the largest input of N is provided by the dissolved organic matter. Dissolved organic nitrate delivered by the Amazon is thought to become bioavailable in the offshore fraction of the plume through bacterial and photochemical transformations (Medeiros et al 2015). For the Amazon, this flux appears to regularly increase from 2004, apart from maxima in years 2007 and 2008. Observations for the Orinoco suggest a doubling of this flux over the last 15 years (figure 1). The particulate fluxes of N, estimated from remote sensing, is also expected to contribute to nutrient supply through desorption of the shelf (Demaster and Aller 2000). It is stable for the three rivers. The dissolved inorganic N flux, computed from NO 3 − in situ measurements, show larger values during the last decade for the three rivers (figures 1(a)-(c)). Before 2013, values above the detection limit (0.01 mg l −1 ) were of similar magnitude than independent Amazon (Richey et al 2009, Doherty et al 2017, Orinoco (Lewis and Saunders 1989) and Congo (Descy et al 2017) water analyses. They did not show a marked evolution over this period. From the years 2013-2014, the average concentration of NO 3 − has increased for the three rivers. On the one hand, the scatter of the measured concentrations is so high that it is difficult to determine how significant the NO 3 − increase really is. On the other hand, the more frequent recording of high NO 3 − fluxes is of concern and suggests a potential evolution of the dissolved NO 3 − export that needs to be investigated. However, it should be noted that the marked changes in terms of NO 3 − for the different rivers occurred 2-3 years after the first massive proliferation of 2011.
In the Amazon river, the largest amount of P is delivered in particulate form (figure 1(g)). The importance of the particulate P is in line with observations by Berner and Rao (1994) who conclude that the solubilization of P from bacterial decomposition of river-transported organic matter and desorption from ferric oxide/hydroxide may result in an effective flux of reactive P about three times greater than that carried only in dissolved form. This particulate flux shows a slight decrease over the last two decades, while the inorganic and organic dissolved fluxes remained stable. The P fluxes for the Orinoco and Congo are one order of magnitude smaller than those of the Amazon.
So, observations show different long-term trends of inorganic, organic and particulate fluxes of N and P. No direct and clear relationship with Sargassum growth can be drawn, neither in terms of long-term evolution, nor in terms of interannual variability (e.g.

