Geochemical Investigation of Arsenic in Drinking Water Sources

DEGREE PROJECT IN ENVIRONMENTAL ENGINEERING, SECOND CYCLE, 30 CREDITS STOCKHOLM, SWEDEN 2017

ENRICO LUCCA

KTH ROYAL INSTITUTE OF TECHNOLOGY SCHOOL OF ARCHITECTURE AND THE BUILT ENVIRONMENT

Geochemical Investigation of Arsenic in Drinking Water Sources in Proximity of Gold Mining Areas within the Lake Victoria Basin, in Northern Tanzania

ENRICO LUCCA

Supervisor Prosun Bhattacharya

Examiner Ann-Catrine Norrström

Supervisor at Dept. of Water Resources Engineering University of Dar es Salaam, Tanzania Felix Mtalo

Degree Project in Environmental Engineering and Sustainable Infrastructure KTH Royal Institute of Technology School of Architecture and Built Environment Department of Sustainable Development, Environmental Science and Engineering SE-100 44 Stockholm, Sweden

Summary

Access to safe drinking water is a challenge for rural communities in many developing countries. Drinking contaminated water endangers human health and impairs social and economic development. Arsenic (As) is a metalloid widely distributed into the environment and is highly toxic in its trivalent inorganic form. The natural occurrence of As in groundwater used for drinking has been associated to the insurgence of skin disorders and respiratory diseases in many parts of the world.

Arsenic is frequently found in the form of sulphide in gold deposits. Human exposure to As via drinking water has resulted from gold mining activities in some instances in USA, Asia, South America and Africa.

The problem of As in drinking water has been brought to attention in Tanzania only few years ago and further investigation are therefore needed to enable an early detection of harmful exposures. This study aims to assess occurrence, source and mobilization mechanisms of As in some drinking water sources within the Lake Victoria Basin, in Northern Tanzania. Rural communities living in areas known for artisanal and large-scale gold mining activities were the target of the present study.

Fifty-four water samples were collected from a variety of drinking water sources (spring, borehole, river and shallow well) in Mara and Geita region during October 2016. pH, electrical conductivity (EC), Redox potential (Eh) and As were measured in situ. Major ions, dissolved organic carbon (DOC) and trace elements, including As, were analysed in the sampled water at KTH-Royal Institute of Technology, in Sweden.

53% of the sampled water do not comply with the WHO recommended limit of 10 µg/L, representing a serious health risk for some rural communities within the Lake Victoria Basin. The spatial distribution of As in the area under investigation is highly heterogeneous and it is mainly influenced by local geology and proximity to the mining sites (approx. < 5 km). Lower As levels in boreholes than in rivers and shallow wells indicates contamination of surface drainage by mining activities and suggest that deep groundwater ( > 40 m) generally represent a source of safer drinking water.

The field-measured Redox potential indicates oxidising conditions, suggesting that oxidation-dissolution of arsenic sulphide minerals is a major mechanism of arsenic mobilization in groundwater. However, this study reveals that several geochemical processes control fate and mobility of As, once it has been released into the aquatic environment. Large discrepancies between field and laboratory measurements of As indicates a strong partition of the metalloid into the particulate fraction. As revealed by the geochemical modelling, co-precipitation with iron /aluminium hydroxides and adsorption on clay minerals are presumed to be the major sinks for dissolved As. Moreover, a good match between peaks in As and dissolved organic carbon concentrations suggests that complexation by humic acids is responsible for enhanced As mobility.

Overall, the present study has led to a better understanding of the problem of arsenic in proximity of gold mining areas in Tanzania and it calls for the development of affordable and sustainable solutions which would provide safe drinking water to the affected population. Sommario

L’accesso a una fonte sicura di acqua potabile è un prerequisito fondamentale per la salute umana e per uno sviluppo sostenibile in ambito sociale ed economico. L’arsenico (As) è uno dei contaminanti di origine naturale più diffusi e problematici delle risorse idriche su scala globale. L’ingestione di acqua contaminata da As è stata associata all’insurrezione di gravi patologie cutanee, respiratorie e del sistema nervoso in molte aree del mondo, in particolare in paesi in via di sviluppo. L’As è un metalloide ubiquitario nell’ambiente e la sua origine geologica è talvolta associata a depositi solfuri auriferi. Pertanto, le attività minerarie per l’estrazione dell’oro possono essere la causa di un’estensiva contaminazione delle acque superficiali e sotterranee. Nonostante il problema dell’arsenico sia stato studiato dalla comunità scientifica a livello gloabel, ci sono ancora Paesi in cui l’esposizione al mettalloide non è ancora stata adeguatamente investigata. Un esempio è il la Tanzania.

Questa Tesi si propone di chiarire origine, distribuzione e mobilità dell’arsenico in alcune fonti di acqua usate a scopo potabile nel bacino idrografico del Lago Vittoria, in Tanzania. Le comunità rurali che vivono in prossimità di attività minerarie sono state l’oggetto principale di studio.

Cinquantaquattro campioni sono stati prelevati da diverse fonti d’ acqua nelle regioni Mara e Geita durante Ottobre 2016: pozzi profondi (> 40 m), pozzi superficiali, sorgenti e fiumi. Misurazioni di pH, conducibilità elettrica, potenziale Redox e concentrazione di As sono state eseguite in situ. I campioni d’acqua sono stati analizzati al KTH di Stoccolma per la determinazione degli ioni principali, carbonio organico disciolto e altri elementi presenti in traccia, tra cui l’arsenico.

Il 53% delle fonti campionate presenta una concentrazione di arsenico che eccede il limite 10 µg/L raccomandato dall’Organizzazione Mondiale della Sanità, costituendo un grave rischio per la salute umana. La distribuzione dell’arsenico nelle acque campionate è altamente eterogenea ed è principalmente influenzata dall’assetto geologico locale a dalla vicinanza al sito minerario. Concentrazioni di arsenico minori nei pozzi profondi rispetto a fiumi e pozzi superficiali indica la contaminazione del deflusso superficiale da parte delle attività minerarie e suggerisce che i pozzi profondi siano una fonte di acqua potabile più sicura.

Le misurazioni in situ del potenziale Redox indicano un ambiente ossidante, presumendo quindi che l’ossidazione/dissoluzione di minerali solfuri di arsenico sia il principale meccanismo di rilascio di As nell’acque superficiali e sotterranee. Tuttavia, questo studio rivela che numerosi processi geochimici regolano la mobilità e il destino dell’arsenico. Un’ampia discrepanza rilevata tra le misurazioni di As in situ e in laboratorio indica una forte partizione del metalloide sulla frazione solida. La modellazione geochimica mostra la tendenza a precipitare di alcune fasi solide responsabili dell’adsorbimento e co-precipitazione dell’arsenico: idrossidi di ferro ed alluminio e minerali argillosi. Infine una buona corrispondenza tra picchi nelle concentrazioni di As e di carbonio organico disciolto suggerisce che meccanismi di complessazione superficiale con acidi umici sono responsabili di una maggiore mobilità dell’arsenico. Sammanfattning Tillgångentill rent,säkert vatten är en utmaning på landsbygdssamhällen i många utvecklingsländer. Åtgång på förorenat vatten riskerar människors hälsa och skadar social och ekonomisk utveckling. Naturlig förekomst av arsenik (As) i grundvatten är ett globalt miljöproblem, vilket utgör en allvarlig risk för människors hälsa på grund av metalloidens höga toxicitet.

Med tanke på att arsenic sulfids mineraler är en viktigt del av guld insättning, har guldgruva aktiviteter anvisas som en orsak till att föroreningar av dränering och grundvatten i flera länder.

Problemet med As i dricksvatten har uppmärksammats i Tanzania för några år sedan och det krävs ytterligare undersökning för att möjliggöra tidig upptäckt av skadliga exponeringar.

Denna studie syftar till att bedöma förekomsten, källan och mobiliseringsmekanismerna för arsenik i vissa dricksvattenkällor i Lake Victoria Basin, i norra Tanzania. Landsbygdssamhällen som är kända för hantverksmässiga och storskaliga guldgruva arbeten var målet för den nuvarande studien.

Femtiofyra vattenprover samlades från källvatten, borehålsvatten, floder och grundbrunni Mara och Geita-regionen under oktober 2016. pH, redoxpotential (Eh), temperatur och elektrisk konduktivitet (EC) mättes i fält. Vattenprovernas koncentration av an- och katjoner, spårämnen (bl.a. arsenik), As(III) samt löst organiskt kol (DOC) analyserades i Sverige på Kungliga Tekniska Högskolan (KTH)

Femtiotre procent av det provtagna vattnet överensstämmer inte med WHO: s rekommenderade gräns på 10 μg / l, vilket utgör en allvarlig hälsorisk för vissa landsbygdssamhällen i Victoria-sjön.

Den geografiska fördelningen av As i det undersökta området är högst heterogen och påverkas huvudsakligen av lokal geologi och närhet till gruvplatserna (ca <5km). Lägre As-nivåer i borehål än i floder och grunda brunnar visar att föroreningar av dränering på grund av gruvverksamhet och föreslår att djupt grundvatten (> 40m) i allmänhet utgör en källa till säkrare dricksvatten.

Däremot, visar denna studie att flera geokemiska processer kontrollerar förutbestämmelse och rörligheten för As, när det har blivit frisläppts ut i vattenmiljön.

Stora skillnader mellan fält- och labbmätningar av As indikerar en stark partitionav metalloid i partikelfraktionen. Som avslöjas av geokemisk modellering antas, samutfällning med järn / aluminiumhydroxider och adsorption på lermineraler vara de huvudsakliga sänkorna för upplöst As. Dessutom antyder en bra matchning mellan toppar i As och upplösta organiska kolkoncentrationer att komplexbildning med humana och fulviska syror är ansvarig för förbättrad rörlighet. Acknowledgements

I would like to give my greatest thanks to all the people who supported me during the progress of this Master of Science Thesis, especially: my main supervisor Prof. Prosun Bhattacharya for invaluable and inspiring guidance throughout the whole project. His expertise and friendly attitude have helped me to successfully realize this Thesis work in a pleasant work environment. my supervisor in Italy Rajandrea Sethi for enlightening comments and for always showing prompt availability in discussing the Thesis with me. my local supervisor Prof. Dr.-Ing. Felix Mtalo for all the support and scientific advice during my staying in Tanzania.

I gratefully acknowledge “Åforsk Foundation” to have promoted this research through their financial support.

Special thanks also to the staff of UDSM, particularly to Mtamba and Ullomi for technical and organizational help during the fieldwork and to Mwangoge for being such a friendly driver. I felt warmly welcome at UDSM and I really enjoyed the two weeks of fieldwork in Mara and Geita, especially for the “nyama choma” and “Serengeti” nights. Thanks also to Stephen Magohe from Department of Geology for helping me with the geological maps of Tanzania.

I am grateful to the PhD students involved in the DAFWAT project: Julian for his valuable advice on statistics and spatial distribution, Regina, Fanuel and Vivian for great support in Tanzania and help in the chemistry lab.

I gratefully acknowledge Ezekiel and Agnieszka for a warm and friendly atmosphere in the water lab at KTH and for the great help with the water analysis.

Extra thanks to all the Master Thesis’ students, PhD students and professors of the engineering geology division at KTH for making the last months of this Thesis funnier and sweeter: Ricardo, Jenny, Kajsa, Srinidhi, Rajabu, Sara, Alireza, Caroline, Hedi, Xi, Flavio, Liangchao, Robert, Ulla Mörtberg and Bo Olofsson. Last but no least special thanks to my parents who gave me support and encouragement through all the Thesis work. List of Figures

Figure 1. Eh-pH diagram for the system As-O2-H2O at 25°C showing dominant dissolved species...... 5 Figure 2. Map showing predicted concentrations of As within the Lake Victoria Basin, in Tanzania. Ijumulana et al. (2016) ...... 12 Figure 3. Skin lesions associated with use of contaminated water around North Mara Gold Mine (Bitala et al. 2009; Evjen, 2011)...... 13 Figure 4. Maps showing Lake Victoria Basin boundaries, the two regions under investigation and the location of the gold mining activities targeted in the present study...... 15 Figure 5. Geological map of Lake Victoria Basin in Tanzania and location of Geita Greenstone Belt (GGB) and Mara-Musoma Greenstone Belt (MMGB)...... 18 Figure 6. Maps showing elevation and stream network in Mara (left) and Geita(right)...... 20 Figure 7. Small scale mining: (a) Mining pit (b) grinding mills (c) sluice box for panning (d) amalgamation pond ...... 24 Figure 8. Large-scale mining: (a) Open pit (b) Waste rock pile (c) Tailing storage facility (d) portion of the tailing dam...... 26 Figure 9. Source of drinking water in the study area: (a) Spring (b) borehole (c) stream (d) hand dug shallow well...... 27 Figure 10. Map showing location and type of the water sources considered in the present study...... 29 Figure 11. Arsenic test kit: comparison of colou rdeveloped on the paper strip with the reference colour chart...... 33 Figure 12. Box plot and key descriptive parameters...... 36 Figure 13. Number of samples in each sub-dataset...... 39 Figure 14. Box plots showing values of pH, Redox potential (Eh) and electrical conductivity (EC) characterizing different types of water source...... 42 Figure 15. Piper diagrams of the sampled water characterizing different locations (left) and types of water source (right). Red ellipse indicates a cluster of surface water samples with high

2- SO4 levels...... 43 Figure 16. Concentrations of major anions characterizing different types of water source...... 45 Figure 17. Concentrations of major cations in the four locations under investigation...... 47 Figure 18. DOC concentrations characterizing the four types of water source...... 49 Figure 19. Fe, Al and Mn levels in the sampled water considering both Mara and Geita Region ...... 50 Figure 20. Concentrations of Fe (top-left), Al (top-right), Mn (bottom) in unfiltered (red) and filtered samples (blue). The dot lines represent the median concentration of each dataset...... 51 Figure 21. As concentrations characterizing the four types of water source...... 53 Figure 22. Graph showing discrepancies between lab- and field-measured As. The green line represents percent difference...... 58

- Figure 23. Variation of HCO3 /SiO2 ratio across the four locations under investigation. Only groundwater samples are considered. BM: Butiama/Musoma; GR: Geita Rural; GT: Geita Town; NM: North Mara...... 61 Figure 24. Bivariate plots indicating typical ranges of carbonate and silicate weathering...... 62 Figure 25. Ternary diagrams for groundwater samples in Geita Town, Butiama/Musoma and North Mara...... 63 Figure 26. Correlations between silica and sodium in samples from Geita Region...... 64

2+ - Figure 27. Relationship of Ca with HCO3 and with pH...... 65 Figure 28. Stiff diagrams of representative groundwater samples characterizing the four locations under investigation. Samples ID 7 and 22 are from Geita region, samples ID 39 and 49 are from Mara region...... 66

- Figure 29. Correlation between HCO3 and DOC in surface water (yellow), shallow well (light blue) and borehole (dark-blue)...... 67 Figure 30. Relationship of DOC with (top-left) Eh; (top-right) Fe; (bottom-left) Zn; (bottom-right) As in borehole waters...... 68 Figure 31. Relationship of As with pH in surface water (yellow), shallow well (light blue), spring (green) and borehole (dark-blue) ...... 69 Figure 32. pH-Eh diagram for the As-O-H system with plot of the sampled water ...... 70

2- Figure 33. Correlation between As and SO4 characterizing different types of water source. Note that x-axis is in log scale ...... 71 Figure 34. Relationship of As with (left) Na+ and (right) Ca2+ in silicate (red) and carbonate (blue) rocks...... 72 Figure 35. Correlations between As and DOC characterizing different types of water source ... 72 Figure 36. Graph showing DOC levels (green line), field measurements of As (blue line) and lab measurements of As (red line) in the sampled water...... 73 Figure 37. Relationship between As and (top) Fe; (middle) Al; (bottom) Mn. Note the log scale in the x axis...... 74 Figure 38. Relantioship between As and Ba in Geita and Mara regions...... 75 Figure 39. Saturation indexes for selected solid phases controlling fate and mobility of As...... 77 Figure 40. Saturation indexes for selected solid phases providing indications about the local geology...... 78

List of Tables

Table 1. Overview of As concentrations in drinking water sources from selected parts in the world...... 3 Table 2. Meteorological data for two stations located near the study area. (Crul, 1995) ...... 19 Table 3. Source of data layers used in ArcGIS...... 37 Table 4. Summary of geochemical parameters for the entire dataset...... 41 Table 5. Distribution of As concentrations in the study area...... 52 Table 6. Difference between field measurements and laboratory results for As ...... 57 Table 7. Distribution of aqueous chemical species (%) in selected surface water samples. oxs stands for oxidation state...... 79

Abbreviations

As(III): arsenite, reduced form of arsenic

As(V): arsenate, oxidised form of arsenic

BH: Borehole

BIF: Band Iron Formation

BM: Butiama/Musoma

CN: cyanide

EC: Electrical Conductivity

Eh: Redox potential

GR: Geita Rural

GT: Geita Town

LVB: Lake Victoria Basin

MMGB: Mara-Musoma Greenstone Belt

NM: North Mara

NMGM: North Mara Gold Mine

GGM: Geita Gold Mine

GBB: Geita Greenstone Belt

UN: United Nations

SW: Shallow well

TZA: Tanzania

WHO: World Health Organization Table of contents

1. INTRODUCTION ...... 1

1.1 The Problem of Arsenic in Drinking Water ...... 1

1.2 Problem Definition ...... 2

1.3 Aim and Objectives of the Present Study...... 2

2. BACKGROUND ...... 4

2.1 Distribution of Arsenic in The Environment ...... 4

2.2 Sources and Geochemistry of As in Water ...... 4

2.2.1 Arsenic in natural water...... 5

2.2.2 Arsenic in mine waters ...... 6

2.2.3 Weathering of sulphide minerals ...... 7

2.2.4 Adsorption-desorption ...... 8

2.2.5 Reduction of Fe-hydroxides ...... 9

2.3 Health impacts ...... 9

2.4 Arsenic in Tanzania ...... 10

2.4.1 Health impacts cases in Tanzania...... 12

3. THE STUDY AREA ...... 14

3.1 General Characteristics of Lake Victoria Basin ...... 14

3.1.1 Population and economy ...... 15

3.2 Geology ...... 16

3.3 Topography and Hydrological Setting ...... 19

3.4 Hydrogeology ...... 21

3.5 Gold Mining Activities ...... 22

3.5.1 Small Scale Mining ...... 23

3.5.2 Large Scale Mining ...... 24

3.6 Drinking Water Sources ...... 26

3.6.1 National drinking water points map ...... 27

4. METHODOLOGY ...... 28

4.1 Sampling Strategy ...... 28

4.1.1 Location of the drinking water sources ...... 30

4.2 Sampling Methodology ...... 30

4.3 Field Investigations ...... 31

4.3.1 pH, EC, Temperature and Redox Potential ...... 31

4.3.2 Arsenic Test kit ...... 32

4.4 Laboratory analysis ...... 33

4.4.1 Major anions ...... 33

4.4.2 Major cations and trace elements ...... 34

4.4.3 Accuracy of major ions analysis ...... 35

4.4.4 Dissolved Organic Carbon (DOC) ...... 35

4.5 Data Management and Interpretation ...... 35

4.5.1 Excel ...... 35

4.5.2 R-software ...... 36

4.5.3 ArcGIS ...... 36

4.5.4 AquaChem ...... 37

4.5.5 PHREEQC and geochemical modelling...... 37

4.6 Limitations ...... 38

5. RESULTS AND DISCUSSION ...... 39

5.1 Results of Field Measurements and Laboratory Analyses ...... 40

5.1.1 pH, Redox potential and Electrical Conductivity ...... 42

5.1.2 Major ions ...... 43

5.1.3 Other major elements ...... 48

5.1.4 Dissolved Organic Carbon (DOC) ...... 48

5.1.5 Al, Fe, Mn...... 49

5.1.6 Arsenic ...... 52

5.1.7 Other trace elements ...... 59 5.2 Relations Between Different Geochemical Parameters...... 59

5.2.1 Local geology and solubility controls on the distribution of major ions . 60

5.2.3 DOC ...... 67

5.2.4 As ...... 69

5.3 Geochemical Modelling ...... 75

5.3.1 Saturation Indexes ...... 76

5.3.2 Aqueous species distribution ...... 78

6. CONCLUSIONS ...... 80

6.1 Final Recommendations ...... 81

BIBLIOGRAPHY ...... 82

APPENDIX A. Hydrogeological map of Tanzania ...... 87

APPENDIX B. Water sampling procedure ...... 88

APPENDIX C. As concentration maps ...... 90

APPENDIX D. Sampling location and results of field measurements ...... 93

APPENDIX E. Results of water analysis: major ions ...... 95

APPENDIX F. Results of water analysis: trace elements ...... 97

1. INTRODUCTION

Access to safe drinking water is a prerequisite for good human health and for a social and economic development. United Nations (United Nations, 2008) estimates that nearly a billion of people today do not have access to clean water and this figure is expected to increase in the near future because of greater pressures from population growth, climate change and environmental pollution. Projections in global population growth of 2 – 3 billion by 2050 forewarn an increased water demand for drinking purposes and food production, resulting in further exploitation of water resources. By altering the water cycle, global warming represents a serious threat to water quality and quantity, increasing water scarcity in many arid and semi-arid regions. Pollution from human settlements, industrial and agricultural activities is already impairing the quality of many water sources, particularly in developing countries where environmental regulations and awareness are often poor.

