A new numerical technique for tracking chemical species in a multisource, coastal ecosystem applied to nitrogen causing Ulva blooms in the Bay of Brest (France)

A new numerical technique is presented that allows the tracking of any chemical element from any source in a simulated foodweb, for instance assessing the proportion of these sources (river loadings, sea entrances, point sources) in the algal diet for the limiting nutrient. An application is shown for nitrogen in an Ulva bloom occurring in a shallow embayment connected with a strongly tidally stirred ecosystem with various sources of inorganic nitrogen, the Bay of Brest, Brittany, France. In a first step, a biogeochemical three-dimensional model was developed to simulate growth and erosion–transport–deposition of free-floating ulvae; this model was able to converge on a realistic distribution of Ulva deposits after a few months, even though it was initialized with a strongly unrealistic distribution of settled ulvae. In a second step, and unfortunately for recovery plans, the tracking technique, applied in this model to all the nitrogen sources entering the bay, revealed that the small, nitrate-polluted rivers flowing directly into the eutrophicated area had a negligible effect, whereas more distant but stronger sources, a big river and a big urban sewage plant, even after dilution, accounted for about 50% and 20%, respectively, of the algal nitrogen content during summer. Despite its high N flux, open ocean contributes only 15% to Ulva growth. The suppression of only one of the main nitrogen sources would not significantly decrease the Ulva bloom, because of the high nitrogen surplus present in the site. The remaining sources would still saturate the needs of the maximum Ulva biomass the site is able to produce. The tracking technique, however, shows that the N turnover in Ulva is only 4 months. Thus, improvements would occur within a year following large N reductions. Eutrophication has been of increasing concern for a while, first in lakes and rivers, and then in the coastal zone. Many studies have been done to point out which nutrient was the limiting or the controlling one in various eutrophicated water bodies: whereas phosphorus has been recognized for a long time as the limiting factor in most polluted freshwater systems (Vollenweider 1968; Schindler 1975), a more contrasted situation arose in coastal marine environments. Phytoplanktonic massive blooms in the plumes of main rivers are rather phosphorusor silicon-limited near shore, but nitrogen-limited off shore, especially in summer. The continuously changing N : P : Si ratio in enriched coastal waters may 1 Corresponding author ([email protected]). Acknowledgments We thank our IFREMER’s colleague Jacky L’Yavanc, who performed the survey for fine-scale bathymetry of the area of interest, and the municipal analysis laboratory of Brest (‘‘Pôle Analytique des Eaux,’’ belonging to the European Economic Interest Group ‘‘Littoralis,’’ specializing in integrated Coastal-Zone Management) for all the nutrient concentrations measured in tributaries. We also thank the CEVA laboratory for providing the map of its diver-operated survey of submersed Ulva deposits, and the so-called ‘‘SOMLIT-Iroise’’ team of the French Monitoring Service of the INSUCNRS, who kindly provided the 1999 time-series of hydrological data at the Sainte-Anne station. The manuscript was substantially improved thanks to remarks provided by the two reviewers and the editor in chief. This study has been granted by the Brest municipality (Brittany, France). drastically change the conclusions about the controlling factor in a few years (Guillaud et al. 2000). Green macrophyte blooms, on the contrary, were unanimously considered as being always nitrogen controlled. This form of eutrophication is typical of coastal, very shallow areas that have been heavily enriched in inorganic nitrogen, mainly of land runoff origin (nitrate), but sometimes of urban sewage origin (ammonia or nitrate). Both for phytoplankton and macrophytes, when the limiting factor has been well established, the question often arises about the main source to be diminished: operational restoring plans need to identify the most important target, and to know to what extent the nutrient load has to be reduced. Direct field experiments with chemical tracers are unfeasible in many cases, because of the