Synthesis Social networks and the resilience of rural communities in the Global South : a critical review and conceptual reflections

In the last decades, a growing scholarship has outlined the crucial role of social networks as a source of resilience. However, with regard to the Global South, the role of social networks for the resilience of rural communities remains an underresearched and underconceptualized issue, because research remains scattered between different strands and has rarely been integrated from a resilience perspective. To provide common ground for the exchange between disciplines and to identify steps towards a more comprehensive social network perspective on the resilience of rural communities in the Global South, we present a systematic review of contemporary case studies from three strands of research: (i) natural resource management, (ii) agricultural innovation, and (iii) social support. Although studies in each strand have their own particular strengths and weaknesses in addressing aspects of the resilience of rural communities in the Global South, they all share a static view of the outcomes of social networks, tend to emphasize structure over agency, and neglect spatial dimensions of social relations. To address these challenges, we propose a translocal social network perspective on resilience that views rural communities as being embedded in social networks that connect people and facilitate the flow of resources, information, and knowledge between places.


INTRODUCTION
Rural communities can be understood as vulnerable socialecological systems (SES) that need to build resilience to withstand internal and external stresses from social, economic, and political changes (Adger 2000, Wilson et al. 2013).It has been argued that many aspects of adaptive capacity reside in social networks (Adger 2003) and that these are a crucial source of resilience (Folke et al. 2005, Folke 2006, Berkes and Ross 2013).This applies in particular to rural communities in the Global South, where often a lack of access to resources, knowledge, and functioning institutions is a major obstacle to sustainable development (Etzold et al. 2012).However, although investigations into the role of social networks is growing at a fast pace, it remains scattered across different strands of research, with related but separate research agendas (Videras 2013).With this paper, we provide a systematic review of current case studies from three of these strands, addressing different aspects relevant to the resilience of rural communities.By analyzing how case studies conducted between 2000 and 2015 conceptualize and operationalize social networks, we level the ground for the exchange between disciplines.Concluding we identify prospects for more fruitfully employing a social network perspective in investigating the resilience of rural communities in the Global South.
During the last decades, resilience has emerged as a key concept across disciplines for investigating responses to changes in human and ecological systems (Folke et al. 2010), resulting in a variety of ways in which resilience is understood, investigated, and applied (Downes et al. 2013).From a concept originally concerned with the persistence of ecological systems in the context of external disturbances (Holling 1973), resilience has developed through a concept underlining the role of adaptive capacity for navigating coupled SES (Gunderson andHolling 2002, Berkes et al. 2003) to one emphasizing the transformation of SES in the face of global change (Walker et al. 2004, Folke 2006, Folke et al. 2010).Attention has thus widened from the ecological to include also the social dimensions of resilience (Adger 2000, Cote andNightingale 2012).This comprises, for example, human agency, social learning, and the skills and capacities of social actors to cope with, adapt to change, and facilitate transformation (Folke et al. 2010, Moore and Westley 2011, Berkes and Ross 2013, Keck and Sakdapolrak 2013, Skerratt 2013, Cretney 2014, Ifejika Speranza et al. 2014).
Similar to resilience, the concept of social networks has been applied in a wide range of sciences from the social to the physical (Borgatti et al. 2009, Scott 2011).Intermediating between micro and macro levels, the investigation of social networks is expected to provide answers to central challenges pertinent in sustainability science, such as promoting social learning, linking knowledge with action, and enhancing collective action (Henry and Vollan 2014).Social networks have been shown to foster the capacity to buffer, adapt to, and shape change (Moore and Westley 2011) by providing resources needed to cope with external stresses and disturbances (Adger 2003), and fostering humans' ability to initiate social innovations and act collectively (Folke et al. 2005, Newman and Dale 2005, More and Westley 2011).Against this background, resilience scholars are increasingly embracing the study of social networks as a promising way to operationalize social-ecological systems research (Janssen et al. 2006, Bodin et al. 2011, 2014, Bodin and Tengö 2012).
In general, a social network perspective refuses individualistic explanation of human behavior and places emphasis on the study of the relations between individuals and the structure of these relationships (Emirbayer and Goodwin 1994, Wassermann and Faust 1994, Fuhse and Mützel 2010).However, there is no uniform theoretical explanation as to why and how the structure of social relations matters (Borgatti et al. 2009, Fuhse andMützel 2010), which has resulted in various conceptualizations and operational approaches toward social network research.
Ecology and Society 22(1): 10 http://www.ecologyandsociety.org/vol22/iss1/art10/Probably the most common and intuitive conception of social networks is as "pipes" (Podolny 2001), connecting various actors through flows of resources, information, or knowledge (Borgatti et al. 2009).Another popular conception is that of networks as "social capital" (Bourdieu 1986, Coleman 1988, Portes 1998, Putnam 2000, Woolcock and Narayan 2000).In essence, the concept of social capital addresses the value of social connectedness (Borgatti and Foster 2003) in terms of competitive advantages (Burt 2000) derived from resources embedded in social structure (Lin 1999).Finally, the conception of networks as a "form of coordination," as opposed to other principles of coordination such as market or hierarchy (Powell 1990), emphasizes the deliberative character of social networks and their potential to facilitate collective action, self-organization, and cross-scale coordination (Schneider et al. 2003, Olsson et al. 2004, Folke et al. 2005, Carlsson and Sandström 2008, Newig et al. 2010).
With regard to the operationalization, three approaches to social networks can be distinguished: metaphorical, descriptive, and structurally explicit (Bodin et al. 2011).In general, studies following metaphorical approaches treat networks as binary variables, which either do or do not exist, whereas studies following descriptive approaches distinguish key properties of networks, such as size, density, or strength of ties.In contrast, studies following structurally explicit approaches draw on formally defined methods of social network analysis (SNA) to analyze structural patterns of social relations derived from relational data.
Although recent years have witnessed major advances in employing a social network perspective in research on SES in general (Bodin and Prell 2011), the role of social networks for the resilience of rural communities is still under-researched and underconceptualized.An increasing number of studies are applying a social network perspective in addressing diverse issues such as, for example, the diffusion of sustainable agricultural practices (Conley and Udry 2001, Bandiera and Rasul 2006, Isaac 2012), the exchange of financial and material support in times of need (Cassidy and Barnes 2012, Scheffran et al. 2012, Islam and Walkerden 2014), and collective action regarding the sustainable management of natural resources (Tompkins et al. 2002, Crona and Bodin 2006, Ramirez-Sanchez and Pinkerton 2009).However, findings from these studies have rarely been integrated from a resilience perspective because contrasting conceptualizations and operationalization of social networks are hindering the exchange between disciplines.Moreover, a systematic synthesis of current research on social networks in rural communities in the Global South is lacking, as are conceptual reflections about implications for future research on the resilience of rural communities.
To close this gap, we present a systematic review of case studies from three different strands of research: (i) natural resource governance, (ii) agricultural innovation, and (iii) social support.Although studies in these strands do not necessarily refer to the concept of resilience in explicit and theoretically founded ways, we opt for a review of studies from these strands because they provide examples of how a social network perspective can be applied in addressing different aspects relevant to the resilience of rural communities in the Global South.For example, research on governance networks provides insights into how social networks facilitate collective action of stakeholders and the navigation and transformation of management systems; research on agricultural innovation networks reveals how social networks facilitate learning between farmers about improved agricultural crops and practices and therefore foster purposeful adaptation to changing conditions; and research on social support networks addresses the role of social networks as a means for households and communities to cope with changes by providing access to resources in times of need.
Although investigating related things, research in each strand is rooted in a different disciplinary background and hence tends to look at social networks from a different perspective.For example, research in the strand of natural resource governance is influenced by environmental management and SES research and hence focuses on social networks as a means of improving collaboration between stakeholders; research in agricultural innovation is informed by agricultural and development economics and hence perceives networks as a means of improving knowledge diffusion and social learning between farmers; and research on social support networks is shaped by vulnerability and disaster risk research and hence is primarily concerned with networks as a livelihood strategy of households and communities.Accordingly, studies from each strand tend to conceptualize and operationalize social networks differently.
By systematically analyzing how studies across these three strands conceptualize and operationalize social networks, this review aims at critically discussing the viability of current social network research and intends to reflect conceptual implications for future research.In the following sections, we outline the analytical framework and present the findings of our review.Based on this, we discuss strengths and weaknesses of each strand in addressing different aspects of resilience.Finally, we conclude by proposing a translocal social network perspective as a conceptual framework for future research on social networks and the resilience of rural communities in the Global South.To analyze how social network research is conceptualized (a), we applied the following categories (see Table1): a.
(1) Conceptual framing: With this category, we indicate whether studies address resilience implicitly or explicitly, and how they frame social networks and resilience. a.
(2) Network variable: This category indicates whether studies treat social networks as an independent or dependent variable.
Studies treating networks as an independent variable focus on how the structure of social relations impacts social behavior.If the focus is on why people are linked in a particular way, networks are treated as a dependent variable (Bodin andCrona 2009, Henning et al. 2012). a.
(3) Network narrative: We choose this category to address underlying theoretical assumptions about how networks make a difference.This includes the conception of social networks as "pipes" (Podolny 2001), as "social capital" (Bourdieu 1986, Coleman 1988, Putnam 2000), and as a "form of coordination" (Powell 1990).
To analyze how social network research is operationalized (b), we applied the following categories: b.(1) Network approach: In line with Bodin et al., we distinguished between metaphorical approaches, descriptive approaches, and structurally explicit approaches (Bodin et al. 2011). b.
(2) Network definition: This category refers to the definition of actors and the social relations of interest between them (Wassermann and Faust 1994), e.g., farmers, households, or institutions and the exchange of material support, information, or knowledge; as well as to the definition of the scale of interaction (Prell 2011), e.g., cooperation between different levels at different administrative or geographical scales. b.
(3) Network analysis: With this category, we indicate on which network level the analysis focuses, e.g., the individual actor, the subgroup, or the network level (Bodin andCrona 2009, Bodin andPrell 2011), and which specific characteristics are highlighted, e.g., actor, tie, or structural characteristics or network context (Entwisle et al. 2007, Doreian andConti 2012).
General categories were used to be able to account for a broad spectrum of case studies.In reality, categorization is not a clearcut process, and studies could be attributed to more than one category.Hence, except for the network approach, we allowed multiple nominations, for example, a combination of network narratives.At the same time, we took into account that categories might not be applicable in all cases.For example, a study following a metaphorical approach might not be explicit about the network level or characteristics addressed.

