1 COMMUNICATION SOCIAL NETWORKS AND PERCEPTIONS OF WATER AND WILDLIFE IN THE OKAVANGO DELTA, BOTSWANA By DEBORAH JEANNE WOJCIK A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFIL LMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2011
2 2011 Deborah Jeanne Wojcik
3 To my parents and Tim for their unconditional love and support
4 ACKNOWLEDGMENTS I owe an enormous debt of gratitu de to my advisor, Martha Monroe, who has been an outstanding facilitator of intellectual growth and tireless supporter of my academic and professional journey. I am grateful to all of my committee members for their many contributions : Chris McCarty for ins piring my inter est in social network analysis and creating the software tools I needed to do it; Brian Child for providing me a vehicle, contacts, and the op portunity to dive into research in southern Africa head first; Marianne S chmink for helping me see things through an anthropological lens and teaching me about gender ; and Matt Cohen for making every conversation with him an opportunity to innovate and elevate my thinking. I am incredibly grateful for the many opportunities I was afforded while part of the University of Florida. The Adaptive Management: Wise Use of Water, Wetlands, and Watersheds IGERT provided an exciting and intellectually stimulating forum for exploring meaningful engagement in interdisciplinary studies. I am grateful to Mark Brown, t he inspiring faculty team, and Carol Binello and Sharlynn Sweeney for always keeping the wheels turning and for the inspiring trips to the Everglades and southern Africa. The UF Tropical Conservation and Development (TCD) program p rovided an incredible com munity, source of support, and intellectual challenge throughout my time at UF. The Center for African Studies provided valuable professional development opportunities and a community of people who share my p assion for research in Africa Speaking of pass ion, I am incredibly grateful for my fellow graduate students for their input, support, and friendship throughout the process. I will continue to be inspired by the passion and enthusiasm that forever emanates from lab and s tudents, and I a m grateful to my colleagues in the IGERT and the TCD program for
5 allowing me to continually learn with and from you. There are too many names to mention I will count on them knowing who they are. I do need to thank Kelly Biedenweg without who m I may not h ave been inspired to make my way to Gainesville in the first place and the IGERT 2007 cohort for becoming my new family upon arrival. This research would not have been possible without several sources of funding and support T he National Science Foundatio n provided preliminary research travel funding and financial support for my graduate studies through the IGERT program. The School of Forest Resources and Conserva tion provided fellowship funding additional employment opportunities through Ma rtha Monroe, and incredible support through Tim White and his staff when it was most necessary in my graduate school journey. The Working Forests in the Tropics IGERT also provided preliminary research fu nding, and the TCD Program funded both research travel and gradua te studies upon my return from the field The Center for African Studies granted me a FLAS Fellowship so that I could develop Setswana language skills to advance my dissertation fieldwork. I am thankful for the many people in Botswana who made my research possible I am immensely grateful to the chiefs and headmen who allowed me to live and work in their communities and provided me important insights about their villages. My research would not have be en possible without my team of assistants (Daniel, Meck, Seteng, Amos, Kutlwano, Onkgolotse, Onkgopotse, Bolick, Osetse, Tsima, Elmo, Motto, Olebogeng, Gosiame, Ruth, Kegoikantse, Keitikile, and Victoria). The many employees and board members of the Khwai Development Trust, Sankoyo Tshwaragano Management Trust, and the Okavango Community Trust were also extremely helpful throughout my research. Tho ugh I cannot possibly name all o f these incredible people
6 individually, Graham, Raphael, and John Witness provided me a great deal of information and connections to man y other helpful people I am especially thankful for the generosity of all of the people who agreed to participate in this study Many other people in Botswana offered their experience and insights during my research des ign and implementation process I a m grateful to the Ministry of Environment, Wildlife and Tourism and specifically the Department of Wildlife and National Parks for granti ng me permission to conduct my research (Permit EWT 8/36/4 III (10)) and actively participating in it Thank you also to the Okavango Research Institute, University of Botswana, OKACOM, BIOKAVANGO Project, Botswana Tourism Board, Department of Water Affair s, Ministry of Health, Botswana Wildlife Training Institute, and Mr. Tiego Mpho for being among those who provided helpful insights about the research context and the research that was needed. I cannot express enough gratitude for t he amazing support of my Maun hosts, Jan and Virginia and my Seronga host, Jennifer Katchmark thinking and the implementation of this research project in Seronga were invaluable. This research would not have happened without the tireless efforts of Tim Podkul who was an incredible source of intellectual, logistical, and personal support I am forever thankful for the incredible love and support Tim provides as my husband and partner and so glad that we were assigned Finally, I am grateful to my wonderful parents and family for their unfailing support, patience, and kindness.
7 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF FIGURES ................................ ................................ ................................ ........ 11 LIST OF ABBREVIATIONS ................................ ................................ ........................... 13 ABSTRACT ................................ ................................ ................................ ................... 14 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 16 The Botswana Context ................................ ................................ ............................ 16 Community Based Natural Resource Management ................................ ................ 19 Research Framework and Questions ................................ ................................ ...... 21 Methodological Overview ................................ ................................ ........................ 23 2 COMM UNITY SIZE, SOCIAL NETWORKS, AND COMMUNICATION FOR ADAPTATION IN RURAL VILLAGES ................................ ................................ ..... 30 Community Size and Communication ................................ ................................ ..... 30 Communica tion and Adaptation ................................ ................................ .............. 32 Social Networks Approach ................................ ................................ ...................... 33 Methods ................................ ................................ ................................ .................. 35 St udy Site ................................ ................................ ................................ ......... 35 Social Network Surveys ................................ ................................ .................... 38 Results ................................ ................................ ................................ .................... 46 Discussion ................................ ................................ ................................ .............. 50 Impacts of Community Size on Network Characteristics ................................ .. 50 Information Flows, Learning, and Adaptation ................................ ................... 53 Implications ................................ ................................ ................................ ............. 56 3 IMPORTANCE OF GENDER AND ETHNICITY IN THE STRUCTURE AND OUTCOMES OF RURAL COMMUNICATION NETWORKS ................................ .. 67 Introductory Framework ................................ ................................ .......................... 67 Participation and Community Based Conservation ................................ .......... 67 Importance of Communicat ion ................................ ................................ .......... 69 Gender, Ethnicity, and Natural Resources ................................ ....................... 71 Context for this Study ................................ ................................ .............................. 73 Research Setting ................................ ................................ .............................. 73 Environmental Communication Background ................................ ..................... 75 Gender and Ethnicity in Context ................................ ................................ ............. 77 Gender, Natural Resource Use, and Communication ................................ ...... 78
8 Ethnicity, Natural Resource Use, and Communication ................................ ..... 80 Methods ................................ ................................ ................................ .................. 86 Personal Social Network Interviews ................................ ................................ 86 Findings ................................ ................................ ................................ .................. 92 Gender ................................ ................................ ................................ ............. 92 Ethnicity ................................ ................................ ................................ .......... 100 Implications for Environmental Communication ................................ .................... 105 4 PERCEPTIONS OF WATER AND WILDLIFE RESOURCES ............................... 127 Literature Review ................................ ................................ ................................ .. 128 Research Setting, Data, and Methods ................................ ................................ .. 131 Research Setting ................................ ................................ ............................ 132 Assessing Mental Models ................................ ................................ ............... 134 Da ta Collection and Analysis ................................ ................................ .......... 135 Phase 1: Domain definition ................................ ................................ ...... 136 Phase 2: Structured perceptions interviews ................................ ............. 138 Phase 3: Data analysis ................................ ................................ ............ 142 Results ................................ ................................ ................................ .................. 145 Water ................................ ................................ ................................ .............. 145 Wildlife ................................ ................................ ................................ ............ 148 Discussion ................................ ................................ ................................ ............ 154 Implications ................................ ................................ ................................ ........... 158 5 CONCLUSIONS ................................ ................................ ................................ ... 176 Broader Significance and Recommendations ................................ ....................... 180 Validity, Reliability, and Objectivity ................................ ................................ ....... 183 Future Research ................................ ................................ ................................ ... 185 Summary ................................ ................................ ................................ .............. 186 APPENDIX A SOCIAL NETWORK ANALYSIS SURVEY ................................ ........................... 188 B FREE LIST INTERVIEW GUIDE: WATER AND WILDLIFE PERCEPTIONS ....... 196 C GOVERNMENT AGENCY SEMI STRUCTURED INTERVIEW GUIDE ............... 197 D PERCEPTIONS OF WATER AND WILDLIFE: INTERVIEW GUIDE .................... 198 E PERCEPTIONS OF WATER AND WILDLIFE: CONCEPT ELICITATION RECORDING FORM ................................ ................................ ............................ 204 LIST OF REFERENCES ................................ ................................ ............................. 206 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 224
9 LIST OF TABLES Table p age 1 1 Overview of characteristics of study villages ................................ ...................... 28 2 1 Population estimates for study villages ................................ ............................... 59 2 2 Number of personal network interviews used in each village to create whole network outputs ................................ ................................ ................................ .. 59 2 3 Social network measures for each ty pe of whole network output in each village ................................ ................................ ................................ ................. 60 2 4 Comparison of network measures across different whole net work outputs for each village ................................ ................................ ................................ ......... 61 3 1 Dominant ethnic group and number of personal network interviews conduc ted in each village ................................ ................................ ................. 109 3 2 Gender of respondents and homophily by gender between social network survey respondents and alters named in each village ................................ ...... 109 3 3 Descriptive statistics calculated for homophily by gender in each village ......... 110 3 4 Results of paired t tests comparing means for men and wom en of homophily by gender in each village ................................ ................................ .................. 111 3 5 Inclusion of traditional authorities in personal communication networks of men and women ................................ ................................ ............................... 112 3 6 Homophily by ethnicity between social network survey respondents and named alters in each village ................................ ................................ ............. 112 3 7 Descriptive statistics and tests of homogen eity of variance for homophily by ethnicity in each village ................................ ................................ ..................... 113 3 8 ANOVA comparison for homophily by ethnicity in each village ........................ 114 4 1 Rural respondents in Phase 1 free list exercises about water and wildlife ....... 162 4 2 Demographic and social network for structured interview respondents ............ 1 63 4 3 Summary of all concepts mentioned about water during structured perceptions interviews ................................ ................................ ...................... 164 4 4 Summary of all conce pts mentioned about wildlife during structured perceptions interviews ................................ ................................ ...................... 166
10 4 5 Responses for chili peppers in wildlife rating exercise ................................ ...... 168 4 6 Number of concepts mentioned about water and wildlife during open ended interview questions ................................ ................................ ........................... 168 4 7 Correlations between social network measures and the n umber of water and wildlife concepts mentioned during open ended interview questions ............... 168 4 8 Results of cultural consensus analysis for ratings of wildlife concepts ............. 168 4 9 Results of Independent Samples Kruskal Wallis Test to test for differences in cultural consensus agreement scores by multiple independent variables ........ 169 4 10 Correlations between social network measures and level of agreement about the importance of wildlife concepts ................................ ................................ ... 169
11 LIST OF FIGURES Figure p age 1 1 Map of Okavango Delta and research sites ................................ ....................... 29 2 1 Network density f or four types of whole network outputs (Maximum, Majority, Minimum, and Respondent Reported) for each village ................................ ...... 62 2 2 for each village ................................ ................................ ................................ .. 63 2 3 Mean number of people (alters) named from the same village as the respondents ................................ ................................ ................................ ....... 64 2 4 ........ 65 2 5 Visualizations for whole network outputs for Respondent Reported .................. 66 3 1 Percent of social network survey respondents in each gender and village. ...... 115 3 2 Number of social network survey responde nts from each ethnic group. .......... 115 3 3 Mean percent homophily by gender among social network survey respondents ................................ ................................ ................................ ..... 116 3 4 Mean percentage homophily by gender for whole networks, including respondents and all alters named by respondents ................................ .......... 117 3 5 Visualizations of village networks; nodes sized by betweenness ..................... 118 3 6 Mean amount of information received by respondents from alters of about wildlife. ................................ ................................ ................................ .............. 119 3 7 Mean amount of information received by respondents from alters about their local CBNRM program. ................................ ................................ ..................... 120 3 8 Percent of men and women naming a village traditional authority compared with the total number of respondents of that gender in each village. ................ 121 3 9 Mean degree centrality by gender in whole networks, including respondents and all alters named by respondents ................................ ................................ 122 3 10 Mean constraint by gender for whole networks, including respondents and all alters named by respondents ................................ ................................ ........... 123 3 11 Mean percent homophily by ethnic group for respondents in each village. ...... 124
12 3 12 Mean amount of information received by respondents from alters about wildlife. ................................ ................................ ................................ .............. 125 3 13 Mean amount of information received by respondents from alters about the local CBNRM program. ................................ ................................ ..................... 126 4 1 Map of Seronga and Gudigwa ................................ ................................ ......... 170 4 2 Respondents for perceptions interviews ................................ ........................... 171 4 3 Visualization of frequency of concepts mentioned during open ended water perceptions questions ................................ ................................ ...................... 172 4 4 Most frequently mentioned sources of information about water. ...................... 173 4 5 Visualization of frequency of concepts mentioned during open ended wildlife perceptions questions ................................ ................................ ...................... 174 4 6 Box plot of agreement scores for three ethnic groups responding to wildlife rankings. ................................ ................................ ................................ ........... 174 4 7 Most frequently mentioned sources of information about wildlife. ..................... 175
13 LIST OF ABBREVIATION S AGM Annual general meeting CBNRM Community based natural resource m anagement CBO Community based organization CITES Convention on International Trade in Endangered Species of Wild Fauna and Flo ra CSO Central Statistics Office, Republic of Botswana DWNP Department of Wildlife and National Parks GAD Gender and development IPC International Poverty Centre IPCC Intergovernmental Panel on Climate Change KDT Khwai Development Trust NRM Natural resourc e management OCT Okavango Community Trust RADP Rural Area Development Programme RPM Reasonable Person Model SES Social ecological system SNA Social network a nalysis STMT Sankoyo Tshwaragano Management Trust TAC Technical Advisory Committee WID Women in dev elopment WMA Wildlife Management Area
14 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy COMMUNICATION SOCIAL NETWOR KS, AND PERCEPTIONS OF WATER AND WILDLIFE IN THE OKAVANGO DELTA, BOTSWANA By Deborah Jeanne Wojcik December 2011 Chair: Martha C. Monroe Major: Forest Resources and Conservation Information is a critical resource for rural community members who must ada pt to survive and sustain natural resource dependent livelihoods. In rural villages of the Okavango Delta of Botswana climate change is predicted to contribute to changes in natural resource availability, health outcomes, and livelihood strate gies. Rural people rely upon w a ter resources to sustain themselves, their subsistence crops, and the wildlife populations that provide local livelihood benefits through community based natural r esource m anagement (CBNRM) programs. People need to access, integrate, and act upon information about these natural resources in order for them to participate in decision making processes and adapt to changing and uncertain conditions. This research applied social network analysis, ethnographic field observations, and interview s to i nvestigate how certain variables namely community size, gender, and ethnicity, affect information flows and perceptions a bout water and wildlife in rural villages Results revealed that the size of a village affects the stru cture of communication ne tworks and th e ways that people participate in decision making processes. People in smaller villages have more opportunities for face to face interaction, though the extent to which this facilitates information exchange and
15 participation is mediated by the social dynamics prese n t in a particular context. Gender and ethnicity also affect how information about natural resources is exchanged, integrated, and acted upon; c ommunication and natural resource decision making is often dominated by men and majority e thnic groups in rural villages The p erceptions that people hold about natural resources are affec ted by the type of resource, the ethnic background of the individual their village communication network M ental models (co gnitive representations built from experience ) are more robust for wildlife than water reflecting the relative complexity and importance associated with understanding wildlife in this setting Ethnicity which is closely tied to a person wildlife, and the extent to which a person is connected to others also appear to affec t perceptions Understanding how community size, gende r, and ethnicity affect communication netw orks, participation, and perce ptions about water and wildlife is helpful in design ing communication strategies and decision making processes that engage marginalized groups, actively foster information exchange, and facilitate adaptation among rural community members.
16 C HAPTER 1 INTRODUCTION Though many resources are highly valued in natural resource management, perhaps the most critical resource is information. Information is necessa ry for learning, decision making, and adaptation. Complexity, uncertainty, and change in social ecological systems (SES) make adaptat ion, and th erefore timely and relevant information critical In Botswana and throughout southern Africa, where climate change is predicted to contribute to significant changes in natural resource availability, health outcomes, and livelihood strategies (IPCC 2007), it is important to understand patterns of communication and h perceptions. This dissertation focuses on rural communities in the remote Okavango Delta region of Botswana, where wat er and wi ldlife are of paramount importance and their management requires that community members have access to information so that they can adapt to changing environmental conditions and engage in decision making processes important to their livelihoods. I nformati on about the populations and movements of wildlife for example, can help community members protect their homes and fields and maximize the benefits receive d throu gh community based conservation programs. The Botswana Context This research was conducte d in the Okavango Delta region of Ngamiland District Botswana. Compared with its southern African counterparts, Botswana has achieved considerable social, political and economic success According to the CIA Factbook (2011), t he GDP of Botswana is estima ted at $28.49 billion USD; this rank s 110 th in the
17 world and second only to South Africa in s outhern Africa. Aside from 2007 20 08, Botswana has grown economically every year since it gained independence from the British in 1966 Though the country has rise n to the status of a middle income country with a per capita GDP of $13,100 in 2010 (CIA Factbook 2011), inequality in income has increased ( IPC 2005). While the proportion of people in poverty is decreasing in Botswana overall ( IPC 2005, C IA Factbook 2011 ), pockets of sev ere poverty persist (CSO 2006). T he remote rural region of Ngamiland has one of the highest concentrations of poverty in the country (Moseki 2009). Botswana is also second in the world in terms of the prevalence of HIV/AIDS (CIA Factbook 2 011) particularly among those of working age (20 45 years old ) This is predicted to cause a decrease in incomes among households affected by th e disease and contribute to increased poverty and income inequality (Greener et al. 2000). T ourism is an impor tant source of income at the national level, and it is second only to diamond mining as a source of revenue for Botswana ( CIA Factbook 2011). Though the focus of tourism varies by region, it is primarily wildlife based tourism. Seventeen percent o f the cou ntry is protected in national parks and game r eserves, with an additio nal 20% of lands designated as wildlife management a reas (WMAs) The Okavango Delta, the largest inland delta in the world, is considered the most important tourist attraction in the cou ntry (Botswana Tourism Department 2000). International tourists visit Botswana to either hunt for trophy animals (e.g., elephants, antelope) or participate in photographic safaris. The tourism model in Botswana g enerally focuses on tourism to protect natural resources and preserve the safari experience (Stevens and Jansen 2002 ) This means that a small number of people tend
18 to pay large sums of money to engage in tourism in Botswana (Botswana Tourism Department 2000). Like other indu stries, tourism brings financial benefits, but it also creates natural resource management challenges. Considerable w aste is generated, and septic systems can discharge effluent into groundwater, which is the primary human water source in the Okavango Delt a (Mbaiwa 2011). Tourism and the mining industry have also placed an increased demand on water resources C limate change models predict that water resources could decrease by as much as 25% in this semi arid region over the next several decades (IPCC 2007) A model specific to the Okavango Delta shows that the combination of human consumption and climate change, which has the potential to reduce rainfall and river flows and increase temperatures in the region, may contribute to significant drying of the Del ta (Murray Hudson et al. 2006). Botswana is at the nexus of water scarcity, wildlife dependence, and climate change issues. Rahm et al. (2006, p. 161) state: The demand for water coming from human use, agriculture, livestock, mining, and wildlife (touris m) is increasing due to the dev elopment, and population growth Rural people of the Okavango Delta region of Botswana rely on water resources to sustain themselves their crops, and the wildlife populations essen tial to local tourism based livelihoods. Anticipated changes in water availability will not only affect humans but also wildlife populations, their abundance, locations, and migration patterns (Malcolm et al. 2002) This has the potential to increase confl icts between humans and wildlife as competition for limited resources increases with decrease d rainfall. Furthermore, many communities receive benefits from wildlife tourism through c ommunity b ased natural
19 resource m anagement (CBNRM) programs. C hanges in w ildlife population densities and movements could greatly impact livelihoods in communities where benefits from CBNRM are important. Community Based Natural Resource Management CBNRM originated in the late 1980s a s a multifaceted approach integrating conser vation and development goals. CBNRM reflected a paradigm shift in the 1980s that moved conservation efforts in many developing countries from strict preservation or conservatio n that integrat ed people into a relationship of managed use of resources (Brundtland 1987). transfer of the rights to use, manage, and allocate resources, especially high val ue An assumption inherent in this approach is that people in rural areas k now their local natural resources best and are therefore better able to manage these resources ( Brandon and Wells 1992 ) to meet join t goals of biodiversity conservation and poverty reduction ( Brundtland 1987, IUCN/UNEP/WWF 1991, Abbott et al. 2001, Adams and Hulme 2001, Artnzen et al. 2003, McShane and Wells 2004). CBNRM and other c ommunity based conservation approaches focus at least in theory, on the creation of locally specific conservation solutions with goals of returning power, influence and use to the local human users of natural resources through participation (Agrawal and Gibson 1999, Ainslie 1999 Agrawal 2000, Kel lert et al. 2000, Twyman 2000, Mulder and Copolillo 2005 ). The idea that local people should be included as participants and decision makers in the management of natural resources has gained prominence since the emergence
20 of community based conservation ( IIED 1994, Western and Wright 1994, Schmi nk 1999, Adams and Hulme 2001). This of ten proves challenging in practice, particularly when intra community diversity and power structures make it dif ficult for voices of certain groups to be integrated into decisi on making processes (Agrawal and Gibson 1999). In many cases, p articipation involves only the elite, with marginalized groups such as women and ethnic minorities most vulnerable to exclusion (Nelson and Wright 1995). Founded in 1989 with USAID funding t he CBNRM approach began in Botswana as a reaction to severe declines in wildlife and the failure of traditional top down management approaches to protect these resources (Taylor 2007). C ommunity based organizations (CBOs) were established as local implemen ters of CBNRM, charged with managing wildlife (Ar ntzen 2003). CBOs were granted leases to WMAs, along with exclusive rights and responsibility for managing n atural resources in these areas. In return, c ommunities receive benefits from any commercial activi ties, tourism or otherwise, that take place on these lands ( Rozemeijer 2009). Though originally designed with a focus on wildlife, the governance and communication structures established by CBNRM provide a framework for thinking about the management of sh ared natural resources and learning that contributes to adaptation (Child 2009) The dynamic nature of both the biophysical and socio cultural environments in which resources are managed in CBNRM requires that people and governance structures be adaptive a nd respon d to changes and uncertainties inherent in these complex systems (Olsson et al. 2004 Giannini et al. 2008 ). Adaptation requires that people learn about changes in biophysical and socio cultural systems, synthesize
21 new information and if necessar y, take action to accommodate observed changes (Kaplan and Kaplan 1982). The participatory decision making processes of CBNRM have the potential to greatly influence learning and knowledge acquisition through participatory deliberations (Schusler et al. 20 03). R esearch Framework and Questions T his dissertation research is based upon the premise that information is a critical currency for decision making, adaptation, a nd natural resource management It is therefore important to understand how information is exchanged, received, and s perceptions. Many variables have the potential to affect information flows and the integration of information into mental models This dissertation addresses several possible factors in chapters two throug h four, focusing in particular on variables related to the social networks of rural communities. These are presen ted as three distinct studies; this reflect s the anticipated publication of these research results as manuscripts in different peer reviewed pu blications. Variability in community size to access, integrate and share information ( Kaplan and Kaplan 1982 Dunbar 1998 ). C hapter 2 Community size social networks, and communication for adaptation in rural villages applies social network analysis to investigate how population size may impact the communicati on relationships in rural communities. The chapter lays out a methodological approach for collecting and combining social networ k analysis at the village level and addresses the research question: How do social network characteristics related to information flows vary with community size ?
22 Community size is hypothesized to affect the characteristics of communication networks such that people in smal ler villages are more closely connected to one another than people in large communities. This is relevant in Botswana where CBNRM is implemented in communities of a wide range of sizes. Size is one factor that may impact how well information about natural resources and CBNRM reaches different members of those villages and how well people can participate in decision making The third chapter, Importance of gender and ethnicity in the structure and outcomes of rural communication networks addresses gender a nd ethnicity as social variables important to understanding information flows about natural resources and participation in CBNRM in this research setting. In patriarchal societies such as that of Botswana, women may not be fully integrated into participato ry processes (Kabeer 1994, Schmink 1999), and minority ethnic groups may be excluded from political processes due to existing power structures ( Mompati and Prinsen 2002 Bengston 2004 ). Chapter 3 investigates the relationships between gender and ethnicity and the social dynamics around natural resource use, communication patterns and participation in decision making processes. This study addressed the following research questions : To what extent do people tend to communicate more with people of their own gender or ethnicity? How might these social dynamics affect communication and participation in CBNRM? Literature review and observations suggest that people t end to communicate more with people of the same gender and ethnic group, which has the potential to affect the information available to them This, in turn, may affect the extent to which they engage in natural resource decision making in their village. This is relevant to considerations of how to achieve conservation and development goals, specifical ly those that emphasize participation and the inclusion of marginalized groups. It may also offer insights into
23 designing communication strategies that allow people of different genders and ethnic groups to have the opportunity to become more informed and active in CBNRM. Chapter 4 focuses on what happens to information at the individual level and investigates the perceptions people have about water and wildlife resources This study P erceptions of water and wildlife resources, draws on theories of mental models to understand some of the variables that may influence perceptions about water and wildlife This chapter addresses the research question : Do differences exist in the ways rural community members perceive water and wildlife and what facto rs may account for existing differences ? Investigating which concepts are and are not important, as this can provide insights about how people understand their environment and internaliz e information from outside sources These results can be used to improve communication strategies enhance the int egration of government messages into understanding about natural resources The final ch apter brings together the findings of chapters two thr ough four and discusses how these results might be applied to enhance the adaptive capacity of rural communities through improved communication and participation in CBNRM. Methodological Overview This r esearch was conducted in four village communities comprising a range of characteristics: Khwai, Sankoyo, Gudigwa, and Seronga (Figure 1 1). These communities vary in terms of numbers of residents, dominant ethnic groups, geographic proximity to the regiona l center in Maun, and types of CBNRM programs (Table 1 1) Khwai and Sankoyo have single village CBNRM programs, while Gudigwa and Seronga
24 are two of the five villages in the multi village Okavango Community Trust (OCT ); these two villages are also the f a r thest from Maun of the four The four villages were selected to represent a range of village characteristics and CBNRM implementation strategies. This research was conceived and conducted over several field visits between 2007 and 2010. An initial study p rogram in 2007 brought to light the challenges rural communities faced balancing con servation and development goals and introduced ideas about the importance of communication in CBNRM. Several months of p reliminary research in summer 2008 involve d assistin g with participatory community meetings and conducting household surveys field observations, and interviews with key informants. These data, combined with information collected through attendance at the national CBNRM meeting of Botswana led to the devel opment of research questions focused specifically on community size and some of the social variables that potentially impact communication in rural communities. Concerns about community size were of interest to both fellow researchers from the University o f Florida and Botswana officials engaged with CBNRM. The majority of data collection was conducted between September 2009 and May 2010. In all data collect ion efforts, local research assistants were hired and trained as part of a village research team. All research assistants were very familiar with their home village and fluent in at least Setswana and English. Several assistants knew additional local language s and were called upon when necessary to conduct these translations. Assistants worked with the re search team to translate all survey instruments into Setswana The primary investigator had completed two intensive courses in Setswana language training prior to commencing field work in 2009 and was
25 able to understand th e translations to some degree. E ac h translation was ve rified by at least three and up to six additional people who were also bilingual in Setswana and English. Research assistant s were extensively trained over several days to conduct a village census and social network surveys Two assista nts were also trained to conduct perceptions interviews Social network analysis was used as a primary research method throughout this dissertation. Personal social network surveys were built and conducted in EgoNet software (McCarty 2011) and used to unde rstand how people are connected and exchange information (Wellman and Berkowitz 1988, Bernard et al. 1990, Wasserman and Faust 1994). Personal networks can provide information about both compositional variables (characteristics of network members) and stru ctural variables (characteristics of the network itse McCarty et al. 2007). Data from personal network interviews were analyzed in chapter 3 to investigate how communication is related to gender and ethnicity. These personal networks were also combined to create whole networks for each village that included all communication conn ections reported (C. McCarty pers. comm. ). These whole networks were then analyzed using UCINET software (Borgatti et al. 2002). Outputs generated by these analyses were used to select the sample of respondents interviewed in the perceptions study presented in chapter 4. For the perceptions study, r esearchers customized existing assessment methods to develop a research instrument that was culturally ap propriate and helpful in understanding perceptions of water and wildlife in this setting. The design emerged during open ende d interviews and pilot testing and resulted in a combination of op en
26 ended free listing exercises, structured ratings questions an d semi structured interview questions These were analyzed using Anthropac 4.98 software (Borgatti 1998). Part of this analysis applied consensus analysis to u nderstand the extent to which perceptions of important concepts related to wildlife were shared a mong respondents. In addition to quantitative data collection and analysis, researchers qualitatively analyzed field notes and interview data recorded during the study. Meeting observations and informal interviews with community members were important dat a sources for contextualizing the results that were observe d. I nterviews with government officials and tribal au thorities provided insights about what the key messages were at the time of this research, how this information is communicated to rural communi ties, and how people interact with and engage in CBNRM in rural areas. One of the intellectual contributions of this research is methodological. This study used social network analysis data in a new way by creating whole networks from personal network dat a. This approach provide d a means of assessing whole network characteristics in settings where conducting whole network interviews is either difficult or undesirable. It is useful when networks are large and research resources are limited or where setting a bou ndary for a whole network c ould undermine the purpose of the research. In this research the number of interviews that would have been required to conduct a whole network study would have been prohibitive; people tend to move between towns and residen ces, which makes creating a meaningful village boundary challenging; and potentially important sources of outside information would have been excluded, which was undesirable given the nature of this research.
27 Data were analyzed using EgoNet version 3.31.11 UCINET 6.357, NetDraw 2.109, Anthropac 4.98, and SPSS 19.0 software packages. Preliminary research results were presented in meetings and through reports to the four main villages included in this study at the conclusion of the data collection phase. Ini tial analyses were also presented at a conference in the Okavango Delta region of Botswana to academic, government, and community representatives.
28 Table 1 1. Overview of characteristics of study villages Village Population estimate Dominant eth nic group(s) in village Approx. driving distance from Maun (km) Type of CBNRM trust Khwai 300 Bushmen 125 Single village Sankoyo 400 Bayei 85 Single village Gudigwa 700 Bushmen 522 Multi village (OCT) Seronga 1 600 Bayei, Bambukushu* 455 M ulti village (OCT) Many ethnic groups are represented in the large village of Seronga, with Bayei and Bambukushu the two largest groups, in that order according to key informants. The kgosi (chief) and many residents are of Bayei origin, while sever al wards are dominated by Bambukushu.
29 Fig ure 1 1. Map of Okavango Delta and research sites. Khwai, Sankoyo, Gudigwa, and Se ronga are circled in green, and Maun, the regional center, is in the red rectangle M ap courtesy of Masego Dhliwayo of the Ok avango Research Institute University of Botswan a Maun, Botswana
30 CHAPTER 2 COMMUNITY SIZE, SOCI AL NETWORKS, AND COM MUNICATION FOR ADAPTATION IN RURAL VILLAGES Community Size and Communication Soc ial dynamics influence the flow of information and ideas wi thin a community (Belaire et al. 2011). G roup size is an important variable that can affect social dynamics and and share information (Dunbar 1998). A n idea or piece of information may s pread more quickly in sm aller an d more closely connected groups than larger communities (Rogers 1995 ; Valente 1995, 1996) Interactions in smaller groups may lead to enhanced trust and commitment ( Lev in and Cross 2004, Coffe and Geys 2008), more meaningful inf ormation exchanges (Carley 1991, Andrews and Delahay 2000), and great er learning among group members (Millar and Curtis 1999, Bouwen and Taillieu 2004). Small, local groups engaged in natural resource management (NRM) may experience more frequent face to face communications, which can contribute to greater accountability and transparency in NRM (Murphree 1993). The perceived advantages of smaller scales were reflected in a global movement toward decentralized management of common pool natural resources (Ostrom 1998, Agrawal 2000). This was mirrored in development and natural resource policy in the 1980s and 1990s specifically the Brundtland Commission Report of 1987 and United Nations Agenda 21 in 1992. Influenced by the ideas of E. F. Sch u macher, wh o suggested that bigger i s not always better in economic systems, the dominant paradigm in conservation and development became Though there appear to be important advantages associa ted with smaller groups, small size may also bring chall enges (Agrawal 2000, Blaikie 2006). Smaller
31 communities or sub group s within a community may lack the capacity to access outside information, funds, or other resources (Agrawal 2000). People in larger group s may have more opportunities part of the same primary social group (Granovetter 1973) These loose connections may contribute advantages such as new knowledge, a greater diversity of perspectives and e nhanced potential for innovation (Granovet ter 1973, Burt 2004, Coffe and Geys 2008). In contrast, small groups may tend to have denser redundant connections which can lead to the reinforcement of nd norms (Valente et al. 2004) and homogeneity in thinking rather than innovati on ( Granovetter 1983, Levin and Cross 2004). This can be especially problematic if the information being exchanged is inaccurate, as inaccuracies may be reinforced within smaller groups where counter arguments may be less likely to arise. The relationship between communication and community size may have evolutionary roots. The social brain hypothesis 1 asserts relationships are affected by and have, from an evolutionary standpoint, affected their brain capacity. Dunbar (1998, 2010) cont ends that the large brain size of humans evolved to accommodate Yet, since brain size and function are biologically limited, the demands of dealing with ocial worlds to become hierarchically structured and size limited (Byrne 1996, Dunbar 1998, Zhou et al. 2005). In order for the brain to be able to remember and process social relationships, people organize these relationships into multiple successive laye rs with the intensity of relationships 1 The social brain hypothesis has also been called the social intelligence theory/hypothesis. It builds upon what is known as the Machiavellian intelligence hypothes is (Dunbar 2010).
