East African Hydropatriarchies an Analysis of Changing Waterscapes in Smallholder Irrigation Farming

MEDDELANDE FRÅN KULTURGEOGRAFISKA INSTITUTIONEN Nr 150

East African Hydropatriarchies An analysis of changing waterscapes in smallholder irrigation farming

Martina Angela Caretta

©Martina Angela Caretta, Stockholm University 2015 Cover illustrations: © Christina Alm Author portrait: © Eva Dalin ISSN 0585-3508 ISBN 978-91-7649-206-2

Printed in Sweden by Holmbergs, Malmö 2015 Distributor: Publit

A mi amor verdadero

Abstract

This thesis examines the local waterscapes of two smallholder irrigation farming systems in the drylands of East Africa in a context of socio- ecological changes. It focuses on three aspects: institutional arrangements, gender relations and landscape investments. This thesis is based on a reflexive analysis of cross-cultural, cross- language research, particularly focusing on the role of field assistants and interpreters, and on member checking as a method to ensure validity. Flexible irrigation infrastructure in Sibou, Kenya, and Engaruka, Tanza- nia, allow farmers to shift the course of water and to extend or reduce the area cultivated depending on seasonal rainfall patterns. Water conflicts are avoided through a decentralized common property management system. Water rights are continuously renegotiated depending on water supply. Wa- ter is seen as a common good whose management is guided by mutual understanding to prevent conflicts through participation and shared information about water rights. However, participation in water management is a privilege that is endowed mostly to men. Strict patriarchal norms regulate control over water and practically exclude women from irrigation management. The control over water usage for productive means is a manifestation of masculinity. The same gender bias has emerged in recent decades as men have increased their engagement in agriculture by cultivating crops for sale. Women, because of their subordinated position, cannot take advantage of the recent livelihood diversification. Rather, the cultivation of horticultural products for sale has increased the workload for women who already farm most food crops for family consumption. In addition, they now have to weed and harvest the commercial crops that their husbands sell for profit. This agricultural gender divide is mirrored in men´s and women´s response to increased climate variability. Women intercrop as a risk adverting strategy, while men sow more rounds of crops for sale when the rain allows for it. Additionally, while dis- cursively underestimated by men, women´s assistance is materially fundamental to maintain the irrigation infrastructure and to ensure soil fertility that makes the cultivation of crops for sale possible. In sum, this thesis highlights the adaption potentials of contemporary smallholder irrigation systems through local common property regimes that,

while not inclusive towards women, avoid conflicts generated by shifting water supply and increased climate variability. To be able to assess the success and viability of irrigation systems, research must be carried out at a local level. By studying how local water management works, how conflicts are adverted through common property regimes and how these systems adapt to socio-ecological changes, this thesis provides insights that are important both for the planning of current irrigation schemes and the rehabilitation or the extension of older systems. By investigating the factors behind the consistent marginalization of women from water management and their subordinated role in agricultural production, this study also cautions against the reproduction of these discriminatory norms in the planning of irrigation projects.

Keywords: smallholder irrigation farming, local gender contract, landesque capital, common property regimes, drylands, feminist epistemology, member checking, Kenya, Tanzania.

Acknowledgments

Acquiring this PhD title has not only been about research, but it has rather been a five years long journey. This journey, punctuated by fieldwork and conferences, has included: acquiring (hopefully) the art of scientific publication, learning two new languages, getting dual citizenship, settling down in Stockholm, coming to terms with the health consequences of fieldwork, los- ing weight, making new friends and finding the love of my life. An intense and emotional rollercoaster along which I was supported by several people! This study has been possible thanks to the financial support of SIDA – Swedish International Cooperation Agency. Additional funding was received from Kungl. Vitterhetsakademien, Lillemor och Hans W:son Ahlmanns geographical fund, Axel Lagrelius geographical fund, Carl Mannerfelt geographical fund, Albert & Maria Bergström Foundation, Knut & Alice Wal- lenberg Foundation, Josephson HS & Emmy fund, Rhodin fund, the Nordic Africa Institute and the Swedish Society for Anthropology and Geography. “Thanks” is not enough to convey my gratitude to my supervisor Lowe Börjeson for choosing me to undertake this research project, for his continuous support throughout this journey – leaving me free to decide the path that this study took, while also warning me not to burn myself out – and for all our honest conversations. I am the academic that I am today especially thanks to you! Thanks also to Gunnel Forsberg for guiding me conceptually and bureaucratically at the beginning of the PhD. To Lars-Ove Westberg, who I have always considered my supervisor, thanks for sharing with me so many trips to Kenya and Tanzania: I have learnt so much from you, Bwana Udongo! Thanks for the mentorship I got from several feminist scholars and geographers: Yvonne Riaño, Danielle Drozdzewski, Maria Dolors García Ramón, Tovi Fenster, Elena Vacchelli, Marcella Schmidt, Paola Minoia, Ragnhild Lund, Maria Prats, Mireia Baylina, Ann Oberhauser, Pamela Moss and Janice Monk. Meeting you all repeatedly in international conferences has been a source of inspiration, encouragement and has led to professional and personal ties that I am looking forward to maintain in the future! The field study would have literally not have been possible without the gracious and patient collaborations of more than hundreds of participants who kindly took the time to repeatedly answer my questions. I was touched being called Bibi Maswali and Chepkemoi and by the affection you showed me in my last trip. Asante Sana! Ashé! Kongoi! I am indebted to Florence

Jemutai Cheptum, Helena Chepto, Timothy Kipkeu, Luka Baileng'o, Cathe- rine Mollel and Israel Mollel who opened their homes to me, became my friends, assistants, translators, gatekeepers, body guards, saw me crying, made me laugh, shared their knowledge with me and with whom I have made indelible memories. The last night we spent all together in Engaruka will always be in my heart! Asante sana rafiki! Asante sana madereva na kaka Joseph Mutua and Francis Mutua! It was through our never-ending eye opening and funny conversations on those long bumpy rides that I got to know East Africa. Thanks for keeping me company, making me enjoy my time and ensuring my safety; even if it meant locking me up in the car! A special thanks goes also to Pastor Edwin Suter whose intervention was truly providential that evening in July 2013. Writing up this thesis has been equally enjoyable and painful. In the homestretch Anders Wästfelt´s suggestions were essential in improving the kappa: thanks! Special thanks go to my colleagues and friends Natasha Web- ster and Thomas Wimark who were so kind to read through some of my publications and to give me suggestions on the method section of the kappa. I am glad I could share this journey with the two of you. Thomas: “you will always be my best friend because you know too much!” and Natasha: our sisterhood is something I truly rejoice! I would like to thank also my “circle of academic kindness” for sharing our hopes, successes and grievances: Bri- an Kuns, Ida Andersson, Johanna Jokinen, Juliana Porsani and Karen Haandrikman. Thanks to Pontus Hennerdal for creating the most productive working environment and for making up, together with Stefan Ene, for my lacks as a geographer, that is to say: maps! Ann-Charlotte Wistedt thanks for making me feel welcomed at the department back in 2011, accepting my weird matatu receipts from fieldwork and assisting me practically in becoming a Swede in the workplace. Thanks to Johan Cederström for helping me with the many IT related issues that I faced during my research and most of all, for making me realize that I could not keep on working with a laptop whose fan was irremediably damage by the sand of Engaruka! I am also indebted to my family contributing to the coming to life of this book: Alejandro Thomas Guillén for your “sound” knowledge, Christina Alm and Riccardo Maculan for the book cover. Al mio migliore amico Matteo: grazie per avermi scovato ad Engaruka, per avermi sostenuto e fatto ridere anche nei momenti di crisi e incazzatura piú profonda. Mi accompagni ormai da 10 anni in questo viaggio svedese e non solo: é significativo aver condiviso anche l´attacco di un bufalo nella savana africana e aver scampato un ammaraggio a Los Roques! Un sentito grazie alle mie amiche lontane – Giada, Elisabetta e Deborah – e vicine – Marialuisa, Eleonora e Carin – per l´ascolto, gli incoraggiamenti e i consigli. A mi compañero de doctorado y de estudio del paisaje Carlos Cornejo Nieto muchas gracias por nuestras estimulantes conversaciones, risas y por los viajes que hicimos y que haremos en el futuro!

Mamma e papá dopo la pazzia del master in Kenya avete vissuto altri cinque anni di preoccupazioni – Giappone, ripetuti viaggi in Kenya e Tanzania e i problemi di salute che mi hanno azzerato. Nonostante tutto, mi avete appoggiato da lontano e anche da vicino – venendo spesso a trovarmi – anche in Kenya e Tanzania! – e facendomi coraggio anche quando ero a terra. Ah, papá: grazie “par verme portá in Africa!” (cit.) la tua compagnia é stata una distrazione divertente e necessaria per farmi tirare il fiato! Mamma, grazie perché mi hai spesso ricordato che dovevo tirare il fiato, che le cose importanti solo altre e per aver accettato le mie scelte, anche se questo implica essere lontane. Silvia, l´atteggiamento da “ortiga” con cui spesso mi prendevi in giro “ti e i to paper!” (cit.) non solo mi ha aiutato a ritrovare un po´ la prospettiva delle cose importanti quando la perdevo, ma mi é anche servita di spinta! En fin, a mi amor verdadero, Luis Andrés Guillén. Fue la casualidad mejor que me pudo pasar que nos conocimos exactamente la semana después que empecé el doctorado!! Me has acompañado en todo este recorrido con mucho amor y comprensión, enseñándome a “bajarle dos” – todavía me resulta un poquito difícil pero creo que estoy mejorando! – y a tomarme el fin de semana libre. Gracias también porque me introdujiste a la salsa y me has apoyado en todos mis entrenamientos locos y en la corsa. La salsa y el deporte son las dos cosas que me han mantenido cuerda en estos años! Todos nuestros viajes, sobre todo el a África y el a Venezuela, han sido ocasiones estupendas para conocernos más y para convencerme que quiero seguir juntos contigo en el viaje de la vida. Te quiero mucho mucho y este PhD es también tuyo: eres el amor de mi vida!

Stockholm, 4th September 2015 Martina Angela Caretta

Contents

Introduction ...... 17 Aim and Research Questions ...... 18

Conceptual framework: Waterscape ...... 21 Common property regimes ...... 22 Local gender contract ...... 23 Landesque capital ...... 25 Canals, furrows, or channels ...... 25

Methodology ...... 27 Feminist epistemology ...... 27 Case study design ...... 28 Case selection ...... 29 Sibou, Kenya ...... 31 Engaruka, Tanzania ...... 34 Methods ...... 37 Focus groups ...... 37 Interviews ...... 41 Plot survey ...... 43 Participatory mapping ...... 43 Limitations ...... 45 Positionality ...... 47 Data analysis ...... 49 Triangulation ...... 50 Reporting of results ...... 50

Summary of studies ...... 53 Article I: Managing variability and scarcity. An analysis of Engaruka: a Maasai smallholder irrigation farming community...... 53 Article II: Local gender contract and adaptive capacity in smallholder irrigation: a case study from the Kenyan drylands...... 54 Article III: Hydropatriarchies and landesque capital: a local gender contract analysis of two smallholder irrigation systems in East Africa...... 55 Article IV: Member checking: a participatory method to test and analyze preliminary results in cross-cultural, cross-language research...... 56

Article V: Situated knowledge in cross-cultural, cross-language research: a collaborative reflexive analysis of researcher, assistant and participant subjectivities...... 57

Concluding discussion...... 59 Common property regimes: scale and external factors ...... 59 Gender and irrigation ...... 61 Landscape investments ...... 65 Conclusion ...... 66

Future studies ...... 69

Svenska sammanfattning ...... 71

References ...... 73

List of articles

Article I Caretta, M.A. (2015). Managing variability and scarcity. An analysis of Engaruka: a Maasai smallholder irrigation farming community. Agricultural Water Management, 159, 318-330. DOI: doi:10.1016/j.agwat.2015.05.015

Article II Caretta, M.A. & Börjeson, L. (2015). Local gender contract and adaptive capacity in smallholder irrigation: a case study from the Kenyan drylands. Gender, Place and Culture 22(5), 644–661. DOI: 10.1080/0966369X.2014.885888

Article III Caretta, M.A. (2014). Hydropatriarchies and landesque capital: a local gender contract analysis of two smallholder irrigation systems in East Africa. The Geographical Journal. DOI: 10.1111/geoj.12102

Article IV Caretta, M.A. (In press). Member checking: a participatory method to test and analyze preliminary results in cross-cultural, cross-language research. Qualitative Research. DOI: 10.1177/1468794115606495

Article V Caretta, M.A. (2015). Situated knowledge in cross-cultural, cross-language research: a collaborative reflexive analysis of researcher, assistant and participant subjectivities. Qualitative Research, 4, 489-505. DOI: 10.1177/1468794114543404

Introduction

Irrigation makes irrigation possible in dryland environments and is often presented as a means to intensify agriculture and to enhance its productivity. For this reason, irrigation is seen as one of the solutions to the current challenges, most importantly climate change and population growth, faced by African agriculture (You et al., 2010). Agriculture in Sub Saharan Africa is predominantly a smallholder and family enterprise (Wiggings, 2009). While often considered backward, smallholder farming does not, however, automatically mean low yields and subsistence farming. For that matter, intensive farming is already present in the “islands of intensive agriculture” of Eastern Africa. In these sites, irrigation systems and appropriate land management practices have been long put in place by local farmers’ groups to secure food provision for an increasing population (Widgren and Sutton, 2004). While often regarded as technical systems, irrigation systems in this thesis are conceived as inherently socio-ecological organizations (see also Plusquellec, 2009) around which local societal structures are shaped, negotiated and challenged throughout time and space. Gender is one of those societal structures. Women are in fact the backbone of African agriculture. They produce 70% of the staple food and also provide 60-80% of the labor for cash crops (Negin et al., 2009). Yet, because their work is often considered to be reproductive, it is not accounted for in national economic statistics. Hence, the gender division of labor in subsistence agriculture has not radi- cally changed since Boserup published her analysis of women´s contribution to agricultural production in Africa more than 40 years ago (1970). Women continue to lack land tenure, access to credit and extension service. This condition confines them to the production of staple crops, hinders their investments in agriculture and the attainment of high yields (Ransom and Bain, 2011). In irrigation systems land titles relate directly to water control, which is traditionally associated with masculinity. Water is in fact encoded with pow- erful themes of ownership and control when it is used for productive means. These meanings are employed to exclude and dispossess several groups, including women (Strang, 2005) who cannot participate in decision making water forums or use water for productive purposes without men´s consent. The subordination of women, inherent in patriarchal societies, thus becomes apparent in irrigated agriculture (Zwarteveen, 2008). It would, however, be

17 wrong to assume that gender structures, often considered the result of traditional norms, are unchanging. The same can be said for ecological conditions. In this sense climate change triggers increased weather variability which results in a shifting volume, intensity and distribution of rainfall (IPCC, 2014). Given the dependence of African agriculture on rainfall to produce crops, there will be an increased need for irrigation. Moreover, this development will be exacerbated by other factors as population growth and increasing market demand (Rosengrant et al., 2009). What remains to be seen is how local food production systems will adapt. In this context, the study of smallholder farming systems where irrigation has a historically long tradition is relevant because lessons can be drawn on how existing social norms can facilitate or hinder inclusive water management and adaptation to ongoing socio-ecological changes (Stump, 2010). In this thesis water is therefore not only conceptualized as a material resource, but also as a force that shapes the landscape, social relations, practices and institutional arrangements that are determined by local socio- ecological processes over time. Thus, the term “waterscape” (Budds and Hinojosa 2012; Tagseth, 2010), which encompasses the study of water management arrangements, gender relations and landscape investments, lies at the core of this thesis. The relevance of this thesis is to be found in the current debate on African agriculture. Often described as traditional and low in productivity, it is argued that yield gaps should be closed by investing in the creation of new and the expansion of existing irrigation systems (e.g. AGRA, 2014 and 2013). These approaches often do not take into account the inherent value of long- term existing local agricultural practices and tend not to assess existing social relations in order to understand how the new infrastructure can be equally beneficial to the overall population (Moseley et al., 2015; Haug and Hella, 2013; Shah et al., 2013). This thesis sets out to analyze current agricultural practices and gender relations in two communities where irrigation has been undergoing for hundreds of years. In the context of this debate, this study highlights how research at the local level can help to integrate local agricultural practices and gender sensitive measures in the planning of irrigation projects.

Aim and Research Questions Irrigation is functional in intensifying agricultural production and is hence considered a crucial element to make African agriculture more productive, especially in the current context of population growth, increasing market demand and changing climatic patterns. Smallholder irrigation is often managed by local institutions that tend to exclude women. Drawing on one case

18 study with two embedded units of analysis, the aim of this thesis is to exam- ine local waterscapes of smallholder irrigation farming systems in East Afri- ca focusing on three aspects: institutional arrangements, gender relations and landscape investments in a context of socio-ecological changes.

Three theoretical and empirical research articles study:

1. How smallholder irrigation farming common-pool regimes deal with the limited and variable availability of water in dryland environment and mediate water conflict among users (article I);

2. How these common-pool regimes are shaped by the local patriarchal relations and how in turn, men and women put in place different agricultural practices to adapt to climate variability (article II);

3. How local societal norms have shaped the local gender division of labor of these common-pool regimes and how these labor arrangements have been manifested over time in agricultural landscape investments (article III).

The last two research articles deal with the methodological aspects of this study:

4. Member checking – a method used all along this research to countercheck preliminary findings – is discussed in relation to its potential to ensure validity beyond triangulation while acknowledging the need to constantly reflexively and critically assess this method (article IV);

5. The role played by the research assistants in shaping the research process is reflexively and critically investigated to bring to light how the respective positionalities of research assistants and researcher reciprocally have conditioned knowledge construction (article V).