RESULTS
Sixty case studies were selected for in-depth analysis: 22 studies from strand (i) natural resource governance, 17 studies from strand (ii) agricultural innovation, and 21 studies from strand (iii) social support.In the following, we present an overview of how these studies conceptualize and operationalize social networks and summarize key findings for each strand.We refer to general characteristics of each strand and highlight particular case studies only where they are needed to illustrate differences in the conceptualization and operationalization of social networks.Detailed information on each case study is provided in Appendix 1.

Network conceptualization
Conceptual framing: Social networks are conceptualized as key factors for understanding collective action and learning in SES.
Even if studies do not explicitly refer to resilience, they conceptualize social networks as central to the management of natural resources.Particular studies draw on concepts such as adaptive comanagement and hence implicitly refer to the resilience of SES (Marín and Berkes 2010, Stein et al. 2011, Moeliono et al. 2014, Apgar et al. 2015, Mannetti et al. 2015).
Network variable: Studies predominantly focus on the structure of social relations and their impact on management outcomes, treating social networks as an independent variable, though there http://www.ecologyandsociety.org/vol22/iss1/art10/are exceptions that take into account factors impacting social networks, such as ecological (Ramirez-Sanchez and Pinkerton 2009), economic (Rico García-Amado et al. 2012), and political changes (Ireland andThomalla 2011, Sundstrom et al. 2012).
Network narrative: Underlying most studies is the conception of networks as a "form of coordination," either focusing particularly on the communication and knowledge flows between resource users at the community level (Crona andBodin 2006, 2010), or with an emphasis on formal organizational networks (Gelcich et al. 2010, Stein et al. 2011, Cohen et al. 2012, Cárcamo et al. 2014, Moeliono et al. 2014, Nuno et al. 2014, Pietri et al. 2015).Furthermore, several studies refer to social networks as a structural feature of "social capital."In this way, they either explain the performance of particular organizations (Marín andBerkes 2010, Marín et al. 2012) and individual actors (Ramirez-Sanchez andPinkerton 2009, Rico García-Amado et al. 2012) or investigate the potential of collective action for conflict resolution (Sanginga et al. 2007) and disaster risk recovery (Ireland andThomalla 2011, Marín et al. 2015).

Network operationalization
Network approach: Structurally explicit approaches, and in particular the application of SNA techniques, are characteristic of most studies in this strand, however, there are also studies that refer to social networks in metaphorical terms, treating networks as an binary variable (Tompkins et al. 2002, Sanginga et al. 2007, Gelcich et al. 2010, Sundstrom et al. 2012).
Network definition: The social relations emphasized involve information, knowledge exchange, and collaboration between resource users and stakeholders.Network boundaries are defined with reference to management systems with a limited set of stakeholders and with clear ecological, geographical, or administrative boundaries.Whereas social ties in most cases are perceived as facilitating exchange and mutual understanding, few studies point to the restrictive potential of social ties (Marín andBerkes 2010, Marín et al. 2012).Several studies explicitly address cross-scale interactions between various political and administrative stakeholders (Tompkins et al. 2002, Gelcich et al. 2010, Stein et al. 2011, Cohen et al. 2012, Cárcamo et al. 2014, Marín et al. 2015).
Network analysis: The focus of research comprises the network, the subgroup, and the individual level, whereby the focus of analysis is on structural network characteristics such as density, centrality, or fragmentation.Common tie characteristics are those of importance, frequency, or intensity.Several studies distinguish between bonding, bridging, and linking ties (Sanginga et al. 2007, Bodin and Crona 2008, Ramirez-Sanchez and Pinkerton 2009, Stein et al. 2011, Cohen et al. 2012, Marín et al. 2012, 2015, Sundstrom et al. 2012, Cárcamo et al. 2014, Apgar et al. 2015).
Few studies highlight the influence of network context (Tompkins et al. 2002, Rico García-Amado et al. 2012, Sundstrom et al. 2012) or actor characteristics, such as leadership or socioeconomic power (Bodin andCrona 2008, Crona andBodin 2010) to explain agency or lack thereof.

