Determinants of Livelihood Diversification: The Case of Community-Based Ecotourism in Oaxaca, Mexico

Diversification is a livelihood strategy that households use to survive and to absorb disturbances. Community-based ecotourism (CBE) is part of a set of sustainable options that are available to households for diversification. This paper analyses households’ capital assets that increase the probability of a family becoming involved in a CBE compared to other forms of diversification, and empirical evidence is presented for coastal communities located in Oaxaca, Mexico. Based on household surveys and a multinomial logit econometric model, the results show that the probability that a household decides to become involved in CBE increases for households with young people and those that have their basic needs covered by government programs or small agricultural production. Other forms of diversification are positively related to agreements within the community, woodfire use, and the presence of irrigated lands. Thus, CBE is determined by specific assets mentioned previously compared to other forms of diversification. Particularly, to promote CBE, tourism policies need to consider households’ human, social, natural, and financial capital assets and the associated programs in an integrated way to enhance community development.


Introduction
Rural households develop various strategies to be resilient, manage risks and stressors, of which diversification, migration, and the intensification of production are the main ones [1]. Specifically, diversification is recognized as one of the most common methods on which rural livelihoods rely [2][3][4] and is defined as the process that allows households to have different activities, sources of income, or employment to improve their welfare [1]. In Latin America, for 20 to 30% of households, off-farm activities represent 50% of the family economy [5,6]. Similarly, in Mexico, diversification of income represents 49% of the total household income [7]. Additionally, in coastal municipalities, the service sector has grown, especially for urban and touristic regions [8]. Tourism is then a strategy and a substantial source of revenue for rural localities in many countries [9], partly due to the constant increase of tourism arrivals throughout the world, reaching 1.2 billion in 2016, and 1.8 billion were predicted for 2030, contributing to 10% of the world's GDP [10]. However, due to COVID-19, which is partly caused by ecosystem fragmentation [11], tourism arrivals compared to the previous year decreased 72% in 2020 [12]. In relation to environmental deterioration, the United Nations General Assembly declared 2017 to be the International Year of Sustainable Tourism, recognizing tourism's contribution to sustainable development goals and its potential to decrease poverty and promote community development and biodiversity conservation. Sustainable tourism should maintain essential ecological processes, respect the sociocultural authenticity of host communities, ensure fair distribution of long-term economic benefits, and ensure the inclusive participation of all stakeholders [13]. Reynolds and Braithwaite [14] highlighted that approximately 40% of international tourism is enjoying wildlife. In that sense, ecotourism has been recognized as central for sustainable tourism because it promotes ecosystem conservation and the improvement of livelihoods [15] and implies environmental education and equity in its implementation [16]. In particular, community-based ecotourism (CBE), in addition to ecotourism criteria, seeks to "create local enterprises that provide livelihood benefits to communities while protecting indigenous cultures and environments" [17], thus making these enterprises key projects for sustainable tourism. CBE implies that a significant proportion of the benefits must be received by local residents and that the development of ecotourism must be led by them [18,19]. The aims of CBE include equity on participation, ecosystem conservation, and improvements in well-being [19]. However, Iorio and Corsale [18] indicated that communities are not homogenous and involvement in CBE as a part of one of the diversification strategies depends on the resources and skills of the household.
Therefore, CBE as part of a household's livelihood strategy depends on capital assets as many other forms of diversification [1]. Human, physical, and financial capital assets have been widely studied in rural context in poverty assessments or household's economy papers [20,21]; whereas social and natural aspects have been studied from environmental or rural sociology studies [22]. Therefore, capital assets have been scarcely linked to study diversification as a process and even less to CBE. This, in turn, will help to identify key aspects to support, in order to incentive more sustainable practices such as CBE.
Thus, the aim of this research is to analyze the relevance of household characteristics in terms of their social, natural, financial, physical, and human assets to go from a non-diversified livelihood strategy to CBE or other forms of diversification. This paper highlights the meaningful variables to promote CBE as a promising form of sustainable tourism instead of other forms of diversification. For that purpose, empirical evidence from coastal communities in Oaxaca, Mexico, is presented since, in this region, CBE has an enormous potential due to the collective action in the region and natural resources, as well as due to the distance to other touristic destinations that are a source of demand. Assessing the factors that influence households' decisions to diversify into CBE provides information for planner and community development.