Overall, drinking water availability is expected to decrease in many parts of the world with a major stress on areas already affected by water scarcity. It is therefore crucial to globally assess the quality of drinking water sources prior to further exploitation, in order to reduce the risk for human health.

1.1 The Problem of Arsenic in Drinking Water

Arsenic naturally occurs in the environment. Because of its toxicity, sufficiently high concentrations of inorganic As in air, water and soil are harmful to the organisms (Smedley P. K., 2002). The main routes of As exposure to human beings are drinking water, ingestion of food prepared with contaminated water and consumption of crops irrigated with high As water. Among the various sources of drinking water, the highest concentrations of As have been reported in groundwater resources, posing a worldwide threat to human health (Nriagu et al., 2007). Chronic As poisoning has been associated with consumption of contaminated groundwater in several parts of the world: Argentina, Bangladesh, Chile, China, Ghana, Hungary, India, Mexico, Taiwan, Thailand and Vietnam (Bhattacharya et al., 2002).The most affected areas are developing countries, where lack of resources poses a limit to scientific research and,

consequently, to provide access to safe drinking water to the affected population. Table 1 presents a general overview of the problem of As in drinking water in some parts of the world, considering both surface and groundwater resources. In view of the growing health concern, the World Health Organization(WHO) has lowered the recommended value of As concentrations in drinking water from 50 to 10 µg/L in 1993. Whilst many national authorities have lowered their limit in line with WHO guideline values, in some countries, such as Bangladesh, India and Tanzania, the limit is still 50 µg/L .

1.2 Problem Definition

Although the problem of As has been globally addressed in the last decades, there are still areas of the world where it has not been fully investigated. An example is Tanzania, where the problem of As has been brought to attention only few years ago. Recently published studies have addressed the environmental impacts of gold mines and the As contamination of drinking water sources in parts of the Lake Victoria Basin (Ijumulana et al., 2015; Kassenga & Mato, 2008). However, there is a lack of a comprehensive research aiming to provide an exhaustive elucidation about As in mining areas and to enable an early detection of harmful exposures.

1.3 Aim and Objectives of the Present Study

The aim of this study is to investigate the occurrence of As in drinking water sources in the Lake Victoria Basin, in Northern Tanzania. Due to the well-known association between gold mining and As, the present study targets areas in the Lake Victoria region which are historically known for gold extraction at large and artisanal scale.

The objectives of the present study are to:

 Assess the quality of drinking water sources in some rural communities within Lake Victoria Basin.  Understand occurrence, source and mobility of As in gold mining areas.

Overall, this study should lead to a better understanding of the problem of arsenic in gold mining areas and encourage the development of solutions, which would provide safe drinking water to the affected population.

Table 1. Overview of As concentrations in drinking water sources from selected parts in the world.

Source of drinking Location As concentration (µg/L) Source of contamination Reference water

Bangladesh Shallow and deep wells <2 - 900 Reduction of Fe-oxyhydroxides in Naidu & Bhattacharya, alluvial sediments 2006

Reducing conditions in alluvial Taiwan Deep wells 10 - 1820 Smedley, 2002 sediments

Oxidation of arsenopyrite in mine Thailand Surface water 1 - 600 Wlliams, et al 1996 wastes

Dissolution of sediments of volcanic origin Bejarano & Nordberg, Argentina Shallow wells 10 - 4000 Desorption of As from Fe, Mn 2003 hydroxides

Oxidation-dissolution of volcanic Chile Surface water 30 - 3310 IARC, 2004 rocks rich in As

Oxidation of As-containing Mexico Shallow wells 50 - 1100 sulphides in sediments and mine Armienta et al.2007 wastes

Ghana Deep wells <1 - 141 Oxidation of arsenopyrite Smedley et al, 1996

Complexes of As with humic Hungary Deep wells <25 - >50 Varsányi &Fodré, 1991 substances

2. BACKGROUND

This Background chapter provides an overview of the geochemistry of arsenic (As) and of the problem of As in Tanzania. Sources, distribution and mobilization mechanisms of As in natural water are presented in the first two sections, while the health impacts due to As exposure are illustrated in the third subchapter. Finally, previous studies regarding As and drinking water quality in Tanzania will are reported.

2.1 Distribution of Arsenic in The Environment

Arsenic is a natural constituent of the Earth’s crust, with an average abundance of 2 mg/kg (Williams, 2001). From the bedrock, inorganic As enters into the environment through a wide range of natural processes such as weathering of rocks, volcanic eruptions, hydrothermal ore deposits and geothermal activities (Nriagu et al., 2007). Man has had an important role in accelerating the mobilization of As in air, water and soils through mining activities, combustion of fossil fuel, use of pesticides and industrial activities. A combination of natural processes and anthropogenic activities has resulted in As concentrations varying by more than four orders of magnitudes in environmental media. Concentrations in air range between 0.02 and 3 ng/m3 in rural areas, and up to 25 ng/m3 and 100 ng/m3 in cities and in proximity of industrial sites, respectively (Smedley, 2002). Most natural soils contain low concentration of As with background values ranging from 1 to 40 mg/kg; however human activities can cause severe soil contamination, raising the concentrations up to thousands of milligrams per kilogram. Natural occurrence of As in surface and groundwater vary from baseline values of 0.1-2 µg/L to gigantic concentrations of 5 mg/L, depending on geology and hydrology of the surrounding environment. (Gomez-Caminero et al., 2001)

2.2 Sources and Geochemistry of As in Water

Interaction of water with soil, bedrock and sediments is the primary cause for the release of As in natural waters. Once As has entered into the aquatic environment, many natural processes and conditions are responsible for its fate and mobility. As

shown in Table 1, the specific mechanism of As mobilisation varies depending on the location, mainly due to the influence of local climate and geology, as well as on the Redox condition of the environmental compartment. Consequently, extensively varying As concentration are found in surface water, shallow alluvial aquifers and deep hard rock aquifers. Among all the natural sources of contamination, weathering of sulphide minerals and reduction of Fe-hydroxides are considered the major ones.

2.2.1 Arsenic in natural water. Several processes and conditions control mobilization and speciation of As in natural waters: adsorption-desorption, biological activity and Redox conditions. Inorganic As is predominantly found in natural water in the form of oxyanion,as a pentavalent -n n- (arsenate, HnAsO4 ) or trivalent (arsenite, HnAsO3 ) specie. Under oxidizing conditions, the arsenate species are dominant, whereas under reducing conditions arsenite species predominate. Similarly, pH controls the protonation of As species, - 2- - determining the predominance of H2AsO4 and H3AsO3 over HAsO4 and H2AsO3 . Figure 1 shows the combined influence of Redox potential and pH on As speciation. The oxidation of As(III) to As(V) does not naturally reach completion and therefore co-occurrence of As(III) and As(IV) can be found in natural water.

Figure 1. Eh-pH diagram for the system As-O2-H2O at 25°C showing dominant dissolved species.

The aqueous chemistry of As in water substantially differs from most of the heavy metal and metalloids naturally occurring in the environment. Most of the trace elements, e.g. lead (Pb), zinc (Zn), cadmium (Cd), nickel (Ni), copper (Cu), cobalt (Co), are present in water solution in the form of cations , i.e. Pb2+, Zn2+,Cd3+, Ni2+, Cu3+,Co2+, whose solubility drastically decreases with increasing pH. In the pH range of natural waters, the presence of trace elements as dissolved species is strongly limited by precipitation of mineral phases, adsorption to metal oxides and clay, and by complexation to humic acids. In contrast, some elements, e.g. chromium (Cr), As, - uranium (U), vanadium (V), are mostly found in the form of oxyanions, i.e. CrxOyHz - - , HnAsO4 , HnUO3 , which are less strongly adsorbed as the pH increases. These oxyanion-forming elements can therefore persist as dissolved species in water at near-neutral pH values. Moreover, dissolved arsenic persists over different redox conditions. Under reducing conditions As(III) is less strongly adsorbed to metals oxides and clay minerals than As(V), in contrast with most of the oxyanion-forming elements, whose reduced species, e.g. selenite, molybdate, vanadate, are less mobile in reducing conditions. (Smedley, 2002; Manning & Goldberg, 1997). The anoxic conditions in subsurface environments and the greater mobility of As(III) contribute to making groundwater systems more at risk to the occurrence of high As concentrations than surface water (Bhattacharya et al., 2002).

2.2.2 Arsenic in mine waters Arsenic is found in more than 200 mineral forms, with arsenopyrite (FeAsS), realgar

(As4S4) and orpiment (As2S3) being the major As-containing primary minerals (Nriagu et al, 2007). Because of their capability to adsorb gold from hydrothermal fluids, these minerals are often associated with hydrothermal gold ores. Exploitation of gold deposits has therefore resulted in widespread As contamination of surface and groundwater systems in many parts of the world. Two are the main sources of As contamination in large scale gold mining operations. The first one is the acidic run-off originating from the waste rocks piles, where weathering of As-bearing minerals occur. When exposed to the atmosphere, the sulphide rich rocks undergo oxidation-dissolution, generating free acidity and liberating heavy metals. The oxidation products of the weathering process are washed away from the rock surfaces during heavy rainfall, forming the so called

“acid mine drainage”. This mine effluent is characterized by very low pH (2 ÷ 4) and concentration of dissolved heavy metals ranging up to several milligrams per litre. The second source of contamination is wastewater leaching from the tailing storage facility, also known as cyanide reservoir. During gold ore processing, which involves the use of cyanide solution, a fraction of the process water is discarded and stored in the tailing storage facility, that is an artificial reservoir surrounded by embankment dams. Seepage through the tailing dam is usually enriched in cyanide and heavy metals, and it is characterized by strong alkaline pH (9 ÷ 10). Williams (2001) has studied the occurrence of As in mine waters of 34 gold mines in Asia, South America and Africa with the aim of outlining the main factors controlling As mobility in mining areas. Although the liberation of dissolved As is a strongly Eh/pH dependent phenomenon, the hydro chemical study shows As mobility over a wide range of both pH and Eh, supporting the peculiar geochemistry of this metalloid. High As concentrations are associated with the generation of acid mine drainage (pH 2-5), but they have been recorded also in alkaline mine waters, with pH varying in the range 8-10. High concentrations of As in strongly alkaline waters has been explained by complexation with cyanide, seeping from the wastewater storing facilities. Another important control of As mobility in mine waters is the hydrochemistry of iron, because of the tendency of dissolved As oxyanions to be adsorbed on iron hydroxides. Co-precipitation of As with iron has been inferred to be a major sink of dissolved As in mining areas (Bhattacharya et al., 2002). Site-specific climate and mineralogy have also been recognized as important factors in the mobilization of As from mine wastes (Williams, 2001).

2.2.3 Weathering of sulphide minerals When arsenopyrite (FeAsS), and similarly other As-containing minerals, is exposed to water and dioxygen (O2), it is subjected to oxidation and dissolution processes, which leads to the release of As species and free acidity (H+). The complete reaction of arsenopyrite’s weathering can be represented as follow (Bhattacharya et al., 2002):

4() + 13 + 6 ↔ 4 () + 4 () + 4 () + 12

- 6+ 3+ 5+ Where the presence of O2 causes the oxidation of S to S and of As to As .

In natural conditions, iron (II) (Fe2+) can be further oxidised to Fe3+ and may precipitate depending on pH:

4 () + 10 + ↔ 4()() + 8

Adsorption of As on Fe(OH)3 and the subsequent co-precipitation has been reported as the principal sink for As in mining areas. (Williams, 2001). Important factors influencing the oxidation-dissolution of arsenopyrite are: (a) pH,

(b) dissolved oxygen concentrations, (c) Fe content, which can substitute O2 as oxidising agent, and (d) presence of bacteria, which can catalyse the oxidation. (Bhattacharya et al., 2002) Because of the oxic conditions of the surface environment, weathering of sulphide minerals is the main mechanism responsible for As contamination of surface water and shallow aquifers. Mining activities can accelerate the release of naturally occurring As by exposing huge amount of As-bearing ore and gangue minerals to the atmosphere.

2.2.4 Adsorption-desorption The mobility of As, and of most trace metals, in natural water is largely controlled by sorption processes. Ion exchange and surface complexation are the main mechanisms responsible for the sorption of heavy metals on variable charge surfaces of oxides and organic matter (Appelo & Postma, 2005). The surface charge is influenced by pH and solution composition, making the solids more likely to adsorb positively charged ions (at high pH), such as Cr3+ and Cu2+, or oxyanions (at low pH), such as 2- - H2AsO4 and H2PO4 .

The main solid phases that are able to adsorb arsenic species are iron (Fe), aluminium (Al), manganese (Mn) oxides/hydroxides and clays. The adsorption capacity depends on:

. pH. Arsenic is generally less strongly adsorbed as pH increases, because of 2- - its tendency to be present as oxyanion (HAsO4 , H2AsO3 ) in natural water; . Redox conditions, which affects the As speciation and the stability of the oxide/hydroxides. As(V) is generally more strongly bound than As(III), however at high pH As(III) has higher affinity with goethite and clay minerals (Manning & Goldberg, 1997; Appelo & Postma, 2005).

. Intrinsic characteristics of the solid phase, mainly due the point of zero recharge (PZC), which determines whether the surface is positive or negative charged. 3- . Presence of competitive ions, such as phosphate (PO4 ), molybdate (MoO4) 2- and sulphate (SO4 ).

2.2.5 Reduction of Fe-hydroxides Reduction-dissolution of Fe-hydroxides has been recently identified as a source of As contamination in groundwater systems. (Smedley et al., 1996; Anawar et al., 2003; Bhattacharya et al., 2002). When sediments rich in Fe-hydroxides and organic matter are buried, strong reducing conditions develop in the subsurface environment. This process causes the reductive dissolution of Fe-hydroxides on which As is adsorbed, leading to the release of the metalloid, as shown by the following reaction:

1 1 7 + () − + 2 ↔ + + + 4 () 4 4 () Because of the reducing conditions, As is predominantly found in the reduced form, As(III).

2.3 Health impacts

Arsenic has toxic effects on human health and is a well-documented human carcinogen. Among all the As compounds, the soluble inorganic form is the most harmful one, with As(III) being 80 times more toxic than As(V). Acute effects due to the ingestion of large doses disturb the gastrointestinal, cardiovascular and nervous systems, leading eventually to death. However, the As concentrations that naturally occur in drinking water do not expose people to the risk of acute poisoning (Gomez- Caminero, 2001). On the other hand, links have been observed between long-term exposure to As through drinking water and increased risk of cancer in skin, lungs, bladder and kidney. Cases of skin cancer after exposure via drinking water were published from Argentina, Chile, Mexico and Taiwan (Zaldivar, 1974). Because of the ability of As to accumulate in the organisms, symptoms of chronic exposure might manifest after several years from the ingestion of contaminated water. Arsenic is circulated around the body through the blood supply and it bio accumulates in hair,

nails and skin. Presence of As in human tissue is therefore a good indicator of prolonged As exposure.

2.4 Arsenic in Tanzania

There is a lack of systematic data about quality of drinking water sources in Tanzania. Albeit some studies have assessed the potentiality of groundwater for drinking and agricultural purposes, there is little knowledge about the occurrence of trace elements in groundwater systems, which can seriously impair access to safe drinking water (Sangea et al., 2016).

Groundwater quality with respect to geogenic contaminants, mainly fluoride (F-), salinity and to lesser extent As has been assessed in the Tanzanian regions located along the East African Rift Valley. (Shedafa & Johnston, 2013). Elevated F- concentrations were reported, whereas As was detected in concentrations below the WHO guideline values of 10 ppb. However, primary and secondary data were not collected from the well-known mining areas, where mining related As contamination may occur.

Because of the expansion of the mining sector at large and small scale, there has been a growing interest in assessing the environmental and social impacts of gold mining activities in Tanzania in the last two decades. Mercury (Hg) and heavy metals contamination of surface water and soil has been associated with small and artisanal mining activities in many rural communities within the Lake Victoria Basin (Kahatano & Mnali, 1995) (van Straaten, 2000). These studies report elevated concentrations of trace elements in localized hot spots in close proximity to mining and processing sites. The concentration levels usually decrease to background values away from the gold mining activities, suggesting that human activities have an important role in the mobilization of trace elements. Nyanza et al. (2014) presented the spatial distribution of As in water sources used for human consumption in Rwamagasa village, a community in Geita region with a long history of artisanal gold mining. In areas where no mining activity takes place, concentrations of As are relatively low, (< 1 µg/L ), whereas in proximity to the mining sites levels of up to 110 µg/L were recorded. It is however unclear which are the main factors affecting

the spatial distribution of As and how much As contamination is severe in other small scale mining areas.

Since large-scale mining started in Tanzania at the end of 20th century, few studies have been carried out on the environmental impacts of large gold mines. The most extensive investigation has been published by Almås and Manoko (2012) who analysed trace elements in samples of water, sediments and soils taken in proximity of North Mara Gold Mine (NMGM) and Geita Gold Mine (GGM), the two largest gold mines in Tanzania. Trace element concentrations at GGM were generally low, however high concentrations of Cu, Co and Ni were found at one sampling point located near the waste rock piles. In contrast, a severe environmental impact was identified at NMGM . Gigantic concentrations of As (up to 8 mg/L) and other heavy metals ( U, Zn, Ni, Cu, Cr, Co, Cd) were found in ponds and streams downstream to a leachate pond, where the acid mine drainage from the waste rock piles is collected and stored. In addition, very high As concentrations (up to 300 µg/L ) were reported in proximity of the tailing storage facility, indicating that continuous or sporadic wastewater leaching had occurred. Since it was not in the scope of this study to assess whether the contaminated effluents from the mining sites had impaired the quality of the drinking water sources used by the local communities, further investigations are needed.

A systematic assessment of the concentration levels of As in some drinking water sources in the Lake Victoria Basin was made by Kassenga and Mato (2008) . The pioneering study, which included sampling of 96 drinking water sources, reported that 41% of the water sources exceeded the WHO guideline value of 10 µg/L . Based on these data, Ijumulana et al. (2016) has generated As prediction map, Figure 2, which highlights a large spatial variability of As over the investigated regions and within each district unit.

µg/L

Figure 2. Map showing predicted concentrations of As within the Lake Victoria Basin, in Tanzania. Ijumulana et al. (2016)

The coverage of this assessment is not complete for all the Lake Victoria Basin, and some districts well known for gold mining activities were not investigated. Moreover, the study did not aim to fully understand the sources of As and its mobility, suggesting that a more detailed geochemical investigation is required.

2.4.1 Health impacts cases in Tanzania Human exposure to As has been cause of concern only recently in Tanzania. A part from hospital reports showing cases of skin cancer, respiratory and gastrointestinal problems in several mining areas around the Lake Victoria Basin, no epidemiological studies have been done to assess whether these diseases are associated with prolonged As exposure.

Life threatening diseases were reported in some villages surrounding North Mara Gold Mine after an acid spill in May 2009 contaminated streams and rivers, commonly used as source of water for domestic and drinking purposes. Cases of skin

disorders (Figure 3) were reported among the villagers and they have been associated with use of water contaminated by acid mine drainage.

Figure 3. Skin lesions associated with use of contaminated water around North Mara Gold Mine (Bitala et al. 2009; Evjen, 2011).

A biological investigation was carried out by Evjen (2011) who collected samples of hair, nail and blood from 63 subjects living in the villages around North Mara Gold Mine. Content of As and other trace elements (Pb, Sb, Zn, Mn, Mo, Cu, Th) were measured in human tissues in order to identify whether the villagers are exposed to high trace element concentrations. The study shows higher concentrations of As, Mn and Th in hair and nails in the villagers around NMGM than in a reference group from Dar Es Salaam. These concentrations might be associated with other health effects caused by As toxicity. It is therefore important to identify the sources of As exposure in the surroundings of the gold mine.

3. THE STUDY AREA

This chapter gives a general overview of the area under investigation. Firstly, general characteristics of Lake Victoria Basin are given. Then, a description of geology, hydrology and hydrogeology is provided. Finally, methods used in large and small scale and mining activities are presented.

3.1 General Characteristics of Lake Victoria Basin

The Lake Victoria Basin is located in the upper reaches of the Nile River Basin and occupies a total area of about 251,000 km2, extending between latitude 1° N and 3° S. The lake itself is the second largest freshwater lake in the world with a surface area of 69000 km2, shared among Tanzania (51%), Uganda (43%) and Kenya (6%) (UNEP, 2006). Favourable climate conditions for agriculture and livestock and the abundance of natural resources have supported the growth of one of the densest and poorest rural population on earth, counting over 35 million people.

The present study focuses on two of the five Tanzanian regions sharing the Lake Victoria Basin: Geita and Mara region, Figure 4. The first one occupies part of the southern shores of the lake, whereas Mara region is situated to the East and it borders with Kenya to the North.