Key findings
Studies identify heterogeneity, cross-scale interaction, network density, and actor centrality as key factors influencing the resilience of governance networks.For the latter two factors, dense and centralized networks with strong bonding ties are shown to be effective in managing simple tasks (Rico García-Amado et al. 2012), while fragmented networks are shown to limit communication and hence adaptive capacity (Cárcamo et al. 2014, Mannetti et al. 2015).Furthermore, for successful transformation toward sustainable governance, studies indicate the need for decentralized and heterogeneous networks that entail bridging ties between administrative and institutional scales of management (Gelcich et al. 2010, Cohen et al. 2012, Marín et al. 2012, 2015) and, in particular, the need for brokers who facilitate collaboration between these scales (Stein et al. 2011, Cárcamo et al. 2014, Moeliono et al. 2014, Nuno et al. 2014, Pietri et al. 2015).
Studies elaborating on possible reasons for inertia in governance processes reveal homogeneity among centrally positioned opinion-leaders as a potential barrier to collective action (Crona and Bodin 2006, 2010, Bodin and Crona 2008, Ramirez-Sanchez and Pinkerton 2009, Moeliono et al. 2014, Mannetti et al. 2015).Social capital based on linking ties (Marín et al. 2012) and flexible arrangements with changing roles and responsibilities are suggested as better suited to meeting the challenges of adaptive management (Apgar et al. 2015).Particular findings show that, successful governance networks can enhance resilience to natural disasters (Marín et al. 2015), but favorable structures alone might not be sufficient to promote proactive resilience building if resource users are excluded from formal institution building (Ramirez-Sanchez and Pinkerton 2009).

Strand (ii): Agricultural innovation
Research in this strand is concerned with processes of social learning in the context of rural transformation.The case studies reviewed, for example, range from acceptance of improved crop varieties (Bandiera and Rasul 2006, Van den Broeck and Dercon 2011, Tatlonghari et al. 2012, Thuo et al. 2014) through the implementation of sustainable and risk-mitigating agricultural practices (Mazzucato and Niemeijer 2000, Conley and Udry 2001, Isaac et al. 2007, 2014, Matuschke and Qaim 2009, Arora 2012, van Rijn et al. 2012, Matouš et al. 2013, Wossen et al. 2013) to the use of modern information and communication technologies (Butt 2015).

Network conceptualization
Conceptual framing: In contrast to studies in the other two research strands, the majority of literature in this strand does not address resilience explicitly.However, from a development economics' perspective, studies perceive of social networks as factors shaping social learning and adaptive changes in the context of agrarian change, and hence implicitly address aspects relevant to resilience.A system's perspective is rare, although there are attempts to embed it in the context of innovation systems (Spielman et al. 2011, Arora 2012, Isaac 2012).A few studies link agricultural innovation with adaptive management (Isaac et al. 2007) or conceptualize social networks as a form of social memory contributing to resilience (Isaac et al. 2014).
Network variable: The majority of studies focus on the outcomes of networks-here the adoption of agricultural practices or technologies-treating social networks as an independent variable.Studies that address the impact of external changes on social network structure are the exception (Mazzucato and Niemeijer 2000, Arora 2012, Isaac et al. 2014, Butt 2015).
Network narrative: Underlying most studies is the conception of networks as pipes through which "flows" of information, knowledge, and advice are transferred and circulated between http://www.ecologyandsociety.org/vol22/iss1/art10/actors.Less frequently, social networks are conceptualized as social capital, explaining differences in adaptation processes between different groups of farmers (Hoang et al. 2006, Tatlonghari et al. 2012, van Rijn et al. 2012).
Network definition: The most frequently investigated social relation is information and advice sharing between farmers and external actors such as extension staff and NGOs (Arora 2012, Matouš et al. 2013, Wossen et al. 2013) Tatlonghari et al. 2012), while relations to external actors and institutions are referred to as weak or bridging ties (Wossen et al. 2013, Thuo et al. 2014).Structurally explicit studies focus on structural measures for explaining information diffusion, such as network density and fragmentation, as well as on actor centrality for identifying brokers of agro-ecological knowledge (Isaac et al. 2007, 2014, Isaac 2012).Network context is addressed by a few studies highlighting the roles played by institutions (Hoang et al. 2006, Spielman et al. 2011, Arora 2012), information technologies (Butt 2015), or migration (Isaac et al. 2014).

Key findings
Studies do not explicitly elaborate on the link between network features and resilience.However, they identify key factors influencing social learning and decision-making processes and hence provide insights into adaptive processes crucial for the resilience of SES.In this regard, studies highlight actor and tie characteristics rather than network structure.Challenging the simple assumption that having more actors in a network increases the likelihood of adopting new technologies, studies reveal that decisions are based on imperfect knowledge and are oriented toward the experience and adoption behavior of network members (Conley and Udry 2001, Matuschke and Qaim 2009, Wossen et al. 2013), and are often subject to strategic considerations (Bandiera and Rasul 2006).Regarding tie characteristics, two groups of studies can be distinguished.The first group identifies social and geographical proximity as conducive to information diffusion: Strong and homophilous ties, for example kinship ties, are shown to facilitate information diffusion (Bandiera and Rasul 2006, Hoang et al. 2006, Matuschke and Qaim 2009, Van den Broeck and Dercon 2011, Tatlonghari et al. 2012).In contrast, the second group of studies emphasizes the role of bridging and linking ties between diverse actors from civil society, public extensions, and the private sector, which provide farmers with access to external sources of information and experiences (Arora 2012, van Rijn et al. 2012, Matouš et al. 2013, Wossen et al. 2013, Isaac et al. 2014, Thuo et al. 2014).Within this group, particular studies highlight geographical factors.First, ties to geographically distant actors increase the likelihood that farmers will gain access to new information (Wossen et al. 2013), and second, experiences gained from farming in different agro-ecological settings can help to build social memory (Isaac et al. 2014).With regard to network structure, those studies applying structurally explicit approaches argue that, unlike governance, innovation requires sparse but efficient networks with a few central actors acting as brokers between formal and informal networks (Isaac et al. 2007, 2014, Isaac 2012).Particular studies point to the critical roles played by elite actors linking external actors and the community, and the danger of reproducing power imbalances through external interventions (Hoang et al. 2006, Spielman et al. 2011, Arora 2012).

Network conceptualization
Conceptual framing: The majority of studies in this strand conceptualize social networks as a source of resources supportive to the resilience of rural households and communities.Even studies not explicitly addressing resilience share the conceptualization of networks as coping strategy in times of need (Kadigi et al. 2007, Torkelsson 2007, Nygren and Myatt-Hirvonen 2009, Rindfuss et al. 2012, Gallego and Mendola 2013, Lyle and Smith 2014, Abizaid et al. 2015, Katikiro et al. 2015) and hence refer to particular aspects of resilience.Unlike studies focusing on governance and innovation, studies on social support more frequently take a community perspective conceiving of social networks as a means for communities to deal with external shocks and risks (Cassidy and Barnes 2012, Ekblom 2012, Baird and Gray 2014, Islam and Walkerden 2014, 2015).
Network variable: Studies tend to focus on the outcomes of social networks and therefore treat social networks as an independent variable.However, there are also studies treating social networks as a dependent variable, emphasizing how social support networks are influenced by the impact of socioeconomic factors, such as livelihood diversification (Baird and Gray 2014, Orchard (Torkelsson 2007), caste influence (Bosher et al. 2007), and migration (Scheffran et al. 2012, Gallego and Mendola 2013, Zimmerer 2014).
Network narrative: Most studies in this strand refer to the notion of networks as social capital explaining differences in the vulnerability of households due to their different embeddedness.Unlike literature on agricultural innovations and governance, few studies in this strand build on the notion of networks functioning as pipes for the exchange of different types of support (Rindfuss et al. 2012, da Costa et al. 2013, Zimmerer 2014, Abizaid et al. 2015, Katikiro et al. 2015) or as a form of coordination (Downey 2010).

