Sources of pollution at the northern end of Lake Tanganyika are concentrated around Bujumbura, Burundi. This article studied the water quality characteristics of seven inflowing rivers emptying into Lake Tanganyika. Chemical analyses were carried out during rainy and dry seasons. The nutrients and suspended solids at some sites were above the critical level for healthy river water. Contamination was concentrated mainly around urban areas which have been newly developed to accommodate the rapidly growing population in the country. Compared to previous studies, the concentrations of nutrients in the river water were relatively high, indicating increasing pollution in Bujumbura's rivers. The mean values of physico-chemical parameters, with the exception of suspended solids, were found to be higher in the dry season than in the rainy season. This result is consistent with previous studies. Principal components analysis and factor analysis were employed for numerical analyses, and proven to be useful to assess the source of pollution in specific river sites, where monitoring data is missing. Routine monitoring of basic physical and chemical indicators of water quality is recommended for all urban lakes and rivers in order to protect both the aquatic ecosystem and health of the local residents living near the lake.

Introduction

Lake Tanganyika is the world's second deepest lake after Lake Baikal in Siberia. It has more than two thousand species of aquatic plants and animals, including more than 600 endemic species, which gives it an irreplaceable status in global biodiversity (Coulter, 1991). With a draining time of 1500 years and filling time of 600 years (Bergonzini, 2002), the ecology of Lake Tanganyika is susceptible to pollution, particularly from the catchment nutrient loads, human induced wastes from growing population in the cities, and climate induced transboundary atmospheric deposition (Bootsma et al., 1996; IMF, 2010).

The Lake Tanganyika basin has a relatively low degree of land use change and development. Agriculture and animal husbandry are the two major land use activities. Burundi's capital, Bujumbura, is the largest city around the lake and is regarded as the major source of pollution for the lake. In addition, the water and soil losses caused by deforestation around the city have increased the amount of sediment discharge, threatening the environment and biodiversity in the littoral regions of Lake Tanganyika (Chen and Kimirei, 2015). With the population growth of ca. 4% per year in the riparian cities (Yu, 2016), human activities are likely to increase influencing the quality of inflow rivers.

The current understanding of the nutrient content in the river inflows into Lake Tanganyika is very limited. Only some large rivers including Rusizi, Malagarasi, and Lufubu were selected for an investigation of the external nutrient sources of the lake (Langenberg et al., 2003a; Brion et al., 2006). Sporadic reports can be found on smaller river inflows in some literature. For example, a monthly survey was conducted at the river mouths of Ntahangwa and Mugere from November 1992 to October 1993 for the water quality indicators such as dissolved oxygen (DO), temperature, pH, conductivity (EC), suspended solids (SS), and nutrients (phosphorous, nitrogen) (Vandelannoote et al., 1996). From March 1994 to July 1995 ten sampling surveys were carried out at the Rusizi River, where eighteen physical and chemical indicators were measured (Vandelannoote et al., 1999). At almost the same time, Langenberg et al. (2003b) measured similar indicators at the outlet of Rusizi River every three-months from August 1994 to August 1995. The latest sampling was completed by Brion et al. (2006) from late August 1998 to November 1999 where samples were collected at the river mouths of Rusizi, Mutimbuzi and Ntahangwa for nutrient analysis. All the reported data were produced before or around 2000 when a pollution study was conducted in the lake (Bailey-Watts, 2000).

The objective of this study was to investigate the water quality status and trends under the pressure of urban and agricultural development, and to give important references for future comparisons. Therein, this article addresses water quality of the seven river inflows of Lake Tanganyika in Bujumbura. The differences in nutrient delivery among these rivers were investigated in order to identify the influence of urban development. Seasonal changes in nutrients were also studied and compared with previous reports aiming at monitoring and managing river water quality in the Basin.

Materials and methodology

Study area

Burundi's capital, Bujumbura, is located at the northern end of Lake Tanganyika (29°36′ E, 3°40′ S) in the western part of the country. Burundi has a tropical highland climate, with the mean annual temperature varying between 23 and 25°C. The mean annual precipitation is between 800 and 1 300 mm, but unevenly distributed. During the dry season, from May to September, the average monthly rainfall is below 50 mm.