32 decreasing as one moves from inner to outer circles (Dunbar 2008 2010). This may impact beyond their immediate social circles (Dunbar 1998). S ince people must be able to access and act upon new information in order to adapt (Kaplan and Kaplan 1982), investigating how community size affects communication patterns is important to understanding communication and adaptation. Communication and Adapt ation The dynamic nature of the biophysical and socio cultural environments in which resources are managed requires that individuals and natural resource governance structures are adaptive (Olsson et al. 2004). C limate change is predicted to contribute to changes in natural resource availability, health outcomes, and livelihood strategies i n sou thern Africa (IPCC 2007 Wilk and Kgathi 2008). In social e cological systems like this one where complexity, uncertainty, and change over time pre sent formidable ch allenges to rural populations attempting to sustain their livelihoods ( Wilk and Kgathi 2008 ), it is critically important to understand the processes underlying adaptation Adaptation may be necessary at multiple levels. At the individual level, adaptation requires people to learn about changes in biophysical and socio cultural systems, synthesize new information, and take action as a result of observed changes (Kaplan and Kaplan 1982). While the predominant focus of adaptation studies in Africa and other de veloping countries has been on trying to understand responses to predicted changes, recent studies have begun to focus more on the process aspects of adaptation (Tscha kert and Dietrich 2010). S ome studies have focus ed specifically on learning as an essenti al part of the adaptation process (Fabricius et al. 2007, McGray et al. 2007, Leary et al. 2008, Tschakert and Dietrich 2010), and others have specifically
33 addressed the importance of communication channels and information exchange as essential components of learning and adaptation (Fabricius et al 2007, McGray et al. 2007, Osbahr 2007). Groups and organizations also need to adapt to changing environments. The terms adaptive management ( Holling 1978, Walters 1986 ( Di etz et al. 2003, Folke et al. 2005) are used to describe intentionally designed, adaptive learning processes at the organizational level that acknowledge complexity, change and uncertainty in social ecological systems (Holling 1978, Walters 1986, Lee 1993 Gunderson and H olling 2002, Berkes et al. 2003 ). These learning processes are often l inked with discussions of social learning (e.g., Berkes et al. 2000, Keen et al. 2005, Pahl Wostl et al. 2007) a set of theoretical constructs that address how individu als learn from others and suggest that learning is particular group membership or social context (Bandura 1963, 1977; Wenger 1999, Wals 2007). Though the definition and implications of social learning are still contested in the literatu re (Reed et al. 2010), studies have shown that participatory decision making processes have the potential to greatly influence learning processes and knowledge acquisition (Schusler et al. 2003 ), and a to their learning experience and potential for adaptation (Armitage 2005, Keen et al. 2005, Pahl Wostl et al. 2007, Fernandez Gimenez et al. 2008). Social Networks Approach by their social context (Belaire et al. 2011) and information must be accessed and integrated in order to learn, and learning is essential for adaptation (Kaplan and Kaplan 1982) Thus, investigating social structures and information flows can provide valuable
34 informatio n toward understanding communication and adaptation. Social network analysis (SNA) provides a theoretical and methodological approach for qua nt itatively analyzing these connections and community characteristics (Borgatti et al. 2009). A social network is comprised of all actors in a network (be they individuals, groups, organizations, countries, etc.) and the relationships among the se actors. In SNA, t he actors and relationships are referred to as nodes and ties, respectively (Scott 2000). Research applyin g the social networks perspective places importance on structural characteristics of a social context that affect interactions and flows of resources (e.g., information) among interdependent actors (Wasserman and Faust 1994). The term and concepts related to social networks have been used in a diversity of social sciences since the 1930s 2 (Moreno 1934, Borgatti et al. 2009), and SNA has been applied with increasing frequency to understand natural resource governance and adaptive capacity ( e.g., Carpenter e t al. 20 01, Schneider et al. 2003, Tompk ins and Adger 2004, Newman and Dale 2005, Bodin et al. 2006 Crona and Bodin 2006, Ramirez Sanchez and Pinkerton 2009 ). In 2010, Ecology and Society published a 15, Issue 4), highlighting the potential impacts of social networks on resource management including the importance of networks in learning processes. Within this issue, Newig et al. (2010) hypothesize d relationships between net work characteristics and various aspects of learning and governance. These authors presented a general hypothesis that as network size increases, information transmission within the network 2 For a more comprehensive overview of the history of network studies in the social sciences, see Borgatti et al. 2009.
35 would also increase ; however, this is likely to be an oversimplifi cation since many factors mediate information flows They also state that while this relationship may apply to very small networks where face to face contact is possible and probable, communication and deliberation may be more difficult in larger groups (N ewig et al. 2010). This study contributes empirical evidence related to these questions about group size and the potential for information exchange. The literature contains many studies and hypotheses about how community size, social networks, informatio n, learning, and adaptation relate to one another. This study focused spe cifically on the relationship between community size and the stru cture of communication networks and addressed the research question: How do social network characteristics related to information flows vary with community size? The researchers hypothesized that as community size increased, network members would become less closely connected to one another which could in turn affect the rate and quality of communication exchange This s tudy provides empirical evidence from four rural communities of rural Botswana and is part of a larger research project addressing information flows, learning, and adaptation in rural communities. Methods Study Site Research was conducted in the Okavango Delta region of Ngamiland, Botswana. The Okavango Delta is a flood approximately 15,000 km 2 Ngamiland District is in the northwest of the country and is sparsely populated with approximately 26,000 h ouseholds and 125,000 people (CSO 2005a), including the regional population center of Maun, which is home to
36 approximately 50,000 people (CSO 2008b). The District wide population density is 0.5 1 person per square kilometer (CSO 2005b). Data were collected in four rural villages with different population size s (Figure 1 1). In order of increasing number of people the villages are: Khwai, Sankoyo, Gudigwa, and Seronga 3 (Table 2 1). While the relative size of villages is generally understood, more precise me asurements of population numbers are challenging to obtain for several reasons. First, 2001 Botswana Census data are not available for all villages. Khwai was gazetted, or officially named, as a village after the 2001 census; official population numbers ar e not yet available. Gudigwa, though larger in population than Khwai, remains ungazetted. The definition of a village can be vague and problematic in Botswana: Although there is no official definition of a village, the Census Office retained the status of village for all settlements officially designated as village is usually identified by the presence of some administrative or social facilities such as the presence of a tribal administration office and availability of facilities such as schools, cl inics or health centres, water reticulation etc. (CSO 2005a p. iii ). While some small villages are included, the Population and Housing Census Guide states that it focuses on villages of 1,000 people or more (CSO 2005a). In this study, is used as it is most commonly used in the Okavango Delta context to describe organized settlements with populations in the hundreds to thousands. Large villages are rare in the remote Okavango Delta region, where access and services are limited, water res ources are highly variable, and conflicts with wildlife are common. It is 3 Village names in this region often have multiple spellings (e.g., Sankoyo is also spelled Sankuyo, Gudigwa is also s pelled Gudikwa). The spellings used here were reported by local research assistants to be the most locally appropriate.
37 in these remote areas, where villages tend to be small, that research about communication and natural resource management is so important. Another challenge with census data is tha t there is a reasonable chance that many people who spend the majority of their time in a particular village will not be CSO 2010). Ethn ographic observations revealed that many people move frequently between residences, especially between homes in remote villages where they spend a majority of their time and shared family homes in larger villages, towns, or regional centers where services are available. Others move between residences in a main village and nearby cattle posts. Hence, an individual may spend a majority of time in one village but either go uncounted or be counted as part of another residence in the census data. This residence pattern and transitory nature presents challenges for census accuracy and for researchers developing a sampling frame. In this research, a dditional information sources were used to supplement census data about population size and then create sampling fram es from which to sample potential respondents in each village. Tribal authorities, local research assistants, and other residents were asked about the population size of their resident village. Respondents often offered a range of values, captured in Table 2 1. In Khwai, a roster of residents was obtained from the Khwai Development Trust (KD T), the local organization tasked with managing the community based natural resource management (CBNRM) program in that village. This list of 271 adults roughly matched the population estimates given by village residents (approximately 300). After checking it for accuracy
38 with local research assistants, it was used as the sampling frame for that village. No such lists were available in Sankoyo, Gudigwa, or Seronga. To cre ate these lists, researchers conducted a census by visiting all households and recording all adult residents. The criterion for inclusion in the sampling frame was that the individual spends at least six months of the year in residence in the village. Give n the previously mentioned transitory nature of the population, this was considered by residents to be a reasonable estimation of residency. Social Network Surveys There are two main types of social network analyses personal and whole that vary in the way they are implemented and in the benefits and challenges they pose. within their social network and report about their ties with these well as the ties between these alters. Alter names are solicited using a specific question based on the research question and type of network of interest. Respondents can be asked to limit their list of alters to a particular boundary (geographic or otherwise), or the solicitation can be left unbounded. Whole network studies, in contrast, ask individual respondents to report about their own ties to others but not to report on interactions between others. In many whole network studies, r esearchers provide a list with a ll of the actors of interest and go through each in turn and ask about the relationship the respondent has with each, focused on a specific type of tie of interest to the research. To create the master list, a clear boundary must be drawn around those who are included in the network and those who are outside.
39 In this study, an innovative approach was used to combine the advantages of personal network data collection with the benefits of whole network analysis. This technique allowed researchers to balance t he realities of the study site with the ability to use multiple analyses important to understanding information flows. Whole networks are often chosen in areas where group membership is well defined; however, the transitory nature of residents in and out o f villages in this study made creating a bounded network difficult. Bounding would have also prevented individual respondents from reporting about people who could be important sources of information but live outside of the village. Keeping the boundary op en was critical to understanding how information, including adaptive innovations, might come from external sources before spreading through a village. Finally, from a logistical standpoint, conducting whole network interviews, though typically shorter in l ength than personal network interviews, would have been impossible for the lead researcher to complete in the four study villages with the time and human resources available. The difficulty of this was compounded by the fact that many people who do spend a majority of their time in the village were not in the village at certain times, making it extremely difficult to reach them for interviews particularly during the planting and harvest season s Despite these challenges, whole network analysis offers impo rtant advantages, particularly when researchers are trying to understand village level dynamics affecting information exchanges in rural communities of different sizes. To analyze the data at the community wide level, EgoNet software was used to combine al l of the personal network data from each village to create a whole network for each village. Analyses of these resultant networks provided valuable data about the structural characteristics of
40 each village network, and they allowed the researcher to gain a better sense of com munication at the village level and make comparisons across villages of different sizes. The data collection and analysis protocol, which is described in more detail below, allowed the researcher to take advantage of the advantages of b oth personal and whole network analyses. Personal network interviews. A master personal network interview study was created in EgoNet software (McCarty 2011) and then customized for each village (Appendix A). The study consists of four parts: (1) questions about the respondent, (2) a question asking for a list of people with whom the respondent communicates regularly, (3) questions about each of the people the respondent named, and (4) questions about the extent to which the people named communicated with o ne another. Interviews began with the respondent answering questions about themselves, including demographic variables (e.g., gender, ethnic background) 4 Twenty eight ego survey questions 5 asked the respondent about their personal demographic characterist ics (e.g., gender, ethnic background) and their sources of information about water and wildlife resources. Respondents were then asked to name 35 people in the alter solicitation question: Please name 35 people aged 18 or older with whom you have communic ated during the last month. Include first and last names. I will let you know when you have finished. 4 The social network survey included a maximum number of 28 questions about the respondent, 1 alter prompt question that was answered 35 times, 14 questions about each alter named (490 questions) and 595 alter pair questions, for a total of 1,148 total questions. 5 In Seronga, these survey questions were adjusted slightly to accommodate an additional question about the local water source requested by individuals in that village and relevant to that community specifically. This adjustment does not impact the social network portions of the survey instrument.
41 All respondents provided exactly 35 names. Respondents were asked demographic information about each of the 35 alters they had named, including the alters ethnicity, and place of residence (inside or outside of that village) The respondents the 595 possible alter pairs. EgoNet created a personal network output for each respondent. Interviews took approximately two hours to complete. EgoNet survey data were coll ected on netbook laptops. This wa s essential for collecting alter data in a survey of ove r 1,000 questions, and it allowed the researcher and assistants to present respondents with graphical representations from their personal network results immediately following their interview. Surveys were written into the software in both Setswana, the national language of Botswana, and English, the official language. Surveys were conducted mainly in Setswana by local research assistants who were trained in each of the study villag es and were under the supervision of the lead researcher. Sampling s trategy. More than 30 personal network surveys were conducted in each of the four villages (Table 2 2). The overall research goal of this study was to understand the communication networks present in these communities ; the sample was selected accordingly A random sampling protocol was used to identify most of the potential respondents. All known adult residents in each village were listed in a roster and numbered. The random number generator function in Microsoft Excel was used to create a list of number s for each village, and these numbers were matched to roster names. Daily, each research assistant was provided with a list of people who were
42 selected based on the random number generator and lived in a certain geographic area of the village Since most t ravel between potential respondents was done by foot on sand paths, often over considerable distances, anyone provided on the daily list could be interviewed, so long as male and female respondents were equally sampled so that gender could be a basis for l ater comparison. The research team attempted to locate each potential respondent three times before removin g his or her name from the list; people who refused interviews were taken off of the list of people with whom to initiate contact. 6 Since many reside nts divide their time between their home residences and subsistence field sites, interviews were also conducted in the more remote field sites when possible with the assistance of a research vehicle. In addition to the randomly selected respondents, rese archers also included a small number of purposively selected respondents. When possible, traditional authorities were purposi vely included in this study. T hese interviews provided valuable information about communication networks both internal and external to these villages and built trust with the leaders. In one village, a traditional authority who did not appear on the original potential respondent list was included in the study; in another, a traditional authority who was on the list was preferentially approached and included. small stores that sell basic items where many people visit, might also be importan t conduits of information. T uck shop employees were interviewed in three of the villages. 6 Researchers were informed by traditional authorities, before this study, that some community members were particularly sensitive about participating in research studies based on negative past experiences. Out of respect for their wishes, researchers considered a refusal final, though a few people who initially refused did seek out membe rs of the research team at a later time and asked to participate. Since these individuals had been included on the initial list, they were included in the study.
43 Steps were also taken to accommodate the geographical and logistical considerations of the research sites. When applicable, an approximately equal number of respondents were sought in each locally defined geographic or administra tive area of the village. In Khwai, there are no such divisions made by residents or authorities; whereas in Sankoyo, for example, the village is generally reported by residents to be divided into two sub sections by the kgotla or traditional meeting plac e. The roster created by the research team incorporated geographic information, allowing the researcher to monitor the geographic distribution of respondents between the two sides of the village. Similar approaches were taken in Gudigwa, where there are fo ur distinct wards with traditional headmen (i.e., sub chiefs) in each, and in Seronga, where there are seven ward areas as well as several cattle posts outside of the main village but included within village authority. Generating whole networks from perso nal network data Personal network data can be aggregated to create whole networks for each village, which can then be used to compare the overall network characteristics of different villages ( C. McCarty pers. comm.) 7 For each village, all the data and n ames of the respondents and the people they named as alters were brought together in the same data set. Two steps were taken when processing each data set to ensure that each individual was represented by only one node in each network. First, an algorithm built into EgoNet matched the same names or names that were similar but differed slightly by spelling or capitalization. Second, a local research assistant from each village went through each of these matches for their village. They checked the names line by line, making 7 All of the steps used to combine personal network data into a whole network output are part of the
44 corrections where the algorithm matched names incorrectly and making additional matches when there were multiple names for the same individual. Multiple naming conventions and a predominance of nicknames made the latter step extremely impor tant for obtai ning an accurate set of names for each village. After the list of names was completed, four iterations of each village whole network were generated using different procedures in EgoNet. One included all ties that were provided by at least on e respondent; even if only one respondent said that two alters were tied and other respondents disagreed, the tie was included. This whole network Another protocol included or excluded ties based on the maj ority of responses given A more conservative method included only those ties agreed upon by all respondents. This The final output included only those ties that the respondents reported about th emselves and the alters they named; it excluded any ties reported between one alter and another alter. This output is referred to as Respondent Reported Analysis of whole networks. UCINET software (version 6.357; Borgotti et al. 2002) was used to analy ze the data sets created for each whole network output (Table 2 3). A number of measures were calculated for each of the output networks generated names in each network observed between actors in the network divided by the maximum number of ties that would be possible if eve
45 sub group s of actors within the overall network structure. Two measures of centralization were also calculated and included in Tabl e 2 3. Measures of n etwork centralization calculate the extent to which the network revolves around a node or a few nodes with regard to a particular centrality measure Values of centralization range from zero to 100 percent. High centralization means that a rel atively small number of individuals in the network are highly connected and others are less connected, whereas low centralization means individuals tend to have similar numbers of ties to other actors in the network (Scott 2000). compares the betweenness centrality values across all of the actors in a network. Betweenness centrality is calculated based on t he number of shortest paths that an actor lies on within the network (Freeman 1977) Measures of betweenness tend to be asso ciated with information flows, as they quantify how quickly information might move from one part of a network to another. A low betweenness centralization value indicates that betweenness centrality is equally distributed among the actors in a network, and many individuals are lie on the shortest paths present within the network. A high er betweenness centralization value means that a small er number of individuals lie along the shortest paths in a network, and many other actors are not as well connected. T o provide a basis for visual comparison, models of the networks were created using NetDraw software (Borgatti 2002; version 2.114). These were developed for two whole network outputs for each village: (1) the output in which ties were determined by the maj ority and (2) the output that only included the ties between the respondents and
46 their alters (and not any alter alter ties). This allowed for comparisons among villages across the same type of output and between two different types of outputs. Network cen trality measures for each actor in the network were added as attribute data. Attributes are specific variables, or characteristics, of nodes in a network that can be analyzed and visualized. A spring embedding algorithm which works in iterations to attrac t like nodes (i.e., nodes that share connections) and repel unlike nodes, was used to create the visualizations presented. Results Key network measures for each of the four different whole network outputs are presented for each village in Table 2 3. While the number of nodes generally increases between the smallest (Khwai) and largest (Seronga) village in terms of population, Gudigwa presents an exception. Despite being larger in population than Sankoyo, the number of nodes in the whole networks generated f or Gudigwa is 17% lower than the smaller village of Sankoyo. The number of components in the whole network outputs is one in all but two cases, meaning that everyone in those networks is connected to everyone else through some communication pathway. There are two exceptions: there are two components in the Respondent Reported output for Khwai and 12 components for the same type of output in Seronga. In Khwai, 98% of the nodes are within one component, and just seven out of 354 individuals are included in t he second small component. Though Seronga has 12 components, most (72%) of the nodes are found wi thin the largest component; the other individuals are divided among the other 11 components. A line graph for each of the four whole network outputs i s presen ted in Figure 2 1 A s hypothesized, a general trend toward lower network density is evident from the
47 smallest (Khwai) to largest (Seronga) village. Values for Sankoyo and Gudi gwa are similar to one another. The networks created based on only Respondent Re ported ties were predictably different from the other three types of whole network outputs. For each village, a m ean value was calculated for the number of ties, density, degree centralization, and betweenness centralization values associated with the Maxi mum, Majority, and Minimum outputs. These mean values were then compared with the values for the Respondent Reported ties only output (Table 2 4). The difference and percent difference between the averaged values and Respondent Reported value were very hig h and for all but two calculations of betweenness centralization, they were greater than 100%. This indicates that as one would expect network outputs that include i nformation about how alters are tied to one another (Maximum, Majority, and Minimum) prov ide considerably more tie information overall than Respondent Reported outputs. The extent to which a whole network demonstrates centralization is important to understanding information flows (Figure 2 2 ) Data for degree and betweenness centralizations a re included in Tables 2 3 and 2 4 The trends in these two data sets are similar across each of the network output options; therefore, the Majority whole network output is presented here since it consistently falls between the Minimum and Maximum output va lues and is a reasonable approach for determining what should be included in the whole network. As commun ity size increased degree centralization decreased, meaning that there was less dominance by a small set of highly connected individuals. When values for degree centralization are lower, it indicates that the connections among nodes in the networ k are more equally distributed. Betweenness centralization
48 s hows the opposite general trend, with the values for betweenness centralization increasing with comm unity size (Khwai=6.72%, Sankoyo=8.58%, Seronga 12.74%). This means that as villages get larger, a smaller proportion of the people lie along the shortest paths within the network and it suggests that fewer members of larger networks are in positions to s erve as information conduits. The exception to the trend is Gudigwa, which is the second largest village but has the lowest betweenness centralization value ( 3.73%). In Gudigwa, there are more people who have similar chances of being along the shortest pat hs between other people in this network. Network characteristics may be impacted by the extent to which people name people from within their own village or from other places. Respondents in Khwai and Gudigwa named a significantly higher number of people f rom their own village, on average, than people from Sankoyo or Seronga (Figure 2 3). People from Seronga were the most likely to name people who lived outside of their village. Visualizations of the whole network outputs are useful for qualitatively compa ring village communication networks. All four of the Majority whole network outputs are h ighly dense (Figure 2 4 ). The visualization for the smallest village, Khwai (A), is a compact sphere, which is associated with a highly dense network. There are only a few small sub group s and pendants, which are individual nodes that are linked through a single tie to the network. The Sankoyo network (B), though not much larger in population size, appears to i nclude many more sub group s than Khwai external to the dense center of the network. The network includes o ne sub group that appears to be fairly dense itself (located at 6:00 on the diagram) Sankoyo and Gudigwa (C) are similar in appearance, which correspond s with the similarity in values for density,
49 degree centr alization, and other network measures between these two villages presented in Table 2 3 and Figures 2 1 and 2 2 The visualization for Seronga (D) reflects the lower density and degree centralization of this network compared to the others. Though all of th e nodes are still connected to one another, these are less densely connected, and there appear to be a number of smaller sub groupings evident within this larger connected whole. In each of the visualizations, betweenness centrality was added as an attrib ute and used to determine the size of each node. Larger nodes indicate greater betweenness centrality. In Khwai and Sankoyo, the smallest villages, one node is clearly larger than the rest within their respective villages. Seronga also shows one dominant n ode; however, there are several additional less large but sizeable nodes in terms of betweenness centrality. As is also indicated by the data presented in Figure 2 2 Gudigwa demonstrates the le ast betweenness centralization, in that it has several similar ly large and medium sized nodes. As expected given the differences in network measures for Majority and Respondent Reported networks, visualizations for Respondent Reported networks appear much less dense than their Majority counterparts (Figure 2 5 ). Khw ai (A) is composed of one large and one small component. Betweenness centrality for Khwai is more distributed among multiple nodes in the Respondent Reported network than the Majority network. The Sankoyo (B) and Gudigwa (C) network visualizations are simi lar in appearance to each other, as was the case for the Majority networks in Figure 2 4 There are 11 small and one large component in the Seronga (D) network visualization.
50 The largest node by betweenness also appears to be central within the large compo nent. Discussion Impacts of Community Size on Network Characteristics Evidence from these rural villages indicates that as hypothesized, community size impacts communication network characteristics Speci fically, as group size increased network densi ty and degree centralization decrease d In other words, smaller communities tend ed to have denser networks, where there are more direct contacts among members. There is also more redundancy in smaller networks m any people are connected to the same set of p eople. As communi ty size increased the resulting networks tend ed to be less centralized around a few highly connected individuals and connections were more evenly distributed among network members. This makes intuitive sense, since someone i n a less popu lous village has a greater chance of being in direct contact with a larger proportion of their whole network than someone in a more populous village. Also, in this particular study region, community population size tends to be generally correlated with geo graphic size, making a higher proportion of direct contacts with other network members more logistically possible in smaller communities. People simply have a greater chance of bumping into one another when they are walking and working in a smaller area. Density and degree centralization values may also be higher in small er villages because both informal gatherings and formal meetings tend to take place in fewer, more centralized locations. F ormal meetings tend to be held in one central location in smaller villages like Khwai and Sankoyo, whereas meetings may be more distributed when there is a ward system like in the larger communities of Gudigwa and Seronga In these
51 larger communities, wards function effectively as smaller villages within villages Many matters are still discussed by the community as a whole in centralized community wide meetings; however, according to key informants many people opt out of these meetings in favor of attending more localized, ward specific meetings with their individual h eadmen (sub chiefs) or other leaders. Larger communities typically rely on a more distributed syst em of information dissemination. This is likely to occur both intentionally (i.e., through the traditional authority ward system) and unintentionally, as even smaller sub group s are likely to self organize for any number of reasons including but not limited to a sha red gender, kins hip relationship, or water source It appears that in Seronga, more people may be involved as information conduits to reach the dif ferent ward sub groups. The low b etweenness centralization values for Gudigwa in dicate th at information exchanges are likely to be less centralized and more widely distributed in this village than others in the study Many people have similar betweenness c entralization values and are therefore similarly able to take advantage of the shortest paths for information flows, despite the relatively large size of this community. On the other h and, despite a ward system similar to that in Gudigwa betweenness centr alization is highest in Seronga. This indicates that a relatively small er proportion of people lie on the shortest paths for information flows in Seronga than in any other village in the study and therefore information may take longer to reach people in d ifferent parts of this network. A comparison of the village and social network characteristics of Gudigwa and Seronga reveals that se veral factors may be c ontributing to different betweenness centralization values First, despite its relatively large size, the whole network outputs
52 for Gudigwa are made up of a relatively small number of actors. As Table 2 3 shows, respondents in Gudigwa named a much smaller set of people overall 417 nodes This means that it is more likely that any one individual will fall on the shortest path within the smaller, denser Gudigwa network ; therefore betweenness centrality will be more equally distributed within the network and there will be a lower betweenness centralization value The ethnic diversity present in Seronga may also contribute to the betweenness centralization results There are many et hnic groups present in Seronga; it is more like a small scale city in this regard This is very different from Gudigwa, which is strongly dominated by just one ethnic group. Ethnicity is further explored as a social variable affecting communication network characteristics in chapter 3. An additional factor that may be affecting these findings is that in Seronga the boundary of the community includes both the main village and several smaller cattle posts These settlements are outside of the main village (they were created as places for people to keep cattl e away from population centers), but they are technically the administrative responsibility of th e village with which they are associated. Information distributed to residents of Ser onga by government agencies, for example, is assumed to reach cattle post residents; however, the chance of these individuals attending formal meetings or casually encount ering individuals in other parts of the network is minimal. Many are at least an hour or more from Seronga by foot, and most residents do not have access to a vehicle. Those cattle post residents who are able to travel into the main town are limited in num be r and may be more likely to fall along the shortest paths within the overall Seronga network.
53 As population increases, it follows that the chance of sub group s forming will also increase. Evidence for this can be seen in the increase in the number of co mponents in Seronga compared with the smaller villages for the Respondent Reported whole network outputs. The formation of sub group s supports the Social Brain Hypothesis (Dunbar 1998). The number of people we know personally, whom we can trust, whom we f has been 150 for as long as we have been a species. And it is 150 because our minds lack the capacity to make it any larger (Dunbar 2010, p. 4) Dunbar sets a maximum numbe r of people an individual may have within his or her network. The actu al number is less important than the idea that a maximum personal network size exis ts no matter the overall community size. This means that as community size increases, people are increa singly likely to form communication sub group s. Also t he proportion of ties that any one individual has in proportion to the total number of possible network ties decreases. Two individuals with the same number of ties in Khwai and Seronga for example, w ould be connected to very different proportions of their respective whole network s It follows that forming sub groups would be more likely in larger villages than smaller ones. Data from this study indicate that, taking these two factors together, increas ing group size is likely to result in an increase in the complexity of the structure of the communication networks (Byrne 1996, Dunbar 1998). Information Flows, Learning, and Adaptation Community size is an independent variable that impacts information flo ws and social organization. The tendency of people to form sub group s, reflected in this context in the ward system of larger villages, for example, has the potential to impact the
54 communication networks of ward members. This may in turn affect the informa tion exchanges, learning, and adaptation that occur since s maller groups have more opportunities for social bridges that build trust and commitment among group members from repeated interactions (Coffe and Geys 2008 ). These exchanges can foster collectiv e action (Putnam 1995, Schneider et al. 2003) and allow groups to adopt innovations more quickly (Rogers 1995) The ward system and other informal sub groupings help people living in larger communities to overcome some communication challenges, but other challenge s remain At the most basic level, information intended for all village residents may have a harder time reaching all people living in larger villages This is particularly the case where face to face communication is required, which is the main a nd most effective form of communication in these rural Botswana communities and similarly remote rural are as of developing countries where cell phon e use is limited or nonexistent and illiteracy rates are high. If people are unable to attend meetings thems elves, they must rely upon others to relay messages in person While this was not specifically studied here, any relayed message has the potential for distortion or misinterpretation when information flows through intermediary individuals to the eventual receivers and interpreters of that information. Field observations i ndicate something akin to the telephone game can sometimes occur, in which a specific message that is passed through a series of individuals becomes a very diff erent message than the origi nal. Smaller networks are denser and contain more direct connections, making the information transmission lines shorter. The extensive oral tradition in rural Botswana aids greatly in the transmission of information; however,
55 each individual along the info rmation path brings pre conceived notions and interpretations of the information that have the potential to influence the message. Despite the challenges of being part of a larger community larger size may provide some important advantages and opportuniti es for learning and adaptation The number of people included in the Seronga whole n etwork was three times th e number in Khwai and when all ties are considered (as with the Maximum, Majority, and Minimum whole network outputs), everyone in Seronga is conn ected through some path to everyone else Some of these conn ections are closer and more direct than others, which provide opportunities for people to both gain from being part of their own immediate social circles and from weak ties in other parts of the n etwork. This can lead to a greater diversity of information, new perspectives and the potential f or innovation and adaptation (Granovetter 1973, Burt 2004, Coffe and Geys 2008). Weak ties are far more difficult to find in more densely connec ted village s like Khwai This could make it more challenging to take advantage of a diversity of ideas that could be helpful for innovation and adaptation. That said, community size is one of multiple factors affecting the availability of weak ties and diverse ideas. Sankoyo, for example, is not much larger, but the visualizations of Sankoyo show that it is less dense than Khwai. Geography, and specifically its closer proximity to the regional center of Maun, makes it easier for people to access outside sources of info rmation, including as Figure 2 3 illustrates, people from outside of their village People from Seronga tend to position as a main village in the region, its proximity to other villages, and its access to outside information and internet connections.
56 Gudigwa appeared not to follow the trends related to community size and social network characteristics exhibited by the other three villages. Despite its relatively large si ze, the whole networks for Gudigwa had fewer nodes and ties, lower density, and lower degree and betweenness centralization than Sankoyo. The betweenness centralization value was the lowest for Gudigwa of all of the villages, and the visualization similarl y shows that there are many people with about the same betweenness centrality values. Betweenness centrality was more distributed in this village, which may indicate that other factors are particularly important to how communication networks work in this v illage. One factor affecting these results may be, like in the case of Sankoyo, the geographic location of the village. Whereas in Sankoyo proximity to the regional center could provide access to people and ideas outside of the village, the opposite may be the case in Gudigwa most remote of the villages in the study. Another possible factor is the fact that Gudigwa is strongly dominated by one ethnic group, which may lead to a more tightly connected network. Implications The empirical evidence presented here contributes to the growing body of knowledge about social network measures and their relationship with variables critical to learning and adaptation Information is an essential component of the learning process. Access to information is affected by here, the size of the community in which a person lives. It is clear that communication strategies need to be designed to take advantage of the benefits of both small and large groups Small villages like Khwai and Sankoyo can take advantage of greater face to face contact amo ng village residents; however, to enhance the potential for innovations important to adaptation, it may be important to
57 deliberately bring in outside information and perspectives to a void the pitfalls of redundant ideas and homogeneous thinking often associated with densely connected networks Yet, this presents challenge and complexity, since outside information may not be trusted. Since it is easier for people to meet face to face in smaller communities, these opportunities should be continued and enhanced, with individuals able to deliberate and learn from one another to access and integrate new information. In larger communities like Gudigwa and Seronga, more attention can be paid t o disseminating information and engaging people directly through the wards. Additional attention could be focused on communicating with sub groups that may be less central within the village networks and individuals who may not be able to attend meetings a nd access information that way Within larger villages o pportunities for information exchange, deliberation and participation within sub groups operating at smaller scales has the potential to build trust within these sub groups and allow community membe rs to become more directly informed and engaged. Data from this study suggest that c ommunity size is an important variable with the potential to affect information dissemination, learning, and adaptation. Additional factors that also likely affect these o utcomes are variables contributing to intra community diversity such as gender and ethnicity, geography, and access to outside information sources. Networks are often more complex than they may first appear, with nuanced interactions related to particular attributes of individuals. The role that ethnicity and gender may play in communication n etworks is explored further in c hapter 3. Understanding how community size and other variables affect social networks is extremely important when trying to determine how to enhance adaptive capacity in light
58 of current and projected environmental changes. It is also critical as we consider how messages are conveyed to and move through communities in general. A wide range of fields from public health to emergency respon se and beyond can benefit from research investigating social networks and their implications.
59 Table 2 1. Population estimates for study villages Village 2001 Census total population (including children) P opulation reported by local reside nts and other sources Adult population counted by researchers Khwai 2 50 300 ^ Sankoyo 372 350 600 158 Gudigwa 600 800 290 Seronga 1 641 1 500 5 000 1 335 Data specific to this village are not available in official 2001 B otswana Census documents. Residents of these villages were included in aggregated population ^ Census was not conducted because a roster of village residents was provided by th e community based organization. Table 2 2. Number of personal network interviews used in each village to create whole network outputs Village Number of completed surveys (n) Khwai 3 1 Sankoyo 31 Gudigwa 3 5 Seronga 39
60 Table 2 3. Social n etwork measures for each type of whole network output in each v illage Villages listed in order of population size with the smallest village first Village Whole network output Number of nodes Number of ties Density Number of components* Proportion of nodes in largest c omponent Degree c entralization (%) Betweenness centralization (%) Khwai Maximum Majority Minimum Respondent Reported 354 354 354 354 23,604 22,098 21,256 1,236 0.1889 0.1768 0.1701 0.0099 1 1 1 2 1.00 1.00 1.00 0.98 48.67 45.33 38.34 8.95 6.82 6.80 6.55 20.76 Sankoyo Maximum Majority 493 493 26,628 26,082 0.1098 0.1075 1 1 1.00 1.00 43.79 39.33 8.91 8.48 Minimum 493 25,940 0.1069 1 1.00 37.55 8.35 Respondent Reported 493 1,514 0.0062 1 1.00 6.71 16.57 Gudigwa Maximum Majority Minimum Respondent Reported 417 417 417 417 21,292 19,332 18,636 1,442 0 .1227 0.1114 0.1074 0.0083 1 1 1 1 1.00 1.00 1.00 1.00 39.26 36.54 32.36 7.60 3.68 3.75 3.75 12.91 Seronga Maximum 1,106 38,544 0.0315 1 1.00 18.04 12.57 Majority 1,106 38,426 0.0314 1 1.00 17.87 12.83 Minimum 1,106 38,426 0.0314 1 1.00 17.87 12.83 Respondent Reported 1,106 2,292 0.0019 12 0.7 2 3.16 30.85 Components calculated using following settings in UCINET (Borgatti et al. valued at 2.