Network operationalization
Network approach: In comparison to the other two strands, this strand is characterized by a more equal presence of all three operational approaches.Descriptive approaches, characterizing social networks according to the nature of the ties involved, account for the majority.Metaphorical approaches that refer to either the existence or the decline of social networks as an explanatory variable of resilience are more frequent than in other strands (Kadigi et al. 2007, Ekblom 2012, Scheffran et al. 2012, da Costa et al. 2013, Zimmerer 2014, Katikiro et al. 2015).
Structurally explicit approaches, drawing on methods of SNA (Downey 2010, Cassidy and Barnes 2012, Lyle and Smith 2014, Abizaid et al. 2015, Orchard et al. 2015) are more frequent than in the strand of agricultural innovation but less frequent than in the governance strand.
Network definition: The dominant social relation of interest is the exchange of material, financial, and emotional support between rural households at the village level.Studies tend to concentrate on the village level (Cassidy and Barnes 2012, Islam and Walkerden 2014, Lyle and Smith 2014, Abizaid et al. 2015), though there are also studies emphasizing the role of social ties that extend beyond the community (Ekblom 2012, Rindfuss et al. 2012, Scheffran et al. 2012, Gallego and Mendola 2013, Islam and Walkerden 2015, Orchard et al. 2015).
Network analysis: The main level of analysis is that of the individual, in this case, households, whereas structurally explicit studies also give attention to the network level (Cassidy andBarnes 2012, Lyle andSmith 2014).Frequently addressed characteristics include tie reciprocity and tie strength, the latter being operationalized either as bonding ties of kinship and bridging ties of neighborhood and friendship (Islam andWalkerden 2014, 2015), or as bonding ties within the community and bridging ties to actors outside the community (Rotberg 2010, Baird and Gray 2014, Islam and Walkerden 2014, 2015, Orchard et al. 2015).Compared to other strands, there is a stronger focus on network context, including social institutions and socioeconomic changes (Torkelsson 2007, Nygren and Myatt-Hirvonen 2009, Baird and Gray 2014, Katikiro et al. 2015, Orchard et al. 2015).Structurally explicit studies focus on structural characteristics such as density, hierarchy, and the centrality of particular households (Downey 2010, Cassidy and Barnes 2012, Lyle and Smith 2014, Orchard et al. 2015).

Key findings
Regarding the coping aspect of social networks, several studies emphasize the importance of strong ties of reciprocity and trust at the community level (Kadigi et al. 2007, da Costa et al. 2013, Goulden et al. 2013, Katikiro et al. 2015) or, more specifically, to the combination of strong and weak ties (Rotberg 2010, Islam and Walkerden 2014, 2015).Particular studies point to the importance of temporal dynamics by revealing that the composition and viability of bridging and bonding ties is not fixed but changes over time (Baird andGray 2014, Islam andWalkerden 2014).Network transitions from traditional support systems to diversified market-oriented networks are shown to have ambiguous implications for community resilience.For example, transitions might foster the capacity to cope with high-incidence/ low-severity impacts, while at the same time reduce the ability to manage low-incidence/high-severity shocks (Baird andGray 2014, Orchard et al. 2015).With regard to actor characteristics, studies show that gender and socioeconomic status determine access to and ability to utilize social networks (Bosher et al. 2007, Torkelsson 2007, Cassidy and Barnes 2012, Rindfuss et al. 2012, Abizaid et al. 2015).Taking into account network structure, some studies conclude that more central households are more resilient because they can access more resources (Cassidy andBarnes 2012, Lyle andSmith 2014).Dense networks are shown to have higher redundancy and hence better opportunities to mobilize resources and act collectively, while larger and less redundant networks might yield greater returns (Orchard et al. 2015).Taking into account the effect of external factors on support networks, a small group of studies indicates the effects of migration either on participation in community networks (Gallego and Mendola 2013), or on livelihoods and resilience in the places of origin (Ekblom 2012, Rindfuss et al. 2012, Scheffran et al. 2012).

DISCUSSION
This review of case studies on the role of social networks in the Global South extends beyond disciplinary boundaries.Its categorization system permits the different strands to be compared and thus allows similarities, differences, and blind spots to be revealed.This opens up the opportunity to critically assess the viability of a social network perspective for addressing the resilience of rural communities in the Global South as well as to discuss implications for future research.

Strand (i): Natural resource governance
Approaching social networks from a systems perspective, this strand addresses the capacity of social networks to navigate the transformation of SES toward sustainable resource use and resilience.A particular strength of this approach lies in linking social network patterns with particular resilience features (Newman andDale 2005, Janssen et al. 2006).Against this background, studies provide instructions for strengthening the resilience of governance systems; they offer opportunities to identify cross-scale mismatches and barriers in governance processes (e.g., Crona and Bodin 2006, Stein et al. 2011, Moeliono et al. 2014, Nuno et al. 2014), to recognize potential change agents (e.g., Crona and Bodin 2010, Cárcamo et al. 2014, Moeliono et al. 2014), to design more sustainable governance regimes (e.g., Gelcich et al. 2010, Cohen et al. 2012, Marín et al. 2012, 2015, Pietri et al. 2015).In doing so, they can draw on the elaborated toolkit of SNA, which is increasingly applied not only in the context of resource governance in the Global South but around Ecology and Society 22(1): 10 http://www.ecologyandsociety.org/vol22/iss1/art10/ the globe (Bodin and Prell 2011).Particularly promising in this regard is the application of SNA for disentangling coupled SES and for investigating the alignment of social and ecological structures and processes (Bodin and Tengö 2012, Bodin et al. 2014, Roldán et al. 2015).
Structurally explicit approaches, as applied in most studies, have their drawbacks, however.Formal assessment of network structure requires clearly defined network boundaries (Scott 2013), a methodological restriction in the context of dynamic rural societies.As well, the focus on definable management systems tends to ignore particularities of resource governance in the Global South, such as social, economic, and political conditions impacting the livelihoods, needs, and rationalities of stakeholders.Another drawback stems from the underlying assumption that exchange and communication between various actors inevitably increases understanding and the willingness to act collectively (Schneider et al. 2003, Carlsson and Sandström 2008, Newig et al. 2010).This collaborative bias tends to downplay conflicts underlying many current resource management issues in the Global South (McNeish 2010), in particular the role of power asymmetries (Crona and Bodin 2010).A technical and apolitical understanding of governance is problematic because it portrays resource conflicts as a manageable task involving the modification of network patterns (Zimmer andSakdapolrak 2012, Scott 2015).In the same way, any argument that SNA can be used as a tool to improve governance processes and hence contribute to resilience runs the risk of reducing resilience building to a mere technical challenge (Scott 2015).
Instead of assuming that favorable network patterns will "lubricate the machinery of natural resource governance" (Crona and Hubacek 2010), more attention should be paid to the skills, means, and motivation of centrally positioned actors to promote new ideas and prompt collective action (Crona andBodin 2010, Moore andWestley 2011).A critical approach to governance ought to address the question of what mode of social-ecological interactions promotes specific governance systems and whose resilience this might foster or impede (Cretney 2014).This would also entail a stronger emphasis on the social and historical context of resource governance in the Global South.