Figure 4. Maps showing Lake Victoria Basin boundaries, the two regions under investigation and the location of the gold mining activities targeted in the present study.

3.1.1 Population and economy The rural population in Lake Victoria Basin is mainly constituted by indigenous people of diverse ethnic groups, whereas a composition of indigenous people and settlers, especially Arabs and Asians, is found in the urban centres. Around 75% of the population lives under US$ 3.10 a day and the average adult literacy rate is 70% (The World Bank , 2016). Fishing, subsistence agriculture and pastoralism are the main sources of livelihoods followed by other economic activities such as quarrying, trading and mining. Tourism is growing fast and it has the potentiality of becoming a major economic sector.

The Lake Victoria and its basin are endowed with abundant natural resources with water, fisheries, and terrestrial biodiversity being the major ones (UNEP, 2006). Nevertheless, large exploration of ore, oil and gas deposits have started in the second half of 20th century thanks to foreign investments, boosting oil extraction and mining activities. The principal minerals extracted are gold and diamond followed by copper and lead.

The rapidly growing population and the interests of foreign investors are causing an uncontrolled exploitation of the natural resources with several implications. Land and

natural resource-related rights are the drivers of many conflicts occurring at country and community levels. Environmental degradation is accelerating and it is leading to an increased exposure of the population to anthropogenic and natural hazards (UNEP, 2006). These factors cause an increment of poverty and a decline in the population’s health and in food security. Recent discoveries of new ore and oil deposits in Uganda and Kenya prefigure further exploitation and an increasing human vulnerability in the basin.

3.2 Geology

The geology of the Tanzanian Lake Victoria Basin consists of Archean granitoids- greenstone belts hosted in the Tanzania Craton. The stratigraphy of the Tanzanian Craton comprises three main geological units: the Dodoma Supergroup, the Nyanzian successions and the overlying Kavirondian system. The Dodoma supergroup, which is constituted by high-grade metamorphic rocks, forms the basement to the Nyanzian greenstone belts sequences. The Nyanzian unit is unconformably overlain by Kavirondian rocks, which are dominated by coarse clastic sediments and associated with horizons of volcanic rocks (Sanislav et al., 2016). Because of the heterogeneous distribution of this geological unit, outcropping of Nyanzian belts are widely present across the Tanzania craton. The greenstone belts are laterally enveloped by granitic, gneissic and migmatitic rocks occurring in the Tanzania cratons and along the shield areas, Figure 5. Finally, Quaternary alluvial and eluvial sediments have deposited around Lake Victoria´s shorelines, developing the largest deposits in proximity to the alluvial plains of Mara and Grumeti rivers in Mara region.

After their formation during the Archean era (3600-2500 mln), the greenstone belts have been subjected to subsequent geotecnic factors, leading to a continuous evolution. Widespread deformation and metamorphism resulted in an enrichment of lithologies and in the deposition of diverse rock types. The main lithologic units found in the Northern Tanzania greenstone belts are felsic volcanic rocks, ferruginous rocks (Banded Iron Formation) and meta-sediments (Kabete et al. , 2011).

Within the greenstones-BIF series, epigenetic mineralization has occurred in fold and fracture zones, leading to the formation of gold deposits, which are found in different geological settings (Anhaeusser, 2014): (1)Auriferous quartz veins (Au-quartz reefs). These deposits occur as small-to large- quartz veins filling simple fractures in meta- volcanic rocks and it consists of gold hosted in sulphide minerals including pyrite and arsenopyrite. (2)Mineralized shear zones. In this setting, gold and sulphide hosted gold occur in large shear zones in BIF and meta-sediments. (3) Disseminated deposits. These deposits consist of gold-sulphide impregnations in various host rocks (BIF, tuff) in carbonatized shear zones (Kahatano & Mnali, 1995). Most of the shear and fractures zones where gold mineralization occurs are located along the contact zones between the greenstone belt and the surrounding gneiss/granitoid units, as shown by the location of gold mines in the geological map, Figure 5.

MMGB

GGB

Figure 5. Geological map of Lake Victoria Basin in Tanzania and location of Geita Greenstone Belt (GGB) and Mara-Musoma Greenstone Belt (MMGB).

The artisanal and large scale gold mining activities considered in the present study insist on two Nyanzian greenstone belts: (1) Geita Greenstone Belt (GGB) and(2) Musoma-Mara Greenstone Belt (MMGB) (Figure 5).

Geita Greenstone Belt constitutes an E–W trending segment of greenstone units situated south of Lake Victoria. To the north, west and east the greenstone is bounded by granitoids while to the south it comes in contact with gneiss along a E-W trending shear zone. The GGB unit consists of banded ironstones intercalated with sedimentary units, pillowed basalts and intruded by diorite and granitoids. The largest gold deposit is Nyankanga, where Geita Gold Mine was established in 1998. A detailed description of the geological setting of this deposit is provided by Sanislav

et al. (2014). The deposit is hosted in a magnetite-rich sedimentary succession, constituted by series of sandstones, chert, turbidities and intruded by a group of intrusive rocks, mainly diorite, feldspar and quartz. The gold mineralization is hosted within a NW dipping deformation zone, which has developed along the ironstone- diorite contacts. Gold occurs mostly as disseminated sulphides in diorite and replacement textures in ironstones. The MMG Belt runs in SW-NE direction in the eastern part of Lake Victoria. It comprises Nyanzian sequences of mafic and felsic volcanic rocks overlain by Kavirondian sedimentary rocks and it is intruded by granites, gneisses and amphibolites (Gray & Macdonald, 1964). The main geochemical features of MMGB are high-Mg andesite and Na-K granites (Manya & Maboko, 2015). Gold deposits are widely distributed along this greenstone belt, with the major ones located in the Mnanka area, where North Mara Gold Mine has started activities in 2002. According to Kavana (2015), the Mnanka area is characterized by a sequence of rhyolitic volcanic rocks, chert and meta-sediments, with rhyolite, diorite, phonolite and andesite being the main lithological units. The gold mineralization is associated with andesite and rhyolite and occurs as gold nuggets and fine disseminations in sulphides.

3.3 Topography and Hydrological Setting

The Lake Victoria Basin is characterized by an equatorial hot and humid climate with a bi-modal rainfall pattern with long rains from March to May and short rains from October to December (FDMT, 2016). Table 2 shows climatologic elements for two meteorological stations located in Mara and Geita regions. Table 2. Meteorological data for two stations located near the study area. (Crul, 1995)

Region Rainfall(mm) Temperature (°C) Evaporation(mm)

Mara 900 14 - 28 -

Geita 1100 17-28 2100

River regimes are strongly influenced by the climate conditions. They are characterized by two annual peaks, the first one in January/February and the highest one in April. Lakes’ levels start rising in November/December and they generally reach the annual peak in April-May before a dry period from June to October. Figure 6 shows elevation and stream networks in Mara and Geita region. The upper part of Geita region drains into Lake Victoria, whereas the lower part drains SW into Lake Tanganyka through Nikonga river. Because of the flat topography and the vicinity to the Lake Victoria shorelines, most of the streams in the upper part of Geita separately flow into wetlands before discharging into Lake Victoria. In contrast, the whole Mara region drains into Lake Victoria and its elevation gradually decreases from the East African Rift Valley to the lake’s shores. Most of the streams originating in the East part of the Region drain into major rivers, with the biggest ones being Mara and Grumeti, before discharging into the lake.

Figure 6. Maps showing elevation and stream network in Mara (left) and Geita(right).

The main land cover in the study area is open grassland, followed by woodland and wetland. The latter one is widely distributed along the Lake Victoria shorelines and in proximity of the outlet of major rivers. One of the largest wetland is fed by Mara River, which origins in Kenya and drains a large part of Northern Mara Region before discharging into the lake. Subsistence agriculture and small-scale livestock farming largely predominate over any other land use. Although mining is not a major land use, it is extensively practiced, especially at artisanal scale (GeoNode, 2015).

3.4 Hydrogeology

There is limited and inadequate information about near surface geology and hydrogeology of Tanzania. Few studies have been carried out with the aim of identifying the aquifer systems and their potentiality for drinking purpose. In most of the cases, the only available information has been extracted from existing boreholes data providing a broad classification of the aquifer formations at regional level. Based on geologic map and boreholes database, the Southern African Development Community has published in 2009 one of the most recent hydrogeological map of Tanzania, providing an overview of type and productivity of the main aquifers, at 1:5 000 000 scale, see Appendix A (Sangea et al., 2016). Two types of aquifers have been identified in the portion of the Lake Victoria Basin considered in the present study: Precambrian basement formation and unconsolidated deposits. An attempt to systematically assess the water resources in rural areas of Geita and Mara regions has been made by the Ministry of Water in collaboration with the Japan International Cooperation Agency in 2005 and 2006. One of the outcomes has been a well inventory reporting location of deep and shallow wells in 11 rural districts around Lake Victoria. Information about depth of the well, water static level, subsurface geology and field measurements (pH and conductivity) are also available in some cases. Although the coverage of well inventory is heterogeneous, considerations can be made about the groundwater systems in the area under study:

(1) The dominant water bearing formations in Geita and Mara region are weathered and fractured basement rocks. Groundwater potential depends on degree and depth of the weathered zone, resulting generally in discontinuous

aquifers with low-moderate productivity (0.5 - 2 L/s). Boreholes drilled in this category usually reach a depth varying between 40 and 80 m and they present a typical stratigraphy of unconsolidated superficial deposits, weathered/fractured rocks and solid bedrock. Recharge occurs through fractures zones and faults. Isolated shallow aquifer systems can be found in the unconsolidated material overlaying the weathered bedrock and they are often exploited through hand-dug wells. However, the productivity of these shallow wells is usually low and affected by season variability. (2) In proximity of the Lake Victoria’s southern and eastern shores, there are thick deposits of alluvial and lacustrine Quaternary sediments, with gravel and sand being the main lithologic units. Aquifers have developed in these formations, usually leading to high productivity (5 - 20 L/s). Boreholes drilled in this category generally present a depth up to 70 m and a homogenous stratigraphy.

3.5 Gold Mining Activities

The mining sector is a major component of the Tanzanian economy, accounting for approx. 4% of GDP and close to half of the country’s exports. Gold is the leading mineral, showing an increase of exports from US$39.8 million in 1999 to US$ 504.1 million in 2003, followed by diamond and copper. (United Nations, 2008) The first gold discoveries were made by Germans during the colonial times with mining activities starting as early as the 1890’s. However, mining had been done only at artisanal and small scales until the country´s economic reforms in 1980th and the new mining policies in 1997, which boosted new local and foreign private investments. Consequently, hundreds of new mining licenses were issued to local and foreign companies and six large scale gold mines were established between 1999 and 2007. Among these mines, all owned by foreign companies, two are considered in the present study: the Geita Gold Mine and the North Mara Gold Mine. Gold exploration is still active resulting in continuous discoveries of high potential prospects and in a steady increase of foreign investment. The largest and richest gold mineralization in Tanzania occurs within the Lake Victoria Gold Field greenstone belts, where surface and subsurface mining is

conducted both at small and large scale. Methods of extraction and processing of gold greatly differ between small scale and large-scale mining.

3.5.1 Small Scale Mining Artisanal mining is performed through excavation of the gold bearing rocks in shallow and deep pits, which are haphazardly dug in the claim area. The deeper pits are usually timbered (Figure 7a) and dewatering is normally done by buckets or by using small water pumps. Once the rocks are hauled to the surface, they are manually crushed using shovel, chisel and hammer before being fed into the grinding mills (Figure 7b). The finely crushed material then undergoes concentration stages of gold recovery, which normally consist of panning and amalgamation. During panning (Figure 7c) a slurry of mined ore and water is passed through an inclined sluice box whose bottom is covered with a carpet that traps the heavy gold particles. The carpets are then washed in a bucket of water to collect the concentrate and mercury is added for the amalgamation stage (Figure 7d). By mixing, often barehanded, the mercury into the concentrate, mercury-gold alloys form and precipitate at the bottom of the bucket. Finally, the amalgam bullion is heated in open air to evaporate the mercury and recover the gold. Rudimentary equipment and harmful working conditions characterize small scale mining, which is usually within rural communities in rural areas. The discovery of a new ore deposit triggers a mine rush and the consequent establishment of settlements, which are usually overpopulated and in poor hygienic conditions. The miners are mostly young, uneducated and without a basic knowledge on geology and mining. The rudimentary level of the artisanal mining poses a serious health hazard to miners and its effects on the environment are severe. The direct contact with mercury onto the skin and the inhalation of vapours and dust during the ore processing seriously harm miners’ health. Moreover, mine wastes are dumped in unprotected piles and the amalgamation ponds often over-flood during heavy rainfall. This causes heavy metal contamination of nearby surface and groundwater, which are often used as drinking water sources by the local communities (Kahatano & Mnali, 1995).

A B

C D

Figure 7. Small scale mining: (a) Mining pit (b) grinding mills (c) sluice box for panning (d) amalgamation pond

3.5.2 Large Scale Mining Gold mining at large scale requires huge initial investments, expertise and modern technology in order to be economically feasible also in area with low-grade ore deposits. Moreover, large mines depend upon the availability of natural resources, especially water and land. The principles of extraction and processing of gold ores are similar all over the world, however methods implemented are site specific and controlled by the type of gold mineralization. The mining methods used in North Mara Gold Mine and Geita Gold Mine are illustrated in Figure 8. The main technique used for the extraction of the hard rock is blasting in open pits (Figure 8a). The ore and the waste rocks are mined separately and hauled to the primary crashers by dump trucks. After rough crushing the overburden is disposed in dumps (Figure 8b) or used to build embankment structures, whereas the ore proceeds firstly to the secondary crashers and then to a grinding mill with lime, water and steel balls, where the size of the ore is reduced to a fine dust. The addition of lime creates alkaline conditions, preventing the formation of cynic gases during the leaching

circuit and in the tailing storage facility. The coarse gold particles are removed by gravitation, whereas a slurry of fine ore particles and water flows into flotation tanks where a mineral rich foam concentrate is formed using chemical conditioning reagents followed by agitation and air sparging. The floatation concentrates are fed in steel agitating tanks where they undergo the so-called carbon-in-leach process. It consists in the simultaneously addition of weak cyanide solution and activated carbon in presence of oxygen and lime. The cyanide dissolves the gold from the solid matrix, forming gold cyanide complexes from which gold is extracted by adsorption on the activated carbon. The loaded carbon is then physically separated from the slurry and, after acid washing, it is sent to the elution column for desorption of gold cyanide and the regeneration of the carbon. The rich eluate solution is passed through electro winning cells where gold and other metals are precipitated onto the cathodes. The loaded cathodes are rinsed to release a gold bearing sludge, which is dried and melted in furnaces. The melted material is poured in a cascade of moulds producing bars of doré bullion, which are flown to smelters in Europe. The tailing slurry from the carbon-in-leach process undergoes thickening for the recovery of process water and the remainder is finally disposed into the tailing storage facility (Figure 8c) which is confined by a tailing dam (Figure 8d). In view of the displacement and processing of huge amount of rocks involved in large-scale operations, a proper management of solid and liquid wastes is essential to prevent extensive contamination of the surrounding environment. Effective collection and treatment of mine effluents must be assured before discharge in rivers and streams.

A B

C D

Figure 8. Large-scale mining: (a) Open pit (b) Waste rock pile (c) Tailing storage facility (d) portion of the tailing dam.

3.6 Drinking Water Sources

Tanzania faces a serious constraint in providing access to safe drinking water. Only 46% of the rural communities have access to improved drinking water services (UNEP, 2006). The biggest challenges facing the drinking water sector include inadequate sanitation, lack of technologies, poor management, and uncontrolled pollution, especially in mining areas (Ministry of Water, 2013). Drinking water is withdrawn from a variety of sources depending on local availability and level of development. The majority of the rural water supplies are based on drilled shallow (< 20 m) and deep wells (30 – 70 m) on which hand pumps are installed (Figure 9b). In most of the cases well’s functioning is irregular because of low aquifer productivity, pump’s breakage and well’s damages. (Ministry of Water, 2013). These factors, combined with a rapidly increasing population, have led to an increased exploitation of shallow groundwater through excavation of hand dug wells (Figure 9d). These wells are usually dug to a depth of 10 m deep by private

owners for domestic water supply and water is withdrawn by buckets. Springs (Figure 9a) and rivers (Figure 9c) are also used for drinking purposes in rural communities where groundwater is not easily accessible. Considering the variety of water sources commonly used by the local population, the water sampling involved collection of samples from groundwater (spring, shallow and deep well) and surface water (rivers, lake) resources.

A B

C D

Figure 9. Source of drinking water in the study area: (a) Spring (b) borehole (c) stream (d) hand dug shallow well.

3.6.1 National drinking water points map The Government of Tanzania, through the Ministry of Water, conducted an extensive study during 2013, with the aim of collecting baseline information about water supply coverage in the country. This generated a National Water Point Map, showing location of more than 65000 drinking water sources in urban and rural areas of Tanzania. With regard to Mara and Geita regions only 50% of the mapped water sources are reported to be functioning, of which the majority are shallow well (44%), followed by boreholes (26%), rain harvesting (13%), surface water (7%), spring (6%) and others (2%).

4. METHODOLOGY

This chapter describes methods used during the progress of the present study. Firstly, the strategy behind the water sampling and the work procedure will be described. Secondly, an overview of field and laboratory measurements will be given and finally, tools and software implemented for data interpretation and analysis will be listed.

4.1 Sampling Strategy

Few studies have been carried out in Tanzania about As exposure in rural communities. Accordingly, a water sampling campaign was conducted in the Lake Victoria Basin during October 2016, with the aim of assessing the occurrence of As in drinking water sources, as it represents the greatest threat to public health. Field measurements were performed and water samples were collected from sources of water used by rural communities for domestic and drinking purposes. Due to the well-known association between As and sulphide-gold ores, areas known for artisanal and large scale gold mining activities were targeted in the present study.

The water sampling strategy was developed based on the objectives of the Thesis:

(a) Evaluate the As exposure of rural communities within the Lake Victoria Basin. Water sources were therefore selected from 2 different regions and 5 different districts.

(b) Understand source and mobilization of As in water. This required a denser sampling in areas affected by high As concentration.

In co-operation with the local supervisors, four areas were finally selected as the targets of the water sampling (Figure 10): (1) artisanal mining in Geita Rural District; (2) artisanal mining in Musoma and Butiama Districts; (3) Geita Gold Mine; (4) North Mara Gold Mine.

Figure 10. Map showing location and type of the water sources considered in the present study.

4.1.1 Location of the drinking water sources A preliminary selection of the drinking water sources to be investigated was made before the actual field trip using the National Water Point Map. During the fieldtrip, support was received from local water authorities, whose personnel are familiar with the study area and the location of major water sources. The sampling campaign, conducted during October 2016, covered five districts in two different regions: Geita Town and Geita rural in Geita region, Musoma Rural, Butiama and Tarime in Mara region (Figure 10). A total amount of 54 water sources were surveyed during the 15 days of the campaign, including five different types of source: surface water (n=15) shallow well (n=12), springs (n=8) and boreholes (n=19). The location of the sources was determined by using the GPS receiver Mobile Mapper 20, developed by Spectra Precision. Latitude, longitude were recorded in the WGS 1984 Geographic Coordinate system and altitude was given in m. The water sampling was conducted in mid-October, hence at the beginning of the short rainy season.

4.2 Sampling Methodology

The work procedure at each water source followed a pre-defined protocol (Appendix B), which has been developed in accordance with previous scientific studies focusing on arsenic (Bhattacharya et al., 2002; Sracek et al., 2005; Bednar et al., 2002) and with groundwater sampling protocols (Triplett, 2006)

Three water samples were collected for laboratory analysis:  50 ml filtered sample for analysis of major anions, i.e. chloride (Cl-), nitrate - 2 - (NO3 ), sulphate (SO4 ) and fluoride (F ).

 50 ml filtered and acidified sample (using suprapure HNO , 70%) for dissolved organic carbon (DOC), major cations, i.e. calcium (Ca2+), potassium (K+), sodium (Na+), magnesium (Mg2+) and trace elements analysis.  25 ml filtered, acidified and As speciated sample. Arsenic speciation was performed using Disposable Cartridges® (MetalSoftCenter, PA). As(V) in the water sample is absorbed by the cartridge, while arsenite, As(III), species remain in the filtrate. The As speciation allows the determination of both arsenate and arsenite in the water sample.

Surface water was sampled using a 1-L polyethylene bottle and it was successively drawn into the 60 mL syringe. When sampling hand-pumped well, water was allowed to flow for some time before collecting the actual sample, in order to get representative water from the aquifer.

Field filtration was done using a 0.45 µm filter to remove suspended solids and colloids which can release or absorb dissolved species altering the free ion activities of the elements present in the water. At each sampling site, one water sample was acidified with suprapure HNO until pH was approx.< 2 in order to stop most of bacterial growth, block oxidation reactions and prevents adsorption or precipitation of cations. All the sampling and storage bottles were kept air tight before the sampling and were rinsed with sample water before the actual sampling. Finally, samples were labelled on spot. During the field trip samples were kept cool by placing them in ice chest and once back in the lab were stored in a refrigerator at 4° C before analysis.

