Cholera outbreaks at Lake Tanganyika induced by climate change ? CHOLTIC- Final Report

Cholera is one of the deadliest diseases in Africa. It reappeared in the area of the African Rift in the late 70’s. The African Rift has been highlighted as major area of cholera propagation (Bompangue et al., 2008a). A link between cholera, phytoplankton blooms and copepod zooplankton has been demonstrated in Asia (Colwell et al., 1996). The African Great Lakes have been suspected to play a role as a reservoir of the bacteria V. cholerae, while human infection and movement are probably involved in the propagation of the disease inland. B. Objectives The objectives of the CHOLTIC project are to better understand the environmental conditions that trigger cholera outbreaks in the Lake Tanganyika region via an interdisciplinary study including the following aspects: (1) In situ monitoring of meteorology, limnology, phytoplankton, zooplankton, fish abundance and bacteriology during a period of three years and collaboration with DRC health authorities and epidemiology researchers. (2) Remote sensing to produce time series of daily images of Chl-a and lakesurface temperatures for the period 2000-2014. (3) Eco-hydrological modeling to investigate links between climate, nutrient mixing and variable abundance of different planktonic groups. (4) Microbiological monitoring and confirmation. (5) Genetic characterization by mass spectra identification of cholera strains. (6) Data analysis of spatio-temporal relationships between environmental factors and health data. C. Conclusions Meteorological monitoring at Uvira and Mpulungu recorded air temperature, atmospheric pressure, rainfall levels, relative humidity, wind speed, wind direction and solar radiation. Limnological monitoring was implemented from 2011 to 2014 at five sites (three coastal and two pelagic). Parameters included T°, pH, dissolved oxygen, conductivity, turbidity, transparency and chl-a concentrations. Marked short-term variability was observed in relation to internal waves and pulse of hypolimnion nutrient. Isotherms in deep waters were found to be good indicators of pulsed fluctuations of parameters. Fish relative abundance of principal species (Stolothrissa tanganicae, Limnothrissa miodon and Lates stappersii) was monitored in two sites. At Uvira in DRC (DR Congo), the main fish catches were the small clupeids: Stolothrissa tanganicae (86%). Peaks and dampening of those fish are generally observed between September and December every year during a secondary upwelling period. Deep isotherms could be indicators of planktonic and fish population changes as well as cholera. At Mpulungu (Zambia), the timing of clupeid increased catches was associated with seasonal changes and planktonic blooms particularly in April/May. Phytoplankton dynamics were assessed at the same sites and periods where the limnology tests were performed. The results include a seasonal succession in the abundance of major algal groups, genera and, where possible, up to the species level. The changes from a cyanobacteria-chrysophyte-chlorophyte community of 1975 (Hecky & Kling 1987) to a cyanobacteria-chlorophyte-diatom community, as observed near Kigoma (Cocquyt & Vyverman 2005), was confirmed for the north. However, dominance of dinophytes was also observed at Uvira during the CHOLTIC survey. At the coast, dinophytes were dominant at the end of the wet season. The cyanobacteriachlorophyte-diatom community was then replaced by a dinophyte-chlorophyte-diatom community. In the south of the lake, a cyanobacteria-chlorophyte-diatom pelagic community was characteristic, while dinophyte levels increased during the wet season. Interannual changes in phytoplankton biomass were mainly due to cyanobacteria in the north and diatoms in the south. Although no significant correlations were found between phytoplankton composition and cholera cases in the North, significant statistical relationships were observed between total phytoplankton abundance (chl-a) and cholera in the integrated analysis. Zooplankton and particularly copepods are important food components for pelagic fish species in Lake Tanganyika. In this region, copepod populations are mainly represented by one species of Calanoida, four species of Cyclopoida and Harpacticoida. Copepods have been identified in oceans as possible reservoirs of V. cholerae bacteria (Colwell et al., 1996). Therefore, the CHOLTIC project involved the monitoring of copepod levels. Copepod abundance was higher in the pelagic environment. Seasonal variations showed peaks in April/May and September/October for the most abundant species of copepods. Tropocyclops tenellus was numerically dominant throughout the study in all post-nauplii stages. Remote sensing-allowed producing time series of Lake Tanganyika surface water temperatures and ocean color products using MODIS data. This assessment covered the period 2002-2014, extending the period covered by previous BELSPO projects by eight years and improving the calibration and validation methods. Raster georeferenced files and lake snapshots of daily- and weekly-aggregated data as well as spatio-temporally aggregated time series in table format were delivered for the entire lake and 12 ecoregions. We briefly present the calibration and validation procedures implemented to derive chl-a concentrations and lake-surface temperatures using in situ data and ARC Lake datasets. Landcover maps of the three sites were also produced from Landsat TM, ETM+ and OLI images. Diachronic classification was applied for the Mpulungu area, and monodate was applied for Uvira and Kalemie. Results of the RS component identified various blooms in the lakes during the 20022014 period. Using these results, we were able to analyze the longest available cholera-case time series. The ecological model developed during the earlier projects on Lake Tanganyika (CLIMLAKE and CLIMFISH) was successfully modified to include the dominant phytoplankton groups present in the lake considered important for the present CHOLTIC project. The model is closed by the predation pressure of planktivorous fish. Simulations were performed using the wind and solar radiation data from the National Centers for Environmental Protection (NCEP) reanalysis 2. The phytoplankton production and their succession in the lake is governed by the availability of nutrients in the surface layer, its depth and light. Chlorophytes, which require low nutrients and high light, dominate in the northern proximity of the lake while the diatoms and cyanobacteria, which require high nutrients and can survive in low light conditions, dominate in the southern basin. The years with stronger wind are accompanied by an increased biomass of diatoms and cyanobacteria, while low wind years have higher biomass of chlorophytes. Various possible climatic scenarios were studied