nonexistence of enough isotopes of the element under study (a n-source problem requires at least n 2 1 isotopes), or the absence of sufficient discrepancy between the natural isotopic signatures of the various sources. Only numerical models can evaluate the effect of each source on the global system, but, up to now, they have been used to test modified situations, e.g., partial or total removal of a nutrient source. This paper shows how to track separately, in the whole, undisturbed, simulated food web, the limiting element coming from a specific input in a multisource context, and to assess the precise proportion of the different existing sources in the actual feeding of proliferating algae. Application of the technique has been done in the case of an Ulva mass bloom arising in a small embayment communicating with a highly mixed tidal ecosystem with many sources of inorganic nitrogen. 592 Ménesguen et al. Fig. 1. Map of the Bay of Brest, with location of the main tributaries and of the marine stations Iroise (I) and Sainte-Anne (S.A). Table 1. Mean annual flow rates of the tributaries and their mean annual ammonia, nitrate, phosphate, and silicate concentrations for year 1999. Flow rate (m3 s21) [NH4] (mmol L21) [NO3] (mmol L21) [PO4] (mmol L21) [Si(OH)4] (mmol L21) Elorn river Aulne river Penfeld river Stang Alar river Costour river 8.1 33.7 0.7 0.1 0.1 6.6 3.2 2.5 5 5 646.6 412.8 631.5 514.2 714.2 2.7 1.3 2.7 2.8 2.8 141.5 113.9 231.5 140 140 Daoulas, Hopital-Camfrout & Faou rivers ;1.7 each 3.6 398.4 1.1 113.9 Zone Industrielle Portuaire (ZIP) & MaisonBlanche sewage plants 0.3 each 1890 56.8 297 140 Iroise Sea 11,000 (neap tide) 33,000 (spring tide) 0.4 4.4 0.2 2.9 Materials and methods Site description—The Bay of Brest, a rather small (180 km2) but productive region of freshwater influence, is situated at the western end of Brittany, France (Fig. 1). It has been intensively studied during the last 30 yr, partly because it represents in a certain sense an extreme estuarine situation, with very high nutrient loadings flowing into a very dispersive macrotidal marine ecosystem. The two main tributaries (Aulne and Elorn Rivers) drain a 2,800 km2-wide watershed, supporting a rather scattered population (360,000 inhabitants) but a highly intensive agricultural activity, mostly cattle breeding. As a result, Aulne and Elorn Rivers exhibit very high nitrate-specific fluxes (i.e., per watershed surface unit), which can be considered as maximum values in the whole North Atlantic region: Aurousseau (2001) gives for the Elorn River during the exceptionally rainy year 2000 a maximum annual specific flux of 9,500 kg km22 yr21 N; and a mean annual specific flux during recent years of 7,200 kg km22 yr21 N for the Elorn River and 5,000 kg km22 yr21 N for the Aulne river. These values are very high compared to the 1,450 kg km22 yr21 N value for the Seine 1 Rhine 1 Elbe ensemble and the 600 kg km22 yr21 N overall North Atlantic mean computed by Howarth et al. (1996). At the end of the 20th century, the corresponding nitrogen loadings amounted to about 2.7 3 106 kg yr21 N for the Elorn River and to 8.6 3 106 kg yr21 N for the Aulne River. Apart from these two main rivers, the Bay of Brest receives six small other rivers, and two main urban sewage plants; all these tributaries are located on Fig. 1, and their annual mean flow rates as well as their mean nitrate, ammonia, and phosphate concentrations during the year 1999 are given in Table 1. Comparatively, and because of the prevailing westerlies coming directly from the North Atlantic Ocean, atmospheric nitrogen inputs are very low in the Bay of Brest, about 80 3 103 kg yr21 (Souchu 1986); they will be neglected in this study. These high nitrogen loadings into a semienclosed bay could be thought to induce a critical eutrophicated status, which is really not the case today. This paradoxical situation has been described and explained by Le Pape et al. (1996) and Le Pape and Ménesguen (1997) as a consequence of the very intense vertical and horizontal mixing by tides (mean amplitude 4 m, spring tide amplitude 7.5 m), which disperse nutrients and growing phytoplankton into a wide and locally deep (up to 40 m) water volume, regularly exchanging with the open sea at its western hydrodynamical margin (the socalled Iroise Sea on Fig. 1). Nevertheless, apart from the general good health status of the central Bay of