Capital Assets on Diversification
Diversification can be measured in terms of a change from farm to off-farm activities [23], as employees or entrepreneurship [24] or as the number of activities independent of the sector [4]. Additionally, vertical diversification has been considered; that is, within the same sector, such as coffee production, the farmer becomes involved in commercialization [25]. In this paper, diversification is defined as a process by which households increase the number of activities depending on the access and possession of capital assets and depends on the vulnerability context and public policies [3]. Context variables such as climate shocks [24,26], road infrastructure proximity [24], and altitude [25] have been shown to be significant. Capital assets are both tangible and immaterial and reveal the capabilities and skills necessary to handle different types of stressors [1,27,28]. There are five types of capital assets: human, social, natural, physical, and financial [1]. Human capital includes education, the number of members in a household, training, and health; natural capital refers to land hectares and natural resources access and use; physical capital is the basic infrastructure and goods needed to carry out production but also house conditions; financial capital is any kind of income and savings; and social capital refers to social networks, participation in associations, and relationships of trust, as well as other elements for collective action. Consequently, scholars have shown that a household deficient in assets will have less welfare; however, if capital assets are accumulated, they can be invested in other activities and induce an enrichment of the livelihood portfolio [29,30]. In addition, capital assets as determinants for diversification have been studied [24,31], including participation in payment for ecosystem services programs [32] or wildlife tourism [33].
Human capital has been widely studied in rural household's economy and poverty, specially showing that higher levels of formal education provide the larger labor opportunities [34], many times associated to non-agricultural activities [35] but also education and traditional knowledge and skills are positive aspects for conservation as showed in Quintana Roo, Mexico [36]. The number of household members and their age is also an important aspect for defining households' strategies, since it represents the workforce [7,37]; however, younger people are many times more interested on migration or other off-arm activities [33]. In the case of Mexico, the empirical evidence suggests that education and age determine household diversification [7,20,21].
In terms of social capital, institutional networks are very important in order to diversify [35,38,39] specially because they represent financial support, capacity building, and market access. Rules, agreements, and customs are also elements for collective action that are generally linked to community organization that often allows access to new forms of income or reduces production costs.
Market access, income, access to credit, and savings are financial elements affecting livelihood decisions [40,41]. Remittances are also an important source of revenues for many households, however, it is not linked to diversification since it is used for food consumption [42]. Government transfers are also important, especially in developing countries such as Mexico [26,43].
Natural capital is made up of the natural resources that the household possess such as its land [20,44], generally associated to agricultural production, but also of wildlife use [22,45] and water that are a result of ecosystem services provided within the region.
The building materials of the house and home appliance are physical assets indicators of welfare but also communication access such as mobiles. In addition, tools and machinery are necessary for many productive activities, and communications infrastructure such as roads and distance to urban centers are also critical to market and service access and are also included in this category [35].

Capital Assets on Tourism and CBE
The capital assets approach has been recognized as a good measure of well-being since it includes not only income as an indicator of poverty but also other tangible and intangible aspects of welfare [46]. Households' capital assets represent a set of characteristics and capacities that allows families to respond to changes and choose a livelihood strategy. Therefore, capital assets are not only taken as determinants to diversification [1], but they are also good indicators of economic activities or public policy outcomes. The vast majority of the literature is focused on capital assets as indicators of success, but some examples exist for explaining participation in tourism planning. Hence, this brief review aims to highlight the importance of different assets on tourism participation and assets as CBE outcomes as a way to understand the role of assets on CBE ( Figure 1) as described in the following paragraphs. However, this paper focuses on demonstrating the role of assets in CBE participation, in order to identify key aspects to promote this activity.