The city encompasses around 146 km2 (Sindayihebura, 2003) with a population of 478,155 (as of 2008; Bigirimana et al., 2012). The formal and informal settlements are extensively situated in the city as shown with impervious surface in Figure 1. Despite a drinking water support system maintained by the Burundian drinking water company (REGIDESO), the residents often resort to using untreated water from the rivers, lakes, shallow wells, and water haulers because REGIDESO lacks the means to satisfy the demand (USAID, 2010). In addition, the wastewater treatment facility is inadequate. The “Laguna” treatment plant is estimated to serve about 40% of the urban area and a large portion of the wastewater is being disposed of in rivers, storm drains and has consequently entered Lake Tanganyika (IMF, 2010). In terms of the health of residents and the lake's ecosystem, the situation of riverine water quality in the region is critical.

We selected seven rivers, notably the Rusizi, Mutimbuzi, Kinyankonge, Ntahangwa, Muha, Kanyosha and Mugere rivers, to monitor and analyze their water quality. The study was undertaken along the north-south gradient of each river (Figure 1). Among them, Rusizi river is the largest with an yearly average flow of 182.4 m3s−1 (Brion et al., 2006). Upstream, the Rusizi River connects with Lake Kivu, passes through the Rusizi National Park and enters Lake Tanganyika in the rural Bujumbura area. The Mutimbuzi and Kanyosha rivers are located at the peri-urban areas in the north and the south of the city, respectively. The Mugere River is situated in the rural area, south of the city. The other three rivers flow through the central urban area (Figure 1), and have a relatively lower flow rate than the Rusizi, which is about 3 m3s−1 (Brion et al., 2006).

Sampling and analytical methods

The samples were collected in the rainy (March) and dry seasons (July) in 2014. The sampling sites were near the river mouth, located 100–200 meters upstream of the bridges across the river. Latitude and longitude of sampling sites were recorded in Table 1. Two parallel samples were collected for each site, including two 500 ml samples of unfiltered water and two 15 ml samples of filtered water, which were filtered through a 0.45 μm cellulose filter. All the water samples were transported in a small incubator with ice to the local laboratory at the National Institute for Nature Conservation and the Environment, Burundi (INECN) within 2 h. Then, except for the samples for suspended solids (SS) measurement, were preserved by freezing for long distance transportation.

Twelve parameters were measured. The values for temperature, pH, DO, EC and salinity were directly determined in situ using a multi-parameter water quality monitoring instrument (YSI Incorporated, Yellow Springs, Ohio, USA). Sensors were calibrated before measurement.

Suspended solids were measured at the INECN immediately after sampling. Total nitrogen (TN) and total phosphorus (TP) were analyzed in a laboratory at Kigoma Centre of the Tanzania Fisheries Research Institute (TAFIRI). For the unfiltered samples, the TN and TP were measured by a sUV-Vis spectrometer (UV-2450PC, Shimadzu). The potassium persulfate oxidation-UV spectrophotometer methodology (APHA, 1998) was adopted. Ammonia–nitrogen (NH4+), Nitrate–nitrogen (NO3), Nitrite–nitrogen (NO2), and orthophosphate (PO43−) were analysed in China by the Flow Continuous Chemistry Analyzer (San++, Skalar) at the State Key Laboratory of the Nanjing Institute of Geography and Limnology Chinese Academy of Sciences (NIGLAS).

In order to improve the reliability of the data quality, two parallel samples were taken for chemical analysis. All the solutions were prepared with analytically pure reagent (99.8%, Sinopharm Chemical Reagent Co., Ltd, Shanghai, China) and ultrapure water.

Statistical analysis method

Factor analysis (FA) has been frequently applied in the analysis of water quality indicators and their spatial variation (Huang et al., 2010). With the dataset consisting of 12 water parameters and their values at seven sites in two seasons, we examined whether the number of samples was sufficient for FA by Kaiser–Meyer–Olkin (KMO) statistics. Once the dataset was tested, then the correlation between variables was determined, whether this was significant or not, by using the Bartlett's test (Shrestha and Kazama, 2007). The KMO result (0.58), was greater than 0.5, allowing us for a Bartlett's test (p < 0.001). The significant result of the Bartlett's test met the requirement to run the FA on our dataset (Wu et al., 2010; Gyawali et al., 2012). Secondly, the individual values were standardized for PCA, with the exception of water temperature. It was eliminated in our analysis due to relatively minor variation. The statistical analysis was then performed using the IBM® SPSS® Statistics Version 21 software.