19 20 Conceptual framework: Waterscape

Water in this thesis is not only a material resource that makes agriculture possible in the East African drylands, but is primarily conceived as a consti- tuting element of social relations. In fact, its materiality and its capacity to flow make it an element to be controlled by some over others; at the same time, its scarcity and its shifting amount forces social arrangements to adapt and to enter into negotiations on its usage. The term that encompasses this hybrid nature/society – a material and social element – is waterscape (Swyngedouw, 1999). Water and social relations are in fact co-produced reciprocally (Budds and Hinojosa, 2012; Loftus, 2009). Social change, according to Swyngedouw in his 1999 work on “the production of the Spanish waterscape”, cannot be grasped without looking at the hydrological process and changes in the hydrology cannot be understood without looking at societal processes. Hence, water and society are mutually constitutive (Swyngedouw, 1999). Different authors have employed this emerging concept according to the different categories of socio- ecological relations that they wanted to analyze, e.g. physical, landscape, institutional (Tagseth, 2010); cross scalar governance (Budds and Hinojosa, 2012); politics and water engineering (Swyngedouw, 1999); and power relations (Loftus, 2009). Knowledge and practice around water determine power relations. For instance, according to Loftus (2009), accumulation of water in the form of control through privatization or waterworks shapes the local social order to the advantage of minorities. In this thesis I argue that the material characteristics of water – flow, forms, volume – define the practices, agricultural routines, landscape investments and institutional arrangements, and the power discourses on who can use water and/or manage it for productive purposes. This study examines the local relations between water and society in smallholder irrigation farming and how they are manifested specifically in three spheres: institutional arrangements, gender relations and landscape investments. While a waterscape can be conceptualized as an even more encompassing notion, these are the three specific elements of waterscape analysis that are at the core of this thesis and that are applied at the local village level over a 30 years period. The local focus was chosen to be able to devote full contextual attention to how power relations are enacted and materialized in the practices and discourses surrounding irrigation. The 30 years focus was relevant for two reasons: first, it was a time frame that participants could relate to in

21 terms of contemporary societal changes and weather variability; second, earlier studies did not cover this time period, but rather studied local phe- nomena longer back in time (e.g. Stump, 2006; Watson et al., 1998; Moore, 1986).

Common property regimes Ostrom (1990) has defined a common property regime as an organization with a limited number of members which through the collective action man- ages an open access resource – be it water, land, forest etc. As a water users´ association fits also under this definition, I applied this concept in my analytical work. Water is a common good that could be depleted by free-riding, but if an institutional arrangement is in place to prevent it from happening through collective action, this risk is adverted. The key to such a robust institution depends, according to Ostrom (1990), on local cultural norms. Moreover, problems with these arrangements, emerging when there is lack of commit- ment and free-riding by some users, can be resolved through constant internal monitoring among members (Ostrom and Ostrom, 1977). The rules that ensure the success – that is to say durability of the institution and sustainability of the resource – of a water users´ association are: 1) boundary rules defining who can benefit from the resource; 2) allocation rules defining the quantity, timing and length of water withdrawal; 3) input rules defining the work that users must contribute with in order to have access to water; 4) penalty rules defining gradual sanctions depending on the rules being broken (Tang, 1992). However, Ostrom (1990) asserts that operational rules and success of the common property regime are highly dependent on the local conditions. For instance, she argues that when the users know each other well and agree to a certain set of rules defining proper behavior in relation to the common to be preserved, sustainability and robustness of the institution are ensured (Ostrom, 1990). It has been debated which characteristics a population should have to put in place a long lasting common property arrangement (for an outline see Cox et al., 2010; Agrawal, 2001). Population homogeneity has been singled out as a crucial element as differences in landholdings, religion, ethnical belonging and wealth can hinder collective action (e.g Bardhan, 2000). On the other hand, heterogeneity may create interdependence among stakeholders who must collaborate to make all have access to the resource at stake (e.g. Komakech et al., 2012). Irrigation systems are grounded in an intrinsic asymmetry: water flows downstream, headenders and tailenders have different endowments. They are nevertheless forced to collaborate in two aspects: appropriation through water allocation for productive purposes

22 and provision through the maintenance of irrigation infrastructures (Ko- makech et al., 2012; Zaag, 2007; Ostrom and Gardner, 1993). If they don’t, they might face the risk that some remove more water, while others hinder water from reaching others´ plots because they do not contribute in the construction of canals and water diversion works. To avoid these incidents from happening a conflict resolution mechanism must be in place, otherwise the arrangement might wreak havoc and the resource might be depleted (Ostrom, 1990). A common property regime often comes into being because of the uncer- tainty surrounding the resource at stake. In the case of water, its supply can be variable and unpredictable due to changing inter/intra-annual rainfall patterns (see also Hillbom, 2012; Ostrom, 1990). In order to reduce the un- certainty surrounding the resource and improving its manageability, communities invest labor in building infrastructure, such as canals and intakes, in order to regulate water flow and distribution (Ostrom, 1990). How labor is defined and who can do what type of labor, and what type of investments that are made by local users, are defined by the next two concepts chosen for the analysis of this thesis.

Local gender contract Gender division of labor can be a useful concept to define who has the upper hand in the local society and particularly in the common property regime. This concept in fact discloses ties of dominance and powerlessness and at- tempts to overturn them (Jackson, 1989). However, the concept of gender division of labor tends to depict a quite static extemporal situation giving most often an unchanging rendering of women´s and men´s tasks in society. Generally speaking it can be stated that men since the beginning of time took care of herds, hunting and working outside the house, and hence became exposed to the public sphere, which they started governing, leaving the private sphere to women. Due to their mother role women remained confined to house work, creating the division between productive and reproductive work (Jackson, 1989; Rosaldo et al., 1974). Because of its rigidity, I did not deem this concept as sufficient to investigate the question of how existing gender norms have influenced and shaped women´s and men´s productive and reproductive tasks and how they are manifested in the agricultural landscape (article II and III). Instead, I needed a concept that better accounted for change over time that could facilitate a place specific analysis and that focused on the underlying causes – i.e. societal norms – that bring about a certain gender division of labor. The importance of highlighting how gender relations are shifting and changing in time and space made me exclude concepts such as “gender regime”, “gender

23 structure” and “doing gender” which tend to give a static picture of gender relations (Risman, 2004). Hence, I employed the concept local gender contract, which I found particularly suitable for my analysis. Gender contract is defined as a process of informal negotiation and reciprocal influence. This is “a pattern of implicit rules on mutual roles and responsibilities, on rights and obligations, and it defines how the social relations between women and men, between the genders and generations, and also between the social production and reproduction are organized in our societies” (Rantalaiho and Heiskanen, 1997, p. 47). These societal power structures, most often unequal, are not static, but rather negotiated according to changes occurring within communities (Rantalaiho and Heiskanen, 1997; Hirdman et al., 1994). Gender contracts are part of these labor regimes, which through daily practices change the landscape. This concept links organically with landesque capital because it has the potential of showing how societal organ- izational changes are continuous and they are invisibly manifested in the dynamic constant process of landscape formation (Mitchell, 2008). Moreover, instead of focusing on the different strata of control that con- cede for women´s subordination, as gender division of labor, gender contract is concerned with illustrating the role of cultural norms in shaping and controlling gender structures at different levels. Additionally, this concept takes into account the small changes that over time lead to the re-negotiation of a gender contract to another one (Hirdman, 1990; see also article II). In geography this concept was introduced by Forsberg (1998) and was defined as local gender contract. Daily practices of production and reproduction are in fact place bound because they depend on the locally distinctive material conditions of the community studied (Forsberg, 2010). Local gender contract is a concept developed in Scandinavia and has only twice been applied in studies outside the Global North (Lindeborg, 2012; Kalabamu, 2005). I chose this concept because through its three layers – cultural overlay (societal norms), social integration (gender division of labor), and socialization (gender division of labor at household level) – I could analyze at which level changes in gender structures were being negotiated and how they influenced the behavior of individuals within their family or gender identities at the societal level. While it is relatively easy to see how the state from the end of the cultural overlay through policy can force partners to renegotiate their roles within the family (e.g. paternal leave in Sweden), it is much more challenging to operationalize the local gender contract negotiation departing from the level of socialization within the household. In practice it would mean to be present when discussions among partners take place in their homes and even then negotiation as such would be hard to prove. Hence, the existence of such negotiation is to be inferred from general community trends or new working arrangements in some families.

24 Landesque capital Gender is the determinant for the local division of labor, which is reflected in the landscape (McDowell and Massey, 1984; Cockburn, 1988). Labor is expressed in the landscape as built morphology where its different functions can be recognized (Widgren, 2004b) as a cultural landscape through the evidence of human modification. A landscape is indeed changing as a manifestation of the evolution of social organization (Mitchell, 2008). In order to analyze the features of an agricultural landscape I employed the concept of landesque capital which refers to “any investment in land with an anticipated life well beyond that of the present crop, or crop cycle” (Blaikie and Brookfield, 1987, p. 9; see also Håkansson and Widgren, 2014). Examples of landesque capital are ditches, irrigation canals and terraces because they change the physical landscape to ease and intensify agricultural production for several seasons to come. This conceptualization has been mostly applied to non-mechanized systems of agricultural production (e.g. Widgren, 2004b; Håkansson and Widgren, 2007; Börjeson, 2004; Blaikie and Brookfield, 1987) because it is “by definition labor-intensive” (Widgren, 2007, p. 68). Investments in land productivity can be systematic and involve coordinated efforts by a group of people, or they can also be the result on an incremental type of labor process, as “cultivation and construction are inex- tricably mixed activities” (Doolittle, 1984, p.135). Landesque capital has been mostly referred to and used in relation to physical systemic investments as irrigation systems or field boundaries (Widgren, 2004b), to its outcomes in terms of soil fertility (i.e. Brookfield, 2001) and to the role played by social organizations (Håkansson and Widgren, 2014; Bayliss-Smith, 2009; Börjeson, 2004; Östberg, 2004; Watson et al., 1998) and kinship (Håkansson and Widgren, 2007) in its formation. It has been less taken into account in terms of incremental and less visible types of investments that result in long term improvement, as is the case of soil fertility (Börjeson, 2014). Even less has the gender dimension been considered (Östberg, 2014; Bayliss-Smith, 2009; Watson et al., 1998).

Canals, furrows, or channels There has been a recent discussion on how the gravitational irrigation systems of the two study sites should be described (Davies et al., 2014; Widgren, 2014). The literature on Sibou, Kenya has consistently described it as a “furrow hill system” (i.e. Davies et al., 2014; Adams et al., 1997). Sut- ton (2000), on the other hand, describes the waterways of Engaruka as canals, while furrows are the trenches dug into the soil to lead the water from the main canal to the field. Moreover, a furrow does not necessarily have to carry water; it can just be a trench dug with a plow (Widgren, 2014).

25 In my dissertational work I decided to use the word canal for two main reasons. First, it is the most commonly used and straightforward engineering term, which makes the topic comprehensible to a bigger audience. Secondly, while the word mfereji is used in Swahili, its origin and etymology are un- clear. On the other hand, the term canal is also used to describe irrigation systems in the Americas (Doolittle, 2001), Europe and Asia (Tang, 1992), so why not in East Africa? Maybe a leftover from the colonial times, not giving the same credit to work of agriculture in Africa as elsewhere? Moreover while the term channel denotes a natural feature, canals are built by people (for further discussion see Widgren, 2014). Davies et al. (2014) use the term channels and also maintain the terminology of furrows because they find that the concept of furrows is helpful in describing the shifting nature of water management and the domestic use of the waterways. I disagree, first, because why should not canals shift and, secondly I could not trace any literature claiming that canals cannot be used for domestic purposes.

26 Methodology

This section presents the epistemological choices, the case study design and selection, and the challenges I faced during the research process. These con- siderations are followed by a detailed description of all methods used during in this study. A discussion of data analysis, triangulation and reporting of results concludes this section.

Feminist epistemology

Feminist epistemology is at the foundations of this thesis. This methodological choice has been based on the fact that there is still a long way to go to do justice to the role of women in producing and reproducing the world. Space, and in the case of this thesis, landscape, are a fundamental component of the world that women are barred from shaping. According to McDowell (1999) feminist geographies are concerned with disclosing the existing connections between gender and space divisions. However, it is fundamental to bear in mind that using the backdrop of feminist geography does not mean solely to investigate the situation of women (Hall, 2002). This thesis presents the relational and structural dynamics of how men and women share spaces and labor practices, while possibly bringing in more of the voices of women, it is based on the analysis of the conditions of both genders. Nevertheless, while the aim is not solely to highlight the position of women, their circumstances undeniably emerge to be characterized by oppression, lack of power and unequal access to resources; thus, the relevance of grounding this study on a feminist epistemology. From a theoretical point of view, this thesis has the potential to scrutinize yet another unjust landscape, including the relations and struggles that created it and the possible changes in these social relations (Mitchell, 2003, 2012). This thesis moreover represents an attempt to overcome the “male gaze” that still defines theorizations of gender and landscape (Rose, 1993). Feminist epistemology also aims at overcoming hierarchical exploitative relationships between researcher and researched (Maynard and Purvis, 1994). While feminist epistemology highlights that power is in the hands of the researcher during the research process, in this thesis I argue that it does not necessarily always needs to be the case. Assistants and translators have

27 in fact a crucial role in deciding the turn and twists that research can take (see article V). Moreover, I came to realize that participants are endowed with the fundamental power to negotiate the character of their involvement and to withhold information. Interviewees might prefer not to disclose certain events to convey an idealized and positive picture of the local conditions to the researcher and they might also decide not to share their knowledge on agricultural practice because they fear being copied by their neighbors who are present during the interviews, as a man once asserted at the end of the interview where a neighbor had been present.

Case study design In this thesis I aimed to study a contemporary phenomenon in a real life context over which I had no control. I had to discern between the context and the phenomenon itself, which were not evident at first (see Yin, 2009). I hence decided to design my research as a case study because it gave me the opportunity to investigate irrigation management and gender relations directly where it was happening using an ethnographic approach, while combining multiple sources of evidence. Before understanding the phenomenon – irrigation management – I had to understand the distinctive context of smallholder irrigation farming in East Africa, such as existing patriarchal norms for example. I chose a single case study design with two embedded units of analysis. In this way I could carry out ethnographic fieldwork in two distinctive, yet complementary and representative locations. This provided me with more data, adding insights and complexity to my research questions while allowing me to aim at analytical generalization given the similarities and the rep- resentativeness of the two study sites. Thanks to the case study design I could expand and further develop the theoretical concepts I had chosen to interpret the data. After the very first exploratory fieldwork I made some theoretical propositions which I verified and/or altered during the following fieldwork sessions. Moreover, the case study design was functional in combining different qualitative methods. Especially multiple repeated focus groups and in-depth interviews gave me the type of data I was looking for to answer the “how” research questions I had in mind (see Yin, 2009). Accordingly, during the course of my research I strived to build personal and trustworthy relationships with informants and participants by going back to Kenya and Tanzania several times, spending a total of seven months in the field. Returning to the field did not only give my assistants, participants and myself something to look forward to, it allowed me to see the changes that had taken place while I was gone and it appeared to reinforce participants´ conviction of my com- mitment to research.

28 Case selection

This research was carried out in two small villages: Sibou, Kenya, and Enga- ruka, Tanzania. The choice of these sites was first guided by the existing contacts with academic and research institutions: Dr. Matthew Davies (Brit- ish Institute in Eastern Africa and University of Cambridge) for Sibou and Dr. Lars-Ove Westerberg (Department of Physical Geography, Stockholm University) for Engaruka. Secondly, both locations are considered to be among a number of places in Eastern Africa with well-functioning indigenous irrigation systems (see Tagseth, 2008). Water availability in this dryland environment made intensive agriculture possible; hence, their definition as “islands of intensive agriculture” (Widgren and Sutton, 2004). Therefore Sibou and Engaruka were representative of a phenomenon that is spread over East Africa, while at the same time they presented differences in terms of environmental (e.g. soils – see Caretta et al., 2014), and national (e.g. economic and political) conditions. Thirdly, studying two locations gave the possibility of examining issues from a wider regional perspective. Logistically it was demanding and complex to carry out research in two countries contemporarily. However, I have been motivated to do so by the fact that while other researchers were also investigating aspects of irrigation in Sibou, in Engaruka, where working conditions were more challenging, I was the only one doing non- archeological research. It was an opportunity to build on previous works done in Sibou and to get novel insights in Engaruka.

Figure 1. Smallholder irrigation farming in Eastern Africa (Tagseth, 2008, p. 62)

Lastly, both irrigation systems have a long history, being dated between the 15th and the 19th century, and have been subject to several archeological studies over the years (e.g. Sutton, 2000; Davies et al., 2014), but their current circumstances have not been investigated in the case of Engaruka and only recently in the case of Sibou (Davies et al., 2014; Adams et al., 1997). Given the uniqueness of the landscape investments visible in these large scale historical systems and their resilience over time, it is relevant to understand how they have adapted to recent changes: i.e. climate variations, population pressure, new market links and introduction of new crops.

Sibou, Kenya

The first location of my study is the village of Sibou, located 35°65' 81″ E, 1°.20' 44″ N in the Tot location of Marakwet District. The village extends for 20.6 km2 (KNBS, 2009) on the slope of the Keryo Valley escarpment, on an altitude between 1000 and 1600 m.a.s.l. The settlement is located on a bowl shaped hill and one of its natural borders on the northern side is the river Embobut. Nowadays the location is more commonly referred to as Tot, which is the settlement along the road on the plain right below the escarpment (Moore, 1986).

Figure 2. The study area around Tot Centre and the different ecological zones. (Caretta et al., 2014, p. 8).