61 Table 2 4. Comparison of network measures across different whole network outputs for each village. Villages listed in order of population size with the smallest village first Village Network measure (1) Mean of Maximum, Majority and Minimum (2) Respondent Reported value Difference between (1) and (2) Percent d ifference (%) Khwai Number of ties Density Degree centralization (%) Betweenness centralization (%) 22,319 0.1786 44.11 6.72 1,236 0.0099 8.95 2 0.76 21,083 0.1687 35.16 (14.04)* 179.01 178.99 132.53 102.15 Sankoyo Number of ties Density Degree centra lization (%) Betweenness central ization (%) 26,217 0.1081 40.22 8.58 1,514 0.0062 6.71 16.57 24,702 0.1019 33.52 (7.99) 178.16 178.30 142 .84 63.54 Gudigwa Seronga Number of ties Density Degree centrali zation (%) Betweenness centralization (%) Number of ties Density Degree centralization (%) Betweenness central ization (%) 19,753 0.1138 36.01 3.73 38,465 0.0314 17.92 12 74 1,442 0.0083 7.60 12.91 2,292 0.0019 3.16 30.85 18,311 0.1055 28.46 (9.18) 36,173 0.0295 1 4.76 (18.11) 172.79 172.82 130.36 110.40 177.51 177.20 139.99 83.07 Parentheses indicate that the difference is negative, which is the case when the Respondent Reported value is greater than the m ean of the Maxi mum, Ma jority, and Minimum
62 Figure 2 1 Network d ensity f or four types of whole network outputs (Maximum, Majority, Minimum and Respondent Reported ) for each village Villages are presented in orde r of increasing population size from left to right. 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 Khwai Sankoyo Gudigwa Seronga Network Density Village Maximum Majority Minimum Respondent-Reported
63 Figur e 2 2 Degree a nd betweenness centralization s for o utputs for each village Villages are presented in order of increasing population size from left to right. 0 5 10 15 20 25 30 35 40 45 50 Khwai Sankoyo Gudigwa Seronga Percent Village Degree centralization Betweenness centralization
64 Figure 2 3. Mean number of people (alters) named from the same v illa ge as the respondents M ea n was calculated for all of the alters named in each village. Original values were: village; 1=Alter lives in same v illage as respondent.
65 A B C D Figure 2 4 Visualizations of whole n Villages included in increasing population size: A) Khwai, B) Sankoyo, C) Gudigwa, and D) Seronga.
66 A B C D Figure 2 5 Visualizations for whole network outputs for Respondent Reported Nodes sized by betweenness. Villages included in increasing population size: A) Khwai, B) Sankoyo, C) Gudigwa, and D) Seronga
67 CHAPTER 3 IMPORTANCE OF GENDER AND ETHNICITY IN THE STRUCTURE AND OUTCOMES OF RURAL CO MMUNICATION NETWORKS Introductory Framework Participation and Community Based Conservation The idea that local people should be included as participants and decision makers in the management of natural resources has gained prominence since the emergence of community based con servation in the late 1980s (IIED 1994, Western and Wright 1994, Schmink 1999, Adams and Hulme 2001). Community based conservation, which in Africa tends to take the form of community based natural resource management (CBNRM), came as a counter narrative t paradigm, which advocated the exclusion of people from lands through the establishment of parks and from political processes through top down policy making (Adams and Hulme 2001, Mulder and Copollilo 2005). The new pa radigm, embodied in CBNRM and other community based strategies, shifted toward the inte gration of people into decision making and the promotion of managed use of natural resources to help meet both conservation and development goals (Brundtland 1987, IUCN/ UNEP/WWF 1991, Abbott et al. 2001, Adams and Hulme 2001, Artnzen et al. 2003, Mulder and Copollilo 2005). The implementation of c ommunity based conservation approaches varies widely around the world. P rograms are generally focusing on the creation of local ly specific conservation solutions and have goals of empowering local human users of natural resources through participation (Agrawal and Gibson 1999, Ainslie 1999 Agrawal 2000, Kellert et al. 2000, Twyman 2000, Mulder and Copolillo 2005 ). This narrative suggests an idealized version of conservation and development; however, the implementation of CBNRM programs presents challenges and has
68 provoked criticism ( Agrawal and Gibson 1999, Leach et al. 1 999 Newmark and Hough 2000 Arntzen 2003, Musumali et al. 2007 Shuster 2007 ) One critique levied by Agrawal and Gibson (1999) is that using the term notion that those participating in a particular community based conservation program are part of a small, integrated unit chara cterized by homogeneity in social structure and norms. Communities are not homogeneous (Butcher et al. 1993) ; they are most often comprised of multiple sub group s that represent intra community diversity in terms of natural resource interests, perceptions, and power (Carlsson and Berkes 2005, Nygren 2005, Crona and Bodin 2006). The concept of community as a unified entity therefore natural resource interests and their a ccess to and engagement in decision making processes (Agrawal and Gibson 1999) articipation has become development orthodoxy ( Cornw all 2003, p. 1325), yet notions of full and equal participation are often challenged when they come into conflict with h istorical, traditional, and social factors that affect the participation of some community members O ften, participatory approaches at the local level tend to be overshadowed by existing inequalities, power structures, and political and economic forces tha t influence management decisions (Mohan and Stokke 2000). Hence, p articipation often involves the elite in a particular communit y and not the general populace, and already marginalized groups like women and ethnic minorities are most vulnerable to exclusio n (Nelson and Wright 1995). Addressing intra community diversity and the social, cultural, and political dynamics associated with this are important to understanding how people engage with conservation (Broscius et al. 2005, Broscius
69 2006, Redford 2011), a nd the extent to which they receive development benefits through community based conservation programs (Agrawal and Gibson 1999). Importance of Communication Communication and access to information is essential to participation; i t a ffect s t ransparency accountability, and the ability of people to effectively participate in decision making processes (Mandondo 2000). Information exchanges are both critical to conservation efforts (Jacobs o n 1999) and inseparable from their cultural contexts (Melkote and S teeves 2001). Social, cultural and political dynamics shape the way information is exchanged, the extent to which certain voices are heard or not heard, and the ways in which traditional and local ecological knowledge are valued, shared, and incorporated into conservation decision making (Berkes and Turner 2006, Davidson Hunt 2006). CBNRM establishes a governance framework intended to promote participatory decision making for the achievement of conservation and development goals. All residents of villages involved in CBNRM are stakeholders in the decision making process and should, by participatory development standards, have a voice in this process. Participation in governance presents many potential benefits to rural residents. Acco r ding to Hickey and Mo han (2004, p. 3), the heart of participatory processes is the Research elsewhere suggests that this can allow people to engage with environmental information firsthand, learn from others the knowledge of others (Schusler et al. 2003, Keen et al. 2005). One social dynamic with the potential to affect communication in community based associate with others who
70 share particular characteristics such as gender or ethnicity (McPherson et al. 2001). This pattern of social organization tends to facilitate communication and reduce conflict since members of a sub group may have common experien ces or backgrounds (Reagans and McEvily 2003, Crona and Bodin 2006). Homophily may foster intra group communication; however, people without connections beyond these immedia te social networks will tend to only learn from those closest to them and may not r eceive new potentially important information from more distant parts of their larger social system (Granovetter 19 83, Levin and Cross 2004). Further, p ower structures at the community level may privilege one group over another such that marginalized group s are effectively excluded from both communication and participation ( Mohan and Stokke 2000 ). This marginalization may reinforce existing social structures, and it may also prevent the incorporation of diverse bodies of knowledge held by different resourc e user groups ( Charnley et al. 2007) It is widely recognized that traditional and local ecological knowledge held by groups engaged in natural resource based livelihoods can be important to conservation and resource management (Colding 1998, Johannes 1998 Berkes et al. 2000, Becker and Ghimire 2003, Moller et al. 2004, Fraser et al. 2006, Charnley et al. 2007, Rist et al. 2010). Resource users often have intimate and specific knowledge about the resources with which they interact directly (Johannes 1981, Hunn et al. 2003), and different knowledge is held by different resource user groups (Ghimire et al. 2004). This diversity of knowledge may not be effectively integrated, however, if certain groups or individuals are excluded from participatory processes. Attention should therefore be paid to the knowledge brought by different groups, and how it is integrated (or not) into management processes (Charnley et al. 2007).
71 Gender, Ethnicity, and Natural Resources Men and women often represent different sta keholder interests with regard to the (Schmink 1999, p. 6). Women in poor rural are as are often responsible for family subsistence ; in female headed households they may be the sole economic providers (Agarwal 1989). In patriarchal systems, inequalities expressed in policy and in differential access to markets may create social, politica l, and economic disadvantages for women (Kabeer 1994, Schmink 1999). These inequalities can hinder progress toward community based conservation goals related t o participation and empowerment if they are not attended to specifically (Mayoux 1995, Schmink 19 99). Important information about the environment as a whole, gained by women through their multiple interactions with the environment (Rocheleau et al. 1996), may be excluded from community based conservation if women are not intentionally incorporated int o decision making processes (Schmink 1999). Though there has been increased attention to gender over the past two decades, it remains an important variable in natural resource management research in rural contexts (Pfeiffer and Butz 2005). Gender and ethn icity are inextricably linked, since gender roles are affected by the ethnic context in which they are defined (Davison 1996). Like gender, e thnicity also affects how people use and value natural resources and how different groups engage in decision making (Mompati and Prinsen 2002). Acknowledging that many designations of ethnicity are externally imposed, politic ally motivated, and subjective (Davison 1996), investigations about how different groups vary in their engagement with natural r esources and their management are important ( McAvoy et al. 2000, Bengston 2004).
72 Differences in traditions, social mores, and language create obstacles to communication and understanding, and histories of exploitation often create profound distrust of government institutio ns and Bengston 2004, p. 48) Further, minority groups may experience physical segregation, political exclusion and prejudice that affect their ability or interest in participating in community processes (Mompati and Prinsen 2002). The literature suggests that gender and ethnicity are among the social variables that can importantly affect natural resource use, communication, and participation in community based conservation programs such as CBNRM. This may especially be the case in remote rural contexts like t he Okavango Delta of Botswana where people often rely more upon their personal networks for information than mass communication sources. Since these personal relationships are highly influenced by social dynamics, attention is n with one another. Researchers addressed the following questions: To what extent do people tend to communicate more with people of their own gender or ethnicity? How may these social dynamics affect communication and participation in CBNRM? The paper begins by providing an overview of the research context and environ mental communication landscape; it then offers a brief review of literature linking gender and ethnicity with natural re source use and communication specific to this research context Researchers then present social network interviews and observation data from field studies, and discuss how these results might contribute to improving communication and participation in CBNRM
73 Context for this Study Research Setting Botswana is a land locked country in southern Africa comprised mainly of semi arid savanna. The Okavango Delta, within the Ngamiland District located in the northwestern part of the country, contrasts with the s urro unding Kalahari Desert The Okavango system is a flood pulsed wetland with a surface area of 5,537,400 hectares, the largest inland delta in the world and the only one in sub Saharan Africa. It is a Wetland of International Importance as designated by the Ramsar Convention, characterized by high biodiversity including several endemic an d endangered species (Ramsar 200 0). The human population density in Botswana is among the lowest in the world (UN 2009). The Okavango Delta is characterized by a low populat ion density of 0.5 1 person/km ( CSO 2005b ); there are appro ximately 26,000 households and 125, 000 people 1 ( CSO 2008b). These numbers include the regional population center of Maun located at the lower reaches of the Delta system, which houses approximately 50,000 people according to 2001 census data (CSO 2008a). The population of Botswana as a whole is divided approximately equally among men and women (UN 2009) T he gender ratio of the Okavango Delta is minimally skewed toward women (CSO 2005b) who comprise approximately 52 to 53 percent of the population in this region ( CSO 2008a, CS O 2008c). The Government of Botswana has not asked questions about tribal affiliation or race since independence and the 1961 census. A general question about languages spoken w as included in the 2001 and 2011 census es ; however, the non 1 The Central Statistics Office of the Government of Botswana divides the country two ways for census purposes by administrative district and by census districts. The areas included are slightly different. Figures given here reflect approximate numbers based on numbers given for both methods.
74 government al organization Reteng which represents minority ethnic groups, contends that this under estimates the numbers of non Tswana speaking groups. A question about mother tongue language was announced for the 2011 census but was not ultimately included in the questionnaire (Gaotlhobogwe 2011). The natural features of the region contribute to local livelihoods. P eople of the Okavango Delta rely on water for consumption, household use, rainfall for subsistence crop cultivation fishing in Delta waters, livestock, and to sustain wildlife populations essential to local tourism based livelihoods. This critical resource, however, is projected to decrease in time. Climate change models vary but gener ally agree that water resources could decrease by as much as 25% over the next several decades in semi arid areas including southern Africa (IPCC 2007). Anticipated changes in precipitation will affect wildlife abundance, locations, and migration patterns. This has the potential to increase conflicts between humans and wildlife as competition for limited resources increases. Furthermore, many communities in the region receive benefits from wildlife tourism through CBNRM. C hanges in wildlife population densi ties and movements could greatly impact livelihoods in communities where benefits from CBNRM are important. Fieldwork was conducted in 2008 and 2009 2010 in several rural villages engaged in CBNRM in the Okavango Delta The villages of Khwai, Sankoyo, Gud igwa, and Seronga vary in location (Figure 1 1) and dominant ethnic groups (Table 3 1) ; all based organizations (CBOs) responsible f or implementing CBNRM at the local level Khwai and Gudigwa are dominated by Bugakhwe pe ople, also referred to as Bushmen. The largest ethnic group in Sankoyo is Bayeyi. Bayeyi are also one of two dominant groups in Seronga, along with the
75 Bambukushu. Seronga is the largest village and has considerable diversity and several additional ethnic group s represented within the larger population. Khwai and Sankoyo are relatively small in size, and they are single village CBNRM programs. In contrast, Seronga and Gudigwa are two villages in a multi village CBNRM program. They are the most distant from one a nother in the Okavango Community Trust (OCT), a CBO that also includes the villages of Beetsha, Eretsha, and Gunutsoga. Among the villages included in this study, Gudigwa is second in population to Seronga. Environmental Communication Background T he focus on personal networks in this study can be enhanced with a brief overview of the environmental communicat ion context based on observations and interviews. Direct contact is rare between residents of this region and the scientists or government officials wh o possess environmental data due to the remoteness of these areas and limited funding for outreach. Environmental information that comes into rural villages from these sources is sometimes conveyed in formal meeting settings through presentations to group s of rural residents at the kgotla the traditional meeting place in each village, or it may be provided directly to the kgosi or chief who then passes this information along to the community at large during kgotla meetings. Many factors may influence th e dissemination and receipt of this information among community members, of kgotla meetings, the chief of the original information, the transmission and interpretation of the information, the means of information transmission (e.g., language used), and the extent to which information is passed through personal networks outside of meetings.
76 Environmental commun ication can, in some cases, be supplemented by mass media sources, though this is also inconsistent. Radio broadcasts can be accessed by many, though broadcasts usually provide generic information more likely to be relevant to the nation as a whole than to the Okavango Delta region. Messages about water conservation, and some national environmental policies, were broadcast on the radio during fieldwork. The closest regular (weekly) newspaper is printed in the regional center of Maun, which is between 85 and 420 km away from the villages included in this study (see Table 1 1). Availability of this and other news sources is sporadic and untimely, particu larly in more remote areas. I t may take weeks for the Maun newspaper to reach the most remote village of Gud igwa, if it arrives at all. Remoteness is an important factor in environmental communication. Gudigwa is the most disadvantaged of the villages in this study in this regard; it is about a 12 hour drive from Maun on paved and sand roads. Seronga, though lo cated nearly as far away from Maun, is closer to the ferry river crossing at Shak a we, a village with electricity at the end of the paved road access to Maun. Seronga is also the largest villa ge in the Okavango Panhandle. It has primary and junior secondary schools, an airstrip, police station, and health clinic. Electricity is supplied to some public buildings by generators. Landline as well as cellular phones and some Internet access make communication far easier in Seronga than Gudigwa, where none of thes e electronic options are available. Some environmental messages are communicated to the more remote villages in OCT through word of mouth, particularly when vehicles go between the villages, or via two way radios based in the OCT offices in each village. T hough available, these means of communication are perceived as inadequate by many Gudigwa residents, who
77 frequently comment about their lack of information about the CBNRM program and other topics of concern. Remoteness is less of an issue for residents of villages closer to Maun like Sankoyo and Khwai (about 1.5 and 3 hours of driving time, respectively). Some residents tend to travel back and forth from Sankoyo and Khwai to Maun fairly regularly, which gives them more firsthand access to public informatio n resources. This differential access to information may contribute to inequities and create or exacerbate existing power dynamics. In addition to environmental information brought into rural communities from outside sources, knowledge about natural resour ces is also generated locally. An alternative set of values and information about local environmental conditions is held by those living in the place, and these community members may be key sources of natural resource information. By tradition and culture elders are highly respected in Botswana society ; they are perceived as valuable sources of information about natural resources. Local elders and others relying directly on natural resources in their rural environments have a sense of changes taking place o ver time and possess specific information about that particular place. Though it is clearly valuable, th ere is no institutional process in place to specifically integrate this knowledge into decision making about natural resources and CBNRM Gender and Et hnicity in Context Building upon the general literature presented at the beginning of the paper, this section delves more specifically into how gender and ethnicity have been shown to interact with natural resources and communication in the Okavango Delta, Botswana
78 Gender Natural Resource Use, and Communication Gender has been an active part of development discourse since the 1970s. Ester based on research in sub Saharan Africa, increased understan ding about the multiple roles of women beyond their reprod uctive roles related to child bearing and household activities. This book brought global attention to the role of women in economic activities like agriculture, and it was one of several factors cont ributing to the conception of the development sub field centered development approach and placed specific emphasis on equality of the sexes in economic development project s (Moser 1989, Kabeer 1994). economic systems but did not question the social structures in place that caused women to be seen and treated as unequal to men. The Gender and Dev elopment (GAD) approach emerged in the late 1970s and 1980s as an alternative to WID GAD focuses on the dynamics of gender relationships as central to social processes and en within a particular social context (Spring 1993, March et al. 1999). Understanding gender dynamics is valuable to understanding about how gender interacts with natural resource use and communication in the Okavango Delta of Botswana. At the broadest le vel, Botswana is a patriarchal society. Cassidy (2001, p. 11) states: Men dominate society in Botswana. Positions of leadership are, by tradition, held by men. Men are always assumed to be the head of the household and rights to land are mainly given to m en Nevertheless, there is growing awareness in Botswana of the need to address both legal and social inequalities facing women (Republic of Botswana et al. 1998). Both
79 Government and NGOs have taken proactive steps in reversing the disadvantages and discri mination faced by women. Labour and Home Affairs that addresses challenges and opportunities related to gender at the policy level. Changes to the status and roles of women in Botswana have occurred in the past several decades, though this is more observable in urban than rural settings where strong traditional roles tend to persist (Cassidy 2001). At the local level, gender role differentiation with regard to natural resources has historical roots and present implications. By tradition, dominant male activities inc luded hunting and fishing; women collected and used veld products 2 for food and crafts (Cassidy 2001, Mbaiwa 2008). This role differentiation creates opportunities fo r people to communicate with those of their own gender. Government of Botswana policies have seasons); however, policies have also changed the relative value placed o n different natural resources through CBNRM. Wildlife can bring communities engaged in CBNRM millions of Botswana Pula per year in international tourism income, which results in a resources (veld products) (Cassidy 2001). Meetings about CBNRM take place in the kgotla, a Batswana political institution where men traditionally came together to discuss matters of community impo rtance (Denbow and Thebe 2006). This history contributes ch allenges to what are meant to be inclusive participatory processes which the sway of village politics. Social factors such as ethnic group and whether one is a 2 collected from forested areas.
80 (C assidy 2001, p. 27). Ethnicity, Natural Resource Use, and Communication The concept of ethnicity is complex (Mazonde 2002) and a in Botswana (Davison 1996, p. 20 ) While this paper does not attempt to problematize ethnicity in its con ceptual entirety, it does explore how concepts of ethnicity may reflect socio political challenges faced by certain groups (Mompati and Prinsen 2002, Nyati Ramahobo 2002). I has often been us ed by Europeans and Westerners to describe a group that share beliefs, language, customs, and other observable characteristics (Banks 1996, Shermerhorn 1996). Though this designation often oversimplifies the variation that exists within such a group, the t was used during colonial times and carries a pejorative connotation (Davison 1996). At rinsen 2002, p. 95). Perceived differences can influence social dynamics, power relationships (Mompati and Prinsen 2002), and conservation efforts (Bengston 2004 ) Though imperfect, one way that groups may be differentiated in Botswana is by mother tongue language, which is particularly relevant in a discussion about communication. Setswana from the British in 196 6, the new government decided that the language Setswana the indigenous language spoken by the majority group but only one of many indigenous languages in Botswana would be a national language taught in schools, and English
81 would be used for official gov ernment business (Mooko 2006). Differentiation by language was seen as potentially divisive and damaging to the new nation, and other languages that had previously been taught in school were banned from education, the media, and the court system (Nyati Ram ahobo 2002). Since independence and through today, a citizen of Botswana is called a Motswana and all citizens collectively are Batswana, 3 Ramahobo 2002, p. 17). Though Setswana is the mother t ongue language of about 2001), 4 this creates confusion and reinforces power dynamics as the Setswana speakers are but one of many ethnic and language groups in Botswana (Cheb anne 2002). There are in fact two major language groups in Botswana Bantu and Khoisan. Setswana is one of more than 20 Bantu languages in Botswana, including distinct language traditions of the Batawana, Bay eyi, Bagalagadi, and Bambukushu which are domin ant in different parts of the Okavango region (Cassidy 2001, Bolaane 2004, Denbow and Thebe 2006). People of the distinct Khoisan language group preceded the Bantu in what is now Botswana. This language group is characterized by distinctive clicks and a di fferent grammatical structure than the Bantu languages. Speakers of the Khoisan language group (known generally as San Bushmen or Basarwa) are believed to be the first inhabitants of Botswana (Mazonde 2002, Denbow and Thebe 2006, 3 Tswana. 4 The Government of Botswana does not colle ct statistics by ethnic group (Mompati and Prinsen 2002, Nyati Ramahobo 2002, Gaotlhobogwe 2011). Population numbers presented here are derived from the studies cited.
82 Mooko 2006). In the Okav ango Delta, speakers of the Khoisan language called Bugakhwe are among those sometimes referred to as due to their proximity to riverine systems, separating them from those Khoisan language groups residing in the desert areas o f Botswana (Mooko 2006). The Bushmen comprise the dominant group in Khwai and Gudigwa villages. These groupings are of great consequence when contemplating communication about natural resources for two main reasons. First, language differences can create barriers to communication at several levels and create or reinforce power dynamics (Mompati and Prinsen 2002). Natural resource information providers coming from outside a community, such as government officials, usually present information in public meeti ngs in English or Setswana (Mooko 2006). State radio and newspaper media convey information only in Setswana and English (Nyati Ramahobo 2002). Those community members who might benefit from this information but speak a different mother tongue may not unde rstand the information being conveyed or feel empowered to participate (Mompati and Prinsen 2002). Within a community, one ethnic group may be most populous and/or powerful in a village and tend to conduct business in that language; however, there is ofte n a diversity of languages spoken (Mompati and Prinsen 2002). In places where multiple languages are spoken, multilingual fluency is also common (Denbow and Thebe 2006); however, it is not universal, and both formal and informal communications may occur in a language exacerbated by physical stratification evident in many rural villages, where subordinate
83 ethnic groups are in separate areas from dominant groups, and differen t languages are used for regular communication (Mompati and Prinsen 2002). importance on cattle rearing and agro pastoralism ha ve traditional roots and important implications for natural resource use. Cattle are a sign of status and wealth among Batswana citizens. This originates from the cultural value placed on livestock and farming by Bantu groups including t he Ba t swana Bayei, and Bambuk u shu. The importance of cattle is illustrated by Paul Mmolotsi Rantao (2006, p.35): The mooing of a cow, the bellowing of an ox or a bull, and even the braying of a donkey are the sound of life in Tswana society. This is preci sely so because livestock provide every basic need of a Motswana food, clothing and shelter. There is virtually nothing that a Motswana ever does that does not have something to do with the cattle (e.g., feeding, clothing, mobility, celebrating, recreation burial as well as demonstrating friendship and solidarity). This strong emphasis on livestock herding is evident in Seronga, where Bayei and Bambukushu groups are dominant (Mbaiwa 2004). Given their emphasis on agro pastoralism major natural resource co ncerns include the availability of water and grazing vegetation, and wildlife is a problem for livestock (see c hapter 4). Of cattle post settlements, where most herders reside, are officially par t of Seronga but located along a sand road hours away from the main part of town by foot. Some cattle post residents move between these more remote dwellings and family homes in Seronga proper, which has the potential to facilitate information flows. This is limited, however, and the chief responsible for reaching this geographically distributed population expressed concern about the difficulty in reaching everyone.
84 name reflective of the different livelihood strategies historically undertaken by Bushmen who traditionally focused on hunting wildlife and gathering veld products (Phuthego and Chanda 2004), focus on hunting and gathering veld products is inferior to the dominant cattle paradigm. of all gro independence land policies that displaced Bushmen from their traditional lands, combined with the institution of regulations and CBNRM programs that restrict game hunting, have reduced or eli minated legal hunting options for many Bushmen (Bolaane 2004). In some cases, this has led to conflicts over natural resource management, and in others, to shifts away from traditional hunting in the livelihood strategies (Mbaiwa 1999, Bolaane 2004, Mbaiwa 2005, Mbaiwa 2011 Mbaiwa et al. 2011 ). Most Bushmen in the Okavango Delta are now located away from their original lands and have adapted their livelihoods to current policies The Government of Botswana established Gudigwa through the Rural Area Devel opment Programme (RADP) in 1987 and encouraged Bushmen to settle there and take on more sedentary livelihood activities (e.g., farming), by offering incentives including a borehole water source, health clinic, and primary school (Bolaane 2004, Cohen 2008). Though there is ethnic diversity throughout OCT, Gudigwa is the only one of the five villages in OCT where Bushmen dominate.
85 Further south in the Delta, Bushmen who had long lived in different parts of what became Moremi Game Reserve were relocated to no rth of the Moremi gate under the RADP when the Reserve boundaries were plotted in 1963. The settlement was moved again to the current village of Khwai when the Reserve boundaries were extended in 1992 (Bolaane 2004). Khwai residents no longer have access t o natural resources within the park and are prohibited from keeping cattle in the village to avoid the potential of spreading disease s ( e.g., cattle lung disease, foot and mouth disease) to cattle intended for sale in European markets Though some people k eep cattle in other places, livestock herding was not a dominant livelihood strategy among the Bushmen before settling in Khwai. Residents collect and sell thatching grass to lodges, and some Khwai residents find employment from tourism operations and the local CBNRM program, the Khwai Development Trust (KDT). Tourism related opportunities brought additional settlers to the village, including some Bayei from villages south of Moremi Game Reserve (Bolaane 2004). Though efforts were initially made by the earl iest residents to keep these later arrivals from receiving benefits through KDT activities, all residents of Kh wai are now eligible to benefit if (Bolaane 200 4 ). Marriages between Bushmen and Bayei in Khwai and near by Sankoyo are common and have made defining ethnicity more challenging, particularly since Bushmen kinship traditions are patrilineal and Bayei are traditionally matrilineal (Larson 1992). Though defining ethnicity is complex and multi faceted, Khwai cont inues to self identify as a Bushman community (Bolaane 2004). The fourth village in this study, Sankoyo, is predominantly Bayey i Basubiya, Bushmen and Bambukushu people also reside in Sankoyo (Mbaiwa 1999) Sankoyo is
86 similar to Khwai in that it is loca ted within a fenced livestock free zone that prevents resident wildlife from interacting with livestock According to information gathered from Sankoyo residents, cattle have not been in the area since the 1950s when livestock were moved elsewhere to avoi d tsetse fly which has only recently been eradicated through spraying. Before CBNRM was established in this village, residents grew crops and traded them with people in the regional center of Maun, which is 85 km to the south of Sankoyo (B. Child pers. co mm. ) (see Table 1 1 and Figure 1 livelihoods have since shifted toward tourism, and c ommunication about natural resources now focuses mostly on wildlife (Cassidy 2001). Methods Researchers hypothesized, based on the literature presented, that gender and ethnicity were important variables affecting communication patterns and participation in CBNRM in the Okavango Delta. This was tested using a mixed method, cross sectional study design that addressed the following questions: To what extent do people tend to communicate more with people of their own gender or ethnicity, thereby demonstrating homophily in their social network? How might these social dynamics be reflected in and affect communication and participation in CBNRM? Data were collect ed through structured social network interviews, meeting and general field observations, semi structured interviews with key informants, and informal conversations with community members du ring fieldwork in 2008 and 2009 2010. Personal Social N etwork I nte rviews Sampling frame and sample selection. Social network data were collected during structured interviews in Khwai, Sankoyo, Gudigwa, and Seronga. To establish a sampling frame, a census of all adult residents by household was conducted by
87 researchers in all villages but Khwai ; there, a roster of residents was obtained from the CBNRM trust organization. The criteria for inclusion in the sampling frame was that the individual be 18 years old or older and spend at least six months of the year in residence i n the village. The researchers selected the individual as the unit of analysis, rather than the household, which is often used in rural studies, because information exchange is a process undertaken by individuals and may vary within a household. A combinat ion of randomly and purposively selected individuals was included in the sample of potential respondents. All known adult residents in each village were included in the roster, with a subset selected by a random number generator as potential respondents. T o accommodate challenging field conditions, research assistants were provided with lists of potential respondents based on the random number generator each day. Since most travel between potential respondents was done by foot on sand paths, often over cons iderable distances, these lists were divided geographically and assistants were asked to cover a particular area in a designated time period. Assistants were not expected to maintain the exact order of these lists as they sought potential respondents; anyo ne listed could be interviewed provided attention was paid to balancing the number of male and female respondents. The research team attempted to locate each potential respondent three times before removing his or her name from the list, and people who ref used interviews were taken off of the list of people with whom to initiate contact. 5 Since many residents divide their time between their home residences and subsistence field sites, interviews were also 5 Researchers were informed by traditional authorities, before this study, that some com munity members were particularly sensitive about participating in research studies based on negative past experiences. Out of respect for their wishes, researchers considered a refusal final, though a few people who initially refused did seek out members o f the research team at a later time and asked to participate. Since these individuals had been included on the initial list, they were included in the study.
88 conducted in the more remote field sites with the as sistance of a research vehicle. Whenever possible, traditiona l and other local authorities (particularly tribal leadership) were purposively included in the sample, as this built trust between researchers, leaders, and residents and provided valuable infor mation about communication networks. A total of 136 social network interviews were conducted (Table 3 1). Gender was used as a c riterion for the sample; an equal number of male and female respondents were sought in each village The sample was comprised o f: 47% males and 53% females in Khwai, 68% males and 32% females in Sankoyo, 42% males and 58% females in Gudigwa, and 49% males and 51% females in Seronga (Figure 3 1 ). Though not used as a criterion for the sample, ethnicity was recorded among the demogr aphic characteristics for each respondent, and the sample of respondents approximated the ethnic group composition in each village, particularly with regard to the dominant ethnic groups in each village listed in Table 3 1: Bushmen in Khwai, Bayeyi in Sank oyo, Bushmen in Gudigwa, and Bayeyi and B ambukushu in Seronga (Figure 3 2 ). Structured social network interview instrument. Personal (or ego) network report about their ti alters. Alter names are solicited using a specific question based on the research question and type of network of interest. A master personal network interview study was created in EgoNet software (McCarty 2011) and then customized for each village (Appendix A). The study consists of four parts: (1) questions about the respondent, (2) a question asking for a list of people with whom the respondent communicates regularly,
89 (3) questions abou t each of the people the respondent named, and (4) questions about the extent to which the people named communicated with one another. Interviews began with the respondent answering questions about themselves, including demographic variables (e.g., gender, ethnic background). Respondents were then asked: Please name 35 people aged 18 or older with whom you have communicated during the last month. Include first and last names. I will let you know when you have finished. No restrictions were placed on whom t he respondent named, and a full list of 35 alters was provided by each respondent. After this list of alters was created, respondents were asked demographic information about the alters including gender and ethnicity, and about the e xtent to which the res pondent received information about wildlife and the local CBNRM trust from each of the 35 alters In the final stage of the interviews, respondents were with each other when you are no possible alter pairs to create a personal network for each respondent. Each interview included up to 1,148 questions (depending on skip logic) and took approximately two ho urs to complete. Survey data were collected using netbook laptops. 6 Questions and possible responses were programmed into EgoNet software (McCarty 2011) in both Setswana and English. Most interviews were conducted in Setswana, but when necessary, local research assistants fluent in additi onal local languages translated the surveys into those languages. 6 The social network survey included a maximum number of 28 questions about the respondent, 1 alt er prompt question that was answered 35 times, 14 questions about each alter named (490 questions) and 595 alter pair questions, for a total of 1,148 total questions. Using EgoNet software, these questions come up in order (including skip logic) automatica lly on the computer.