Results

The characteristics of water quality and seasonal variation

For each of the seven riverine inflows into the lake, the physical and chemical water quality parameters were analyzed and the results were presented in Table 2. Temperature showed less variation among sites or between seasons. Conductivity, which is a good measure of the total dissolved solids (i.e., Ca, Mg, Na, K, carbonate, bicarbonate, sulphate, chloride, phosphate, etc.) varied from 49.8 μs cm−1 to 1272.0 μs cm−1 with a mean value of 371.7μs cm−1. The mean pH value was 8.53, ranging from 7.02 to 10.54 indicating a relatively alkaline nature. The mean dissolved oxygen (DO) concentration was 6.36 mg l−1, with a range of 2.31 mg l−1 to 9.42 mg l−1. In order to have a general assessment of the indicator values, reference values have been provided at the bottom of the Table 2, which were reported as background parameters of water quality for the lakeshore (Vandelannoote et al., 1996). When compared with the reference value, all indicator values varied from site to site but were not far from the reference values except for the nutrient and SS values. Their means were about ten times higher than that of the reference values. This revealed that riverine flow was an important nutrient source for the nearshore of the lake. The dissolved oxygen concentration reached critical levels for healthy water (>4 mg l−1; EPA, 1986) for all seven sites with the exception of one value in the dry season. The mean concentration of SS was 404 mg l−1, with a range from 105.3 mg l−1 to 643.5 mg l−1 and was higher than the critical level of concentration of 80 mg l−1. The comparative results revealed a distinct possibility of pollution in these rivers.

A substantial difference was observed between the dry and rainy seasons, particularly in the nutrient indicators. The nutrient values of the seven sites for the two seasons were shown in Figure 2, which illustrates that TN, NH4+, NO3, and PO43− had relatively higher values in the dry season. For SS, the average value of the seven sites was higher in the rainy season than that in the dry season. TP didn't show an obvious difference between two seasons. This trend is consistent with the previous study in three rivers from this region (Brion et al., 2006). Langenberg et al. (2003b) also observed higher dissolved inorganic nitrogen (DIN) concentrations during the dry season compared with the wet season in the euphotic zone of the lake. Nevertheless, more data is needed to determine any seasonal differences.

River water quality changes along the rural-urban section

For the seven rivers located in rural and urban Bujumbura, the results of two observations were presented in Figure 2. The sampling sites on the x-axis from left to right represent the site locations from the north to the south along the lakeshore on the map, and so the urban sites (S3, S4, S5) are situated in the middle and the peri-urban and the rural ones at the sides. The bar chart shows a spatial trend with low concentrations of TP, PO43−, TN, NH4+ and NO3 at the two ends and a high concentration in the middle. Total phosphorous, a popular indicator for nutrient assessment, reached its peak at site 3 (Kinyankonge river) with 3.6 mg l−1 in the dry season, about 115 times the lowest value at site 7 (Mugere river). Total nitrogen also reached its peak value of 14.27 mg l−1 at site 3, which is about seven times higher than the lowest value at site 1 (Rusizi river). Ammonium reached its peak at site 3 as well, at 4.3 mg l−1. Changes in DO exhibited the opposite pattern of the nutrients, reaching its lowest point at site 3. This indicates that the rivers running through the urban area have a significantly higher level of nutrients than those running through the rural area, indicating that some urban sites have been polluted.

Discussion

Factors affecting seasonal variation in water quality

Seasonal water quality changes were accounted for by various factors. The most direct influences included precipitation, surface run-off, interflow, and groundwater flow (Vega et al., 1998). During the rainy season, rainfall is intense (high precipitation in a short time), leading to rapid increases in flow rate, suspension, and transport of sediment. Substantial erosion further triggers a high discharge of brownish, sediment-loaded water, so the value of SS was increased (Vandelannoote et al., 1999). However, during the dry season, the river water mainly relies on groundwater recharge, and the volume becomes low. The subsurface flow, which is rich in ionic concentration as a result of the higher rock soil dissolution as well as increased evapo-transpiration, elevates the levels of ionic concentration in the water (Bellos et al., 2004). Moreover, the dilution of the pollutants entering the river is less in the dry season when water volumes are low. This has also been recognized as an important reason for the high nutrient concentration in dry seasons (Vandelannoote et al., 1999).