Brest, some peripheral and confined eutrophication phenomena have been recorded for several years, which are restricted to estuaries proper (e.g., Prorocentrum micans red-brown waters in summer at the mouth of Elorn River) or to embayments somewhat disconnected from the central intense tidal circulation. The most typical one is a ‘‘green tide,’’ made of free-floating Ulva rotundata, which grow and settle in the shallow Moulin Blanc cove, containing both the Brest beach and yachting port (Fig 2). Annual statistics from the Brest public utilities, which give the annual volume of ulvae collected by bulldozers on the beach, reveal a rather stable biomass production of the site: from 1989 until 1999, the annually collected volume fluctuated between 635 m3 and 1,694 m3, with a mean of 1,048 m3. To reduce this nuisance, the Brest municipality launched in 1999 a targeted action aiming at making an inventory of all the potential causes of 593 Coastal eutrophication modeling Fig. 2. Map of submersed benthic deposits of free ulvae, in June 2000 (from CEVA 2000). Fig. 3. Rectangular nonuniform grid used for the model (inset: in gray, area retained for summation of Ulva deposits in the Moulin Blanc cove). Fig. 4. Splitting the water column into layers referenced to the zero hydrographic level. this Ulva bloom, with a particular interest in identifying the main nutrient sources feeding the algae. In June 2000, a diver-operated survey was conducted by the CEVA lab (CEVA 2000), providing the map of submersed bottom deposits of ulvae in the whole Elorn estuarine area (Fig. 2). The work detailed hereafter consisted in building a mathematical model of algal primary production in the Bay of Brest, able to explain in a robust fashion the actual location and biomass of this very restricted algal mass bloom, and to assess the respective roles of the various tributaries in this eutrophication process. Model equations—The three-dimensional (3D) ecological model is based on the phytoplankton model previously applied to the Bay of Seine area and described elsewhere (Cugier et al. 2005), adapted to the Bay of Brest problem by adding Ulva equations modified from Ménesguen and Salomon (1988) and Ménesguen (1992). Model of the physical environment—The hydrodynamical context is furnished by SiAM3D model (Cugier and Le Hir 2002), a 3D hydrodynamic model that solves the so-called ‘‘shallow water’’ equations, using a finite difference technique on a nonuniform rectangular horizontal computational grid. As shown in Fig. 3, this allows one to define small meshes in an area of interest, for instance here the Elorn estuary and its eutrophicated cove (mesh size 5 150 m), while keeping large meshes in far regions (mesh size 5 1,000 m at the western entrance of the Bay). The model provides water surface elevation and velocities in the three space directions at each node of the grid, and solves an advection–dispersion equation for temperature (XT), salinity (XS), inorganic suspended matter (XSM), and, more generally, any dissolved or particulate variable. The model uses real depth coordinates on the vertical axis, which is split into a maximum of 10 layers of given thickness (2 m for the four layers above the zero hydrographic level, 3 m, 3 m, 4 m, 5 m, 10 m, and up to 20 m beneath) as explained in Fig. 4. The model is forced with tidal harmonic components at the marine boundaries, with measured flows and concentrations at river boundaries and with wind-induced stresses at the surface. The two main tributaries (Elorn and Aulne Rivers) were meshed up to their upstream tidal limit propagation, but the six other small rivers and two sewage outlets were considered as simple freshwater inputs in the marine meshes containing their outlet. The model also takes into account one or several sediment layers. Bottom exchanges are formulated according to Partheniades for erosion and to Krone for deposition (Cugier and Le Hir 2000). Harmonic components of tide at the sea boundary and bathymetry were provided by the Service Hydrographique et Océanographique de la Marine (S.H.O.M.). The year 1999, for which the most complete data base was available, was chosen as the reference simulation year, even if the Ulva biomass survey was done only in June 2000. Flow rates and 594 Ménesguen et al. Fig. 5. Conceptual diagram of the biogeochemical part of the model. Table 2. State variables of the ecological model.