Capital Assets as Indicators of CBE Impacts
The effects of CBE on livelihoods have shown interesting results regarding social capital. In particular, external institutional support [47], leadership [48], and reciprocity [49] are key issues for its success.
In terms of income, a case study presented in Cambodia showed a small difference between CBE members and non-members [50] because indirect effects were not considered; members could also perceive benefits if they had homestays, became a tour guide, or cooked in the restaurant. Other studies have shown more integrated effects considering environmental and socioeconomic assets, showing positive effects in CBE in Mexico [51], China [52], and Botswana [53]. Qian et al. [52], for example, analyzed the impact that community-based tourism (CBT) and lease-operation tourism have on assets, showing that in CBT, the overall value of assets is greater and that this form of tourism thus contributes to well-being. Similarly, Simpson [17] showed in two case studies in South Africa that tourism can have a positive effect on communities related to employment generation, infrastructure building, biodiversity, and gender involvement.

Capital Assets as Indicators of CBE Impacts
The effects of CBE on livelihoods have shown interesting results regarding social capital. In particular, external institutional support [47], leadership [48], and reciprocity [49] are key issues for its success.
In terms of income, a case study presented in Cambodia showed a small difference between CBE members and non-members [50] because indirect effects were not considered; members could also perceive benefits if they had homestays, became a tour guide, or cooked in the restaurant. Other studies have shown more integrated effects considering environmental and socioeconomic assets, showing positive effects in CBE in Mexico [51], China [52], and Botswana [53]. Qian et al. [52], for example, analyzed the impact that community-based tourism (CBT) and lease-operation tourism have on assets, showing that in CBT, the overall value of assets is greater and that this form of tourism thus contributes to well-being. Similarly, Simpson [17] showed in two case studies in South Africa that tourism can have a positive effect on communities related to employment generation, infrastructure building, biodiversity, and gender involvement.
In addition, assets are widely used to measure ecotourism impacts (in general) on livelihoods, showing an [15,54] increase in income, employment availability [55][56][57][58], education, skills, and infrastructure improvement [58]. Ecotourism has also been mentioned as a driver for small enterprises [56] and in many cases as a positive instrument for natural capital conservation [58]. However, ecotourism can cause inequality in economic distribution [55,57] and differences in household welfare [56], as well as other trade-offs [59]. The positive and negative effects of CBE and ecotourism have been shown in the literature, but the capital assets approach allows us to assess tangible and intangible aspects of well-being.

The Role of Capital Assets in Community Participation in Tourism
Capital assets, as determinants of CBE participation, have not been studied, but they have been used to analyze community participation in tourism development. Bello et al. [60] determined that community participation in tourism planning is linked to community members' capital assets, specifically, education, training, financial resources, and other factors, such as benefit distribution and coordination. Similarly, Bennett et al. [58] confirm that human, financial, natural, and physical capital are important to assess the community's capacity to participate in tourism in protected aboriginal areas in Canada. Moreover, social capital, power relationships, and institutional arrangements have been identified as being critical to the development of tourism in communities [48,58,60,61]. In addition, assets are widely used to measure ecotourism impacts (in general) on livelihoods, showing an [15,54] increase in income, employment availability [55][56][57][58], education, skills, and infrastructure improvement [58]. Ecotourism has also been mentioned as a driver for small enterprises [56] and in many cases as a positive instrument for natural capital conservation [58]. However, ecotourism can cause inequality in economic distribution [55,57] and differences in household welfare [56], as well as other trade-offs [59]. The positive and negative effects of CBE and ecotourism have been shown in the literature, but the capital assets approach allows us to assess tangible and intangible aspects of well-being.

The Role of Capital Assets in Community Participation in Tourism
Capital assets, as determinants of CBE participation, have not been studied, but they have been used to analyze community participation in tourism development. Bello et al. [60] determined that community participation in tourism planning is linked to community members' capital assets, specifically, education, training, financial resources, and other factors, such as benefit distribution and coordination. Similarly, Bennett et al. [58] confirm that human, financial, natural, and physical capital are important to assess the community's capacity to participate in tourism in protected aboriginal areas in Canada. Moreover, social capital, power relationships, and institutional arrangements have been identified as being critical to the development of tourism in communities [48,58,60,61]. Similarly, social networks, external institutional support [48,58], and the history of the community [47] have also been recognized as relevant. In addition, the intention to participate in community tourism has been linked to perceived impacts on social, economic, and natural dimensions [62]. Gender and women's empowerment also influence tourism development [63]. Gender has an influence on the intention to participate in communitybased tourism [64] and to support community-based tourism development [65].
The literature shows that capital assets influence participation in tourism and have been used as indicators for CBE impacts. That is why the focus of this research is on household assets as determinants of CBE.