In addition to the hydrological factors, atmospheric deposition which was driven by land use activity may be another factor influencing the seasonal variation. In a previous study, transboundary atmospheric deposition was widely identified as an important source of the elevated nitrogen concentration in the dry season (Bootsma et al., 1996; Langenberg et al., 2003b). During the second half of the dry season of each year, the grass and bushes in this area are burned to make way for new growth in rainy seasons. When the fires become uncontrollable, woodlands may also be burned, as fuel wood burning is a common practice of local people in the entire region (Thevenon et al., 2003). These vast areas of burning biomass give rise to a high level of atmospheric nitrogen, which may finally enter into rivers through dry deposition and run-off and could influence seasonal variation in the area.

Factors affecting spatial variations of water quality

Among the seven rivers Kinyankonge River (S3) has the highest value for TN, NH4+, TP and PO43− and the lowest DO values. This river has two tributaries, which run through the northern part of urban Bujumbura. The large “Laguna” wastewater treatment ponds are situated at the joint point of the two tributaries. It discharges the tail water from the disposal system into the river. Together with the discharges of raw sewage from households, this constitutes a major reason for the poor water quality in site 3. The effects of pollutant discharge were exacerbated in the dry season by the limited dilution caused by the low hydrological flow and water volume. The second most polluted site in this study was in the Muha River (S5). It lies in the southern part of the Bujumbura city center where a large number of large housing plots are situated on both sides of the river. Garbage was observed to be floating in the river, including beverage bottles, rag sheets, and plastic bags. About 200 meters upstream of the sampling site garbage piled into a dam (Figure 3). Chemicals from leaking sewers, landfills, and other activities produced a high concentration of nitrogen-enriched nutrients. According to South African water quality guidelines (1996), the concentration of ammonia at S3 and S5 exceed the criteria for suitable human consumption (NH4+, <1 mg l−1; Department of Water Affairs and Forestry, 1996).

The peri-urban and rural rivers (S1, S2, S6 and S7) showed relatively low nutrient concentrations with the exception of nitrate. Less contaminated water with lower anthropogenic activities in the area might account for the results. The high concentration of nitrate could also be due to fertilizer use on agricultural land as well as waste water and atmospheric nitrogen deposition into the rivers (Brion et al., 2006). Under oxidizing conditions, the ammonia nitrogen in the fertilizer can be transformed into nitrite nitrogen and finally enter into the rivers through rain wash and run-off (Shen et al., 2011).

Pollution source assessment with FA/PCA

We obtained the principle components (PCs) and their eigenvalues from the observed water quality parameters. The first three components contributed over 94% of the variation of the original water quality variables. The factor loadings for each observed variable were calculated as shown in Table 3. The first factor (F1), which explained 61.1% of the total variance, exhibited strong positive loadings for nutrient related parameters (TP, TN, PO43−, NH4+ and NO2). As reported, the riverine phosphorous and the nitrogen in the form of NH4+ were mainly caused by domestic sewage (Vandelannoote et al., 1996), and we, therefore, inferred that F1 represented the contribution of urban sewage and industrial activities. F2, which explained 21.7% of the total variance, had strong positive loadings on EC and salinity as well as strong negative loading on NO3. Considering the significant influence of fertilizing and cultivation activities on NO3 concentration in the surface water and runoff (Shen et al., 2011), in our case, it is likely that F2 represented the contribution of agricultural activities. While F3, which accounted for 11.9% of the total variance, could have associated with a strong positive loading on suspended solids and implied soil erosion during the rainy season as a contributing factor. This evidence has also been shown by a previous study that the anthropogenic activity in the catchment of Lake Tanganyika is significant in that deforestation has caused higher rates of soil erosion over the past century (Cohen et al., 2005).