Seven samples were not filtered in the field because of the high content of solids, which made unfeasible the filtration trough the 0.45 µm filter. These samples were therefore filtered once back in the lab. Nevertheless, these samples were acidified in situ. Difficulties were also encountered in keeping the cold chain between the sampling and the analysis, especially during the transport from the field to the lab and the shipment to Sweden.

4.3 Field Investigations

4.3.1 pH, EC, Temperature and Redox Potential pH, water temperature, electrical conductivity (EC) and Redox Potential (Eh) were measured in situ immediately after the sampling. Measurements were done by the use of HQ440 dual input radiometer coupled with:  pH electrode (IntelliCAL™ PHC201) for pH and Temperature.  Conductivity electrode (IntelliCAL™ CDC401).  Standard Gel Filled ORP Electrode (IntelliCAL™ MTC101) equipped with a 3M Ag/AgCl reference electrode.

To assure a good quality of the measurements, field equipment was calibrated before the sampling campaign and it was regularly tested against standard solution to ensure the calibration. The electrodes were rinsed with de-ionized water before and after use and were appropriately stored.

4.3.2 Arsenic Test kit The Low Range Arsenic Kit, developed by Hach, was used at each water source to quantify the trace amounts of total inorganic As. This field method allows the detection of As concentrations in the range 10 - 500 µg/L directly on site, giving a first indication of the contamination of the study area. This information resulted particularly valuable during the fieldwork since few data are available on the spatial distribution of As in the Lake Victoria Basin. The sampling strategy was therefore daily adjusted, based on the preliminary As concentrations. The field test kit includes five powdered reagents, As test strips and a reaction vessel. The procedure was as follows: the reaction vessel was filled with 50 ml of water sample; the powdered reagents were added one at a time and the vessel was gently swirled to allow mixing; the test strip was placed in the cap of the vessel and the vessel was then capped and left to react for 30 - 35 minutes. The arsine gas generated

(AsH3) during this time interval reacts with the test strip in the cap causing a change in colour depending on the As concentration. The test strip was then removed and the developed colour was immediately compared to the reference colour chart, see Figure 11. A light yellow colour corresponds to low concentration (10µg/L ), whereas dark orange indicates high concentrations (500 µg/L ).

Figure 11. Arsenic test kit: comparison of colour developed on the paper strip with the reference colour chart.

4.4 Laboratory analysis

The laboratory analysis were conducted at KTH-Royal Institute of Technology and at Linköping University, in Sweden. The samples were analysed for major ions, F-, 3- PO4 , DOC and trace elements, including As (V) and As(III).

4.4.1 Major anions

4.4.1.1 Alkalinity Alkalinity was measured in the filtered samples using Mettler Toledo Titrator. The method consists of titrating 10 ml of sample with 0.02 M HCl () to pH=5.4. The output from the instrument is the volume of HCL ( ) necessary to bring the sample to pH 5.4. Alkalinity was then calculated in accordance to the following equation:

∗ =

Between pH 6.3 and 10.3 the predominant specie in the measured alkalinity is - bicarbonate HCO3 , whose concentration can be calculated by multiplying alkalinity - by 61 mg/L, which is the molar mass of HCO3 .

4.4.1.2 Major anions and fluoride - - 2- - Concentration of remaining major anions, namely Cl , NO3 , and SO4 , and F were measured using Dionex DX-120 Ion Chromatography at the Department of Land and Water Resources at KTH. Ion chromatography is a laboratory technique based on the ability of an ion exchanger to separate major ions present in water. Once the ions are distributed over different time intervals, a conductivity meter measures the electrical conductivity, which is then converted into the concentration of the specific ion. The Dionex DX-120 is equipped with an IonPacAS14 column and a Dionex ASRS suppressor. The filtered and not-acidified samples were used for this analysis. A method for high concentrations of inorganic ions was used in consistency with the U.S. EPA Method 300.0.3. A three points calibration standard was implemented with standard solutions prepared from single anion stock solutions of 1000 mg/L. Standard multi-anions solutions were instead used as quality checks every 15 samples.

4.4.1.3 Phosphate The SEAL Analytical AutoAnalyser 3 (AA3) was used to measure the concentration 3- of PO4 in the filtered and acidified water samples. The instrument consists of an autosampler, a peristaltic pump, a chemistry module and a data acquisition software. Based on colorimetric technique, AA3 allows low detection limits of dissolved nutrients in environmental samples (SEAL Analytical, 2014)

4.4.2 Major cations and trace elements Major cations, namely Ca2+, Na+, K+, Mg2+, and trace elements, including As, were analysed using PerkinElmer Inductively coupled plasma Mass Spectrometry (ICP- MS) at the Department of Environmental Science in Linköping University. Inductively coupled plasma mass spectrometry consists of ionizing the water sample with a plasma source and then using a mass spectrometry to separate and detect the generated ions. The filtered and acidified samples were used for this analysis. Firstly, a method for major cations was run requiring sample dilution (dilution factor = 25) and the use of a high concentration detector. Secondly, a low concentration detector

was used for analysis of trace elements, during which water samples were not diluted. In both the methods, calibration was done before and after the analysis and quality check controls were made every twenty samples.

4.4.3 Accuracy of major ions analysis The accuracy of the major ions analysis was estimated by determining the electro neutrality of water, since the sum of positive charges should be equal to the sum of negative charges (Appelo & Postma, 2005):

( − | |) ℎ (, %) = × 100 ( + | |)

Where the sum of cations and anions are expressed in meq/L and with their charge sign. Deviations in electrical neutrality larger than 10% usually indicate poor sampling and/or analytical procedures.

Another check on the chemical analysis was made by comparing the field measured electrical conductivity with the sum of anions and cations, in accordance with the following relationship:

= (/) ≈ ⁄100 (⁄)

4.4.4 Dissolved Organic Carbon (DOC) The filtered and acidified samples were analysed at Department of Land and Water Resources at KTH using a Total Organic Carbon Analyzer TOC-L. A method for non-purgeable organic carbon measurements was implemented. A small amount of

HCl is added to the sample to convert all the inorganic carbon into CO2, which is then removed through sparging. The treated sample is heated to 680°C in an oxygen- rich environment and the generated CO2 is detected by an infrared gas analyzer.

Finally, CO2 levels are converted to DOC concentrations.

4.5 Data Management and Interpretation

4.5.1 Excel The physical and chemical characteristics of the collected water samples were firstly displayed in Excel environment, providing a first insight into the dataset. Ion water

balance and quality checks were calculated, and scatter plots were used to preliminary investigate the correlation between different geochemical factors.

4.5.2 R-software R-software was implemented for basic statistics analysis. Histograms, box plots and data plots were created to understand the distribution of the investigated parameters and the presence of outliers. Particularly, box plot is a useful plotting tool to study the distributional characteristic of a group of values, since it divides the dataset in quartiles, as shown in Figure 12.

Figure 12. Box plot and key descriptive parameters.

Values falling 1.5 times outside the inter quartile range above the upper quartile and below the lower quartile are considered outliers. Correlation factors among the physical and chemical characteristics of the water samples were also calculated in R environment.

4.5.3 ArcGIS ArcGIS 10.3.1 was used to spatially visualize data and to create maps showing water sources location, geology, topography and hydrography of the area under study. In Table 3 the data used in ArcGIS are listed and their sources are reported.

Table 3.Source of data layers used in ArcGIS.

Data Source Reference

Geological Survey of (Geological Survey of Geology of Tanzania Tanzania Tanzania, 2015)

TZA Administration Tanzania National Bureau (National Bureau of units of Statistics Statistics, 2015)

(Ministry of Water, Water points map Ministry of Water - TZA 2013)

LVB boundary RCMRD GeoPortal (GeoNode, 2015)

SRTM Elevation - 1Arc (U.S. Department of the US Geological Survey - Interior, 2015)

4.5.4 AquaChem The software AquaChem (version 4.0) was used to identify the type of water in each sample and to construct piper plots and stiff diagrams.

4.5.5 PHREEQC and geochemical modelling. PHREEQC is a computer model performing a wide variety of geochemical calculations: (i) speciation and saturation indexes, (ii) batch-reaction (iii) one dimensional transport calculations and (iv) inverse modelling. (Pankhurst & Appelo, 1999). In this study, PHREEQC (Version 3 for Windows) was used to calculate species’ distribution and saturation indexes of relevant solid phases. Saturation index SI is defined as:

= log/

Where Ksp is the solubility product for a given temperature and IAP is the ion activity product. Solubility products of solid phases were taken from “Phreeqc_Master” database, developed by Mahoney (2016) and based on

“WATEQF” database. SI=0 indicates that water is at equilibrium with respect to the mineral. When SI > 0, water is supersaturated and the mineral should precipitate. On the other hand, if SI < 0, water is under saturated and the mineral should dissolve. Coupled with PhreePlot, PHREEQC was also used to create the Eh-pH diagram of As species in natural water.

4.6 Limitations

The following limitations were encountered during the progress of the present study:

 Limited information about groundwater distribution and flow in Lake Victoria Basin.  Little information about depth and design of the sampled wells.  Small amount of samples (n=54). This has partly limited the intention of investigating a large portion of the Lake Victoria Basin.

All these limiting factors have affected the outcomes of the present study, without however impeding the accomplishment of the designated objectives.

5. RESULTS AND DISCUSSION

The Result and Discussion chapter consists of three sections: i) “Results of field measurements and laboratory analyses; ii) “Relations between different geochemical parameters”, where relationships between different geochemical parameters are investigated; iii) “ Geochemical modeling”, which illustrates the outcomes of the geochemical modeling. In line with the objectives of the present study, water samples were collected at different locations within the Lake Victoria Basin and from different types of water source. This has led to the need of considering sub-datasets, discerning the samples among four locations, namely Geita Town (GT), Geita Rural (GR), Butiama/Musoma(BM) and North Mara (NM), and among four types of water source, i.e. Boreholes (BH), Shallow wells (SW), Surface water (Surface) and Spring. Figure 13 shows number of samples within each sub-dataset.

North Mara 15

Musoma/Buti… 14

Geita Rural 9

Geita Town 16

0 5 10 15 20

Spring 8

Surface 15

Shallow wells 12

Boreholes 19

0 5 10 15 20

Figure 13. Number of samples in each sub-dataset.

5.1 Results of Field Measurements and Laboratory Analyses

This section provides the results of lab and field measurements in the following order: (i) pH, Redox potential (Eh),electrical conductivity (E.C.); (ii) major ions; (iii) other major elements; (iv) DOC; (v) Al, Fe, Mn; (vi) As and (vii) other trace elements.

Field measurements taken in Tanzania and the results of the laboratory analyses were firstly merged into one single dataset. Table 4 gives a first insight into the database, providing key statistical characteristics for each investigated parameter.

Table 4. Summary of geochemical parameters for the entire dataset.

% WHO Parameter Unit Typicalrange Extreme values Compliant guideline samples

Minimum Median Maximum

Temp °C 23.4 26.55 29,3 33 pH 6.81 5.33 8.37 µS/c E.C. m 49 372 1590 4040 Eh mV 261 433 508 201 to 225 As field ppb 0 20 300 10 46 DOC mg/L 1.1 1.9 4.3 6.2 to 8.7 - HCO3 mg/L 16.8 155.7 659.0 Cl- mg/L 0.8 12.8 104.9 250* 100 - NO3 mg/L <0.1 8.1 74.4 104.7 to140.0 50 83 2- SO4 mg/L <0.1 5.7 230.65 619,2 250* 96 3- PO4 µg/L 3.0 9.6 96.5 177.0to 318.7 Na+ mg/L 3.5 28.1 203.9 494,8 200* 98 Mg2+ mg/L 0.6 7.4 44.8 62.3 K+ mg/L 0.2 2.3 13.2 79.4 Ca2+ mg/L 1.8 26.3 118.4 429.6 F- mg/L 0.09 0.25 2.44 1.5 98 Si mg/L 3,8 20.2 47.2 Li µg/L 0.3 2.0 49.4 B µg/L 1.9 10.7 58.4 193.6 2400 100 Co µg/L <0.1 0.2 8.9 33.8 to 34.1 Sr µg/L 22.1 202.4 808.8 2064.8to 2550.1 Zr µg/L <0.1 <0.1 1.2 2.6 to 4.4 Mo µg/L <0.1 0.4 5.0 7.8 to 8.2 70 100 Cd µg/L <0.1 <0.1 0.2 3 100 Ba µg/L 9.9 65.5 321.2 478 700 100 Pb µg/L <0.1 0.2 4.9 11.9 10 98 U µg/L <0.1 0.3 7.7 10.8 to 14.2 30 As tot** µg/L <0.1 0.5 18.2 37.9 to 56.3 10 87 As (III) µg/L <0.1 0.4 3.7 Al µg/L 10.0 23.1 668.6 668.6to 1193.7 200* 91 Ti µg/L 0,1 0,9 11.1 25.4 V µg/L <0.1 2.7 24.3 Cr µg/L <0.1 0.1 0.6 1.3 and 1.7 50 100 Mn µg/L 1.1 22.1 237.6 327.0 400* 100 Ni µg/L 0.3 1.0 7.2 9.04 70 100 Cu µg/L 0.8 1.8 11.5 2000 100 Zn µg/L 0.5 11.9 74.4 90.1 Fe µg/L 2.6 42.3 722.9 2075 *WHO limit that may give rise to complaints from consumers (WHO, 2011) ** laboratory measurements of As

5.1.1 pH, Redox potential and Electrical Conductivity pH ranges from moderate acidic (pH=5.3) to alkaline (pH=8.4) and it visibly varies among the types of water, Figure 14. Surface water (Surf) shows the highest pH, ranging from neutral (pH=7.1) to alkaline (pH=8.4), whereas spring water is from moderate acidic (pH=5.3) to neutral (pH=7.0). Water from boreholes (BH) and shallow wells (SW) present a wide pH variation in both acidic and alkaline conditions.

Figure 14. Box plots showing values of pH, Redox potential (Eh) and electrical conductivity (EC) characterizing different types of water source.

Eh measurements vary between 200 mV up to 508 mV, with 75% of the values falling in the range 398 –508 mV. Seven water sources, i.e. six boreholes and one surface water, show the lowest Eh, ranging between 210 and 250 mV. Reliable Eh-

measurements are often difficult to obtain, therefore they can only be used as preliminary indicator of the Redox status of the groundwater systems. Nevertheless, these values imply oxidising conditions. The typical range of EC lies between 50 µS/cm and 1200 µS/cm. However, in two occasions, both of them from surface water, values of 1590 µS/cm and 4040 µS/cm were measured. In average, spring water shows lower EC values, suggesting lower concentrations of total dissolved solids.

5.1.2 Major ions The concentrations of Total Dissolved Solids (TDS) are all under the WHO limit of 1000 mg/L, except from two samples where they reach 1168 and 1354 mg/L. Both samples were taken from streams flowing in close proximity to mining sites. Major ions composition and type of water of the sampled sources are illustrated in the piper diagrams, in Figure 15. The first one (right-side) illustrates the distribution of major ions discerning among the locations of the water sources (Geita Town, Geita Rural, Musoma/Butiama and North Mara), whereas the second one (left-side) proposes a classification of the samples according to the type of source (borehole, shallow well, spring, surface water).

SW BH Spring Surface

Figure 15. Piper diagrams of the sampled water characterizing different locations (left) and types of water source (right). Red ellipse indicates a cluster of surface water samples with 2- high SO4 levels.

Regarding to cations, location of the water source is the factor influencing most the type of water. Almost all the samples from Geita Town present Na+ as the

dominating cation, followed by Ca2+ and Mg+. In contrast, Geita Rural and Butiama/Musoma has Ca2+ - Mg+ as the predominant couple with Na+ being the third most present cation. Finally, samples from North Mara Golda Mine are to an equal extent of Ca2+ - Na+ and Na+ - Ca2+ type, with Mg as the third most present cation. - According to anions, 85% of the samples are of HCO3 type, regardless of the type of water source and location. However, a cluster of four surface water samples (ID 14, 2- 34, 40, 51) presentSO4 as the dominant anion. Most of the groundwater samples are clustered near the left corner of the diamond, 2+ + - indicating temporary hardness, i.e. water rich in Ca , Mg and HCO3 . This suggests low retention times of water in the aquifer systems. However, according to anions, - - 35% of the boreholes are characterized by HCO3 - Cl water type, indicating inception of groundwater ageing. Three shallow wells fall in the top corner of the diamond, which is the region of - - water of permanent hardness. These samples have NO3 and chloride Cl as the predominant anions, alleging to the possibility of contamination from human activities. Most of the shallow wells are in fact hand dug and unprotected.

5.1.2.1 Anions Although 85% of the samples are of bicarbonate type, differences can be seen in the anion distribution depending on the type of water source, Figure 16.

Figure 16. Concentrations of major anions characterizing different types of water source.

- HCO3 is the most predominant anion, ranging from 17 mg/L up to 660 mg/L. Water - from boreholes has the highest concentrations of HCO3 and shows the largest variability, followed by shallow wells and surface water. Rock weathering and - CaCO3 dissolution is expected to be the source of high HCO3 levels in deep groundwater. Spring water has the lowest concentrations. 2- SO4 is the second most prevalent anion in surface water, with a typical range between 1 mg/L and 220 mg/L. In one occasion (ID 14) a value of 619 mg/L was measured, representing the only sample exceeding the WHO recommended limit of 2- 250 mg/L. SO4 is found to a lesser extent in the other types of water source, where 2- the typical range lies between 1 and 45 mg/L. High SO4 levels in streams, rivers

and dams give a first indication of surface water contamination from mining activities. - - - NO3 and Cl constitute the most dominant anions after HCO3 in shallow wells, - springs and boreholes. The acute WHO limit of 50 mg/L for NO3 is not complied by 8 water samples, of which 2 are from springs, 3 from boreholes and 3 from shallow wells. These high concentrations can be explained by contamination of the water sources due to human activities, especially use of fertilizers in agriculture and poor sanitation practises. Cl- concentrations illustrate full compliance with the WHO limit of 250 mg/L. 3- PO4 is the least abundant anion in all the water sources, ranging between 0.003 and 0.4 mg/L.

5.1.2.2 Cations Ca2+ and Na+ are the most abundant cations, ranging 2 - 120 mg/L and 4 - 115 mg/L respectively. One sample (ID 14) shows Na+ concentration equal to 203 mg/L, which 2- is slightly higher than the WHO limit of 200 mg/L. In the same sample, SO4 and Ca2+concentrations reached 600 mg/L and 103 mg/L respectively. Magnesium is the third predominant cation, with concentrations lying between 1 and 45 mg/L. Finally, K+ is the least prevalent cation in water, presenting a range going from 0.2 to 13 mg/L. The distribution of cations significantly varies across the four locations, Figure 17.

Figure 17. Concentrations of major cations in the four locations under investigation.

Ca2+ and Mg+ show similar variability, with notably lower concentrations in Geita Town than in the other three locations. In contrast Na+ and K+ have the lowest concentrations in Geita Rural, with higher values found in Geita Town, Musoma/Butiama and North Mara. Higher concentrations of K+ and Mg+ in Musoma/Butiama and North Mara may be due to high-Mg andesite and K-granites occurring in the Musoma-Mara Greenstone belt (Manya & Maboko, 2015), whereas K+ concentrations in Geita Town are explained by the intrusions of feldspar in the greenstone belt unit (Sanislav et al., 2016).

Ultimately, it is therefore apparent that the major ion chemistry is controlled by the geological characteristics of the terrain varying across the sampling locations.

5.1.2.3 Accuracy of major anions analysis Electrical balance was examined and 90% of the sample show a deviation smaller than 10 %. This implies that the results of the water sampling are reliable. Improper sampling and/or inaccurate analytical procedure might have affected the seven samples showing a deviation larger than 10%, and they were therefore excluded from the subsequent analysis. When comparing to the electrical conductivity, a better agreement was found with anions than with cations, suggesting that concentrations of cations were generally too low.

5.1.3 Other major elements After major ions, the most abundant element in the sampled water is silica (Si) with concentrations ranging between 3.8 and 47.2 mg/L. The highest concentrations are found in boreholes water where most of the silica content is due to weathering of rocks containing silicate minerals. Among the four locations Geita Rural has the highest Si levels, suggesting a local geology enriched in silicate rocks. In contrast, Musoma/Butiama and Mara show the lowest Si concentrations.

5.1.4 Dissolved Organic Carbon (DOC) DOC gives an indication of the content of organic matter in natural water. In this study the typical range of DOC concentrations lie between 1.14 and 4.35 mg/L, however an excess range is observed between 6 and 9 mg/L. As shown in Figure 18, DOC levels in surface water are notably higher than in groundwater, probably because of the runoff from agriculture and grassland soils.

Surface

Figure 18. DOC concentrations characterizing the four types of water source.