Study Area
In Mexico, the government uses nature tourism as a general name for all types of tourisms having a direct interaction with nature and includes wildlife watching, ecotourism, CBE, and adventure tourism such as rafting. In this country, 36% of natural tourists are wildlife watchers [66] and provide 26.5% of the revenues from this type of tourism. Moreover, ecotourism has been proven to be successful in terms of environmental and socioeconomic outcomes in case studies within the country [67]. In addition, the federal government has established certain programs to promote this activity, such as the National Program of Ecotourism and Rural Tourism that was created in 2010, the Strategy for Biodiversity Mainstreaming in the Tourism Sector, and the Strategy for the development of Nature Tourism in Mexico [68], which include ecotourism, wildlife watching, CBE, and adventure tourism.
In Oaxaca state, there are several communities taking tourists to observe wildlife and experiencing nature due to the ecological and cultural diversity existing in the state, despite poverty conditions [69]. The Oaxaca tourism office promotes natural tourism routes, including the coastal region, and statistics show that 39% of Oaxaca visitors enjoy wildlife [70]. However, the priority at the national and state levels is traditional beach and sun tourism, and CBE is only promoted as a one-day trip as a complementary offer.
The localities studied in this research are located on the Pacific coast and are rich in mangroves and dry forest due to its tropical sub-humid climate that have been declared Ramsar sites but with poverty conditions, and in response, households adopted migration (through remittances) and diversification into the tertiary sector, including CBE, as their main livelihood strategies [42]. This region was affected by the hurricane Paulina in 1997 and in 2012 Carlotta made landfall in these communities.
Ventanilla, Vainilla, and Escobilla localities are located in the Santa María Tonameca municipality and Barra de Navidad in Santa María Colotepec ( Figure 2). They are situated within a touristic corridor limited by Puerto Escondido and Huatulco. Puerto Escondido used to be a small fishing village and is currently a sun and beach touristic destination, where surf is one of the main attractions. In contrast, in Huatulco, the original population was evicted by the government to build a tourist resort to increase employment in the region and economic development based on international hotel chains promoting allinclusive packages [71]. In Huatulco, the main touristic attractions are coral reefs and visits to Copalita cascades and coffee plantations. Both cities attract international and national tourism depending on the month of the year, and guided tours are promoted by hotels and other tour operators to visit other communities in the region, such as those studied in this paper.
The selection criteria for these localities are that (a) they have similar environmental and socioeconomic contexts and (b) they have similar forms of diversification but with different levels of consolidation. Thus, these localities are sufficiently homogeneous while also having some differences.
According to the National Population Council [72], the socioeconomic characteristics of the localities show that households have not even finished primary level and marginalization is high. Agriculture and fisheries contribute to local income and self-consumption [71], and government transfer is also a significant part of the economy. CBE started after the marine turtle ban (declared in 1992 due to an overexploitation of this resource) as a new form of revenue [51,73] promoted by a nongovernmental organization. Marine turtle exploitation remains prohibited; thus, tourism and agriculture are the main sources of revenue within the area. CBE consists of guided tours to coastal lagoons to observe birds, crocodiles, mangroves, and marine turtles nesting in the beach. Tourists can also enjoy local food and purchase handmade artwork.
Ventanilla enjoys more tourists than the other localities since they have an agreement with tours coming from Huatulco and have built an interesting social network for support. As a consequence, the majority of households depend on this activity economically [74]. Ventanilla community have always preserved their independence in relation to their community-based decision making. Although the community has suffered from certain social conflicts [73], causing the creation of a new eco-touristic cooperative, they have co-existed for at least 10 years sharing the coastal lagoon for guided tours. Therefore, this community is an example for many other in the region concerning community-based ecotourism [51]. According to the National Population Council [72], the socioeconomic characteristics of the localities show that households have not even finished primary level and marginalization is high. Agriculture and fisheries contribute to local income and selfconsumption [71], and government transfer is also a significant part of the economy. CBE started after the marine turtle ban (declared in 1992 due to an overexploitation of this resource) as a new form of revenue [51,73] promoted by a nongovernmental organization. Marine turtle exploitation remains prohibited; thus, tourism and agriculture are the main sources of revenue within the area. CBE consists of guided tours to coastal lagoons to observe birds, crocodiles, mangroves, and marine turtles nesting in the beach. Tourists can also enjoy local food and purchase handmade artwork.
Ventanilla enjoys more tourists than the other localities since they have an agreement with tours coming from Huatulco and have built an interesting social network for support. As a consequence, the majority of households depend on this activity economically [74]. Ventanilla community have always preserved their independence in relation to their community-based decision making. Although the community has suffered from certain social conflicts [73], causing the creation of a new eco-touristic cooperative, they have co-existed for at least 10 years sharing the coastal lagoon for guided tours. Therefore, this community is an example for many other in the region concerning community-based ecotourism [51].
Escobilla beach is a marine turtle federal sanctuary, which covers 149 ha, and monitoring is one of the main objectives of the government team located in this park. The sanctuary was established after the marine turtle ban, and due to the prohibition for extracting those species, Escobilla members diversified offering guided tours to a coastal lagoon and the beach to replace local people's income from marine turtle exploitation, creating an ecotourism cooperative. In addition, some households have restaurants and cabanas for tourists, but agriculture remains the main activity. In contrast, Vainilla has recently started offering guided tours to a small coastal lagoon, but its main source of revenue is subsistence agriculture. Escobilla beach is a marine turtle federal sanctuary, which covers 149 ha, and monitoring is one of the main objectives of the government team located in this park. The sanctuary was established after the marine turtle ban, and due to the prohibition for extracting those species, Escobilla members diversified offering guided tours to a coastal lagoon and the beach to replace local people's income from marine turtle exploitation, creating an ecotourism cooperative. In addition, some households have restaurants and cabanas for tourists, but agriculture remains the main activity. In contrast, Vainilla has recently started offering guided tours to a small coastal lagoon, but its main source of revenue is subsistence agriculture.
Barra de Navidad has also recently started CBE into a coastal lagoon and has built a restaurant and cabanas, but agriculture represents the main source of income [74].