90 Social network analysis. Personal network data were analyzed using EgoNet (version 3.31.11 ; McCarty 2011 ) and UCINET ( version 6.357; Borgatti et al. 2002 ) social network software. The names listed by respondents were used to assess the extent to which each personal network reflected homophily with regard to gender and ethnic share the characteristics of th e res pondent (gender and ethnicity); results were compiled for each village. Next, since communicating with people in positions of power may influence the type and quality of information a person can access, and oftentimes chiefs are provided information a nd expected to disseminate it to community members, an analysis was done to investigate the extent to which individuals included traditional authorities in their lists of alters In Khwai and Sankoyo, the tradit ional authority is a single chief, while in t he larger villages of Gudigwa and Seronga, this included both the main chief and the multiple headmen who report to the chief and are responsible for specific geographically designated areas, or wards, within the village. To expand this analysis of comm unication patterns to the village le vel, personal network data were joined together in EgoNet to create a comprehensive communication network for each village ( C. McCarty p ers. comm. ). Compositional and structural characteristics of these whole networks we re analyzed using UCINET software. Two indicators of social structure were used to better understand these communities: degree centrality and constraint. Degree centrality measure s the number of direct connections a person has to other people in the networ k A high degree centrality i ndicates that someone is active within the ir network and well connected to other network members (Wasserman and Faust 1994) Constraint is a network measure used in association with
91 structural holes. Structural holes are places where there are limited, non redundant connections between network sub group s. In other words, where there are structural holes, one or a few individuals may serve to connect sub group s, and in doing so, (Burt 1977 1992, 2004; Marsden 1982, Hanneman and Riddle 2005) A high value for constraint means that a high proportion network connections lead back to the same people, indicating that they are unlikely to be in a position to span structural holes; whereas a low value means that the person has less redundant connections and therefore a greater potential to serve as a bridge across structural holes (Burt 1992). Network data were exported to NetDraw software (version 2.114; Borgatti 2002) for visualiza tion and IBM SPSS Statistics version 19 for comparison with other social variables. Observations and semi structured interviews To supplement and contextualize the social network data, field notes were qualitatively analyzed in this study. Data included m eeting observations, semi structured interviews with key informants, and informal conversations with community members. Meeting observations from the CBNRM Annual General Meetings in four out of the five villages in OCT in 2009 and Khwai in 2010, along wit h several additional CBNRM meetings in OCT, Khwai, and Sankoyo were indexed and analyzed with regard to the relationship between gender and communication in participatory and decision making processes related to natural resources. Semi structured and info rmal interviews with village and community based conservation efforts, and observations provided data about
92 natural resource use and communication. Researchers qualit atively analyzed field notes to reveal overarching themes related to communication and CBNRM in this context. Findings Data analysis revealed several interesting findings related to the relationships between gender and ethnicity and natural resource use, c ommunication, and engagement in natural resource management. Relevant findings for gender and ethnicity are discussed in turn. Gender Natural resource use and communication. Observations and interviews revealed that natural resource use var ied by gender i n these villages in ways similar to those presented in the literature. Women often used natural resources to make craft prod ucts such as curios and baskets and talked while working together on crafts or other household projects Women in Sankoyo, for examp le, traveled together by truck to the nearby river to collect supplies for baskets. Some small groups of women gathered under trees in a particular yard while weaving and doing other household tasks. Women also worked together to collect and sell thatching grass, which is in high demand. In Khwai, a truck owned by KDT was regularly hired by a group of women who worke d together and sold the grass in quantity to tourism operators and other buyers outside of the village. Trust employees revealed that this work was consistently done by women, and joked that the only man involved was the truck driver. Women and men also took on different roles in the subsistence farming cycle. Often men and women were active in fa mily fields at different times but they came toge ther at certain times in the planting and harvesting cycle to work. These types of activities, where women demonstrated homophily with regard to resource use, promoted gender specific organization and
93 created spaces where women communicated separately from men about natural resources. Communication networks. The role differentiation observed with respect to natural re source use reflected homophily, and the results of social network analyses communication networks r eflected homophily as well In all four villages, both male and female respondents named people of the same gender more often than people of the other gender (Table 3 2) Among respondents in Khwai, Gudigwa, and Sankoyo, males demonstrated greater homophi l y than females in their communication networks (Khwai: M ales= 61% F emales=54%; Sankoyo: M ales=68% Females=55%; Gudigwa: Males=75%, F emales=51%) ( Table 3 2). In other words, men tended to name more men in their networks than women named women; however the only village where this difference was sig nificant was Gudigwa (Figure 3 3 ). The exception to this was in Seronga, where women, on average, named women 62% of the time and men name d men 55%. People also tended to be connected to more people of their own gender when homophily was calculated across the whole network outputs for each village (Figure 3 4) The descriptive statistics show that these means are quite similar to those found for respondents only (Table 3 3). The means were the same for both men an d women in Khwai, Gudigwa, and Seronga In Sankoyo, the average for the men was the same and the average for women was nearly the same ( 54% for the whole network compared to 55% for respondents only ) Results of paired t test comparison s of means across t he whole networks generated for each village i ndicated that men exhibited significantly greater homophily by gender compared to women (p=.05) (Table 3 4)
94 These data suggest that in three of the villages, associations with one anoth er, men may still be viewed as more valuable sources of information given their political and social status within village society. When sized by betweenness centrality, there are many more men than women among the prominent nodes in each of the villages ( Figure 3 5 ). Men assume more positions of power and authority in these villages, making them particularly valuable contacts when attempting to secure resources, natural and otherwise. The tendency for both men and women to name men as part of their communi cation networks may reflect a tendency among both male and female community members to either ask directly for access to desired resources, or to engage in communications that enhance relationships and social capital with male individuals perceived to hold power and influence in order to indirectly secure access to those resources. These contacts may be with people in positions of actual authority, or those of perceived power due to their status as men in this patriarchal society. Seronga does not follow th e same trend as the other villages, and women exhibit more homophily than men. Though the difference is not statistically significant (Figure s 3 3 and 3 4), this indicates that women may be less likely to favor men in Seronga. With more people in the villa ge in general, perhaps there are also more women seen as valuable as information sources. To investigate this further, researchers compared m en and women with regard to the extent to which they reported receiving information about wildlife and the local CBNRM trust from their own gender compared with the other gender. When the mean of these reported amounts of information were calculated across all respondents of a particular gender, b oth men and women report receiving significantly more information
95 abou t wildlife from men than women in Sankoyo, Gudigwa, and Seronga (Figure 3 6 ). The trend is the same in Khwai; however, while women still receive more information about wildlife from men, men receive approximately the same amount of information about wildli fe from men and women. In the case of information about the local CBNRM trust, the trend is the same, though t he differences are less and not significant at a 95% confidence interval except in Seronga, where men get significantly more information about the CBNRM trust from other men (Figure 3 7 ). The lower amount of information received about the CBNRM trusts in general may contribute to this result. In sum mary both genders report receiving more of the information that they receive about wildlif e and their local CBNRM trust from men than women Women are talking to women, but it appears that there are not talking as much to women as they are to men a bout wildlife and the CBNRM program. Additional analysis focused on the extent to which respondents named tr aditional authorities in their personal networks Personality and accessibility may affect the extent to which community members interact with the chief and if information is disseminated through these leaders. In Khwai, for example, the chief which he sits in the kgotla and makes himself available to discuss issues and concerns brought by residents. In this small village, the chief is well known and highly accessible. In Sankoyo, by contrast, only one person named the chief, wh ich suggests that the dynamics surrounding communication between the chief and village residents are different in this village. Communication outside the village, however, is something this chief does well, as he is known to be effective in working with pe ople at higher levels of governance. While the types of direct interactions observed in Khwai do occur in larger
96 villages to some extent, the sheer number of people for whom leaders are responsible may make this type of direct contact with the main chief l ess likely. Instead, r esidents of Seronga and Gudigwa may be more likely to communicate with n, the traditional authority subordinate to the main chief who is responsible for a specific ward. The yard of one headman in Gudigwa was a popu lar meeting place for people resident within that ward, and there was a similar meeting place near the home of a headman in a ward in Seronga. Accounting for communication connections with both the chief and headmen of a village, t raditional authorities w ere named more often by men than women overall, though this varied by village (Table 3 5 Figure 3 8 ). In Khwai, the proportion of respondents of each gender naming the chief was nearly equal (Males= 43%, Females= 41%). Among the subset of the total respon dents in Khwai who named the chief, a higher proportion of women (54%) than men (46%) named the chief In contrast, in Gudigwa, 6 5% of male respondents named a traditional authority, compared with 27% of women, and 83% of those naming the traditional autho rities were men Gudigwa demonstrated the greatest contrast in responses between men and women. This difference was present but less extreme in Seronga, where 58% of male and 45% of female respondents nam ed the traditional authorities, and men comprised 55 % of those naming the traditional authorities from that village. In Seronga, more than half (11 out of 20) respondents who named a traditional authority in their network named more than one, indicating that those who were connected to those in power may ac tually be connected to multiple sources of power and information.
97 Structural network measures, which address how people are connected, can help investigate the functioning of these networks. The whole networks in each village were analyzed to see if there was a difference between men and women in terms of how many connections people had and the extent to which each gender might be able to take advantage of structural holes and become information brokers with in the ir network. Degree centrality is a measure of the number of ties one node has to other nodes in the network. In this study, a person with high degree centrality had a high number of direct the potential to directly communicate with a large number of other people within his or her village network. This means that they may have more direct access to information because of their large number of personal contacts, and/o r they may hold power since they could also decide whether to share information they have with those with whom they are connected. In each village men had more connections in the ir personal networks than women; however, this difference was only sig nificant in Gudigwa (F igure 3 9 ). With regard to the potential for community members to take advantage of structural holes, i n Khwai, Gudigw a, and Seronga, men had lower constraint values than wo men, indicating that men would therefore be more likely than women to have the opportunity to take advantage of structural holes and gain from the potential power associated with them (Figure 3 10 ). In Gudigwa, this difference was significant. There was also a significant difference in Sankoyo; however, the relationship there was the opposite, and men were significantly more constrained than the women. Further research on why this might be so and whethe r or not this is affecting the ability of
98 women to take advantage of their structural position in Seronga could provide important insights about the social dynamics in this village Communication and p articipation in CBNRM. Attendance at community kgotla meetings that are focused on natural resources varies in terms of overall attendance and the proportion of men and women present. Each community trust responsible for implementing CBNRM holds an Annual General Meeting (AGM). In OCT, the AGM is repeated in each of the five member villages. Observations at four of five of these OCT meetings in October 2009 and several additional CBNRM meetings in OCT, Khwai, and Sankoyo in 2009 and 2010 revealed that the discussions in these meetings are heavily dominated by men. Several examples illustrate this trend. In Gudigwa, the benches close to the front of the kgotla were occupied by men, and all women wer e on the ground around a tree fa rther from where discussion took place. Women are accustomed to sitting on the gro und and were observed to be engaging in short private discussions during the course of the meeting; however, they were physically separated from the official discussion and not contributing. The only woman who spoke after introductions of guests at the mee ting was from the Department of Wildlife and Natural Parks (DWNP), one of the invited officials at the meeting. In Beetsha, there was a great deal of discussion surrounding accounting for funds by OCT leadership. In that discussion, 5 6 comments were noted, only five of which were made by women. Of these, three were by the external auditor, a women based in Maun, and two were made by the same DWNP official who spoke in Gudigwa. The meeting in Gunutsoga was attended by 10 men and eight women. Fourteen out of 60 comments noted were contributed by
99 women, half of which came from female officials. Several of the audience comments made by women cam e from the same outspoken woman There was considerably more discussion at the fourth village meeting in Eretsha, where 86 total comments were offered during the discussion, 31 of which were made by women. However nearly all of the comments by women came from officials in the kgotla, particularly from the female master of ceremony facilitating the meeting, the auditor, an d the DWNP staff person. Only one comment came from a woman in the audience. All community members are stakeholders under CBNRM, but participation in decisions and communications surrounding CBNRM clearly vary by gender. Women are less involved than men i n these formal communications; however, additional observations and interview data indicate that women do engage in a great deal of communication about natural resources that may indirectly affect deci sion making. These are in different forms than what may gleaned by attendees at meetings can be shared with non attendees through informal but regular communication networks, like tho se observed among women gathering grasses. If t his information is discussed within a household, it to be represented through their husbands in formal meetings. By the same token, female headed households may be marginalized because they do not have the same opportunity to influence natural resource decisions through these household discussions (Cassidy 2001). Women may also gain information and influence through employment or elected leadership in their local trust organization, or with a safari operator. The Manager of the
100 Sankoyo Tshwaragano Management Trust (STMT) in Sankoyo indicated that he and the T rust made a concerted effort to include women in elected positions of Trust leadership, and many women served as employees in Trust tourism operations as well. All four villages employed women as trust employees, and both women and men have opportunities to be employed by safari operators. The opportunities in tourism operations, however, are generally differentiated by gender specifi c roles, which may affect the information to which employees have access. Interestingly, in OCT, where radios connect the villages in this multi village trust, women are nearly always employed as operators of the radios that are used for CBNRM as well as g eneral communications. This gives them direct access to information and puts them in the positio n to be communication conduits ; however, most of the information conveyed by two way radio is either of a logistical nature or of a personal one since this is a lso the means of communication between community members in the villages and those in safari camps. Ethnicity Natural resource use and communication. Unlike gender where strong gender role differentiation was observed, natural resource uses were relative ly homogeneous across ethnic groups. Interviews and observations suggest that although historical differences affected the perceptions and opinions about natural resources held by pe ople of different ethnic groups; these generally were not reflected in pra ctice. As stated in the literature review, B ushmen have taken on more sedentary livelihood strategi es compared with previous hunter gatherer traditions People of all ethnicities across the four villages engaged in subsistence farming activities, and some people of each different ethnic group raised livestock either in their home villages or in other towns. Engagement in these activities appeared to be related more to government
101 regulations than ethnic background. Natural resource communications tended to f ocus on farming activities, the availability and quality of water, and wildlife management in CBNRM. In all of the villages, observational data and informal interviews indicated that ethnicity was an important factor in communication relationships among vi llage residents. Seronga, where many ethnic groups come together present s a level of diversity similar to what one might find in a small city. Wards in this and similar villages were traditionally organized so that people of the same ethnic background wer e located together ( Mompati and Prinsen 2002) Though this is not always true, the tendency toward ethnic homophily persists in the geographic arrangement of households within Seronga today H ouseholds of people of the same ethnic group were often found li ving close to one another. Ethnic homophily is hence often expressed in physical proximity, which creates spaces where people like one another are also more likely to communicate with one another on a regular basis. On several occasions there, when queried about the home locations of other villagers, respondents were able to provide this information only for me mbers of their own ethnic group but not f or others, even when members of the other ethnic group lived much closer. People also tended to know the nam es and more information about people in their own ethnic group. Ethnicity is complex and challenging to define, yet it appears to be an important factor in who is talking wit h whom, and about what, in these villages. Communication networks. Social network communication networks reflect homophily with regard to ethnicity (Table 3 6, Figure 3 1 1). R espondents from all villages showed a strong tendency to name members of their
102 own ethnic group. Seronga is the most diverse among the villages studied, and respondents there showed marked homophily. Of the 64% Bayei respondents, 87% of the people they named were also Bayei. The 28% of Bambukush u respondents named other Bambukushu people 72% of the time. Bushmen are a small group with in Seronga, and represented only 3% of respondents. This subset of respondents showed a relatively low level of homophily by ethnicity with same ethnicity names given only 37% of the time. This lower level of homophily may reflect the relatively small numb er of people in their own ethnic group, or it may reflect a tendency for people in this group to seek information from those in majority groups, who also hold positions of power. Expanding this to the whole network outputs for each village, ANOVA results indicate that the dominant ethnic groups in each village exhibit a significantly higher level of homophily by ethnicity (Tables 3 7 and 3 8). In Khwai, members of the dominant ethnic group (Bushmen) have a mean homophily value of 92%, the Bayeyi in Sankoyo have a value of 93%, and the Bushmen in Gudigwa exhibit 98% homophily. In Seronga, there is considerable ethnic diversity and, at least theoretically, greater opportunities to interact with people of different ethnic groups. Yet, members of the two domina nt ethnic groups in Seronga the Bayeyi and Bambukushu, had homophily values of 88% and 75%, respectively, which reflects the tendency people have to associate more with people like themselves. I t is clear from Fig u r e 3 5 that members of the dominant ethni c group or groups in each village also dominate as the largest nodes and are highly central within their village communication network. Researchers found that people reported receiving more i nformation about wildlife and the CBNRM trust from people of the i r ethnic group compared with the amount of
103 information received from people of other groups For information about wildlife, this difference is present in Sankoyo, Gudigwa, and Seronga and significant in the first two at a 95% confidence interval (Figure 3 1 2 ). The same trend is observed for information about the CBNRM trust; however, in this case, the difference is only sig nificant in Sankoyo (Figure 3 13 ). These results suggest that if some members of a particular ethnic group are receiving important info rmation, this information may be disseminated through communication channels within that particular ethnic group. The challenge comes, however, if one or more ethnic groups is marginalized and not receiving the information at the outset. In these cases, th e chances that members of that ethnic group will receive the information are diminished. K hwai presents an interesting contrast to this challenge in that people appear t o get about the same information from those inside and outside of their ethnic gr oup; t here is no apparent difference between these sources in terms of the amount of information that is reported to be gained These data suggest that in the small, densely connected village of Khwai, information tends to be well distributed which may mean tha t most people have the potential to be sources of valuable information. Communication and p articipation in CBNRM During several interviews and informal conversations during the study, members of minority ethnic groups reported feeling relatively uninfor med, not only about environmental and CBNRM details, but also about the CBNRM decision making processes themselves. One challenge is that the language barriers that often accompany ethnic diversity in this setting present concerns about the extent to which speakers of different mother tongue languages can actively participate in CBNRM. W hile young people tend to know Setswana and some English,
104 elders often have little or no schooling and speak only their mother tongue language. These elders are relied upon as primary sources of information (see chapter 4), but they are often not integrated into CBNRM decision making processes. This becomes particularly problematic when considering the role that traditional ecological knowledge can play in effectively managin g natural resources. While ecological knowledge based upon their personal experiences and historical perspectives may still be shared, language barriers may limit the interaction and integration of this knowledge with other sources of information. 7 Ethnic minorities of any age may also face challenges when it comes to participation in CBNRM, particularly in a leadership capacity. Democratic elections favor the dominant groups in each village, since representatives on the CBNRM boards are elected by majority vote, which tends to reinforce existing power structures. In OCT, this is somewhat overcome by electing board members from each village, which allows different villages to gain some representation on the board. At the time of this study, the board chairma n was from Seronga and Bayeyi, which had been the case for several years. When he spoke at the AGM, he did in his mother tongue or read in English. Though someone translate d his statements into Setswana, this meant that only Bayeyi audience members were re ceiving this information firsthand. 7 It would be interesting to compare elders to other age groups; however, it was difficult to do so quantitatively in this study. Age data was collected for social network analysis interview respondents (though many elderly people do n ot know their age and do not have identification that provides this information ) A fter many attempts during pilot testing to collect data about the ages of people in unable or unwilling to offer this information about other people As such, the data and findings specific to elders in these villages was collected through ethnographic field observations and informal interviews.
105 Implications for Environmental Communication Communication is essential to the conservation and effective management of natural resources. To acquire the information they need about natural resources, rural residents in the Okavango Delta of Botswana often rely upon communication with local leaders and members of their personal social networks, which are highly influenced by social dynamics. Evidence from this study shows that, by and large, rural community members exhib it homoph ily by gender and ethnicity. It suggests that focused attention should be placed upon understanding these social variables and the ir potential impacts on communication about natural resources. I nformation is powerful. Its exchange can build trust, and the process of participation can contribute to individual empowerment (Kelly and van der Riet 2001), giving individuals a voice desire and willingness to participate, however, is affected by his or her perceptions of the type of reaction expected from other participants, as well as the long range actions (Kaplan and Kaplan 1982 Rahmena 1992, Reid et al. 2008). This has important implications for members of marginalized groups, such as women and individuals of minority ethnic groups, in Botswana. Over the past several decades, considerable efforts have focused on engaging women and minority groups, particularly those from the Khoisan language group, in va rious social and political processes in Botswana society. Yet, evidence collected in this study suggests that minority groups, as defined within a particular village, continue to face challenges One area where efforts may be able to be enhanced is in the representation of women and members of minority ethnic groups on natural resource governing bodies. This may be challenging to implement at the local level, where
106 village level politics and social dynamics tend to prevail and this is a delica te and challe nging process. While policies may provide a framework for addressing these challenges, great attention must be paid to understanding the social dynamics, local politics, and potential impacts of policies at the local scale Such changes are likely to take a long time and require considerable effort at multiple levels of governance. In the shorter term, in formation dissemination and training activities about natural resources a nd decision making processes could be targeted to reach, and facilitate informatio n exchanges with, leaders and members of minority groups. The Technical Advisory Committee (TAC) of Botswana is responsible for working with local C BNRM programs to build capacity, and they could help enhance engagement with members of minority ethnic grou ps and women in CBNRM participatory processes. They might also, work to facilitate communication channels and information exchange s between wo men and traditional authorities This of course assumes that those in power want to share their information, which may vary considerably among chiefs. The chief of Khwai provides an excellent examples of how being available for open dialogue with all community members aids in information exchange. Information exchanges could specifically target activities that traditionally engage women (e.g., crafts, veld products). Efforts could be focused on helping groups that are alread y organized to build their capacity to impact natural resource decision making, and those who are informally associated to become more formalized in an effort to secure greater access to information and influence as a collective group.
107 I nformation exchang e at this level has the potential to empower minority groups to find a more informed and cohesive voice when it comes to participation in larger scales of natural resource decision making. Existing trainings and communications around CBNRM and natural reso urces c ould be enhanced to actively incorporate local knowledge, particularly knowledge that is typically found among women (i.e., about veld resources) and minority ethnic groups (i.e., traditional uses and environmental trends over time). Social network analysis and/or key informant interviews may be helpful in i dentify ing those individuals within different minority groups who hold both formal and informal key roles in these communication network s These individuals c an serve as valuable and influential s ources o f information for others and are therefore important to identify and engage with environmental communication efforts. Working to integrate these important members of minority groups into the environmental communication landscape in deliberate and m eaningful ways could provide important benefits to marginalized groups and individuals in rural villages. In the rural Okavango Delta region of Botswana, where nearly all people depend upon natural resources for subsistence and survival, information is im portant for people to be able to effectively participate in decision making processes associated with CBNRM. Yet, facilitating environmental communication is a challenging prospect when social and cultural circumstances, such as one dominant gender and cer tain dominant ethnic groups, create power differences that are expressed in communication networks and participatory processes. These social dynamics present challenges, but also pose opportunities to enhance existing and build new communication networks that are locally relevant, potentially
108 powerful ways of engaging residents in broader communication and governance networks. The strategies suggested here have the potential to enhance success of CBNRM programs and other development programs centered on pa rticipation in these rural areas. Strategic communication should be seen as a means of empowering those whose voices are often left unheard, and as a means to achieving enhanced conservation and development benefits.
109 Table 3 1. Dominant ethnic group and n umber of personal network interviews conducted in each v illage Village Dominant ethnic group(s) in village Bayeyi Bushman Bambukus h u Baherero Batswana Other Total completed surveys (n) Khwai Bushmen 5 2 6 31 Sankoyo Bayei 22 1 1 5 2 31 Gudigwa Bushmen 33 1 35 Seronga Bayei, Bambukushu* 25 1 11 2 39 Many ethnic groups are represented in the large village of Seronga, with Bayei and Bambukushu the two largest groups, in that order according to key informants. The kg osi (chief) and many residents are of Bayei origin, while several wards are dominated by Bambukushu. Table 3 2. Gender of respondents and homophily by gender between social network survey respondents and alters named in each village Khwai Sankoyo Gudigwa Seronga Gender Respondents (%) Homophily* (%) Respondent (%) Homophily (%) Respondents (%) Homophily (%) Respondents (%) Homophily (%) Male Female 47 53 61 54 68 32 75 51 42 58 68 55 49 51 55 62 Homophily is calc ulated as the average percent of alters named in the same gender as the respondent.
110 Table 3 3 Descriptive statistics calculated for homophily by gender in each village Village Gender N Mean homophily by gender Std. deviation St. error mean Khwai Men 490 .61 .489 .022 Women 595 .54 .499 .020 Sankoyo Men 454 .68 .468 .022 Women 630 .54 .498 .020 Gudigwa Men 806 .75 .434 .015 Women 383 .51 .501 .026 Seronga Men 665 .55 .498 .019 W omen 700 .62 .496 .018
111 Table 3 4. Results of paired t test s comparing means for men and women of homophily by gender in each village est for e quality of v ariances F Sig. t test for equality of m eans 95% Confiden ce interval of the difference Village t df Sig. (2 tailed) Mean di fference St d. error d iffere n ce Lower Upper Khwai Equal variances assumed 16.986 .000 2.155 1083 .032 .065 .030 .006 .124 Equal variances not assumed 2.157 10 50.734 .031 .065 .030 .006 .124 Sankoyo Equal variances assumed 71.085 .000 4.403 1082 .000 .132 .030 .073 .190 Equal variances not assumed 4.448 1009.66 .000 .132 .030 .074 .190 Gudigwa Equal variances assumed 124.716 .000 8.524 1187 .00 0 .242 .028 .186 .297 Equal variances not assumed 8.106 663.869 .000 .242 .030 .183 .300 Seronga Equal variances assumed 21.810 .000 2.503 1363 .012 .067 .027 .119 .014 Equal variances not assumed 2.501 1355.320 .012 .067 .027 119 .014
112 Table 3 5 Inclusion of traditional authorities in personal communication networks of men and women Men naming traditional a uthority Women naming traditional a uthority V illage Count Compared with total male respondents in that vi llage (%) Compared with all in v illage w ho mentioned c hief (%) Count Compared with total female r espondents in that v illage (%) Compared with all in village who mentioned c hief (%) K hwai 6 43 46 7 41 54 Sankoyo 0 0 0 1 6 100 Gudigwa 15 65 83 3 27 17 Seronga 11 58 55 9 45 45 Table 3 6 Homophily by ethnicity between social network survey respondents and named alters in each village Khwai Gudigwa Sankoyo Seronga Ethnic Group Respondents (%) H omophily* (%) Respondents (%) Homophily (%) Respondents (%) Homophily (%) Respondents (%) Homophily (%) Bayei Bushman Bambukushu Baherero Batswana Other 16 84 0 0 0 0 0 26 91 0 97 0 0 0 3 0 98 0 71 0 3 3 15 2 0 93 0 0 28 22 64 3 28 0 0 5 0 87 37 72 46 Homophily is calculated as the average percentage of alters named in the same ethnicity as the respondent. I ndicates that there were no respondents of that ethnic group and therefore no calcu lation of homophily for that ethnic group.
113 Table 3 7. Descriptive statistics and tests of homogeneity of variance for homophily by ethnicity in each village 95% Conf idence interval for mean Test of homogeneity of variances Village Ethnic grou p N Mean Std. deviation Std. error Lower Upper Levene statistic df1 df2 Sig. Khwai Bayeyi 175 .26 .438 .033 .19 .32 Bushmen 910 .92 .278 .009 .90 .93 Total 1085 .81 .393 .012 .79 .83 145.051 1 1083 .000 Sankoyo Bayeyi 769 .93 .258 .009 .91 .9 5 Bushmen 35 .00 .000 .000 .00 .00 Baherero 35 .00 .000 .000 .00 .00 Basubiya 175 .28 .450 .034 .21 .35 Other 70 .21 .413 .049 .12 .31 Total 1084 .72 .450 .014 .69 .74 80.717 4 1079 .000 Gudigwa Bushmen 1154 .98 .143 .004 .97 .99 Other 35 .00 .000 .000 .00 .00 Total 1189 .95 .217 .006 .94 .96 3.099 1 1187 .079 Seronga Bayeyi 875 .88 .328 .011 .86 .90 Bushmen 35 .37 .490 .083 .20 .54 Bambukushu 385 .75 .435 .022 .70 .79 Other 70 .46 .502 .060 .34 .58 Total 1365 .81 .395 .011 .79 .83 74.538 3 1361 .000
114 Table 3 8. ANOVA c omparison for homophily by ethnicity in each village Village Sum of Squares df Mean Square F Sig. Khwai Between Groups (Combined) 63.595 1 63.595 662.794 .000 Linea r Term Unweighted 63.595 1 63.595 662.794 .000 Weighted 63.595 1 63.595 662.794 .000 Within Groups 103.913 1083 .096 Total 167.508 1084 Sankoyo Between Groups (Combined) 121.488 4 30.372 333.952 .000 Linear Term Unweighted 102.165 1 1 02.165 1123.342 .000 Weighted 19.323 3 6.441 70.822 .000 Within Groups 98.132 1079 .091 Total 219.620 1083 Gudigwa Between Groups (Combined) 32.571 1 32.571 1645.145 .000 Linear Term Unweighted 32.571 1 32.571 1645.145 .000 Weight ed 32.571 1 32.571 1645.145 .000 Within Groups 23.501 1187 .020 Total 56.072 1188 Seronga Between Groups (Combined) 20.921 3 6.974 49.429 .000 Linear Term Unweighted 14.205 1 14.205 100.681 .000 Weighted 6.717 2 3.358 23.803 .000 Wi thin Groups 192.019 1361 .141 Total 212.941 1364
115 Figure 3 1 Percent of social net work survey respondents in each gender and village. Figure 3 2 Number of social network survey respondents from each ethnic group 14 13 23 19 17 18 11 20 0% 20% 40% 60% 80% 100% Khwai Sankoyo Gudigwa Seronga Village Males Females 2 1 2 5 1 1 11 26 33 1 5 22 25 0 5 10 15 20 25 30 35 40 45 Khwai Sankoyo Gudigwa Seronga Bayei Basarwa/Bushman Bambukushu Baherero Batswana Other
116 Figure 3 3 M ean percent homophily by gend e r among social network survey respondents (n=136)
117 Figure 3 4 Mean percentage homophily by gender for whole networks, including respondents and all alters named by respondents (n=2,276)
118 Figure 3 5 Visualiza tions of village networks; nodes sized by betweenness. Villages are included in increasing population size: A) Khwai, B) Sankoyo, C) Gudigwa, and D) Seronga. Squares are males; circles are females. Red = Bayeyi. Green = Bushmen. Orange = Bambukushu. Light blue = Basubiya. A B C D
119 Figure 3 6 Mean amount of information received by respondents from alters of about wildlife
120 Figure 3 7 Mean amount of information received by respondents from alters about their local CBNRM program.
121 Figure 3 8 Percen t of men and women naming a village traditional authority compared with the total number of respondents of that gender in each village 0 10 20 30 40 50 60 70 Khwai Sankoyo Gudigwa Seronga Percentage of Males Percentage of Females
122 Figure 3 9 Mean degree centrality by gender in whole networks, including respondents and all alters named by res pondents (n=2,276).
123 Figure 3 10 Mean constraint by gender for whole networks, including respondents and all alters named by respondents (n=2,276).
124 Figure 3 1 1 Mean percent homophily by ethnic group for respondents in each village 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Khwai Sankoyo Gudigwa Seronga Bayei Basarwa/Bushman Bambukushu Baherero Batswana Other
125 Figure 3 12 Mean amount of information received by respondents from alters about wildlife
126 Figure 3 1 3 Mean amount of information received by respondents from alters about the local CBNRM program
127 CHAPTER 4 PERCEPTIONS OF WATER AND WILDLIFE RESOURC ES Rural peo ple in developing countries often rely upon natural resources to meet their daily and livelihood needs (Shackleton and Shackleton 2004) yet we know relatively little about the perceptions that rural community members have about these natural resources (Hu nter et al. 2010). Environmental change is predicted to affect resources and rural livelihoods (IPCC 2007), and rural residents will need to adapt to these changing conditions. Adaptation requires that people learn about changes in biophysical and socio cu ltural systems, synthesize new information, and take action to accommodate observed changes (Kaplan and Kaplan 1982). Information is therefore a critical currency for adaptation at both the level of individuals (Kaplan and Kaplan 1982) and groups engaging in natural resource management, where information is essential for transparency, accountability, and participation in decision making processes and the factors affecting into their thinking about natural resources. Botswana is at the nexus of water scarcity, wildlife dependence, and climate change issues. For people living in the Okavango Delta region in the no rthwest of the country, water and wildlife are two natural resources of utmost importance to subsistence and tourism based livelihoods (Wilk and Kgathi 2007, Mbaiwa 2008, 2011). These resources are predicted to change over time with livelihoods affected in turn the factors that may affect these perceptions important in this setting.
128 This study focuses on the question: What factors account for any differences that exist i cultural background and experience with regard to water and wildlife affect their perceptions of these resources? This is investigated in two ways by comparing responses across two villages and by comparing the responses for water and wildlife. A secondary focus of the study is to understand the extent to which people are integrating priority messages of the government agencies responsible for communicating about water and wildl ife into their thinking about these resources. Mental models are used as a theoretical framework for this study. This set of theories asserts that mental models are the cognitive structures people have in their minds that describe the way the world works When an individual reveals their mental model, this can help researchers understand how an individual perceives his or her environment (Arentze et al. 2008). The study provides valuable insights about the assessment and comparison of perceptions of rural residents, the variables that may affect mental models in this context, and considerations for agencies that are responsible for communicating about water and wildlife to rural communities. Literature Review At their most fundamental level, mental models help people make sense of the world around them (Denzau and Douglass 1994) and and Kaplan 1997, p. 579). T he origin of mental models theory tr aces back to Kenneth scale models of the Laird and Goldvard Steingold 2007, p. 169). An individual has many models w ith varied content
129 and interconnections with items in other models that together explain the complexity of environmental conditions and objects in their world (Johnson Laird 1983, Halford 1993, Smith 1999). These representations serve as k nowledge structur es that contain (Kearney and Kaplan 1997). Stephen and Rachel Kaplan (1982) addressed the cont representations allow for: 1) generality, keeping track of similarities to reduce the need to attend to all stimuli as new; 2) economy, the ability to store vast amounts of information an d retrieve what is relevant; and 3) connectedness, the ability to build bridges between points that are known to deal with the unknown. ( Kaplan 1973, Norman 1983) through p rocesses of experiential learning (Kolb 1984) and as cognitive capacity expands with development (Barroui l le t and Grosset 2007). Knowledge is constructed as a result of experiences in which an individual interacts with his or her environment (Piaget 1953). M ental m odels are shaped by a combination of sensory intake (Dretske 1981), perceptions ( Johnson Laird 1983), and preference s (Kaplan and Kaplan 1982). In building mental models, people add and integrate new information into what they already know ( Kaplan and Kaplan 1982, Klimoski and Mohammed 1994) and build upon existing understandings to interpret the world according to past experiences (Jonassen 1991, Fosnot 1996, Seel 2001). Mental models are created about individual experiences and cultural contexts, both of which allow people to make sense of things and act as they see fit in their world (Shore 1995).
