Operationalizing integrated water resource management (IWRM) often involves decentralization of water management via community-based management (CBM). While attention has been given to the components leading to successful CBM, less is known about what factors motivate people's willingness to participate (WTP) in such programs. This study analyzed factors that influence household WTP in CBM in a transboundary watershed located where El Salvador, Guatemala, and Honduras converge – the Trifinio Region. Several variables were hypothesized to influence WTP: sense of community (SOC), dependence on water resources, level of concern for water resources, and socio-economic characteristics. In 2014, quantitative and qualitative data were collected from 62 households in five communities. Most respondents reported high levels of WTP in future CBM initiatives, and multivariate regression analysis revealed that SOC was the most important predictor of WTP, with wealth and perceptions of watershed management also statistically significant. Qualitative analyses revealed water availability was more concerning than water quality, and perceptions of inequitable access to water is an important constraint to developing CBM strategies. Taken together, these results suggest that enhancing SOC and relationships between local and regional levels of governance prior to establishing community-based projects would facilitate more success in implementing IWRM.

Introduction

Access to clean reliable water resources remains a challenge for many low and medium income (LMI) countries. Globally, water resources are degraded through a variety of human induced changes to ecosystems, including direct introduction of human and livestock waste, agricultural contaminants, and deforestation of surrounding vegetation. The latter affects both the quality and quantity of water resources, with deforestation polluting waterways through increased sediment levels and changing the timing and amount of available water (Douglas et al., 2007). These changes affect human well-being, with more than 800 thousand deaths from waterborne illnesses each year, mainly in young children (Prüss-Ustün et al., 2014). Illnesses and resources allocated to the collection of safe drinking water also have significant effects on household labor and economic productivity in LMI countries. In 2010, access to clean and reliable water was declared a basic human right by the United Nations, and innumerable efforts continue to be launched to achieve water access and security by agencies working in LMI countries.

One common approach to improve and conserve water resources in rural areas is by implementing integrated water resource management (IWRM) (Heathcote, 2009). The Global Water Partnership (2012) defined IWRM as a process, which ‘aims to ensure the coordinated development and management of water, land, and related resources by maximizing economic and social welfare without compromising the sustainability of vital environmental systems’. A central component of successful conversion of IWRM principles into practice focuses on a participatory approach, involving local stakeholders in a management capacity. Specifically, community-based management (CBM) is one strategy that can be nested within an integrated management approach where communities are tasked with managing resources locally, which ultimately tie into broader national or basin-level management objectives.

The theoretical reasoning behind CBM is straightforward in that local populations have a more vested interest in the sustainable use of natural resources than higher-level institutions that do not directly rely on those same resources. While this management approach can lead to positive conservation outcomes and ensure that needs of locals are successfully met (Heathcote, 2009), heterogeneity in their impacts and success remains. As a result, considerable efforts have been made to identify the community characteristics associated with successful CBM. Some of the most well recognized literature on this subject comes from Ostrom's (1990) seven design principles, which demonstrate commonalities among enduring self-governing common pool resource institutions. While informative and useful in describing what features are shared among successful long-term CBM institutions, this literature does not shed light on what factors motivate people's willingness to participate (WTP) in such decentralization programs.

Such information can inform efforts prior to the implementation of CBM strategies that enhance local participation and the likelihood of CBM and IWRM success, which is especially important when considering the many obstacles that can arise from failed attempts to establish or maintain such institutions. There is some information on factors that influence participation in CBM in the context of fisheries (Zanetell & Knuth, 2004; Cavalcanti et al., 2010) and forestry management (Coulibaly-Lingani et al., 2011; Dyer et al., 2014). However, it is unknown whether the same factors apply to water resources management. One difference may be that forest and fishery resources are tangible entities that can be counted and collected by users, whereas water resources may be more difficult to observe and form perceptions on degradation or management options.

Therefore, the objective of this study was to assess the factors that influence rural residents WTP in CBM of water resources in a transboundary conservation area in Central America – the Trifinio Region. In the Trifinio Region, IWRM is currently being adopted to safeguard against further water resources degradation (Buch et al., 2009; Green & Daoust, 2012). Within Trifinio, water has been declared a public good; and the implementation of IWRM is in its infancy, which provides the opportunity to assess important household factors that contribute to WTP at the community level before IWRM programs are fully realized. More broadly, this study contributes to the small empirical literature assessing the drivers behind local participation in CBM projects for water resources, providing guidance on what steps can be taken prior to implementation of CBM to encourage more participation and enhance success.

CBM

Decentralization of natural resource management via CBM has been popular in forest, rangeland, and fishery resources in LMI countries since the 1990s, but is still relatively new within watershed management (Heathcote, 2009). In general, such studies suggest that social bonds within a community play a significant role in shaping individual participation in fisheries and forest management – with stronger bonds leading to more participation. In part, this is because rural communities often have an inherent reliance on, and bond between, community members. Additionally, the importance of trust in and among community actors cannot be overstated in the pursuit of collaborative programs or campaigns (Ostrom, 1990). Thus, these connections and relationships are often essential to many forms of collaborative pursuits within communities.

Dependence on natural resources has also been shown to influence participation in CBM. This is primarily because resource dependency often enhances local players' motivation to care for their resources because they directly rely on them for well-being and survival. Generally, the greater the dependence on a given resource the more likely individuals are to participate in management activities designed to protect those same resources (Lise, 2000; Zanetell & Knuth, 2004; Coulibaly-Lingani et al., 2011). In the case of water resources, this dependence may be less tangible than for resource products such as fisheries or forests.

Level of concern for a given resource can also impact participatory mechanisms. While intuitively a higher level of concern might seem to suggest more participation, Zanetell & Knuth (2004) found that a high level of concern for Venezuelan fisheries had a negative relationship with WTP in fisheries management. They attributed this result to a ‘defeatist attitude’, where people with high levels of concern believed that no intervention could solve the problems with local fisheries.

Finally, many studies cite the importance of socio-economic variables in participatory management (Lise, 2000; Dungumaro & Madulu, 2003; Coulibaly-Lingani et al., 2011). Wealth is frequently a source of heterogeneity among community members, which influences social dynamics, CBM, and programmatic intervention (e.g., Dungumaro & Madulu, 2003; Araral, 2009). Additionally, education level is a widely recognized factor that impacts participation (Lise, 2000; Coulibaly-Lingani et al., 2011); generally, as education levels increase, the more likely people are to participate in CBM. Finally, gender is an important indicator of participation because, in most places, men are considered the ‘decision-makers' leaving women to tend to household matters (Lise, 2000; Dungumaro & Madulu, 2003). This social inequity towards women is frequently cited within international research and campaigns regarding CBM, but may be different in a context like water resources management where women are often more involved with water collection and usage.