Data Collection
A total of 212 heads of household were surveyed in 2014, representing 73% of the total households of the four rural coastal communities (Ventanilla, Vainilla, Escobilla, and Barra de Navidad).
A pilot survey was carried out previously to assess the consistency of questions and appropriate language. The survey was applied dividing the communities into quadrants, and all households were asked to answer the questions (such as a census), but some houses were uninhabited or did not want to answer. The socioeconomic characteristics of households and information on each type of capital (human, social, financial, physical, and natural), as shown in Table 1, were the sections of the questionnaire and based on the literature presented in previous section.  [34] Community infrastructure and regional context [35,75,77] Data on education, gender, number of members, and age of each member of the household were collected as part of human capital. Access to health services and the presence of chronic diseases were also asked because they influence employment and social development. Training was considered important since many new projects require new knowledge in specific areas, such as biodiversity or tourism attention. In terms of social capital, it is important to assess the rules and agreements within the community as proxies of collective action. Additionally, social cohesion and relationships with external institutions such as nongovernmental organizations are important for diversification. It has also been proven that households need a minimum of financial and physical capital in terms of income, credits, or productive infrastructure to diversify [7]. Finally, natural capital is relevant for food security and ecotourism. The survey length was an hour maximum. The question format for each capital was dichotomous, open ended, or codified depending on the variable.
For the analysis, three categories of household diversification were considered: nondiversified households (dedicated only to one activity), households that are diversified into different activities, including CBE, and households that are diversified into many activities other than ecotourism. This information was obtained by asking about household activities.