Strand (ii): Agricultural innovation
In contrast to the system perspective of the governance strand, this strand adopts an actor-based perspective on the capacity of social networks to promote adaptive capacity through social learning and the adoption of technology in the context of agrarian change.Its strength lies in accounting for a variety of actor and tie characteristics (e.g., Conley and Udry 2001, Bandiera and Rasul 2006, Tatlonghari et al. 2012, Thuo et al. 2014) as well as social, political, and religious factors (e.g., Mazzucato and Niemeijer 2000, Matouš et al. 2013).Building on descriptive approaches utilizing sophisticated econometric methods, studies in this strand are less restricted by defining boundaries and are more conscious of the relevance of relations that cross geographic scales (e.g., Matuschke and Qaim 2009, Isaac 2012, van Rijn et al. 2012, Wossen et al. 2013, Isaac et al. 2014).Although studies do not explicitly address the links between social networks and resilience, they provide valuable information about how sustainable innovations, and hence adaptive capacity, can be promoted by research, development, and policy.(e.g., Hoang et al. 2006, Spielman et al. 2011, Van den Broeck and Dercon 2011).
Although the descriptive approaches offer greater flexibility in the network definition than structurally explicit approaches, they are limited in their ability to assess network structure.Furthermore, using network proxies such as group membership involves methodological problems.First, farmers might adopt or choose group membership because of unobserved individual characteristics or hidden variables.Second, the behavior of the group might influence the individual, who in turn might influence the group (Manski 1993).Ways of circumventing these problems have been suggested (Bandiera andRasul 2006, Matuschke andQaim 2009).However, these adjustments do not account for the simplistic conceptualization of networks as pipes, which tends to oversimplify decision-making processes in rural contexts.This omission is of particular relevance because work on social contagion (Burt 1987) suggests that social actors align their behavior with reference not only according to directly available information but also according to perceived norms and roles regarding their positions within a given network structure (Burt 1987, Grabher 2006).Of further concern is the strand's bias toward economic explanations of decision making, which downplay the roles of social, political, and environmental aspects in mediating the social and economic values of innovations.
Seen from a systemic perspective, a major factor that stands in the way of understanding resilience is the strand's focus on decisions at the individual level.Understanding how social networks facilitate or impede the adoption of more sustainable agricultural practices is a major, but not a sufficient, basis upon which to make claims about the resilience of SES (Carpenter et al. 2001).As a way forward, studies that approach innovation networks from a systems' perspective (Spielman et al. 2011, Isaac 2012) and link them to concepts such as adaptive management and social memory (Isaac et al. 2014) might be instructive in addressing multiscale interactions and positioning them in social, political, and cultural contexts (Atwell et al. 2008).

Strand (iii): Social support
This strand provides insights into the ways in which households employ their social networks as a strategy to cope with and recover from risks.Its strengths lie in providing a community perspective on household coping strategies and in employing a multimethod mix comprising quantitative and qualitative aspects of social networks.This combination offers the flexibility to take into account actor, tie, and network characteristics, as well as the impact of network context such as social institutions (e.g., Bosher et al. 2007, Torkelsson 2007, Nygren and Myatt-Hirvonen 2009) and socioeconomic changes (e.g., Baird and Gray 2014, Zimmerer 2014, Katikiro et. al 2015, Orchard et al. 2015).Furthermore, in contrast to the strands of governance and innovation research, studies in this strand more explicitly account for the temporal dynamics of social networks (e.g., Rindfuss et al. 2012, Goulden et al. 2013, Islam and Walkerden 2014), and hence provide a more nuanced understanding of how the resilience of rural households evolves in the context of rural transformation (Rigg 2006).
One particular issue of concern, however, stems from reducing social networks to "assets" that households have at their disposal.A tendency to reiterate tautological assumptions about the positive role of social capital (Nygren and Myatt-Hirvonen 2009) is particularly prevalent in metaphorical approaches that consider the mere existence of networks.This is a one-sided perspective, because networks are not necessarily solely beneficial but may http://www.ecologyandsociety.org/vol22/iss1/art10/also exclude actors from community resources and reinforce dependencies and differences between the actors (Bohle 2006, Torkelsson 2007, Steinbrink 2009).Furthermore, networks are not always readily available but involve time and resources to maintain (Nygren andMyatt-Hirvonen 2009, Lyle andSmith 2014), and their effectiveness might be limited by risks faced by its members (Gallego and Mendola 2013).Accordingly, participation in community networks is an ambiguous proxy for resilience, not least because it excludes those who cannot afford to be part of the network (Torkelsson 2007, Cassidy andBarnes 2012).Another drawback of studies in this strand is their tendency to conceive of social networks as consisting of strong bonding ties as relations of reciprocity and trust.Indeed, a bias toward reciprocal ties neglects the fact that norms of reciprocity, in particular between close family and kin, can exert high social pressure, and hence weak ties might be prioritized when seeking support (Nygren and Myatt-Hirvonen 2009).
In terms of the resilience of rural communities, it is not only problematic to confuse social proximity with the degree of support but also problematic to narrowly focus on the community level as the primary level of social interaction.Studies taking into account the impact of external actors on the resilience of rural households (e.g., Islam andWalkerden 2014, 2015) are providing important insights on the impact of external factors but might not be sufficient to address the mobility of rural livelihoods in the Global South (Ellis 2003, Rigg 2006).Rather studies should shift attention toward social ties spanning different locations to address coping capacity in an increasingly connected world (Scheffran et al. 2012).