With regard to the borehole sub-dataset, four samples present DOC content higher than the typical range, with concentrations varying between 2.6 and 3.2 mg/L. The same samples show the lowest values of the field measured Redox potential and concentrations of Fe and Zn significantly higher than the average. Based on the field measurements of the Eh, it can be inferred that in these boreholes microbial oxidation of organic matter is responsible for the depletion of dissolved oxygen (lower Eh values) and the consequent reductive dissolution of Fe hydroxides. This can promote the mobilization of metals, such as Zn and As. (Anawar et al., 2003)

5.1.5 Al, Fe, Mn The results for Al, Fe and Mn are grouped together because of the tendency of these metals to form oxides/hydroxides in the pH range of natural water, largely influencing the mobilization of heavy metals, such As, Cu, Cd and Pb through sorption/desorption processes. Figure 19 shows the distribution of Fe, Al and Mn levels in the sampled water.

Figure 19. Fe, Al and Mn levels in the sampled water considering both Mara and Geita Region .

Iron levels typically vary between 2.6 and 328 µg/L . However, in four occasions, specifically in two shallow wells and in two boreholes, values of 548, 644, 722 and 2075 µg/L were measured. Al shows an excess range between 100 and 1194 µg/L , in addition to a typical range going from 10 to 77 µg/L . Four samples do not comply with the WHO limit of 200 µg/L. Typical Mn concentrations vary from 1 to 88 µg/L , however 6 samples fall in the range 113-327 mg/L.

In seven occasions field filtration had not been performed. However, nitric acid was added for ensuring sample preservation and samples were filtered once back to the lab prior to any analysis. Acidifying the sample before the filtration has lead to a drastic change in the chemistry of the water sample. The largest variations are observed in the concentrations of Al, Fe and Mn (Figure 20).

Figure 20. Concentrations of Fe (top-left), Al (top-right), Mn (bottom) in unfiltered (red) and filtered samples (blue). The dot lines represent the median concentration of each dataset.

The median concentration of Fe, Al and Mn in the unfiltered samples (red dot line in Figure 20) are from 10 to 100 times higher than in the filtered samples (blue dot line in Figure 20). These results indicate that these metals show greater concentrations in the particulate fraction than in the dissolved one. If field filtration is not performed, all the suspended solids will remain in the sample and the subsequent acidification will cause the dissolution of hydroxides and the subsequent release of metals from the solid phases. The discrepancies are larger in the concentrations of Fe, Al and Mn because these metals tend to occur in natural water as suspended oxides/hydroxides, which are readily filtered out when field filtration is performed.

These findings are in agreement with the study conducted by Bowell et al. (1995), who investigated heavy metal pollution of Orangi Rive, in North Tanzania, caused by effluents from a medium scale gold mine. By comparing metal concentrations between filtered and unfiltered samples, the study reveals a strong partition into the particulate fraction of Fe, Al and Pb.

5.1.6 Arsenic The first two sections of the “Arsenic” section deals with the results of the field and laboratory measurements of As. Secondly, the speciation of As in presented and, finally, the spatial distribution of As levels is illustrated and described. Maps showing As concentrations are shown in Appendix C.

5.1.6.1 Field-measured As Field measurements of As were conducted using the Hach Low Range Arsenic Field Test Kit. The accuracy of these measurements are strongly influenced by fieldwork conditions (exposure to sunlight and dust) and by accurateness in comparing the developed colour on the test strip with the reference colour chart (see Figure 9). Arsenic concentrations vary from under detection limit (10 µg/L ) up to 300 µg/L . The distribution of As levels is presented in Table 5, proposing the same subdivision as in the As field test kit.

Table 5. Distribution of As concentrations in the study area.

µg/L 0 10 30 50 70 300 500 Samples within 47 17 9 15 12 0

the range % % % % % %

53% of the drinking water sources present As levels exceeding the WHO recommended limit of 10 µg/L. When considering the Tanzanian limit of 50 µg/L , only 27% of the sampled sources result unsafe. Arsenic concentrations greatly differ among the four types of water source, see Figure 21. Except from three occasions (ID 43-46-53), all borehole waters show As content below or equal to 10 µg/L , whereas surface water is the most contaminated source with a median concentration equal to 70 µg/L . The observed decreasing concentrations with depth at which water is withdrawn (surface water – shallow well

– borehole) demonstrates that As contamination in mining areas is a surface phenomenon.

Surface

Figure 21. As concentrations characterizing the four types of water source.

When considering the borehole sub-dataset, two of the three samples (ID 40, 46) exceeding the As WHO limit have the highest DOC levels and they are characterized by relatively low Redox potential, i.e. 260 and 220 mV. The combination of high DOC and low Redox probably indicates microbial oxidation of organic matter, reductive dissolution of Fe/Al hydroxides and the subsequent release of dissolved As (Anawar et al., 2003). On the other hand, the highest As concentrations in surface waters are associated with high SO4 concentrations, suggesting that oxidation/dissolution of As bearing minerals is a major source of As.

5.1.6.2 Spatial variation of As The spatial variation of As across the four locations under investigation is heterogeneous. None distribution patterns can be identified at large scale and site- specific conditions have to be considered in order to understand the occurrence of As. Generally, Mara region is affected by significantly higher As concentrations than

Geita. Only 30% of the drinking water sources in Geita region exceed the recommended limit of 10 µg/L , whereas in Mara region the non-compliant sources are 73%. A substantial difference in the As levels between the two regions has been previously pointed out by Kassenga and Mato (2008) and it is in agreement with the prediction map developed by Ijumulana et al., (2016), Figure 2. As described in the “Geology” section, auriferous mineralization in Mara and Geita regions is bound to greenstones rocks which presents a complex lithology composed of meta-basic/meta- rhyolitic felsic volcanic rocks (lavas, tuffs and tuffites),ferruginous rocks (BIF) and meta-sediments. Evident differences in the concentrations of major ions (section 5.1.2) and of As between Mara and Geita regions suggest that local geology might play a key role in the occurrence and distribution of As. However, an relationship between As levels and a specific type of rock has not been identified, requiring further investigations. An attempt to explain the link between local geological setting and As content in the sampled water is given in section 5.2 Maps showing As concentrations at each of the four locations are presented in Appendix C, and a description is provided below, distinguishing between large scale and small scale mines.

Large scale mines Mining slags or mine wastes generated due to extensive mining can cause extensive pollution of the surrounding surface and ground waters. The drinking water sources in the vicinity of the two large scale mines targeted in the present study show contrasting As contamination levels. The population living in the surrounding of North Mara Gold Mine (NMGM) are exposed to As concentrations varying between 10 µg/L and 300 µg/L (median value = 70 µg/L ). In contrast, As levels in Geita Town and in the villages around Geita Gold Mine (GGM) exceed the recommended limit of 10 µg/L only in three cases out of the sixteen sampled water sources, with the highest concentrations being 40 µg/L .

North Mara Gold Mine Arsenic contamination around NMGM is extensive and affects both surface and groundwater resources. The highest concentrations, i.e. 300 and 150µg/L , were reported in streams flowing through the mining site, alleging to pollution by mine effluents. High As levels in Tighite River, 150 and 300 µg/L (ID 40 and 47

respectively), suggest drainage of contaminated run-off from the open pit and the waste rocks piles. Although the present study does not provide evidences of acid mine drainage, such as low pH and high heavy metals’ concentrations, it is likely that surface run-off enriched in trace metals drains from the waste rock piles into the river. When the acidic overland flow is mixed with river water, the high alkalinity and pH of the latter cause precipitation of metals and partition into sediments. The buffering action of pH and alkalinity is therefore the reason for low concentration of trace elements in the samples collected from Tighite River. Differently from most of the other heavy metals, As is mobile, as oxyanion, also at high pH, initially explaining why high As concentrations were recorded where the content of other trace elements is low. The claim for acid mine drainage polluting streams and river is supported by Almås and Manoko (2012), who found severe trace elements contamination of surface water and sediments downstream to a leachate tank one month after an accidental acid had occurred in 2009. In addition, high As level in samples ID 51, 52 and 54 indicates seepage from the tailing storage facility, where the process water enriched in cyanide and heavy metals is stored and decanted. Given that similar results were obtained by Almås in 2009, leakage of the tailing dam seems to be a continuous source of As contamination of the surrounding streams and rivers. Besides the evident signs of drinking water contamination caused by mine effluents, high As levels were also found in area not directly affected by mining operations. High As levels in shallow wells and boreholes in Nyamongo village suggest that bedrock and overlying sediments in North Mara are naturally enriched in As-bearing minerals. The oxidising conditions of the groundwater systems enhance weathering of these mineral and the subsequent release of As in the aquatic environment. In conclusion, effluents from the mining site and an As-rich local geology are all concurrent causes for extensive As pollution in the drinking water sources around North Mara Gold Mine, representing a serious risk for human health. Nevertheless, signs of health impacts in the villagers living around the mine have been already reported by (Evjen, 2011), who highlighted higher As concentrations in human tissues in people from Nyamongo village than in the reference group from Dar es Salaam.

Geita Gold Mine and Town With regard to As and other trace elements, the environmental impact in the surrounding of Geita Gold Mine is negligible. None indications of contamination from waste rock piles and tailing storage facility were found and only three of the sixteen sampled drinking water sources show As content higher than the recommended limit. Although a proper management of waste rocks and tailing dam is one of the main reasons for such a low impact, density and location of settlements play an important role in exposing the population to likely contaminations. Geita Town and most of the surrounding villages are located upstream to the mining site. Thus, mine effluents and surface run-off from the claiming area do not impact the drinking water sources used by the local communities. Only one village is located in close proximity to the tailing storage facility and the surface water sampled at this location show As level equal to 30 µg/L (sample ID 14). However, since artisanal gold mining is largely conducted at this village, the As contamination is probably due to this practice, rather than indicating seepage from the tailing dam. In Geita Town, shallow wells and boreholes present low As concentrations, with the highest value, 40 µg/L , being from a shallow well. This result advocates that the local geology in Geita is less enriched in As-bearing minerals than the one in North Mara. Differences in gold mineralization and in the hosting geological setting may give a feasible explanation for lower As values in Geita region.

Small scale mining Arsenic concentrations greatly vary across the artisanal mining areas under investigation and the extension of the contamination is generally isolated to the close proximity of the mining and processing sites, as indicated in previous studies (Kahatano & Mnali; 1995; Nyanzaet al., 2014; Van Straaten, 2000).The highest concentrations, 100 and 70 µg/L , (ID 30, 29) were measured in shallow mining pits (< 20 m) from which drinking water is withdrawn after they have been dismantled for the mining purpose. Although the pits are exhausted with regard to gold minerals, the coarse sediments and the hard rock in which they are dug remain in contact with groundwater and the atmosphere, promoting oxidation/dissolution of As-containing minerals. Signs of surface water contamination were also found in vicinity of artisanal gold mines, suggesting contaminated run-off from waste rocks dumps and

occasional flooding of amalgamation ponds (ID 21, 27, 33). Water sampled from deep boreholes show the lowest concentrations and in none, occasions exceed the recommended limit of 10 µg/L . This indicates that deep groundwater may represent a source of safe drinking water in rural communities practising artisanal mining.

5.1.6.3 Laboratory-measured As Arsenic concentrations were analysed in laboratory with ICP-MS. There is a large discrepancy between the field measurements and the results of the water analysis see Table 6. According to the lab results, only 15% of the samples show As levels higher than the recommended limit of 10 µg/L , instead of the 81% indicated by the field measurements.

Table 6. Difference between field measurements and laboratory results for As

As field (µg/L ) As lab (µg/L )

Min <10 <0.1

Median 20 0.5

Max 300 56.3

Firstly, the accuracy of the two measuring methods was questioned. The results from the ICP-MS are considered reliable because of good recovery and reproducibility of quality checks and of additional standards, which were run together with the samples. Regarding field measurements, scientific studies have been made with the aim of assessing the accuracy of field test kit for As determination. (Rahman et al., 2002) has shown that reliability of As field kits tends be poor at low range (< 50µg/L ), because of ambiguous identification of the colour developed on the test strip. Although both measurements are reliable, using different methods in situ and in laboratory may have contributed to the observed differences in the lab and field measured As. However, this reason alone cannot explain the discrepancies observed in the present study, where field values of 200 – 300 µg/L correspond to values of 5–10 µg/L in the lab results. Secondly, the sampling and analytical procedure were reviewed. The As field test kit was used in situ on an unfiltered sample of water, whereas the lab analyses were

performed on the filtered and acidified sample. Consequently, the 0.45µm field filtration seems to be responsible for the removal of most As. Thus, large discrepancies between field measurements and water analysis indicate that most of the As is adsorbed on the particulate fraction, which are readily filtered out during filtration, rather than occurring as dissolved species. This hypothesis is supported by a good correlation between field and lab measurements of As, as shown in Figure 22. Peaks in lab-measured As correspond to peaks in the field-measured As, suggesting that only a small fraction of the total As occur as dissolved species.

Field As Lab As Difference 350 120

300 100

250 80

200 60

As (µg/l)As 150

40 % Difference 100

20 50

0 0 3GT 8GT 9GT 12GT 19GR 29BM 30BM 32BM 34BM 39BM 40NM 42NM 43NM 44NM 49NM 50NM 51NM 52NM 54NM ID Sample Figure 22. Graph showing discrepancies between lab- and field-measured As. The green line represents percent difference.

Similar results were obtained by Bowell et al. (1995), who relates strong partition of Al and Fe in particulate fraction with large discrepancy between As concentrations in filtered and unfiltered samples. In addition to Al/Fe and Mn-oxi/hydroxides, As can be strongly adsorbed on clay minerals (Manning & Goldberg, 1997) and it is subjected to surface complexation with fulvic substances (Buschmann, et al. 2006), which are both filtered out when field filtration is performed.

5.1.6.4 As (III) As (III) concentrations are very low, ranging between < 0.1 and 3.7 µg/L . When compared to the total dissolved As levels, i.e. the lab measurements of As, it is evident that As(V) species largely predominate ( > 80%) over the reduced species As(III). This result is explained by the oxidising conditions of the sampled surface and groundwater.

5.1.7 Other trace elements After major ions and Si, the most predominant element is Strontium (Sr) ranging from 22.1 µg/L up to 808.8 µg/L. In two occasions (ID 14, 54) values of 2064 and 2550 µg/L were recorded. Trace amounts of Sr are widely distributed in limestone and in igneous rocks. Barium (Ba) results quite abundant with concentrations ranging between 9.9 and 478 µg/L . The highest concentrations, >150 mg/L, were all measured in borehole water from Geita Town. These concentrations are probably due to the granite-like igneous rocks which dominate the geology in Geita Town. All the samples comply with the recommended limit of 700 µg/L . U and Mo concentrations in North Mara Gold Mine are one order of magnitude higher (6-12 µg/L ) than in the other three locations (< 1µg/L ), indicating a peculiar geology.

5.2 Relations Between Different Geochemical Parameters

Relationships between different geochemical parameters are investigated in this subchapter. In the first section (5.2.1) the analysis of relationships between major ions and silica is presented and considerations about the local geological setting in Mara and Geita regions are made. The second section (5.2.2) regards relationship of DOC with other geochemical parameters and finally, the third section (5.2.3) deals with correlations between As and other elements. Considering that drinking water samples were collected from four different types of water source and across four locations, analysis of correlations between different hydrogeochemical parameters were performed at sub-dataset level (Figure 13). Nevertheless, the amount of samples is not homogenous across the sub-datasets, partially affecting the relevance of the correlations.

5.2.1 Local geology and solubility controls on the distribution of major ions It has been previously highlighted that concentrations of major ions greatly differ across the four areas under investigation. Significant discrepancies in the ion composition of water may indicate different geological settings and groundwater processes. By comparing concentration ratios between major ions and silicate an insight into the rock source of the water was obtained and the results are listed below:

2- 2+  Four samples (ID 14, 34, 40, 51) show SO4 /Ca >1, suggesting pyrite oxidation or calcite precipitation. These samples correspond to the cluster of surface water identified in the piper diagram in Figure 15. As they were collected in close proximity to mining sites, pyrite oxidation is expected to be the main process affecting these waters.  Except from few cases, all the groundwater samples from Geita Town and - Geita Rural are characterized by HCO3 - SiO2 ratio <5, suggesting silicate

weathering as the dominant process, see Figure 23. In contrast, HCO3 /SiO2 >10 in Musoma/Butiama and North Mara indicates weathering of carbonate minerals. However, silicate weathering is also occurring in North Mara.

30 40 30

/ SiO 3 20 HCO

Carbonate

Silicate

0 BM GR GT NM

Location - Figure 23. Variation of HCO3 /SiO2 ratio across the four locations under investigation. Only groundwater samples are considered. BM: Butiama/Musoma; GR: Geita Rural; GT: Geita Town; NM: North Mara.

Bivariate plots are used to better understand the source of major solutes in the sampled groundwater (Raychowdhury et al., 2013). In Figure 24 the Na-normalized - Ca versus HCO3 plot shows that groundwater samples from Geita region (Geita Town and Geita Rural) are influenced by silicate weathering, whereas samples from Mara region present a trend towards the typical range of carbonate dissolution. Similar results are derived also from the Na-normalized Mg versus Ca plot.

carbonate dissolution 10 Geita Mara /Na 3 1 HCO

silicate weathering

0.1 0.1 1 10 Ca/Na

10 carbonate dissolution

Geita Mara

1 Mg/Na

silicate weathering

0.1 0.1 1 10 Ca/Na

Figure 24. Bivariate plots indicating typical ranges of carbonate and silicate weathering. A closer examination of the Piper plot for groundwater samples (borehole and shallow well) in Geita Town, Butiama/Musoma and North Mara Gold Mine is given in Figure 25. When a series of water analysis shows a linear trend passing through the corner of one or both the ternary diagrams, then precipitation/solution is likely to be a dominant process in the aquifer matrix (Hounslow, 1995).

Figure 25. Ternary diagrams for groundwater samples in Geita Town, Butiama/Musoma and North Mara.

+ - In this study, linear trends towards Na and HCO3 in Geita Town and North Mara reveal weathering of albite, a sodic plagioclase widely distributed in felsic rocks. On the other hand, in Butiama/Musoma (Mara region) the linear trend towards Ca2+ may indicate calcite precipitation/solution; however this conclusion is not supported by a - trend towards HCO3 in the anion triangle. It is important to highlight that the above deductions give a general overview of the possible groundwater matrix reactions occurring at regional level. If a more detailed investigation has to be conducted, the groundwater flow paths should be locally determined. However, this information is not available for the present study. Moreover, in addition to rock weathering, other groundwater processes, such as ion exchange and mixing of waters with different composition, occur in the aquifer matrix. + Nevertheless, a weak correlation between Na and SiO2 (Figure 26) supports that albite weathering is one of the dominant process in Geita region.

30 R² = 0.40

Na (mg/l) Na 20

0 0 10 20 30 40 50 SiO2 (mg/l)

Figure 26. Correlations between silica and sodium in samples from Geita Region. The albite weathering reaction is as follow:

+ + 2NaAlSi3O8 + 2H +9H2O → Al2Si2O5(OH)4 kaolinite +2Na + 4H4SiO4

Where the monosiallization of the plagioclase release sodium and silica, which is however partly retained in the solid phase, forming kaolinite. (Raychowdhury et al., 2013). A similar relationship is not observed in samples from Mara region. 2+ - Calcite precipitation is deduced when plotting Ca versus HCO3 and pH, (Figure 27) in accordance with the following reaction:

2+ Ca + 2HCO3 → CO2(g) + H2O + CaCO3

160 160 carbonate carbonate

120 120 R² = 0.40

(mg/l) 80 (mg/l) 80 2+ 2+

Ca R² = 0.30 Ca

40 40

0 0 300 400 500 600 700 6.5 7.0 7.5 8.0 - HCO3 (mg/l) pH 2+ - Figure 27. Relationship of Ca with HCO3 and with pH. Further considerations about the main rock types occurring at the four locations can me made when looking at the stiff diagrams of the sampled groundwater. In fact, the shape of a stiff diagram can be linked to a specific type of rock (Hounslow, 1995). Figure 28 shows four stiffs diagrams which are representative of the areas under investigation. Samples from Geita Town and Rural (ID 7 and 22) indicate rhyolite- granite type of rock, whereas in Butiama/Musoma rocks containing carbonate minerals seem to predominate (samples ID 39 and 49).

2 0 2 2 0 2

meq/kg meq/kg

- - Na+ Cl Na+ Cl

++ - ++ - Ca HCO3 Ca HCO3

++ -- ++ -- Mg SO4 Mg SO4

7 22 8 4 0 4 8 8 4 0 4 8 meq/kg meq/kg

- + - Na+ Cl Na Cl

- ++ - ++ HCO Ca HCO3 Ca 3

-- ++ -- ++ SO Mg SO4 Mg 4

39 49

Figure 28. Stiff diagrams of representative groundwater samples characterizing the four locations under investigation. Samples ID 7 and 22 are from Geita region, samples ID 39 and 49 are from Mara region. To sum up, differences in the local geology and in the gold mineralization at the four locations under investigation can be deduced when looking at the ion composition of the sampled water. The spatial variability in the geology concurs with the intrinsic characteristics of the Tanzanian Greenstone belts, which are characterised by variety of lithological assemblages: felsic volcanic rocks, ferruginous rocks and sedimentary units (Kabete et al., 2011). Geita Greenstone Belt, on which Geita Rural and Geita Town are located, is characterized by a more homogeneous geology dominated by intrusive and extrusive silicate rocks, i.e. diorite and basalts. This is concordantly indicated by bivariate plots, ternary and stiff diagrams and the relationship between + Na and SiO2. In this setting gold occur as small-to large- quartz veins filling simple fractures in diorite and ironstones.