Model Description
Diversification (DIV), the dependent variable of the model, is defined as follows: i.
Not diversified = households that engage in only one economic activity, independent of the type and characteristic, such as agriculture, livestock, fishing, or other nonagricultural activity, self-employment, or employment; ii. Diversified = households that engage in more than one activity but not ecotourism; iii. Diversified including ecotourism = households that engage in more than one economic activity and are part of the CBE (as a cooperative member).
The independent variables presented in Table 2 were first analyzed using factorial analysis and a correlation test between the variables of each type of capital to select the variables that explained the best diversification. For example, the average age of the households and the average age of the head of the households are two variables that could be correlated, and there is no need to include both variables in the model; thus, this kind of selection was conducted. Table 2 presents the variables obtained in the previous statistical analysis; these are the best variables to explain the model. Then, a multinomial logit model was run in STATA. Multicollinearity and heteroskedasticity were assessed, and any problems that were identified were corrected by the inflation factor of variance (FIV) and the Breusch-Pagan test, respectively. A multinomial econometric model is an extension of a binary model when the dependent variables have more than two options. Let us consider that (Yi, Xi) is a random sample of the population under study, where i represents the number of households (I = 1, 2, 3, . . . , n) and Yi is the decision to diversify, which has three options, (1) Not to diversify, (2) Diversify, or (3) Diversify into ecotourism, and Xi represents the capital variables. We are interested in knowing how changes in Xi affect the possibilities of Yi: The logit multinomial model is estimated using the maximum likelihood to obtain consistent and asymptotically normal parameters. Thus, using the variables from Table 2 the model takes the following form: DIV ij = β 0 + β 1 * age i + β 2 * agreement i + β 3 * support i + β 4 * plot i + β 5 * woodfire i + β 6 * irrigation i + β 7 * infrastructure i + e j = 1 (Not diversified), 2 (Diversified), and 3 (Diversified into CBE).

Household's Socioeconomic Profiles
Diversification is a livelihood strategy for 90% of the households in the area of study, and 23.6% have chosen ecotourism (Figure 3) to complement farm and other off-farm activities and 66% of the households possess other forms of income. Nondiversified households are dedicated to agricultural activities (10.4%) (Figure 3).

Household's Socioeconomic Profiles
Diversification is a livelihood strategy for 90% of the households in the area of study, and 23.6% have chosen ecotourism (Figure 3) to complement farm and other off-farm activities and 66% of the households possess other forms of income. Nondiversified households are dedicated to agricultural activities (10.4%) (Figure 3). The average annual income of the nondiversified households is USD 2153/yr, which is half the income of diversified households, which is on average USD 5225/yr. The diversified households involved in ecotourism earn, on average, USD 5294/yr and without CBE, USD 5225/yr ( Figure 4). Therefore, CBE is competitive with other forms of diversification in terms of household income. The average annual income of the nondiversified households is USD 2153/yr, which is half the income of diversified households, which is on average USD 5225/yr. The diversified households involved in ecotourism earn, on average, USD 5294/yr and without CBE, USD 5225/yr ( Figure 4). Therefore, CBE is competitive with other forms of diversification in terms of household income.  With respect to other social aspects, households diversified into CBE have the lowest average age, and approximately half of the sample believe that community agreements are enforced ( Table 3). The percentage of households receiving institutional support is larger for non-diversified households because they are the most vulnerable ones, since they have lower income as presented previously and also in terms of basic needs, land hectares, liming food production, and security. In contrast, diversified households have a larger proportion of hectares cultivated and irrigated. In terms of natural resources use, a high proportion of the sample uses wood fire for cooking, which is part of the Oaxaca traditional way of cooking. In contrast, there are differences on physical capital assets, since households with the largest infrastructure value are the diversified ones. Thus, diversified families generally have better living conditions.  With respect to other social aspects, households diversified into CBE have the lowest average age, and approximately half of the sample believe that community agreements are enforced ( Table 3). The percentage of households receiving institutional support is larger for non-diversified households because they are the most vulnerable ones, since they have lower income as presented previously and also in terms of basic needs, land hectares, liming food production, and security. In contrast, diversified households have a larger proportion of hectares cultivated and irrigated. In terms of natural resources use, a high proportion of the sample uses wood fire for cooking, which is part of the Oaxaca traditional way of cooking. In contrast, there are differences on physical capital assets, since households with the largest infrastructure value are the diversified ones. Thus, diversified families generally have better living conditions.