Tot was founded in 1949 by the colonial authority as a tax collection point, after the road to get there was built thanks to forced labor between 1913 and 1936. From the 1950s on the settlement acquired a school, a police station and a health center (Moore, 1986). At the end of the 1970s Critchley (1979, p. 6) reported: “There cannot be few places in Kenya as populous as the Kerio Valley with such poor shops”. Nowadays Tot Centre sees the presence of a few shops, an extension service office, the Keryo Valley Development Agency office and the Anglican Church which is in charge of the former World Vision Center, an American non-governmental organization which run campaign against female genital mutilation in the 1980s and early 1990s. During the course of the research in 2012 Tot Centre was connected to the national electricity grid, but no cables provide electricity to the surrounding

31 villages. A matatu (collective transport carrying about a dozen people) leaves Tot at 5 a.m. to climb the escarpment via Chesoi to reach Kapsowar in the highlands at 10 a.m. From there frequent matatus reach Eldoret town. There is also a daily matatu heading through the valley to Piretwo from where matatus for Kabarnet in the Baringo District are available. There are furthermore now numerous motorcycles ferrying persons between the different centers. However, the road remains a problem, needing constant maintenance, especially during the rain period. Politicians have during election campaigns promised that the roads will be tarmacked, but this is yet to happen. Virtually all adults in Tot have a cell phone, often connected to the money transfer system M-pesa. M-pesa has been revolutionary for shopkeepers in Tot: they can order products from Eldoret and have them shipped on the same day through the matatu service. M-pesa has also enabled farmers to be paid directly on delivery. The last census counted 2,453 people living in 660 households (KNBS, 2009), more than twice the number of households that lived here in the 1980s (Moore, 1986). The inhabitants of Sibou are Marakwet, a Kalenjin speaking people. Villages, like most other Marakwet villages, are made up of several patrilineal clan groups, each consisting of one or several lineages (Östberg, 2004). There are four patrilineal clan groups in Sibou: Kaapisyooy, Syapan, Kaapsireen, Kaachēēpsoom (listed in accordance to their respective sizes). Each of these clans has their own canal taking off from Embobut Riv- er. Every clan´s canal is managed by a clan water users group that meets to build, repair canals and to decide on water turns and rights. This group, with solely male members, is headed by a chairman in charge of summoning meetings and distributing water rights (Östberg, 2004). The villages have no formal leaders or governing committees. Instead decision making is vested with local meetings, kok, summoned as a specific issue needs to be solved, and where all initiated men can participate on equal terms. Women can attend, and will do so if they consider an issue to be discussed is of direct concern to them. They are, however, given a subordinate role at the meetings. Water distribution is organised at special meetings for those who control a particular canal, the pik ab per (people of water) (Östberg, 2004). Age sets form another organising principle in Marakwet. Both men and women belong to an age set, depending on when their circumcision ceremony took place. These groups are eight and rotate. Currently elder men belong to the Kaberur age set, with a few surviving men of the Korongoro generation, and when these all have passed away it is time for the young men to become initiated as Korongoro. The Kaberur are in their late 60s to 70s while the Kaplelach, in their 50s, are a politically influential group gradually being replaced in this position by the following age-set, the Kimnyigei. The initiation into age sets is tied to circumcision. Campaigns have been waged against female genital mutilation, and modernized alterna-

32 tive rites of passage promoted most actively by people associated with the African Inland Church (Moore, 2011, p. 43). Farming in the dry Kerio Valley depends on an extensive system of irrigation canals leading water down the steep escarpment to the cultivated fields on the valley floor. Along a stretch of about 40 kilometers of the escarpment, there are some 90 canals with a total length of about 350 kilometers (Davies et al., 2014). The latest account by Davies et al. (2014) reports that in Tot there are eight canals adding up to a total of 43.8 km. Several accounts describing various aspects of this spectacular pre-colonial system are available (e.g. Adams et al., 1997; Watson et al., 1998; Adams and Wat- son, 2003; Östberg, 2004, 2014). The irrigation systems are also referred to in other studies of the Marakwet (e.g., Moore, 1986; Critchley, 1983). This body of research has treated the system’s physical layout, history, labor organization, water distribution, gender issues, rules, and practices. Several studies have been carried out on the origins of this system, and the latest dating by Davies et al. (2014) indicates that the system is between 200 and 300 years old. The first written account by Thomson in 1885 (1887, p. 310) confirms that the irrigation system was in operation already by the end of 19th century. Sibou, as other villages in Marakwet, have three distinct land use areas, with different topographic and environmental characteristics. The valley floor and the Kerio River, in the east, is termed keu. Here land is communal- ly held and farming collectively organized to manage the major work to lead irrigation water to the land. Farms are cleared and fenced by a community cultivating a particular land, while the cultivated plots inside this area are managed by individual households. Livestock graze in keu. On the slopes of the escarpment towards the west is the residential area, lagam, with home gardens and where also the sites for ceremonies are located. Further up the escarpment, in masop, agriculture is rainfed with maize and beans as the most common crops. Here also good grazing areas are found and timber and firewood are collected in the forests. Living in the lagam was particularly justified from the 1990s to the early 2000s when the Marakwet people were involved in a conflict with the neighboring ethic group, the Pokot. This made people abandon the fields in the keu and retreat up to the masop. This conflict, which lasted more than a decade with a low but steady intensity, was triggered by Pokot cattle raiding, and caused counter raiding from the Marakwet side (Bolling and Österle, 2007). However, in the late 1990s it escalated with acts of ethnic cleansing, among which the killing of 15 people at the Tot health center in 2000, in response to the killings of some Pokot women on their way to the market, and the final attack of 2001 to a school in Marakwet leaving 50 pupils dead (Widgren, 2003). The conflict was then settled in 2002 thanks to the negotiation of the central Kenyan authority and then President Moi. Marakwet people could then come back to inhabit the lagam and cultivate the keu (HRW,

33 2002). Nowadays people are moving down to reside in the keu where land is plentiful and the soil is believed to be more fertile (see Östberg and Caretta, forthcoming).

Engaruka, Tanzania

The second location of my study is Engaruka located 35° 57' 45″ E, 2° 59' 20″ S on the foothill of the Rift Valley escarpment. Administratively Enga- ruka ward is within the Monduli district which is part of Arusha region.

Figure 3. Relative location of Engaruka. (Caretta, 2015a, p. 320)

Between 1972 and 1982 during the Ujamaa – or compulsory resettlement – the current villages of Engaruka Juu and Chini, the latter being circa 2,5 km downstream Engaruka River, were created (Sutton, 2004). Here houses are within compounds consisting normally of: 1) a rectangular house made by a wooden structure and with a tin roof where the household members sleep; 2) a smaller round thatched house which serves as a kitchen and 3) an enclosure for goats. Few houses are made of bricks and have a latrine (cf. Coast,

34 2002). There are several small settlements in the surrounding of Engaruka Juu and Chini – the biggest being Olemelepo to the south and Neng´alah to the south east – comprising one or several bomas. These enclosures fenced with thorns include several manyattas, i.e. round houses made of sticks, cow dung and mud with a thatched roof. The inhabitants of Engaruka are a mix of Kisongo and Arusha (Sutton, 1986). While the Kisongo is a core pastoralist section of the Maasai, the Arusha are a small agricultural community of the Maasai who originated from Mount Meru and Pangani River valley. Exchange between these two sections has been intense and fruitful historically, intermarriage was common and the dialects spoken by the two groups were the most similar among all the Maa dialects. The two groups have been complementing each other socially and economically at least since the end of the 1800s (see McCabe et al., 2010; McCabe, 2003; Spear, 1993). The total population of Engaruka ward in 2012 was 11,121 individuals living in 2,171 households (NBS, 2012). Census data (NBS, 2002) shows that the population had increased during the previous 10 year period by 3,825 units as it accounted for 7,985 individuals in 2002. Polygamy is not unusual among Maasai living in Engaruka and households with five or more children are not infrequent (see also Winterbottom et al., 2009). Personal observations confirm a priority for boy’s schooling and the practice of early marriage for daughters (see also Archambault, 2011; Bonini, 2006). Female genital mutilation is prohibited by law in Tanzania yet, interviewees report that girls are circumcised singularly in hideouts or the practice is performed to them when they are babies (see also Winterbottom et al., 2009). The local society is patriarchal. Interviewees tend to give a picture of the Maasai society as having a clear-cut gender division of labor where men are herders while women work in the gardens. Reality is somehow different with children engaged in herding and men also doing a share of agricultural work, while women attend to their productive role in a less visible way (see also Wangui, 2014; McCabe et al., 2010). The life of Maasai men is characterized by four age grades: warriors, jun- ior elders, senior elders, elderly. These are successive life stages that men go through during their life. The initiation through circumcision is the first and most important step to entering the Maasai age set. Women do not have age set, but they become automatically members of their husbands’ age set (Bo- nini, 2006; Spencer, 1993). Both Engaruka Juu and Chini have a madukani – market. The market in Chini is on Thursday afternoon and is divided in two sectors: one for cattle, managed by men, and one for beads, vegetables, clothing, firewood, sugar, tobacco and general petty trading, run prevalently by women. Wholesalers come from the neighboring towns of Babati and Karatu to buy black beans. The market in Engaruka Chini is a busy weekly gathering for farmers and pastoralists living in the surrounding villages and bomas. The market in

35 Engaruka Juu takes place Tuesday and Saturday morning and consists mostly of women who sell small amounts of vegetables from their gardens. Even though Engaruka Chini is on a rather busy route, there are just a couple of kiosks and a mill at its entrance. Most shops are located in Engaruka Juu, which is a relatively older settlement. In the madukani of Juu, school, kitchen (e.g. thermos, cooking fat, sugar) and agricultural supplies (e.g. pesticides, seeds) can be bought. A kiosk and a mill are also found here. Engaruka Chini lies on the road from Mto wa Mbu towards lake Natron and the Kenyan border. Tourists travelling to Lake Natron use this gravel road and sometimes buses coming from Kenya go through this route, along which there are no border controls. It is also the shortest way from Kenya to Mto wa Mbu. The Engaruka Express leaves Engaruka Chini at dawn and returns from Mto wa Mbu at dusk. This bus is the only daily guaranteed communication from and to Engaruka. Stopping in Selela, the bus reaches Mto wa Mbu after roughly four hours. From there, several buses can be caught to Arusha and Dar es Salaam to the east and Karatu to the west. While cellphone networks reach almost everywhere in East Africa, Enga- ruka ward remains one of the few sites where there is no phone network. This situation negatively affects locals who, for instance cannot market their produce or ask for help in case of a medical emergency.

Abandoned Engaruka

Engaruka has long been at the center of archeological interest in Eastern Africa. While the first record goes back to the German explorer G.A. Fischer in 1883 (in Sassoon, 1967), excavations began with German archeologists Reck in 1913, followed by L.B.S. Leakey in 1935. A fundamental contribution was made between 1964 and 1966 by Sasoon, who carried out the first radiocarbon tests and could estimate Engaruka´s origins during the first mil- lennium AD. Archeologists’ interest increased and excavations became more frequent and across vaster areas, notably by Robertshaw in 1982 and by Sut- ton in 1978, 1984 and 2004. Most recent work was carried out in 2003 by Laulumaa (2006) and Stump (2006) who demonstrated that what had been previously identified by Sutton (1998) as seven different villages, was in reality one continuous village surrounded by an area of 2000 ha were irrigated agriculture was carried out (Laulumaa, 2006). The complex irrigation system attracted the attention of archeologists who aimed at disclosing the features of this highly specialized agricultural landscape which is considered the most extensive archeological site of this type in Eastern Africa (Stump, 2006). The abandoned system is conventionally divided in the so-called northern, central and southern fields. While the northern fields are the earliest and the most extended, they share with the

36 southern field a pattern of single lines of stones dividing the plots and a network of canals (Sutton, 2004). On the other hand, the central fields are characterized by a grid of fairly small agricultural plots that together with stone cairns seem to indicate that plots division lines came as a result of stone clearance. Other signs of clearance appear to be the stone circles which, it has been suggested, acted as livestock enclosures. Moreover, as no canals have emerged through excavations in the central fields, it is assumed that irrigation was probably taking place by removing stones from the plot boundaries and let the water spread (Stump, 2006). The irrigation systems and the field systems remains have been the object of numerous studies attempting to trace the origins, the development and the reasons of the abandonment of Engaruka (Westerberg et al., 2010; Stump, 2006; Sutton, 1984). The remains have been dated from 15th to 16th century (Sutton, 1986; Stump, 2006) and, it is estimated that the system was in operation between the 15th and the 19th century (Westerberg et al., 2010).

Methods Methods are presented according to the chronological order in which they were carried out all along the study, but also in each fieldwork session.

Focus groups This method was chosen in line with feminist epistemology as “the focus group is a collectivistic rather than an individualistic research method that focuses on the multivocality of participants’ attitudes, experiences, and be- liefs” (Madriz, 2000, p. 836) and it can reduce the risk of a hierarchical relation between the researcher and the researched most common in an interview. Thanks to the minimal engagement of the researcher, this method allows for the natural social interactions of already existing groups to surface with their internal hierarchy and cultural codes. Moreover, focus groups facilitate the “horizontal interview among participants” (Madriz, 2000, p. 840) who through discussion and internal agreement validate their common experience. In total 11 focus groups were carried out during six weeks of fieldwork in January and February 2012. These groups were gender and age specific and were made up of between 6 and 8 people, in order to give everyone a chance to speak up (Morgan, 1997). The table below (Table 1) shows the characteristics of every single group interviewed.

37 Location Name of the group (if giv- Gender Age Number of en otherwise age set) participants Sibou Saving group Sibou F 35–45 7 Sagat traditional dancing F 40– 6 group 60+ Water management group M 35–50 8 Nyongi – young generation M 20– 35 7 Nyongi – young generation F 20– 35 8 Cesawach traditional danc- F 35–50 9 ing group Kaberur – elders M 40– 8 60+ Water management group F 35– 8 Engaruka Kamati ya maji 60+ Wakinamama F 35– 8 60+ Ol Donyo Nnado – Isiangi- F 20–30 8 kin Olemelepo – Korianga M 20–30 7

Table 1. Summary of focus groups

Participants in the focus groups were selected by the research assistants according to age, gender and occupation criteria. Since the questions of the focus groups revolved around agricultural practices, crops and weather, all had to be active farmers. As both communities are quite small, and in the case of Engaruka groups were formed on the basis on of location, all participants knew each other beforehand. The four groups interviewed in Sibou are formal groups: the water management committee, two women´s traditional group and the women´s merry go round (saving group). They were deemed relevant for the study not only because their internal dynamics would be smoother and everyone would feel freer to talk, as compared to groups made up on the spot. Focus groups in Sibou took place in the house where I was staying where participants could take tea and biscuits during the average two hours long discussion. In Engaruka they were carried out under a tree around the houses where people lived and I would take along biscuits or bread for everyone. After the focus groups, participants were given 200 Ksh (20 SEK) or 500 Tsh (5 SEK), corresponding roughly to the price of a soda or work in the fields for half a day. This sum had been discussed and agreed upon with research assistants who had suggested this amount as fair for participating to a discussion for a couple of hours. It was planned that people should not be informed prior to partaking that they would be rewarded with money, but the

38 research assistants reported that they had to mention it in order to encourage, especially men, to participate. As underscored by Krueger and Casey (2009) and confirmed by research assistants recruiting male participants, a monetary reward had the function to stimulate men´s participation so that they knew they could come away with something from this activity. I had long discussions with my research assistants on why money had to be mentioned to men, while women would never ask. This issue became particularly interest- ing when a couple of groups of men complained that the amount given was too small, while three groups of women decided to keep the money in their common fund. I tend to think that because I knew most of the women per- sonally before carrying out the focus groups, and they were well aware of what my study was about, they were less reluctant to participate. It is true that I knew men less before focus groups, but yet during another discussion on whether to pay cash participants or to just serve tea, sodas and biscuits, which would have cost me more than paying everyone 200 Ksh, my male research assistants motivated men´s behavior by claiming that: “Men don´t care about tea and biscuits, they want to have the money, even if it is worth less, because it is a symbol that they have gained something material”. This statement is dense with meaning. Men´s wish to receive a monetary compen- sation, rather than food, could be part of that cultural overlay of symbols that indicate men´s control over material resources. Such behavior could be connected to the same reason why women in Sibou cannot irrigate autonomously. Control over water might just be another “material symbol” of men´s homosociality and patriarchal societal power. Focus groups literature (e.g. Krueger and Casey, 2009; Morgan, 1997) recommends carrying out focus groups in the mother tongue of participants to minimize the risk of error, misunderstandings and also to be more time effective. Thus, research assistants were trained on how to carry out activities autonomously for a couple of afternoons before starting. While they were already conversant in the topic of my study and during interviews could fill in my questions whenever I was stuck or feeling unwell, conducting a focus group autonomously was perceived by them as a challenge. After rehearsing all the tasks with the four of them in Sibou and with the two of them in Engaruka, they discussed among themselves and addressed questions or doubts the following day. Assistants were also in charge of sketch- ing and setting up prior to the activities the poster used for the focus group. In Sibou assistants carried out the focus groups in couples of two men or two women, while in Engaruka all focus groups were carried out by my female research assistant, as the male counterpart´s level of English was not suffi- ciently high for me to fully rely on him. During focus groups I would always sit in the back so that I would not be a distraction for participants, but I could nod to the assistants if they had any doubt on the order of activities or remind them of some details. Eye contact and understanding each other´s look was crucial during this exercise (see

39 also Leeuw de et al., 2012). I think we managed it quite well probably because our relationship was closer to a friendship rather than a professional one. Most of the times I did not even intervene once during the focus group and I had the perception that participants had fully forgotten my presence as they were quite taken by the questions they had to reflect and agree upon. Moreover, since participants were aware that I could not understand a single word of Kimaasai or Marakwet, the risk of biased answers was minimized (cf. Jakobsen, 2012). The focus groups activities were mainly four:

1. Participants had to list the crops that they cultivate with the timing of planting, weeding, irrigating and harvesting. Crops were not listed in order of importance, but the analysis showed that participants had actually done so, as the crops that were the most important for them, be woman or man, were the ones they knew best. 2. Participants were asked to describe the crop cycle also for the previous two years: 2010 and 2009. Secondly they had to rank the last three years according to crop performance. They were given 20 bot- tle tops or beans that they had to divide and allocate to the three years. This was certainly the most interactions dense and tense moment of all the focus groups. People would raise their voice and even stand up pointing at the poster. As I couldn’t understand the language, a couple of times I worried people were having an argument, but I was reassured by my assistants that it was not the case. 3. Participants had to go through their memories of the weather of the last 30 years and mention the years that were absolutely positive and negative in terms of crops yields. Finally, people were also asked to rank the last 3 decades. Since it is understandably hard to remember weather in the 1980s or in the 1990s, assistants had thought through the important community events of the last three decades and would mention them to participants whether people were unsure. These events included circumcisions’ ceremonies in Sibou, as these happen roughly every 6–8 years when there is a good harvest and in these occasions a new age group is named. In Engaruka the most recent volcanic eruptions of Ol Donyo Lengai were used as a time reference. 4. Lastly, participants had to go through the events that take place on a yearly base in their community. This exercise could be conducted only in Sibou because there is a specific timing every year for different types of traditional ceremonies and elders meet every year in February to decide the timing of cultivation or the clearing of new plots. In Engaruka time is not paced by traditional celebrations as in Sibou: they can happen anytime a family has the means to celebrate a marriage or a circumcision. Asking this question was relevant not

40 only to understand more of the local rituals and culture, but especially to understand if there is a frequency to the agricultural practices.