General challenges
Besides the strengths and weaknesses of each research strand, our review also identifies general challenges: current case studies on the role of social networks tend to provide a static view of network outcomes, emphasize structure over agency, and neglect the spatial dimensions of social relations.A general challenge that has to be met by future social network research in the Global South is the tendency to abstract social structure from network context (Entwisle et al. 2007).Indeed, the majority of studies focus on the outcomes of networks rather than on the question of how social networks evolve in the context of change (Baird and Gray 2014).In most cases, studies addressing the impact of external drivers such as socioeconomic and political factors are following metaphorical or descriptive approaches and thus tend to remain silent about impacts in terms of network structure.Structurally explicit approaches that could provide these insights often fail to make sense of network context.Building on heuristic assumptions about how structural patterns are related to resilience features (Bodin et al. 2006, Janssen et al. 2006), studies following an analytical explicit approach tend to make general judgments about "trade-offs" between structural features and the "right mix" of ties instead of addressing the quality of ties for particular purposes (Videras 2013) and identifying contextual aspects of social interaction (De Nooy 2013).Moreover, studies tend to focus on the assessment of networks at a given point in time.In dynamic contexts, such as that of rural transformation, however, assessing "network snapshots" (Ernstson et al. 2008) is not sufficient to make causal claims about resilience in the long term (Bodin and Prell 2011).This applies in particular when taking into account that SES evolve through adaptive cycles (Gunderson and Holling 2002) and that changing systems configurations might require different social networks (Downey 2010, Goulden et al. 2013).Studies using long-term panel surveys might overcome this challenge but are time and resource consuming (Rindfuss et al. 2012).
A much-discussed issue in network research is its inability to address the dialectical relationship between social structure and agency (Crona et al. 2011).This problem arises in the majority of our sample studies, which implicitly or explicitly build on the assumption that the presence of favorable networks is sufficient to ensure agency, here the ability to identify and enact solutions to sustainable development challenges (Newman and Dale 2007).However, addressing only one part of the iterative cycle between social processes and social structure (Bodin and Prell 2011) fails to make sense of the mechanisms through which social relations are reproduced and configured over time (Emirbayer and Goodwin 1994).Treating social structure "as is" (Bodin and Prell 2011:365) does not reflect how that structure evolves through communicative processes (Fuhse andMützel 2010, Ingram et al. 2014) and neglects the critical role played by the means, skills, and motivation of particular social actors who "make things happen" (Crona et al. 2011:53) and, in particular, how they create social networks conducive to resilience (Moore and Westley 2011).
Finally, a further bias of current network research is its tendency not to take the spatial dimensions of social networks seriously.Indeed, most studies reviewed adopt a network-centric perspective, with a one-sided conception of horizontal and frictionless social "spaces of flows" (Jessop et al. 2008:391).In general, spatial assumptions underpinning social inquiries should be treated with caution (Jessop et al. 2008) to avoid falling into the trap of determinism.In the context of the Global South, where mobility and multiple connections between rural and urban areas are the norm rather than the exception (Ellis 2003, Steinbrink 2009), a spatially blind form of social network research, however, risks losing sight of significant determinants of rural livelihoods.Migration is a major strategy for livelihood diversification (Rigg 2006, World Bank 2011) and climate change adaptation (Black et al. 2011) and hence should be accounted for in studies addressing the role of social networks for the resilience of rural communities (Rindfuss et al. 2012, Scheffran et al. 2012, Gallego and Mendola 2013, Isaac et al. 2014).
Although all three challenges could apply to networks research in general, we argue that they are of particular concern for understanding the role of social networks for the resilience of rural communities in the Global South.Current social network research, with its static focus on network outcomes and its inability to take social agency sufficiently into account, is ill suited to addressing temporal and spatial dynamics in factually highly mobile societies (Ellis 2003, Rigg 2006).Furthermore, it provides an ahistorical perspective on social networks that tends to mask the political nature and colonial history of resource conflicts (McNeish 2010).

Toward a translocal social network perspective
Against the backdrop of these challenges, we envisage a social network perspective on resilience that takes into account the complexity and dynamics of rural livelihoods in an increasingly connected world.As a means to this end, we propose integrating research on social networks and resilience with the concept of translocality (Greiner and Sakdapolrak 2013a).http://www.ecologyandsociety.org/vol22/iss1/art10/The concept of translocality addresses the increasing connectedness of daily life, which is inter alia facilitated by multiple forms of mobility, including everyday movements, and seasonal and long-term migration (Brickell and Datta 2011).By emphasizing the simultaneous embeddedness of social actors in translocal networks spanning different locales, translocality combines the socio-spatial dimensions of both place and social networks (Jessop et al. 2008).It thereby challenges dichotomous geographical conceptions such as space/place, rural/urban, and core/periphery (Steinbrink 2009, Greiner andSakdapolrak 2013b).Instead of conceiving of migration as a singular and unidirectional movement of people, translocality highlights the importance of migration-induced feedback processes between areas of origin and destination.This includes the circulation and flows of ideas, symbols, knowledge, and practices between mobile and nonmobile actors through translocal social networks (Greiner and Sakdapolrak 2013a).Hence, embeddedness in these translocal networks determines the availability of and access to resources and therefore has the potential to strengthen the resilience of its actors (Scheffran et al. 2012, Sakdapolrak 2014).
Accordingly, the notion of translocal resilience points to the role of translocal networks in conditioning the capacity of particular actors, households, and communities to cope with and adapt to changes, transform livelihoods, and explore alternative modes of social-ecological interaction (Sakdapolrak 2014, Sakdapolrak et al. 2016).These capacities in turn impact the means and opportunities to shape and utilize translocal networks.In other words, translocal networks are both preconditions and outcomes of the resilience of rural communities.By acknowledging that different capacities at individual, household, and community level are not necessarily complementary but might compete with each other, the notion of translocal resilience places particular emphasis on the role of social norms and power asymmetries in negotiating and defining desirable resilience outcomes.In other words, it provides a "situated" approach to resilience that broadens the scope of research toward including the processes and social relations that support resilience (Cote and Nightingale 2012).
A translocal network perspective, we claim, holds promise for addressing the challenges faced by current research on social networks and resilience in the Global South.First, by integrating the socio-spatial dimensions of networks and place, a translocal network perspective shifts the research focus from locally bound entities, such as the village, a region, or a management area, to the connectedness between actors at different places, while, at the same time, emphasizing the role of spatiality in social networks.Second, by taking into account mutual feedback processes between areas of origin and destination, it facilitates a dynamic understanding of complex rural transformations that cannot be understood by focusing on locally bound networks only.Third, it draws attention to the dialectic relationship between social structure and agency by revealing how capacities of resilience are related to daily practices of mobile and nonmobile actors in utilizing and shaping their networks.In doing so, it has the potential to contribute to a resilience research "of fine nuances," in the sense of Bourdieu, which takes into account economic and social power relations from the local to the global level (Deffner et al. 2014).Finally, a translocal network perspective would be suited to overcoming the apolitical tendencies of both resilience and network theory through reassessing resilience and social networks from a critical sciences perspective (Scott 2015).

CONCLUSION
This review provides a systematic overview on the conceptualization and operationalization of social networks across three strands of research and a discussion of their strengths and weaknesses in addressing aspects of the resilience of rural communities in the Global South.Research on governance networks, rooted in SES research, predominantly conceptualizes social networks as a form of coordination in the context of management system transformation.With its bias toward methods of formal network analysis, studies are powerful in providing insights into how networks can facilitate cross-scale adaptive management and how structural patterns relate to key system features relevant for the resilience of SES.However, because of methodological constraints, studies are limited to clearly identifiable management systems and tend to underestimate the role of human agency and power asymmetries.Contrastingly, research on innovation networks, informed by development economics, centers around the conception of social networks as pipes of information exchange required for the adaptation to changing conditions.Econometric methods provide opportunities for assessing a wide range of factors relevant, for example, to the purposeful changes of crops and practices; however, they remain descriptive in nature and vague with regard to the impact of these changes on resilience on higher levels.Studies on social support, rooted in vulnerability and disaster research, address the role of social networks as a means of coping with shock.By conceptualizing social networks as social capital, studies in this strand help to broaden the scope of vulnerability and livelihoods research.However, they tend to focus on social networks as assets at the community level thereby omitting the role of migration-induced feedback processes between areas of origin and destination.
Opportunities for sharpening and developing future research agendas include inter alia a critical approach to governance networks that reconsiders the role of actors' differential agency and power asymmetries; an integration of actor-and systemsbased approaches to agricultural innovation networks; and finally a shift away from stressing reciprocal and trusting relations at the community level toward addressing support networks spanning multiple locales in the context of mobility and social, economic, and political changes.More specifically, the review points to central challenges to be met in future research on social networks and resilience in the Global South.These particularly include the tendency of current network research to focus on network outcomes and the difficulties involved in assessing network dynamics, an overemphasis of network structure while undertheorizing the role of agency in shaping and reproducing social networks, and the tendency to neglect spatial dimensions of social relations despite the highly mobile character of many rural societies.
To address these challenges, we propose linking future research with the concept of translocality.A translocal social network perspective on the resilience of rural communities addressing embeddedness in and connectedness between places shifts the focus of research from bounded entities toward the connections between places; it takes into account the dynamic http://www.ecologyandsociety.org/vol22/iss1/art10/interrelationship between structure and agency and provides a multidimensional conception of social relations.Hence, it offers a framework well suited to the complexity of rural-urban realities in the Global South.Actors: agencies and organizations.