Marginal Effects of Capital Assets on Diversification
The results of the multinomial model show the probability that capital assets influence the decision of a non-diversified household to diversify both in general (Model A) and into ecotourism (Model B) ( Table 4). The models have a statistical significance of 99% (Prob > chi 2 ), and 65% of the variables are statistically significant (Prob-Ind).
In model A (general diversification), all the variables included in social capital are significant. If households consider that agreements have been made, the probability that they will diversify increases by 18.3% (prob < 1%), while if households have received support from the government or from a nongovernmental organization, the probability is reduced by 16% (prob < 5%). In addition, the number of hectares irrigated for agriculture increases the probability that the household will decide to diversify by 13% (prob < 5%), and the use of wood fire for cooking is a significant variable of the natural capital that increases the probability of diversification by 34% (prob < 1%). However, if the household has more land, the probability decreases by 1.8%. Finally, the average age of the household (who are more than 15 years old) and the infrastructure value both positively affect the household's decision to diversify its activities but by a low proportion compared to the other variables.
Model B (CBE) shows the variables that are significant for a nondiversified household to diversify into ecotourism. All capitals have significant variables in the model except physical capital. In social capital, if households have received support from the government or from a nongovernmental organization, the probability of diversifying into ecotourism increases by 11.6% (prob < 10%), whereas if households consider that agreements are accomplished, the probability is reduced by 7%. In the case of natural capital, as the number of parcels increases and if the household uses wood fires, the probability of ecotourism increases by 5.5% (prob < 10%) and 2.7%, respectively. In contrast, if the average age of the household increases (for persons of more than 15 years old), the probability is reduced by less 1% (prob < 10%), opposite of the results shown in the first model. Finally, the number of hectares with irrigation has a negative effect on the probability of doing ecotourism, and the probability decreases by 11.5% (prob < 10%).