Before starting the discussion participants were asked if everything could be recorded. Usually the focus group would take place in the morning. Since the crop calendars and the results of the rankings were easy to grasp on the spot, at the end of the discussion I would address some extra questions to participants in relation to labor dynamics during cultivation or the crops that they knew most of. At the very end, I would open up for questions about myself or the study and stress that anytime they would remember something more about the topic of the focus group they could contact my assistants and tell them what they came up with. Days later, when I interviewed some of the participants, several times they told me that they had thought through things and they had some details to add. In the afternoon, one assistant would translate the tape for me and I would take notes of the details I had missed from the representation in the charts. Moreover, a couple of times during lunch at Tot Centre men and women started questioning each other on why they gave different answers to the rankings and had an animated discussion on what was right and wrong. These incidents were crucial to improve my understanding of the gendered discourse and perceptions around crops and agricultural practices. Retrospectively, I argue that the specific activities and settings used for group interviews borrowed some elements from the workshop (see Caretta & Vacchelli, in press). A workshop is defined as a group activity where participants learn something and is mostly employed in participatory action research (PAR) (Moschitz and Home, 2014; Reid and Frisby, 2008; Chambers, 2002). In PAR workshops are used to engage participants in determining the research aim and in taking stock of results to improve local conditions (Reid and Frisby, 2008). Brainstorming, role playing, collage, sorting of pictures, ranking (usually with the help of stimulus material as pictures), flipcharts, posters, project plans, etc. are common activities in workshops (Chambers, 2002). These tasks partially correspond to the methods I employed in my research. While they were not necessarily meant to trigger a learning process aimed at the participants, they stimulated the production of situated and contextual knowledge (see Caretta & Vacchelli, in press).

Interviews Semi structured interviews were carried out with 39 women and 19 men in Sibou and 13 women and 12 men in Engaruka. Most interviewees were picked from focus groups to deepen understanding in relation to agricultural practices and the local gender contract. As for focus groups and as recommended by Williamson et al. (2011), I instructed field assistants that I did not want them to soften or sanitize answers and we went through the inter-

41 view guide together. Research assistants would arrange autonomously interviews and suggested new interviewees who they deemed to be a reliable source of information. Such convenience and snowball sampling were usually employed when elders or local authorities had to be interviewed, as it was easier to interview someone with whom the assistant had already a direct contact. At the beginning of each interview, which were carried out in people´s home or field, I would greet people in Swahili or introduce myself and the study I was carrying out, depending on whether we had met before. Even though people did not fully understand Swahili, which sometimes was the case in Sibou, I still chose to say a few sentences for them to realize that I was making an effort to facilitate our interaction, instead of requiring the mediation of a translator all the time. Prior to beginning, all interviewees were asked for their consent to record the interview and were reassured that they would remain anonymous. No one ever objected to being recorded. People were not required to sign a form to give their consent as this would have required extra explanation and probably would have made them suspicious and skeptical about participating in the study as the majority of them could not understand written English. Moreover, people in these remote communities are rarely asked to sign any- thing and no one has a written proof of the ownership of their land, thus some can be afraid that signing an official document may entail giving away their land. The irrigated landscape, or waterscape, was a crucial component of my research, not only because the study of its changes is at the core of my inves- tigation, but also because it is a component of all the methods I employ. In fact, observation, in depth and open ended interviews and participatory mapping were all taking place in the waterscape or while walking through it. Carrying out fieldwork in a landscape is not just observing or walking or interviewing or taking pictures: it is all of them together. This is a performance because body, subjectivity and mobility shape the research we conduct (Setten, 2013). In fact, walking in the landscape is a strategy to get ac- quainted with the surroundings, meet people and be seen (Årlin et al., in press; Börjeson, 2004; Widgren, 2004a). So called “walk along interviews” (Carpiano, 2009) are functional to witness events and have accidental conversations with people about events that they consider important. In this case the interviewer follows the interviewee in the places of his/her daily life like his/her plot or his/her neighborhood. Interviews were conducted according to several themes (i.e. agricultural practices, gender division of labor, crop types, weather predictions and perception), rather than precise questions, which I had sketched in my notebook. This was the preferred method because it avoided truncating the conversation in order to follow a precise interview guide (Falconer and Kawabata, 2002). In fact, following the answers the interview would take different turns, while still trying not to lose com-

42 pletely the focus and return to the main themes. Because of the unstructured nature of the interviews, their up to two hours length and the diversity and richness of responses, I could only pen down in my notebook few keywords while on the spot. In the afternoons, when the temperature was too hot to walk around and keep on interviewing, I would listen to the tapes again and make sense of my keywords by noting the precise statements that interviewees had made. Transcripts, whether the result of listening again to the tape by myself or the direct translation by an assistant, were analyzed and used according to an interpretivist approach, which takes into account the context in which interviews were recorded and transcribed. In this way the partiality of transcripts was acknowledged because they were considered a reconstruction of the actual interview and not the objective rendering of the discussion that had taken place (Lapadat, 2000 in Witcher, 2010). In fact, transcripts have their shortcomings as “nonverbal clues and emotional context of interviews are difficult to depict” (Witcher, 2010, p.124).

Plot survey A household survey was designed together with assistants, who conducted it autonomously every three months between January 2012 and November 2013. Seven farmers were selected in each Sibou and Engaruka, according to the different characteristics of their plot: crops cultivated, age of plot, location and type of soil. They were informed of the survey and gave their consent for it to be carried out for at least one year. In the end, complete data from 12 plots was gathered. Soil samples were taken from each plot and analyzed at Jomo Kenyatta University during January 2012 and February 2013. Water samples were also taken and analyzed in February 2014. This survey was primarily designed to complement a study of soil formation processes in the two sites, led by Lars-Ove Westerberg, at the Department of Physical Geography, Stockholm University, that was carried out parallel to my own work. I have only to a limited extent used data from this survey, e.g. information on cropping practices and on gender division of labor.

Participatory mapping Participatory mapping in this study consists of discussing land use changes, field and community histories together with farmers on the basis of satellite images and historical aerial photographs (Chilvers, 2009). This method served the purpose of acquiring data on the irrigation systems expansion and on the chronology and developments of the surveyed plots. Together with member checking (see article IV), participatory mapping was the most engaging method employed. At first, most farmers would get

43 confused by the satellite image and have a hard time grasping it. Some of them even manifested the fear of being somehow spied from the sky, so it was challenging to explain what a satellite was and that it would only take images of physical details in the landscape, not of people. When the assistants and I started pointing at details in the images that surrounded the interviewees in that precise moment: an irrigation channel, a fence, a tree, they would overcome their initial surprise and start reading the images with an interest. This turned out to be a very successful method because by being in the fields, asking questions with the support of the satellite image, people could make reference to features surrounding them and link them to their memories of the changes that had taken place (see also Årlin et al, in press). The technique of using a marker to single out plot borders was especially helpful because it added another level of understanding between the interviewee and me as it showed that we could fully comprehend each other, even though we would not speak the same language. During the last fieldwork participatory mapping was carried out with groups who were selected by the research assistants according to their age and their occupation: they all had to be active farmers (see Table 2).

Location Name of the group Gender Age Number of and location participants Sibou Women- Cheboryot F 35–45 9 Water management M 40– 5 group- Kapsimotwo 60+ Water management M 30–50 9 group- Arron tree Nyongi – young gen- M/F 25– 40 10 eration - Kimaina Nyongi – young gen- F 20– 45 7 eration – Kapkutung Kaberur – elders M/F 50+ 6 Engaruka Water management F 35– 8 group - Kamati ya maji 60+ Ol Donyo Nnado M/F 20–40 7 Ndimi Juu- Neng´alah M/F 25 –40 8 Ndimi chini/ Baraka – M 40– 5 Wazee 60+ Arushani M/F 25–35 10

Table 2. Summary of participatory mapping groups

Groups were mostly gender mixed. In a couple of instances in Sibou groups were made up by only women as they were all part of an already existing

44 women only saving group in Sibou. In Engaruka, even though women sat in those groups, they hardly ever spoke of because the local patriarchal gender structure which endows men with the role of community leaders and speak- ers on behalf of their wives. When interrogated women would most commonly said they did not know the answer to the question. It was typically challenging for them to even get to the see the satellite image as men would literally and physically take control of it. At the same time, it was not feasible to engage women only groups in Engaruka. Women were reluctant to participate and would claim that their husbands were the ones owning the land and knowing best. These mapping exercises took place in different outdoor spots from where people could recognize landscape feature and be able to locate themselves in the satellite image. As focus groups, the mapping was carried out autonomously by assistants who had previously been instructed. I would sit in the back, without interrupting; only taking some pictures at times. After having explained what the image represented the assistants would proceed to single out on the image the location where the exercise was taking place and from there guiding participants to single out their plot and circle it with a colored marker. Participants were asked to differentiate whether they cultivated that plot on a continuous base or on a shifting one. Singling out one plot was often a very complicated task for most participants who would require for others assistance in finding their plot on the image. Overall the mapping exercise was time consuming for people who were compensated with a 50Ksh (4 SEK) in Sibou and a soda in Engaruka, according to assistants’ suggestions. At the end of the exercise participants had many questions about how the data would be used, whether they could access the picture again and how exactly the image was taken. I would reiterate that the images would be left to my assistants and that they could be considered some sort of gifts to their communities from Stockholm University. Walking around Engaruka and Sibou with these big rolls of satellite images called for people´s attention. People would come to me and ask if they could participate and check the picture and give their input. It was the first time that this happened during my study: I did not have to look for participants, people were volunteering because everyone was curious to understand what the image represented and how it was taken.

Limitations Sibou is probably the ideal place to carry out fieldwork in a remote location. Even though four hours from the closest tarmac road and five hours away from the closest hospital in Iten, I could keep communication with my family and work in Sweden. Moreover, I had the luxury of living in a small house by myself located in a guarded compound with running water and electricity three hours a day. I also had someone cooking and washing clothes for me

45 because, as much as I wished I could cater for myself, I simply had no clue where to start from. Those simple tasks would have taken so much time off my research, that I preferred paying some ladies to do it for me, especially after I had heard they were single mothers with several children who they could not all send to school. Most importantly in Sibou I could count on the help and support of four assistants who were available to sort out any problem that would arise. They were very well organized among themselves, shared tasks and knew everyone in the community, so getting an interview or visiting someone´s field was just a phone call away. The situation in Engaruka was practically the opposite. I could only count on the help of one assistant. Catherine Mollel, who is a single Maasai ele- mentary school teacher in Engaruka Juu, is the only one in the community that speaks good enough English to be able to carry out interviews and translate. Engaruka ward is distributed on a much bigger area than Sibou and it was not feasible to walk to all locations, especially given that the temperatures usually are above 35C. Consequently, interviews and focus groups were usually concentrated in the areas relatively closer to where I was living. But on Tuesdays we would walk more than one hour under the sun to Enga- ruka Chini for the market and arrange for other interviews. Beside Catherine, I could count on a Maasai man, Israel Mollel, who had previously assisted archeologists and physical geographers from the UK and Sweden. However, the nature of those studies did not require a proper translator and his English is quite limited. In my first fieldwork we tried to work together for some interviews, but it soon turned out to be too stressful for both of us. He had a hard time getting through the questions I wanted to ask and I could not acquire the details I had planned to. Misunderstandings between us were quite frequent due to the language barrier that his limited English and my limited Swahili could not overcome. Moreover, his latent chauvinism did not go well along with my will to be treated as equal and my continuous questioning and attention to details differed from his interpreta- tions of things and wish for quick answers. I sometimes had to take a quite strong stand and ask him to let Catherine and I do our work – when she was present – while he would go off to somewhere else. While we never confronted each other, Catherine had to mediate between us sometimes. It is, thus, quite evident that my state of mind during fieldwork was not always characterized by peace and concentration, having to carry out work in a new context with new people with whom I would not always get along. Neverthe- less, his help remained important given the contacts he had in the community and the motorbike he owned, which was useful in arranging interviews or taking me and my assistant around to meet respondents. Having learnt Swahili represented certainly an advantage because I could communicate directly to people and my thoughts did not need to be mediated through my assistants. Being able to understand directly what interviewees would say without the need for translation was certainly a big plus. I under-

46 stood what was being said and I did not worry or doubt the translation I would get. Nevertheless, carrying out interviews autonomously or not needing translations of the answers made the interviews faster, which gave me less time to think about the following questions. To avoid jumping from one question to the other, I would take some notes here and there and carry out communicative validation at the end of the interviews, checking also whether I had some more questions that had remained unanswered. There were challenges in carrying out fieldwork in two different countries: more coordination to arrange the logistics of fieldwork was required, especially considering that in Engaruka there is no phone network. It was practically impossible for me to communicate with Catherine when I was in Sweden and plan for the following fieldwork or even just know how she was doing or how the weather was in Engaruka. She rarely had time to take the one and a half hour bus to Mtu wa Mbo, the closest town with functioning communication networks. The lack of a communication network, which apparently can only be reached on top of a tree forty minutes walk away from Engaruka Juu, did undeniably influenced my fieldwork in Engaruka. Being isolated for weeks was not of great help for my mood at times, but I managed to channel my energies and concentration to work as hard as I could in the least amount of time. In this context, Catherine and I became close friends, not only because we spent time together all the day, but also because we became eager to understand more and more of each other´s culture, to share our reciprocal histories and frustrations. At the end of my fieldwork sessions, I always felt a great sense of accomplishment, not only because I thought I would have got enough good data for my research, but because I had found the inner strength to carry through in such an opposite, and sometimes adverse, situation as compared to my daily life.

Positionality Positioning yourself critically and reflexively, explaining your partiality and providing connections rather than final answers is crucial in an ethnographic research process, particularly under the lenses of feminist epistemology (Cook et al., 2005; Rose, 1997; England, 1994). While in this thesis I want to unveil unjust patriarchal power structures, I am aware that I am in no better position to understand women in East Africa than a man, just because I am a woman. I have also tried to be cautious in not reproducing a vision of women as powerless, subdued, poor and unedu- cated (Cornwall et al., 2009; Mohanty, 1984). On the other hand, my presence and my being openly “on women´s side”, as one of my assistants put it, conditioned the research process and its outcomes. I transgressed local social norms as white, solo, unmarried and childless woman. As a researcher, I was challenged by locals who manifested their

47 expectations for the results that will be produced from the study, particularly on the aspect of soil fertility. I came to the realization that “ethnography is both a process and a produce, ethnographers’ lives are embedded within their field experiences in such a way that all of their interactions involve moral choices” (Tedlock, 2000, p. 455). I reflected through dilemmas and difficult circumstances with the method of autoethnography, which com- bines the research process “graphy” with culture “ethno” and the self “auto” and is considered a crucial performative and emancipatory exercise within feminist methodology. This method brings to light the limitations, struggles and vulnerability of the researcher (Billo and Hiemstra, 2013; Ellis and Bochner, 2005). I was aware that being an outsider to a community would require a certain level of flexibility and that, given the gap between my positionality and the local culture, I could not perform my research objectively (see also Rama- zanoglu and Holland, 2002; Rose, 1997). I faced some situations and saw some scenes that haunted me and that made me wonder “What should I do?” while being afraid of doing something because I felt vulnerable in my position as white, western and lone female (see also Kannen, 2013; Acker, 2000). The dilemmas I faced can be roughly categorized in two categories: violence and money. I witnessed violence against school pupils in numerous occasions and I was once threatened by a drunkard, who came running after my assistant and me. As for money, I had rehearsed my answer: “I am a student and I do not have enough money myself, so it is hard for me to help, but I am very sorry to hear what you are going through.” As recommended by Nash (1994), I wanted to both detach myself to make sure people did not profit from me because they would ask for help also the next time we would meet or would withdraw their collaboration because I was not assisting them. However, the frustration grew stronger and stronger repeating this sentence, especially in some cases which really looked like they could use some money. During my second fieldwork, I made a donation for a 10 years old boy to reach the nearest hospital, get an X-ray and then a cast on his arm which he had broken falling from a papaya tree while attempting to get the fruit because he was hungry, according to the mother. Following this cir- cumstance, I did not get more requests or critiques from people because, according to my assistants, they had realized it was an urgent and unique case that required immediate attention. I never gave money directly to people, but I supported certain institutions in the local communities. During the first fieldworks project money were used to pay for books for the primary school in Engaruka and for cement to build a laboratory in the secondary school in Sibou. I also started a fund raiser among friends and relatives to pay secondary school to one boy in Engaruka and two girls in Sibou. They all had promising grades in primary school, but could not continue to secondary school because of lack of funds as they are orphans. Special dona-

48 tions by friends and relatives went for primary school books in Sibou two times. Moreover, at the end of every fieldwork I would leave a certain amount to my assistants to distribute it to those people, especially widows with several nieces and nephews to take care of, single mothers with small children or women groups who wanted to rent a plot, whose situation had moved me. I am not going to pretend that emotions did not play an important role in my fieldwork. While during interviews I somehow learnt to conceal my emotions when interviewees would bring up sensitive issues, I fully dis- closed my happiness when the boy came back to show me the casted arm and when I told two girls and their single mothers that I would support them in finishing secondary school.

Data analysis It should be noted that the data gathered on crops was not used to do a direct analysis of agricultural production, but used to analyze changes in agricultural practices, with a focus on gendered tasks. Data gathered in this study was analyzed in two ways. First, explanation building was done through narratives (see Yin, 2009). This technique was used for articles I, IV and V. Interviews were first transcribed by me or with the help of assistants and then were read through to extrapolate the data that had to be dealt with in the specific article. Data was then arranged in charts, timelines, chronologies and matrixes to categorize the evidence gathered. The categorization helped to get the general picture and single out the general findings which were reported in the articles also in the form of citations. The second way in which data was analyzed was by doing explanation building by testing theoretical propositions (see Yin, 2009). This technique was used in articles II and III. Data was organized in charts and chronologies containing variables deducted from the two theoretical concepts used (see figures in article II and III). Hence, chronologies were not just used as de- scriptive means, but rather as analytical tools (Yin, 2009). Proposition were tested through these matrixes and chronologies. For instance, was it true that women mainly would carry out incremental work? The data in form of citations and charts was piled and organized to verify this proposition. This process had to be done over and over to see whether the data fit or not into the theoretical proposition and to which extent. It should also be noted that data analysis did not only take place in the post fieldwork phase. Member checking (see article IV) jumpstarted the analysis already in the field since the data gathered could be counterchecked with participants and assistants who would clarify how I should interpret it, adding more details to previous information they had given me.

49 Triangulation Triangulation of results was ensured by the multiple sources of evidence and the array of methods used in the course of the research. First, these methods very successively used to get a more detailed picture of the phenomenon under attention from the general to the specific (focus group to in-depth interview; participatory mapping to walk around interview). Secondly, these data was taken all together and counterchecked through the survey and through member checking (see article IV) to go from the specific to the general again. Triangulating these multiple sources allowed for the construction of validity, which was additionally ensured through member checking as participants and assistants could review preliminary data, correct it and help already in the analysis (see Årlin et al., in press; Yin, 2009; Cho and Trent, 2006). This process was applied to every single article of this thesis. Finally, articles were triangulated using data that could not be included in the articles as well as secondary material in the form of academic publications and international reports in order to discuss the overall findings of the thesis and to be able to draw the general theoretical and empirical conclusions to be presented and discussed later in this chapter.