130 Because mental models are based on individual experience, they are ultimately unique, based upon and creating different ways of engaging with the world (Ritchie Dunham and Puente 2008). When people are familiar with an environment or context, they have more nodes (items) of informati on c reating a more robust mental model, which makes him or her less dependent upon outside information (Kaplan and Kaplan 1 982, Rickheit and Sichelschmidt 1999) When new information is available, people tend to first fit it into their already existing models Human brains have evolved to deal with a level of uncertainty and complexity in our environment (Beratan 2007), and me ntal models provide an important benefit in economy of mental energy and efficiency in thinking (Kaplan and Kaplan 1982, Barrouillet and Grosset 2007). This becomes challenging, however, as diversity complex ity, and uncertai nty in the environment increase In these cases, existing mental models may be inadequate to make sense of the new information, and more information and more robust mental models may be necessary to reason, make predictions, reach conclusions and adapt to changing conditions ( Kaplan an d Kaplan 1982). Th e Reasonable Person Model (RPM) offers additional insights into the factors that affect whether and to what extent mental models are built. Mental model building is dependent not only upon the availability and understandability of inform ation in a but also on whether or not he or she has opportunities to take meaningful action and be effective at using this information (Kaplan and Kaplan 2009). To invest the mental energy required to build rich mental models about so mething,
131 In other words, people do not typically create rich mental structures about topics they h ave no way of using. People need to feel as if they are heard and respected when they participate (Kaplan and Kaplan 2009); however, participation is not enough. People need to also feel as though can reach effective ends and not be left feeling helpless in their attempts to take action based on their mental models (Kaplan 2000). Mental models are important to consider both in terms of adaptation and in any context where people, and therefore their mental representations of the world, come together. There has been increased attention about how mental models may be shared or not within groups, and how this can affect various aspects of participatory natural resource management proc esses (Jones et al. 2011). Ritchie Dunham and Puente hat you see, what you advocate, and what you ultimately decide information may be received diff erently by different members of a group, and improve communication (Abel et al. 1998). L earn ing more about backgrounds and perspectives may also allow messages to be tailor ed so that they can be integrated into the variety of mental models present (Morgan et al. 2002). Research Setting, Data, and Methods wildlife and discusses how experi ential and cultural factors may contribute to the observed similarities and differences in these models. It was conducted as part of a larger research project investigating rural communication networks and perceptions of
132 water and wildlife resources in the Okavango Delta, Botswana (Figure 4 1). Data for this study were collected in March and April of 2010. The primary investigator worked with two local research assistants (one in each rural village included in this study) who had several months of experienc e on the research team and were specifically trained to conduct this study. Interview guides for structured perceptions interviews were written in both Setswana (the national language) and English (the official language of Botswana). Most interviews were c ompleted in Setswana while being translated into English for note taking and analysis. When necessary, research assistants fluent in several languages verbally translated parts of the interviews into additional local languages. Research Setting The Okavan go Delta region of Ngamiland, Botswana is a semi arid region of southern Africa where environmental change is predicted to alter water and wildlife resources (IPCC 2007). M any individuals and communities in Botswana currently rely on scarce and declining n atural resources (Thakadu and Schuster 2007). The situation is complex and highly uncertain (Giannini et al. 2008), and changes are likely to impact the livelihood strategies of rural residents dependent upon natural resources (Hunter et al. 2010). Resear chers collected data in two villages participating in a community based natural resource management (CBNRM) program (Figure 4 1) The CBNRM approach began in Botswana in the late 1980s as a reaction to severe declines in wildlife and the failure of traditi onal top down management approaches to protect these resources resulted in a shift from national control over wildlife management to roughly 90 local community based programs that encompass a range of sizes, program characteristics,
133 and outcomes (Shuster 2007). CBNRM programs are now a fixture in natural resource management in Botswana (Taylor 2007) and are seen as a potential way to balance conservation and development objecti ves (Arntzen 2003). The two villages in this study are within the same multi village CBNRM program, the Okavango Community Trust (OCT). OCT was registered in 1996 as the community based organization (CBO) responsible for implementing CBNRM in this locatio n (Schuster 2007). Seronga is the largest of the villages in OCT. It houses the main OCT staff office and is closest to the river, paved road, and larger villages. Gudigwa is the most remote of the five villages and is farthest from Seronga. The ethnic bac kgrounds and experiences with water and particularly wildlife differ considerably between Seronga and Gudigwa Y et, for the purposes of the CBNRM program, these two villages are C ommunity members from all five villages are tasked with managing wildlife resources and the financial benefits derive d from these resources together despite inherent intra village as well as inter village diversity (Agrawal and Gibson 1999). For participation in CBNRM to be meaningful and effective, community members must be able to access and integrate information and draw upon their mental models when they engage in natural resource management decision making processes (Jones et al. 2011). Conversely, for mental models to be rich and useful, opportunities must exist for effective actions that use that knowledge. S eronga and Gudigwa were selected for this study because they provided a basis of comparison for how available information was integrated into the mental models of people with different ethnic backgrounds and experiences with natural resources.
134 Assessing Mental Models An advantage of taking a mental models approach is that there are several tools se research techniques are predicated on theories of cognitive psychology asserting that mental models guide how people think about, perceive, and react to their environment (Kaplan and Kaplan 1982). It is important to acknowledge that since mental models are internal to the respondent, they cannot through any method be observed directly (Rowe and Cooke 1995). Some aspects of mental models are not easily expressed in outward language, and elicitation techniques will tend to oversimplify the truly complex na ture of the way people think (Johnson Laird 1983). There are several methods, however, that can help respondents to externalize their personal, internal representations (Jones et al. 2011). Researchers combined several elicitation techniques to design a s tudy protocol effective and appropriate in this research setting. The approach included the use of open and close ended interview questions, photographic concept cards, and consensus analysis. O pen ended interview questions offer depth in data collection about perceptions; they allow people to express themselves and explain through words the way they think about a subject or problem in more complex ways (Morgan et al. 2002). Using cards with concepts on them can be used to help respondents generate more co mprehensive sets of ideas about a topic (Spradley 1979), and cards can be used to help people organize concepts according to particular criteria (Kearney and Kaplan 1997, Ozesmi and Ozesmi 2004). Consensus analysis allows researchers to understand the exte nt to which there may be a shared mental model (Stone Jovanovich et al. 2011) such that particular knowledge or perceptions are shared by a group
135 the views of their gro up (Caulkins and Hyatt 1999). In all of these techniques, respondents were asked to speak about concepts and relations between concepts. These are considered indirect elicitation techniques, since they require researchers to extract and analyze data about provided (Carley and Palmq uist 1992, Jones et al. 2011). Researchers used these methods to address the following questions: What differences exist in the ways rural community members perceive water and wildlif e, and what factors may account for these differences? Researchers hypothesized that due to perceptions about these resources would be different. Mental models abou t wildlife would be more complex than those for water, since wildlife issues are complex and people have more of an opportunity to make decisions and take meaningful action about wildlife. Additionally, researchers predict e d that people more central within their village communication network would have more robust mental models since they have the potential to access more information from a networks standpoint. Data Collection and Analysis Data were collected and analyzed across respondents to assess their mental models about two domains, water and wildlife. Quantitative analyses were supplemented by observations and ethnographic field notes collected throughout the fieldwork period. Extended observations provided important context helpful for interpreting t he data.
136 Phase 1: Domain definition The first step in the process of understanding perceptions about water and wildlife minds about these two topics, or domains. Initial free listing interviews. Twenty one rural community members participate d in individual or small group free listing exercises (Table 4 1). Respondents were asked, list as man y things as they could think of related to one domain (water or wildlife) and then asked to do the same for the second. Probing and redundant questions were used to increase recall and generate more comprehensive lists of words and concepts for each topic (Brewer et al. 2002, Bernard 2006). The order of domains addressed (water or wildlife) was alternated between respondents in an attempt to reduce the potential for consistent respondent fatigue regarding one topic. The interview guide used for this free li sting exercise is included in Appendix B. The goal of this exercise was to determine the range of concepts that sh ould be included in a set of cards used in the second phase of the process This was accomplished by having respondents name as many items as possible in each of the two domains. A grounded theory approach using open coding was used to identify categories and concepts that emerged from the responses provided (Glaser and Strauss 1967). This inductive approach allowed researchers to determine a l ist of concepts that were then included in pilot studies of structured perception interviews that followed. Results were coded and tabulated by hand in the field on an ongoing basis, and when there was general agreement on the concepts offered by responden ts,
137 researchers closed the free listing interview portion of the study (Weller and Romney 1988). Interviews with government officials. In addition to the list of concepts gained through free listing exercises with rural residents, the research team wanted to understand what messages and concepts government agencies thought were important to rural residents of the Okavango Delta. Interviews were conducted with 9 officials at multiple agencies working on issues related to water and wildlife resources in Bots wana: Department of Environmental Affairs, Department of Wildlife and National Parks, Department of Water Affairs, Department of Public Health, Department of Tourism, the Biokavango Project, 1 and the Botswana OKACOM 2 Office. Key informants and a snowball s ampling technique were used to find the most appropriate contact in each office (Heckathorn 1997, Bernard 2006). One respondent served as the primary respondent in each office, with the exception of the Department of Wildlife and National Parks (DWNP), whe re one interview was conducted in the regional office in Maun and another was conducted at the local office in Seronga. These semi structured interviews were used to collect data about what government agencies expected communities to have heard and integra ted into their thinking about water and wildlife. Priority concepts were identified, discussed, and added to the list of possible mental model concepts to be included in the structured perceptions interviews in Phase 2 of the study. Data were also co llecte d about how 1 The Biokavago Project is a funded by the Global Environmental Facility (GEF) and the Government of Botswana, and is implemented by the Okavango Research Institute of the University of Botswana in n the http://www.orc.ub.bw/biokavango/ ). 2 OKACOM is the Permanent Okavango River Basin Water Commission comprised of representatives from Angola, Namibia, and Botswana ( http://www.okacom.org/index.htm ).
138 these messages are conveyed to rural residents. Discussions emerging from these inte rviews ranged in focus and depth ; the basic interview guide is included in Appendix C. Data were analyzed qualitatively for themes (Bernard 2006), specifically focused on messages from government to rural villages about water and wildlife and the means of information dissemination used by government to communicate with rural villages. Phase 2: Structured perceptions interviews Sample selection and respondents. D ata about social networks related to communication collected during an earlier phase of the larger research project information that person receives. Information will, in turn, influence his or her mental models about water and wildlife. To investigate the relationship between social network position and perceptions, the sample of interview respondents was purposively selected based on previously coll ected social network d ata (see c hapter 2), specifically degree centrality and betweenness centrality. Centrality measures generally measure the importance or prominence of actors within a network (Wasserman and Faust 1994). In this s tudy, degree centrality measured the number of ties a person ha communication network High degree centrality means that the person has many direct connections to other people; a low value indicates that the person has less direct connect ions Someone with high degree centrality has the potential to access a lot of people, and if some of those people have important information, they may have a better chance of accessing it. Betweenness centrality is calculated based upon the number of shortest paths a perso n lies on within the network The value indicates the tendency of that acto r to be in
139 the middle of others within the network. Measures of betweenness tend to be associated with information flows, as they can help quantify how quickly information could mov e from one part of a network to another. A low betweenness centrality value indicates that the person lies on relatively few shortest paths between other people; a higher value means that the person is more often found between and along the shortest paths between other people. While any information exchanges are ultimately determined by the decisions of an individual, a person with high betweenness centrality may be able to control the extent to which and speed with which information is exchanged, since the y tend to sit in the middle of and on the shortest paths between many other people in the network. People with high and low values for the two centrality measures, degree and betweenness, were included in a list of potential respondents and represented in the final set of respondents (Table 4 2 Figure 4 2 ). The respondents represented a range of values for both measures. This was of interest because someon e may have high degree centrality, which means that they have a lot of connections, but if his or her betweenness centrality was low, this person may only have access to information from his or her direct contacts and not be connected through those contacts to others in the network. Someone with high values for both degree and between n ess centrality, in c ontrast is likely to be connected to many people who are also connected to people themselves. This may be helpful when trying to access information within a network. People who are low on both measures may not be able to access information from their pers onal network like those more central within the network can. A total of 30 people completed the perceptions interviews, 17 in Seronga and 13 in Gudigwa.
140 Research instrument and procedure. Pilot tests conducted with rural community members revealed that man y respondents did not have experience with categorization tasks and were unable to categorize concepts into meaningful categories. This made direct mental model elicitation techniques, which require respondents to categorize and provide visual representati ons of their mental structures (Jones et al. 2011), inappropriate for this research setting. Instead, an indirect elicitation method was used, about water and wildlife. The fina l research instrument incorporated free listing and rating techniques in a structured interview protocol that incorporated both open ended and closed response questions and prompts with photographic cards (Appendix D). Concept cards were an important part of this design. Sets of cards were created for water and wildlife in advance of the perceptions interviews from items elicited during the free listing exercises with rural community members and semi structured interviews with government officials. Photogr aphs and Setswana words were included on each card. These were pilot tested in an iterative manner to ensure the photos and words used represented the same concept to each respondent, and the cards represented the range of responses for water and wildlife. Researchers intentionally included cards for items mentioned with high and low frequency, which is a requirement of the consensus analysis later performed on these data (Romney et al. 1986). The same cards were used with each respondent to allow for stati stical comparison among respondents. A total of 34 and 23 cards were used for water and wildlife, respectively. The structured interviews were divided into three parts. Respondents were first asked seven basic demographic questions about themselves ( e.g. ethnicity, education,
141 research assistant asked questions of the respondents in Setswa na and translated the full responses aloud into English. He also marked items, in the order they were given, on an interview guide spreadsheet (Appendix E). This spreadsheet included the items on the concept cards as well as space to add additional topics mentioned by respondents for which cards had not been created. Concurrently, the primary investigator placed cards for the concepts mentioned on the table facing the respondent. She also took notes abou t the responses given and audio recorded the interview s. These notes supplemented quantitative data and allowed researchers to and Geiselman 1992). After the concept cards were laid out, they were used as one of several prompt perceptions (Brewer 2002). Additional questions focused specifically on river water and to provide clarifica tion about the responses given. The responses to these questions were treated as successive free lists and were used in combination to gain a more conclude the section about w ater, respondents were asked to name the people they would go to if they were seeking more information about water. In the final section of the interview, respondents were asked a similar but expanded set of questions about wildlife. Data collected in Phas e 1 of the study and pilot tests of the survey instrument revealed that there was little variation among the
142 responses about water, and responses about wildlife were far more varied and reflected considerable differences in perceptions and opinions. Becaus e of this and since the main focus of CBNRM programs in the region is wildlife, researchers placed a greater emphasis on wildlife in these interviews. The wildlife section began with questions parallel to those about water, and respondents were asked abo ut what came to mind when thinking about wildlife. Successive free lists and question and card prompts were used to elicit comprehensive lists of concepts. Following these questions about their perceptions of wildlife, respondents were handed the cards for the items they had mentioned and asked to rate them on a scale of values: 1=Very Important, 2=Somewhat Important, and 3=Not as Important. Researchers then handed respondents the remaining cards for wildlife, representing concepts they had not yet mentione d during the interview, and gave them the option to rate these items on the same scale. These ratings were recorded in the interview guide; items that respondents chose not to rate were marked as 4=Not Selected. Respondents were encouraged to comment on th eir ratings as they went through the activity, and notes were recorded. The final questions of the interview focused on government messages about wildlife and the local CBNRM program. Phase 3: Data analysis Mental model assessment. Interview data were ana lyzed in several ways. Researchers analyzed the responses elicited about perceptions of water and wildlife to ance, familiarity, or study, researchers sought to understand the salience, or importance, of particular
143 concepts within the perceptual domains of water and wildlif e. The Freelist function in Anthropac 4.98 software (Borgatti 1998) was used for this analysis, which provided measure that combined frequency and rank (Smith 1993). The free lists inputted were generated from the responses to the initial open ended questions separately by topic, water and wildlife. The order of items was based on the orde r in which concepts were mentioned by respondents; concepts mentioned multiple times were listed in the position that they were mentioned first. Qualitative data from interview notes and ongoing field observations were analyzed for themes and supplemented the quantitative analyses. Consensus analysis. To investigate the extent to which there was agreement and perhaps a shared mental model among respondents about the importance of wildlife concepts (Stone Jovanovich et al. 2011), an informal consensus analys is 3 was conducted in Anthropac. Consensus analysis is a statistical procedure used to ascertain if there are shared group beliefs about a topic or domain and the extent to which individual responses reflected this shared understanding (Romney et al. 1986, Weller and Romney 1998, Weller 2007). In other words, it helps to determine whether or not there is a shared mental model about a particular topic (Stone Jovanovich et al. 2011). (Weller and Romney 1988, p. 77). 3 Informal consensus analysis is used for ordinal, interval, and ration scaled responses, whereas formal consensus analysis is limited to categorical response data (Weller 2007).
144 In this study, the cultural consensus model statistically analyzed the rating responses that were given to systematically asked interview questions and estimated 986). The model then determined if there was in fact consensus among respondents about the domain, and calculated a to which his or her responses matched those of the group (Weller 2007). In this study, collective group. Researchers wanted to explicitly avo id the mistaken impression that a lower score might reflect less knowledge about wildlife, when it simply indicated a lack of agreement with the most commonly held ideas about the importance of various wildlife concepts and perhaps reflected a different kn owledge. The analysis was run using the ratings provided in each interview for all 23 concept cards. The cultural consensus model was run with three data sets: (1) all respondents from Seronga and Gudigwa (n=30), (2) for Seronga only (n=17), and (3) for Gu digwa only (n=13). These analyses were used to see if there was agreement across the two villages that are part of the same CBNRM program in OCT, and if there was agreement or disagreement at the village level. The agreement values produced by the cultural consensus model for each individual were compared with several independent variables to determine if these accounted for the observed variability in agreement scores. Variables included gender, ethnicity, education level, and the number of village groups in which the individual participated. These relationships were tested with the nonparamet ric Independent
145 Samples Kruskal Wallis Test in SPSS. Researchers also tested the extent to which a es of degree centrality and betweenness centrality. Results Water Government messages. The major and consistent theme that emerged from agency interviews was the importance of rural residents using standpipes connected to boreholes as their source of water rather than river water, to avoid human health risks hu man animal conflict, and disease. Over the past several decades, the Government of Botswana has invested heavily in providing boreholes, and more recently communal standpipes, in rural villages. In parallel, they have invested in conveying messages about the enhanced health, cleanliness, and safety of these water sources for humans. These messages were conveyed both directly at traditional community meetings in each village, called kgotla meetings, a nd through the kgosi, or chief. Messages have more recently started shifting toward conservation, since water scarcity is becoming a bigger concern throughout Botswana. Mental model concepts. Free listing and perceptions interviews with residents reveale d that these government messages about using standpipes had in fact reached and been integrated into the thinking of rural residents. Many talked about gathering water for cooking from standpipes (50%), with 5 respondents mentioning standpipes and 4 mentio ning boreholes specifically (Table 4 3). The most often mentioned topics were about personal water for drinking, washing, bathing, and cooking. Many respondents also spoke about water as important to food, mentioning fish in the river and rainwater as a re quirement for growing crops. Respondents from Seronga tended
146 to offer more information about the river than respondents from Gudigwa. For example, water lilies, which are collected from river water as a food source, were mentioned by 65% of respondents fro m Seronga and only 23% from Gudigwa. Several individuals from Seronga mentioned the threat of water animals such as hippopotamuses and crocodiles when collecting fish or reeds from the river, which may have been influenced by a recent drowning of a Seronga resident from such activities. A visualization of the 72 distinct concepts given by respondents is in Figure 4 3 ; the size of the word or concept represents how frequently it was mentioned by respondents. Interviews brought to light serious concerns about water quality and quantity in several. In Seronga, though people reported using standpipes, they also expressed concerns about the quality of that water. Rust was noted as a problem with unknown consequences by several respondents, and personal observations revealed that rust was often visib le in standpipe water. Gasoline, which leaked into the water supply during pumping was also often prese nt in the water from standpipes In general, people used the standpipes as their water source whenever it was working; however, some elderly women stated that river water had always been safe, and it was safe for them as well. Several respondents in both villages mentioned that the avail ability of water in standpipes was a concern. The primary researcher observed several water outages lasting hours to days while in Seronga. In Gudigwa, where only one individual was given the authority to operate the borehole, there was a period of several days without water in the standpipes and another period when the village water tank was overflowing. In
147 both cases, the borehole operator, who was also a member of the OCT board, was out of town and therefore unavailable to fix the problem. Sources of i nformation. When asked to name who they would go to for information about water, responses varied a great deal but focused on government officials (Figure 4 4 ). No single information source was offered by a majority of respondents. Forty three percent of a ll respondents (13) named the Department of Water Affairs Eight of the respondents were from Seronga and five from Gudigwa Though based in Maun, representatives travel to villages to deliver information According to interviews with government officials, these visits had been quite frequent during the time when standpipes were being installed several years earlier. The Department is also ultimately responsible for managing the operation of village boreholes and standpipes, and it hires local people to man age these activities. Thirty seven percent of all respondents mentioned the borehole employees (operators). The borehole operator was the information source named most frequently in Gudigwa by 46% of respondents; however, the same information source was on ly mentioned by 29% of Seronga respondents. Respondents mentioned elders as sources of water information the same number of times as borehole operators. Among respondents from Seronga, elders were named most often as the source of information about water, though the average rank (or position in an individual free list) was 3.111, indicating that this source was usually mentioned after the Department of Water Affairs, which had an average rank of 1.625 (Table 4 3). The rank value indicates the average posit concepts mentioned. Elders as a general category were only mentioned twice in
148 Gudigwa, though individual elders were named specifically in some responses. It was interesting to note that the kgosi, or chief, was named (either b only five respondents out of 30 (17%). Some respondents may consider the kgosi specific names, respondents tended to name individuals other than the kgo si. Many respondents did mention that they tended to hear about water issues from the government during kgotla meetings, which matched the information dissemination methods mentioned by government officials during those interviews. Wildlife Government m essages. The two dominant theme s that emerged during interviews with government officials were: (1) efforts to encourage rural people to use chili peppers as a deterrent to elephants who raid crops from fields and (2) the importance of wildlife and benefit s of wildlife based tourism Crop raiding is a major complaint among rural residents, who are mostly dependent upon annual crops for subsistence. Elephants were mentioned by 77% of respondents (Table 4 4), most often in relation to their impact on crops an d local people. Yet, not a single person mentioned chili peppers when speaking about wildlife during the perceptions interview before being prompted by the photo cards. Once presented with the cards, a majority of respondents (19) did rate chili peppers as a very important concept related to wildlife six rated it somewhat important, three stating it was not very important, and two electing not to consider this concept in their ratings (Table 4 5). Wildlife officials indicated that they use similar methods of communication as those mentioned in interviews with water officials including presentations at kgotla meetings and communication directly with chiefs. They also employ demonstration projects that allow local farmers to implement chili pepper
149 deterrent m ethods in their fields. While these demonstrations were believed to be successful by government officials, this success seems to be very locally specific. Some residents recalled hearing something about these projects, but they had not seen it personally n or implemented it in their own fields. A related theme that emerged from interviews with government officials was their focus on communicating with rural communities about the importance of wildlife as a source of income through tourism. DWNP was the ori ginal agency responsible for CBNRM in Botswana, 4 and it continues to be an important agency in wildlife management and an adviser to CBNRM projects. Money and jobs were mentioned by a majority of respondents (70% and 63%, respectively), and four respondent s mentioned that wildlife was important in general ; however, the context of these responses were more linked to benefits observed directly than to government agency efforts. Mental model concepts. The concepts mentioned regarding wildlif e reflect both benefits and concerns associated with wildlife. As Table 4 mentioned most often and by 93% of respondents. Respondents often talked both about elephants, killing of livestock by wildlife, wildlife presenting dangers), while also mentioning the benefits wildlife provided local people through tourism. Nearly two thirds of respondents discussed jobs coming from wildlife; many associated these with safari operators and/or OCT. A majority of people mentio ned 4 Responsibili ty for CBNRM projects has since moved to the Botswana Tourism Board, a para statal organization responsible for marketing and managing tourism operations, community based and otherwise, in Botswana.
150 OCT (63%); close to a third of respondents mentioned development benefits that came from wildlife, either through OCT or in general. Many people mentioned the e, and two elders in particular mentioned changes that had occurred and how the government has become more involved than they wished regarding wildlife management. Though not among the most common responses, it was interesting to note that 23% of responden ts mentioned the beauty of wildlife specifically. A visualization of the 79 distinct concepts given by respondents is in Figure 4 5 ; the size of the word or concept represents how frequently it was mentioned by respondents. Comparing per c eptions of water and wildlife I nterview data revealed that expressed about water. Table 4 6 shows that both the number of distinct concepts mentioned and the total number of responses offered w as greater for wildlife than water during the open ended portions of the structured perceptions interviews. This difference is difficult to quantify, however, because the open coding method used by researchers meant that the complexity of each distinct con cept could vary depending on the coding scheme. Qualitative data were therefore essential to understanding this apparent difference. Researchers first noted that the responses given for water during Phase 1 free list interviews demonstrated relatively litt le diversity and complexity when compared with wildlife and focused most often on household daily uses and supply concerns. The qualitative data from structured perceptions interviews offered similar results. Wildlife questions elicited more personal stori es and reflected a greater diversity of backgrounds and experience with regard to wildlife than water. Many people mentioned either
151 specifically having mixed feelings about wildlife or provided stories about how wildlife brought the community both tourism benefits and subsistence challenges. Some respondents from Gudigwa, in particular, spoke about their past relationships with wildlife and the changes that have occurred since CBNRM was instituted. All respondents from Gudigwa self identified as part of th e same River Bushman ethnic group, who were traditionally migratory hunters and gatherers (Bolaane 2004) before settling in Gudigwa under government land policies in 1987 (Cohen 2008) These results prompted the researchers to investigate the extent this apparent complexity might be correlated with the social network variables used in this study. Though it is an imperfect proxy of mental model robustness, researchers compared a count of the number of concepts respondents offered during the structured perce ptions interviews with social network measures of degree centrality, betweenness centrality, and a ranking value. The ranking value was used because the centrality values were relative to the village where the respondent lived, and the rankings allowed for more appropriate comparison across the two villages. Results showed that respondents who were more central within their network expressed a greater number of concepts. The number of concepts offered by respondents was significantly correlated at a 95% con fidence interval with betweenness centrality for both water ( r =0.533) and wildlife ( r= 0.457) and to the ranking index for water only ( r= 0.430 5 ) (Table 4 7). C onsensus analysis. Results generated by the consensus analysis revealed that there was agreement about the concepts perceived to be most important about wildlife 5 Respondents with the highest degree centrality was ran ked number 1, the second highest was ranked 2, and so forth. The negative correlation between ranking and the number of water concepts means that those with lower numbers (and therefore higher centrality values) tended to mention a greater number of concep ts, as hypothesized.
152 (Table 4 8 ). This was determined by evaluating the model outputs for the ratio of Factor 1 to Factor 2, where eigenvector ratios over 3:1 indicate that a consensus exists (Weller 2007). When treated as a single sample for all OCT respondents, the ratio of factor 1 to factor 2 was 4.883. The ratio for the Seronga respondents only was 6.515, indicating that there was also consensus within Seronga. In contrast, the ratio for Gudigwa was 1.672, i ndicating that consensus did not exist among the 13 respondents from that village. The Seronga consensus is estimated to be at the 95% confidence level, since the average level of cultural knowledge or agreement, was 0.561, which is greater than the 0.5 t hreshold required, among 17 respondents (Weller and Romney 1998, p.77). This average agreement value was the highest of the three models, with the full group at 0.448 and Gudigwa at 0.317. Agreement values were compared to see if there were differences in responses based on several social variables, including ethnicity, gender, education, and the number of village groups in whic h a person participated (Table 4 9) as well as social network measures (Table 4 10) Results of the Independent Samples Kruskal W allis Test showed that the only signifi cant variable was ethnicity. A box plot of these values illustrates that there are three different means and standard deviations for the three ethnic groups presented (Fi gure 4 6 ). It must be noted that scores in this figure reported most often; it is not a measure of absolute knowledge or competency. Although the standard deviations overlap across the three ethnic groups, the large ra nge for Bushmen suggests that this group may have a unique knowledge set about wildlife resources distinct fr om the other two ethnic groups. Correlations between social
153 network measures and level of agreement were weak within this sample (Table 4 10), thou gh the direction of correlations suggest that social network position could play an important role and further investigation is warranted. Source s of information. The top several sources of informa tion are presented in Figure 4 7 The government agency res ponsible for wildlife management, DWNP, was Gudigwa (54%), which might be explained by the p resence of a DWNP office in Seronga. Similar to the results for water, elders were among the most named sources of The general category of elders was mentioned by only 15 % of Gudigwa respondents; however, many individuals were mentioned by name, including some elders. Safari operators, and specifically the safari operator in a joint venture partnership with OCT at the time of the study, were named by 28% of the total respo ndents. The kgosi in each village was named as often as elders in general, with 24% of respondents naming the kgosi as an information source. This was higher for wildlife than for water, where only 17% of respondents named the kgosi. OCT was named as an in formation source about wildlife by only 4 of the total respondents, all of whom were from Seronga where the OCT main office is located Though not indicating an overall trend, it was interesting to note that one respondent from Seronga mentioned that Bushm en from Gudigwa were good sources of information about wildlife, and named two Gudigwa elders specifically in this regard.
154 Discussion These results bring to light several interesting points. Results indicated that there were differences in the responses gi ven by people from Seronga and Gudigwa and in the responses for water versus wildlife Theories of how and why mental models are formed offer a framework for understanding the differences observed. Mental models are tools that people build and use to under stand their world and make decisions, and the content and personal experience and the decision making demands placed on those models. At a basic level, people in Seronga probably offered more respons es related to the river because of their proximity and regular interaction with river resources. The geographic location of people in Seronga provided particular experiences not shared by those in Gudigwa, who live further from the main river and experienc e only seasonal flooding. Though Gudigwa residents are aware of some of the same concepts, they have less direct experience with them on a regular basis. Thus, the salience of concepts related to the river was higher among respondents in Seronga than Gudig wa. In contrast to their experience with river resources, Gudigwa residents were very expressive about water shortage issues and reported concerns similar to those in Seronga about water availability With only one person holding the key, Gudigwa residents regularly experience challenges associated with water shortages, with little power to influence either the situation or the outcomes from these shortages. Access to water is highly mediated by the government, first through the Department of Water Affairs and then th rough local borehole operators. This contributes to similarities in the experiences people have in rural areas with regard to water resources. In addition, most rural community members in both Seronga and Gudigwa use standpipe water for the
155 same household purposes (e.g., drinking, washing dishes and clothes, bathing, cooking). Mental models theories suggest that these similarities in experience can help account for the general consistency and lack of variation evident in water perceptions in thes e two villages. The relative simplicity in these perceptions is also explained by RPM, since people are unlikely to invest mental energy in the building of complex mental models where they have little ability to participate in decision making processes or make meaningful changes. Water access is now mediated by the government and borehole operators, meaning that very little individual decision making is required. When rural villages first received standpipes, government officials faced considerable challen ges and invested diseases from river water in the messages. These messages appear to be well wants water for household tasks, they simply decide to go to the standpipe. As suggested by RPM, since people do not need to make complex decisions about water and have little ability to make a difference, they are unlikely to invest heavily in building m ental models about this topic. The relatively simple models they have are sufficient for the decision making conditions they face with regard to water. far more varied and individualized, which ma y help explain the richness and diversity of perceptions noted in present hun ting experience, direct contact (good or bad) with wildlife, growing crops,
156 raising livestock, and direct interactions with OCT, safari companies, and DWNP officers. Engagement in a CBNRM program mediates the relationships individuals have with wildlife an d the government agencies that regulate it (e.g., limiting hunting, models included many personal, unique stories and experiences related to wildlife. T here are also many decisions that might have to be made about wildlife, and these decisions tend to be made more quickly. For example, if an elephant enters a subsistence crops, or the relativ e consequences of crop raiding versu s shooting the animal A separate set of decisions may need to be made for conflicts with livestock, and another for decisions related to the CBNRM program. Wild animals are far less predictable, and people have far more opportunities to make decisions and take action with regard to wildlife than water in their personal actions and, at least in theory, through CBNRM participatory decision making processes. People have relatively more capacity and power to make decisions a bout wildlife than water, and as RPM suggests, this is reflected in the complexity of their mental models. They need more knowledge in order to take action that can have significant consequences. This complexity in decision making may also contribute to th e relatively low level of success of DWNP in conveying their priority message about the value of chili peppers as a crop raiding deterrent. Despite employing communication strategies similar to the Department of Water Affairs, this message has not yet been initial responses. One factor that may contribute to this is that while the community as a whole is impacted by crop raiding, personal experience varies. People may also be
157 challenged to integrate a new approach into their mental model, which may already contain strategies for deterring elephants, effective or not. As stated earlier, the complexity related to wildlife requires both a great deal of mental energy and a robust mental model. The new information may not be easily fit in models, which provide important benefits in economizing of use of mental energy and allowing for greater efficiency in thinking (Kaplan and Kaplan 1982, Barrouillet and Grosset 2007). People may also have more preconceived notio ns and/or misconceptions built into their mental models that are difficult to overcome. It will likely require experience, and thus may take more time and strategic communication approaches to allow people to better integrate the new chili pepper concept i nto their current mental models. Consensus analysis revealed that there are differences among respondents with regard to the perceived importance of different wildlife concepts. There was consensus, according to the model, among the full group of responde nts and within Seronga; however, this was not the case for Gudigwa. When analyzed separately, the diversity of responses among Gudigwa respondents became clear. This was interesting, given that all Gudigwa respondents came from the same ethnic group. Three factors may help explain these findings. First, it may be a matter of the information that is available to each person. The information a person is able to access could be related to the specific people with whom they communicate The second factor is the integration of that information. There may be another variable or variables that are mediating the central in this network better able to take in information about wildlife than those less
158 connected. Though the small sample size in this study makes this difficult to assert quantitatively and results were not statistically significant, extended research in Gudigwa indicated that there appeared to be a qualitative difference i n the perceptions and attitudes of some older community members compared with younger ones. Younger informants in Gudigwa, who may have been born in the village, only understand wildlife policy after the establishment of CBNRM, may not have experienced pas t hunting traditions, and spoke more often about benefits of tourism and OCT. This contrasted with elders from that village who often relayed stories about past hunting experiences and their d esire to go back to old ways. They expressed a high degree of fa miliarity with the topic of wildlife in general. Though both young and old people discussed aspects of these ideas during perceptions interviews, the lack of consensus may i ndicate that these divisions exist based on d ifferences in experience. Implicati ons Rural people rely on several sources of information their personal experiences, trusted elders, and government agencies. For both water and wildlife, the government agencies responsible for disseminating information about those resources are the most f requently mentioned sources of information. This dependence is most evident for wildlife, as respondents named DWNP nearly three times more than any other single information source Yet, the same people did not mention the highest priority message of the D epartment the use of chili peppers as a deterrent for elephants, in their interviews We suggest that this is related to wildlife and t o fill this gap and increase the chances of this message being integrated int
159 more opportunities for people to gain personal experience with the chili pepper elephant deterring strategy Agencies responsible for communicating about natural resources wi th rural communities may also need to tailor their strategies to the type of message being conveyed. When it comes to water, since people generally have the same mental models and experience with the resource, presenting in the usual manner at the kgotla i s likely to continue to be effective. As long as people are not asked to deliberate or act on water issues, their minimal mental models are sufficient. In contrast, information about wildlife may be better communicated more directly to sub group s of people within the village, such as subsistence farmers or elders. Given the greater complexity of decisions to be made and existing mental models about wildlife, it may take more time models about things related to wildlife and its management. If people feel that they empowered to participate and make a difference through their actions, they are more likely to invest the mental en ergy to incorporate new information into their thinking. related to the richness of their mental models. People who are less central within their social network demonstrate less complexity in their mental models. This suggests t hat more attention needs to be paid to reaching all members of a community, not just those who attend particular meetings where information may be conveyed directly. This is particularly important for wildlife. Some people may be more or less able to atten d meetings to hear fr om government officials or CBNRM leaders directly with those unable to attend relying upon their personal connections for information. In this research
160 settin g, access to information is one of several factors that may impact how peopl e perceive their ability to effectively engage in p articipatory decision making processes for CBNRM. In OCT where multiple villages come together, communication must therefore be deliberate and comprehensive. In Gudigwa, which is already geographically dis tant from the center of OCT activity, people often complained about their relative lack of information about wildlife and CBNRM activities compared with other villages, and the data in this study suggest that those less central within Gudigwa are in fact n ot integrating the same concepts into their mental models. I nformation is not only a benefit of CBNRM, it is also a necessity and a concerted effort should be made to better communicate with people in different social positions and with different backgrou nds present in all five of the OCT villages. One opportunity to focus communication efforts about wildlife is to actively engage elders in each village. Elders are clearly a trusted and important source of information. They are also under represented on CBN RM boards. The chairs of CBNRM boards tend to be powerful older men, but o ne of the perceived successes of CBNRM is the ability to retain young people in their home villages through engagement in CBNRM leadership positions (Schuster 2007). Whether or not e lders are interested in becoming part of the decision making structure of CBNRM, they should be actively engaged in the information exchange about environmental history that is invaluable to cur rent natural resource management and future adaptation. Mental models research provides valuable insights about how people perceive natural resources, how messages might be better communicated for integration into
161 how opportu nities for participation and effective action and how the diversity of mental models might impact participatory decision making processes. Further investigation into mental models may help researchers better understand how people perceive and learn about changes in biophysical and socio cultural systems and synthesize this information to adapt.