Methods

Study area

The Trifinio Region is located at the borders of Guatemala, Honduras, and El Salvador (Figure 1). The leaders of these three countries designated this tri-national region in 1987 as part of the Central American peace process. This transboundary region is the headwaters to three major river systems – the Lempa, Motagua, and Ulua Rivers – providing an estimated 4.5 million downstream with water (German Society for International Cooperation (GIZ), 2011; Green & Daoust, 2012). Because these river systems provide water not only to local communities, but also to major cities within these countries, their management and protection have consequences that span far beyond the Trifinio border.
Fig. 1.

Map of Trifinio and sampled communities.

Fig. 1.

Map of Trifinio and sampled communities.

Water and forest resources are under increasing stress in Trifinio due to population growth, economic activities, and land use change. An estimated 408 thousand rural residents in the region are living in poverty – per capita income less than $1.50 US per day (Nelson & Chomitz, 2004); while population growth rates are estimated at 1.6–3% per year (López, 2004). Agricultural production and tourism are the main economic activities in Trifinio (López, 2004). Agricultural activity occurs primarily through small, single-family plots of approximately five hectares (ha), and most of this land is devoted to subsistence crops – corn and beans – and coffee for commercial purposes (GIZ, 2011). As a result of conversions of forested land to agriculture and urban uses, the Trifinio Region lost approximately 19% of existing forest cover between 1986 and 2010 (Clemente et al., 2011).

Changes in water resources have also been documented. In drought years, the flow of the Lempa River has decreased by over 70% (Alvarado-Cruz, 2013). In addition, surface water quality has diminished throughout the region, primarily due to fecal coliform, agricultural and industrial contamination, and suspended sediment (Alvarado-Cruz, 2013). Today, historical ‘top-down’ management of water resources in Trifinio is slowly changing to a ‘bottom-up’ integrated management approach, with the creation of IWRM models listed as a primary objective for the Tri-National Water Agenda. This policy document discusses the need for synchronized management efforts among varying societal levels, including a focus on enhancing community involvement in managing water resources at the local level (Buch et al., 2009), which requires a better understanding of the perspectives, concerns, and knowledge of the rural populations.

This study was conducted in five communities, three in Guatemala (Figure 1): (1) Veguitas, (2) La Libertad, and (3) La Majada; and two communities in Honduras: (1) Sesesmil and (2) Nueva Estanzuela. Communities ranged from 30 households to 1,000 households. No communities were sampled in El Salvador due to time constraints. Communities were selected based on their accessibility and because they were priority areas for previous water, forest, and rural development programs in the region.

Survey instrument

A household survey was developed to gather information on household demographics, WTP, independent variables hypothesized to influence WTP, and two open-ended questions about perceptions of water resources and management. A bi-lingual, native Spanish speaker translated the survey instrument into Spanish. Each of the survey constructs and how they were operationalized, including the open-ended questions, is described in more detail below. Based on previous studies, four primary variables were initially hypothesized to influence individual WTP in CBM: (1) sense of community (SOC), (2) dependence on natural resources (DEPEND), (3) level of concern regarding water resources (CONCERN), and (4) socio-economic predictors (WEALTH, GENDER, EDUCATION). Table 1 provides a list of acronyms and measurement for WTP and primary variables.

Table 1.

Acronyms, descriptive statistics, and internal reliability for WTP and regression covariates.

Acronym Variable Measurement Mean (Std Dev) Cronbach's αa 
WTP Willingness to participate 1-‘very unwilling’ to 5-‘very willing’ 4.02 (0.65) 0.79 
SOC Sense of community 1-‘strongly disagree’ to 5-‘strongly agree’ 4.23 (0.51) 0.85 
DEPEND Dependence on water resources 1-‘strongly disagree’ to 5-‘strongly agree’ 4.18 (0.55) 0.69 
CONCERN Level of concern for water resources 1-‘strongly disagree’ to 5-‘strongly agree’ 3.63 (0.76) 0.36 
MANAGEMENTb Perceptions of current water resource management 1- indicates a 4 and 5 responses, 0- indicates a 3, 2, or 1 responses 0.53 (0.50) – 
WEALTH1 First tier wealth items Owned asset mean: 1-‘owns item’, 0-‘doesn't own item’ 0.15 (0.26) 0.79 
WEALTH2 Second tier wealth items  0.76 (0.26) 0.74 
GENDER Reported gender 1-‘male’, 0-‘female’ 0.61 (0.49) – 
EDUCATION Level of education 0-‘no school’, 1-‘some primary’, 2-‘complete primary’, 3-‘some secondary’, 4-‘complete secondary’, 5-‘some college’ 1.15 (1.15) – 
LAND Mean farm plot size Mean of two reported farming parcels in hectares 3.43 (6.85) – 
CDs Community dummy variables 1- indicates community response, 0- indicates non-community response – – 
Acronym Variable Measurement Mean (Std Dev) Cronbach's αa 
WTP Willingness to participate 1-‘very unwilling’ to 5-‘very willing’ 4.02 (0.65) 0.79 
SOC Sense of community 1-‘strongly disagree’ to 5-‘strongly agree’ 4.23 (0.51) 0.85 
DEPEND Dependence on water resources 1-‘strongly disagree’ to 5-‘strongly agree’ 4.18 (0.55) 0.69 
CONCERN Level of concern for water resources 1-‘strongly disagree’ to 5-‘strongly agree’ 3.63 (0.76) 0.36 
MANAGEMENTb Perceptions of current water resource management 1- indicates a 4 and 5 responses, 0- indicates a 3, 2, or 1 responses 0.53 (0.50) – 
WEALTH1 First tier wealth items Owned asset mean: 1-‘owns item’, 0-‘doesn't own item’ 0.15 (0.26) 0.79 
WEALTH2 Second tier wealth items  0.76 (0.26) 0.74 
GENDER Reported gender 1-‘male’, 0-‘female’ 0.61 (0.49) – 
EDUCATION Level of education 0-‘no school’, 1-‘some primary’, 2-‘complete primary’, 3-‘some secondary’, 4-‘complete secondary’, 5-‘some college’ 1.15 (1.15) – 
LAND Mean farm plot size Mean of two reported farming parcels in hectares 3.43 (6.85) – 
CDs Community dummy variables 1- indicates community response, 0- indicates non-community response – – 

aInternal reliability of scaled data.

bOriginally a part of the CONCERN scale.