In contrast, samples from NMGB suggest a more complex geological setting, where silicate weathering does not seem to be the dominant weathering process. Bivariate

2+ - plots, stiff diagrams and relationships between Ca , pH and HCO3 suggest the presence of rocks containing carbonate minerals. An explanation can be found in the type of gold mineralization within the North Mara Greenstone Belt, where gold occur as impregnations in various host rocks (BIF, tuff) in carbonatized shear zones. - Carbonitazion of the shear zones can be the reason of higher HCO3 -SiO2 ration and the trend towards carbonate dissolution in the bivariate plots, when compared to samples from Geita region.

5.2.3 DOC - Degradation of organic matter may be a major source of HCO3 in natural water (Hasan et al., 2007). Evidences for this process are found in the surface water of the - present study, which shows weak correlation between DOC and HCO3 , Figure 29. Similar correlation is not observed in shallow well and borehole water, since the - major source of HCO3 in groundwater is dissolution of mineral such as calcite and Ca-feldspar

700 surface BH SW 600

500

400 (mg/l) - 3 300

HCO R² = 0.30 200

100

0 0 2 4 6 8 10 DOC (mg/l)

- Figure 29. Correlation between HCO3 and DOC in surface water (yellow), shallow well (light blue) and borehole (dark-blue).

DOC is also an important factor influencing the mobility of metals in aquatic systems. In our study, signs of metal mobilization enhanced by DOC were found in borehole waters. As shown in Figure 30 high DOC levels in boreholes are associated with the lowest field measured Redox potential, advocating to depletion of oxygen

caused by microbial degradation of organic matter. Consequently, the less strong oxidising conditions trigger partial dissolution of Fe oxides/hydroxides, rising dissolved Fe concentrations in water. Moreover, the clear linear trend between Zn and DOC indicates that Zn is being mobilized by humic and fulvic acids through the formation of surface complexes. A clear relation among Redox, DOC, Fe and As cannot be observed; however, the highest concentration of As in borehole water is associated with the highest DOC level, one of the lowest Redox value and high Fe concentration. Although data are inconclusive, the importance of the DOC-Fe-As system in groundwater should be taken into consideration.

600 800 ID 19

600 400 ID 48

400 ID 43 ID 46 Eh (mV) Eh ID 19 ID 46 (µg/l) Fe 200 ID 48 200 ID 43

0 0 1 2 3 4 1 2 3 4 DOC (mg/l) DOC (mg/l) 100 160 ID 46

80 ID 46 120 R² = 0.427 60 80 ID 19 ID 48

40 (µg/l)As Zn (µg/l)Zn ID 43 40 20 ID 43

0 0 1 2 3 4 1 1.5 2 2.5 3 3.5 DOC (mg/l) DOC (mg/l) Figure 30. Relationship of DOC with (top-left) Eh; (top-right) Fe; (bottom-left) Zn; (bottom- right) As in borehole waters.

5.2.4 As Correlations between As and other geochemical parameters are analysed considering the field measurement of As, as they represent concentration of total arsenic (adsorbed and dissolved species). Moreover, lab-measured concentrations of As lower than 1 µg/L (50% of the samples) are considered not fully reliable and they are therefore not suitable for correlations’ analysis.

5.2.4.1 pH-Eh In accordance with its peculiar geochemistry, As was found at varying pH, from moderate acidic spring water (pH 5.5 – 6.5) to alkaline river water (pH 7.5 – 8.5), see Figure 31 .The mobility of As at high pH as oxyanion can explain the slightly positive trend of As with pH, which is however significant only in shallow wells. A cluster of four surface water samples show the highest As concentrations, associated with pH in the alkaline range 7.7 – 8.3. These samples (ID 40, 47, 51, 52) were collected in the surrounding of North Mara Gold Mine, where drainage from the mining site and seepage from the tailing storage facility are contaminating streams and rivers. High As levels and high pH in the samples confirms the mobility of the metalloid as oxyanion.

400 BH SW spring Surface

300

200 As field (µg/l)field As 100 R² = 0.413

0 5.00 5.50 6.00 6.50 7.00 7.50 8.00 8.50 9.00 pH

Figure 31. Relationship of As with pH in surface water (yellow), shallow well (light blue), spring (green) and borehole (dark-blue)

In Figure 32 the combined effect of pH and Eh on As dissolved species is proposed with the plot of the water samples collected during the present study.

Figure 32. pH-Eh diagram for the As-O-H system with plot of the sampled water

All the samples fall in the stability field of the oxidised form of As, arsenate (As V), suggesting the predominance of this specie over the reduced form, arsenite (As III). The analysis of the field speciated samples also report equivalent results.

5.2.4.2 Major ions Since the geological occurrence of As in the study area is presumed to be associated 2- with sulphide minerals and gold mineralization, the correlation between As and SO4 2- was investigated, Figure 33. As levels show a weak positive trend with SO4 only in shallow wells. Weak correlation in shallow wells and lack of similar trend in borehole and surface water infer the importance of other geochemical processes and conditions in As mobility once the metalloid has entered into the aquatic system

250 BH SW Surface 200

150 As (µg/l)As 100

0 0 1 10 100 1000

2- SO4 (mg/l) 2- Figure 33. Correlation between As and SO4 characterizing different types of water source. Note that x-axis is in log scale

Arsenic shows a strong negative correlation with Ca in carbonate rocks, which may indicate sequestration of As by calcite. The hypothesis of As co-precipitation with calcite is supported by the precipitation of calcite shown above (Figure 27) and by a - 2 weak negative trends between As and HCO3 (R =0.25). Although co-precipitation of As with calcite has not been fully addressed by the scientific community, few studies have shown that calcite could immobilize a large part of dissolved As (Winkel et al., 2013).

160

carbonate 120

As (µg/l)As R² = 0.54 40

0 40 60 80 100 120 140

Ca (mg/l)

Figure 34. Relationship of As with (left) Na+ and (right) Ca2+ in silicate (red) and carbonate (blue) rocks.

5.2.4.3 DOC The importance of DOC in the mobility of As and of other metals has been already highlighted for borehole waters. A weak correlation between DOC and As is also found in shallow wells, Figure 37.

250 BH SW Surface 200

150

100

As (µg/l)As R² = 0.35

0 1 2 3 4 5 6 7 8 9 DOC (mg/l) Figure 35. Correlations between As and DOC characterizing different types of water source

A better understanding of the relation between DOC and As is provided by Figure 38, where DOC levels are plotted together with field and lab measurements of As.

Field As Lab As DOC 350 10

9 300 8

250 7

6 200 5 150

4 (mg/l)DOC As (µg/l)As

100 3 2 50 1

0 0 3GT 8GT 9GT 12GT 19GR 29BM 30BM 32BM 34BM 39BM 40NM 42NM 43NM 44NM 49NM 50NM 51NM 52NM 54NM ID Sample Figure 36. Graph showing DOC levels (green line), field measurements of As (blue line) and lab measurements of As (red line) in the sampled water.

A good match of peaks and valleys in Figure 38 between As and DOC levels emphasizes the role of humic acid in As mobility and transport. Number of observation in experimental studies show evidence for binding of As (V) by natural organic matter through surface complexation (Pickering & Thanabalasingam, 1986).

5.2.4.4 Al, Fe, Mn Al, Fe and Mn oxide-hydroxides play an important role in the mobility of arsenic through adsorption processes and co-precipitation. No significant trends are observed between these metals and As, Figure 39. Considering that Fe, Al and Mn concentrations refer to the dissolved species, and not to the particulate fraction, lack of correlation do not exclude that co-precipitation with oxide-hydroxides is occurring and that it represents a major sink for dissolved As. As pointed out from the analysis of the seven unfiltered samples, Al, Fe and Mn are mostly found in the particulate fraction, which is the one responsible in binding heavy metals.

BH SW Spring Surface 350

300

250

200

As (µg/l)As 150

100

0 1 10 100 1000 Fe (µg/l) 350 300 250 ) 200 150 As (µg/lAs 100 50 0 10 100 1000 Al (µg/l) 350 300 250 200 150 As (µg/l)As 100 50 0 1 10 100 1000 Mn (µg/l)

Figure 37. Relationship between As and (top) Fe; (middle) Al; (bottom) Mn. Note the log scale in the x axis.

5.2.4.5 Other trace elements Correlations between As and other trace elements were only preliminary investigated during the present study and no meaningful trends were observed. However, when plotting As vs Ba a strong difference is observed between Geita and Mara, confirming once more a different geological setting in the two regions (Figure 40). High Ba concentrations in Geita are associated with low As levels, indicating granite-like type of rock rich in Ba and poor in As. In contrast, Mara is characterized by an As-enriched geological setting with low occurrence of Ba, indicating that granite-like rocks do not predominate in this region.

350

300 Geita Musoma/Mara 250

200

150 As (µg/l)As

100

0 0 200 400 600 Ba (µg/l)

Figure 38. Relationship between As and Ba in Geita and Mara regions.

5.3 Geochemical Modelling

Geochemical modelling was performed using the cluster of surface water samples showing SO4 as the predominant anion, see Piper diagram in Figure 15. Proximity to mining sites and high SO4 concentrations indicate that contamination by mine effluents is likely to occur in these surface water sources. Accordingly, they represent a meaningful set of samples to understand source and mobilization of arsenic in mining areas.

The result and interpretation of the modelling should be considered as an estimate of the distribution of aqueous chemical species and of the major dissolution/precipitation processes occurring in water.

5.3.1 Saturation Indexes Among all the solid phases included in Phreeqc database, only few are relevant to the present study. Figure 39 focuses on solid phases playing an important role in the mobilization of arsenic species.

All As minerals are highly under saturated, except from scorodite, which is near to equilibrium. Scorodite is a common weathering product of arsenopyrite and its persistence in nature has been shown to control As concentrations in water (Rimstidt & Dove, 1985). Slightly negative values of SI may therefore indicate dissolution of this mineral, giving goethite and aqueous arsenate. Once As is released into the water, it tends to be readily adsorbed on other solid phases, such as Fe, Al, Mn oxide-hydroxides and clay minerals. The positive SI for Fe (III) oxide-hydroxides (ferrihydrite, and goethite) and for Al oxide-hydroxides (gibbsite, boehmite and diaspore) reveals that water is supersaturated with regard to these solid phases, suggesting their tendency to precipitate. Co-precipitation of As with Fe and Al oxide- hydroxides is a major sink for dissolved As and it therefore represents a feasible explanation for the surprising difference between field and laboratory measurements of As. Mn oxides-hydroxides are all highly under saturated, with exception of manganite, that is close to equilibrium in sample ID 34, 40 and 51. The water samples are also oversaturated with respect to clay minerals like kaolinite, montmorillonite and illite, which all show strong sorption behaviour towards As(V) (Manning & Goldberg, 1997).

9 Sample 14 7 Sample 34 Sample 40 Sample 51 5

1 K

-1 - Illite Saturation Index (S.I.) Index Saturation Al(OH)3 Gibbsite Goethite Diaspore Kaolinite Scorodite Boehmite -3 Jarosite Magnetite _Scorodite Manganite Ferrihydrite Al Rhodocrosite AlFerryhydrite

-5 Montmorillonite

-7

Figure 39. Saturation indexes for selected solid phases controlling fate and mobility of As.

When looking at the saturation indexes of some selected solid phases, which can give indication about the local geology (Figure 40), a substantial difference is seen among the four samples. The carbonate minerals (aragonite, calcite and dolomite) are oversaturated in samples ID 34 and 51 and quartz is under saturated only in sample 34. This is line with the previous findings which indicate silicate weathering in Geita (Sample ID 14) and a more complex geology in Mara (samples ID 34, 40, 51), where both silicate and carbonate weathering occur. In addition, oversaturation of calcite, dolomite and aragonite in sample ID 51 may be an indicator of seepage from the tailing storage facility at North Mara Gold Mine. Since the process water stored in the artificial reservoir is rich in lime, a leakage from the tailing dam would result in high carbonate concentrations and in the subsequent precipitation of carbonate minerals in the stream from which sample ID 51 was collected. However, data are inconclusive and the local geology surely contributes to the high carbonate levels in stream water around North Mara Gold Mine.

3 Sample 14 2 Sample 34 Sample 40 1 Sample 51 0

-1 Albite Barite Calcite Quartz Gypsum Celestite Dolomite Aragonite Anhydrite

-2 Magnesite Chalcedony

Saturation Index (S.I.) Index - Saturation 3

-4

-5

Figure 40. Saturation indexes for selected solid phases providing indications about the local geology.

5.3.2 Aqueous species distribution The modelled aqueous speciation for the four samples is presented in Table 7, with samples rearranged according to pH, from the lowest value (7.43 in ID 14) to the highest (8.37 in ID 34). The speciation in the four samples is roughly similar. Among - the aluminium phases, Al(OH)4 predominates in all the samples ranging from 98% to 100 % with increasing pH. Similar trend is observed in the distribution of iron + species in samples ID 34 – 40 – 51, where Fe(OH)3 and Fe(OH)2 tend to be less dominant with increasing pH, in favour of Fe(OH)4.However, sample ID 14 shows significant occurrence of Fe(II) species, contributing to approx. 45 % of total dissolved Fe. Co-occurrence of Fe(III) and Fe(II) species and high concentration of Mn, present as Mn+2 , suggest that Fe and Mn are the predominant Redox couples in sample ID 14. This hypothesis is supported by field measurements of Redox potential, which indicate a value of approx. 200 mV in this sample and values above

400 mV in the other three samples, where O2 reduction is the main Redox process.

- Finally, a diverse speciation of As species is observed. HAsO4 predominate, ranging from 31 to 60 %, followed by arsenate complexes with Ca and Mg. pH regulates the - presence of H2AsO4 and the ration between hydrogen and not- hydrogen Ca/Mg-As complexes. Different studies report the presence of Ca and Mg arsenates in their precipitates; however, it is not clear whether all these species really exist in natural

waters. (Bowell et al., 2015) . Nevertheless these findings give indication of the role of Ca and Mg in the mobilization and speciation of As, and suggests further investigation.

Table 7. Distribution of aqueous chemical species (%) in selected surface water samples. oxs stands for oxidation state.

ID 14 ID 40 ID 51 ID 34

pH : 7.43 pH : 7.82 pH : 8.02 pH : 8.37

specie oxs % specie oxs % specie oxs % specie oxs %

Al Al Al Al - - - - Al(OH)4 (III) 98 Al(OH)4 (III) 99,2 Al(OH)4 (III) 99,5 Al(OH)4 (III) 100

Al(OH)3 (III) 1,6 Al(OH)3 (III) 0,7 Al(OH)3 (III) 0,5 Al(OH)3 (III) 0 + + + + Al(OH)2 (III) 0,4 Al(OH)2 (III) 0,1 Al(OH)2 (III) 0 Al(OH)2 (III) 0 As As As As -2 -2 -2 HAsO4 (V) 50,1 HAsO4 (V) 60 HAsO4 (V) 53,8 HAsO4-2 (V) 31,2 - - CaHAsO4 (V) 14,8 CaHAsO4 (V) 10,2 CaAsO4 (V) 15,7 MgAsO4 (V) 28,1 - - - H2AsO4 (V) 11 CaAsO4 (V) 7,9 CaHAsO4 (V) 12,5 CaAsO4 (V) 20,3 - - MgHAsO4 (V) 8,3 H2AsO4 (V) 6,3 MgAsO4 (V) 5,9 MgHAsO4 (V) 11,3

FeAsO4 (V) 5,5 MgHAsO4 (V) 6,2 MgHAsO4 (V) 5,3 CaHAsO4 (V) 7,3 - - - CaAsO4 (V) 5 MgAsO4 (V) 4,3 H2AsO4 (V) 3,4

H3AsO3 (III) 0,0 H3AsO3 (III) 0 H3AsO3 (III) 0 H3AsO3 (III) 0 Fe Fe Fe Fe

Fe(OH)3 (III) 40,4 Fe(OH)3 (III) 82,6 Fe(OH)3 (III) 83,2 Fe(OH)3 (III) 77,9 +2 + - - Fe (II) 29,4 Fe(OH)2 (III) 10,5 Fe(OH)4 (III) 8,5 Fe(OH)4 (III) 19,4 + - + + Fe(OH)2 (III) 13,2 Fe(OH)4 (III) 5,5 Fe(OH)2 (III) 7,2 Fe(OH)2 (III) 2,7 +2 +2 +2 FeSO4 (II) 8,9 Fe (II) 0 Fe (II) 0 Fe (II) 0 + FeHCO3 (II) 4,9 Mn Mn Mn Mn +2 +2 +2 Mn (II) 62,6 Mn (II) 65,8 Mn (II) 54,1 MnCO3 (II) 53,3 +2 MnSO4 (II) 19 MnCO3 (II) 20,1 MnCO3 (II) 29,4 Mn (II) 35,8 + + + MnHCO3 (II) 9,2 MnHCO3 (II) 7,9 MnSO4 (II) 8,4 MnHCO3 (II) 5,9 + MnCO3 (II) 9 MnSO4 (II) 5,9 MnHCO3 (II) 7,6 MnSO4 (II) 4,8 S(6) S(6) S(6) S(6) -2 -2 -2 -2 SO4 (VI) 83 SO4 (VI) 89,9 SO4 (VI) 87,7 SO4 (VI) 81,8

CaSO4 (VI) 10,1 CaSO4 (VI) 6,3 CaSO4 (VI) 8,4 MgSO4 (VI) 9,9

MgSO4 (VI) 4,6 MgSO4 (VI) 3,05 MgSO4 (VI) 2,8 CaSO4 (VI) 7,9

6. CONCLUSIONS

Overall, the present study has accomplished the goal of providing a better understanding of the problem of arsenic in drinking water sources in proximity of gold mining areas, in Tanzania. Hypotheses about occurrence, source and fate of As are proposed and signs for evidence are shown.

The occurrence of As in drinking water is of great concern in parts of the Lake Victoria Basin in Tanzania. 53% of the sampled water do not comply with the WHO recommended limit of 10 µg/L , posing a severe health risk, which should be promptly addressed. The spatial variation of arsenic in the study area is highly heterogeneous and it mainly depends on local geology and proximity to gold mining activities. Low arsenic levels in water withdrawn from drilled boreholes suggest that deep groundwater (> 40m) generally represent a source of safe drinking water.

The source of As is deduced to be weathering of sulphide minerals naturally occurring in zones of gold mineralization within the Greenstone Belts. Oxidising conditions (Eh>350 mV) of the sampled water, high sulphate concentrations in stream and rivers, and positive As-SO4 correlation in shallow wells and surface water, support this hypothesis. Moreover, high As levels associated with high Fe and Al concentrations in some surface water samples and saturation indexes near to equilibrium for scorodite and Al-scordite suggest that solubility of weathering products plays a key role in controlling arsenate levels in water, rather than dissolution of primary minerals (arsenopyrite).

Once arsenic has entered into the aquatic environment, several geochemical processes control its mobility. Large discrepancies between field and lab measurements indicate a strong partition of As into the particulate fraction. The geochemical modelling shows over-saturation with regard to Al/Fe oxide-hydroxides and clay minerals, suggesting the tendency of these solid phases to precipitate. Adsorption on clay minerals and co-precipitation with oxide-hydroxides are therefore presumed to immobilize most of the dissolved As. In addition, a strong match between peaks in As and DOC concentrations reveals the importance of surface complexation by humic acids in the mobility of the metalloid. Finally, sequestration

of As by calcite seems to be a major sink for dissolved As in groundwater, which should be further investigated.

In conclusion, this project has opened the doors to new stimulating research questions which should be further investigated.

6.1 Final Recommendations

Before any further research, it is highly recommended to collect baseline information about groundwater systems in the Lake Victoria Basin. It is in fact essential in order to understand the distribution of As in the subsurface environment. Moreover, the present study has pointed out a substantial difference in the local geological setting between Mara and Geita Greenstone Belt and its influence in the occurrence of As, which should be further investigated. It is therefore recommended to conduct a more extensive water sampling with the aim of covering a larger portion of the Lake Victoria Basin and of better understanding the local geology. A more extensive water sampling is also crucial to asses As exposure via drinking water within rural communities which were not considered in the present study. The leading goal should be to provide an evaluation on As contamination of drinking water source at district level in the whole Lake Victoria Basin in Tanzania. Since surprising discrepancies between field and laboratory measurements of As were encountered during the present study, it is recommended to review the water sampling procedure. In particular, it is suggested to collect an unfiltered and acidified sample in addition to the filtered and acidified one. In this way the partition of As and other heavy metals into the particulate fraction could be determined. This consideration is especially valid when sampling in environments characterized by oxidising conditions, since adsorption to hydroxides and clay minerals is enhanced in these conditions.