Responses to this
Scale: localinternational governance system.
Network: governance network.
Individual: bridging organizations or institutions.
Actor: organization type, scale of operation, preference for learning.
Cross-scale communication transcending formal governance systems facilitates coordination and learning.
A trade-off exists between facilitating flows of knowledge between various levels and promoting learning on local level.

Crona and Bodin 2006
Investigate if a lack of collective action to remedy unsustainable use of fishery resources may be attributed to the structures of the social networks in a Kenyan coastal village.

Implicit reference to resilience:
Social networks as factor determining successful implementation of community based comanagement. Independent: Influence of networks on collective action towards the comanagement of common-pool resources.
Form of coordination: Homogeneity among key players is likely to reduce communities' ability to synthesize new knowledge.
Low levels of ties among members of the network, higher fragmentation and overall low network cohesion hamper the opportunities for collective action.

Marín and Berkes 2010 †
Assess the Chilean shellfish comanagement system from an organizational network perspective including networks of actors, the functions of these actors, and fishers' perceptions about Chile's comanagement arrangement.

Implicit reference to resilience:
Co-management as alternative form of governance that is comprised of multiple and complex social networks. Independent: Influence of networks on management outcome.

Form of coordination:
Social networks as formal arrangements between complex sets of actors.

Structurally explicit:
Focus groups, indepth interviews, and questionnaire with representatives of fishery associations in two administrative regions.
Social relation: facilitation, hindrance of co-management.
Actos: small-scale fisher associations, public or private counterparts.
Scale: local -national governance system.
Subgroup: functional groups in hindering and facilitating networks.
Studies of co-management should not focus only on collaborative (or facilitating), but also on hindering relationships.
Highly centralized governance systems hamper participation of grassroots organizations.
More autonomy and ability to use existing linking social capital between local organizations and actors at other scales would provide an opportunity to experiment and learn.

Implicit reference to resilience:
Adaptive management of socialecological systems requires networks fostering the combination of local knowledge, crossscale coordination, and social learning.

Independent:
Influence of network structure on information exchange and governance outcome.

Form of coordination:
Governance networks as purposefully created networks of inter-organizational linkages.

Structurally explicit:
Semi-structured interviews with representatives of organizations involved in REDD+ Indonesia.
Social relation: regular exchange of information.
Actors: organizations from academic, private, and public sector.
Network: governance network.
Actor: number of employees, headquarter locations, membership in organizations.
Homophily and power generate a network that lacks the integration between diverse groups.
Power asymmetries can hinder effective and inclusive governance.
Brokers able to connect different clusters of information seeking will be crucial for effective and inclusive management.

Nuno et al. 2014
Analyze the multiple roles played by different institutions in the management system of the Serengeti and describe the interactions between different actor types.

Explicit reference to resilience:
Social networks as a means to characterize collaborations and social relationships facilitating conservation and promoting resilience of governance. Independent: Influence of social networks on management outcome.

Form of coordination:
Management systems as networks of organizations involved in resource management.

Structurally explicit:
Scenario building, institutional analysis, and semi-structured interviews with representatives of organizations responsible for the regulation of resource use in the Serengeti.
Context: market changes and requirements.
Market requirements shape networks Organic coffee commercialization is the main source of bridging ties that have resulted in more connectivity and resilience.
Despite power asymmetries and internal conflicts, the local network facilitates an effective management of common pool resources.
Institution-building is required, because highly centralized networks may not be appropriate for governing socialecological systems in the long term.

Implicit reference to resilience:
Social networks matter for the adaptive co-management of natural resources. Independent: Influence of social networks on governance outcome.

Form of coordination:
Governance networks consisting of formal and informal relations between political, social, and economic organizations and institutions.
Structurally explicit: Questionnaire and semi-structured interviews with stakeholders in the catchment area (including leaders in four villages).
Social relation: funding, information and knowledge exchange, collaboration.
Actors: organizations and institutions involved in water management.
Scale: local -regional governance system.
Network: governance network.
Village leaders play a crucial role in linking otherwise disconnected actors but are not integrated in the management system.
Instead of imposing top-down institutional arrangements, it is more promising to build on existing social structures.Actors: institutions involved in the management of coastal resources.
Scale: local -international governance system.
Network: spaces of dependency (between localized organizations), spaces of engagement (between local stakeholders and external interests).
Context: institutional setup / legislation and regulations.
Cross-scale networks may permit an institutional shift towards more integrated and inclusive approaches.
There are winners and losers in any strengthening of networks for the comanagement of resources.Thus there is a need to understand the institutional form of networks facilitating inclusive decision making at various scales.† We included studies from Chile, despite its high-income status, in order to take account of representative studies.Individual: farmers.
Actor: adoption rate of agroforestry practices.
Tie: number of ties, farmer-to-farmer ties, organizational ties.
The increase of organizational ties is related to a decrease of network density in personal producer networks as they replace producer-to-producer ties.
Producers with ties to organizations are likely to be positioned in more sparse and effective farm management, specifically the management of agrodiversity in Ghana.
to the adoption of innovations and influences the advancement or weakening of sustainable farm management.information rate rural areas.SNA: ego-networks.tension, NGOs).
Scale: local / community + linkages to external actors.
efficient information networks as identified by a low level of redundant ties correlated to higher reported on-farm agrodiversity.
Unlike common-pool resource management which requires dense networks, exchange of complex information requires diffuse but efficient social networks.

Isaac et al. 2014
Investigate the role of migrant farmers in agricultural information networks and explore implications on the adoption of pro-environmental management practices in Ghana.

Explicit reference to resilience:
Environmentally induced migration as a livelihood strategy shaping agricultural information networks and therefore the sustainable management of agroecosystems and the resilience of the region of destination. Dependent: Influence of migration on social networks and their ability to promote the adoption of environmental management practices. Pipe: Networks as conduit for the exchange of knowledge on agricultural practices.
Structurally explicit: Famer interviews in three rural communities.
Social relation: exchanging and seeking information on agricultural practices.
Actors: farmers in the community and farmers in other communities.
Scale: local / community + linkages between communities.
Network: agricultural communication network.
Actor: settlement categories, origin and location, demographic data, land tenure, adaptive practice, brokerage roles.
Tie; tie-frequency between communities.
Cohesion of the communication network is dependent on a few strategic bridging ties initiated by migrant farmers.
Migrant farmers have larger networks, act as brokers between socially and spatial distant groups, and tend to apply pro-environmental agricultural methods.This can be conceptualized as socialecological memory.
Migrant farmers are potential agents of innovation and adaptive management.

Matouš et al. 2013
Explore the roles of social networks and extension networks for the adoption of resource conserving practices among Ethiopian farmers.