Dependent variable: willingness to participate

WTP was defined as an individual's interest and disposition towards contributing to future IWRM initiatives at the community level through a CBM mechanism. It was measured using a hypothetical scenario designed to determine whether respondents would be willing to participate in CBM, which was based on the narrative used in Zanetell & Knuth (2004) study. The hypothetical scenario was framed by asking respondents to imagine there was a program to empower his/her community to design and implement a plan for managing water resources. Furthermore, the respondents were informed that in this hypothetical program it would be the community's responsibility to take charge of the plan, to implement changes, and to monitor and enforce these changes. After such framing, nine separate statements about participation in this hypothetical program were posed and respondents answered using a five-point Likert scale, which ranged from not willing to very willing. These nine items related to the respondents' individual interest in participating, taking a leadership role, ability/interest in changing water collection routines, and working collaboratively within and outside the community; it also included two items related to respondents' perceptions of other community members' interest in participating in a CBM program.

Independent variable: sense of community

Community as a concept is multi-faceted and complex, linking together many social and place-based phenomena. Socially, ‘community’ incorporates the shared history of its residents and their overall cohesiveness, producing sentiment about the location and its people. Personally, ‘community’ can include individual connections to place that generate contentment or other emotions that are bound by location. In the field of community psychology, this is often referred to as SOC (Sarason, 1974). Thus, for this study, we defined SOC to include both personal connections and social characteristics of communities using the following definition: the inherent level of connection, trust, belongingness, and cohesiveness established through living and working in shared physical spaces.

We utilized an adaptation of Buckner (1988) 18-item neighborhood cohesion scale, using a five-point Likert scale ranging from strongly disagree to strongly agree. Originally, Buckner's scale was designed to evaluate neighborhood SOC in Western settings. The scale measures individual SOC, but can be extrapolated to the community level, which allows researchers to deduce community cohesiveness. Zanetell & Knuth (2004) successfully applied the scale in Venezuelan fishing-based communities, and suggested these communities were comparable to a neighborhood because residents interact with nearly everyone on a daily basis within the physical boundaries of the community. A similar logic holds for the communities sampled in this study.

Independent variable: dependence on water resources

Resource ‘dependence’ contains both functional and emotional components. For this study both components were drawn upon through five Likert statements. From the functional perspective, respondents were asked about their reliance on local water sources for income, food, and health. Emotional dependence was captured using two statements: (1) it is important for you to live near water sources and (2) living near water contributes to your personal happiness. These statements were measured using the same five-point scale as SOC.

Independent variable: level of concern

For this study, CONCERN was defined as the reported level of concern regarding the current state of water resources within a community. To capture this construct, we used three statements, which focused on perceptions regarding current: (1) management practices, (2) water protection levels, and (3) water quantity levels. Respondents answered using the same five-point Likert scale as SOC.

Independent variables: socio-economic information

Socio-economic information relevant to WTP, along with additional household information regarding livelihood strategies, were gathered via a short series of questions. GENDER was measured as a binary variable, and EDUCATION as a five-level categorical variable. WEALTH was assessed using binary responses to a list of household assets. The asset list included indicators of high wealth, moderate wealth, and low wealth.

Preferred level of management

Although not hypothesized to impact WTP, the preferred level of water resource management – family, community, or government – was also included within our survey to inform policy recommendations. This was assessed via three, five-point Likert items ranging from strongly disagree to strongly agree.

Open-ended questions

The two qualitative questions focused on household perceptions of water resources in their community and their management. The first question asked about changes to water resources in the past 10 years; and the second inquired about existing water management protocols in the community.

Survey sampling and implementation

Within the five communities, systematic random sampling was utilized. Sampling began with the household of the community leader, after which enumerators selected every third household. If the head of household was unavailable, enumerators went to the subsequent household. Only two households refused to be surveyed; thus, non-response bias is not a concern for this study. Enumerators introduced the study and determined the head of household by a series of scoping questions about farm, water, and forest resource ownership. Based on responses to these questions, enumerators decided whether to proceed with the questionnaire. In total, 62 households were surveyed.

Surveys were conducted verbally with native Spanish speakers. Enumerators were hired from the Tropical Agricultural Research and Higher Education Center's (Spanish acronym CATIE) Trifinio field office and were accompanied by the research team. Before survey implementation, half-day enumerator training and survey pilot tests were conducted.

Data analysis

Quantitative analysis

Quantitative analyses were conducting using IBM's SPSS 22 (IBM Corp., 2013). Likert scales WTP, SOC, CONCERN, and DEPEND were analyzed as interval type data because they each: (1) are comprised of multiple Likert type items (Carifio & Perla, 2008; Boone & Boone, 2012) and (2) meet normality, homogeneity of variance, and independence assumptions required of parametric analyses (Field, 2009). Cronbach's alpha1 determined the internal consistency reliability of scaled variables. Exploratory factor analysis2 (EFA) was conducted on the WEALTH items to determine approximations of wealth levels. To test for multicollinearity among independent variables, bivariate Pearson correlations3 were conducted. To analyze the effect of each independent variable on WTP, the following linear multivariate cross-sectional regression model was estimated: 
formula
1
where CDs are community dummy variables used to control for community-level heterogeneity. Standardized beta coefficients were used, robust standard errors were estimated, and adjusted R2 values were reported.

Qualitative analysis

Qualitative information was analyzed using directed content analysis (Berg & Lune, 2012). Themes were identified both a priori and through summative content analysis, which counts the number of repetitions of emergent themes. Furthermore, linguistic connectors were utilized to understand both casual relationships and conditional relationships among themes (Ryan & Bernard, 2003).

Validity and reliability

Our small sample size and lack of El Salvadorian communities limits the generalizability of these results. Additionally, the five communities included in this study were chosen based on accessibility and presence of previous environmentally-focused projects and therefore may represent an above average WTP. These elements impact this study's external validity; however, the sampled communities have similar livelihood characteristics to other studies conducted in Trifinio, and systematic random sampling was utilized, which affords a suitable level of randomization within communities. Internal validity was maintained through construct validity of pre-designed scales and through the systematic random sampling design. Additionally, internal consistency reliability of the survey instrument was tested and maintained using Cronbach's alpha.

Results

Quantitative results

Household characteristics

Most respondents surveyed were male, between 30 and 45 years of age, and had little to no formal education (Table 2). Over half of the surveyed households reported farm related activities as their predominant source of income. The primary subsistence crops planted were corn and beans. The 62 respondents reported having a total of 104 farm parcels (Table 2). When compared more broadly to the population of Trifinio, these communities are similar in make-up based on their livelihood strategies, typical farm plot size, and number of people per household (GIZ, 2011). However, the level of education in sampled communities was lower than average for the region (GIZ, 2011), which may be related to the rural nature of the communities sampled.