BIBLIOGRAPHY

Almås, Å. R., & Manoko, M. L. (2012). Trace elements concentrations in soil, sediments, and waters in the vicinity of Geita Gold Mines and North Mara Gold Mines in northwest Tanzania. Soil and Sediment Contamination , 21(2), 135-159.

Anawar, H. M., Akai, J., Komaki, K., Terao, H., Yoshioka, T., Ishizuka, T., et al. (2003). Geocehmical occurence of arsenic in groundwater of Bangladesh: sources and mobilization processes. Geochemical Exploration , 77, 109-131.

Anhaeusser, C. R. (2014). Archean greenstone belts and associated granitic rocksv - A riview. Journal of African Earth Sciences , 684-732.

Appelo, C., & Postma, D. (2005). Geochemistry, groundwater and pollution (2nd ed.). Amsterdam: A.A. Balkema Publishers.

Armienta, M. A., Rodriguez, C. R., Ongley, L. K., Brust, H., Morales, F., Aguayo, A., et al. (2007). Origin and fate of arsenic in a historic mining area of Mexico. In P. Bhattacharya, A. B. Mukherjee, J. Bundschuh, R. Zevenhoven, & R. H. Loeppert (Eds.), Arsenic in Soid and Groundwater Environment (pp. 473-498). Ann Arbor, Michigan: Elsevier B.V.

Bartha, A., Varsányi, I., & Fodré, Z. (1991). Arsenic in drinking water and mortality in the Southern Great Plain, Hungary. Environmental Geochemistry and Helath , 13, 14-22.

Bednar, A. J., Garbarino, J. R., Ranville, J. F., & Wildeman, T. R. (2002). Preserving the distribution of inorganic arsenic species in groundwater and acid mine drainage samples. Envrionmental Science and Technology , 36, 2213-2218.

Bejarano, G. S., & Nordberg, E. (2003). Mobilisation of Arsenic in the Rio Dulce Alluvial Cone, Santiago del Estero Province, Argentina. Stockholm: Dept. of Land and Water Resources Engineering.

Bhattacharya, P., Frisbie, S. H., Naidu, R., Jacks, G., & Sarkar, B. (2002). Arsenic in the Environment: A Global Prospective. In B. Sarkar (Ed.), Handbook of Heavy Metals in the Environment (Chapter 6) (pp. 147-215). New York: Marcell Dekker.

Bitala, M. F., Kweyunga, C., & Manoko, M. L. (2009). Levels of heavy metals and cyanide in soil, sediment and water from the vicinity of North Mara Gold Mine in Tarime District, Tanzania. Christian Council of Tanzania.

Bowell, R. J., Alpers, C. N., Jamieson, H., Nordstrom, D. K., & Majzlan, J. (2015). Arsenic: environemntal geochemistry, mineralogy and microbiology (Vol. 79). U.S.A.: De Gruyter.

Bowell, R. J., Warren, A., Minjera, H. A., & Kimaro, N. (1995). Environmental impact of former gold mining on the Orangi river, Serengeti N.P., Tanzania. Biogeochemistry , 28, 131- 160.

Buschmann, J., Kappeller, A., Lindauer, U., Kistler, D., Berg, M., & Sigg, L. (2006). Arsenite and arsenate binding to dissolved humic acids: influence of pH, type of humic acid, and aluminium. Environmental Science Technology , 40, 6015-6020.

Crul, R. C. (1995). Limnology and hydrology of Lake Victoria. Paris: UNESCO.

Dove, P. M., & Rimstidt, D. J. (1985). The solubility and stability of scorodite, FeAsO4:2H2O. American Mineralogist , 70, 838-844.

Evjen, C. (2011). Arsenic and trace metals in hair, nails and blood of villagers from the vicinity of a gold mine in Tanzania. Oslo: The Department of Biology.

FDMT. (2016, 12 15). Flood & Drought Management Tools. Retrieved 1 6, 2017, from http://fdmt.iwlearn.org/en

Geological Survey of Tanzania. (2015). Geological Survey of Tanzania. Retrieved October 10, 2016, from http://www.gst.go.tz/

GeoNode. (2015). RCMRD GeoPortal. Retrieved December 20, 2016, from http://servirportal.rcmrd.org/

Gomez-Caminero, A., P., H., M., H., E., K., D.R., L., M., M., et al. (2001). Arsenic and arsenic compounds. (2nd edition ed.). Geneve: World Health Organization.

Gray, I. M., & Macdonald, A. S. (1964). Tanzania Quarter Degree Sheets 4NE and 5NW. Dodoma.

Hasan, M. A., Ahmed, K. M., Sracek, O., Bhattacharya, P., von Brömssen, M., Broms, S., et al. (2007). Arsenic in shallow groundwater of Bangladesh: investigations from three different physiographic settings. Hydrogeology Journal , 15, 1507-1522.

Hounslow, A. W. (1995). Water Quality Data: Analysis and Interpretation. Florida: CRC Press.

IARC. (2004). Some drinking-water Disinfectants and Contaminats, including Arsenic. Lyon: WHO.

Ijumulana, J., Bhattacharya, P., & Mtalo, F. (2016). Arsenic occurence in groundwater sources of Lake Victoria basin in Tanzania. In P. Bhattacharya, M. Vahter, J. Jarsjö, J. Kumpiene, A. Ahmad, C. Sparrenbom, et al. (Eds.), rsenic Research and Global Sustainability: Proceedings of the Sixth International Congress on Arsenic in the Environment (pp. 86-87). Stockholm: CRC Press.

Kabete, J. M., Groves, D. I., McNaughton, N. J., & Mruma, A. H. (2011). A new tectonic and temporal framework for the Tanzania Shield: Implications for gold metallogeny and undiscovered endowment. Ore Geology Reviews , 48, 88-124.

Kahatano, J. M., & Mnali, S. R. (1995). Heavy metal contamination due to artisanal gold mining in the Tanzanian Lake Victoria Gold Fields. Environment and mining in eastern and southern Africa - selected papers from an international conference 23-27th October 1995, (pp. 66-81). Mwanza, Tanzania.

Kavana, E. (2015). Geolgoical Mapping, Structural setting andPetrographic Description of the Archean Volcanic Rocks of Mnanka Area, North Mara. Melbourne: World Geothermal Congress.

Lake Victoria, B. (2014). Lake Victoria Basin Coimmission: One peopl, One destiny. Retrieved January 13, 2017, from https://www.lvbcom.org/

Mahoney, J. (2016). Phreeqc Database.

Manning, B., & Goldberg, S. (1997). Adsorption and stability of arsenic (III) at the clay mineral-water interface. Environ. Sci. Technol. , 31, 2005-2011.

Manya, S., & Maboko, M. (2015). The Neoarchean Greenstone Belts of northern Tanzania. Dar Es Salaam: Department of Geology, University of Dar Es Salaam.

Ministry of Water. (2013). Water Point Mapping Project. Retrieved October 2, 2016, from http://wpm.maji.go.tz/

Naidu, R., & Bhattacharya, P. (2006). Management and remediation or arsenic from contaminated water (Chapter 18). In R. Naidu, E. Smith, G. Owens, P. Bhattacharya, & P. Nadedaum (Eds.), Managing Arsenic in the Environemnt: From Soil to Human Health (pp. 331-354). Melbourne, Australia: CSIRO Publishing.

National Bureau of Statistics. (2015). Tanzania National Bureau of Statistics. Retrieved October 20, 2016, from http://www.nbs.go.tz/

Nriagu, J. O., Bhattacharya, P., Mukherjee, A. B., Bundschuh, J., Zevenhoven, R., & Loeppert, R. H. (2007). Arsenic in soil and groundwater: an overview. In P. Bhattacharya, A. B. Mukherjee, J. Bundschuh, R. Zevenhoven, & R. H. Loeppert, Arsenic in Soil and Groundwater Environment. Elsevier B.V.

Nyanza, E. C., Dewey, D., Thomas, D. S., Davey, M., & Ngallaba, S. E. (2014). Spatial distribution of mercury and arsenic levels in water, soil and cassava plants in a community with long history of gold mining in Tanzania. Bulletinof Environmental Contamination and Toxicology , 93(6), 716-721.

Pankhurst, D. L., & Appelo, C. L. (1999). Description of input and examples for PHREEQC Version 3 - A computer program for speciation, batch-reaction, one-dimensional transport, and inverse geochemical calculations. Water Resources Investigations Report. Denver: U.S. Department of Interior.

Pickering, W. F., & Thanabalasingam, P. (1986). Arsenic Sorption by Humic Acids. Environmenta Pollution , 233-246.

R. Kassenga, G., & R. Mato, R. (2008). Arsenic contaminations levels in drinking water sources in mining areas in Lake Victoria Basin, Tanzania, and its removal usign stebilized ferrasols. Int. J. Biol. Chem. Sci. 2(4) , 389-400.

Rahman, M. M., Mukherjee, D., Sengupta, M., Chowdhury, U., Chanda, C., Selim, M., et al. (2002). Effectiveness and reliability of arsenic field testing kits: are the million dollar screening projects effective or not. Environ.Sci.Technology , 36, 5385-5394.

Raychowdhury, N., Mukeherjee, A., Bhattacharya, P., Johannesson, K., Bundusch, J., Sifuentes, B., et al. (2013). Prvenance and fate or arsenic and other solutes in the Chaco- Pampean Plain of the Andean foreland, Argentina: From prospectives of hydrogeochemical modeling and regional tectonic setting. Journal of Hydrology , 518, 300-316.

Rimstidt, D., & Dove, P. M. (1985). The solubility and stability of scorodite, FeAsO4:2H2O. American Mineralogist , 70, 838-844.

Sangea, Upton, & Ó Dochartaigh. (2016). Africa Groundwater Atlas: Hydrogeology of Tanzania. British Geological Survey. Retrieved 03 01, 2017, from http://earthwise.bgs.ac.uk/index.php/Hydrogeology_of_Tanzania

Sanislav, I. V., Brayshaw, M., Kolling, S. L., Dirks, P. G., Cook, Y. A., & Blenkinsop, T. G. (2016). The structural history and mineralization controls of the world-class Geita Hill gold deposit, Geita Greenstone Belt, Tanzania. Mineralium Deposita , 1-23.

Sanislav, I. V., Dirks, P. G., Cook, Y. A., Blenkinsop, T. G., & Kolling, S. L. (2014). A giant gold system, Geita greenstone belt, Tanzania. Acta Geologica Sinica (English Edition) , 88(2), 110- 111.

SEAB. (2017). Seab Geam ltd. Retrieved January 3, 2017, from http://www.seabgem.com/ebrochure/musomaMaraGreenstone.html

SEAL Analytical. (2014). SEAL Analytical. Retrieved December 18, 2016, from http://www.seal-us.com/Products/AA3SFAAnalyzer/tabid/59/Default.aspx

Shedafa, M. H., & Johnston, R. (2013). Groundwater vulnerability to geogenic contaminats: a case study, Tanzania. 36th WEDC International Conference. Nakuru, Kenya.

Skougstad, M. W., & Horr, A. C. (1963). Occurrence and Distribution of Strontium in natural water. Whashington: U.S. Department of Interior.

Smedley, P. K. (2002). A review of the source, behaviour and distribution of arsenic in natural waters. Applied Geocehmistry , 17, 517-568.

Smedley, P. L., Edmunds, W. M., & Pelig-Ba, K. B. (1996). Mobility of arsenic in groundwater in the Obuasi gold-mining area of Ghana: some implications for human health. In R. Fuge, &

J. McCall (Eds.), Environmental geochemistry and health: with special reference to developing countries (pp. 163-181). London: Chapman and Hall.

Sracek, O., Bhattacharya, P., Brömssen, M., & Jacks, G. (2005). Natural enrichment of arsenic in groundwaters of Brahamanbaria district, Bangladesh: geochemistry, speciation modeling and multivariate statistics. Natural Arsenic in Groundwater: Occurence, Remediation and Management (pp. 133-143). London: Taylor & Francis Group.

The World Bank . (2016). The World Bank Group. Retrieved November 23, 2016, from http://data.worldbank.org/country/tanzania

Triplett, L. (2006). Sampling procedures for groundwater monitoring wells. Minneota: Minnesota Pollution Control Agency.

U.S. Department of the Interior. (2015). United States Geological Survey. Retrieved September 20, 2016, from https://answers.usgs.gov/

UNEP. (2006). Lake Victoria Basin Environment Outlook: Environment and Development. Nairobi: UNEP.

United Nations. (2008). Promoting mineral cluters: The case of Tanzania. United Nations. van Straaten, P. (2000). Mercury contamination associated with small-scale gold mining in Tanzania and Zimbabwe. The Science of Total Environment , 259, 105-113.

Varsányi, I., Fodré, Z., & Bartha, A. (1991). Arsenic in drinking water and mortality in Southern Great Plain, Hungary. Environmental Geochemistry and Health , 13, 14-22.

WHO. (2011). Guidelines for drinking-water quality. Geneva: WHO Library Cataloguing.

Williams, M. (2001). Arsenic in mine waters: an international study. Environmental geology , 40(3), 267-278.

Winkel, L., Casentini, B., Bardelli, F., Voegelin, A., Nikolaidis, N., & Charlet, L. (2013). Speciation of arsenic in Greek travertines:Co-precipitation of arsenate with calcite. Geochemica et Cosmochimica Acta , 106, 99-110.

Wlliams, M., Fordyce, F., Pajiitprapapon, F., & Charoenchaisiri, P. (1996). Arsenic contamination in surfcae drainage and groundwater in part of the southeast Asian tin belt, Nakhon Si Thammarat Province, southern Thailand. Environemntal Geology , 27, 16-33.

Zaldivar, R. (1974). Arsenic contamination of drinking water and foodstuffs causing endemic chronic poisoning. Beiträge zur Pathologie , 151 (4), 384-400.

APPENDIX A. Hydrogeological map of Tanzania (Sangea et al.,2016)

APPENDIX B. Water sampling data sheet used during the water sampling

APPENDIX B.(cont) Work procedure adopted during the water sampling

APPENDIX C. Figure 1. Arsenic concentrations in small scale gold mining areas: Geita Rural and Butiama/Musoma.

APPENDIX C. (cont) Figure 2. Arsenic concentrations around Geita Gold Mine.

APPENDIX C. (cont) Figure 3. Arsenic concentrations around North Mara Gold Mine.

APPENDIX D. Table listing location, type of water source and field measurements for the sampled waters (from ID 1 to ID 27).

Water Filtered/Not ID Location Latitude (S) Longitude (E) Altitude Temp pH EC Redox Asfield Source Filtered

m °C µS/cm mV ppb 1 Geita Town -2,872 32,216 1303,4 SW 24,2 5,9 93,3 448,6 0 Filtered 2 Geita Town -2,886 32,210 1219 Spring 25,2 6,47 179,1 441 5 Filtered 3 Geita Town -2,889 32,212 1218 Surf 27,2 7,4 258 393 30 Filtered 4 Geita Town -2,873 32,234 1248,9 SW 26,8 6,14 217 422,8 40 Filtered 5 Geita Town -2,877 32,230 1239 BH 27,6 5,83 256 473,3 10 Filtered 6 Geita Town -2,878 32,227 1237 Spring 25,8 5,87 105,8 504,1 10 Filtered 7 Geita Town -2,878 32,246 1244,7 BH 26,8 6,77 314 452,3 0 Filtered 8 Geita Town -2,871 32,238 1243,7 SW 24,3 6,41 285 309 10 Filtered 9 Geita Town -2,854 32,230 1377,4 Spring 23,4 6,09 49,8 397,4 10 Filtered 10 Geita Town -2,874 32,229 1259 BH 25,9 5,58 144,2 479,4 0 Filtered 11 Geita Town -2,874 32,229 1259 BH 26,8 5,9 167,1 508,2 0 Filtered 12 Geita Town -2,884 32,233 1227,8 BH 26,8 6,69 398 441,9 10 Filtered 13 Geita Town -2,842 32,146 1160,6 Spring 27,1 5,33 63,7 503,7 0 Filtered 14 Geita Town -2,837 32,155 1217,09 Surf 25,5 7,43 1590 214,7 30 Filtered 15 Geita Town -2,900 32,152 1217,09 Spring 26,9 5,81 126,2 447,6 20 Filtered 16 Geita Town -2,919 32,162 1236,1 BH 25,3 6,81 517 412,5 0 Filtered 17 Geita Rural -3,112 32,235 1324 BH 25,7 6,65 760 449,8 0 Filtered 18 Geita Rural -3,094 32,235 1250,3 Spring 26,9 5,76 97,1 456,5 20 Filtered 19 Geita Rural -3,113 32,239 1311,6 BH 27,6 6,85 555 225,1 10 Filtered 20 Geita Rural -3,153 32,257 1271,01 BH 28 6,23 169,2 423,3 0 Filtered 21 Geita Rural -3,155 32,257 1271 Surf 29,3 7,24 207 419,2 70 Not Filtered 22 Geita Rural -3,195 32,315 1337,8 BH 27,1 5,86 156,4 480,7 10 Filtered 23 Geita Rural n.t. n.t n.t. SW n.t. n.t n.t. n.t n.t. Not Filtered 24 Geita Rural -3,121 32,038 1240 BH 24,6 6,79 332 415,9 10 Filtered 25 Geita Rural -3,127 32,041 1232 SW 24,8 5,89 107,7 427,6 20 Filtered 26 Butiama/Musoma -1,790 33,750 1185 SW 26,2 6,55 575 436,6 10 Filtered 27 Butiama/Musoma -1,805 33,776 1160,4 Surf 27,7 7,73 378 422,9 70 Not Filtered

APPENDIX D. (cont) Table listing location, type of water source and field measurements for the sampled waters (from ID 28 to ID 54).

Water Filtered/Not ID Location Latitude (S) Longitude (E) Altitude Temp pH EC Redox Asfield Source Filtered

m °C µS/cm mV ppb 28 Butiama/Musoma -1,811 33,762 1189,9 BH 29,3 6,9 1058 433,8 0 Filtered 29 Butiama/Musoma -1,810 33,762 1196,2 SW 25,6 7,46 607 404,5 50 Filtered 30 Butiama/Musoma -1,814 33,759 1176,3 SW 7,82 761 410,7 100 Filtered

31 Butiama/Musoma -1,874 33,511 1173,3 SW 26,2 6,62 114,5 455,7 30 Filtered 32 Butiama/Musoma -1,828 33,587 1201 SW 26,3 7,25 955 458,4 40 Filtered 33 Butiama/Musoma -1,821 33,980 1303,8 Surf 26,1 7,81 346 405,8 50 Filtered 34 Butiama/Musoma -1,792 34,097 1352 Surf 26,2 8,37 651 397,9 70 Filtered 35 Butiama/Musoma -1,803 34,100 1391,9 SW 26,5 6,56 741 455,5 70 Filtered 36 Butiama/Musoma -1,800 34,105 1398,8 BH 26,7 7,47 795 446,7 10 Filtered 37 Butiama/Musoma -1,740 34,148 1243,3 Surf 28,3 8,35 1014 391,5 10 Filtered 38 Butiama/Musoma -1,664 34,134 1181,8 Surf 32,8 8,33 156,7 364,4 10 Not Filtered 39 Butiama/Musoma -1,590 33,936 1171,6 BH 28 7,17 1023 433,9 10 Filtered 40 North Mara -1,447 34,538 1279 Surf 25,4 7,82 509 410,8 150 Filtered 41 North Mara -1,522 34,566 1173 Surf 25,5 7,64 477 432,2 70 Not Filtered 42 North Mara -1,522 34,565 1170 Surf 27,6 7,77 121,2 430 30 Filtered 43 North Mara -1,464 34,534 1286,8 BH 25,5 6,99 1016 260,9 30 Filtered 44 North Mara -1,458 34,525 1287,6 Spring 26,5 6,06 307 403,4 50 Filtered 45 North Mara -1,459 34,521 1290,7 SW 25,8 6,39 448 411 70 Not Filtered 46 North Mara -1,462 34,472 1184 BH 26,7 7,01 687 211,1 150 Filtered 47 North Mara -1,460 34,471 1182,8 Surf 26,2 7,97 542 373,1 300 Not Filtered 48 North Mara -1,470 34,472 1209,1 BH 27,3 7,02 1155 201 10 Filtered 49 North Mara -1,466 34,527 1306 BH 27,4 7,15 715 413,7 60 Filtered 50 North Mara -1,480 34,524 1242 Spring 25,3 6,93 864 431 20 Filtered 51 North Mara -1,491 34,508 1170 Surf 24 8,02 771 395,4 220 Filtered 52 North Mara -1,499 34,513 1160 Surf 26,6 7,71 130,2 397,2 160 Filtered 53 North Mara -1,488 34,504 1216 BH 27,3 7,21 1059 224,5 10 Filtered 54 North Mara -1,475 34,480 1194 Surf 33 8,02 4040 307 300 Filtered

APPENDIX E. Table listing location, type of water source, major ions, DOC and Si content for the sampled waters (from ID 1 to ID 27)..