162 Table 4 1. Rural respondents in Phase 1 free list exercises about water and wildlife Responde nt ID R espondent g roup Gender Year b orn 1 2 3 A A B Male Male Female 1982 1983 1984 4 5 C D Male Female 1980 1982 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 D D D D E E F F G H I I J K L M Male Male Female Female Female Female Male Male Male Male Fe male Male Female Male Female Male 1987 1982 1986 1987 1984 1971 1968 1960 1961 1979 1956 1942 1973 1985 1984 Indicates birth year unknown. This is sometimes the case with elders who do not remember the year of their birth and whose ID cards do not st ate a year.
163 Table 4 2. Demographic and social network for s tructured interview respondents ID Gender Ethnicity Age Years in v illage Education Number of village g roups Degree c entrality Betweenness c entrality Se r1 Female Bayei 27 27 Junior S econdary 1 0.065 0.023 Ser2 Male Bayei 29 29 Junior Secondary 0 0.026 0 Ser3 Male Bayei 26 26 Cambridge Diploma 1 0.061 0.008 Ser4 Male Bambukushu 52 52 Primary 1 0.051 0.037 Ser5 Male Bayei 66 66 Primary 2 0.162 0.061 Ser6 Female Bayei 44 44 Primary 0 0.075 0.019 Ser7 Female Bayei 28 28 Junior Secondary 1 0.085 0.017 Ser8 Male Bambukushu 58 58 None 1 0.092 0.036 Ser9 Male Bambukushu 30 30 Senior Secondary 0 0.079 0.015 Ser10 Male Bayei 33 33 Vocational Training 0 0.059 0.023 Ser11 Male Bayei 34 3 4 Vocational Training 1 0.068 0.032 Ser12 Male Bambukushu 59 59 Primary 1 0.142 0.076 Ser13 Female Bayei 25 25 Primary 1 0.033 0 Ser14 Female Bambukushu 23 23 Junior Secondary 1 0.010 0 Ser15 Female Bayei 66 66 None 0 0.054 0.002 Ser16 Female Bayei 75 75 None 0 0.033 0.002 Ser17 Male Bayei 70 70 Primary 1 0.215 0.119 Gud1 Male Bushman 26 26 Cambridge Diploma 0 0.527 0.028 Gud2 Male Bushman 18 18 Junior Secondary 0 0.066 0 Gud3 Female Bushman 31 12 Primary 0 0.108 0.008 Gud4 Male Bushman 27 19 Juni or Secondary 2 0.449 0.032 Gud5 Male Bushman 86 19 None 0 0.130 0.008 Gud6 Female Bushman 18 18 Junior Secondary 1 0.064 0 Gud7 Male Bushman 28 21 Cambridge Diploma 0 0.360 0.014 Gud8 Male Bushman 39 23 Junior Secondary 2 0.483 0.022 Gud9 Female Bushm an 35 31 Senior Secondary 1 0.218 0.016 Gud10 Male Bushman 34 34 Senior Secondary 2 0.365 0.027 Gud11 Female Bushman 53 24 None 0 0.373 0.010 Gud12 Female Bushman 22 22 Senior Secondary 0 0.162 0.008 Gud13 Female Bushman 53 23 None 0 0.025 0 Note: Hig her values for both degree centrality and betweenness centrality reflect greater connectivity. These numbers are relative to the village that there are in, however, and should not be taken as absolute values.
164 Table 4 3. Summary of all concepts mentio ned about water during structured perceptions interviews F requency of r esponse Percent of r espondents (%) Average rank in l ist Smith's S v alue Item Both (B) Ser onga (S) Gud igwa (G) B S G B S G B S G BASIC_NEED 10 4 6 33 24 46 1 .0 1 .0 1 0.333 0.235 0.462 BEING_WET 1 1 0 3 6 0 5 .0 5 .0 0.024 0.042 BIRDS 2 1 1 7 6 8 9 .0 10 .0 8 0.029 0.029 0.028 BOREHOLE 4 1 3 13 6 23 10 .0 14 .0 8.7 0.036 0.008 0.073 BRICKS 1 1 0 3 6 0 5 .0 5 .0 0.026 0.046 BUFFALO 1 1 0 3 6 0 13 .0 13 .0 0.003 0.005 BUILDING_CONSTRUCTION 11 6 5 37 35 38 5. 3 4.8 5.8 0.226 0.252 0.191 BUSINESS 2 1 1 7 6 8 12.5 8 .0 17 .0 0.024 0.036 0.009 CHITA 2 2 0 7 12 0 5 .0 5 .0 0.038 0.066 CLEANING_VEHICLES 1 1 0 3 6 0 8 .0 8 .0 0.017 0.029 CONSERV ATION 2 1 1 7 6 8 6.5 4 .0 9 .0 0.041 0.045 0.036 COOKING 15 6 9 50 35 69 3.1 2.8 3.3 0.412 0.305 0.552 COUNTRY 1 0 1 3 0 8 3 .0 3 .0 0.029 0.067 CROPS 17 7 10 57 41 77 4.6 4.6 4.6 0.415 0.322 0.538 DEVELOPMENT 1 1 0 3 6 0 5.0 5.0 0.02 4 0.043 DIRTY_WATER 8 6 2 27 35 15 10.6 9.8 13.0 0.082 0.118 0.036 DISEASE 6 3 3 20 18 23 8.7 8.7 8.7 0.092 0.079 0.108 DRINKING 21 11 10 70 65 77 4.0 4.1 3.9 0.531 0.502 0.570 DROWNING 1 0 1 3 0 8 10.0 10.0 0.003 0.008 DWNP 1 0 1 3 0 8 9.0 9.0 0.007 0.015 ECOSYSTEM 1 1 0 3 6 0 6.0 6.0 0.023 0.040 ELEPHANTS 1 1 0 3 6 0 11.0 11.0 0.008 0.014 EVAPORATION 1 1 0 3 6 0 11.0 11.0 0.017 0.031 FISH 21 12 9 70 71 69 9.0 8.3 9.9 0.294 0.350 0.221 FLOODING 2 0 2 7 0 15 14.5 14.5 0.007 0.015 FOOD 6 1 5 20 6 38 8.3 12.0 7.6 0.084 0.016 0.174 FROGS 2 2 0 7 12 0 14.0 14.0 0.014 0.024 FRUITS 2 2 0 7 12 0 13.0 13.0 0.015 0.027 GOVERNMENT 4 3 1 13 18 8 9.8 9.0 12.0 0.047 0.073 0.012 GRA SSES_REEDS 15 12 3 50 71 23 9.3 9.3 9.0 0.194 0.258 0.110 GRAZING 1 1 0 3 6 0 11.0 11.0 0.015 0.026 HEALTH 2 2 0 7 12 0 4.0 4.0 0.053 0.093 HYDROGEN_OXYGEN 1 1 0 3 6 0 3.0 3.0 0.029 0.051 INSECTS 1 1 0 3 6 0 7.0 7.0 0.018 0 .032 JOBS 2 1 1 7 6 8 4.0 7.0 1.0 0.050 0.029 0.077 KUDU 1 1 0 3 6 0 12.0 12.0 0.005 0.009 LIVESTOCK 14 6 8 47 35 62 7.1 7.3 6.9 0.269 0.208 0.349 LIVING_CREATURES 1 0 1 3 0 8 2.0 2.0 0.030 0.069
165 Table 4 3. Continued Frequency of r esponse Percent of r espondents (%) Average rank in list Item Both (B) Seronga (S) Gudigwa (G) B S G B S G B S G LOCAL_PEOPLE 17 9 8 57 53 62 6.7 7.9 5.4 0.301 0.237 0.385 MINES 2 2 0 7 12 0 7.5 7.5 0.038 0. 066 MONEY 3 1 2 10 6 15 11.0 11.0 11.0 0.035 0.022 0.051 MOSQUITO 1 1 0 3 6 0 17.0 17.0 0.004 0.007 MUD 1 0 1 3 0 8 15.0 15.0 0.002 0.005 OKAVANGO_DELTA 3 3 0 10 18 0 8.3 8.3 0.043 0.075 PAPYRUS 7 3 4 23 18 31 8.1 5.0 10.5 0.110 0.112 0.106 RAIN 4 4 0 13 24 0 8.0 8.0 0.081 0.143 RESPIRATION 1 1 0 3 6 0 5.0 5.0 0.025 0.044 RIVER 14 10 4 47 59 31 5.8 4.6 8.8 0.278 0.395 0.124 RIVER_PALM 1 1 0 3 6 0 7.0 7.0 0.013 0.024 ROAD_CONSTRUCTION 1 1 0 3 6 0 5.0 5.0 0.024 0.042 RUST 1 1 0 3 6 0 13.0 13.0 0.003 0.005 SAFARI_CAMP 1 1 0 3 6 0 12.0 12.0 0.003 0.005 SALVINIA 1 0 1 3 0 8 14.0 14.0 0.009 0.021 SHORTAGE 6 0 6 20 0 46 6.5 6.5 0.105 0.242 SNAKE 5 5 0 17 29 0 13 .4 13.4 0.033 0.058 STANDPIPE 6 5 1 20 29 8 9.5 9.2 11.0 0.069 0.108 0.018 TOILETS 2 2 0 7 12 0 5.0 5.0 0.049 0.087 TORTOISES 1 1 0 3 6 0 16.0 16.0 0.006 0.010 TOURISM 10 7 3 33 41 23 8.1 8.0 8.3 0.132 0.158 0.098 TRANSPORT 12 8 4 40 47 31 11.4 11.1 12.0 0.096 0.132 0.048 VEGETATION 8 5 3 27 29 23 9.3 8.8 10.0 0.140 0.170 0.102 WASHING_BAT HING 21 9 12 70 53 92 3.3 3.0 3.5 0.560 0.453 0.700 WATERING_GARDEN 3 3 0 10 18 0 4.0 4.0 0.079 0.140 WATER_ANIMALS 19 12 7 63 71 54 8.2 8.8 7.3 0.297 0.311 0.280 WATER_LILIES 14 11 3 47 65 23 8.3 8.6 7.0 0.219 0.302 0.110 WATER_POOL 3 2 1 10 12 8 8.0 8.5 7.0 0.048 0.061 0.031 WATER_TANK 1 0 1 3 0 8 13.0 13.0 0.003 0.006 WEATHER 1 1 0 3 6 0 12.0 12.0 0.016 0.028 WILDLIFE 12 5 7 40 29 54 7.5 7.6 7.4 0.216 0.165 0.283
166 Table 4 4. Summary of all concepts mentioned about wildlife during structured perceptions interviews Frequency of response Percent of respondents (%) Average rank in list Smi Item Both (B) Seronga (S) Gudigwa (G) B S G B S G B S G ANTELOPE 1 1 0 3 6 0 4 .0 4 .0 0.027 0.047 BEAUTY 7 6 1 23 35 8 4.3 4.2 5 .0 0.184 0.279 0.059 BIRDS 1 1 0 3 6 38 9 .0 9 .0 6.2 0.007 0.012 0.21 0 BUFFALO 11 6 5 37 35 0 7 .0 7.7 0.222 0.231 CITES 1 1 0 3 6 0 5 .0 5 .0 0.023 0.041 COMPENSATION 1 1 0 3 6 0 12 .0 12 .0 0.009 0.016 CONSERVATION 7 2 5 23 12 38 4.1 2.5 4.8 0.18 0 0.103 0.28 0 CONSUMPTION 1 0 1 3 0 8 2 .0 2 .0 0.031 0.072 COUNTRY_BENEFIT S 1 1 0 3 6 0 15 .0 15 .0 0.015 0.026 CROPS 16 13 3 53 76 23 8.1 8.2 7. 7 0.256 0.368 0.11 0 CULLING 2 2 0 7 12 0 12.5 12.5 0.012 0.022 DAMAGE 1 1 0 3 6 0 6 .0 6 .0 0.021 0.036 DANGEROUS 9 5 4 30 29 31 4.8 4.2 5.5 0.213 0.229 0.192 DESTROY_FEEDING_PLACE 1 0 1 3 0 8 5 .0 5 .0 0.024 0.056 DEVELOPMENT 9 4 5 30 24 38 10. 7 7.3 13.4 0.108 0.152 0.049 DISTANCE 1 0 1 3 0 8 3 .0 3 .0 0.029 0.067 DWNP 8 5 3 27 29 23 8.9 9.2 8.3 0.116 0.136 0.09 0 EDUCATION 6 3 3 20 18 23 11.5 10 .0 13 .0 0.077 0.092 0.056 EDUCATIONAL_PARKS 1 0 1 3 0 8 10 .0 10 .0 0.006 0.014 ELEPHANT 23 14 9 77 82 69 6.2 6. 1 6.4 0.473 0.546 0.377 FOOT_MOUTH 4 3 1 13 18 8 13 .0 14 .0 10 .0 0.037 0.059 0.008 FUTURE_GENERATIONS 1 1 0 3 6 0 7 .0 7 .0 0.015 0.027 GEMSBOK 1 1 0 3 6 0 4 .0 4 .0 0.026 0.045 GIRAFFE 3 1 2 10 6 15 3 .0 2 .0 3.5 0.082 0.055 0.118 GOVERNMENT 15 11 4 50 65 31 7.9 8.2 7 .0 0.276 0.349 0.181 GRASSES_REEDS 1 1 0 3 6 0 10 .0 10 .0 0.015 0.026 HUNTING_LICEN SES 3 1 2 10 6 15 13. 7 14 .0 13.5 0.021 0.011 0.034 HYENA 3 3 0 10 18 0 11 .0 11 .0 0.035 0.062 IMPORTANCE 4 1 3 13 6 23 1.3 1 .0 1.3 0.131 0.059 0.226 IVORY 2 1 1 7 6 8 7 .0 7 .0 7 .0 0.039 0.035 0.044 JOBS 19 10 9 63 59 69 10.1 11.1 9 .0 0 .236 0.206 0.274 KILLING_WILDLIFE 2 0 2 7 0 15 1.5 1.5 0.063 0.146 KUDU 2 1 1 7 6 8 7 .0 9 .0 5 .0 0.029 0.023 0.038 LEOPARD 3 2 1 10 12 8 7.0 7.5 6.0 0.050 0.059 0.038 LION 15 7 8 50 41 62 8.2 10.7 6.0 0.255 0.165 0.373 LIVESTOCK 15 11 4 50 65 31 9.5 10.6 6.3 0.207 0.236 0.169 LOCAL_PEOPLE 28 17 11 93 100 85 6.7 6.5 6.9 0.528 0.589 0.448
167 Table 4 4. Continued Frequency of response Percent of respondents (%) Average rank in list Item Both (B) Seronga (S) G udigwa (G) B S G B S G B S G MEAT 19 11 8 63 65 62 5.8 7.5 3.8 0.413 0.346 0.501 MINES 1 1 0 3 6 0 8.0 8.0 0.024 0.042 MIXED_FEELINGS 2 1 1 7 6 8 1.0 1.0 1.0 0.067 0.059 0.077 MONEY 21 13 8 70 76 62 11.2 7.9 8.1 0.352 0.410 0.276 OC T 19 11 8 63 65 62 1.0 10.9 11.6 0.169 0.201 0.126 OKAVANGO_DELTA 3 2 1 10 12 8 8.7 6.5 13.0 0.042 0.057 0.023 OPT 1 1 0 3 6 0 24.0 24.0 0.003 0.005 PIGERY 1 1 0 3 6 0 25.0 25.0 0.001 0.002 POACHING 9 3 6 30 18 46 6.9 6.3 7.2 0.1 73 0.117 0.246 POISONING_ANIMALS 1 0 1 3 0 8 4.0 4.0 0.027 0.062 PORCUPINES 1 1 0 3 6 0 10.0 10.0 0.015 0.026 PROMOTES_NAME 1 0 1 3 0 8 8.0 8.0 0.017 0.038 RED_LECHWE 2 2 0 7 7 0 5.0 5.0 0.048 0.084 RELATIONSHIP_COUNTRIES 1 1 0 3 3 0 6.0 6.0 0.027 0.047 RHINO 2 2 0 7 7 0 5.5 5.5 0.045 0.079 RIVER 6 4 2 20 20 15 9.0 9.3 8.5 0.092 0.104 0.077 SAFARI_OPERATOR 14 8 6 47 47 46 9.1 11.9 5.5 0.208 0.138 0.299 SALTS 2 0 2 7 7 15 3.0 3.0 0.055 0.126 SE DIBA 1 0 1 3 3 8 9.0 9.0 0.016 0.036 SELLING 1 0 1 3 3 8 2.0 2.0 0.031 0.071 SKINS_HIDES 9 4 5 30 30 38 5.6 6.5 4.8 0.204 0.149 0.277 SMALL_ANIMALS 1 0 1 3 3 8 7.0 7.0 0.013 0.031 SNAKES 1 1 0 3 3 0 6.0 6.0 0.022 0.039 S PACE_VEGETATION 1 1 0 3 3 0 3.0 3.0 0.031 0.054 TOURISM 17 12 5 57 57 38 4.6 4.4 5.0 0.427 0.551 0.266 TRANSPORT 5 1 4 17 17 31 13.4 13.0 13.5 0.019 0.005 0.037 TROPHY_HUNTING 5 1 4 17 17 31 7.2 9.0 6.8 0.086 0.020 0.174 VEGETATION 3 3 0 10 10 0 4.7 4.7 0.068 0.120 WARTHOG 1 1 0 3 3 0 3.0 3.0 0.029 0.051 WATRE 1 1 0 3 3 0 5.0 5.0 0.025 0.044 WATER_ANIMALS 7 7 0 23 23 0 9.4 9.4 0.113 0.199 WATER_LILIES 1 1 0 3 3 0 11.0 11.0 0.013 0.022 WATER_POOL 2 0 2 7 7 15 7.0 7.0 0.034 0.079 WILDLIFE_KILLS 2 0 2 7 7 15 7.0 7.0 0.035 0.080 WILD_ANIMALS 2 0 2 7 7 15 5.0 5.0 0.044 0.102 WILD_DOG 1 1 0 3 3 0 7.0 7.0 0.018 0.032
168 Table 4 5. Responses for chili peppers in wildlife r ating exercise Rating Frequency of response Weighted frequency 1 Very important 19 20.8 2 Somewhat important 6 6.07 3 Not very important 3 2.01 4 Not chosen for rating 2 1.12 Table 4 6. Number of concepts mentioned about water and wildlife during open ended interview questions Water: Number of distinct concepts Water: Total number of concepts mentioned Wildlife: Number of distinct concepts Wildlife: T otal number of c oncepts mentioned OCT 6 9 368 73 396 Seronga 60 216 59 234 Gudigwa 39 152 47 162 Table 4 7. Correlation s between social network measures and the n umber of water and wildlife concepts mentioned during open ended interview questions Social network m easure Statistical t est Number of water c oncepts Number of wildlife c oncepts Degree Centrality Pearson Correlation 0.216 0.145 Sig. (2 tailed) 0.251 0.445 Betweenness Centrality Pearson Correlation 0 .533* 0 .457* Sig. (2 tailed) 0.002 0.11 0 Ranking by Degree^ Pearson Correlation 0.430* 0.220 Sig. (2 tailed) 0.018 0.242 p= 0.05 ^ Rankings were assigned based on rel ative degree centrality values ( 1 17 in Seronga and 1 13 in Gudigwa ) This was done to account for the fact that values for centrality measures are relative to the village of the respondent. The negative value refl ects the fact that lower rankings reflect higher values for degree centrality. Table 4 8. Results of cultural consensus analysis for ratings of wildlife concepts Data s et Number of r espondents Ratio of F actor 1:Factor 2 Consensus Average a greement St d eviation a greement OCT 30 4.833 Yes 0.448 0.244 Seronga 17 6.515 Yes 0.561 0.195 Gudigwa 13 1.672 No 0.317 0.207
169 Table 4 9. Results of Independent Samples Kruskal Wallis Test to test for differences in cultural consens us agreement scores by multiple independent variables Social variable Asymptotic s ignificance Decision Ethnicity 0.006 R eject null hypothesis Gender 0.414 Retain null hypothesis Education 0.588 Retain null hypothesis N umber of village g roups 0.179 Retain null hypothesis p=0.05 Table 4 10. Correlation s between social network measures and level of agreement about the importance of wildlife concepts Social network m easure Statistical t est Level of a gre ement about w ildlife c oncepts Degree c entrality Pearson Correlation 0.068 Sig. (2 tailed) 0.722 Betweenness c entrality Pearson Correlation 0.233 Sig. (2 tailed) 0.215 Ranking by d egree ^ Pearson Correlation 0.209 Sig. (2 tailed) 0.267 p = 0.05 ^ Rankings were assigned based on relative degree centrality values ( 1 17 in Seronga and 1 13 in Gudigwa ) This was done to account for the fact that values for centrality measures are relative to the village of the respondent. The negative value reflects the fact that lower rankings reflect higher values for degree centrality.
170 Figure 4 1. Map of Seronga and Gudi gwa courtesy of Masego Dhliwayo of the Okavango Research Institute Un iversity of Botswana, Maun Botswana Khwai Sank
171 Figure 4 2. Respondents for perceptions interviews. Respondents were selected to represent a range of degree and betweenness centrality values.
172 Figure 4 3 Visualization of frequency of concepts menti oned during open ended water perceptions questions. Created at http://www.wordle.net
173 Figure 4 4 Most frequently mentioned sources of information about water.
174 Figure 4 5 Visualization of frequency of concepts mentioned during open ended wildlife perceptions questions. Created at http://www.wordle.net Figure 4 6 Box plot of agreement scores for three ethnic groups responding to wildlife rankings
175 Figu re 4 7 Mos t frequently mentioned sources of information about wildlife.
176 C HAPTER 5 CONCLUSIONS We as a people should be encouraging and educating each other about conserving wildlife. S eronga resident 2010 Water and wildlife are both of gr eat importance to the people of the Okavango Delta though for different reasons. W ater is seen a basic necessity and part of everyday tasks W ildlife which was on c e a source of food for subsistence hunting families is now seen as a resource for those be nefiting from CBNRM and/or as a menace among those whose crops are raided, livestock killed, or lives threatened. Regardless of the natural resource in question, communication is important to ability to understand and participate in decision maki ng and adapt to changing resource conditions. People need information to learn, adapt, and act (Kaplan and Kaplan 1982), and in remote, rural settings, individuals tend to rely on personal communication networks to get the information they need. These stud ies investigated information flows and perceptions about water and wildlife to better understand how different variables may impact communication and the integration of available information into the mental models of rural community members. In chapter 2, researchers found that community size is an important variable affecting the characteristics of communication networks in rural villages. Personal network data were combined to create whole networks for each village, from which multiple network measures we re calculated. Results indicated that larger villages were less densely connected than smaller ones, in which most people tended to see and interact with one anot her. As community size increased the measure of degree
177 centralization decreased indicating t hat networks tend to be less centralized around a few highly connected individuals in larger villages and communication was more evenly distributed over the network as a whole T he ward system which essential creates villages within large villages that h a ve their own traditional authorities and meetings, is likely to contribute to this finding. Though community size is an important variable affecting network characteristics, the measures for Sankoyo resembled the considerably larger Gudigwa in some cases m ore so than t he similarly sized Khwai. This suggests additional factors affect network outcomes such as proximi ty to the regional center and intra community diversity that can make smaller communities communicate as much larger villages do In sum, commu nity size is an important variable in the structure of communication networks, and communication strategies can be specifically targeted to accommodate the characteristics of different community sizes. S maller more densely connected rural villages may ben efit from interventions that introduc e outside ideas and adaptive strategies This may allow smaller communities to overcome tendencies toward homogeneity of thinking and integrate new information that might otherwise come from a larger number of perspecti ves and the presence of weak ties that tend to be characteristic of larger villages. Different strategies can be used i n larger villages, where people tend to break into sub groups both formally, through the ward system, and informally by gender, geographi c area, ethnic group, kinship, or shared water sources Communication can be targeted sp ecifically to these sub gr oups, and when possible, decision making processes that may be best addressed through face to face
178 interactions could be delegated to wards to enhance overall participation through these opportunities at smaller scales. The results presented in chapter 3 reveal the importance of gender and ethnicity in communication networks. Analyses of homophily revealed that men and women tend to have more me mbers of their own gender in their communication networks, and this level of homophily is significantly higher for men than women, particularly in Gudigwa. Despite t he fact that women work with and use natural resources in sub group s separate from men, men have the information about wildlife. B oth men and women seek more information about wildlife resources and the local CBNRM program from men. This holds true among those consulting the traditional authorities, where men received more information from the c hiefs and headmen than women do. Social network characteristics differ by gender also, with men having more connections, on average, to other people in their network. It appears that men are more likely to be able to bridge different sub groups in all vill ages except Sankoyo indicating that they may be better able to take positions of power by bro kering information across groups Women participated less in village meetings related to CBNRM, which focus far more on wildlife than natural resources usually as sociated with women like grasses and reeds. Ethn icity was also found to be an important variable in this study. T he propor tion of people that respondents named from within their own ethnic group was disproportionate ly hig h, and p eople received sig nificant l y more information from those in the same ethnic group Members of the dominant ethnic group in each village tended to be better connected withi n the village and had more opportunities to broker information across sub groups than members of minority ethnic groups
179 The empirical evidence presented in chapter 3 make s it clear that gender and ethnicity are impor tant variables to consider when designing strategies to more effectively communicate with and foster participation in decision making in rural villages I nterventions that actively and deliberately engage minority ethnic groups and women in information exchanges and decision making processes may help these groups overcome the communication and participation challenges that were observed. Finding ways to add these voices to participatory processes may provide insights and ideas valuable to innovation and adaptation. Results presented in chapter 4 indicated that there were important differences between the ways people perceived water and wildlife. Mental m odels appeared to be more c omplex for wildlife than water, which reflects the more complex nature of decisions surrounding wildlife resources compa red with water in this context. It also reflects the needs for individuals to make more complex decisions Wh en there is more of an opportunity to make a difference through decision making, as was the case for wildlife, people tend to build more complex mental models Further investigation of perceptions about wildlife revealed that t here was general agreement a cross all of the respondents in terms of which concepts were most important about wildlife; however, when compared village by village, this consensus did not exist among Gudigwa respondents T hese results and additional ethnographic study suggest that ther e may be a division in thinki ng within Gudigwa between older community members who had experience with wildlife before the establishment of CBNRM and those who only came to know about wildlife concepts and issues since CBNRM was established. The extent to
180 of different wildlife concepts was related to two factors. Ethnicity was found to be a significant factor, as were p social network measures. Specifically, people with more direct connections to others w ithin the ir vi llage network were more likely to provide responses that represented the consensus answers for that village. This may mean that unique and potentially innovative voices are not being heard and integrated where they could provide important ben efits for shared lear ning and participatory decision making. These results suggest that the complexity of wildlife in this context is reflected in though the formation of these mental models may be mediated by social factors in thes e rural villages. As in previous chapters, these results suggest that to effectively communicate messages considered important to rural communities, more may need to be done to reach people in minority ethnic groups and those in less central positions with in these village networks. Also important are concerted efforts to listen to the voices of those whose perspectives differ from the most common ones, since these people could contribute new and diverse perspectives based on different experience and knowled ge. This could, i n turn, contribute to innovative thinking about adaptive strategies if these voices are effectively brought into the fold. Broader Significance and Recommendations n c ommunication strategies specifically for certain villages, or types of villages, could bring great improvements to r ural communication. These strategies can take advantage of the benefits of both small and large networks. In small villages like Khwai, w here the community is highly connected to one another and not very connected to people outside of the village, concerted efforts to incorporate information and perspectives from outsi de of the community may help facilitate innovative thinking and aid in th e long term
181 adaptive capacity of these villages Since it is easier for people to meet face to face in smaller communities, op portunities for social learning from participatory processes allow people to deli berate and learn from one another to access and i ntegrate new information. Intervention efforts can be spent working to integrate minority voices, both women and those of minority ethnic groups. In larger communities such as Gudigwa and Seronga, it is difficult to effectively reach everyone While kgotl a meetings are important information dissemination forums, efforts to reach identified sub group s within the village (e.g., by ward, ethnic group, gender) could greatly facilitate people receiving information directly who might not otherwise have access to reliable information. These s maller sub group s provide m ore opportunities for face to face deliberation which can build trust, facilitate learning and allow for v oices that may not otherwise be heard to be shared While this idea is not new, it can be m ore deliberately built into how decision making processes are structured in larger villages, particularly within CBNRM, which tends to operate at the village or multi village level. In the latter case, when several villages like those in OCT come together to make decisions about natural resources, the majority of people have little chance of having their voice heard and learning from decision making processes firsthand. Moving some of these discussions down to smaller scales and sub groups within the whole could provide great benefits for NRM learning, and adaptation. Facilitating communication and participation is a challenging prospect in remote rural settings. This is made all the more difficult when social and cultural circumstances create centers o f power with one dominant gender and certain ethnic groups. CBNRM, and rural communities in general can benefit from deliberate attention paid to gender
182 and ethnicity. Networks are often more complex than they may first appear, and information disseminati on and training activities about natural resources and decision making processes can be targeted specifically to minority groups. Acknowledging the current social realities these minorities face, environmental communication and engagement strategies can ai m to specifically reach and facilitate information exchanges with these groups Information is a right of all community members with regard to CBNRM, and if existing social structures prevent equal distribution and exchange of information, specific efforts may be necessary to ameliorate this. Working to integrate these mem bers of minority groups into participatory decision making processes in deliberate and meaningful ways is a targeted means of reaching marginalized groups and individuals in rural villages Though this research was not an evaluation of CBNRM in Botswana by intent or implementation, the findings offer suggest ions that can be applied to improve the CBNRM program. This research is particularly important in light of climate change threats, whi ch and affect human health, ag riculture and food security, ecosystems, and wildlife. Research findings about communication in these rural contexts where people are vu lnerable to these changes can be applied across several fields to design better strategies for informing and engaging rural community members in decision making processes. Understanding the communication connections among people, their perceptions and the ir decisions, will provide vital information about how individuals can learn from others, become empowered through enhanced communication, build
183 adaptive capacity for greater social ecological resilience, and ultimately make more informed decisions. Valid ity, Reliability, and Objectivity The research design and instruments used in this project resulted in strong internal, c ontent, and construct validity. Triangulation of a variety of data sources, including data collected from rural community members, tra ditional authorities, and government officials provided multiple data sources that were used to cross check across respondents. The methods were also triangulated, such that qualitative data from interviews and field observations were used to check and co ntextualize the results from quantitative methods such as social network surveys and structured perceptions interview segments. This mix of quantitative and qualitative data allowed researchers to increase internal validity while gaining further insights i nt o the reasons behind responses Repeated pil ot testing and preliminary studies enhanced the content validity of the instruments used. In nearly all cases, theory driven hypotheses were confirmed by data, indicating strong construct validity. The purposi ve selection of villages included in these studies limit s the external validity of the results somewhat. T he villages selected represented key characteristics important to this study, as was confirmed by an expert on CBNRM in the region to be the best for this study (J. Mbaiwa pers. comm. ); however, this site selection was purposive, limiting the generalizability of the results. At the village level, the people selected for social network surveys were, for the most part random, with the exception of a small number of purposively included leaders in all but one of the villages. This increased the validity of these data and allowed results to be generalized to each
184 village. The relatively small, non random sample used in the perceptions study (chapter 4) limit s the external validity of this study in particular. The reliability of data was supported in several ways. First, respondents with similar characteristics tended to respond in similar ways to the questions. Also, on a few occasions when technical problems required that some questions be answered again, these were observed to be consistent. To maximize reliability, question wording was adjusted on many occasions for both the social network survey and perceptions interview to reflect changes that needed to b e made based on pilot testing. In each instrument, options were given for many questions that allowed respondents to indicate that they did not know the answer to a question, preventing people from answering when they had insufficient data. This was not t h e case, however, for the alter alter pair questions in the social network surveys that, by design, require some quantification of the relationship between two people respondents named within their network. This may have prompted some people to respond that the two people did not speak at all, when it might be that the respondent did not know. Researchers worked with local research assistants to enhance the o bjectivity of data recorded. Meeting notes were indexed and evaluated using objective criteria, when possible (e.g., gender of speaker). Local research assistants were extremely helpful in enhancing objectivity. These r elationships developed considerably over the course of fieldwork, and trust and openness in communication increased as well. Conversations with assistants throughout the field research period shed important light on the context of observations and how these observations might be bes t interpreted and recorded to maintain objectivity. Assistants were often able to express both their own though ts, and
185 the thoughts of those holding other perspectives and positions within their village. These individuals were invaluable in providing information and contacts that allowed the researchers to understand the historical and political contexts in which o bservations were made. Meetings with communities at the end of fieldwork provided an additional opportunity to learn from the community as they reflected on the preliminary findings, which was also extremely valuable in contextualizing observations and dat a. Future Research This research provides insights about some of the many variables affecting communication networks and perceptions about natural resources in the Okavango Delta. Further study in additional villages, both those engaged in CBNRM and not, may reveal interesting information about the extent to which CBNRM affects information flows and perceptions about natural resources, and wildlife in particular. R esearch involving other single and multi v illage trusts might provide additional empirical ev idence supporting the results presented here, or it may reveal additional factors affec ting communication. A larger sample size from across the Okavango Delta or throughout Botswana would allow for greater generalizability of the results Though the CBNRM program in each country has unique features, a comparison across countries in southern Africa might reveal how different program governance structures and local conditions lead to variations in communication and participation about CBNRM. Research into ad ditional variables affecting communication and participation in CBNRM may also provide results that could inform strategies for enhancing engagement in decision making processes. Power and elite capture were addressed in this study indirectly by investigat ing gender and ethnicity; a more explicit investigation of how power structures affect commu nication and perceptions would help build
186 understanding about the complexity in these rural village systems. Elite capture studies often look at the extent to which resources are distributed within a community. The research presented in this dissertation suggests that linking elite capture directly with information may yield important insights about how and why power structures are reinforced in rural communities, pa rticularly those engaged in CBNRM. With regard to the perceptions study, future research could focus on continued methodological improvement s and additional theory testing The research instrument and analysis techniques could be tested in addi tional locat ions in the Okavango Delta and Botswana, as well as in other rural contexts. From a theoretical standpoint, fu ture study could focus on testing hypotheses specifically related to t he Reasonable Personal Model, and could also move beyond perceptions to deci sions and actions, to investigate their decisions and actions. Summary The empirical evidence presented in this dissertation contributes to the growing body of knowl edge about social network measures and their relationship with variables critical to learning and adaptive capacity. Understanding how people access, exchange, integrate, and apply information about natural resources is critically important to e n ha n cing pa rticipation in c ommunity conservati on, providing information in ways that may help build adaptive capacity, and achieving conservation and development goals. People in rural communities often rely on personal social networks for in formation and c ommunity size, gender, and ethnicity are variables important to these information flows. Designing communication strategies and participatory decision making processes with these factors in mind can help improve engagement among rural community
187 members and build th e capacity of marginalized groups. This is not only important for the effectiveness of cu rr ent CBNRM programs in the Okavango Delta, but also for community based conservation programs elsewhere that have similar goals of incorporating local voices into dec isions and management.