Each factor can be modeled with a multiple linear regression equation for the observed variables. The factor scores were calculated for the river sites (Table 4). The results showed that the F1 had high positive values at site 3 and site 5. Because it stands for urban and industrial pollution sources, the results indicated that these sites were distinctly characterized by urban wastewater pollution. F2 had a negative value due to agricultural pollution in the form of NO3, indicating that sites 4, 5, 6 and 7 all suffered from agricultural pollution. F3, which is positively related with SS, indicated that sites 2, 4, 5 and 6 all were strongly influenced by soil erosion. With the factor scores of each site we identified specific pollution source types. The result for the Kinyankonge River (S3) is consistent with the observation based on the site; it was heavily polluted by urban wastewater. For the Muha River (S5), Ntahangwa River (S4) and Kanyosha River (S6) which are located at urban or peri-urban areas, agricultural pollution and erosion influences have been identified by the high scores of F2 and F3, respectively. This might be related to agricultural activities, deforestation and soil erosion in the upstream regions. Our study of the land use structure has revealed that the cultivated land occupied 35–56% in the three river catchments (Yu, 2016). For the rural rivers, Mugere River (S7) was distinguished by agricultural pollution, while Rusizi River (S1) did not show a discernible pollution trend. Above all, with the FA/PCA method, three factors were derived from the observed eleven water quality indicators, endowed with meanings of different pollution sources, and accordingly used for quantitative assessment of the river sites with the limited monitoring data.

Pollution trends of river water

Despite the limited data in this area, we assembled the river water parameters reported in the literature (Table 5). All the previous sampling sites were located at the river and near the lakeshore. In the literature, sampling sites for each river differed little over time. The Rusizi River had been reported more frequently than the others. The measured data in this research is in the range of previous studies for DIN, SS, pH, EC, and DO, but not for TP. The four records for DIN in Rusizi River exhibited an upward trend from 1994, when the first value was recorded, to the present study conducted in 2014. The DIN also increased for Mutimbuzi and Ntahangwa, but decreased for Mugere compared with previous reports. This revealed that at least three inflows to the north end of Lake Tanganyika have become more contaminated. Although the water of Kinyankonge River and Muha River were not measured before, they could be assessed as heavily polluted by their relatively higher nitrogen concentration compared with Ntahangwa, Mutimbuzi and Rusizi measured in this study. Since riverine nutrient loading depends on the availability of nutrients in the drainage basin, the change observed was primarily intensified by human activity across the watershed. Although the Burundian domestic industries and manufacturing companies are relatively low in number, the COTEBU (a textile industry) and Brarudi brewery S.A. discharged significant amounts of wastewater into the rivers, 2350 and 2100 cubic per day, respectively (West, 2001). According to the 2008 census, the population of Bujumbura was 0.5 million inhabitants (ISTEEBU, 2008). The city experienced a rapid increase of 400% from 1960 to 1990, and 103% from 1990 to 2008 when the population density reached to 3276 persons per km2 (Bigirimana et al., 2012). In some neighborhoods such as Buyenze and Bwiza, the population density exceeded 33,000 persons per km2 and 27,000 persons per km2 respectively (ISTEEBU, 2008).

Because of the low capacity for environmental management of the country, garbage dumps and sewage effluent discharges were extensively identified along the sides of the rivers (Figure 3). The Municipal Technical Services (SETEMU) of Bujumbura have set up a system of weekly garbage collection using trucks and disposal the garbage at the municipal dump in Buterere, but SETEMU only covers 38% of the Bujumbura's need (USAID, 2010). As we investigated in 2014, people living around the rivers threw garbage into the rivers or piled it up on the river banks; so when the rainy season begins, the garbage is washed into the rivers. Many people near the city have been living with insufficient wastewater treatment facilities inevitably leading to water quality deterioration and increased sanitation and health hazards.

Conclusions

Our investigation of various water quality parameters of seven rivers flowing into the Lake Tanganyika provides some baseline evidence of the latest status of the lake catchment area, particularly across the Bujumbura, Burundi region. The water quality parameters of most rivers are largely contaminated by increased nutrient discharge from the catchment. The rivers in close proximity to the City of Burundi are heavily affected by various forms of water pollution. Growing urbanization and limited or no waste water treatment facilities are the main drivers of water pollution. Apart from this, increased land use change, including the application of fertilizers, deforestation, firewood burning, and atmospheric deposition can also cause water pollution in the region. Primarily, we recommend that the areas of increased discharge of wastewater in the river should be identified as focal areas for infrastructure development for wastewater treatment plants. Secondly, the introduction of sustainable development and incorporation of integrated water resource management principles (jointly by scientists, community groups and water stakeholders) in urban planning is essential in the region. Lake Tanganyika Authority (LTA) which is the secretariat established by the contracting states has the duty to co-ordinate the implementation of IWRM in the four riparian states around the lake, it harmonizes management policies, laws, regulations monitoring and data exchange, and provides a forum for the countries to engage in lake management discussions. However, given the poor economic and funding deficiencies of the LTA and the riparian countries, it would be difficult to achieve a more comprehensive database development and water resource management in the study region.