Reporting of results Results were presented to participants and informants in the form of a book- let in English, Swahili and Marakwet during a fieldwork in January 2015. Copies were distributed to local authorities and to schools, where it would be used in civics. The booklets reception was exceptionally positive. Partici- pants were eager to get a copy of it and were thankful to get a chance to read the results of the study that they had taken part in during the course of four years. It was surprising to hear that my supervisors and I were the first researchers to travel back and take the time to report back our findings, as numerous researchers have been conducting studies in these two localities since the 1970s. For me it was an ethical imperative to materially show my gratitude to participants for having taken the time to share information with me. The fact that this is not a common praxis within science, especially within social science, shows the superficial attitude to ethics that appears to be prevalent within academia. The second aim of the last fieldwork was to have assistants and farmers from both study sites to travel to the other study location, to get to know each other and exchange knowledge on agricultural practices and social norms. Farmers exchanging seeds and discussing irrigation schedules, debat- ing why women are excluded from irrigation and different women groups encouraging with each other to pursue bigger projects are just some exam-

50 ples of how this outcome was successful. Ending, like this, with action research, I argue, was needed to show respect and appreciation to all of those that have contributed to this research and for the results of this work not to be confined to the academia (see also Årlin et al., in press), but to be spread among those that can benefit most from a critical perspective on irrigation management and women´s participation in it.

52 Summary of studies

This thesis is comprised of five articles that are presented in the order out- lined by the research questions. The first three articles focus on the three aspects of waterscape – institutional arrangement, gender relations and landscape investment – at the topical core of the thesis, while the final two articles deal with methodological issues related to the case study design of the research. The first article examines the common-pool regime of the local irrigation system in Engaruka, Tanzania, in order to understand how water conflicts are mediated while ensuring agricultural productivity against the background of limited and variable water supply and population growth. The second article investigates how the local patriarchal structure has shaped the common-pool water regime of Sibou, Kenya, and the gender dimension of adaptive capacity to climate variability. The third article studies the role played by cultural norms in determining the gender division of labor both in Engaruka, Tanzania, and Sibou, Kenya, and how this has materialized in agricultural landscape investments over time. The fourth article is a methodological paper which presents member checking: a method used in the research to ensure validity beyond triangulation in the context of cross cultural and cross language research. The fifth article addresses another aspect of cross cultural and cross language research: the role played by research assistants in shaping the process of knowledge construction.

Article I: Managing variability and scarcity. An analysis of Engaruka: a Maasai smallholder irrigation farming community.

This article examines local water management and agricultural practices currently employed in Engaruka. Management of limited and fluctuating water supply is a fundamental challenge facing many irrigation systems. Water allocation in such a context can generate conflicts, especially when

53 the irrigation system is the most important source of livelihood for the local community. These circumstances are further strained when population pressure increases demand for an already limited resource. In Engaruka water use is regulated through boundary, allocation, input and penalty rules by a users’ association that constantly controls and negotiates water allocation to avoid conflicts among headenders and tailenders. As different crops – maize and beans, bananas and vegetables – are cultivated different watering schemes are applied depending on the water requirements of every single crop. Farmers benefit from different irrigation schedules and from different soil characteristics by having their plots both downstream and upstream. In fact, depending on water supply, cultivation is opportunistically extended or reduced. Engaruka is an ethnical homogeneous and interdependent community where headenders and tailenders are often the same people and are hence inhibited to carry out unilateral action. Drawing on common-pool resource literature, this study argues that in a context of population pressure and limited and fluctuating water availability, non-equilibrium behavior, consisting in the negotiation of water rights and modification of the size of the irrigated area through water supply demand management, is crucial for the satisfaction of basic and productive needs in Engaruka and for the avoid- ance of water conflicts.

Article II: Local gender contract and adaptive capacity in smallholder irrigation: a case study from the Kenyan drylands.

This co-authored article investigates adaptive capacity to climate variability in Sibou, Kenya and how this is dictated by local patriarchal norms. By taking into account the local societal and cultural preconditions that bring about a certain gender division of labor, while not excluding ongoing change in these power dynamics, we argue that the concept of local gender contract can serve as an analytical tool in the study of (changing) gender structures. We show how the local gender division of labor is mirrored in men’s and women’s different strategies to adapt to erratic and changing climatic patterns. The local patriarchal gender structure give men the control over irrigation water and lately also the control over cash crops. Women are hence responsible for cultivating traditional crops and ensuring the family´s day to day food security. We define this as local resource contract. The relatively recent introduction of improved seeds in Sibou has allowed for the diversification of livelihoods and the cultivation of crops for sale. Programs giving out commercial crops seeds were directed to men who have since then ac-

54 quired directly the income generated from commercial crops. Women and men nowadays tend to collaborate more in sowing and harvesting than they used to do before. We denominate this as household income contract. The gender division of labor is being re-negotiated and, we argue, the local gender contract is also being renegotiated at the family level. However, we note, the new local gender contract, while featuring more shared tasks, does not entail a more equal division of labor, but rather adds the burden of women through the sowing, weeding and harvesting of cash crops on top of their usual work with food crops. On the other hand, women´s knowledge on food crops seeds is acknowledged by men and there are signs that a new gender contract could emerge in the future. Some women have started to challenge the current local gender contract by washing clothes in the canals and by controlling the income generated through petty trading. Finally, the article is one of the few using the concept of local gender contract outside its Scandi- navian originating context and argues for its usefulness in the analysis of gender division of labor in subsistence farming. For this co-authored article I acquired and analyzed data and I did a substan- tial part of the writing and of the production of figures.

Article III: Hydropatriarchies and landesque capital: a local gender contract analysis of two smallholder irrigation systems in East Africa.

This article studies how local gender structures are mirrored in men´s and women´s respective contribution to landscape investments and how they have changed over time in Engaruka, Tanzania and in Sibou, Kenya. It is often said that while women are the owners of the crops, men own the land. The study of landesque capital mirrors this dichotomy by focusing on the visible systematic investments in the form of irrigation canals or terraces usually made by men. In this view, the work of women has largely remained invisible. By employing a local gender contract analysis, I analyze the societal norms underlying the local gender division of labor and how it is visible in the landscape. Hence, this article shows that due to the local patriarchal structure that puts women in a subordinate position, their participation in the realization of systematic tasks is not acknowledged. Moreover, incremental landscape capital, which includes less apparent, and hence less researched, forms of landscape investments as soil fertility, is often the main result of women´s work. While a clear line cannot be drawn between the tasks of men and women and their related landscape features, a connection can be drawn between the local gender contract and how it is physically and rhetorically manifested in processes of land use intensification.

Article IV: Member checking: a participatory method to test and analyze preliminary results in cross-cultural, cross-language research.

This article presents a method that I have used to improve the trustworthi- ness, transferability and confirmability of my PhD research, while facilitat- ing a less hierarchical relationship between the researcher and the researched. Based on similar methods developed in the 1990s by colleagues, I created a pamphlet in Swahili summarizing the findings from the previous fieldwork session, which was used to engage participants, to complement and correct data. This method is defined as member checking as it consists in counter checking preliminary findings with informants and participants to ensure their validity. Several photos were included in the pamphlet, so that analphabets could recognize familiar faces and recall the content following the images. Photos attracted family members’ attention helping the task of explaining the content to them, while spreading the knowledge of the ongoing research and my presence in the community. This pamphlet was used to start off interviews and focus groups from where they had been left off. Par- ticipants and interviewees were asked to assess whether the content of the pamphlet was in line with what they had been discussing the previous time. The employment of this method represented also an attempt to subvert the power hierarchy that is inherently present in cross-cultural cross-language studies. Moreover, I capitalized on my decision of coming back to the field several times both to build trust with participants and to deepen my understanding of local conditions. While triangulation would probably not have made emerge this, distributing the leaflet during the second fieldwork dis- closed my erroneous grasp of community practices. In the spirit of reflexivity, it should, however, be noted that the pamphlet was not always an effective tool. For instance, not all participants felt at ease contradicting me or speaking their mind. Finally, in this article I argue that employing a plethora of methods can ensure validity beyond triangulation, a higher quality in data gathered and that innovative methods should be tried out repeatedly in the field to harness their full usefulness.

Article V: Situated knowledge in cross-cultural, cross- language research: a collaborative reflexive analysis of researcher, assistant and participant subjectivities.

Data gathering for this thesis was done with the help of assistants and interpreters. Their presence and collaboration is, however, rarely acknowledged in cross-cultural, cross-language social science research. This article tries to close this methodological gap by showing a threefold perspective: the researcher, the research assistants and the participant. Diaries kept by myself and three research assistants as well as field notes are used as material for this article. By analyzing a series of situations – featuring conflicts, emotions, anxiety and misunderstandings – between these three subjects, it emerges how power hierarchy shapes the research process. First, the ideal of feminist epistemology to facilitate equal relations between researcher and research participants is hardly attainable even when the researcher is open to be questioned and confronted. Secondly, assistants play a crucial role in the process of knowledge production, not only by translating, but also by orient- ing the researcher towards a specific focus, ensuring the researcher´s safety and pursuing their own research agendas. Finally, validity of cross-cultural studies is enhanced through reflexive methodological analysis of the process of knowledge construction involving all the subjects engaged.

57 58 Concluding discussion

In this section I take stock of the results of this study and of the conceptual framework used and attempt to situate them in the ongoing debate on the present and future of African agriculture. The discussion is presented according to the three elements of waterscape that I brought forward in my work – institutional arrangements, gender relations and landscape investments – by making reference to previous ethnographic and conceptual studies.

Common property regimes: scale and external factors In this thesis, given the limited and variable availability of water in dryland environment, water management has been studied in Engaruka and to a less- er extent in Sibou. While local contexts and agricultural practices might differ, it emerged that both systems have been able to adapt to growing population pressure, increasing market demand and changing climatic conditions. Flexible irrigation infrastructures allow farmers to shift the course of water and to extend and retract cultivation depending on seasonal rainfall patterns. Moreover, diversification of livelihoods, play an increasing role in both study sites. Water conflict in Engaruka is adverted through a decentralized common-pool resource management system particularly through the figure of the water divider mgawa maji, as discussed in article I. Water rights are continuously renegotiated depending on water supply. Size of plots is increased or reduced through water supply demand management, which can be described as non-equilibrium behavior. In the face of these challenges, it remains, however, to be seen how and for how long this conflict-free condition can be maintained in Engaruka. The challenges faced by the study sites are common for Sub Saharan Af- rica where population is growing fast and where poverty is prevalent (You et al., 2010; Diao et al., 2007). Food demand is expected to rise by 50% by 2030 (Wheeler and von Braun, 2013). Climate change will also play into this future scenario with a location dependent increase or decrease of precipita- tions and temperatures (IPCC, 2014). In this context, irrigation has been heralded as a solution by several organizations and scholars. Irrigation is part of a package of measures meant to improve the productivity by intensifying African agriculture, now promoted

59 as New African Green Revolution1. National initiatives, international think- tanks and development organizations have been arguing that smallholder farmers´ market inclusion will trigger a wider economic growth in Sub Sa- haran Africa (AGRA, 2013 and 2014; Haug and Hella, 2013; Lawrence, 2010; Diao et al., 2007). The overall aim of this public-private joint venture is to increase land productivity by closing the yield gap, that is to say reduc- ing the gap between produced and potentially producible yields in countries with limited land through a technological package of seeds, fertilizer and irrigation coupled with machinery (Moseley et al., 2015; Foley et al., 2011; Diao et al., 2007). A glimpse of such changes is already visible in the two study sites, in the form of new and improved crops that are cultivated by few, but an increasing number of men. While local agriculture was based on intercropping, mono- cropping is required for these new crops due to the risk that pests might de- stroy or spoil them and hence decrease their sale value. Participants bring as an example tomatoes which are prone to pests when they are planted among or by beans (see article II). Pesticides, mostly in spray, are also used, especially on tomatoes and watermelon to avoid the insurgence of pests. These crops are produced mostly for sale and are only to a limited extent consumed locally and their production is intensive – with as many as three or four har- vests every year. While full capitalist market integration is hampered by poor road infrastructures, the currently conflict-free common property regimes might be faced with external factors that may make the system more rigid – due to production requirement – and thus break their constant non- equilibrium shift. New actors, as wholesalers, might change the homogeneity of the local populations, which according to Ostrom (1990) have made the system durable, and may thus put it into jeopardy.

1 The New African Green Revolution is not a new idea. Already during the British Empire irrigation was theorized as part of a package of enhancements for African agriculture which were generated through external inputs (Adams and Andersson, 1988). The Green Revolution was then promoted during the 1970s in Africa, Asia and Latin America. While in South Asia between 1966 and 2005 yields went up by 240% (FAO STAT, 2009 in Nengin et al., 2009), equal progress was not made in Sub Saharan Africa. Corrupted African governments, defi- cient organization and lack of longsighted coordination among different development goals were singled out as the cause for failure (Collier and Dercon, 2014; Lawrence, 2010; Dorward et al., 2005). Foreign aid to agriculture was diminished after the 1970s and led to the collapse of the extension service (Lawrence, 2010) and the discontinuation of several infrastructural projects (Wiggings, 2009; Bahiigwa et al., 2005). More recently, the elimination of subsidies, which were previously ensured through foreign aid, drew down the prices of crops and consequently discouraged farmers from producing hybrid maize and horticultural products in East- ern Africa which had previously taken off (Wiggings, 2009; Kherallah et al., 2000; for a complete analysis on Kenya see Poulton and Kanyinga, 2014). The failure of the Green Revo- lution in Africa led to so-called agro-pessimism as it was questioned during the 1990s and early 2000s whether agriculture could tow African economic growth (Holmen and Hyden, 2011).

60 At that point, the common property regime concept will not be fully ap- plicable anymore. In fact, as critics of Ostrom´s framework have argued (e.g. Cox et al., 2010; Tucker et al., 2007; Cinner and McClanahan, 2006; Berkes, 2006; Young, 2002) these institutional arrangements work under only certain conditions among which population size and resource boundaries are the most crucial. Hence, common property regimes, may only function in small scale systems. They can eventually be nested in larger systems due to the cross-scalar characteristic of the resource at stake (Cox et al., 2010). Howev- er, that is not the case in the study sites as the local communities are the main appropriators of the Engaruka River and of the Embobut River. They seem to fit into the case described by Bardhan (2000, p. 861) that “coopera- tive behavior in an irrigation community is by and large significantly related negatively to….urban or market connections”. It is plausible to infer that, as in other studies (Tucker et al., 2007; Cinner and McClanahan, 2006) capitalist market integration can therefore lead to overexploitation of water and rising inequality among farmers, depending on whom takes up cash cropping first. Conversely, increased market dependence in presence of an improved road infrastructure, might also trigger livelihood diversification and de- creased dependence on the local irrigation system (Gibson, 2001; Pinkerton and Weinestein, 1995). However, because “the Marakwet…do not just ´manage irrigation´, but rather manage ´water as part of the technologies of daily life” (Davies et al., 2014, p. 2), and the same can be said for the Maasai of Engaruka, less dependence on irrigation will mean that the glue that keeps the social order together will start to dissolve. Nevertheless, I would argue that because of the similarities between the two study sites the theory of the commons can be scaled up in this case from local-level to regional scale, given the presence of several other similar irrigation systems (see Tagseth, 2008). The characteristics allowing for the functioning of these common property regimes can be singled out: homogeneity of the local population, local monitoring, agreed sanctioning, as well as clear appropriation, provision and boundary rules. Population growth and climate change are the two main factors with which these systems have currently to come to terms with and adapt to; capitalist market integration, highly dependent on the quality of existing infrastructures, at the moment appears to be an emerging but distant challenge, which certainly requires future studies.

Gender and irrigation The common property regimes of Engaruka and Sibou appear to be relatively conflict-free because of their socio-economic homogeneity. However, elements of difference internal to common property regimes are often not taken in full account. Gender is one of them. Different gender identities in

61 fact equal different use and perceptions of water (Cleaver, 2000). How men and women relate to the local waterscape is determined by the local social norms that require negotiation. Local gender structures strongly reflect on the limited reproductive use that women can make of water in both locations. Whether explicitly regulated, as in Sibou, or informally normalized, as in Engaruka, women have no decision making power in relation to water management and, while carrying out strenuous agricultural practices, are often described as “helping” their husbands. While at first it could have appeared that women´s conditions in both case study locations were similar, as none could access water directly or take part in reparation and construction work, a local gender contract analysis brought to light the underlying local societal norms that formally or informally regulate women´s exclusion from water management. Likewise, a gender contract analysis made emerge how men and women adapt differently to climate variability and how the local context, both from an environmental and a social point of view, is decisive in shaping their response to shifting weather patterns. Women exclusion from water management and irrigation has been widely documented in almost all continents.2 While women´s role in water provision and management was recognized during the UN decade on water 1981- 90, through the decades the focus on women in the water sector at the international level has been inconsistent (Coles, 2009). This thesis adds to this existing body of knowledge. Women use water both in productive and reproductive ways and are hardworking farmers in the fields, but due to taboos and cultural norms (Coles, 2009; Joshi and Fawcett, 2005; Watson et al., 1998) they are excluded from irrigation and from water management decision making arenas (Ahlers and Zwarteveen, 2009). Be it a large scale projects managed by engineers or smallholder irrigators, water as a means of agricultural production is in the hand of men: hence the term “hydropatriar- chy”. The subordination of women, inherent in patriarchal societies, is therefore reflected in the water sector (Zwarteveen, 2008). Nevertheless, by employing the concept of local gender contract outside its originating context, this thesis shows that gender relations are not static, but rather have been undergoing small but significant renegotiations, especially in relation to the advent of cash cropping and horticulture. Here again, due to their implicit role of head of the household and farmer, men have started to take advantage of their position to capitalize on the introduction of new remunerative crops. This gender bias is confirmed by the fact that, after

2 Some examples: Turkey by Harris, 2008; in the Andes by Zwarteveen et al., 2010; Caretta et al., 2015; in Nepal by Zwarteveen et al., 2010; in India by Coles, 2009 and Joshi and Faw- cett, 2005; in Sudan by Coles, 2009; in East Africa by Watson et al., 1998; Njuki et al., 2014; in Swaziland by Peter et al., 2008; in Zambia and Ghana by Van Koppen et al., 2013; in Can- ada by Anderson et al., 2013; in England by Strang, 2005; in Brasil by Ferreira and de Moraes, 2015.

