Abstract Half a century after the Tanganyika sardine (Limnothrissa miodon Boulenger) was introduced into Lake Kivu, several aspects of metazooplankton ecology were investigated from January 2002 to June 2005 in the pelagic zone of the lake. As in other large lakes of the region, zooplankton in Lake Kivu is species-poor, and is dominated by copepods. In addition to three cyclopoid species, four cladoceran and 12 rotifer taxa were recorded. Zooplankton showed marked seasonal variation. Total crustacean abundance increased to a distinct dry season maximum (August-September), following the rise of phytoplankton production associated with deep vertical mixing. The three copepods and the most important cladoceran species exhibited different patterns of vertical migration, depending on their feeding habits, life stages and body size at the adult stage. The relatively small Tropocyclops confinis Kiefer was permanently present in the euphotic layer, while the largest copepod species, Thermocyclops consimilis Kiefer and Mesocyclops aequatorialis Kiefer, as well as the cladoceran Diaphanosoma excisum Sars, exhibited a typical diel vertical migration, with some differences among life stages. Total biomass of copepods in Lake Kivu (Mean= 0.3 g C m-2) is lower than in lakes Tanganyika and Malawi. Similarly, mean annual total production (8.3 g C m-2 y-1) is about three times as low as in lakes Malawi and Tanganyika. The ratio between phytoplankton production and zooplankton production is low (about 1.6 %), suggesting a low transfer efficiency at this food web level. As primary production is in the range found in the other large lakes of the region, it is suggested that the low total zooplankton biomass in Lake Kivu is related to the disappearance of Daphnia curvirostris Eylmann, which was likely the key grazer before the planktivorous fish introduction. The absence in Lake Kivu of a calanoid copepod species, which can more efficiently exploit phytoplankton production, may be another reason why carbon transfer efficiency between phytoplankton and zooplankton is low. Data analysis using multivariate methods showed that seasonal variations of the ratio mixed layer depth: euphotic depth (Zm:Zeu) is the key driving force influencing plankton dynamics via its influence on quantity and quality of zooplankton food resources. This suggests that mesozooplankton dynamics in Lake Kivu is essentially bottom-up controlled. Presently, the sole indication of a significant impact of L. miodon predation on zooplankton is the decrease of average body size of the cladoceran Diaphanosoma over time. Résumé Un demi-siècle après l’introduction de la sardine Limnothrissa miodon Boulenger, endémique du lac Tanganyika, plusieurs aspects de l’écologie du métazooplancton pélagique du lac Kivu ont fait l’objet d’investigations dans ce lac de janvier 2002 à juin 2005. Comme dans d’autres grands lacs de la région, le métazooplancton du lac Kivu présente une faible diversité spécifique. Il est dominé par trois espèces de copépodes cyclopoïdes. Quatre espèces de cladocères et 12 taxa de rotifères ont également été observés. Le zooplancton de ce grand lac présente des variations saisonnières marquées. L’abondance totale des crustacés présente un pic distinct vers la fin de la saison sèche (août - septembre), qui succède à un pic de phytoplancton. Les 4 espèces les plus abondantes montrent différents patrons de migration verticale, dépendant de leur régime alimentaire, du stade de développement et de la taille du stade adulte. Le plus petit copépode, Tropocyclops confinis Kiefer reste en permanence dans la zone photique alors que les 2 grandes espèces, Thermocyclops consimilis Kiefer et Mesocyclops aequatorialis Kiefer, ainsi que le cladocère Diaphanosoma excisum Sars, réalisent des migrations verticales journalières, dont l’amplitude varie selon le stade de développement des individus. La biomasse totale des copépodes (moyenne = 0.3 g C m-2) est plus faible que dans les lacs Tanganyika et Malawi. Une différence semblable a été observée pour la production totale annuelle qui est de 8.3 g C m-2 j-1, soit environ 3 fois moins que dans les deux autres grands lacs. Le rapport entre la production primaire phytoplanctonique et la production zooplanctonique est également faible (environ 1.6 %) ; indiquant une faible efficience de transfert à ce niveau du réseau trophique. Le fait que la production primaire du Lac Kivu soit du même ordre de grandeur que dans les autres grands lacs de la région suggère que les faibles valeurs de la biomasse totale du zooplancton sont une conséquence de la disparition de Daphnia curvirostris Eylmann, qui était sans doute le brouteur le plus efficace avant l’introduction de la sardine. L’absence au Lac Kivu d’une espèce de copépode calanoide, qui peut exploiter le phytoplancton de façon optimale, est sans doute une autre raison de la faible efficience du transfert de carbone entre le phyto- et le zooplancton. L’analyse des données par des méthodes multivariées a montré que la variation saisonnière du rapport zone de mélange: zone photique (Zm:Zeu) est le facteur clé influençant la production du métazooplancton via la quantité et la qualité de ses ressources alimentaires. La dynamique zooplanctonique dans ce lac subit donc essentiellement un contrôle « bottom-up ». La diminution de la taille des individus de Diaphanosoma au cours du temps est actuellement la seule indication de l’impact de la prédation de Limnothrissa miodon sur le zooplancton.

With a volcanic origin, Lake Kivu is deep and meromictic, and shows a very particular limnology and some astonishing features. The data available on its limnology and phytoplanktic communities are limited, dispersed or outdated. This is the first deep, long term study (2002-2004) on limnology and phytoplankton ecology of Lake Kivu, combining different techniques: HPLC analysis of marker pigments, flow cytometry, light, epifluorescence and electron microscopy. Lake Kivu combines a relatively shallow euphotic layer (~18m) usually smaller than its mixed layer (20 – 60 m), and with a weak thermal gradient in the mixolimnion. With an annual average chlorophyll a in the mixed layer of 2.2 mg m-3 and low nutrient levels in the euphotic zone, the lake is clearly oligotrophic. Concerning its phytoplanktonic composition, the most common species were the pennate diatoms Nitzschia bacata Hust. and Fragilaria danica (Kütz.) Lange Bert., and the cyanobacteria Planktolyngbya limnetica (Lemm.) Komárková-Legnerová and Cronberg and Synechococcus sp. Diatoms were the dominant group in the lake, particularly during the dry season episodes of deep mixing. During the rainy season, the stratified water column, with high light and lower nutrient availability, favoured dominance of cyanobacteria. Phycoerythrin-rich phototrophic picoplankton and heterotrophic bacteria cell numbers were constantly high, with relatively subtle spatial, seasonal and vertical variations. In open lake waters, where allochthonous carbon inputs are most probably inconsequential, HNA heterotrophic bacteria abundance is strongly correlated with chlorophyll a. Recent investigations revealed an increasing methane production in the deep waters during the past three decades, leading to an accumulation of gas and the subsequent lowering of the energy required to trigger a devastating release of gasses. The role of primary producers and the probable changes on the export ratio of the organic matter into deep waters after the introduction of the planktivorous sardine from Lake Tanganyika Limnothrissa miodon, is discussed. The actual primary production, 0.71 g C m-2 d-1 (~ 260 g C m-2 y-1), doesn’t seem substantially different from some punctual measurements made in the past, which discards the hypothesis of anthropogenic eutrophication. We believe that the export ratio of organic matter into the deep waters was largely affected by biological changes that could explain part of the methane increase in the past 30 years in Lake Kivu.