Table 2.

Summary of household socio-demographic and farm characteristics.

Characteristic Entire samplea Guatemala Honduras 
Socio-demographic characteristics 
Gender 
 Male 61% 48% 73% 
 Female 39% 52% 27% 
Education level 
 Incomplete Primary 78% 72% 82% 
 Complete Primary 16% 28% 6% 
 Complete Secondary 1% 3% 
 Incomplete University 5% 9% 
 Observations 62 29 33 
Farm parcels 
Subsistence crops (# of plot)b 
 Beans 63 31 32 
 Corn 64 30 34 
 Coffee 12 10 
 Plantains 
 Mean size of primary plot (ha) 4.08 0.89 7.13 
 Mean size of secondary plot (ha) 3.11 0.56 4.85 
 Observations 104 45 59 
Characteristic Entire samplea Guatemala Honduras 
Socio-demographic characteristics 
Gender 
 Male 61% 48% 73% 
 Female 39% 52% 27% 
Education level 
 Incomplete Primary 78% 72% 82% 
 Complete Primary 16% 28% 6% 
 Complete Secondary 1% 3% 
 Incomplete University 5% 9% 
 Observations 62 29 33 
Farm parcels 
Subsistence crops (# of plot)b 
 Beans 63 31 32 
 Corn 64 30 34 
 Coffee 12 10 
 Plantains 
 Mean size of primary plot (ha) 4.08 0.89 7.13 
 Mean size of secondary plot (ha) 3.11 0.56 4.85 
 Observations 104 45 59 

aThe entire sample includes the 29 households in Guatemala and the 33 households in Honduras.

bSome plots contain multiple crop types.

Sesesmil was the wealthiest community in our sample, based on the mean farm plot size and the highest community owned asset mean (Table 3). Veguitas and La Libertad were in the middle of the wealth spectrum, having similar mean farm plot sizes, and similar owned asset means. La Majada and Nueva Estanzuela represented the lower end of the wealth spectrum. Nueva Estanzuela had a higher mean farm plot size, but La Majada had a greater owned asset mean.

Table 3.

Community-level wealth summaries and WTP.

Characteristic Veguitas La Libertad La Majada Sesesmil Nueva Estanzuela 
Mean size primary farm plot (ha) 1.12 1.05 0.54 10.51 0.72 
Mean of asset quantitya 0.43 0.47 0.37 0.60 0.26 
Mean WTPb 3.80 4.55 4.00 3.75 4.35 
Characteristic Veguitas La Libertad La Majada Sesesmil Nueva Estanzuela 
Mean size primary farm plot (ha) 1.12 1.05 0.54 10.51 0.72 
Mean of asset quantitya 0.43 0.47 0.37 0.60 0.26 
Mean WTPb 3.80 4.55 4.00 3.75 4.35 

aAverage quantity of assets owned out of 18 items.

bLikert scale ranging from one to five.

Exploratory factor analysis

Our EFA met all of the established criteria discussed previously. The EFA of WEALTH items revealed three separate wealth indicators, which are adopted in future analyses (Table 4). The pattern component matrix revealed that WEALTH1 loaded high status symbols while WEALTH2 is comprised of tier-two assets, which included items that most households have. WEALTH3 contained two items related to water and sanitation.

Table 4.

The pattern matrix of the exploratory factor analysis on WEALTH.

Wealth indicator WEALTH1 (3.824) WEALTH2 (1.737) WEALTH3 (1.465) 
Vehicle 0.888   
Motorcycle 0.757   
Horses 0.732   
Chainsaw 0.671   
Internet 0.636   
Electricity  0.835  
Cellphone  0.784  
TV  0.724  
Latrine   0.767 
Septic   0.754 
Wealth indicator WEALTH1 (3.824) WEALTH2 (1.737) WEALTH3 (1.465) 
Vehicle 0.888   
Motorcycle 0.757   
Horses 0.732   
Chainsaw 0.671   
Internet 0.636   
Electricity  0.835  
Cellphone  0.784  
TV  0.724  
Latrine   0.767 
Septic   0.754 

Note: n = 62; KMO statistic = 0.57; Bartlett's test of sphericity: X2 = 266.09, df = 91, p <0.01; Direct oblimin rotation, delta = 0; Eigen values reported in parentheses adjacent to each WEALTH factor.

Cronbach's alpha and Pearson correlations

Cronbach's alpha revealed WTP, DEPEND, SOC, WEALTH1, and WEALTH2 all had acceptable internal consistency reliability (Table 1). However, WEALTH3 and CONCERN had low internal consistency reliability, and, as complete variables, were excluded from further analyses. Instead, the three individual items from CONCERN were transformed into binary responses and included within Equation (1) independently to determine if individual components of this variable influenced WTP. This transformation of a Likert statement to a dichotomous variable is appropriate when there is a bimodal distribution with a clear distinction between positive and negative perceptions, with few neutral responses4. Thus, in the final regression analyses, CONCERN is captured by one binary variable: perceptions of current water resource management practices (MANAGEMENT). A Pearson correlation matrix revealed no collinearity problems among independent variables.

Descriptive statistics on WTP variables

Approximately 60% of respondents were strongly willing to participate in future CBM initiatives (scores ≥4), 32% of the respondents felt mildly willing or neutral (scores 3–3.9), while 8% were not willing to participate (scores < 3). The mean WTP value was around four (Table 1). Interestingly, stated WTP varies across communities (Table 3). Mean WTP scores revealed that three of the five communities are strongly willing to participate, with Sesesmil and Veguitas – two of the wealthiest communities – only mildly willing. Utilizing an ANOVA procedure, we determined that there is a statistically significant difference (p < 0.05) between Sesesmil and the two communities with the highest reported WTP (La Libertad and Nueva Estanzuela).

SOC and DEPEND both had mean scores greater than four (Table 1), indicating high social and personal connections within communities and high levels of water resource dependence. The mean value for MANAGEMENT was 0.53, indicating a split disposition towards the quality of current water resource management practices. WEALTH1 had a mean score of 0.15 out of one. This low value reflects that most people within this sample did not own any of these high-dollar items. In contrast, over three quarters of respondents owned items in the WEALTH2 category.

Regression output

The full regression model (i.e., Equation (1)), explained 50% of the variation in this sample (Table 5). Specifically, SOC was positive and statistically significant (p < 0.01), indicating that higher SOC lead to higher WTP. WEALTH1 was also statistically significant (p < 0.01), with wealth positively correlated with WTP. Additionally, one of the original CONCERN items, perceptions of current water resource management (i.e., MANAGEMENT), had a negative impact on WTP (p < 0.05). This suggests that respondents who viewed their water resources as being poorly managed were more willing to participate. All CDs were statistically significant5. DEPEND, WEATLH2, GENDER, and EDUCATION were not statistically significant and did not contribute to the overall explained variance in WTP.