62 decades of studies showing women´s exclusion from the provision of extension service, it is still more probable for trial plots to be men´s plots rather than women´s (Gĩthĩnji et al., 2014; O´Sullivan et al., 2014). This condition is closely linked to the fact that women in Africa own only 15% of the land on average (5% in Kenya) and consequently receive just 10% of the overall farm credits. Therefore, the yields of women farmers are generally lower than men, independently if they are heads of household or sharing cultivation responsibilities with their husbands (Ransom and Bain, 2011). This situation is also due to their crops choice. Women tend to choose food crops, not just to feed their family, but also because they cannot access credit to acquire commercial crops seedlings and seeds. Accordingly, in Si- bou and Engaruka women continue to play a pivotal role in subsistence farming, while the extra burden of weeding and often harvesting their husbands’ crops for sale is added to their tasks. It should, however, be noted that without intercropping, mainly done by women, soil fertility would not be ensured, as it may risk to be depleted by the mono-cropping required for cash crops. While men are taking most advantage from the dawning capitalist market integration, it is important to bear in mind that without women´s maintenance of soil fertility and food production for their families, this shift to a mixed agricultural production – subsistence and market oriented – would not have been possible. Moreover, both the old and the emerging new gender division of labor are mirrored in women´s and men´s strategies to adapt to increased climate variability. Men tend to invest most of their time and energy in crops for sale, while women differentiate the risk by mixing food crops with different growing cycles. The latter strategy is an insurance against abrupt and unexpected weather events. Even in case of too much or too little rain some of the women´s crops will survive, while men´s crops will only develop well and therefore be attractive to wholesalers in a certain rain regime, making them more vulnerable to the increased climate variability expected (IPCC, 2014). Women´s burden in agriculture is being exacerbated by the introduction of new crops as they are generally expected to weed and harvest also their husbands´ crops. The returns are often meager for women. Men often contribute to school expenses for children, but they also spend money on leisure activities instead of reinvesting them in the welfare of their families. Conse- quently, women do not only have to keep on providing food for their families, but they must work extra or count on groups´ savings if an emergency occurs. Gender disparities have increased with the new local gender contract and given women´s exclusion from water management, it can be envisioned that this social order will not be sustainable in the long run. Of the approaches put in place by the New African Green Revolution few draw inspiration from small scale practices – e.g. intercropping, agroforestry, composting – which are closer to women´s farming. On the contrary, capital intensive strategies and privatization of productive means – e.g. seeds, ferti-

63 lizers – are promoted in an attempt to implement supply chain management and consequent commercialization of African agricultural produce (Moseley et al., 2015). Women, as the cheapest source of labor and the least engaged group, risk becoming the losers in this process of globalization of African agriculture (Tacoli et al., 2014). In fact, in the eventuality of new irrigation technologies being adopted or new crops being taken up, women will often find themselves in a disadvantaged position and will not be able to profit from these new changes (see also Ferreira and de Moraes, 2015; Joshi and Fawcett, 2005). A review made by the World Bank or irrigation projects put in place after 1997 highlighted that only 23% addressed the needs of female farmers and only 11% had special provisions for the poor (World Bank, 2002 in Njuki et al., 2014). The low adoption rate of treadle pumps by women (10%) in a project in Kenya and Tanzania was, for instance, due to the fact that the pumps required the use of legs and showing thighs is considered inappropriate for women (Njuki et al., 2014). This example shows how the introduction of new irrigation technologies tends to be biased towards men (Njuki et al., 2014; van Koppen et al., 2013) not only because they are taken as the physical reference, but also because it is erroneously assumed that the household is a unitary model where both partners share the same benefits from development intervention. For instance, moving from rain fed agriculture to irrigation can reduce women´s access to agriculture (Whitehead, 1998); moving from animal husbandry and self-sufficiency farming to cash cropping through irrigation does not ensure the improvement of living standard for all members of the family, but might rather in- duce a more marked exploitation of women´s work (Harris, 2008) or the inclusion of men into what previously was women´s realm, as high-value horticulture, might push women out of their productive domain (van Koppen et al., 2013). In Swaziland, for example, the introduction of an irrigation scheme made women lose their cash crop cotton against sugar cane that, thanks to the available irrigation, became men´s crop. In this case the irrigation scheme did not improve the food security of any family and actually worsened the living standards of female headed households (Peter et al., 2008). Women´s sense of alienation from decision making has clearly emerged in my ethnographical work. Women are aware that their stand is not taken into account by the local water users’ associations. They might also be afraid to speak up due to illiteracy or to the sole male composition of water committees where addressing other people is seen as a male trait and is perceived as negative in women (Ahlers and Zwarteveen, 2009). On the other hand, the participation of women in water users´ associations or the implementation of “women friendly” irrigation technologies, as piped water on premises to avoid long walks, does not alone ensure the empowerment of women or the improvement of their living standards. For instance, their inclusion in meetings might result in the reproduction of gender structures with women serv-

64 ing food to men. There is no easy solution. That is why it is important, as this study does, to analyze local conditions and the underlying social norms to understand which measures should be taken to facilitate women´s inclusion in water management and to give them a fairer share of productive and reproductive burdens. This study shows that the gender division of labor is being re-negotiated and the local gender contract is also being bargained at the family level. While the new local gender contract does not entail a more equal division of labor, there are signs that a new gender contract could emerge in the future. Some women have started to challenge the current local gender contract by washing clothes in the canals and by controlling the income generated through petty trading. These findings show that whether external development interventions are planned and implemented, they must not disregard the commonness of patriarchal gender structures and their influence on everyday irrigation management. There is in fact a risk in overlooking these relational and evolving dimensions of farming. To benefit the whole local community, investments in agricultural intensification should be gender sensitive and avoid reproducing preconceived notions and assumptions about male head of the family and intra-household unitary model.

Landscape investments The expansion of irrigated land in Africa from 12.6 to 20 million ha by 2015 has been supported by FAO and NEPAD since 2002 (You et al., 2010). Con- sidering that irrigated land in Sub Saharan Africa makes up only for 3.5% of the region´s land but produces 25% of the agricultural output (Shah et al., 2013), it is easier to see why an expansion of irrigation can trigger economic growth and enhance food security. The type of irrigation that is foreseen entails precision agriculture, motor pumps or drip irrigation that will ensure water efficiency and productivity (Shah et al., 2013; Bossio et al., 2010). However, this type of infrastructure and landscape investments can hardly be afforded by smallholders lacking access to financing (Ransom and Bain, 2011; Dorward et al., 2005) or access to an off farm income decisive for extra capital accumulation to be re-invested into irrigated agriculture (Cous- ins, 2013). This prospective type of agriculture appears to be rather directed towards large farms that can bear the costs of these new technologies, while the average African farm is less than 2ha. Commercialization of African agriculture through large farms of at least 10ha is, in fact, seen as the way to pursue sustained economic growth. Big commercial farms will have access to financing, logistics, marketing, storage and technology and will make them successful. In fact, large scale farming is successful thanks to specialization

65 in high value products for exports as tea, fruits, horticulture or flowers (Col- lier and Dercon, 2014). In this sense, large scale irrigation is juxtaposed with what is believed to be traditional – and implicitly backward – smallholder irrigation farming (You et al., 2010; Plusquellec, 2009). The mix between shifting cultivation, semi-permanent fields, household fields and riverine cultivation attest the complexity of these systems that have withstood and adapted to centuries of socio-ecological changes (see also Davies et al., 2014). In support of this statement, it has been seen that small farms are “more efficient than large farms in countries with a surplus in labor but a scarcity of land and capital” (Diao et al, 2007, p. 39). For instance, in Ghana economic growth has happened in the last 20 years mainly through smallholder farming (Wiggings, 2009). Additionally, large scale irrigation projects have not proven successful in bringing about food security (Shah et al., 2013; You et al., 2010; Hu- chon and Maisonhaute, 2010). Rather, poverty appeared to be more prevalent in these schemes (Namara et al. 2010). Landscape investments in the two study sites, as in the other regional examples of smallholder irrigation farming (i.e. Watson, 2009; Tenge, 2004), have been possible thanks to a profound knowledge of the water, a deep understanding of local environmental conditions and a definite preparedness for adapting to changes. Large scale projects often do not draw any lessons – whether social, practical or ecological – from the experiences of small scale common property regimes and have recurrently failed (on Kenya see e.g. Davies et al., 2014; Östberg and Caretta, forthcoming). On the contrary, the implementation of such schemes should ideally be compatible with and add to already existing local irrigation systems, while challenging their discriminatory practices against women.

Conclusion Against the background of the current debate on African agriculture, this thesis highlights the potentials of contemporary smallholder irrigation systems to adapt to ongoing socio-ecological changes through common property regimes that, while not gender inclusive, avoid conflict generated by shifting water supply and heightened climate variability. Often described as traditional and low in productivity, it is argued that African yield gaps should be closed by investing in the creation of new and the expansion of existing irrigation systems (e.g. AGRA, 2014 and 2013). By analyzing current agricultural practices and gender relations in two communities where irrigation has been undergoing for hundreds of years, this study emphasizes how research at the local level can help to integrate local agricultural practices and gender sensitive measures in the planning of irrigation projects.

66 By studying how local management works, how conflicts are adverted through common property regimes and how these systems adapt to socio- ecological changes, this thesis shows that when external interventions are implemented to improve agricultural productivity they should build upon and be integrated with existing local agricultural knowledge. By exploring the factors behind the consistent exclusion of women from the water sector, this study stresses that local gender relations should not be overlooked, as they are often unequal. Assessing existing social relations is fundamental in order to understand how new infrastructures can be equally beneficial to the overall population and to suggest gender sensitive measures in the planning of irrigation projects. Finally, by studying the role of assistants in the process of situated knowledge production and by making use of member checking techniques, this thesis emphasizes the importance of reflexive analysis to ensure validity in cross-cultural, cross-language research.

67 68 Future studies

Examining the waterscape of East Africa smallholder irrigation farming has brought forward several questions both at a conceptual level and on the future of African agriculture. I briefly summarize them in this section. First of all, this thesis has shown the adaptability and flexibility of these irrigation systems to fluctuating water supply and changing weather patterns. These common property regimes have worked for centuries, the same cannot be said for national initiatives put in place, especially in Marakwet, to increase agricultural production. In 1979 the Kerio Valley Development Au- thority (KVDA) was established with the aim to trigger major transfor- mations, including new permanent and improved irrigation infrastructures, a railway line for export, improved agricultural extension service, etc. (Dietz et al., 1987). The establishment in 1986 of a tree nursery has so far been the only successful project. A demonstration plot was abandoned due to the conflict in the area, the cement canals constructed were also abandoned (Davies et al., 2014). A project to plant 20 acres with maize was given up in 2005 after only two seasons. In 2007 a project to grow sugar cane was discontin- ued because of lacking transportation to urban areas. In 2008 about 125 acres of bush, 45 minutes’ walk from Tot Centre, was cleared in keew to produce seeds for Kenya Seeds Company. Water was never lead onto the site, which members of the Kaapisyooy clan eventually started to cultivate as communal land. In 2011 rice was planted along the Embobut riverbanks and KVDA invested in a rice shelling facility in Tot Center. In 2013 it was not in operation and no one in Sibou cultivated rice. Why did all of these projects fail? Can these dynamics be connected to the failure of the Green Revolution in Africa and to the inconsistent strategies of the Kenyan Government? Some of these questions will be explored in a forthcoming publication (Östberg and Caretta, forthcoming). Secondly, as explored in the discussion, it is debated whether the concept of common property regimes can be applied outside small scale systems (see Cox et al., 2010; Tucker et al., 2007; Cinner and McClanahan, 2006; Berkes et al., 2006). What will happen if some external factors will jeopardize existing systems? Market integration is certainly one external driver that will play an increasingly important role for smallholder irrigation farming communities in East Africa provided that road infrastructure will also develop. Much of this improvement will depend on government policies (Berkes et al.,

69 2006). The case of Engaruka, where an increasing amount of young farmers is cultivating horticultural products for sale and where tenure arrangements are evolving, with the government gearing up to register land, is very relevant and should be studied longitudinally during a 10-15 years’ period. Mar- ket integration in the case of Sibou is already happening particularly thanks to M-Pesa, the Safaricom system of money transfer through cellphone. Cell- phone distribution is equal among men and women and women make wide use of M-Pesa as well. However, studies are lacking on the gendered impacts of the usage of this technology and the increasing number of agricultural smartphone applications (Duncombe, 2015). And, again it will be worthwhile studying the use of these technologies longitudinally (10-15 years) in order to see whether they have effectively triggered market integration and changed crop routines or if the lack of infrastructure has hampered such developments. Third, an irrigation scheme is already in place and functioning in the former communal land of Sibou. Financed between 2012 and 2015 by the Ca- nadian Red Cross in collaboration with the Kenyan Red Cross, this project is advertised as a peace building initiative between the Marakwet and the East Pokot communities as the land to be irrigated is situated on both sides of the Kerio River. 250 acres in Marakwet and 250 acres in East Pokot are irrigated through sprinklers with water diverted from Embobut River. While at first it was stated that the least well off families will be provided with land, this has effectively not happened as the land is in the Kaapisyooy communal land. This clan, who is among the most well off groups in the area, has claimed ownership over the project. The objective of the Red Cross, which provided hybrid seeds, pesticides and fertilizers, was that farmers should not any longer have to depend on the rains: they will be able to cultivate remunerative cash crops – onions, green gram, cowpeas – all year around. By January 2015 a first season of maize cultivation had successfully been concluded and the resident manager was enthusiastic: ”You should have seen the yellow landscape. It was wonderful.” (field notes). However, at the onset of the second season he was disappointed that some fields were not cultivated and farmers were planting a mix of crops (millets, tomatoes and watermelons, in addition to maize), which did not, he argued, make economic sense. Kenya Seed was in fact planning to take over the marketing support, but they feared that the quantities would be too low to be commercially viable if farmers would not all be cultivating maize. Will this project survive? Or will farmers make use of sprinklers to cultivate the crops that they prefer and they know they can market directly? How to reconcile centuries old smallholder irrigation farming knowledge with a technologically advanced medium scale irrigation system? How to resolve the conflict between investors and authorities’ short-term expectations on the one hand and local practices and experiences on the other? These questions can be explored by studying this project.

70 Svenska sammanfattning

Den här avhandlingen undersöker de lokala landskap som skapas av småska- ligt bevattnade jordbrukssystem på två platser i Östafrikas torrområde. Stu- dien lägger fokus på tre aspekter: institutionella arrangemang, genusrelation- er och landskapsinvesteringar. Avhandlingen tar utgångspunkt i en reflexiv analys av forskning i en fler- kulturell och flerspråkig miljö, med särskilt fokus på betydelsen av fältassi- stenter och tolkar, samt på återrapportering av preliminära resultat som en metod för att säkerställa validiteten. De flexibla bevattningssystemen i Sibou, Kenya, och Engaruka, Tanzania, gör det möjligt för jordbrukarna att styra vattenflödet och utvidga eller minska den odlade arealen beroende på säsongsnederbörden. Jordbrukarna löser konflikter om vattnet genom ett decentraliserat system för gemensam resursförvaltning. Bevattningsrättigheterna omförhandlas löpande och an- passas till vattentillgången. Vatten ses som en gemensam nyttighet och för- valtningen är präglad av en ömsesidig vilja att undvika konflikter. Förvalt- ningen sker genom delaktighet och delad information om vattenrättigheterna. Avhandlingen visar dock att vattenförvaltningen är ett privilegium nästan enbart förbehållet av män. Strikta patriarkala normer reglerar kontrollen av vattnet och utesluter i praktiken kvinnor från förvaltningen av bevattningssystemen. Kontrollen över vattenanvändningen för produktiva ändamål är en manifestation av manlighet. Samma könsobalans har under de senaste de- cennierna uppstått när män har ökat sitt engagemang i jordbruket genom att odla avsalugrödor. Kvinnorna kan på grund av sin underordnade ställning inte utnyttja denna möjlighet till inkomstdiversifiering. Odlingen av avsa- lugrödor har snarare ökat arbetsbelastningen för kvinnorna, som redan pro- ducerar huvuddelen av livsmedelsgrödorna för hushållskonsumtion. Förutom odling av livsmedelsgrödor måste de nu rensa ogräs och skörda de grödor som deras män säljer. Denna klyfta mellan könen inom jordbruket avspeglas även i mäns respektive kvinnors attityder till den ökade klimatvariationen. Kvinnor samodlar flera grödor för att försäkra sig om att säkerställa hus- hållskonsumtionen, medan män istället sår flera omgångar avsalugrödor när regnet så tillåter. Trots att de är underordnade män, är kvinnornas insatser i praktiken helt avgörande för underhållet av bevattningssystemen och för att bibehålla genom samodling den bördighet som möjliggör odling av avsa- lugrödor.

71 Sammanfattningsvis belyser avhandlingen anpassningsförmågan hos bevattningssystem som förvaltas av småbrukare, som visserligen inte inklude- rar kvinnor, men som lyckas undvika de konflikter som uppstår genom för- ändringar i vattenförsörjningen och ökad klimatvariation. För att kunna bedöma hållbarheten och effektiviteten hos ett bevattningssystem, måste forskningen utföras på lokal nivå. Genom att studera anpass- ningen till socio-ekologiska förändringar och hur konflikter undviks med hjälp av gemensamma förvaltningssystem, ger denna studie insikter som är viktiga både för planering av nya bevattningssystem och för att rusta upp eller utvidga äldre system. Genom att undersöka faktorerna bakom den kon- sekventa uteslutningen av kvinnor från förvaltningen av bevattningssystemen, liksom deras underordnade roll i jordbruksproduktionen, varnar denna studie också för att reproducera denna diskriminering i planeringen av be- vattningsprojekt.

Nyckelord: bevattningsjordbruk, lokala könskontrakt, markbundet kapital, resursförvaltning, torrområden, feministisk kunskapsteori, Kenya, Tanzania.

