Les lacs Tanganyika et Malawi sont, de par leur volume, les deux plus grands lacs africains. Ces réservoirs semblent pour l'instant épargnés par la pollution en éléments en trace. Il est toutefois crucial, en raison de leurs caractéristiques hydrologiques, de poser la question du temps de réponse de ces systèmes à une pollution chronique potentielle véhiculée par les affluents. Cet article simule ainsi cette réponse dans la fraction dissoute suite à l'introduction pendant 50 ans de polluant par tous les affluents. Cette démarche s'appuie sur un modèle hydrologique intégrant les trois compartiments des colonnes d'eau (épi-, méta- et hypolimnion) et sur la prise en compte de la réactivité des éléments dissous dans ces compartiments par l'intermédiaire du taux de rétention élémentaire. Ainsi quatre types d'éléments sont considérés, (i) le type Cl, non réactif, (ii) le type Si, réactif-nutritif, (iii) le type Mn et (iv) le type V tous deux réactifs sensibles aux conditions d'oxydo-réduction. La réactivité de l'élément, l'efficacité du mélange vertical ainsi que la position de l'oxycline dans la colonne d'eau conditionnent l'amplitude et la cinétique de réponse des systèmes ainsi que le temps de retour à la situation initiale après l'arrêt des apports polluants. Ces caractéristiques propres à l'élément et au lac influent sur le risque potentiel encouru par l'écosystème et l'homme. Ainsi la pollution affecte principalement les eaux de surface (types Cl et V), les réseaux trophiques (type Si), les eaux profondes (types Si et Mn) et le compartiment sédimentaire (types Mn et V).
Lakes Tanganyika and Malawi are the largest African lakes as measured by volume. They constitute essential water and protein resources for the surrounding populations. These aquatic systems have become stressed due to high human population density, growth and associated activities. While eutrophication was apparent locally and organic pollutants were detected in fish and water, concentrations of several dissolved trace elements of potential concern corresponded to uncontaminated systems. However, due to their hydrological features, it was important to characterise the lake response time to chronic contamination loaded by the tributaries. This paper presents two simulations of this response, in the dissolved fraction, following 50 years of pollutant input by the tributaries. The first simulation corresponded to an annual pollutant input that was the same for both lakes, resulting in mean river input concentrations of 5.0 U L-1 and 3.7 U L 1 (where U is a weight or molar unit), respectively, for lakes Tanganyika and the Malawi. The second simulation corresponded to an annual input proportional to the lake volume, with mean river input concentrations of 5.0 U L 1 and 1.5 U L 1, respectively, for lakes Tanganyika and the Malawi. The polluted input was loaded by the dissolved fraction with the exception of Mn-type elements, which were carried by the particulate fraction. This approach was based on an annual hydrological model of three water column compartments (epi-, meta- and hypolimnion) of these meromictic lakes. In addition, the reactivity of dissolved elements in the water column was taken into consideration. The reactivity was characterised by the elemental retention rate that quantifies dissolved-particulate interactions linked to biological and physico-chemical processes. The reactivity of trace elements was assessed through their concentration distribution profile in the water column. Four element types were considered: the non-reactive elements characterised by homogenous concentrations in the water column (Cl-like); the micronutrient-type elements (Si-like) characterised by a strong positive concentration gradient below the thermocline; redox-sensitive elements (Mn-like) characterised by a strong positive concentration gradient below the oxycline and other redox-sensitive elements (V-like) characterised by a strong negative concentration gradient below the oxycline. Trace elements (F, Al, Fe, Mn, V, Ba, Sr, Mo, Cr, Ni, Co, Cu and Pb) in both lakes were associated with these element types but they did not necessarily belong to the same type in both lakes. Other elemental types likely occurred (e.g., carbonate type and Fe types) but they were not clearly identified. After 50 years, surface concentrations ranged from 0 to 1.15 U L 1 in Lake Tanganyika and from 0 to 2.40 U L 1 in Lake Malawi. The difference between the lakes was linked to the greater volume of Lake Tanganyika, mainly in its hypolimnion, and to the longer vertical water exchange time in Lake Tanganyika. For Cl-type elements the concentration response decreased for both lakes from the epi- to the hypolimnion with similar kinetics for the epi- and metalimnion and a delay for the hypolimnion. For Si-type elements the response decreased in Lake Malawi from the hypo- to the epilimnion and for Lake Tanganyika the maximal concentration was calculated in the metalimnion. The concentration range was higher in Lake Malawi than in Lake Tanganyika. For the Mn-type elements, the maximum concentration was calculated in the hypolimnion with a higher response in Lake Malawi. The metalimnetic water concentration of Lake Tanganyika increased slightly and epilimnetic and metalimnetic waters of Lake Malawi did not react. For V-type elements the epilimnetic waters were more sensitive to the increase, with a higher response for Lake Malawi. In Lake Malawi concentrations also increased in the metalimnion. Concentrations in the hypolimnetic zone of both lakes and metalimnetic zone in Lake Tanganyika remained zero. Depending on the element type and on the lake, the time required to return to initial conditions, when contaminant inputs stopped, varied from 30 to 7 300 years. In the epilimnetic zone of both lakes the intensity of reaction and the pollution persistence were higher for Cl-type elements. For Si-type elements, mainly in Lake Malawi, the vertical input from deep waters was sufficient to sustain productivity even after the input of pollutants was stopped. For these elements the dissolved contamination was mainly stored in deep waters. For Mn-type elements the contamination was also stored in deep waters with a relatively slow net transfer to the sedimentary compartment. V-type pollutants were transferred from the dissolved to the particulate phase in deep waters leading to a relatively rapid net transfer to the sediment. Once the pollutant was in the system and until its evacuation to the outlet or to sediment, the risk for the ecosystem and for the population was associated with its presence in the dissolved phase of the surface water. The risk was then higher for Cl- and V-type elements as well as for the Si-type elements that were introduced into the web food. For the Si- and Mn-type elements that were mainly stored in deep waters, the associated risk was linked to a breaking of the thermo-haline stratification or to a reinforcement of vertical mixing. For the V-type elements and also for the sedimentary fraction of the Mn-type elements, the risk was also associated with possible remobilization from the sediments due to physico-chemical changes at the water-sediment interface.Element reactivity, efficiency of the vertical mixing and the depth of the oxycline control the importance and the kinetic response. They also controlled the time to attain initial conditions once contaminant inputs were stopped. These features, relative to the element and to the lake, were key parameters in the assessment of the potential risk for both the ecosystem and people that rely on these lakes. Even if the elemental typology was the same for both lakes, elements can be considered a different type from one lake to another. Contamination from the same pollutant would then have different consequences, for instance regarding the associated risk. Computed hydrochemical budgets were simple but realistic, illustrating the behaviour of elements in the water column. Computation of this budget requires the knowledge of global water column fluxes, which have to be improved mainly for Lake Tanganyika. The element's reactivity was mainly linked to liquid-solid reactions. It would be interesting in future studies to characterise particulate phases and their reactivity and to introduce such processes in hydro-geochemical models. Computations of chronic contamination response indicate that for both lakes, due to the inertia of the hydrochemical system, the lack of lake water contamination does not imply a systematic lack of pollution in the tributaries. Once pollution is detected, it will be persistent. A global watershed monitoring program should be organised in the near future. Monitored parameters should be relevant to metallic and organic pollutants, as well as eutrophication.