Table 5.

Regression results.

Predictors Full model Parsimonious model Parsimonious model substituting LAND for WEALTH 
SOC 0.49*** (0.16) 0.49*** (0.13) 0.50*** (0.14) 
DEPEND 0.017 (0.13) – – 
MANAGEMENT −0.25** (0.14) −0.23** (0.13) −0.18* (0.14) 
WEALTH1 0.50*** (0.39) 0.55*** (0.34) – 
WEALTH2 0.08 (0.17) – – 
LAND – – 0.28** (0.01) 
GENDER −0.03 (0.16) – – 
EDUCATION 0.08 (0.07) – – 
CD1 (Veguitas) 0.28** (0.26) 0.30** (0.24) 0.12 (0.22) 
CD2 (La Libertad) 0.66*** (0.25) 0.66*** (0.24) 0.47*** (0.22) 
CD3 (La Majada) 0.62*** (0.24) 0.65*** (0.22) 0.42*** (0.20) 
CD4 (Nueva Estanzuela) 0.61*** (0.24) 0.60*** (0.23) 0.39*** (0.21) 
Adjusted R2 0.501*** (0.46) 0.526*** (0.45) 0.453*** (0.49) 
Predictors Full model Parsimonious model Parsimonious model substituting LAND for WEALTH 
SOC 0.49*** (0.16) 0.49*** (0.13) 0.50*** (0.14) 
DEPEND 0.017 (0.13) – – 
MANAGEMENT −0.25** (0.14) −0.23** (0.13) −0.18* (0.14) 
WEALTH1 0.50*** (0.39) 0.55*** (0.34) – 
WEALTH2 0.08 (0.17) – – 
LAND – – 0.28** (0.01) 
GENDER −0.03 (0.16) – – 
EDUCATION 0.08 (0.07) – – 
CD1 (Veguitas) 0.28** (0.26) 0.30** (0.24) 0.12 (0.22) 
CD2 (La Libertad) 0.66*** (0.25) 0.66*** (0.24) 0.47*** (0.22) 
CD3 (La Majada) 0.62*** (0.24) 0.65*** (0.22) 0.42*** (0.20) 
CD4 (Nueva Estanzuela) 0.61*** (0.24) 0.60*** (0.23) 0.39*** (0.21) 
Adjusted R2 0.501*** (0.46) 0.526*** (0.45) 0.453*** (0.49) 

*p ≤ 0.10, **p ≤ 0.05, ***p ≤ 0.01; n = 62.

Note: Standardized beta coefficients and robust standard errors reported.

A more parsimonious model that includes only covariates that were statistically significant in the full model – SOC, MANAGEMENT, WEALTH1 and CDs – explained 52.6% of the variation in this sample (Table 5). The magnitude and significance of the covariates in the parsimonious model are similar to those in the full model.

As a robustness check on our wealth variable we substituted a measure of the mean size of land parcels (LAND) for WEALTH1 (Table 5). This substitution did not meaningfully alter the results within our regression model – both WEALTH1 and LAND were statistically significant with positive effects on WTP. However, the explained variance dropped to 45.3% when LAND was included instead of WEALTH1.

Management preferences

When asked who should manage water resources, respondents favorably supported household, community, and government-level management institutions. However, respondents preferred management at the family or community level, with mean scores of 4.3 and 4.5, respectively, while government-level management was the least preferred, with a mean score of 3.4.

Qualitative results

Nearly three quarters of this sample reported that there had been a change in water quantity in the last 10 years for their primary water source. Most respondents (52%) mentioned water scarcity as the observed change. Qualitative analysis revealed that respondents who discussed water quantity issues commonly linked increases in population (24%) and/or deforestation (34%) to their observed changes. For example, one respondent stated, ‘the water source is not enough for the future population, and also the water sources are being deforested so the community faces drought’. Only 11% of respondents reported improvements in the amount of available water, primarily due to the installation of pumps within homes.

A smaller subset of respondents (21%) identified changes in water quality as a concern over the last 10 years. Most respondents who reported water quality issues linked it to health concerns. In this context, water was referred to as ‘contaminated;’ yet most respondents did not identify a cause of contamination. The three respondents who attributed contamination to a specific source reported that mining, agriculture, and ‘wastes' affected water quality. Additionally, some improvements in water systems occurred in these communities; approximately 30% of households noted that they had piped water and 68% had latrine access.

Parsed by community, 75% or more of respondents from four of the five communities reported water scarcity as a major concern (Figure 2); while in Sesesmil, 52% of the respondents acknowledged water scarcity as a primary concern. This may be due to the fact that Sesesmil is the wealthiest community in our sample, and more of the families in this community had access to private wells and pumped water. Furthermore, when looking at water quality concerns, the two poorer communities in our sample – Nueva Estanzuela and La Majada – had the most respondents reporting concern (Figure 2). One woman from Nueva Estanzuela reported that as a community, ‘we do not have options to get better water quality. We do not know what to do.’ Conversely, one respondent from Sesesmil reported, ‘I have to protect my springs. It's other's business to do as they see fit.’
Fig. 2.

Concern for water scarcity and quality by community. Sourced from qualitative responses; percentages are based on responses per community.

Fig. 2.

Concern for water scarcity and quality by community. Sourced from qualitative responses; percentages are based on responses per community.

In terms of water resource management, respondents within all communities reported the presence of a water management committee administered by local community leaders or the local government (i.e., the municipal government in Trifinio). Of the respondents that mentioned local committees (n = 38), 16% reported some degree of mismanagement or corruption on these boards. This theme was consistent across all five communities. Additionally, several respondents stated access to water was unfair despite having a local water committee.

Discussion

Factors associated with willingness to participate in CBM

SOC

Based on its predictive power in our regression models, findings from this study indicate that SOC is one of the most important predictors of WTP in CBM. The importance of social connections in conservation and CBM is well recognized (e.g., Zanetell & Knuth, 2004; Gruber, 2010), but researchers, policy-makers, and development institutions should be cautious in accepting the idea of ‘community’ as a panacea to natural resource conservation practices (Agrawal & Gibson, 1999). Simply recognizing the importance social dynamics play in conservation is no longer sufficient, yet applying such information is easier said than done, especially when considering how to best enhance, create, or extend SOC to the broader regional scope.