72 References

Acker, S. (2000). In/out/side: Positioning the researcher in feminist qualitative research. Resources for Feminist Research, 28(1-2), 189–208 Adams, W. M, Watson, E., & Mutiso, S. (1997). Water, Rules and Gender: Water Rights in an Indigenous Irrigation System, Marakwet, Kenya. Development and Change, 28(4), 707–730. http://doi.org/10.1111/1467-7660.00061 Adams, W. M., & Anderson, D. M. (1988). Irrigation before Development: Indige- nous and Induced Change in Agricultural Water Management in East Africa. Adams, W. M., & Watson, E. E. (2003). Soil erosion, indigenous irrigation and environmental sustainability, Marakwet, Kenya. Land Degradation & Develop- ment, 14(1), 109–122. http://doi.org/10.1002/ldr.528 African Affairs, 87(349), 519–535. http://doi.org/10.2307/722892 AGRA – Alliance for a Green Revolution in Africa. (2013). Africa agriculture sta- tus report: focus on staple crops. Nairobi, Kenya. Retrieved from http://agra- alliance.org/resources/resources/ AGRA – Alliance for a Green Revolution in Africa. (2014). Africa agriculture sta- tus report: climate change and smallholder agriculture in Sub Saharan Africa. Nairobi, Kenya. Retrieved from http://agra-alliance.org/resources/resources/ Agrawal, A. (2001). Common Property Institutions and Sustainable Governance of Resources. World Development, 29(10), 1649–1672. http://doi.org/10.1016/S0305-750X(01)00063-8 Ahlers, R., & Zwarteveen, M. (2009). The water question in feminism: water control and gender inequities in a neo-liberal era. Gender, Place & Culture, 16(4), 409– 426. http://doi.org/10.1080/09663690903003926 Anderson, K., Clow, B., & Haworth-Brockman, M. (2013). Carriers of water: abo- riginal women’s experiences, relationships, and reflections. Journal of Cleaner Production, 60, 11–17. http://doi.org/10.1016/j.jclepro.2011.10.023 Archambault, C. (2011). Ethnographic empathy and the social context of rights: “rescuing” Maasai girls from early marriage. American Anthropologist, 113(4), 632–643. http://doi.org/http://dx.doi.org/10.1111/j.1548-1433.2011.01375.x Årlin, C., Börjeson, L. & Östberg, W. (in press). Participatory Checking and the Temporality of Landscapes: Increasing Trust and Relevance in Qualitative Re- search. In D. Stump and C. Isendahl. (Eds.) Handbook of Historical Ecology and Applied Archaeology, Oxford: Oxford University Press. Bahiigwa, G., Mdoe, N., & Ellis, F. (2005). Livelihoods Research Findings and Agriculture-led Growth. IDS Bulletin, 36(2), 115–120. http://doi.org/10.1111/j.1759-5436.2005.tb00207.x Bardhan, P. (2000). Irrigation and Cooperation: An Empirical Analysis of 48 Irriga- tion Communities in South India. Economic Development and Cultural Change, 48(4), 847–865. http://doi.org/10.1086/452480 Bayliss-Smith, T. (2009). The meaning of ditches: interpreting the archeological record from New Guinea using insights from ethnography. In Denham, T. P., Iriarte, J. & Vrydaghs, L. (Eds.) Rethinking Agriculture: Archaeological and

73 Ethnoarchaeological Perspectives (126-48). Left Coast Press, Walnut Creek CA. Berkes, F. (2006). From community-based resource management to complex systems: the scale issue and marine commons. Ecology and Society, 11(1), 45. Billo, E., & Hiemstra, N. (2013). Mediating messiness: expanding ideas of flexibility, reflexivity, and embodiment in fieldwork. Gender, Place & Culture, 20(3), 313–328. http://doi.org/10.1080/0966369X.2012.674929 Blaikie, P. M., & Brookfield, H. C. (1987). Land Degradation and Society. Taylor & Francis. Bollig, M., & Österle, M. (2007). “We turned our enemies into baboons”: warfare, ritual and pastoral identity among the Pokot of Northern Kenya. In Rao A., Bol- lig M. & Böck M. (Eds.) The Practice of War: Production, Reproduction and Communication of Armed Violence, 23. Bonini, N. (2006). The pencil and the shepherd´s crook. Ethnography of Maasai education. Ethnography and Education, 1(3), 379–382. http://doi.org/10.1080/17457820600837036 Börjeson, L. (2004). A history under siege: intensive agriculture in the Mbulu High- lands, Tanzania, 19th century to the present. Dept. of Human Geography, Stockholm University. Börjeson, L. (2014). The antithesis of degraded land: toward a green conceptualization of landesque capital. In Håkansson, N. T., & Widgren, M. (Eds.) Landesque Capital: The Historical Ecology of Enduring Landscape Modifications (251- 268). Left Coast Press, Walnut Creek CA. Boserup, E. (1970). Women in Economic Development. London: Earthscan. Bossio, D., Geheb, K., & Critchley, W. (2010). Managing water by managing land: Addressing land degradation to improve water productivity and rural livelihoods. Agricultural Water Management, 97(4), 536–542. http://doi.org/10.1016/j.agwat.2008.12.001 Brookfield, H. (2001). Intensification, and Alternative Approaches to Agricultural Change. Asia Pacific Viewpoint, 42(2-3), 181–192. http://doi.org/10.1111/1467- 8373.00143 Budds, J., & Hinojosa-Valencia, L. (2012). Restructuring and rescaling water governance in mining contexts: the co-production of waterscapes in Peru. Water Al- ternatives, 5(1), 119–137. Caretta, M. A. (2014). Hydropatriarchies and landesque capital: a local gender contract analysis of two smallholder irrigation systems in East Africa. The Geo- graphical Journal, http://doi.org/10.1111/geoj.12102 Caretta, M. A. (2015a). Managing variability and scarcity. An analysis of Engaruka: A Maasai smallholder irrigation farming community. Agricultural Water Man- agement, 159, 318–330. http://doi.org/10.1016/j.agwat.2015.05.015 Caretta, M. A. (2015b). Situated knowledge in cross-cultural, cross-language research: a collaborative reflexive analysis of researcher, assistant and participant subjectivities. Qualitative Research, 4, 489–505. http://doi.org/10.1177/1468794114543404 Caretta, M. A., & Börjeson, L. (2015). Local gender contract and adaptive capacity in smallholder irrigation farming: a case study from the Kenyan drylands. Gender, Place & Culture, 22(5), 644–661. http://doi.org/10.1080/0966369X.2014.885888 Caretta, M. A., Cadena Montero, G.Y., Sulbarán, L., Sandoval, R. (2015). ¿La revolución tiene cara de campesina?” Un caso de estudio de la participación activa de las mujeres en el riego del páramo venezolano. Revista Latino

74 Americana de Genero y Geografia, Vol. 6 , n. 2. pp. 3- 23. doi: 10.5212/Rlagg.v.6.i2.0001 Caretta, M.A., & Vacchelli, E. (In press). Re-thinking the boundaries of the focus group. A reflexive analysis on the use and legitimacy of group methodologies in qualitative research. Sociological Research Online. Caretta, M. A., Westerberg, L.-O., Börjeson, L., & Östberg, W. (2014). Labour, climate perceptions and soils in the irrigation systems in Sibou, Kenya & Engaruka, Tanzania. Department of Human Geography, Stock- holm University. Carpiano, R. M. (2009). Come take a walk with me: The “Go-Along” interview as a novel method for studying the implications of place for health and well-being. Health & Place, 15(1), 263–272. http://doi.org/10.1016/j.healthplace.2008.05.003 Chambers, R. (2002). Participatory workshops : a sourcebook of 21 sets of ideas and activities. London: Earthscan. Chilvers, J. (2009). Deliberative and Participatory Approaches in Environmental Geography. In N. Castreeessor, D. Demerittessor, D. L. Director, & B. R. Chair (Eds.), A Companion to Environmental Geography (400–417). Wiley- Blackwell. Cho, J., & Trent, A. (2006). Validity in qualitative research revisited. Qualitative Research, 6(3), 319–340. http://doi.org/10.1177/1468794106065006 Cinner, J. E.e., & McClanahan, T. R. (2006). Socioeconomic factors that lead to overfishing in small-scale coral reef fisheries of Papua New Guinea. Environ- mental Conservation, 33 (01), 73–80. http://doi.org/10.1017/S0376892906002748 Cleaver, F. (2000). Moral Ecological Rationality, Institutions and the Management of Common Property Resources. Development and Change, 31(2), 361–383. http://doi.org/10.1111/1467-7660.00158 Coast, E. (2002). Maasai Socioeconomic Conditions: A Cross-Border Comparison. Human Ecology, 30(1), 79–105. http://doi.org/10.1023/A:1014567029853 Cockburn, C. (1988). Machinery Of Dominance: Women, Men, and Technical Know-How (Northeastern University Press). Northeastern. Coles, A. (2009). Domestic water provision and gender roles in drylands. HAGAR: Studies in Culture, Polity & Identities, 9(1). Collier, P., & Dercon, S. (2014). African Agriculture in 50 Years: Smallholders in a Rapidly Changing World? World Development. 63, 92– 101http://doi.org/10.1016/j.worlddev.2013.10.001 Cook, I. et al. (2005). Positionality – situated knowledge. In D. Sibley P. Jackson, D. Atkinson & N. Washbourne (Eds.) Cultural Geography: A Critical Dictionary of Key Ideas (16-26). I.B.Tauris. Cornwall, A., Harrison, E., & Whitehead, A. (2009). Gender Myths and Feminist Fables: The Struggle for Interpretive Power in Gender and Development. Wiley. Cousins, B. (2013). Smallholder Irrigation Schemes, Agrarian Reform and “Accu- mulation from Above and from Below” in South Africa. Journal of Agrarian Change, 13(1), 116–139. http://doi.org/10.1111/joac.12000 Cox, M., Arnold, G., & Tomás, S. V. (2010). A review of design principles for community-based natural resource management. Ecology and Society, 15(4), 38. Critchley, W. (1979). Chesongoch Agricultural Project 1977 -1979. Final Report. Chesongoch Catholic Mission.

75 Critchley, W. (1983). Agricultural Developments in Marakwet: Some Controversial Issues. In B. Kipkorir, R. Soper, J. Ssennyonga, (Eds.) Kerio Valley: Past, Pre- sent and Future (19-26). University of Nairobi: Institute of African Studies. Davies, M. I. J., Kipruto, T. K., & Moore, H. L. (2014). Revisiting the irrigated agricultural landscape of the Marakwet, Kenya: tracing local technology and knowledge over the recent past. Azania: Archaeological Research in Africa, 0(0), 1–38. http://doi.org/10.1080/0067270X.2014.979527 Diao, X., Hazell, P. B., Resnick, D., & Thurlow, J. (2007). The role of agriculture in development: Implications for Sub-Saharan Africa (Vol. 153). International Food Policy Research Institute (IFPRI). Dietz, T., van Haastrecht, A. and Moore, H. (1987). Endo, Mokoro, and Embobut locations, Elgeyo Marakwet District. Iten: Regional Development Research for the Arid and Semi-Arid lands Programme Elgeyo Marakwet. Djurfeldt, G., Aryeetey, E., & Isinika, A. C. (Eds.) (2011). African Smallholders: Food Crops, Markets and Policy. CABI. Djurfeldt, G., Holmen, H., Jirstrom, M., & Larsson, R. (Eds.) (2005). The African Food Crisis: Lessons from the Asian Green Revolution. CABI. Doolittle, W. E. (1984). Agricultural Change as an Incremental Process. Annals of the Association of American Geographers, 74(1), 124–137. http://doi.org/10.1111/j.1467-8306.1984.tb01438.x Doolittle, W. E. (2001). Learning to See the Impacts of Individuals. Geographical Review, 91(1/2), 423. Dorward, A., Kydd, J., & Poulton, C. (2005). Beyond Liberalisation: “Developmen- tal Coordination” Policies for African Smallholder Agriculture. IDS Bulletin, 36(2), 80–85. http://doi.org/10.1111/j.1759-5436.2005.tb00201.x Duncombe, R. (2015). Mobile Phones for Agricultural and Rural Development: A Literature Review and Suggestions for Future Research. European Journal of Development Research. http://doi.org/10.1057/ejdr.2014.60 Ellis, C.& Bochner, A.P. (2005). Autoetnography, personal narrative, reflexivity: researcher as subject. In Denzin, N.K., Lincoln, Y.S. (Eds.), The SAGE Hand- book of Qualitative Research (733-768), 3rd ed. SAGE. England, K. V. L. (1994). Getting Personal: Reflexivity, Positionality, and Feminist Research. The Professional Geographer, 46(1), 80–89. http://doi.org/10.1111/j.0033-0124.1994.00080.x Falconer, A. H. K., & Kawabata, H. (2002). Toward a more fully reflexive feminist geography. In P. Moss, P (Ed.), Feminist Geography in Practice: Research and methods. Oxford, Blackwell. Foley, J. A., Ramankutty, N., Brauman, K. A., Cassidy, E. S., Gerber, J. S., John- ston, M., … Zaks, D. P. M. (2011). Solutions for a cultivated planet. Nature, 478(7369), 337–342. http://doi.org/10.1038/nature10452 Forsberg, G. (1998). Regional variations in the gender contract: Gendered relations in labour markets, local politics and everyday life in Swedish regions. Innova- tion: The European Journal of Social Science Research, 11(2), 191–209. http://doi.org/10.1080/13511610.1998.9968561 Forsberg, G. (2010). Gender, geography and spatial practice. In B. Hermelin, and U. Jansson (Eds.) Placing human geography: Sweden through time and space. Stockholm: SSAG. Gibson, C. C. (2001). Forest resources: Institutions for local governance in Guate- mala. In Berger, L., Ostrom, E., Norgaard, R. B., Policansky, D., & Goldstein, B. D. (Eds.), Protecting the Commons: A Framework for Resource Management in the Americas (71–90), Island, Washington, DC, .

76 Gĩthĩnji, M. wa, Konstantinidis, C., & Barenberg, A. (2014). Small and Productive: Kenyan Women and Crop Choice. Feminist Economics, 20(1), 101–129. http://doi.org/10.1080/13545701.2013.878467 Håkansson, N. T. & Widgren, M. (2007). Labour and landscapes: the political economy of landesque capital in nineteenth century Tanganyika. Geografiska Annal- er. Series A. Physical Geography, 89(3), 233. Håkansson, N. T., & Widgren, M. (2014). Landesque Capital: The Historical Ecol- ogy of Enduring Landscape Modifications. Left Coast Press, Walnut Creek CA. Hall, J. (2002). The next generation: can there be a feminist geography without gender? The Great Lakes Geographer, 9(1), 20-27. Harris, L. M. (2008). Water Rich, Resource Poor: Intersections of Gender, Poverty, and Vulnerability in Newly Irrigated Areas of Southeastern Turkey. World De- velopment, 36(12), 2643–2662. http://doi.org/10.1016/j.worlddev.2008.03.004 Haug, R., & Hella, J. (2013). The art of balancing food security: securing availability and affordability of food in Tanzania. Food Security, 5(3), 415–426. http://doi.org/10.1007/s12571-013-0266-8 Hillbom, E. (2012). When water is from God: formation of property rights governing communal irrigation furrows in Meru, Tanzania, c. 1890–2011. Journal of Eastern African Studies, 6(3), 423–443. http://doi.org/10.1080/17531055.2012.696891 Hirdman, Y. (1990). The gender system : theoretical reflections on the social subordination of women. Uppsala: Maktutredningen. Hirdman, Y., Hartman, S. R., & Hartman, S. (1994). Women-from possibility to problem? Gender conflict in the welfare state-the Swedish model. Arbetslivs- centrum. Holmen, H. & Hyden, G. (2011). African agriculture: from crisis to development? In G. Djurfeldt, E. Aryeetey, & A.C. Isinika (Eds.) African Smallholders: Food Crops, Markets and Policy (23-34). CABI. HRW – Human Rights Watch. (2002). Time of transition. Retrieved from http://www.un.org/womenwatch/daw/ws2009/ Huchon, J., Maisonhaute, J. (2010). Local irrigation projects in north Western Ken- ya, conceptual frameworks and development practices: the missing links. In Ca- las, B., Martinon, C.A.M., (Eds.). Shared Waters, Shared Opportunities: Hy- dropolitics in East Africa. African Books Collective. IPCC – Intergovernmental Panel on Climate Change. (2014). Fifth Assessment Re- port: What’s in it for Africa? - Executive summary. Retrieved from http://reliefweb.int/report/world/ipcc-s-fifth-assessment-report-what-s-it-africa- executive-summary Jackson, P. (1989). Maps of meaning: An introduction to cultural geography. Psy- chology Press. Jakobsen, H. (2012). Focus groups and methodological rigour outside the minority world: making the method work to its strengths in Tanzania. Qualitative Re- search, 12(2), 111–130. http://doi.org/10.1177/1468794111416145 Joshi, D., & Fawcett, B. (2005). The role of water in an unequal social order in In- dia. In A. Colesand T. Wallace (Eds.) Gender, Water and Development (39-56). Basingstoke, GB, BERG. Kalabamu, F. (2005). Changing gender contracts in self-help housing construction in Botswana: the case of Lobatse. Habitat International, 29(2), 245–268. http://doi.org/10.1016/j.habitatint.2003.09.005

77 Kannen, V. (2013). Pregnant, privileged and PhDing: exploring embodiments in qualitative research. Journal of Gender Studies, 22(2), 178–191. http://doi.org/10.1080/09589236.2012.745681 Kedir, A. M. (2003). Rural poverty report 2001: the challenge of ending rural poverty edited by the IFAD. (Oxford: Oxford University Press, 2001, pp. 266). Journal of International Development, 15(5), 667–668. http://doi.org/10.1002/jid.1013 Kherallah, M., Delgado, C. L., Gabre-Madhin, E., Minot, N., & Johnson, M. (2000). The Road Half Traveled: Agricultural Market Reform in Sub-Saharan Africa. International Food Policy Research Institute (IFPRI). KNBS – Kenya National Bureau of Statistics. (2009). Kenya Population and Hous- ing Census, vol. IB. August 2010. Nairobi: Government Press. Komakech, H. C., Van Der Zaag, P., & Van Koppen, B. (2012). The dynamics between water asymmetry, inequality and heterogeneity sustaining canal institutions in the Makanya catchment, Tanzania. Water Policy, 14(5), 800. Koppen, B. van, Hope, L., & Colenbrander, W. (2013). Gender Aspects of Small- scale Private Irrigation in Africa. IWMI. Krueger, R. A., & Casey, M. A. (2009). Focus groups : a practical guide for applied research. Thousand Oaks, Calif.: Sage Publications. Laulumaa, V. (2006). Estimation of the population of ancient Engaruka - a new approach. Azania: Archaeological Research in Africa, 41(1), 95-102. http://doi.org/10.1080/00672700609480436 Lawrence, P. (2010). The African Tragedy. International and National Roots. In Padayachee, V. (Ed.) The Political Economy of Africa. New York: Routledge, (19-38). Leeuw, S. de, Cameron, E. S., & Greenwood, M. L. (2012). Participatory and community-based research, Indigenous geographies, and the spaces of friendship: A critical engagement. The Canadian Geographer / Le Géographe Canadien, 56(2), 180–194. http://doi.org/10.1111/j.1541-0064.2012.00434.x Lindeborg, A.-K. (2012). Where gendered spaces bend : the rubber phenomenon in northern Laos. Uppsala: Kulturgeografiska institutionen, Uppsala Universitet. Loftus, A. (2009). Rethinking Political Ecologies of Water. Third World Quarterly, 30(5), 953–968. http://doi.org/10.1080/01436590902959198 Madriz, E. (2000). Focus groups in feminist research. Handbook of Qualitative Re- search, 2, 835–850. Maynard, M., & Purvis, J. (1994). Researching Women’s Lives From A Feminist Perspective. Taylor & Francis. McCabe, J. T. (2003). Sustainability and Livelihood Diversification among the Maa- sai of Northern Tanzania. Human Organization, 62(2), 100–111. McCabe, J. T., Leslie, P. W., & DeLuca, L. (2010). Adopting Cultivation to Remain Pastoralists: The Diversification of Maasai Livelihoods in Northern Tanzania. Human Ecology, 38(3), 321–334. http://doi.org/10.1007/s10745-010-9312-8 McDowell, L. (1999). Gender, Identity, and Place: Understanding Feminist Geog- raphies. University of Minnesota Press. McDowell, L., & Massey, D. (1984). A woman’s place? Geography Matters, 128– 147. Mitchell, D. (2003). Cultural landscapes: just landscapes or landscapes of justice? Progress in Human Geography, 27(6), 787–796. http://doi.org/10.1191/0309132503ph464pr Mitchell, D. (2008). New Axioms for Reading the Landscape: Paying Attention to Political Economy and Social Justice. In Political Economies of Landscape Change (29–50). Springer Netherlands.