Link with changes in plume productivity and Sargassum distribution
The diversity of the nutrient trends and the lack of knowledge on the lability of the dissolved and particulate riverine material render uncertain the assessment of the long-term evolution of the riverine fertilization of the ocean. As an independent marker of possible changes in the nutrient export by the large Tropical Atlantic rivers, the long-term evolution of surface chlorophyll estimated from satellite ocean color is now analyzed. Chlorophyll is the main pigment in phytoplankton and here we use chlorophyll as a proxy of phytoplankton biomass. As it has been evidenced for the Mississippi in the northern Gulf of Mexico (Lohrenz et al 1997, Rabalais et al 2002, Wysocki et al 2006, we expect that fluctuations in riverine nutrients alter the dynamics of phytoplankton growth and thus phytoplankton biomass in the large tropical river plumes. The difference between the 'Sargassum period ' (2011-2018) and the years before (2003-2010) reveals an overall decrease of the chlorophyll concentration in the Tropical Atlantic ( figure 2(b)). This decline is sharper in the Amazon, Orinoco, and Congo plume regions. Since Chlorophyll retrieval from space is subject to large discrepancies between the different available products, we compared five monthly chlorophyll products from three different groups (GlobColour, NOAA, and CCI). For the three rivers considered, four out of the five different products show a consistent decrease of chlorophyll concentration in the plume areas The basin scale decrease of chlorophyll evidenced in figure 2(b) is in line with the study by Gregg and Rousseau (2019) that suggested that global net ocean primary production has experienced a small but significant decline in the 18 year satellite records from 1998 to 2015, in response to shallowing surface mixed layer depth, decreasing nitrate supply and changes in the phytoplankton communities. Chlorophyll concentrations in river plumes exhibit a larger decrease. The underlying cause of these changes in the chlorophyll content of the plumes is difficult to ascertain from observations only. It is worth mentioning that (a) colored detrital material contributes to total light attenuation in the blue region of the spectrum where chlorophyll-a also absorbs strongly (Fournier et al 2015) which could lead to large errors in ocean color retrievals, (b) the response of the productive plumes may not only depend on the riverine nutrient flux but on other variables such as temperature, stratification, turbidity, or dust deposition. But this decrease, whether it is caused by a decrease of plume productivity or weaker discharge of dissolved colored material (which is not observed in SO-HYBAM observations of organic and particulate nutrient fluxes, figure 1) is difficult to reconcile with the hypothesis of an overall increase in fertilization by tropical rivers in recent years. A better understanding of the river plume biogeochemistry is required, together with analysis of possible competing growth dynamics between phytoplankton and Sargassum.
The seasonal distribution of Sargassum for year 2017 is shown in figure 3 together with the chlorophyll concentrations. The Sargassum bloom during the first 6 months of the year occurs preferentially in the Intertropical Convergence Zone (ITCZ; located between the equator and 10 • N), where chlorophyll is relatively high compared to the surrounding subtropical oligotrophic area. To our knowledge, the causes of the high chlorophyll level have not been identified, but could be the result of diatomsdiazotroph assemblages (Subramaniam et al 2008, Schlosser et al 2014, atmospheric deposition of dust (Yu et al 2015), or biomass burning emissions (Barkley et al 2019). Yet, the presence of relatively high chlorophyll concentration indicates nutrient availability that may participate to sustain Sargassum growth.
Interestingly, we remark that during September-October, when the North Brazil Current retroflects and transports the Amazon riverine freshwater to the east, the abundance of Sargassum in the plume area between 60 • W and 40 • W, is drastically reduced relative to the two previous months. The North Brazil Current is mainly fed by waters originating from the equatorial area and the southern Tropical Atlantic (Johns et al 1998) where no massive proliferation of Sargassum was observed in the previous months. Our interpretation is that the weak abundance of Sargassum in the plume at this time is mainly controlled by advection of low Sargassum water in the region. The low salinity of the plume could also limit the proliferation of Sargassum there. Indeed, culture experiments of Sargassum natans and Sargassum fluitans described in Hanisak and Samuel (1987) revealed some dependence of their growth rate to salinity. A reduction in salinity from 36 to 30 caused a reduction in the growth rates by almost half, and no growth was observed for salinity below 18. This effect may likely limit the fertilizing effect of the nutrient rich river plumes. The river plume dispersion numerical experiment (figure 3) also reveals that the central Atlantic is not under the influence of the Amazon plume during the first half of the year. The largest coincidence between the plume and Sargassum distribution occurs in June-July-August (figure 4), when the Amazon plume extends toward the Lesser Antilles. This is in line with the analysis by Gouveia et al (2019) that showed that the Amazon plume fingerprints on oceanic primary productivity spatio-temporal variability are restricted to the western Tropical Atlantic. The first 6 months of the year appear to be crucial for the occurrence of Sargassum along the south American and Caribbean coasts a few months later (Wang and Hu 2017, Putman et al 2018, Wang et al 2019, Berline et al 2020. Even if Amazon river fertilization could contribute to the seasonal growth in the portion of western Tropical Atlantic under seasonal influence of the Amazon plume (an area between 60 • W and 40 • W and between 0 • N and 20 • N), this analysis further suggests that it does not drive the largescale seasonal bloom. At the annual scale, we found that only 9% of the Sargassum biomass occurred in the river plume area in 2017, with occurrence below 5% from September to May and peak at 23% in July when the plume is well extended toward the Lesser Antilles. It is even more unlikely that the Congo and Orinoco rivers could contribute to the large-scale bloom due to the limited imprint of the plumes on the chlorophyll distribution and remoteness of the river plumes from the main Sargassum bloom areas.
As a conclusion, while increasing inputs of N and P in the watershed from human activity, predominantly from land-based activities, are thought to have the potential to significantly increase the nutrient fluxes toward the ocean and have been proposed as contributors of the Sargassum proliferation, this analysis suggests that riverine fertilization is unlikely a key controlling factor of both seasonal and interannual variability of the Sargassum biomass. In agreement with recent findings by Johns et al (2020), it fails to explain the Sargassum distribution shift that occurred after 2010. Instead, Johns et al (2020) proposed that an extreme negative phase of the North Atlantic Oscillation triggered the 2011 event and that vertical mixing dynamics below the ITCZ sustains Sargassum growth in the Central Tropical Atlantic. This is in line with the enhanced chlorophyll concentrations observed below the ITCZ (figure 4). However, the forcing processes sustaining the productivity there remain to be clarified. This study also reminds us that advection is instrumental in controlling the seasonal distribution of Sargassum, as already revealed by several studies (Brooks et al 2018, Wang et al 2019, Berline et al 2020. Although much progress has been made recently on how Sargassum advection responds to currents and winds (Berline et al 2020, Miron et al 2020, this issue has yet to be fully evaluated and understood. That key aspects of growth and movement are missing from for our ability to understand and forecast spatiotemporal variability in the distribution of pelagic Sargassum.

Methods
Methods and associated references are available in the supplementary material.

Data availability statement
The data that support the findings of this study are openly available at the following URL/DOI: https://hybam.obs-mip.fr.