188 APPENDIX A SOCIAL NETWORK ANALYSIS SURVEY General Information The following is a list of interview questions included in the social network analysis survey Questions were programmed into EgoNet software on field laptop c omputers. During interviews, questions appear ed automatically on the computer screen in Setswana and English. Questions follow ed the question logic assigned in EgoNet Questions about the Ego ( Respondent ) 1. Bong jwa mmodiwa/monna kapa mosadi? What is the gender of the re s pondent? Monna Male Mosadi Female 2. O tshotswe leng? What year were you born? Answer: 3. O nale lebaka le le kae o nna mo motseng? How long have you lived in the village (in years)? Answer: 4. O dirile kgotsa o badile bokae kwa sekolong? What level or certificate did you complete in school? Answer: 5. Le nna le le kae mo lwapeng? How many people live in your household, including yourself? Drop down menu: 1 20 or more 6. Letso la gago ke mang? What ethnic /language group do you come from? Bayeyi Basarwa Bambukushu Baherero Basubiya Tse dingwe Othe r If response to #6 is Tse dingwe Other #7 7. Letso la gago ke mang? What ethnic /language group do you come from?
189 Answer: 8. Ba lelwapa la gago ba ga kae metsi? Where does your household get water for drinking? Pompo ya b othe Public standpipe or tap Pompo ya mo lwapeng Household standpipe Sediba sa batho Bor ehole/well used only by humans Sediba sa diphologolo Boreholle/well also used by livestock Segelo sa batho mo nokeng River area used only by humans Manwelo a diphologolo mo nokeng River area also used by livestock Pula Rainwater Mo go ngwe Other source Ga ke itse I don't know 9. A go nnile thata kana mothofo go bona metsi a a nowang mo dingwageng tse di sa tswang go feta? Has it become harder or easier to access drink ing water in the last few years? Thata/bokete Harder Mothofo Easier Tswana fela le pele Remained the same Ga ke itse I don't know 10. A metsi a a nowang a phepha? Is your drinking water clean and healthy to drink? No Nyaa Sometimes Nako d ingwe Yes Ee 11. Ko o gelelang teng metsi go tshepega go le kae? How reliable is your main water source? Ga go tshepege gotlhelele Not at all reliable Go a tshepega nako tse dingwe Somewhat reliable Go a tshepega Very reliable 12. Mo lebakeng la ngwaga le le fitileng a o kile wa amogela o berekile golo gongwe? Over the past year, have you been paid for any type of employment? No Nyaa Yes Ee 13. A tiro ya gago ya mo bogaufing e ne e le e e sennela ruri kana ya nakwana, o ne o berekela kae ? Was your most recent employment permanent or temporary, and where was the work?
190 Tiro ya sennelaruri mo motseng Permanent wage employment in village Tiro ya sennelaruri mo lefelong la lekgotla/OCT Permanent wage employment in OCT concession area Tiro ya sennelaruri mo Maun Permanent wage employment in Maun Tiro ya sennelaruri ko ntle ga maun kana mo motseng Permanent wage employment someplace other than village or Maun Tiro ya nakwana mo motseng Temporary/ocassional work in village Tiro ya nakwana mo lefelong la lekgotla/OCT Temporary/ocassional work in OCT concession area Tiro ya nakwana mo Maun Temporary/ocassional work in Maun Tiro ya nakwana ko ntle ga Maun kana mo motseng Temporary/ocassional work someplace other than village or Maun 14. Ke efe kompone e e neng e go duelela tiro e? What type of company or organization paid you for this work? Lekgotla la sechaba OCT Kgosi kgotsa makgotlana a mangwe a motse Village organization or Chief Mogwebi mo lefelong Private Safa ri Operator Bagwebi ba bangwe Other Private Company Goromente Government Tse dingwe Other 15. A o leloko la lekgotla la [trust name] ? Are you a member of [the community CBNRM trust] ? Nyaa No Ee Yes Ga ke itse I don't know 16. A o na le m aemo mo lekgotleng? What is your role or position within OCT? Ga gona, leloko fela None, member only Modulasetilo Board Chairperson Mokwaledi Board Secretary Motshwari wa madi Board Treasurer Leloko la komiti Board Member Mookamedi Trus t Manager Motshwara dibuka Trust Accountant Mogolwane wa bathokomedi ba lekgotla Chief Escort Guide Motlhokomedi wa lekgotla Escort Guide Mookamedi wa camp site Campsite Manager Mokgweetsi Driver Lebereki la lekgotla lengwe fela Other Tru st Employee Maloko a komiti mo motseng Village Trust Committee Member 17. A o ne o tseneletse phuthego ya bothe ya OCT ngwaga (AGM) ya bofelo ya
191 phalane ngwaga o o fetileng? Did you attend OCT's Annual General Meeting (AGM) last October? Nyaa No Ee Yes Ga ke itse I don't know 18. A go nale mongwe ko nte ga gago yo otseneletseng phuthego ya bothe ya OCT (AGM) ya bofelo ya phalane ngwaga o o fetileng? Did someone other than yourself attend OCT's AGM this year? Nyaa No Ee Yes Ga ke itse I don't know 19. Fa re lebeletse palo ya diphuthego tse di tseneng, o ka tswa o tseneletse dile kae -lefela, ko tlase ga sephatlo, sephatlo, go feta sephatlo, kana tsotlhe? Compared to the number of kgotla meetings that were held, about how many meetin gs did you attend in the last year -none, less than half, about half, more than half, or all? Lefela None Ko tlase ga sephatlo Less than half Sephatlo About half Go feta sephatlo More than half Tsotlhe All Ga ke itse I don't know 20. A o t saya karolo mo dikomiting kana mo ditlhopheng dingwe tsa motse? Are you involved in any village based committees or groups? Nyaa No Ee Yes 21. Ke dife ditlhopha tse o tsayang karolo mo go tsone mo motseng? What committees or groups are you invol ved with in the village? Answer: 22. Gantsi o tsaya jang kitso ka kgaolo le lefatshe ka kakaretso? How do you usually get information about what is going on in the region and in the country? Ga ke bone kitsiso epe I do not get any information Ke bolele lwa ke batho ba bangwe Other people tell me Ka dipampiri tsa dikgang Newspapers Ka seramamowa Radio Ditshwantsho tsa motsikenyego Television Radio ya ditlhaeletsano Two way radio Ka tse dingwe Other source If response to #22 is Tse dingwe Other #23
192 23. Gantsi o tsaya jang kitso ka kgaolo le lefatshe ka kakaretso? How do you usually get information about what is going on in the region and in the country? Answer: 24. Gantsi o tsaya jang kitso ka metsi? How do you usually get informatio n about water resources? Ga ke bone kitsiso epe I do not get any information Ke bolelelwa ke batho ba bangwe Other people tell me Ka dipampiri tsa dikgang Newspapers Ka seramamowa Radio Ditshwantsho tsa motsikenyego Television Radio ya ditlh aeletsano Two way radio Ka tse dingwe Other source If response to #24 is Tse dingwe Other #25 25. Gantsi o tsaya jang kitso ka metsi? How do you usually get information about water resources? Answer: 26. Gantsi o tsaya jang kitso ka diphologolo tsa naga? How do you usually get information about wildlife? Ga ke bone kitsiso epe I do not get any information Ke bolelelwa ke batho ba bangwe Other people tell me Ka dipampiri tsa dikgang Newspapers Ka seramamowa Radio Ditshwantsho tsa mots ikenyego Television Radio ya ditlhaeletsano Two way radio Ka tse dingwe Other source If response to #2 6 is Tse dingwe Other # 27 27. Gantsi o tsaya jang kitso ka diphologolo tsa naga? How do you usually get information about wildlife? Answer:
193 A lter Prompt Question Bolela batho ba le masome a mararo le botlhano ba dingwaga tse di lesome le borobabobedi le go feta ba o buileng le bone mo kgweding e e fitlileng. Go bua ka kakaretso e seng fela ka molomo. Tsenya maina a bone le difane. Ke tla a go bolelela fa a lekane. Please name 35 people aged 18 or older w ith whom you have communicated with during the last month. Include first and last names. I will let you know when you have finished. Alter Questions In each of the follo w ing questions, $$ is replaced by an alter name. Each of the 35 alter names is substituted in turn with all questions asked about each of the alters 28. A $$ ke monna kana mosadi? Is $$ a man or a woman? Monna/Rre Man Mosadi/Mme Woman 29. $$ o wa lets o lefe? What ethnic /language group does $$ come from? Bayeyi Basarwa Bambukushu Baherero Basubiya Tse dingwe Other Ga ke itse I don't know If response to # 30 is Tse dingwe Other # 31 30. $$ o wa letso lefe? What ethnic /language group d oes $$ come from? Answer: 31. A $$ o nna mo motseng bogolo sephato sa ngwaga? Does $$ live in the village for at least half of the year? Nyaa No Ee Yes If response to #31 is Nyaa No #32 32. $$ o nna kae? Where does $$ live? Answer:
194 33. O tsaya kitso ee kae ko go $$ ka metsi? How much information do you get from $$ about water? Ga gona None E e seng kalo Some E e ntsi A lot If response to #33 is E e seng kalo Some or E e ntsi A lot #34 If response to #33 is Ga gona None #35 34. O t shepa go le kae dikgang kgotsa dikitsiso tse o di tsayang mo go $$ ka metsi? How much do you trust the information you receive from $$ about water? Ga gona None E e seng kalo Some E e ntsi A lot 35. O fa $$ kitso ee kae ka metsi? How much info rmation do you give $$ about water? Ga gona None E e seng kalo Some E e ntsi A lot 36. O tsaya kitso ee kae ko go $$ ka diphologolo tsa naga ? How much information do you get from $$ about wildlife ? Ga gona None E e seng kalo Some E e ntsi A lot If response to #3 6 is E e seng kalo Some or E e ntsi A lot #37 If response to #36 is Ga gona None #38 37. O tshepa go le kae dikgang kgotsa dikitsiso tse o di tsayang mo go $$ ka diphologolo tsa naga ? How much do you trust the information you receive from $$ about wildlife ? Ga gona None E e seng kalo Some E e ntsi A lot 38. O fa $$ kitso ee kae ka diphologolo tsa naga ? How much information do you give $$ about wildlife ? Ga gona None E e seng kalo Some E e ntsi A lot
195 39. O ts aya kitso ee kae ko go $$ ka [the community CBNRM trust] ? How much information do you get from $$ about [the community CBNRM trust] ? Ga gona None E e seng kalo Some E e ntsi A lot If response to #3 9 is E e seng kalo Some or E e ntsi A lot #40 If response to #39 is Ga gona None #41 40. O tshepa go le kae dikgang kgotsa dikitsiso tse o di tsayang mo go $$ ka [the community CBNRM trust]? How much do you trust the information you receive from $$ about [the community CBNRM trust] ? Ga gona None E e seng kalo Some E e ntsi A lot 41. O fa $$ kitso ee kae ka [the community CBNRM trust]? How much information do you give $$ about [the community CBNRM trust] ? Ga gona None E e seng kalo Some E e ntsi A lot Alter Pair Question T he alter pair question is asked for each combination of alters (e.g., alter 1 and alter 2, 1. Go kgonafala jang gore ($$ 1) ______________ le ( $$ 2) ______________ ba buisane fa o seyo? How likely is it that ( $$ 1) ______________ and ( $$ 2) ______________ would communicate with each other when you are not around? Ga go kgonafale g otlhelele Not at all likely Go na le kgonafalo Somewhat likely Go a kgonafala Very likely
196 APPENDIX B FREE LIST INTERVIEW GUIDE: WATER AND WIL DLIFE PERCEPTIONS Name: Gender: Birth Year: Main Question (Wildlife): What are the things you think of when you think about wildlife? Pr obing Questions : Can you think of any (other) problems associated with wildlife? Can you think of any (other) benefits associated with wildlife? Can you think of any (other) uses for wildlife? Does anything else come to mind when you think about wildlife in your daily life? Is there anything more that comes to your mind about wildlife? Main Question (Water): What are the things you think of when you think about wildlife? Probing Questions : Can you think of any (ot her) problems associated with water ? Can you think of any (other) benefits associated with water ? Can you think of any (other) uses for water? Does anything else come to mind when you think about water in your daily life? Is there anything more that comes to your mind about water ? [ Note: Order of free lists (wildlife or water) was alternated between respondents. ]
197 APPENDIX C GOVERNMENT AGENCY SE MI STRUCTURED INTERVIEW GUIDE The following questions were used to guide interviews with government agency off icials and understand priority messages and rural communication strategies with regard to water and wildlife. What are the primary responsibilities of your agency and/or depa r tment? In what ways and to what extent do you work on issues related to water a nd/or wildlife? I n what ways and to what extent /how often does your agency interact with rural communities? Does your agency have any specific techniques or strategies it uses to communicate with rural communities? What are the priority messages your a gency is trying to convey to rural communities right now? Specifically to Khwai, Sankoyo, Gudigwa, and Seronga? Do you think that your agency has been effective in meeting your communication goals? Why or why not? Are there any messages that you have tri ed to convey in the past? How successful has this been? Why do you think this is so? What other government agencies work with water or wildlife issues? Could you please provide me with contact information for an appropriate official in that/those office( s)? Is there anything else I should be thinking about with regard how the Botswana government is working on water and wildlife issues in rural communities?
198 APPENDIX D PERCEPTIONS OF WATER AND WILDLIFE: INTERV IEW GUIDE Opening to Interview : Transla and carried in interview kits. 1. Introduce selves 2. Explain that we are conducting a study about water and wildlife resources in this village. We will ask a few brief questions about water and t hen, separately, ask more detailed questions about wildlife (diphologolo tsa naga). 3. Their participation is voluntary. There is no penalty for stopping. We are happy to answer any questions at any time that will help them complete the interview; please feel free to ask. 4. This will take about an hour. 5. We will take a photo at the end of the interview, and they will receive a printed copy as a gift of thanks for their time soon. 6. There are no right answers or wrong answers to the questions we a sk. We want to know what YOU think. We appreciate your honest answers to all of these questions. If you think of anything as we are talking, please say it, even if it is not the answer to the specific question we are asking. The ans wers to the followin g questions should be entered on the interview guide sheet. Opening Questions about Respondents: Leina: Name of r espondent: Bong jwa mmodiwa/monna kapa mosadi? What is the gender of the respondent? O tshotswe leng? What year were you born? O n ale lebaka le le kae o nna mo motseng? How long have you lived in the village (in years)? O dirile kgotsa o badile bokae kwa sekolong? What level or certificate did you complete in school? Letso la gago ke mang? What ethnic group do you come from? A o tsaya karolo mo dikomiting kana mo ditlhopheng dingwe tsa motse? Are you involved in any village based committees or groups?
199 M etsi / Water : As you go through questions #1 5, lay out cards that match was the respondent has said, after he or she completes each question. When you lay them out, group them according to the explanations they give, so that things that were mentioned together are placed on the table in a group. This section should be done quickly relative to the wildlife section. 1. Ke eng se se tlang mo tlhaloganyong ya gago fa o akanya ka metsi? What things come to mind when you think about water? Put a number on the paper next to each thing mentioned, in the order that they are mentioned. If they say something that is not already on the list, add it at the bottom and put a number next to it in the correct order. Translate any explanations about the things mentioned into English. If the respondent says only the things and does not explain what they mean, ask question #2. 2. O raya jang fa o re ___________? O ne o akantse dilo dife fa o ne o re_______________? Ke kopa gore o mpolelele gore o raya fa o re__________? What do you mean by _____________? What things were you thinking of when you said __________? Can you pl ease tell me wh at you mean by __________ ? you mean by that? Do the same thing you did for question #1 above, by marking each additional thing mentioned in the same column as question #1. 3. Jaanong o nale dikarata fa pele ga gago.Ke dife dilo gape tse di tlang mo tlhaloganyong ya gago fa o akanya ka metsi? Now you have some cards in fro nt of you. What other things come to mind when you think about water? Do the same thing you did for question #1 above but in a new column, by marking each additional thing mentioned. 4. [Dingwe tsa dilo tse o di umakileng di amana le noka.] A go nale dingwe gape tse o di akantseng tse di amanang le noka? [ Some of the things you mentioned are related to the river.] Are there any other things you think of related to the river?
200 A o tla batl a go bua sengwe ka metsi a a mo nokeng? Would you like to say anything related to water in the river? 5. [Dingwe tsa dilo tse o di umakileng dingongorego/mathata a metsi.] A o nale dingongorego/mathata(mangwe) ka metsi? [ Some of the things you mentioned are concerns about water.] Do you have any (other) concerns about water? concerns. 6. Fa obatla go itse sengwe ka metsi o ka botsa mang? If you wanted more informatio n about water, from who m would you get the information ? Write down a list of names on interview guide When you are finished with this initial list, mark // 7. A go nale batho ba bangwe gape ba ba o ka tsayang kitso mo go bone ka metsi? Ke kopa gore o bue batho botlhe ba o ba akantseng, e ka nna ba mono kana ba ba sa nneng mono. Are there any other people you get information from about water? Please list all people you can think of, both inside and outside of the village. 8. Ke eng se goromente a b atlang lo se itse ka metsi? A o gakologelwa sengwe se goromente a kileng a se bua mabapi le metsi? What does the government want you to know about water? Can you remember any specific messages from the government about water? 9. Ke dife ditsela tse goromente a di dirisang go gorosa melaetsa /dikitsiso tsa metsi? How/In what ways does the government tell you about water? After they answer this question, take up all of the cards and start on wildlife section. Diphologolo tsa N aga / Wildlife : As you go through questions #1 10, lay out cards that match was the respondent has said, after he or she completes each question. When you lay them out, group them according to the explanations they give, so that things that were mentioned together are placed on the table in a group.
201 1. Ke dife dilo tse di tlang mo tlhaloganyong ya gago fa o akanya ka diphologolo tsa naga? What things come to mind when you think about wildlife (diphologolo tsa naga)? Put a number on the paper next to each thing mentioned, in t he order that they are mentioned. If they say something that is not already on the list, add it at the bottom and put a number next to it in the correct order. Translate any explanations about the things mentioned into English. If the respondent says on ly the things and does not explain what they mean, ask question #2. Ga gago. 2. O raya jang fa o re ___________? O ne o akantse dilo dife fa o ne o re_______________? Ke kopa gore o mpolelele gore o raya fa o re__________? What do you mean by _________ ____? What things were you thinking of when you said __________? Can you please tell me what you mean by _________________? 3. Jaanong o nale dikarata fa pele ga gago. Ke dife dilo gape tse di tlang mo tlhaloganyong ya gago fa o akanya ka diphologolo tsa naga? Now you have some cards in front of you. What other things come to mind when you think about wildlife (diphologolo tsa naga)? 4. [Dingwe tsa dilo tse o di umakileng ke mathata/dingongorego mabapi le diphologolo tsa naga./ A o nale mathata/ding ongorego (dingwe) ka diphologolo tsa naga? [ Some of the things you mentioned are concerns about wildlife (diphologolo tsa naga).] Do you have any (other) concerns about wildlife (diphologolo tsa naga)? 5. A motes wa lona o tswelwa mosola (mongwe) ka diphologolo tsa naga? Does your village get any (other) benefits from wildlife (diphologolo tsa naga)? 6. A ka bowena go nale mosola mongwe o o o tsayang mo diphologolong tsa naga? Do you personally get any (other) benefits from wildlife (diphologolo tsa naga)? 7. A go nale mosola (mongwe) o motse wag ago o o boning mo makgobokgobo(OCT)? Does your village get any (other) benefits from the Trust? 8. A go nale mosola(mongwe) o wena o o bonang mo makgobokgobo(OCT)? Do you personally get any (other) bene fits from the Trust? Hand the respondent all of the cards they have mentioned.
202 9. Ke kopa gore o farologanye dikarata tse go ya mosola wa tsone mo go wena. Di farologanye jaana: Di mosola thata, Di mosola go se kae, Ga di mosola mo go kalo. Please pu t each of the cards you have in a group: Very Important, Somewhat important, or Not as important. 10. Ke lebogetse dikarabo tsa gago. Go na le dingwe tse di ka nnang mosola kana tsa seka tsa nna mosola ka diphologolo tsa naga mo go wena. Ke nale dikarata tse dingwe tse di setseng tse ke tla di go bontshang. Fa nngwe ya tsone e le mosola mo go wena, mpolelela gore e mosola jang le gore e amana jang le diphologolo tsa naga. Thank you very much for your answers. There are some other things that may or may not be important to you about wildlife (diphologolo tsa naga). I have cards with some additional things on them to show you. If any of these are important to you, please tell me in a few words why they are important and how they are related to wildlife (diphologolo tsa naga ). Go through each of the cards, saying the word and showing the picture. Lay each card directly on t op of the last, so only the most recent card is showing. After talking about each card, ask: A o bona go le mosola thata, go le mosola go sekae kana go se mosola mo go kalo? Do you think this is Very, Somewhat or Not So Important? Interviewer or respo ndent then places each card in the correct group. 11. Fa o batla go itse thata ka ga diphologolotsa naga o o ka botsa mang? If you wanted more information about wildlife (diphologolo tsa naga), who would you get the information from? Write down a list of names. When you are finished with this initial list, mark // 12. A go nale batho gape ba ba go fang kitso ka diphologolo tsa naga? Ke kopa gore o bitse batho botlhe ba o ba akantseng, e ka nna ba mono kana go sele. Are there any other people you g et information from about wildlife (diphologolo tsa naga)? Please list all people you can think of, both inside and outside of the village. 13. Ke eng se goromente a batlang lo se itse ka diphologolo tsa naga? A o gakologelwa molaetsa mongwe o gorome nte a kileng a o fa o o mabapi le diphologolo? Go nale mo go ngwe gape?
203 What does the government want you to know about wildlife (diphologolo tsa naga)? Can you remember any specific messages from government about wildlife (diphologolo tsa naga)? Anythi ng else? 14. Goromente o dirisa ditsela dife / gorosa jang melaetsa ka ga diphologolo tsa naga? How/In what ways does the government tell you about wildlife (diphologolo tsa naga)? After they answer this question, take up all of the cards. 15. A o kile wa utlwa sengwe ka ka go dirisa pepere/phirihiri go koba ditlou mo masimong? Have you heard anything about using chili peppers to keep elephants away from crops? O utlule gotweng? O ne o utlwa mo go mang? What did you hear? Who did you hear it from? 16. A o lemile ngwaga ono? Did you plough this year? 17. If they ploughed, ask: A o ne wa lema kana wa dirisa pepere/phirihiri go koba ditlou mo dijalong tsa gago? Did you plant any chili peppers or use chili peppers in other ways to protect your cro ps from elephants? 18. A o ruile dikgomo kana dipudi? Do you own cattle or goats? 19. Fa e le ee, dikgomo di le kae? Dipudi di le kae? If yes, how many cattle? How many goals? 20. A go nale sengwe gape se o ka se buang mabapi le diphologolo tsa n aga? Is there anything else you can think of that you would like to share about wildlife (diphologolo tsa naga)? Thank them, take a photo, and say farewell!
204 APPENDIX E PERCEPTIONS OF WATER AND WILDLIFE: CONCEPT ELICITATION RECORDING FORM Met si / Water (Questions 1 5) Setswana English Q1 Q2 Q3 Q4 Q5 Batho ba motse Local People Bojanala Tourism Bojang le Letlhaka Grasses and Reeds Bolwetse Disease Dijalo Crops Dimela Vegetation Diphologolo tsa Metsi Water Animals Diphologolo tsa Naga Wildlife Ditiro Jobs Ditlatlana/Diteko Baskets Ditlhapi Fish Diurwa Livestock Go apaya Cooking Go nwa Drinking Go tlhatswa le go tlhapa Washing/Bathing Goromente Government Koma Papyrus Leuba Drought Madi Money Makgobokgobo a Okavango Okavango Delta Metsi a a Leswe Dirty Water/Pollution Mogobe Water Pool Monang Mosquito Morwalela Flooding Mosepele Transport Noga Snake Noka River Pompo/Thepe Standpipe Pula Rain Rusi Rust in Water Sediba Bor ehole Tanka ya Metsi Water Tank Tladi le Legadima Thunder & Lightning Tswii Water Lillies
205 Diphologolo tsa Naga / Wildlife (Questions 1 5, 9 10) Setswana English Q1 Q2 Q3 Q4 Q5 Q9 Q10 Batho ba mo tse Local People Bogodu jwa Diphologolo Poaching Bojanala Tourism Dijalo Crops Dimela Vegetation Diphologolo tsa Metsi Water Animals Diruiwa Livestock Ditiro Jobs Go Tsomela Mekgabisa (Trophy) Hunting Goromente Government Lephata la Diphologolo DWNP Madi Money Makgobokgobo Trust OCT Makgobokgobo a Okavango Okavang o Delta Matlalo a Diphologolo Skins/Hides Mogobe Water Pool Mogwei wa Safari/di Kampa Safari Operator Nama Meat Nare Buffalo Noka River Phirihiri/Peper e Chili Pepper Tau Lion Tlou Elephant Note: The a dditional rows at the bottom of the recording form were included to capture any topics/concepts mentioned by res pondents for which concept cards had not been created.
206 LIST OF REFERENCES Abbot, J. I. O., D. H. L. Thomas, A. A. Gardne r S. E. Neba, and M. W. Khen. 2001. Understanding the links between conservation and development in the Bamenda Highlands, Cameroon. W orld Development 29 (7):1115 1136. Abel, N., H. Ross, and P. Walker. 1998. Mental models in rangeland research, communication and management. Rangeland Journal 20: 77 91. Adams, W., and D. Hulme. 2001. Conservation and community: changing narratives, policie s and practices in African conservation. Pages 9 23 in Hulme, D. and M. Murphree, editors. African wildlife and livelihoods: the promise and performance of community conservation. Heinemann, Portsmouth, New Hampshire, USA. Agrawal, A. 2000. Small is beauti ful, but is larger better? Forest management institutions in the Kumaon Himalaya, India. Pages 57 85 in Gibson, C. C., M. A. McKean, and E. Ostrom, E. editors. People and forests: communities, institutions, and g overnance MIT Press, Cambridge, Mass achuse tts, USA Agrawal, A., and C. C. Gibson. 1999. Enchantment and disenchantment: the role of community in natural resource conservation. World Development 27 (4):629 649. reso urces in rural South Africa. Development Southern Africa 16: 375 401. Andrews K. M., and B. L. Delahay 2000 Influences on knowledge processes in organizational learning: the psychosocial filter. Journal of Management Studies 37 :797 810. Arentze, T.A., B. G. C. Dellaert, and H. J. P. Timmermans. 2008. Modeling and measuring i mental representations of complex spatio temporal decision p roblems. Environment and Behavior 40 :843 869. Armitage, D. 2005. Adaptive capacity and community based natural r esource management. Environmental Management 35 (6) : 703 715. Arntzen, J. W. 2003. Main findings of the review of CBNRM in Botswana. IUCN/SNV CBNRM Support Programme, Gaborone, Botswana. Bandura, A. 1963. Social learning and personality development. Holt, Ri nehart, and Winston, New York, New York, USA. Bandura, A. 1977. Social l earning t heory. Prentice Hall, Englewood Cliffs, New Jersey, USA.
207 Barrouillet, P., and N. Grosset. 2007. Memory retrieval and content effects in conditional reasoning: A developmental models account. Pages 1 25 in S c ha e ken W., A. Vandierendonck, W. Schroyens, and G. D'Ydewalle, editors. The mental models theory of reasoning: r efinements a nd e xtensions. Lawrence Erlbaum Associates, Mahway, New Jersey, USA. Becker, C. D., and K. Ghimire 2003. Synergy between traditional ecological knowledge and conservation science supports forest preservation in Ecuador. Conservation Ecology 8 (1): 1 [online] URL: http://www.consecol.org/vol8/iss1/ art1/ Belaire, J. A., A. K. Dribin, D. P. Johnston, D. J. Lynch, and E. S. Minor. 2011. Mapping stewardship networks in urban ecosystems. Conservation Letters 00: 1 10. Bengston, D. N. 2004. Listening to neglected voices: American Indian perspectives on na tural resource management. Journal of Forestry 102 (1):48 52. Beratan, K. K. 2007. A cognition based view of decision processes in complex social ecological systems. Ecology and Society 12 (1): 27. [online] URL: http://www.ecologyandsociety.org/vol12/iss1/art27/ Berkes, P., J. Colding, and C. Folke. 2000. Rediscovery of traditional ecological knowledge as adaptive management. Ecological Applications 10 (5):1251 1262. Berkes, P., J. Colding, and C. Folke, editors. 2003. Navigating social ecological systems: building resilience for complexity and change Cambridge University Press, Cambridge, UK. Berkes, P., and Turner. 2006. Knowledge, learning and the evolution of conservation practice for socia l ecological system resilience. Human Ecology 34 (4):479 494. Bernard, H. R. 2006. Research methods in anthropology: qualitative and quantitative a pproaches f ourth edition. AltaMira Press, Lanham, Maryland, USA. Bernard, H. R., E. G. Johnsen, P. D. Killwor th, C. McCarty, G. A. Shelly, and S. Robinson. 1990. Comparing four different methods for measuring personal social networks. Social Networks 12: 179 215. Biggs, H., D. du Toit, M. Etienne, N. Jones, A. Leitch, T. Lynam, S. Pollard, and S. Stone Jovicich. 2 008 A preliminary exploration of two approaches for documenting 'mental models' held by stakeholders in the Crocodile Catchment, South Africa. Water Research Commission, South Africa. Bodin, ., B. Crona, and H. Ernstson. 2006. Social networks in natural resource management: what is there to learn from a structural perspective? Ecology and Society 11 (2):r2. [online] URL: http://www.ecologyandsociety.org/vol11/iss2/resp2/ Bolaane, M. 2004. The impact of game reserve policy on the River BaSarwwa / Bushmen o f Botswana. Social Policy and Administration 38 (4):399 417.
208 Borgatti, S. P. 1996. AnthroPac 4.98 Analytic Technologies, Harvard, Massachusetts, USA. Borgatti, S. P. 2002. NetD raw network visualization. Analytic Technologies, Harvard, Massachusetts, USA. B orgatti, S. P., M. G. Everett, and L. C. Freeman. 2002. Ucinet 6 for Windows: software for social n etwo rk a nalysis (version 6.357). Analytic Technologies, Harvard, Massachusetts, USA. Borgatti, S. P., A. Mehra, D. J. Brass, and G. Labianca. 2009. Network a nalysis in the social sciences. Science 323 (5916):892 895. Bosreup E. 1970 Allen & Unwin, London UK. Botswana Department of Tourism, Ministry of Commerce and Industry, Republic of Botswana. 2000. Botswana Tourism Mas ter Plan: Final Report. Botswana Tourism Development Programme (Fo undation Phase) Project No. 7 ACP. BT. 4/No. 6 ACP BT 44. [online] URL: http://www.botswanatourism.co.bw/asse sts/tourism_master_plan.pdf Bouwen, R., and T. Taillieu. 2004. Multi party collaboration as social learning for interdependence: developing relational knowing for sustainable natural resource management. Journal of Community and Applied Social Psychology 14 :137 153. Brandon, K. E., and M. Wells. 1992. Planning for people and parks: Design dilemmas. World Development 20 (4):557 570. Brewer D. D., S. B. Garrett, and G. Rinaldi. 2002. Free listed items are effective cues for eliciting additional items in sem antic domains. Applied Cognitive Psychology 16 :343 358. Broscius, J. P. 2005. Common ground between anthropology and conservation biology. Conservation Biology 20 (3):683 685. Brosius, J. P., A. Tsing, and C. Zerner. 2005. Communities and conservation: hist ories and politics of community based natural resource management. Altamira Press, Lanham, Maryland. Brundtland, G. H. 1987 Our common f uture: w orld c ommission on e nvironment and d evelopment. Oxford University Press, Oxford, UK. Burt, R. S. 1977. Position s in multiple network systems, Part one: A general conception of stratification and prestige in a system of actors case as a social topology. Social Forces 36 (1):106 131. Burt, R. S. 1992. Structural holes: the social structure of competition. Harvard Univ ersity Press, Cambridge, Massachusetts, USA and London, UK.