78 Mitchell, D. (2012). They saved the crops: labor, landscape, and the struggle over industrial farming in Bracero-era California (Vol. 10). University of Georgia Press. Mohanty, C. T. (1984). Under Western eyes: Feminist scholarship and colonial discourses. Boundary 2, 12, 333–358. Moore, H. L. (1986). Space, Text, and Gender: Anthropological Study of the Marakwet of Kenya. CUP Archive. Moore, H. L. (2011). Still life: hopes, desires and satisfactions.Cambridge: Polity Press. Ferreira, A., and Moraes, A. F. J de. (2015). Advances and setbacks in women’s participation in water management in Brazil. In Buechler, S., & Hanson, A.-M. S. (Eds.) A Political Ecology of Women, Water and Global Environmental Change. Routledge. (77-96) A Political Ecology of Women, Water and Global Environmental Change, 77. Morgan, D. L. (1997). The Focus Group Guidebook. SAGE. Moschitz, H., & Home, R. (2014). The challenges of innovation for sustainable agriculture and rural development: Integrating local actions into European policies with the Reflective Learning Methodology. Action Research, 12(4), 392– 409. http://doi.org/10.1177/1476750314539356 Moseley, W., Schnurr, M., & Kerr, R. B. (2015). Interrogating the technocratic (ne- oliberal) agenda for agricultural development and hunger alleviation in Africa. African Geographical Review, 34(1), 1–7. http://doi.org/10.1080/19376812.2014.1003308 Namara, R. E., Hanjra, M. A., Castillo, G. E., Ravnborg, H. M., Smith, L., & Van Koppen, B. (2010). Agricultural water management and poverty linkages. Agri- cultural Water Management, 97(4), 520–527. http://doi.org/10.1016/j.agwat.2009.05.007 Nash, K. (1994). The Feminist Production of Knowledge: Is Deconstruction a Prac- tice for Women? Feminist Review, (47), 65–77. http://doi.org/10.2307/1395254 NBS – National Bureau of Statistics, (2002). Population data. Tanzania. Retrieved from http://www.nbs.go.tz/ NBS – National Bureau of Statistics, (2012). Population data. Tanzania. Retrieved from http://www.nbs.go.tz/ Negin, J., Remans, R., Karuti, S., & Fanzo, J. C. (2009). Integrating a broader notion of food security and gender empowerment into the African Green Revolution. Food Security, 1(3), 351–360. http://doi.org/10.1007/s12571-009-0025-z Njuki, J., Waithanji, E., Sakwa, B., Kariuki, J., Mukewa, E., & Ngige, J. (2014). A Qualitative Assessment of Gender and Irrigation Technology in Kenya and Tan- zania. Gender, Technology and Development, 18(3), 303–340. http://doi.org/10.1177/0971852414544010 Östberg, W. & Caretta, M.A. (Forthcoming) Kerio Valley, 1973-2013: A case study of Kenyan smallholder agriculture. Re-submitted to African Studies. Östberg, W. (2004) The Expansion of Marakwet Hill-Furrow Irrigation in the Kerio Valley of Kenya. In Widgren, M. & J. E.G. Sutton (Eds.), Islands of Intensive Agriculture in Eastern Africa: Past & Present. (19– 48). Oxford: James Currey. Östberg, W. (2014). Irrigated fields are wives: indigenous irrigation in Marakwet, Kenya. In Håkansson, N. T., & Widgren, M. (Eds.). Landesque Capital: The Historical Ecology of Enduring Landscape Modifications (197– 214). Left Coast Press, Walnut Creek CA. Ostrom, E. (1990). Governing the Commons: The Evolution of Institutions for Col- lective Action. Cambridge University Press.

79 Ostrom, E., & Gardner, R. (1993). Coping with Asymmetries in the Commons: Self- Governing Irrigation Systems Can Work. The Journal of Economic Perspec- tives, 7(4), 93–112. Ostrom, V. & Ostrom, E. (1970) A theory for institutional analysis of common pool problems. In G. Hardin and J. Baden. (Eds.) Managing the commons. San Fran- cisco, CA: W.H. Freeman. (157–172). O'Sullivan, M., Rao, A., Banerjee, R., Gulati, K. & Vinez, M. (2014). Levelling the field: improving opportunities for women farmers in Africa. Washington DC; World Bank Group. Peter, G., Simelane, N. O., & Matondo, J. I. (2008). Socioeconomic impacts of small-scale irrigation schemes on women in Swaziland. Physics and Chemistry of the Earth, Parts A/B/C, 33(8–13), 850–858. http://doi.org/10.1016/j.pce.2008.06.020 Pinkerton, E., & Weinstein, M. (1995). Fisheries that work: sustainability through community-based management. The David Suzuki Foundation. Vancouver. Plusquellec, H. (2009). Modernization of large-scale irrigation systems: is it an achievable objective or a lost cause. Irrigation and Drainage, 58(S1), 104-120. http://doi.org/10.1002/ird.488 Poulton, C., & Kanyinga, K. (2014). The politics of revitalising agriculture in Ken- ya. Development Policy Review, 32 (2), 151–172. http://doi.org/10.1111/dpr.12080 Ramazanoglu, C., & Holland, J. (2002). Feminist Methodology: Challenges and Choices (1st ed.). SAGE Publications Ltd. Ransom, E., & Bain, C. (2011). Gendering Agricultural Aid An Analysis of Whether International Development Assistance Targets Women and Gender. Gender & Society, 25(1), 48–74. http://doi.org/10.1177/0891243210392571 Rantalaiho, L., & Heiskanen, T. (1997). Gendered practices in working life. Mac- millan press. Reid, C., & Frisby, W. (2008). Continuing the Journey: Articulating Dimensions of Feminist Participatory Action Research (FPAR). In P. Reason & H. Bradbury, The SAGE Handbook of Action Research (pp. 93–105).United Kingdom: SAGE Publications Ltd. Risman, B. J. (2004). Gender as a Social Structure: Theory Wrestling with Activism. Gender and Society, 18(4), 429–450. Robertshaw, P. (1986). Engaruka Revisited: Excavations of 1982. Azania: Archaeo- logical Research in Africa, 21(1), 1–26. http://doi.org/10.1080/00672708609511365 Rosaldo, M. Z., Lamphere, L., & Bamberger, J. (1974). Woman, Culture, and Socie- ty. Stanford University Press. Rose, G. (1993). Feminism & Geography: The Limits of Geographical Knowledge. Univ. of Minnesota Press. Rose, G. (1997). Situating knowledges: positionality, reflexivities and other tactics. Progress in Human Geography, 21(3), 305–320. http://doi.org/10.1191/030913297673302122 Rosegrant, M. W., Ringler, C., & Zhu, T. (2009). Water for Agriculture: Maintain- ing Food Security under Growing Scarcity. Annual Review of Environment and Resources, 34(1), 205–222. http://doi.org/10.1146/annurev.environ.030308.090351 Sassoon, H. (1967). New Views on Engaruka, Northern Tanzania. The Journal of African History 8(02): 201–217.

80 Setten, G. (2013). ’Til fots’ – et metodeblikk på landskap og praksis. In Førde, A. et al. (Eds.) Å finne sted. Metodologiske perspektiver på stedsanalyser. Tapir Aka- demisk Forlag, Trondheim. Shah, T., Verma, S., & Pavelic, P. (2013). Understanding smallholder irrigation in Sub-Saharan Africa: results of a sample survey from nine countries. Water In- ternational, 38(6), 809–826. http://doi.org/10.1080/02508060.2013.843843 Spear, T. (1993). Being “Maasai”, but not “people of cattle”: Arusha agricultural Maasai in the nineteenth century. In T. Spear and R. Waller (Eds.) Being Maasai (120–136), James Currey Ltd., London. Spencer, Paul (1993) Becoming Maasai, Being in Time. In T. Spearand R. Waler(Eds.) Being Maasai (140-156). London: James Currey, Strang, V. (2005). Taking the waters. In A. Colesand T.Wallace, (Eds.) Gender, Water and Development. Basingstoke, GB, BERG. (21-38). Stump, D. (2006). The development and expansion of the field and irrigation systems at Engaruka, Tanzania. Azania: Archaeological Research in Africa, 41(1), 69–94. http://doi.org/10.1080/00672700609480435 Stump, D. (2010). “Ancient and backward or long-lived and sustainable?” The role of the past in debates concerning rural livelihoods and resource conservation in Eastern Africa. World Development, 38(9), 1251-1262. http://doi.org/10.1016/j.worlddev.2010.02.007 Sutton, J. E. G. (1984). Irrigation and Soil-Conservation in African Agricultural History. The Journal of African History, 25(01), 25–41. http://doi.org/10.1017/S0021853700022544 Sutton, J. E. G. (1986). The Irrigation and Manuring of the Engaruka Field System. Azania: Archaeological Research in Africa, 21(1), 27–51. http://doi.org/10.1080/00672708609511366 Sutton, J. E. G. (1998). Engaruka. Irrigation agriculture in the Northern Tanzanian Rift Valley before the Maasai era. Azania, 33, 1–37. Sutton, J. E. G. (2000). Engaruka: An irrigation agricultural community in northern Tanzania before the Maasai. British Institute in Eastern Africa. Retrieved from http://www.bcin.ca/Interface/openbcin.cgi?submit=submit&Chinkey=194887 Sutton, J. E. G. (2004). Engaruka: The success and abandonment of an integrated irrigation system in an arid part of the Rift Valley, c. 15th to 17th centuries. In Widgren, M. & Sutton, J. E. G. (Eds.) Islands of Intensive Agriculture in East- ern Africa (114–132). James Currey Publishers. Swyngedouw, E. (1999). Modernity and hybridity: Nature, regeneracionismo, and the production of the Spanish waterscape. Annals of the Association of Ameri- can Geographers, 89(3), 443. Tacoli, C., Polack, E. Nhantumbo, I. Tenzing, J. (2014). Building resilience to environmental change by transforming gender relations. IIED. Retrieved from http://pubs.iied.org/17237IIED.html Tagseth, M. (2008). Oral history and the development of indigenous irrigation. Methods and examples from Kilimanjaro, Tanzania. Norsk Geografisk Tidsskrift - Norwegian Journal of Geography, 62(1), 9–22. http://doi.org/10.1080/00291950701864898 Tagseth, M. (2010). Studies of the waterscape of Kilimanjaro, Tanzania Water management in hill furrow irrigation. Dissertation NTNU. Retrieved from http://www.diva-portal.org/smash/record.jsf?pid=diva2:355649 Tang, S. Y. (1992). Institutions and collective action: self-governance in irrigation. San Francisco: ICS Press. Tedlock, B. (2000). Ethnography and ethnographic representation. In N.K Denzin and Y.S. Lincoln (Eds) Handbook of qualitative research. SAGE. 455- 486

81 Tenge, A. J., De Graaff, J., & Hella, J. P. (2004). Social and economic factors affect- ing the adoption of soil and water conservation in West Usambara highlands, Tanzania. Land Degradation & Development, 15(2), 99–114. Thompson, C. B. (2012). Alliance for a Green Revolution in Africa (AGRA): ad- vancing the theft of African genetic wealth. Review of African Political Econo- my, 39(132), 345–350. http://doi.org/10.1080/03056244.2012.688647 Thomson, J. (1887). Through Masai Land: a Journey of Exploration Among the Snowclad Volcanic Mountains and Strange Tribes of Eastern Equatorial Africa. Sampson Low, Marston, Searle, & Rivington. Tucker, C. M., Randolph, J. C., & Castellanos, E. J. (2007). Institutions, Biophysical Factors and History: An Integrative Analysis of Private and Common Property Forests in Guatemala and Honduras. Human Ecology, 35(3), 259–274. http://doi.org/10.1007/s10745-006-9087-0 Wangui, E. E. (2014). Livelihood Shifts and Gender Performances: Space and the Negotiation for Labor among East Africa´s Pastoralists. Annals of the Associa- tion of American Geographers, 0(0), 1–14. http://doi.org/10.1080/00045608.2014.924734 Watson, E. E. (2009). Living terraces in Ethiopia : Konso landscape, culture & development. Woodbridge, UK ; James Currey. Watson, E. E., Adams, W. M., & Mutiso, S. K. (1998). Indigenous Irrigation, Agri- culture and Development, Marakwet, Kenya. The Geographical Journal, 164(1), 67–84. http://doi.org/10.2307/3060546 Westerberg, L.-O., Holmgren, K., Börjeson, L., Håkansson, N. T., Laulumaa, V., Ryner, M., & Öberg, H. (2010). The development of the ancient irrigation system at Engaruka, northern Tanzania: physical and societal factors. Geographical Journal, 176(4), 304–318. http://doi.org/10.1111/j.1475-4959.2010.00370.x Wheeler, T., & Braun, J. von. (2013). Climate Change Impacts on Global Food Security. Science, 341(6145), 508–513. http://doi.org/10.1126/science.1239402 Whitehead, A. (1998). Gender, Poverty and Intra-Household Relations in Sub Sa- haran Small Holder Households: Some Lessons from Two Case Examples. Background Paper prepared for the 1998 SPA Report on Poverty and Gender in sub-Saharan Africa. World Bank. Widgren, M. & Sutton, J. E. G. (Eds.) (2004). Islands of Intensive Agriculture in Eastern Africa: Past & Present. James Currey Publishers. Widgren, M. (2003). bin Laden och bevattningstunnlarna – två perspektiv på hållbar samhällsutveckling. I K.E. Knutsson & F. Lundmark (red) Kultur, säkerhet och hållbar samhällsutveckling efter 11 september. Stockholm: Riksbankens jubile- umsfond & Gidlunds förlag. (193-206). Widgren, M. (2004a). Can Landscapes be Read? In H. Palang, H. Sooväli, M. An- trop, & G. Setten (Eds.) European Rural Landscapes: Persistence and Change in a Globalising Environment (455–465). Springer Netherlands. Widgren, M. (2004b). Towards a historical geography of intensive farming in eastern Africa. In M. Widgren & J.E.G Sutton (Eds.) Islands of Intensive Agricul- ture in Eastern Africa (1–18). James Currey Publishers. Widgren, M. (2007) Pre-colonial landesque capital: a global perspective. In A. Hornborg,J. Martinez-Alier and J.R. Mcneill, (Eds.) Rethinking Environmental History: World-System History and Global Environmental Change. Alta Mi- ra Press, Walnut Creek. (61–77). Widgren, M. (2014). Furrows in Africa - canals in the Americas? Azania: Archaeo- logical Research in Africa, 49(4), 524–529. http://doi.org/10.1080/0067270X.2014.968389

82 Wiggins, S. (2009). Can the smallholder model deliver poverty reduction and food security FAC Working Paper 08, Brighton: Future Agricultures Consortium Williamson, D. L., Choi, J., Charchuk, M., Rempel, G. R., Pitre, N., Breitkreuz, R., & Kushner, K. E. (2011). Interpreter-facilitated cross-language interviews: a research note. Qualitative Research, 11(4), 381–394. http://doi.org/10.1177/1468794111404319 Winterbottom, A., Koomen, J., & Burford, G. (2009). Female Genital Cutting: Cul- tural Rights and Rites of Defiance in Northern Tanzania. African Studies Re- view, 52(01), 47–71. http://doi.org/10.1353/arw.0.0142 Witcher, C. S. G. (2010). Negotiating Transcription as a Relative Insider: Implica- tions for Rigor. International Journal of Qualitative Methods, 9(2), 122–132. Yin, R. K. (2009). Case Study Research: Design and Methods. SAGE. You, L., Ringler, C., Nelson, G., Wood-Sichra, U., Robertson, R., Wood, S.Sun, Y. (2010). What is the irrigation potential for Africa? (IFPRI discussion paper No. 993). International Food Policy Research Institute (IFPRI). Young, O. R. (2002). The Institutional Dimensions of Environmental Change: Fit, Interplay, and Scale. MIT Press. Zaag, P. van der. (2007). Asymmetry and Equity in Water Resources Management; Critical Institutional Issues for Southern Africa. Water Resources Management, 21(12), 1993–2004. http://doi.org/10.1007/s11269-006-9124-1 Zwarteveen, M. (2008). Men, masculinities and water powers in irrigation. Water Alternatives, 1(1), 111–130. Zwarteveen, M. Z., Udas, P. B., & Delgado, J. V. (2010). Gendered Dynamics of Participation in Water Management in Nepal and Peru: Revisiting the Linkages between Membership and Power. In Berry, K. A., & Mollard, E. (Eds.) Social Participation in Water Governance and Management: Critical and Global Per- spectives. Earthscan. (69–92).