Implicit reference to resilience:
Farmers' decisions are influenced by other farmers and the nature of social relationships.Informal networks can play a larger role in the protection of natural resources than formal institutions. Independent: Influence of social networks on the adoption of agricultural techniques. Pipe: Networks as conduit for the exchange of information. Descriptive: Household survey and semi-structured interviews with local extension staff in three villages in three agro-ecological zones.

Econometric analysis.
Social relation: information exchange on resource conserving agricultural techniques.
Actors: households in the village and external actors (extension staff).
Scale: local / village + linkages to external actors.
Network: community information network.
Actor: wealth, ethnoreligious group, knowledge about resource conserving practices, geographic location.
Farmers living close to the village center, with big land, bigger overall networks, and same ethnicity are more likely to be included in the extension system.
Same religion and ethnicity between farmers and extension staff increases likelihood of learning from extension.
Farmers with larger networks are more likely to know an extension agent but those are receptive to extension are those who are less socially connected.

Matuschke and Qaim 2009
Examine the impact of social networks on the adoption of modern seed technologies among smallholder farmers in rural India.

Implicit reference to resilience:
Social networks impact the adoption of agricultural innovations and contribute to increased agricultural productivity and reduced poverty. Independent: Influence of social networks on social learning and technology adoption. Pipe: Networks as a conduit enabling interaction and knowledge exchange between farmers.

Descriptive:
Household interviews in seven villages in four rural districts.
Econometric modeling / analysis.Communication takes place along homophilious rather than along heterogeneous lines.
The behavior of members in the individual network has bigger impact as their characteristics.

Mazzucato and Niemeijer 2000
Explore the role of social institutions in guiding decisions regarding the use of technologies drawing on a case study from Eastern Burkina Faso.

Implicit reference to resilience:
In (not specified) Context: socioeconomic changes (increasing market integration).
Mixing of market and social aspects has led to changing use and composition of networks.
Networks extending beyond the village level are used to access a repertoire of resources and technologies.This has increased the ability to adapt to changes.
Technologies are more attractive if they help maintain existing networks (lending qualities of a technology).Intervention should not only focus on technologies but on how farmers' abilities to network can be increased.

Spielman et al. 2010
Examine how Ethiopian smallholders innovate, how their social networks contribute to innovation processes, and how they are influenced by policies and market factors.

Implicit reference to resilience:
The way how smallholders innovate and improve their livelihoods is under constant change.Innovation systems comprise actors linked by networks of knowledge and others exchanges. Independent: Influence of social networks on innovation adoption. Pipe: Networks as a conduit for the exchange of information, inputs, and credit within the innovation system.

Structurally explicit:
Household interviews, focus group discussions, and semistructured interviews with key actors in 10 case study locations.
Social relation: exchange of knowledge, inputs, credit, finance, price and market information.
Actors: actors of the innovation system: farmers, extension staff, private and public sector representatives.
Network: national innovation system.
Heterogeneous and integrated networks provide farmers with greater livelihood options.
Interconnected public organizations play a central role, while market and civil society actors play only a peripheral role.
Public extension and administration exert a strong influence over smallholder networks.By crowding out market-based and civil society actors the public sectors potentially limits beneficial innovation processes.Structure: size of subgroups.
Farmers are strongly influenced by their kin and friends.Having more relatives increases the likelihood of having a big network.
Information opportunities of men and women vary in terms of exposure to and control of information.Men tend to have larger networks than women.
Gender should be accounted for when investigating the determinants of social networks.

Thuo et al. 2014
Examine the effects of social network factors on information acquisition and adoption of new seed varieties among groundnut farmers in Uganda and Kenya.

Implicit reference to resilience:
Social networks play a key role for social learning, and hence the adoption of new agricultural practices.

Independent:
Influence of social networks on information acquisition and the adoption of new seeds.

Pipe
Networks as a conduit for the exchange of information. Descriptive: Household interviews in two study sites.
Social network size and type of ties (e.g.kinship) play a significant role in enhancing the adoption of natural resource management practices.
Adoption of new natural resource management practices is more frequent among farmers whose plot is located next to adopters ("learning by watching").
Distance between network members has a positive effect on the adoption of new technologies.External sources of information such as extension play a crucial role in enhancing the adoption of resource management practices.Individual: households.
Caste is a decisive factor determining the access to particular supportive networks.
The poor and powerless castes (those with poor access to political social networks) are dependent on their informal social networks.These networks are typically accessed via the involvement with NGOs.

Cassidy and Barnes 2012
Explore the relationship between household connectivity and household resilience to shocks such as illness, crop damage, and livestock diseases in a rural community in Botswana.

Explicit reference to resilience:
Social networks as a strategy of vulnerable rural households to deal with external and internal stresses and shocks and to increase their resilience. Independent: Influence Context: decline of resources, market integration, inmigration.
Informal social relations have become loose and changeable due to perceived ecological changes and the influx of people without a fishing culture background.
This has led to an erosion of mutual help, solidarity, and social exchange.However, few primary bonds such as family and clan networks have managed to survive and even flourish within new situations.

Lyle and Smith 2014
Explore the linkages between participation in community collective activities and access to adaptive support networks in a Peruvian highland community.

Implicit reference to resilience:
Adaptive Social networks are not a "capital" or "asset" that poor always can draw from.Rather networks are based on dynamic and negotiated transactions that cannot be mechanically stored or accumulated.
Networks are not necessarily available and free of charge but are based upon complex norms of reciprocity.
Cultivation of networks requires time, effort and money which the poor peasants lack.Instead, social networks tend to reinforce the existing differences.

Orchard et al. 2015
Assess the association between aquaculture, livelihoods, and social networks in coastal communities of North Vietnam.

Explicit reference to resilience:
Aspects of resilience reside in the social networks of natural resource dependent communities: They facilitate access to livelihood resources in order to respond to change and increase the ability of communities to self-organize.
Independent / Dependent: Influence of aquaculture on social networks.
Influence of social networks on the resilience of local communities.

Social capital:
The structure and function of social networks is a crucial aspect of social capital enabling people to act together, and to pursue shared benefits.
Structurally explicit: Household survey in three coastal communities.
Livelihood diversity index.
Social relation: communication about mangrove system related issues.
Scale: local / village + linkages to external actors.
Network: mangrove communication network.
Context: market integration.
Economic transition alters mangrove system governance through the increasing influence of market mechanisms on the structure of social networks.
Traditional dense social networks (bonding capital) are replaced by larger and less dense networks extending the village level (bridging capital).
By reducing redundancy and connectedness market integration negatively impacts the capacity of communities to buffer the loss of ties and to selforganize.This way, market integration impacts community resilience.

Explicit reference to resilience:
Migration alters social networks central to insitu conservation of agro-diversity and hence impacts social ecological resilience. Dependent: Influence of migration on social networks. Pipe: Livelihood networks connecting various groups through the exchange of information and influence.

Metaphorical:
Survey on land use and livelihood activities (including migration and social networks) in a highland region.
Social relation: kinship, social relations between livelihood groups (not specified).
Scale: (not specified) (not specified) Context: migration.Social networks of migration related livelihood groups are powerfully shaped through international and national migration, while at the same time supporting agrobiodiversity use and in-situ conservation.

Table 1 .
Categories applied for the literature review.

Table A1 .
3: Strand iii: Social support Intra-family exchanges are influenced by marital status of the migrant, the presence of children and parents in the household of origin, and by having siblings depart from it.The location of the spouse is of relevance as well.