Threats to SOC can come from inequity, lack of trust, or generally weak social relationships within communities, greatly diminishing likelihood of CBM success. Results from Dyer et al. (2014) support that communities that lacked trust and consensus led to a reduction in successful participation outcomes in community-based forestry management in Southern Africa. In addition, Ravnborg (2008) study corroborates that inequity of forest resource access in Nicaragua arose from a lack of fair monitoring due to corrupt management. Within our study, qualitative references to corruption in existing water management boards were reported from a small subset of respondents. This suggests that these issues need to be identified and addressed before broader IWRM approaches are implemented in Trifinio to avoid concerns of inequality or corruption undermining any new CBM initiatives.

Current water management

Household perceptions regarding water resource management (MANAGEMENT) were also a statistically significant predictor of WTP – respondents were more willing to participate in CBM when they reported their water resources were not managed well. Intuitively this seems reasonable; an individual may be motivated to participate in CBM if current management systems are not meeting their needs. In this case, individuals' entry into such management models may be requisite upon a negative perception of resource quality or availability to provoke their initial buy-in.

However, our results vary from those of Zanetell & Knuth (2004), who found that a defeatist attitude caused respondents to be less willing to participate if their concerns for the fishery were high. Thus, there may be a threshold where high levels of concern about management of a resource will begin to detract from WTP – stemming from a belief that change in policy and/or management is ‘hopeless' versus still possible. More research on this relationship between management perceptions and WTP could shed insight on whether a threshold exists and why.

Dependence on water resources

While place dependence (DEPEND) is often cited as an important component of participatory management (Lise, 2000; Coulibaly-Lingani et al., 2011), it was not a significant factor in predicting WTP within this study. One explanation is that in our study area there was little variation reported in dependence on water resources, with most households reporting high dependence – mean DEPEND scores were greater than four for each community. This high dependence is further supported by our qualitative results, where 52% of respondents reported water availability was their primary water resource concern.

An alternative explanation is that DEPEND was mismeasured in our study because most crops in Trifinio are rain fed, yet two items in the DEPEND scale were specific to dependence on water resources for income and food, with values for these statements averaging to less than four. Thus, we may not have captured the desired functional dimension of DEPEND we sought. Future research implementing the use of a water dependence variable should try to incorporate questions more specific to the region of study.

Socio-economic factors

In many other CBM studies, GENDER and EDUCATION affect participation. For example, Lise (2000) found higher levels of education enhanced participation in joint forestry management in India because respondents were more aware of local forest resource issues. It follows that environmental issue awareness may be a result of education. In our case, EDUCATION may not be significant due to little variation in education background, with most respondents not completing primary school. GENDER was also not significant; this is surprising given the more active roles women play in water collection, and a potentially greater awareness of water resource degradation. However, in our sample, women were no more likely than men to be willing to participate.

WEALTH1, which included only high-value assets, had a positive relationship with WTP, but WEALTH2 was not statistically significant. This relationship between WTP and WEALTH1 suggests that when higher value items are owned, household levels of WTP increase. Yet, in many CBM studies higher levels of wealth lead to lower levels of participation (e.g., Dungumaro & Madulu, 2003; Araral, 2009). In these cases, wealth had a negative relationship with participation in CBM practices because wealthier individuals could remove themselves from participation – they may already own a well or forest plot – and do not need to contribute to a collectively-managed resource. However, Lise (2000) found that results concerning wealth's contribution in joint forestry management in India were mixed. In one community, wealth negatively affected participation, but in another, wealth positively impacted the likelihood of participation. In our sample, only three respondents outside of the Sesesmil community possessed any of the WEALTH1 assets. When we dropped these three respondents from regression models we found the results in Table 5 unchanged, meaning that the variation in WEALTH1 represents a within-community variation in household wealth in Sesesmil, not variation at the household level across all five communities.

Community dummy variables

Community-level differences played an important role in explaining WTP in this study (Table 5). Community-level factors related to WTP could be related to SOC, perceptions of water resources management, or previous participation in rural development or CBM programs. Based on the large differences in wealth across communities, we suspect that community wealth is partly responsible for the statistically significant and positive impact that CDs had on WTP – relative to the wealthiest community, all other communities are more likely to participate. Our qualitative data support this conclusion, with wealthier respondents preferring privatization of water access. Such preference for individual management of water resources occurred only within Sesesmil, reflecting findings from other CBM studies where with high-wealth levels a household can substitute purchased resources for degraded resources and thus have little need to participate in CBM.

In light of these community-level differences, and variation in the WEALTH1 variable described above, the covariate WEALTH1 in Table 5 can be interpreted as how wealth across households in Sesesmil is related to WTP, where the wealthiest respondents appeared more likely to participate than the moderately wealthy. This is interesting but not something we can tease out completely in this study. It suggests that the true relationship between wealth and WTP may be non-linear and more similar to a U-shaped curve, with participation rates decreasing as wealth levels increase, due to substitution of resources, but turning upwards again after a certain threshold of wealth is passed.

Overall, the CDs seem to indicate more about the relationship between wealth and WTP across communities than our intended measure using assets at the household level; this is because our sample has more heterogeneity in assets – and land resources as indicated by farm plot size (Table 3) – across communities than within communities at the household level.

Perceptions of changes in water resources

One emergent theme from qualitative analysis was that water quantity issues were more frequently mentioned than water quality issues. Furthermore, it was far more likely for respondents to link a cause to water quantity issues (e.g., deforestation, population increases). These results suggest that communities were more concerned about water scarcity than water quality, or were more likely to perceive these types of changes. Changes in timing and availability of water impact residents in the Trifinio Region and there is evidence to support that deforestation is one of the drivers of these changes (Nelson & Chomitz, 2004).

Respondents who identified water availability as a primary concern also tended to include an observation of such trends, often reported as decreases in the amount of rainfall or changes in rivers, wells, and springs. Yet, when water quality was listed as a concern, the descriptions were often more vague, indicating ‘contamination’ as the culprit. This may suggest that because water quantity decreases are more tangible they are reported more often. The causes of water quality do not appear well understood among respondents, but health-related consequences from contamination are frequently reported. Overall, variation exists in access to consistent and clean water resources in our sample, and should be viewed as an important source of heterogeneity among communities when determining strategies for IWRM and CBM implementation and for targeting education and outreach strategies.

Policy implications for CBM

Taken together, the quantitative and qualitative results from this study point toward three steps that could be taken prior to the implementation of IWRM to increase the likelihood of successful CBM projects: (1) strengthen SOC within communities and among communities; (2) start with small-scale education focused campaigns specific to local resource issues and conservation; and (3) strengthen relationships between communities and intermediate-level governance.