The FA/PCA tools used in our study suggest that the sources of water quality pollution in the region can be identified more reliably by using this low-cost, efficient technique which can aid the development of a comprehensive monitoring and management strategy of the river and lake systems in the region. As the current results are still insufficient to discern whether the types of water quality problems are ubiquitous in the Lake Tanganyika Basin, finally, the increase in unplanned development, its causal factors and contribution suggest that river water quality parameters should be investigated further. In general, a routine monitoring of basic physical and chemical indicators of water quality for urban rivers and lakes in less studied regions of Africa and beyond is essential for protection of the lakes and river systems as well as the well-being of the local residents.

Acknowledgements

The authors are grateful to Dr. Ismael Kimirei of the Kigoma Centre at the Tanzania Fisheries Research Institute (TAFIRI) for his kind help with lab work and to Mr. Gabriel Hakizimana of the Lake Tanganyika Authority for his coordination efforts, and to Professor Giri R Kattel of Federation University of Australia for his comments on this article. Many thanks to Mr. Sixte Sakubu for his guidance in Bujumbura and to Dr. Chenjie Tang and Dr. Xiaolong Yao for their help with water sample analysis.

Funding

This study was funded by the Sino-Africa Joint Research Center, Chinese Academy of Sciences (NO. SAJC201319).

The text of this article is only available as a PDF.