Water ID Location Alkalinity HCO - Cl- NO - SO 2- PO 3- Na+ Mg2+ K+ Ca2+ DOC Si Source 3 3 4 4 mmol mg/L mg/L mg/L mg/L µg/L mg/L mg/L mg/L mg/L mg/L mg/L 1 Geita Town SW 0,81 49,13 0,76 0,34 1,11 11,8 10,24 1,73 0,72 3,91 1,741 22,27 2 Geita Town Spring 0,77 47,01 12,12 29,43 2,73 8,49 17,27 3,84 1,12 8,55 1,395 19,13 3 Geita Town Surf 1,98 120,54 19,27 0,2 4,61 3,76 25,02 4,43 6,67 15,44 7,583 5,48 4 Geita Town SW 0,99 60,51 24,99 21 3,53 3,61 27,22 2,24 1,57 8,86 3,321 28,67 5 Geita Town BH 0,65 39,61 21,89 55,05 4,46 3,35 23,42 3,76 7,95 14,08 1,521 17,44 6 Geita Town Spring 0,54 33,18 3,54 17,96 0,98 7,4 10,54 1,96 0,4 5,7 1,252 14,18 7 Geita Town BH 3,25 198,21 4,27 0,17 7,38 66,15 40,08 5,13 1,24 17,26 1,322 45,6 8 Geita Town SW 1,49 90,93 23,93 24,18 6,57 27,13 25,15 4,99 5,99 14,65 3,321 17,38 9 Geita Town Spring 0,32 19,52 1,51 0,2 1,54 3 3,54 1,08 0,78 1,75 1,367 6,29 10 Geita Town BH 0,77 47,21 8,03 22,14 1,12 3 12,52 2,39 1,32 7,03 1,325 16,05 11 Geita Town BH 0,65 39,53 11,08 36,79 1 13,26 18,08 2,63 0,78 7,66 1,646 20,21 12 Geita Town BH 4,23 257,79 3,34 0,13 16,1 22,81 55,33 6 1,1 23,26 1,586 47,2 13 Geita Town Spring 0,28 16,84 2,59 14,74 0,28 3 6,51 1 0,92 3 2,904 11 14 Geita Town Surf 3,04 185,2 23,09 0,15 619,21 15,72 203,91 24,79 8,28 103,32 4,352 9,44 15 Geita Town Spring 0,61 37,09 8,5 19,81 0,79 3 12,33 2,8 0,6 7,02 2,531 13,39 16 Geita Town BH 5,66 345,26 11,38 0,04 2,08 31,38 36,06 15,52 2,09 49,19 1,973 36,84 17 Geita Rural BH 3,54 216,06 104,92 73,35 11,68 176,94 29,58 30,74 2,32 63,49 1,843 34,47 18 Geita Rural Spring 0,61 37,09 11,23 14,05 0,39 9,3 12,22 0,55 0,86 3,54 2,225 19,39 19 Geita Rural BH 5,29 322,45 35,42 3,61 2,24 12,96 28,91 16,52 0,97 55,84 2,649 36,44 20 Geita Rural BH 1,41 86,13 12,78 0,3 0,97 183,58 11,46 6,54 0,84 10,84 1,861 33,77 21 Geita Rural Surf 1,23 75,15 21,42 0,05 0,98 624,21 24,96 6,02 2,94 30,13 5,725 26,87 22 Geita Rural BH 0,95 58,19 12,83 4,2 0,27 26,1 17,07 1,1 2,45 5,73 1,566 38,39 23 Geita Rural SW 5,42 330,62 9,29 0,52 5,29 342,21 17,24 13,13 0,45 76,72 1,641 30,28 24 Geita Rural BH 3,43 209,11 6,27 1,12 0,09 38,93 10,94 16,1 0,8 25,49 1,467 42,16 25 Geita Rural SW 0,44 26,84 4,48 5,52 0,11 5,43 6,51 1,98 0,24 3,5 1,511 12,58 26 Butiama/Musoma SW 0,82 50,26 62,39 139,99 15,49 81,18 28,89 15,18 6,41 38,11 1,956 38,47 27 Butiama/Musoma Surf 2,47 150,55 7,92 5,07 47,12 97,18 39,5 8,15 7,03 23,93 5,885 16,04

APPENDIX E .(cont) Table listing location, type of water source, major ions, DOC and Si content for the sampled waters (from ID 28 to ID 54)

Water ID Location Alkalinity HCO - Cl- NO - SO 2- PO 3- Na+ Mg2+ K+ Ca2+ DOC Si Source 3 3 4 4

mmol mg/L mg/L mg/L mg/L µg/L mg/L mg/L mg/L mg/L mg/L mg/L 28 Butiama/Musoma BH 10,8 659,04 12,84 23,19 35 6,83 53,24 38,03 1,31 118,41 1,14 16,95 29 Butiama/Musoma SW 6,36 388,08 6,9 10,21 5,65 11,93 25,86 25,16 3,7 55,51 1,749 22,18 30 Butiama/Musoma SW 7,5 457,38 5,33 5,6 45,29 9,81 47,7 44,78 1,9 46,46 1,507 14,56 31 Butiama/Musoma SW 0,39 23,91 18,46 5,42 1,2 14,51 8,83 2,65 2,44 6,39 1,87 24,58 32 Butiama/Musoma SW 8,07 492,03 29,47 74,39 24,27 28,9 41,61 38,48 5,37 88,21 1,978 41,72 33 Butiama/Musoma Surf 3,39 206,67 6,16 1,29 6,9 28,03 30,56 6,73 4,35 27,02 6,214 3,82 34 Butiama/Musoma Surf 3,04 185,32 3,26 0,12 193,22 4,09 22,26 32,95 4,05 51,44 3,369 3,85 35 Butiama/Musoma SW 1,17 71,61 91,96 43,68 121,21 14,29 40,92 18,39 7,87 56,21 2,506 26,4 36 Butiama/Musoma BH 8,72 532,04 14,55 3,3 4,77 20,8 50,15 27,88 0,99 72,16 3,08 23,84 37 Butiama/Musoma Surf 10,01 610,85 13,59 1,05 63,64 8,6 111,21 40,31 6,65 54,61 7,547 11,01 38 Butiama/Musoma Surf 1,37 83,69 3,03 4,17 1,02 58,44 20,66 2,71 6,92 8,53 11,44 18,35 39 Butiama/Musoma BH 10,04 612,56 45,02 8,13 23,45 3 57,14 37,37 13,15 80 1,41 34,82 40 North Mara Surf 2,09 127,61 4,99 26,27 110,05 318,71 49,02 8,05 8,63 31,96 3,99 18,11 41 North Mara Surf 3,84 234,48 30,36 5,79 9,91 393,7 84,1 5,53 10,82 17,51 15,1 40,06 42 North Mara Surf 1 61 7,46 4,63 5,75 73,66 12,73 1,12 5,34 6,42 3,871 15,52 43 North Mara BH 7,26 442,62 77,21 67,16 34,25 7,99 102,1 23,76 3,26 73,08 2,837 32,76 44 North Mara Spring 0,92 55,88 21,71 69,58 10,77 46,79 19,47 5,88 10,05 20,09 2,422 33,67 45 North Mara SW 1,08 65,64 36,67 81,39 35,59 137,35 36,9 6,57 10,49 28,05 3,48 28,83 46 North Mara BH 7,26 442,86 14,6 8,18 10,22 9,17 62,27 22,93 2,36 51,67 3,21 26,71 47 North Mara Surf n.t. n.t. n.t. n.t. n.t. 23,34 51,93 9,12 9,39 35,61 3,33 20,56 48 North Mara BH 8,08 492,64 86,82 104,69 32,17 4,99 73,01 33,58 6,7 111,26 1,811 23,47 49 North Mara BH 8,53 520,45 3,3 5,81 5,63 3 33,48 26,8 5,92 79,47 1,918 30,42 50 North Mara Spring 6 366 44,98 56,88 75,73 14,53 57,11 22,37 0,64 85 2,664 28,73 51 North Mara Surf 2,55 155,67 9,05 15,66 230,65 3 89,67 9,63 3,54 54,4 8,168 11,83 52 North Mara Surf 0,77 47,09 8,1 4,17 6,74 96,46 20,31 1,15 7,95 7,12 2,947 12,13 53 North Mara BH 10,02 610,98 33,84 34,49 31,68 3 108,49 23,88 2,22 86,3 1,496 19,62 54 North Mara Surf 3,81 232,53 52,92 0,59 0,1 3 494,86 62,27 79,4 429,57 8,684 11,66

APPENDIX F. Table listing trace elements’ content for the sampled waters (from ID 1 to ID 27).

ID F- Li B Co Sr Zr Mo Cd Ba Pb U Aslab Al Ti V Cr Mn Ni Cu Zn Fe AsIII

mg/L µg/L µg/L µg/L µg/L µg/L µg/L µg/L µg/L µg/L µg/L µg/L µg/L µg/L µg/L µg/L µg/L µg/L µg/L µg/L µg/L µg/L 1 0,2 1,2 5,6 0,1 55,7 0,1 0,1 0,0 75,1 0,3 0,0 0,4 158,6 1,5 0,6 0,4 2,3 0,7 1,2 8,4 66,5 n.t. 2 0,4 1,4 4,4 0,0 131,1 0,1 0,3 0,0 146,2 0,2 0,0 0,1 23,1 0,2 0,2 0,1 1,5 0,5 0,8 14,8 10,8 0,3 3 0,6 0,4 14,1 0,2 196,7 0,0 0,4 0,0 81,8 0,2 0,0 0,9 23,1 0,4 0,3 0,1 156,6 0,6 0,9 6,3 205,4 1,3 4 0,3 3,0 6,0 0,6 132,9 0,0 0,2 0,0 140,9 11,9 0,3 0,3 24,4 0,4 0,7 0,1 19,9 2,2 1,5 10,5 23,4 0,3 5 0,2 1,4 193,6 0,3 195,8 0,1 0,0 0,0 321,2 0,1 0,0 0,1 39,4 0,5 0,7 0,2 16,8 1,4 5,8 16,1 43,3 0,2 6 0,2 1,2 4,2 0,2 84,4 0,0 0,0 0,0 81,6 0,0 0,0 0,0 13,8 0,2 0,1 0,1 1,1 1,1 1,1 5,6 2,6 0,1 7 0,6 12,6 12,1 0,2 276,9 0,0 0,4 0,0 253,1 0,2 0,9 1,1 11,6 0,5 6,8 0,0 39,5 0,6 2,5 21,8 11,8 n.t. 8 0,4 0,7 8,1 0,6 198,8 0,0 0,4 0,0 245,0 0,1 0,1 1,3 17,6 0,4 5,3 0,1 88,5 0,8 2,2 56,0 2075,1 1,2 9 0,1 1,0 1,9 0,6 22,1 0,0 0,0 0,0 40,6 0,1 0,0 3,3 50,6 0,9 0,1 0,1 8,1 1,1 1,0 6,3 314,5 3,3 10 0,2 1,4 6,9 0,3 130,8 0,0 0,1 0,0 120,9 0,4 0,1 0,1 11,3 0,1 0,2 0,1 7,0 1,1 2,0 8,4 7,9 n.t. 11 0,2 2,1 5,6 0,4 152,9 0,0 0,1 0,0 127,1 1,2 0,0 0,1 20,5 0,2 0,4 0,1 2,0 1,6 9,5 90,1 9,9 n.t. 12 0,8 12,4 10,9 0,2 384,7 0,1 1,4 0,0 164,3 4,9 1,3 3,1 26,2 0,6 7,4 0,1 37,0 0,4 5,2 11,5 15,2 0,6 13 0,1 1,9 4,5 0,5 46,4 0,0 0,0 0,0 69,8 0,1 0,0 0,1 37,1 0,3 0,1 0,1 12,5 1,2 1,3 5,9 17,6 n.t. 14 0,1 1,6 10,5 33,8 2064,8 0,0 4,0 0,0 83,0 0,0 0,1 0,3 14,3 7,1 0,1 0,1 237,6 0,3 1,1 14,5 231,1 0,4 15 0,1 1,7 5,4 0,9 89,4 0,0 0,2 0,0 108,5 2,1 0,1 0,1 31,2 0,3 0,2 0,1 24,4 1,6 0,8 31,8 75,7 0,3 16 0,4 11,3 8,7 0,1 627,2 0,0 0,1 0,0 478,1 0,2 2,5 0,1 26,9 0,7 2,6 0,1 3,1 0,3 4,9 33,1 48,0 n.t. 17 0,1 14,8 4,6 0,8 471,0 0,0 8,0 0,1 235,2 0,2 0,4 1,9 99,4 2,5 4,5 0,3 114,7 7,2 5,9 13,8 71,8 n.t. 18 0,1 2,0 7,3 2,6 40,8 0,1 0,4 0,0 72,9 0,3 0,1 0,2 167,7 1,2 1,4 0,3 39,8 2,1 1,6 9,5 109,9 0,2 19 0,1 21,4 5,2 0,6 178,0 0,0 2,8 0,2 26,1 0,1 0,5 1,3 17,0 0,4 5,5 0,1 327,0 0,9 7,1 47,8 722,9 0,6 20 0,2 14,6 4,7 0,2 70,8 0,0 0,3 0,0 89,3 0,8 0,1 0,7 48,5 1,1 8,2 0,5 8,5 5,1 8,8 38,6 55,5 n.t. 21* 0,3 1,5 138,3 25,4 361,6 0,2 8,7 23,1 2472,5 43,6 0,7 1,0 1252,5 19,7 7,2 27,1 1303,2 525,9 268,2 3742,1 1992,5 0,9 22 0,2 6,3 4,2 0,4 103,4 0,0 0,2 0,0 104,9 0,2 0,6 0,1 12,9 0,2 0,7 0,1 9,8 0,9 3,3 8,2 10,9 0,1 23* 0,2 15,4 17,2 6,2 184,4 0,0 0,0 0,1 18,1 17,9 0,1 36,7 2047,6 6,3 10,1 7,0 761,5 6,9 13,0 13,7 6312,8 n.t. 24 0,1 3,4 8,1 0,2 93,6 0,0 0,2 0,0 20,6 0,1 0,0 0,3 13,4 1,9 14,8 1,8 113,4 0,8 1,3 5,8 13,3 0,1 25 0,1 1,0 4,9 0,9 31,8 0,0 0,2 0,0 47,4 0,1 0,0 0,1 36,4 1,4 0,6 0,4 36,1 2,6 0,8 10,7 144,6 0,3 26 0,3 1,0 15,5 0,2 351,9 0,0 0,2 0,0 88,9 0,1 0,0 0,3 41,6 1,8 4,9 0,2 29,0 0,8 1,9 13,0 41,3 0,2 27* 1,0 2,1 21,5 3,4 231,3 1,0 0,4 0,1 108,3 4,0 1,7 1,4 3845,7 7,0 17,8 5,5 225,2 8,5 10,8 23,7 3445,9 1,3 * Unfiltered sample

APPENDIX F. (cont) Table listing trace elements’ content for the sampled waters (from ID 27 to ID 54).

ID F- Li B Co Sr Zr Mo Cd Ba Pb U Aslab Al Ti V Cr Mn Ni Cu Zn Fe AsIII

mg/L µg/L µg/L µg/L µg/L µg/L µg/L µg/L µg/L µg/L µg/L µg/L µg/L µg/L µg/L µg/L µg/L µg/L µg/L µg/L µg/L µg/L

28 0,2 3,3 15,9 8,9 277,2 0,1 0,2 0,0 20,2 0,4 0,6 0,6 17,9 0,9 3,2 0,2 2,5 0,5 5,7 10,6 21,3 n.t. 29 0,6 0,7 10,6 0,2 350,7 0,1 1,0 0,0 34,2 0,1 0,9 0,8 17,3 0,6 3,4 0,1 43,1 0,8 1,3 12,4 9,9 0,4 30 0,1 1,9 19,8 1,7 202,4 0,0 0,2 0,0 15,4 0,0 0,4 56,4 30,8 1,0 3,8 0,2 4,6 9,0 1,3 2,8 17,0 3,2 31 0,3 1,7 10,3 0,3 75,6 1,2 0,1 0,0 26,6 0,3 0,1 0,5 1193,8 11,1 2,7 1,1 17,4 3,8 11,5 19,2 644,2 0,3 32 1,1 0,9 50,5 0,2 494,5 0,1 0,8 0,0 26,8 0,1 1,7 0,6 14,5 1,8 24,3 0,6 29,7 3,9 4,5 13,5 4,6 0,2 33 1,1 0,3 25,7 0,2 252,5 0,1 2,3 0,0 66,7 0,1 0,7 0,5 30,8 0,5 1,4 0,1 28,5 0,9 2,0 10,3 12,8 0,4 34 0,6 0,8 16,2 0,2 260,5 0,1 1,4 0,0 17,6 0,2 0,3 15,4 19,4 2,7 4,1 0,1 7,6 1,8 1,5 13,1 15,8 2,3 35 0,1 1,3 18,6 2,0 593,2 0,1 0,3 0,0 28,7 0,2 0,0 0,3 28,1 1,8 2,1 0,1 135,7 4,7 2,0 8,7 20,8 0,2 36 0,1 3,8 28,2 0,1 292,8 0,0 0,4 0,0 9,9 0,4 1,1 0,1 10,3 0,7 9,7 0,0 1,2 0,5 2,2 69,8 18,9 0,2 37 1,5 0,9 43,0 1,3 492,2 0,3 5,0 0,0 108,0 0,2 3,8 1,2 18,1 2,1 16,8 0,1 40,7 2,7 4,1 3,2 62,6 0,7 38* 0,8 2,5 16,8 2,5 116,1 2,2 0,4 0,0 98,9 6,3 0,8 1,0 3015,7 5,3 9,7 1,9 264,1 3,1 3,1 9,3 4746,1 0,0 39 1,3 33,4 58,4 0,1 808,8 0,0 8,2 0,0 64,3 0,0 13,1 1,4 14,2 0,9 15,0 0,1 7,2 0,5 0,9 0,5 21,2 0,0 40 1,0 6,5 15,5 0,3 314,3 4,4 8,2 0,0 52,9 0,3 2,4 37,9 668,6 10,0 1,6 0,3 59,0 2,5 1,2 5,3 273,9 3,2 41* 1,6 4,6 31,4 7,4 206,4 4,7 0,3 0,3 229,3 10,5 2,3 2,9 8517,0 10,1 18,7 6,8 1421,7 11,6 15,2 34,9 12511,3 2,6 42 0,5 3,2 9,5 0,1 47,0 4,1 1,1 0,0 43,1 0,5 0,3 0,6 337,1 9,4 1,7 0,2 27,0 0,6 1,4 36,9 326,8 0,1 43 1,9 3,5 14,2 0,4 272,7 0,3 1,2 0,0 17,5 0,1 3,7 12,7 30,8 1,1 4,4 0,1 37,7 0,7 1,8 36,2 161,4 2,0 44 0,6 2,0 13,0 0,2 185,3 1,0 0,3 0,0 56,9 0,3 0,2 3,3 372,5 3,1 1,2 0,3 6,7 1,5 1,5 8,7 180,8 n.t. 45* 0,6 2,7 14,9 0,6 303,4 2,0 0,3 0,0 49,8 0,9 0,3 5,5 916,7 3,0 2,0 0,6 61,3 3,4 2,1 9,1 558,9 0,6 46 0,1 16,5 18,0 0,2 332,0 0,3 4,2 0,0 76,4 0,1 7,7 0,4 21,9 0,6 3,4 0,1 48,5 0,8 1,0 74,4 328,0 0,1 47* NA 5,7 14,7 0,6 339,5 1,7 3,0 0,0 81,9 1,5 2,4 19,3 1602,1 14,0 2,9 0,9 94,8 2,6 2,1 6,8 961,1 2,0 48 1,5 20,9 28,9 0,2 609,7 0,1 2,0 0,0 49,5 0,2 10,8 0,4 18,5 1,0 6,3 0,1 19,0 0,5 2,0 19,9 548,8 0,2 49 0,1 11,9 14,3 0,2 348,6 0,0 1,3 0,0 31,0 0,1 6,1 2,8 10,1 0,4 3,2 0,0 10,2 0,5 1,3 56,2 13,7 0,6 50 0,1 1,2 15,6 0,2 490,0 0,0 1,2 0,0 43,3 0,0 1,9 18,2 10,0 1,4 9,2 0,1 74,3 1,8 1,3 3,7 8,0 n.t. 51 2,4 0,6 11,6 1,0 477,4 0,6 7,8 0,0 57,5 0,1 4,6 18,1 77,0 3,5 5,0 0,2 45,2 1,8 1,8 3,1 36,2 3,7 52 0,5 2,7 8,8 0,2 49,4 2,6 1,2 0,1 39,2 0,5 0,2 0,7 155,2 4,6 1,8 0,3 15,8 1,8 5,4 67,8 145,5 0,4 53 1,9 49,4 33,2 0,1 598,9 0,1 2,8 0,0 36,0 0,1 14,2 0,2 17,5 0,7 1,1 0,1 14,2 0,5 1,1 30,2 50,6 0,2 54 1,5 2,6 30,6 34,1 2550,1 0,8 4,6 0,0 66,9 0,2 1,7 9,5 139,0 25,4 2,7 0,2 174,2 6,6 2,1 4,6 96,8 3,1 *Unfiltered sample

TRITA SEED-EX 2017:25