First, despite the common findings that social connections and trust are important in successful CBM programs, Zanetell & Knuth (2004) affirmed that further knowledge on how to foment community cohesion is required. This recommendation is echoed by Pronyk et al. (2008), who studied whether social capital could be generated in rural South Africa in the context of microfinance loans paired with participatory HIV training for women. Throughout the 2-year period that groups of women worked together, social capital between the members of the various community groups increased; this held true between groups from different communities as well. These results indicate that ‘social capital can be intentionally generated in relatively short programmatic time frames' (Pronyk et al., 2008, p. 1567). However, when considering how to scale up such efforts to a regional scale, the problem becomes more complex.

Flint et al. (2010) argued that the conceptions of ‘community’ and ‘region’ are a process and not simply a product or outcome, and referred to them as ‘fields’. The development of a community field concept ‘emphasizes dynamic, emergent processes of interaction’ (Flint et al., 2010, p. 25). Community fields are comprised of many social fields – social connections, shared awareness, and history – making them an ideal place to start when considering how to scale up social bonds. This raises an interesting epistemological question when establishing CBM: how do we best utilize existing social fields within communities to create a broader regional field?

One suggestion is to make use of existing common themes within existing social fields and work toward scaling up this shared identity (Theodori, 2005). Thus, the process of determining these shared points of identity is paramount when considering how to generate or broaden community into a regional scale. For this study, one clear point of shared identity is local concern regarding water scarcity. Additionally, it would be prudent to consider within which existing community fields would the likelihood of mobilization occur. This direction would require an understanding of the scope of collective community norms, history of shared endeavors and their successes and failures.

Furthermore, results from this study suggest that people are more willing to participate when they believe their water resources are not being managed well, which implies that investment in education around water resources degradation could be useful. Qualitative results indicate that people were most concerned about the amount of available water. Green & Daoust (2012) suggested that for transboundary integrated management projects it is pragmatic to pursue focused objectives with narrow scopes, intentionally crafted to build on successes. In the case of the Trifinio Region, future programmatic interventions could concentrate on educating the populace on water scarcity and its causes, and secondarily, on education about regional-level problems related to water quality, given the higher percentage of respondents that readily identify with water quantity issues.

Finally, restructuring of resource governance should include strengthening of intermediate levels of government (i.e., the municipality level in Trifinio) and the relationship between these levels of government and local communities. This can provide the necessary oversight and assistance to support long-term CBM practices. Evans (1997) indicated that social connections may be enhanced via institutional partnerships and by working at multiple levels simultaneously in a nested institutional structure.

Within our study, several respondents did not view current management of water resources as sufficient (i.e., MANAGEMENT). It is difficult for high-level government to provide the necessary support to local-level management institutions. Instead, efforts to enhance intermediate-level water management could influence how successful CBM institutions are in the long term. This is because they can provide the necessary support in monitoring water resource utilization, and ensure equitable establishment, or re-structuring of management institutions to include all relevant stakeholders in the decision-making processes related to water resources. Both principles are recognized as important components in successful CBM regimes (e.g., Ostrom, 1990; Dyer et al., 2014). This notion of enhancing intermediary levels of government is also supported by Berkes (2007), who stated ‘community-based conservation as a panacea, like government-based conservation as a panacea, ignores the necessity of managing commons at multiple levels, with vertical and horizontal interplay among institutions' (p. 15188).

This necessity of effectively nesting governance is further supported by our qualitative analysis of current water resources management practices, which identified perceptions of corruption and inequality in some existing community management boards. This is despite the fact that the majority of respondents felt that the community was the appropriate level for managing water resources, indicating that there was also some reluctance or distrust in governmental management. Furthermore, results indicate that government was the least preferred option in managing water resources, when compared to community and family management options. Thus, strengthening trust between communities and intermediate levels of governance via small-scale, water scarcity focused initiatives may be the best way to build ‘community’ more broadly in this region.

Conclusions

Providing communities with the ability to collectively manage their natural resources should, theoretically, increase the likelihood that those resources will be utilized in a sustainable manner and the needs of people are met (Heathcote, 2009). However, in practice, studies often find that simply creating CBM institutions does not lead to equitable access or sustainable uses of a given resource (e.g., Ostrom, 1990; Ravnborg, 2008; Araral, 2009). Addressing the underlying factors that motivate participation in CBM programs before they are launched is likely to lead to more effective and fair CBM practices in the long run because it can identify the conditions that need to be enhanced before engaging in CBM (Zanetell & Knuth, 2004).

This study's primary objective was to investigate the underlying factors that motivate WTP in CBM in the Trifinio Region, and to understand household perceptions of water resources and management. Results from this research contribute to the empirical literature on the factors that drive participation within CBM institutions. In the Trifinio Region, we found that social and personal connections within communities, perceptions of water resource management, and wealth all influence WTP in CBM. These findings contribute empirical evidence on WTP in CBM specific to water resources, an understudied area for WTP. Findings from this study corroborate many of the theoretical findings of past studies related to relationships between SOC and WTP, but also uncover some different and contradictory results on why rural residents are willing to participate in CBM, and consequentially IWRM. These results also point to future research directions in exploring non-linear relationships between dependence on water resources, wealth levels, and WTP.

Program recommendations for Trifinio policy-makers, and more broadly for the IWRM audience, include enhancing social connections at the community and intermediate levels of government through small-scale programs focused on specific local issues regarding water resources. Once established, these small-scale initiatives can build into an integrated management model that uses mid-level governance to convey the needs of IWRM to the transnational governance level. These results indicate that enhancing social connections in local communities and nesting CBM programmatic design into mid-level governance may enhance efforts to establish CBM within integrated management models.

Acknowledgements

This research would have not been possible without field support from CATIE's Trifinio field office. Additionally, we thank Carlos Muñoz-Brenes and Sam Dupree for their help with data collection. We appreciate guidance and comments on previous versions of this paper from Drs. Sanyal, K. Eitel, Putsche, and Paveglio. We acknowledge funding support from NASA's Land Cover Land Use Change Project, Grant Number NNX13AC70G.

1

A coefficient of >0.6 served as the cut-off point for reliability analysis.

2

Direct oblimin rotation was utilized; factors were extracted based on: (1) Eigen values >1, (2) Kaiser-Meyer-Olkin Measure of Sampling Adequacy (KMO) >0.5, and (3) item loadings >0.5.

3

Correlation coefficients between independent variables (i.e., >0.8) excluded both predictors in the same regression model.

4

When neutral responses were present they were included with the dichotomization of ones and twos.

5

The omitted CD was for the wealthiest community, Sesesmil, and relative to this community, all other communities have a higher WTP.