by changing the surface layer depth, its temperature and the wind stress. Different phytoplankton groups responded differently to the changes in the model forcing. Epidemiological monitoring of cholera by teams involved in the CHOLTIC project (DLM/DRC and UFC/France) was implemented as foreseen. In the context of epidemiological cholera data collected in the DRC, particular attention was focused on the South-Kivu and Katanga Provinces near Lake Tanganyika. Among the databases of the 515 DRC health zones, the project took advantage of the development of finescale databases for two target health zones: Kalemie and Uvira (2008 to 2014). Using these data, we constructed health zone attack rate maps. Time series were elaborated for these target health zones and main zones along the shores of Lake Tanganyika (Nundu, Fizi, Nyemba, Kansimba and Moba). The decompositions of time series were carried out for the two target health zones and two health areas with an endemic profile. Predominance of the most significant attack rates during rainy seasons was confirmed. Microbiological screening was successfully implemented in Zambia, and for the DRC, clinical samples were obtained via the national surveillance system. A total of 47 environmental samples from water, plankton and fish were collected in Mpulungu, Zambia. Between August 2012 and October 2014, the results indicated seasonal fecal contamination of Lake Tanganyika surface water. Furthermore, four environmental Vibrio cholerae non-O1 strains were isolated from these environmental samples. Additionally, clinical samples were collected in Mpulungu (27 V. cholerae O1, Inaba, during an outbreak in 2012), Katanga, DRC (73 V. cholerae O1, Inaba), and SouthKivu, DRC (28 V. cholerae O1 isolates: Inaba (3) and Ogawa (25)). The implementation of microbiological analysis of water and clinical samples in field laboratories was challenging and complex due to insecurity and instability in the Lake Tanganyika region and high staff turn-over at the field sites. Nevertheless, we were able to reinforce the capacities and collect a set of environmental and clinical V. cholerae isolates for further molecular analysis. Genetic analyses of 531 Vibrio cholerae isolates from patients and environmental samples from the DRC and Zambia were analyzed via MLVA (Multiple-Locus Variable number tandem repeat Analysis). The results were then compared to those of a panel of V. cholerae isolates from Guinea, Ghana and Togo. Overall, we identified 118 unique MLVA haplotypes. MLVA typing revealed the short-term divergence and microevolution of these V. cholerae populations, thereby providing insight into the dynamics of cholera outbreaks. The results revealed strong geographical clustering. Isolates from the African Great Lakes Region formed a closely related group, while West African isolates constituted a separate cluster. Interestingly, certain strains in the DRC have circulated in the region over a period of several years, occasionally giving rise to expansive epidemics. The six environmental isolates in our panel were unrelated to the epidemic isolates. Data analysis Time series (TS) of data were analyzed for the period 2002 to 2014 and the projectmonitoring period (2011-2014). The cholera epidemiological data were analyzed as a dependent variable in relation to environmental variables. All statistical analysis applied aimed to take into account delayed responses, autocorrelation, nonstationarity and the rare event nature of the epidemiological dependent variable. Binomial, ARIMA-remainders and ARIMA-modeled time series analysis enabled us to observe the following: - Spatiotemporal interactions were identified between cholera outbreaks and chl-a concentrations in the lake from remote sensing time series (aggregated by week and by ecoregion). - Two main cholera periods were identified in relation with the main meteorological seasons: wet season peaks and end of wet/dry season peaks. The different wind direction at seasonal change induces hydrodynamic fluctuations comparable to modes of development of cholera outbreaks. Damped oscillation of cholera cases was similar to those observed in various lacustrine related variables (e.g., T°, chl-a and clupeid catches), which suggests a probable lacustrine relationship with outbreaks. - Unusual high pelagic surface temperatures appeared to have a strong and positive two-week delayed effect on cholera cases with regard to the expected number of cases. - Rainfall levels were not found to have a direct statistical relationship. Seasonality was linked to a variety of other meteorological and environmental changes, which points to focalized attention on detailed study particularly in relation with initial cases at the beginning of outbreak conditions. The H0 hypothesis concerning environmental relationships with cholera was thus supported by the statistical and ecological relationships at Lake Tanganyika. However, an alternative hypothesis associated with human impacts at the origin of the outbreaks may not be rejected as lacustrine and meteorological conditions could favor the development of Vibrio cholerae, particularly as the lake presents high pH values (>9.5) during periods of planktonic blooms. In the last 40 years, various ecological changes have taken place in the lake, which may be favorable for the establishment of Vibrio cholerae. D. Support to a sustainable development policy. CHOLTIC has enabled the establishment of a new collaboration between health and environmental stakeholder sectors. Multidisciplinary collaboration is a necessary step to understand cholera outbreaks, warn populations, provide advice, and develop appropriate measures and warning systems to decrease the transmission of epidemics. CHOLTIC has helped to reinforce capacities in three stations around the lake in the field of meteorology, limnology, plankton studies and microbiology. This involved both training and logistical aspects. Various tools continue to allow local teams to prolong the monitoring for cholera studies and other fields of investigation. Various environmental and bacteriological field databases generated during the CHOLTIC project represented an important baseline, not only for further analysis with these data, but also for further investigations and comparisons. An ecological model has been considerably improved by detailing new ecological components, thereby enabling precise investigations of various planktonic groups associated with climate changes and its impact on the hydrodynamic and ecological characteristics of the lake. Analyses of CHOLTIC data have enabled us to identify interesting correlations between cholera outbreaks and environmental data, which researchers may focalize on for targeted sampling design and analysis.

Jaar van publicatie:2015

Trefwoorden:cholera, East Africa, lake Tanganyika, climate impact