References

American Public Health Association (APHA)
,
1998
.
Standard methods for examination of water and wastewater
.
American Water Works Association and Water and Environmental Federation
,
Washington, DC
.
Bailey-Watts, A.,
2000
.
Summary Technical Report. Pollution Special Study. Pollution Control and Other Measures to Protect Biodiversity in Lake Tanganyika
.
UNDP/GEF/RAF/92/G32:72
.
Bellos, D., Sawidis, T., Tsekos, I.,
2004
.
Nutrient chemistry of river pinios (Thessalia, Greece)
.
Environ. Int.
30
,
105
115
.
Bergonzini L.,
2002
. Computed mean monthly water balance of a large lake: the case of Lake Tanganyika. In: J., Klerkx, B. Imanackunov (Eds.),
Lake Issyk-Kul: Its Natural Environment Part II
, pp.
217
244
.
Springer
,
Netherlands
.
Bigirimana, J., Bogaert, J., De Cannière, C., Bigendako, M. J., Parmentier, I.,
2012
.
Domestic garden plant diversity in Bujumbura, Burundi: Role of the socio-economical status of the neighborhood and alien species invasion risk
.
Landscape Urban Plan
.
107
,
118
126
.
Bootsma, H., Bootsma, M., Hecky, R.,
1996
. The chemical composition of precipitation and its significance to the nutrient budget of Lake Malawi. In: A. I. Johnson (Ed.),
The limnology, climatology and paleoclimatology of the East African lakes
, pp.
251
265
.
CRC Press, Florida
.
Brion, N., Nzeyimana, E., Goeyens, L., Nahimana, D., Tungaraza, C., Baeyens, W.,
2006
.
Inorganic nitrogen uptake and river inputs in Northern Lake Tanganyika
.
J. Great Lakes Res.
32
,
553
564
.
Chen, S., Kimirei, I.,
2015
.
Demonstration research on comprehensive water quality monitoring in the Lake Tanganyika basin. MOST/UNEP joint project report
. http://www.protect-lake-tanganyika.cn/
Cohen, A. S., Palacios-Fest, M. R., McGill, J., Swarzensk, P. W., Verschuren, D., Sinyinza, R., Songori, T., Kakagozo, B., Syampila, M., O'Reilly, C. M., Alin, S. R.,
2005
.
Paleolimnological investigations of anthropogenic environmental change in Lake Tanganyika: I. An introduction to the project
.
J. Paleolimnol
34
,
1
18
.
Coulter, G. W.,
1991
.
Lake Tanganyika and its life. Natural History Museum Publications
,
Oxford University Press
,
London, Oxford, New York
.
Department of Water Affairs and Forestry
,
1996
.
South African Water Quality Guidelines
(second edition). Volume
1
:
Domestic Use
.
Gyawali, S., Techato, K., Yuangyai, C., Monprapusson, S.,
2012
.
Evaluation of surface water-quality using multivariate statistical technique: A case study of U-tapao river basin, Thailand
.
KMITL Sci Techol J
.
12
,
7
20
.
Huang, F., Wang, X., Lou, L., Zhou, Z., Wu, J.,
2010
.
Spatial variation and source apportionment of water pollution in Qiantang River (China) using statistical techniques
.
Water Res
.
44
,
1562
1572
.
International Monetary Fund (IMF)
,
2010
.
Burundi: Poverty Reduction Strategy Paper: Second Implementation Report. IMF Country Report No.10/312
.
Institut de Statistiques et d'Etudes Economiques du Burundi (ISTEEBU)
,
2008
.
Troisieme recensement général de la population et de l'habitat de 2008 (Third General Census of Population and Housing, 2008)
.
Langenberg, V. T., Nyamushahu, S., Roijackers, R., Koelmans, A. A.,
2003a
.
External nutrient sources for Lake Tanganyika
.
J. Great Lakes Res.
29
,
169
180
.
Langenberg, V. T., Sarvala, J., Roijackers, R.,
2003b
.
Effect of wind induced water movements on nutrients, chlorophyll-a, and primary production in Lake Tanganyika
.
Aquat. Ecosyst. Health.
6
,
279
288
.
Shen, Y., Lei, H., Yang, D., Kanae, S.,
2011
. Effects of agricultural activities on nitrate contamination of groundwater in a Yellow River irrigated region. In:
Water Quality: Current Trends and Expected Climate Change Impacts
.
348
,
73
80
.
Melbourne, Australia
.
Shrestha, S., Kazama, F.,
2007
.
Assessment of surface water quality using multivariate statistical techniques: A case study of the Fuji river basin, Japan
.
Environ. Modell. Softw.
22
,
464
475
.
Sindayihebura, B.,
2003
.
Périmètre urbain de Bujumbura (Map of urban area of Bujumbura)
.
Direction générale de l'urbanisme Bujumbura
.
Thevenon, F., Williamson, D., Vincens, A., Taieb, M., Merdaci, O., Decobert, M., Buchet, G.,
2003
.
A late-Holocene charcoal record from Lake Masoko, SW Tanzania: climatic and anthropologic implications
.
Holocene
13
(
5
),
785
792
.
U.S. Agency for international development (USAID)
,
2010
.
Burundi Water and Sanitation Profile
. Retrieved from http://www.usaid.gov.
US EPA
,
1986
.
Quality Criteria for Water. EPA 440/5-86-001, 1986
.
United States Environmental Protection Agency, Office of Water
,
Washington, DC
.
Vandelannoote, A., Robberecht, H., Deelstra, H., Vyumvuhore, F., Bitetera, L., Ollevier, F.,
1996
.
The impact of the River Ntahangwa, the most polluted Burundian affluent of Lake Tanganyika, on the water quality of the lake
.
Hydrobiologia
.
328
,
161
171
.
Vandelannoote, A., Deelstra, H., Ollevier, F.,
1999
. The inflow of the Rusizi river to Lake Tanganyika. In: O. Lindqvist, H. Molsa K. Solonen, J. Sarvala, (eds.),
From Limnology to Fisheries: Lake Tanganyika and Other Large Lakes
.
Springer
,
Netherlands
.
Vega, M., Pardo, R., Barrado, E., Debán, L.,
1998
.
Assessment of seasonal and polluting effects on the quality of river water by exploratory data analysis
.
Water Res
.
32
,
3581
3592
.
West, K.,
2001
.
Lake Tanganyika: Results and Experience of the UNDP/GEF Conservation Initiative (RAF/92/G32) in Burundi, Democratic Republic of Congo, Tanzania and Zambia
.
United Nations Development Programme (UNDP), Global Environment Facility (GEF) and UNOPS
.
Wu, M. L., Wang, Y. S., Sun, C. C., Wang, H., Dong, J. D., Yin, J. P., Han, S. H.,
2010
.
Identification of coastal water quality by statistical analysis methods in Daya Bay, South China Sea
.
Mar. Pollut. Bull.
60
,
852
860
.
Yu, C.,
2016
.
Impact of land use on water environment in the Northeast area along Lake Tanganyika in Africa
. Ph.D. dissertation,
Nanjing Institute of Geography and Limnology, Chinese Academy of Sciences
,
Nanjing, China
.