References

References
Alvarado-Cruz
A. P.
(
2013
).
Inventory of Information and Identification of Gaps for Characterization and Assessment of the Esquipulas-Ocotepeque-Citalá Aquifer
,
Report for the IUCN
,
Livelihoods and Climate Change Unit
,
Guatemala
.
Berg
B. L.
Lune
H.
(
2012
).
Qualitative Research Methods for the Social Sciences
.
8th edn
.
Pearson
,
Boston
,
USA
.
Berkes
F.
(
2007
).
Community-based conservation in a globalized world
.
Proceedings of the National Academy of Sciences of the United States of America
104
(
39
),
15188
15193
.
Boone
H. N.
Boone
D. A.
(
2012
).
Analyzing Likert data
.
Journal of Extension
50
(
2
),
1
5
.
Buch
M. S.
Hernández-Vela
O.
Jiménez
A.
Orellana
J. U.
(
2009
).
Agenda Hídrica Trinacional: Una Propuesta Participativa para la Gestión Integrada de los Recursos Hídricos en la Parte Alta de la Cuenca del Río Lempa (Trinational water agenda: A participatory proposal for integrated management of water resources in upper reaches of the Lempa River)
,
Report for the Comisión Trinacional del Plan Trifinio
,
San Salvador
,
El Salvador
.
Buckner
J. C.
(
1988
).
The development of an instrument to measure neighborhood cohesion
.
American Journal of Community Psychology
16
(
6
),
771
779
.
Clemente
A. H.
Hernandez
B. E.
Cherrington
E. A.
(
2011
).
Cobertura Uso de la Tierra la Región del Trifinio: Estudio de los Anos 1986, 2001 y 2010 Mediante Metodos de Teledetetección (Land use coverage of the Trifinio Region: Study of the years 1986, 2001 and 2010 using remote sensing methods)
.
Report for CATHALAC
,
Centro del Agua del Trópico Húmedo para América Latina y el Caribe
,
Panama
.
Coulibaly-Lingani
P.
Savadogo
P.
Tigabu
M.
Oden
P. C.
(
2011
).
Factors influencing people's participation in the forest management program in Burkina Faso, West Africa
.
Forest Policy and Economics
13
(
4
),
292
302
.
Douglas
E. M.
Wood
S.
Sebastian
K.
Charles
J. V.
Chomitz
K. M.
Tomich
T. P.
(
2007
).
Policy implications of a pan-tropic assessment of the simultaneous hydrological and biodiversity impacts of deforestation
.
Water Resource Management
21
,
211
232
.
Dungumaro
E. W.
Madulu
N. F.
(
2003
).
Public participation in integrated water resources management: the case of Tanzania
.
Physics and Chemistry of the Earth, Parts A/B/C
28
(
20–27
),
1009
1014
.
Dyer
J.
Stringer
L. C.
Dougill
A. L.
Leventon
J.
Nshimbi
M.
Chama
F.
Kafwifwi
A.
Muledi
J. I.
Kaumbu
J.
Falcao
M.
Muhorro
S.
Munyemba
F.
Kalaba
G. M.
Syampungani
S.
(
2014
).
Assessing participatory practices in community-based natural resource management: experiences in community engagement from southern Africa
.
Journal of Environmental Management
137
,
137
145
.
Evans
P.
(ed.) (
1997
).
State-Society Synergy: Government and Social Capital in Development
.
University of California international and area studies digital collection
,
Berkley
,
USA
.
Field
A.
(
2009
).
Discovering Statistics Using SPSS
.
3rd edn
.
SAGE
,
London
,
United Kingdom
.
Flint
C. G.
Luloff
A. E.
Theodori
G. L.
(
2010
).
Extending the concept of community interaction to explore regional community fields
.
Journal of Rural Social Sciences
25
(
1
),
22
36
.
German Society for International Cooperation (GIZ)
(
2011
).
Estado de la Región Trifinio 2010: Datos Socioeconómicos y Ambientales de los Municipios (State of the Trifinio Region 2010: Socioeconomic and environmental data of municipalities)
.
Report for the Ministerio Federal de Cooperación Económica y Desarrollo
,
Eschborn
,
Germany
.
Global Water Partnership
(
2012
).
What is integrated water resource management? http://www.gwp.org/The-Challenge/What-is-IWRM/
(Accessed December 19 2015)
.
Green
E.
Daoust
A.
(
2012
).
Final Evaluation of the Integrated Management of the Montecristo Trinational Protected Area Project
,
Report for the Global Environmental Facility
,
Quebec
,
Canada
.
Heathcote
I. W.
(
2009
).
Integrated Watershed Management – Principles & Practice
.
John Wiley and Sons
,
New York
,
USA
.
IBM Corp.
(
2013
).
IBM SPSS Statistics, Version 22.0
.
IBM Corp
,
Armonk
,
USA
.
López
A.
(
2004
).
Environmental Conflicts and Regional Cooperation in the Lempa River Basin: The Role of Central America‘s Plan Trifinio
.
Report for Aldelphi Research
,
Oktoberdruck
,
Germany
.
Nelson
A.
Chomitz
K. M.
(
2004
). .
Ostrom
E.
(
1990
).
Governing the Commons: The Evolution of Institutions for Collective Action
.
University Press
,
Cambridge
,
UK
.
Pronyk
P. M.
Harpham
T.
Busza
J.
Phetla
G.
Morison
L. A.
Hargreaves
J. R.
Kim
J. C.
Watts
C. H.
Porter
J. D.
(
2008
).
Can social capital be intentionally generated? A randomized trial from rural South Africa
.
Social Science & Medicine
67
(
10
),
1559
1570
.
Prüss-Ustün
A.
Bartram
J.
Clasen
T.
Colford
J. M.
Cumming
O.
Curtis
V.
Bonjour
S.
Dangour
A. D.
De France
J.
Fewtrell
L.
Freeman
M. C.
Gordon
B.
Hunter
P. R.
Johnston
R. B.
Mathers
C.
Mäusezahl
D.
Medlicott
K.
Neira
M.
Stocks
M.
Wolf
J.
Cairncross
S.
(
2014
).
Burden of disease from inadequate water, sanitation and hygiene in low- and middle-income settings: a retrospective analysis of data from 145 countries
.
Tropical Medicine and International Health
19
(
8
),
894
905
.
Ryan
G. W.
Bernard
H. R.
(
2003
).
Techniques to identify themes
.
Field Methods
15
(
1
),
85
109
.
Sarason
S. B.
(
1974
).
The Psychological Sense of Community: Prospects for a Community Psychology
.
Jossey-Bass
,
San Francisco, CA
.