209 Burt R. S. 2004 Structural holes and good ideas. American Journal of Sociology 110 (2):349 399. Byrne R. W. 1996 Machiavellian intelligence. Evolutionary Anthropology: Issues, News, and Reviews 5 (5):172 180. Carley K. 1991 A theory of group stability. American Sociological Review 56 (3):331 354. Carlsson, L., and F. Berkes. 2005. Comanagement: concepts and methodological implications. Journal of Environmental Management 75 :65 76. Carpenter, S. B. Walker, J. M. Anderies, and N. Abel. 2001. From metaphor to measurement: r esilience of what to what? Ecosystems 4 (8):765 781. Carpenter S. R., and W. A. Brock. 2008. Adaptive capacity and traps. Ecology and Society 13 (2): 40. [online] URL: http://www.ecologyandsociety.org/vol13/iss2/art40/ Cassidy, L. 2001. CBNRM Support Programme Occasional Paper No. 5. IUCN / SNV CBNRM Support Prog ramme, Gaborone, Botswana. Cassidy L., K. Good, I. Mazonde, and R. Rivers 2001 Regional assessment of the status of the San in southern Africa: an a ssessment of the status of the San in Botswana. Legal Assistance Center, Windhoek, Namibia. Caulkins, D., and S. B. Hyatt. 1999. Using consensus analysis to measure cultural diversity in organizations and social movements. Field Methods 11: 5 26. Chebanne, A. 2002. Minority languages and minority peoples: issues on linguistic, cultural and ethnic death in Bot swana. Pages 47 55 in Mazonde, I. N., editor. Minorities in the millennium: perspectives from Botswana. University of Botswana, Gaborone, Botswana. Child, B. 2009. Community conservation in southern Africa: rights based natural resource management. Pages 1 87 200 in Suich, H., and B. Child, editors. Evolution and innovation in wildlife conservation: parks and game ranches to transfrontier conservation areas. Earthscan, London, UK, and Sterling, Virginia, USA. Charnley, S., A. P. Fischer, and E. T. Jones. 200 7. Integrating traditional and local ecological knowledge into forest biodiversity conservation in the Pacific Northwest. Forest Ecology and Management 246: 14 28. CIA (Central Intelligence Agency). 2011. The CIA factbook: Botswana. [online] URL: https://www.cia.gov/library/publications/the world factbook/geos/bc.html
210 Coffe, H., and B. Geys. 2008. Measuring the bridging nature of voluntary o rganizations: t he i mportance of a ssociation s ize. Sociology 42 (2):357 369. Cohen, S. 2008. When a push comes to hush: promoting mainstream views and silencing alternatives through conservation narratives. Current Conservation 3 (3):14 17. Colding, J. 1998. Analysis of hunting options by the use of general food taboos. Ecological Modelling 110 (1):5 17. Cornwall, A. 2003. Whose voices? Whose choices? Reflections on gender and participatory development. World Development 31 ( 8 ) :1325 1342. Craik, K. 1943. The nature of explanation. Cambridg e University Press, Cambridge, UK. Crona, B., and Bodin. 2006. What you know is who you know? Communication patterns among resource users as a prerequisite for co management. Ecology and Society 11 (2): 7. [online] URL: http://www.ecologyandsociety.org/vol11/iss2/art7/ CSO ( Central Statistics Office ) Republic of Botswana. 2001a. 2001 Population and housing census: g uide to the villages of Botswana Government Printer, Gaborone, Botswana. C SO ( Central Statistics Office ) Republic of Botswana. 2001b. 2001 Population census atlas: Botswana. Department of Printing and Publishing Services, Gaborone, Botswana. [online] URL: http://www.cso.gov.bw/images/stories/Census_Publication/pop%20atlas.pdf C SO (Central Statistics Office), Republic of Botswana. 2006. Botswana Demographic Survey 2006. Department of Printing and Publishing Services, Gaborone, Botswana. [online] URL: http://www.cso.gov.bw/images/stories/HouseHold/2006_bdsrprt.pdf CSO ( Central St atistics Office ) Republic of Botswana. 2008a. Distribution of population by sex by villages and their associated localities. [online] URL: http:// www.cso.gov.bw/index2.php?option=com_content&task=view&id=159&pop= 1&page=0&Itemid=106 CSO ( Central Statistics Office ) Republic of Botswana. 2008b. Table 1.2. Population and household distribution by administrative district (1991 and 2001). [online] URL: http://www.cso.gov.bw/index.php?option=com_content&task=view&id=143&Itemid =90 CSO ( Central Statistics Office ) Republic of Botswana. 2008c. Distribution of popula tion by sex by village and their associated localities: 2001 population. [online] URL: http://www.cso.gov.bw/index2.php?option=com_content&task=view &id=159&pop= 1&page=0&Itemid=106
211 CSO ( Central Statistics Office ) Republic of Botswana. 2010 Population and housing census 2011: the first census question. Central Statistics Office, Gaborone, Botswana. [online] URL: http://www.cso.gov.bw/images/stories/PHC_2011/1stcensus_question.pdf Davidson Hunt, I. J. 2006. Adaptive learning networks: developing resource management knowledge through social learning forums. Human Ecolo gy 34 (4):593 614. Davison J. 1996. Gender, lineage, and ethnicity in southern Africa. Westview Press, Boulder, Colorado, USA. Denbow J., and P. C. Thebe 2006 Culture and customs of Botswana. Greenwood Press, Westport, Connecticut USA Denzau, A. T. an d D. C. North. 1 994 Shared Mental Models: Ideologies and Institutions. Kyklos 47 (1):3 31. Dietz, T., E. Ostrom, and P. C. Stern. 2003. The struggle to govern the commons. Science 302 (5652):1907 1912. Doak, D. F., J. A. Estes, B. S. Halpern., U. Jacob, D. R. Lindberg, J. Loworn, D. H. Monson, M. T. Tinker, T. M. Williams, J. T. Wootton, I. Carroll, M. Emmerson, F. Michell, and M. Novak. 2008. Understanding and predicting ecological dynamics: are major surprises inevitable? Ecology 89 (4):952 961. Downs, R., and D. Stea. 1977. Maps in m inds. Harper and Row, New York, New York, USA. Dretske, F. I. 1981. Knowledge & the flow of i nformation. MIT Press, Cambridge, Massachusetts, USA. Dunbar R. I. M. 1998 The social brain hypothesis. Evolutionary Anthropology 6 ( 5):178 190. Dunbar R. I. M. 2008 Cognitive constraints on the structure and dynamics of social networks. Group Dynamics Theory Research and Practice 12 (1):7 16. Dunbar, R. I. M. 2010. and other evolu tionary quirks. Harvard University Press, Cambridge, Massachusetts, USA. Fabricius, C., C. Folke, G. Cundill, and L. Schultz. 2007. Powerless spectators, coping actors, and adaptive co managers: a synthesis of the role of communities in ecosystem managemen t. Ecology and Society 12 (1): 29. [online] URL: http://www.ecologyandsociety.org/vol12/iss1/art29/
212 Fernandez Gimenez, M. E., H. L. Ballard, and V. E. Sturtevant. 2008. Adaptive management and social learning in collaborative and community based monitoring: a study of five community based forestry organizations in the western USA. Ecology and Society 13 (2): 4. [online] URL: http://www.ecologyandsociety.org/vol13/iss2/art4/ Fisher, R. E., and R. E. Geiselman. 1992. Memory enhancing techniques for investigative interviewing : t he cognitive interview. Thomas, Springfield, Illinois, USA. Folke, C., J. Colding, and F. Berkes. 2003. Synthesis: Building resilience and adaptive capacity in social ecolo gical systems. Pages 352 287 in Berkes, F., J. Colding, and C. Folke, editors. Navigating social ecological s ystems : b uilding resilience for complexity and change. Cambridge University Press, Cambridge, UK and New York, New York, USA. Folke, C., T. Hahn, P Olsson, and J. Norberg. 2005. Adaptive governance of social ecological systems. Annual Review of Environmental Resources 30: 441 473. Fosnot, C. T., editor. 1996. Constructivism: theory, perspectives, and practice. Teachers College Press, New York, New Yo rk, USA. Fraser, D. J., T. Coon, M. R. Prince, R. Dion, and L. Bernatchez. 2006. Integrating traditional and evolutionary knowledge in biodiversity conservation: a population level case study. Ecology and Society 11 (2): 4. [online] URL: http://www.ecologyandsociety.org/vol11/iss2/art4/ Freeman, Linton. 1977. A set of measures of centrality based upon betweenness. Sociometry 40 :35 41. Gaotlhobogwe, M. 2011. Mother tongue: another opportunity goes begging. Mmegi online 28 (109). July 27, 2011. [online] URL: http://www.mmegi.bw/index.php?sid=1&aid=542&dir=2011/July/Wednesday27 Ghimire, S. K., D. McKey, and Y. A umeeruddy Thomas. 2004. Heterogeneity in ethnoecological knowledge and management of medicinal plants in the Himalayas of Nepal: implications for conservation. Ecology and Society 9 (3): 8. [online] URL: http://www.ecologyandsociety.org/vol9/iss3/art6/ Giannini, A., M. Biasutti, I. M. Held, and A. H. Sobel. 2008. A global perspective on African climate. Climatic Change 90 :359 383. Glaser, B. G., and A. Strauss. 1967. The discovery of ground e d theory: strategies ofr qualitative research. Aldine, New York, New York, USA. G ranovetter, M. S. 1973 The s trength of w eak t ies. The American Journal of Sociology 78 (6):1360 1380.
213 Granovette r, M. S. 1983 The s trength of w eak ties: a n etwork t heory r evisi ted. Sociological Theory 1 :201 233. Greener, R., K. Jefferis, H. Siphambe. 2000. The impact of HIV/AIDS on poverty and inequality in Botswana. South African Journal of Economics 68 (5):393 404. Gunderson, L. H. 2000. Ecological resilience in theory and appl ication. Annual Review of Ecological Systems 31 : 425 439. Gunderson, L. H. 2003. Adaptive dancing: i nteractions between social resilience and ecological crises. Pages 33 54 i n Berkes, F., J. Colding, and C. Folke, editors. Navigating social e cological syste ms: b uilding resilience for complexity and change. Cambridge University Press, Cambridge, UK and New York, NY, USA. Gunderson, L. H., and C. S. Holling. 2002. Panarchy: understanding transformations in human and natural systems. Island Press, Washington, D .C., USA. Halford, G. S. 1993. Children's understanding: the development of mental m odels. Lawrence Erlbaum Associates, Hillsdale, New Jersey, USA. Heckathorn, D. D. 1997. Respondent driven sampling: a new approach to the study of hidden populations. Soci al Problems 44: 174 199. Hickey S., and G. Mohan. 2004 Towards participation as transformation: critical themes and challenges. Pages 3 24 in Hickey, S. and G. Mohan, editors. Participation: from tyranny to transformation? Zed Books, London, UK. Holling, C. S., editor. 1978 Adaptive environmental assessment and management. Wiley and International Institute for Applied Systems Analysis, Chichester, New York, USA. Holling, C. S., and G. K. Meffe, G. K. 1996. Command and Control and the Pathology of Natural Resource Management. Conservation Biology 10 (2): 328 337. Hunn, E. S., D. R., Johnson, P. N. Russel, and T. F. Thornton. 2003. Huna Tinglit traditional environmental knowledge, conservation, and the management of a 'wilderness' park. Current Anthropology 4 4 (Supplement): S79 S103 Hunter, L. M., S. Strife, and W. Twine. 2010. Environmental perceptions of rural South African residents: the complex nature of environmental concerns. Society and Natural Resources 23 (6):525 541. IIED (International Institute for E nvironment and Development) 1994. Whose Eden? An Overview of Community Approaches to Wildlife Management. IIED, London, UK.
214 IPC (International Poverty Centre) and Botswana Institute for Development Policy Analysis. 2005. Poverty status report for Botswana : incidence, trends, and dynamics. United National Development Programme International Poverty Centre Research Project Report. [online] URL: http://www.ipc undp.org/publications/repor ts/Botswana.pdf IPCC (Intergovernmental Panel on Climate Change). 2007. Climate Change 2007: Synthesis Report. An Assessment of the Intergovernmental Panel on Climate Change. Abdelkader, A., R. Bojariu, S. Diaz, I. Elgizouli, I., D. Griggs, D. Hawkins, O Hohmeyer B. P. Jallow, L. Kajfez Bogataj, N. Leary, H. Lee, and D. Wratt, editors [online] URL: http://www.ipcc.ch/pdf/assessment report/ar4/syr/ar4_syr.pdf IUCN/UNEP/WWF. 19 91. Caring for the earth: a strategy for sustainable l iving. Gland, Switzerland. Jacobson S. 1999 Communication skills for conservation professionals. Island Press, Washington, D.C., USA. Johannes, R. E. 1981. Word of the lagoon: fishing and marine lore in the Palau district of Micronesia University of California Press, London, UK. Johnson Laird, P. N. 1983. Mental models: t owards a cognitive science of language, inference, and c onsciousness. Harvard University Press, Cambridge, Massachusetts, USA. Johns on Laird, P. N. and R. M. J. Byrne. 1991 Deduction. Lawrence Erlbaum Associates, Hillsdale, New Jersey, USA. Johnson Laird, P. N. and E. Goldvard Steingold 2007 Models of cause and effect. Pages 1 25 i n Schaeken, W., A. Vandierendonck, W. Sc hroyens, an d G. D'Ydewalle, editors. The m ental m odels t heory of reasoning: refinements and e xtensions. Lawrence Erlbaum Associates, Mahway, New Jersey, USA. Johnson Laird, P. N., P. Legrenzi, P. Girotto, M. S. Legrenzi, and J. P. Caverni 1999. Nave probability: a mental model theory of extensional reasoning. Psychological Review 106 :62 88. Jonassen, D. H. 1991. Objectivism versus constructivism: d o we need a new philos o phical paradigm? Educational Technology Research and Development 39 (3):5 14. Jones, N. A., H. Ros s, T. Lynam, P. Perez, and A. Leitch. 2011. Mental models: an interdisciplinary synthesis of theory and methods. Ecology and Society 16(1): 46. [online] URL: http://www.ecologyandsociety.org /vol16/iss1/art46/ Kabeer N. 1994 Reversed realities: gender hierarchies in development thought. Verso, London, UK and New York, New York, USA.
215 Kaplan, S. 1973. Cognitive maps in perception and thought. Pages 63 78 in Downs, R. M., and D. Stea, editors. Image and environment. Aldine, Chicago, Illinois, USA. Kaplan S., and R. Kaplan 1982 Cognition and environment: f unctioning in an uncertain w orld. Praeger Publishers, New York, New York, USA. Kaplan, S. 2000. Human nature and environmentally responsib le behavior. Journal of Social Issues 56 (3): 491 508 Kaplan, S., and R. Kaplan. 2009. Creating a larger role for environmental psychology: the Reasonable Person Model as an integrative framework. Journal of Environmental Psychology 29: 329 339. Kearney, A. R., G. Bradley, R. Kaplan, and S. Kaplan. 1999. Stakeholder p erspectives o n a ppropriate forest m ana gement in the P acific Northwest. Forest Science 5 (1): 62 73. Kearney, A. R., and S. Kaplan. 1997. Toward a methodology for the measurement of k nowledge struct ures of ordinary people: t he Conceptual Content Cognitive Map (3CM). Environment and Behavior 29 (5): 579 617. Keen M., V. A. Brown, and R. Dyball 2005 Social learning: A new approach to environmental management. Pages 3 21 i n M. Keen, V. A. Brown and R. Dyball, editors. Social learning in environmental management: Building a sustainable future. Earthscan, London, UK and Sterling, Virginia, USA. Kellert, S. R., J. N. Mehta, S. A. Ebbin, and L. L. Lichtenfeld. 2000. Community natural r esource management: p romise, rhetoric, and r eality. Society and Natural Resources 13 (8): 705 715. Kelly K., and M. van der Riet 2001 Participatory research in community settings: Processes, methods and challenges. Pages 159 188 in Seedat, M., N. Duncan, and S. Lazarus, edito rs. Community psychology: t heory, method and practice ; South African and other perspectives Oxford University Press, Cape Town South Africa. King, A. 1995. Avoiding ecological surprise: Lessons from long standing communities. Academy of Management Review 20 (4): 961 985. Klimoski, R. and S. Mohammed. 1994. Team mental model: construct or m etaphor? Journal of Management 20 (2):403 437. Kogovsek, T., and A. Ferligoj. 2005. Effects on reliability and validity of egocentered network measurements. Social Network s 27 (3): 205 229. Kolb, D. A. 1984. Experiential learning: experience as the source of learning and development Prentice Hall, Englewood Cliffs, New Jersey, USA.
216 Larson T. J. 1992 The Bayei of Ngamiland. The Botswana Society, Gaborone, Botswana. Leach, M ., R. Mearns, and I. Scoones. 1999. Environmental entitlements: dynamics and institutions in community based natural resource management. World Development 27 (2):225 247. Lee, K. N. 1993. Compass and gyroscope: integrating science and politics for the envi ronment. Island Press, Washington, D. C., USA. Levin D. Z., and R. Cross 2004 The s trength of w eak t ies y ou c an t rust: t he m ediating r ole of t rust in e ffective k nowledge t ransfer. Management Science 50 (11):1477 1490. Malcolm, J. R., A. Markham R. P. Ne ilson, and M. Garaci 2002 Estimated migration rates under scenarios of global climate change. Journal of Biogeography 29 : 835 849. Mandondo, A. 2000 Resource Governance Systems in History. Occasional Paper No. 31. (CIFOR) Cen ter for International Forestry Research Bogor, Indonesia March, C., I. Smyth, and M. Makhopadhyay, editors. 1999 A guide to gender analysis frameworks. Oxfam, Oxford, UK. Mayoux, L. 1995 Beyond naivety: women, gender inequality and participatory develo pment. Development and Change 26: 235 258. Mazonde I. N. 2002 The San in Botswana and the issue of subjectivities national disintegration or cultural diversity? Pages 57 71 in Mazonde, I. N., editor. Minorities in the millennium: perspectives from Botswan a. University of Botswana Gaborone, Botswana. Mbaiwa J. E. 1999 Prospects for sustainable wildlife resource utilization and management in Botswana: a case study of East Ngamiland District. M.Sc. Thesis. University of Botswana, Gaborone, Botswana. Mbaiwa J.E. 2004 The success and sustainability of community based natural resource management in the Okavango Delta, Botswana. South African Geographical Journal 86 (1) :44 53. Mbaiwa, J. E. 2005 Wildlife resource utilization at Moremi Game Reserve and Khwai c ommunity area in the Okavango Delta, Botswana. Journal of Environmental Management 77: 144 156. Mbaiwa J. E. 2008 Tourism development, rural livelihoods, and conservation in the Okavango Delta, Botswana. Dissertation. Texas A&M University, College Station Texas, USA.
217 Mbaiwa, J. E. 2011. The effects of tourism development on the sustainable utilisation of natural resources in the Okavango Delta, Botswana. Cultural Issues in Tourism 14 (3):251 273. Mbaiwa J. E., B. N. Ngwenya, and D. L. Kgathi 2008 Conte nding with unequal and privileged access to natural resources and land in the Okavango Delta, Botswana. Singapore Journal of Tropical Geography 29: 155 172. Mbaiwa, J. E., A. Stronza and U. Kreuter. 2011. From c ollaboration to conservation: insights f r om t he Okavango Delta, Botswana. Society and Natural Resources 24 ( 4 ): 400 411 McCarty C. 2011 EgoNet software for social network analysis (version 3 31 11). University of Florida, Gainesville, Florida, USA. McCarty, C., P. D. Killworth, and J. Rennell, J. 2007 Impact of methods for reducing respondent burden on personal network structural measures. Social Networks 29 (2): 300 315. McGray H., A. Hammill, and R. Bradley. 2007 Weathering the storm: options for framing adaptation and development. World Resourc es Institute, Washington, D.C., USA. McPherson, M., L. Smith Lovin, and J. M. Cook. 2001. Birds of a feather: homophily in social networks. A nnual Review of Sociology 27 :415 44. McShane, T. O. and M. P. Wells 2004 Integrated conservation and development ? Pages 3 9 i n McShane, T. O. and M. P. Wells, editors. Getting b iodiversity p rojects to w ork. Columbia University Press, New York, New York, USA. Melkote S. R., and H. L. Steeves 2001 Communication for development in the third world: theory and practi ce for empowerment Second edition. Sage Publications, New Delhi, India, Thousand Oaks, California, USA, and London, UK. Millar J., and A. Curtis 1999 Challenging the boundaries of local and scientific knowledge in Australia: opportunities for social l earning in managing temperate upland pastures. Agriculture and Human Values 16 :389 399. Mohan G., and K. Stokke 2000 Participatory development and empowerment: the dangers of localism. Third World Quarterly 21 (2):247 268. Moller, H., F. Berkes, P. O'Bri an Lyver, and M. Kislalioglu. 2004 Combining science and traditional ecological knowledge: monitoring populations for co management. Ecology and Society 9 (3): 2 [online] URL: http://www.ecolo gyandsociety.org/vol9/iss3/art2
218 Mompati, T., and G. Prinsen. 2002 Ethnicity and participatory development methods in Botswana: some participating are to be seen and not heard. Pages 92 109 in Eade, D., editor. Development and Culture. Oxfam, Dorset, UK. Mooko T. 2006 Counteracting the threat of language death: the case of minority languages in Botswana. Journal of Multilingual and Muticultural Developent (27) 2:109 125. Moreno, J. L. 1934 Who shall survive? Nervous and Mental Disease Publishing Company, Washington, D .C. USA Morgan, M. G., B. Fischhoff, A. Bostrom, and C. J. Atman. 2002. Risk communication: a mental models approach. Cambridge University Press, Cambridge, UK. Moseki, T. 2009. Study within Botswana dis tricts. Mmegi Online 26 (163). October 30, 2009. [online] URL: http://mmegi.bw/index.php?sid=4&aid=123&dir=2009/October/Friday30 Moser C. 1989 Gender planning in the third w orld: meeting practical and strategic gender needs. World Development 17 (11):1799 1825. Mulder M. B., and P. Cop p olillo 2005 Conservation: linking ecology, economics, and culture. Princeton University Press, Princeton, New Jersey, USA and Oxford, UK. Mu rray Hudson, M., P. Wolski, and S. Ringrose. 2006 Scenarios of the impact of local and upstream changes in climate and water use on hydro ecology in the Okavango Delta, Botswana. Journal of Hydrology 331: 73 84. Musumali, M. M., T. S. Larsen, and B. P. Kal tenborn. 2007. An impasse in community based natural resource management implementation: the case of Zambia and Botswana. Oryx 41 (03):306 313. Nelson N., and S. Wright editors. 1995 Power and participatory development: theory and practice. ITDG Publishi ng, London, UK. Newig, J., D. Gnther, and C. Pahl Wostl. 2010 Synapses in the network: learning in governance networks in the context of environmental management. Ecology and Society 15 (4): 24. [online] URL: http://www.ecologyandsociety.org/vol15/iss4/ar t24/ Newman, L. L., and A. Dale. 2005. Network structure, diversity, and proactive resilience building: a response to Tompkins and Adger. Ecology and Society 10 (1): r2. [online] URL: http:// www.ecologyandsociety.org/vol10/iss1/resp2/ Newmark, W. D., and J. L. Hough. 2000. Conserving Wildlife in Africa: Integrated Conservation and Development Programs and Beyond. Bioscience 50 (7):585 592.
219 Norman, D. A. 1983. Some observations on mental models Pages 7 14 in Gentner, D., and A. L. Stevens, editors. Mental m odels. Lawrence Erlbaum Associates, Hillsdale, New Jersey, USA. Novak, J. D., and D. B. Gowan. 1984. Learning how to learn. Cambridge University Press, Cambridge, UK. Nyati Ramahobo, L. 2002 Ethnic identity and nationhood in Botswana. Pages 17 27 in Mazonde, I. N., editor. Minorities in the millennium: perspectives from Botswana. University of Botswana, Gaborone, Botswana. Nygren, A. 2005 Community based forest management within the context of institutional decentralization in Honduras. World Development 33 (4):639 655. Olsson, P., C. Folke, and T. Hahn. 2004 Social e cological transformation for ecosystem m anagement: the development of adaptive co management of a wetland landscape in s outher n Sweden. Ecology and Society 9 (4): 2 27. Ozesmi, U. and S. L. Ozesmi. 2004. Ecological a multi step fuzzy cognition mapping approach. Ecological Modeling 176: 43 64. Pahl Wostl, C., M. Craps, A. Dewulf, E. Mostert, D. T abara, and T. Taillieu. 2007 Social learning and water resources management. Ecology and Society 12 (2): 5. [online] URL: http://www.ecologyandsociety.org/vol12/iss2/art5/ Phuthego, T. C., a nd R. Chanda. 2004 Traditional ecological knowledge and community based natural resource management: lessons from a Botswana wildlife management area. Applied Geography 24 (1): 57 76. Piaget, J. 1953 The origins of intelligence in children. Routledge and K egan Paul, London, UK. Plate, R. 2006. Assessing the effectiveness of systems oriented instruction for preparing students to understand complexity. Doctoral Dissertation. University of Florida Gainesville, Florida, USA. [online] URL: http://purl.fcla.edu/ fcla/etd/UFE0017106 Pu Journal of Democracy 6 (1):65 78. Rahm, D., L. Swatuk, and E. Matheny. 2006. Water resource management in Botswana: balancing sustainability and economic develop ment. Environment, Development and Sustainability 8 : 157 183 Rahmena M. 1992 Participation. Pages 116 131 in Sachs, W., editor. The development d ictionary Zed Books, London, UK.
220 Ramirez Sanchez, S., and E. Pinkerton. 2009 The impact of resource scarci ty on sharing networks in Loreto, BCS, Mexico. Ecology and Society 14 (1): 22. [online] URL: http://www.ecologyandsociety .org/vol14/iss1/art22/ R amsar Convention on Wetlands. 2000. The annotated Ramsar list of wetlands of international importance: Botswana. [online] URL: http://www.ramsar.org/cda/en/ramsar pubs annolist anno botswana/main/ramsar/1 30 168%5E16701_4000_0__ Rantao P. M. 2006 Setswana culture and tradition. Pentagon Publishers, Gaborone, Botswana. Reagans, R., and B. McEvily. 2003 Network stru cture and knowledge transfer: the effects of cohesion and range. Administrative Science Quarterly 48 (2):240 267. Redford, K. H. 2011. Misreading the conservation landscape. Oryx 45 (3):324 330. Reed, M. S., A. C. Evely, G. Cundill, I. Fazey, J. Glass, A. La ing, J. Newig, B. Parrish, C. Prell, C. Raymond, and L. C. Stringer. 2010 What is social learning? Ecology and Society 15 (4): r1. [online] URL: http://www.ecologyandsociety.org /vol15/iss4/resp1/ Reid, A., B.B. Jensen, J. Nikel, and V. Simovska, editors. 2 008. Participation and learning: perspectives on education and the environment, health, and sustainability. Springer, New York, New York, USA. Republic of Botswana, UNDP, and UNFPA. 1998 National Gender Programme Framework, TA Publications Botswana, Gabor one, Botswana. Rickheit, G. and L. Sichelschmidt. 1999. Mental models: Some answers, some questions, some suggestions. Pages 9 40 in Rickheit, G. and L. Sichelschmidt, editors. Mental models in discourse processing and reasoning. North Holland, Amsterdam, The Netherlands. Rist, L., R. U. Shaanker, E. J. Milner Gulland, and J. Ghazoul. 2010. The use of traditional ecological knowledge in forest management: an example from India. Ecology and Society 15 (1): 3. [online] URL: http://www.ecologyandsociety.org/vol15/iss1/art3/ Ritchie Dunham, J. L., and L. M. Puente. 2008 Strategic Clarity: Actions for Identifying and Correcting Gaps in Mental Models. Long Range Planning 41 (5):509 529. Rocheleau, D., B. Thomas Slayter, and E. Wangari, editors. 1996 Feminist political ecology: global issues and local experiences. Routledge, London, UK. Rogers E. M. 1995 Diffusion of innovations Fourth edition. Free Press, New York, New York, USA.
221 Romney, A. K., S. C. Weller, and W. H. Batchelder. 1986. Culture as consensus: a theory of culture and informant accuracy. American Anthropologist 88: 164 177. Rowe, A. L., and N. J. Cooke. 1995. Measuring mental models: Choosing the right tools for the job. Human Resourc e Development Quarterly 6 (3) : 243 255. Rozemeijer, N. 2009 CBNRM in Botswana. Pages 243 256 in Suich, H., and B. Child, editors. Evolution and innovation in wildlife conservation: parks and game ranches to transfrontier conservation areas. Earthscan, Londo n, UK, and Sterling, Virginia, USA. Ryan, G. W., J. M. Nolan, and P. S. Yoder. 2000. Successive free listing: using multiple free lists to generate explanatory models. Field Methods 12 (2):83 107. Schermerhorn, R. 1996 Ethnicity and Minority Groups. Page 1 7 in Hutchinson J., and A D. Smith editors. Ethnicity. Oxford University Pres s, Oxford, UK. Schmink, M. 1999 Conceptual framework for gender and community based c onservation MERGE (Managaing Ecosystems and Resources with Gender Emphasis) Case Study No 1. U niversity of Florida, Gainesville, Florida, USA. [online] URL: http://www.tcd.ufl.edu/merge.shtml Schneider, M., J. Scholz, M. Lubell, D. Mindruta, M. Edwardsen. 2003 Building consensual instituti ons: networks and the National Estuary Program. American Journal of Political Science 47 (1) : 143 158. Schusler T M., D. J. Decker, and M. J. Pfeffer. 2003 Social Learning for Collaborative Natural Resource Management. Society and Natural Resources 16 (4 ):309 326. Schuster, B. 2007. Proceedings of the Fourth National CBNRM Conference; 20th to 23rd of November Potential to Contribute and CBNRM Status Report 2006. IUCN CBNRM Support Programme, Gaborone, Botswana. Scott, J. 2000 Social network analysis: a handbook. Sage Publications, London, UK. Seel, N. M. 2001. Epistemology situated cognition and mental models: 'Like a bridge over troubled waters'. Instructional Science 29 (4 5):403 427. Seligman, M. E P. 1975. Helplessness. Freeman, San Francisco, California, USA. Shackleton, C. M., and S. E. Shackleton. 2004 The importance of non timber forest products in rural livelihood security and as safety nets: a review of evidence from South Africa. South Afr ican Journal of Science 100 (11 and 12):658 664. Shore, B. 1998 Culture in mind: cognition, culture, and the problem of meaning. Oxford University Press, New York, New York, USA and Oxford, UK.
222 Smith, E. R. 1999 New connectionist models of mental represe ntation: Implications for survey research. John Wiley & Sons, New York, New York, USA. Smith, J. J. 1993. Using ANTHROPAC 3.5 and a Spreadsheet to Compute a Free List Salience Index. Cultural Anthropology Methods 5 (3):1 3 Smyth, I. 1999 Introduction. Pag es 8 16 in March, C., I. Smyth, and M. Makhopadhyay, editors. A guide to gender analysis frameworks. Oxfam, Oxford, UK. Songorwa, A. N. 1999. Is community based wildlife management gender sensitive? Experiences from Selous Conservation Programme in Tanzani a. Uongozi Journal of Management Development 11 (2):145 166. Spradley, J. S. 1979. The ethnographic interview. Harcourt Brace Jovanovich College Publishers, Orlando, Florida, USA. Spring A. 1993 Guidelines for evaluating the inclusion of women and gender in development projects. Forum Valutazione No. 5 Rome, Italy. Stevens, P. W., and R. Jansen. 2002 Botswana National Ecotourism Strategy. Government of Botswana, Gaborone, Botswana. [online] URL: http://www.ub.bw/ip/documents/2002_Botswana%20National%20Ecotourism%20 Strategy.pdf Taylor, M. 2007. CBNRM for Whose Benefit? A Case Study of Subsistence Hunting on the Boundaries s Northern Protected Are as. Pages 27 33 in Schuster, B. and O. T. editors. CBNRM Support Programme Occasional Paper No. 15: Natural Resources Management and People. IUCD CBNRM Support Programme, Gaborone, Botswana Thakadu, O. T. and B. Schuster 2007 Introduction Pages 9 10 in Schuster, B., and O. T. Thakadu, editors. CBNRM Support Programme Occasional Paper No. 15: Natural Resources Management and People. IUCD CBNRM Support Programme, Gaborone, Botswana. Tompkins, E., and W. Adger. 2004. Does adaptive management of natural resources enhance resilience to climate change? Ecology and Society 9 (2):10. [online] URL: http://www.ecologyandsociety.org/vol9/iss2/art10/ Tshakert P., and K. A. Dietrich 2 010 Anticipatory learning and climate change adaptation and resilience. Ecology and Society 15 (2): 11. [online] URL: http://www.ecologyandsociety.or/vol15/iss2/art11/ Tversky, A., and D. Kahne man. 1974. Judgment under uncertainty: heuristics and biases. Science 185 (4157):1124 1131. Twyman C. 2000 Participatory conservation? Community based natural resource management in Botswana. The Geographic Journal 166 (4):323 335.
223 Valente T. W. 1996 Soc ial network thresholds in the diffusion of innovations Social Networks 18 (1): 69 89. Valente T. W., P. Gallagher, and M. Mouttapa. 2004 Using social networks to understanding and prevent substance use: a transdisciplinary perspective. Substance Use and M isuse 39 (10 12):1685 1712. van der Riet M. 2008 Participatory research and the philosophy of social science: Beyond the moral imperative. Qualitative Inquiry 14 (4):546 565. Walker, B., C. S. Holling, S. R. Carpenter, and A. Kinzig. 2004 Resilience, adap tability and transformability in social ecological systems. Ecology and Society 9 (2): 5. [online] URL: http://www.ecologyandsociety.org/vol9/iss2/art5/ Wals, A. E. J. 2007 Social learning to wards a sustainable world: principles, perspectives, praxis. Wageningen Academic Publishers, Wageningen, The Netherlands. Wasserman, S., and K. Faust. 1994 Social network analysis: methods and applications. Cambridge University Press, New York, New York, USA. Weller, S. C. 2007. Cultural consensus theory: applications and frequently asked questions. Field Methods 19 (4): 339 368. Weller, S. C., and A. K. Romney. 1988. Systematic data collection. Sage, Newbury Park, California, USA. Wellman, B., and S. D. B erkowitz. 1998. Social structures: a network approach Cambridge University Press, Cambridge, UK. Wenger E. 1999 Communities of practice: learning, meaning, and identity. Cambridge University Pres s, Cambridge, UK and New York, New York, USA. Western, D. and R. M. Wright. 1994 The background to community based conservation. Pages 1 12 i n Western, D., R. M. Wright and S. Strum editors. Natural connections: Perspectives in community based conservation Island Press, Washington, D C. USA. Willis, G. B. 1999. Co gnitive interviewing: a how t o g uide. From Reducing Survey Error through Research on the Cognitive and Decision Processes in Surveys. Materials from short course presented at the 1999 Meeting of the American Statistical Association. Research Triang le Institute, Research Triangle Park, NC, USA. Zhou, W. X., D. Sornette, R. A. Hill, and R. I. M. Dunbar. 2005. Disc rete hierarchical organization of social group sizes. Proceedings of the Royal Society of Biology 272 (1561):439 444.
224 BIOGRAPHICAL SKETCH D eborah J. Wojcik is passionate about education and the environment and has worked in environmental education and communication for 15 years. She has worked as a field environmental educator, classroom teacher, program evaluator, workshop facilitator, and p rogram manager in academic, non profit, government, and corporate settings. She has studied, conducted research, and volunteered in the United States, Australia, southern Africa, and the Peruvian Andes. Deborah received her Ph.D. at the University of Flori da from the School of Forest Resources and Conservation with a concentration in Tropical Conservation and Development and Graduate Certificate in Environmental Education and Communications in the fall of 2011 She received Master of Environmental Managemen t, Master of Arts in Teaching, and Bachelor of Scien c e degrees from Duke University.