Discussion
The transition from a non-diversified strategy to a diversified strategy, including CBE or other activities, depends on different capital assets, but the assets that are significant for both types of activity are government support and irrigated land. On the one hand, institutional support positively influences CBE, which is consistent with Ávila-Foucat and Rodríguez-Robayo [33], who showed that for wildlife tourism, government transfer was significant. In contrast, institutional support is negatively related to other types of activities because the most vulnerable families receiving government subsidies, in our case study, are less suitable to diversify since a minimal amount of capital is required [7]. On the other hand, irrigated land is negatively associated with CBE and positively associated with other sources of diversification. This is an interesting result meaning that households having more infrastructure for agriculture become involved in activities probably associated with the transformation of the product, such as a transformation of maíz into tortillas. Similar results have been reported for coffee producers, where some producers decide to diversify within the value chain of coffee [25]. In contrast, CBE households are limited in the amount of land for agriculture and possess fewer possibilities to intensify their agriculture. Institutional support and irrigated land have a significant but opposite effect on the two types of diversification analyzed in this paper. Additionally, agreements, wood fire use and age were only significant in one of the models. This means that CBE is determined by different capital assets compared to other forms of diversification.
Specific factors determining CBE show a positive influence of at least one variable for each type of capital, with the exception of physical capital; notably, institutional support as part of social capital has the largest marginal effect. A household's physical capital is mainly composed of agricultural or other productive infrastructure, but ecotourism infrastructure (such as boats or cabanas) is owned by the cooperative and is not part of the household's assets directly. That is why physical capital is not significant for CBE.
In this paper, we analyze only two forms of social capital: institutional support as an indicator of social networks and community agreements as an indicator of social norms and rules.
Institutional support is mainly linked to government programs for poverty, for agricultural production, or for daily wages for some environmental actions. Except for Ventanilla, non-governmental organization support is minimal. Institutional support has a positive influence on CBE because one of the main programs that households received was Prospera (64% of households), which was created and implemented from 2012 to 2018 from the federal government to reverse poverty conditions by improving access to food, health, and education. Basic needs are covered with this program, which allows households to look for other strategies to diversify. Another important program was PROCAMPO, which subsidizes agricultural production by providing MXN 1,300/ha for each agricultural cycle, allowing production for self-consumption, which allows engagement in other activities by the heads of household [7,26,58]. Additionally, the National Commission for the Development of Indigenous Peoples (CDI by its Spanish abbreviation) provides support for building ecotourism infrastructure, and the Ministry of Environment provides daily employment for conservation activities that can be linked to ecotourism attractions. Additionally, nongovernmental institutions have provided funds to those communities over the years such as Red de humedales, La Ventana, and Fondo Oaxaqueño [79].
The findings are consistent with the CBE literature, which has shown the relevance of financial help for starting this activity [59,60,80]. However, government support has a negative influence when a household wants to diversify into other activities such as small commerce or taxis because, as we stated earlier, the main program aims to provide basic needs, and diversified households are less dependent on external help because, as shown by other studied poorest households, they are not the ones that diversify [7]. Diversified households, as described in Table 3, have a larger proportion of land having the possibility to cover their basic nutritional needs. Therefore, government support does have different effects depending on whether the household wants to diversify into a tertiary activity apart from CBE. This implies that diversified households that depend on government subsidies are more suitable to be involved in CBE. CBE households require having basic nutritional and social needs covered and financial help for building a CBE project.
The previous is confirmed when looking at natural capital. The number of parcels is a significant and positive variable for CBE, but having irrigated hectares has a negative effect, which means that households dedicated to ecotourism practice self-consumption agriculture and need a minimum amount of land. Landowners with irrigated land are dedicated to other off-farm activities, which is consistent with the literature indicating that large producers increase production rather than diversifying their livelihood [35]. In this region, the average size of a parcel of land is 2.6 ha, which is the minimum for self-consumption; therefore, the land is used to secure food rather than boosting income. Thus, the probability that a household decides to become involved in CBE increases when they have their basic needs covered due to government programs or small agricultural production. Ecotourism is an opportunity to enhance the ability of a household to survive, it is more of a push factor than a strategic plan, that is, households are obliged to look for many sources of income, such as CBE, in order to confront many stressors, instead of choosing from many options the best to target a new market as a result of analyzing different options. The maintenance of CBE depends on the ability to build relationships with external institutions for developing infrastructure and markets.
The perception of community agreements is an important variable for diversification, but surprisingly, it is negative for CBE. According to Duchelle et al. [81], households with high social capital can easily develop new activities that are based on trust and assistance from neighbors, which explains the results of the first model. In contrast, for CBE, agreements at the community level are not significant, probably because, for ecotourism, the agreements are taken at a cooperative level; in contrast, institutional support can be provided at the household level. Therefore, some variables that are not significant at the household level might be at the community or cooperative scale. The construction of social capital is not per se an objective of CBE; instead, it is more oriented to an equitable distribution of benefits and respect to the local culture; however, it is implicitly required for good management. The results mean that the transition from nondiversified to CBE is independent of agreements or rules at a community level but probably not at a cooperative scale. Indeed, in Ventanilla, for example, there have been some social conflicts and two cooperatives created in the same village with the same purpose [73]; however, they have coexisted for many years. In addition, coordination between community members, social networks, leaderships, and norms are other variables mentioned in the literature that could be explored in further research [48,58,60].
Regarding human capital, age is not significant for off-farm activities, which is not expected since, in Mexico, age is among the main determinants for diversification [37,43]. In contrast, households with the lowest average age are more likely to be involved in CBE, as mentioned by Bello et al. [60]. This involvement may occur for two reasons: a mangrove guiding tour is physically demanding, and young people are willing to engage in new activities [6]. The average age of the households dedicated to ecotourism is 37 years old; thus, there is a large proportion of the population that can potentially be interested in CBE. However, it is always a challenge to retain young people in rural areas since migration is always a temptation.
Wood fire is another variable that has a positive effect on general diversification, which is not expected since families spend approximately two hours a day collecting wood for the fire, leaving less time for diversification. However, in this particular case, food that is sold in small local restaurants is cooked in a traditional woodfire stove; thus, collecting this natural resource provides revenues and savings [45]. Moreover, it has been demonstrated that wood fire consumption increases if there is good access and a large number of resources and if households have low income, as is the case in these communities [82].
In sum, CBE determinants are institutional support, number of parcels, age, and irrigated land. CBE is considered an important strategy for achieving sustainable tourism; thus, institutions interested in promoting this activity need to take into account that ecotourism is part of a diverse livelihood portfolio that depends on household capacities. Households are embedded in social, financial, natural, physical, and human capital, complementing each other. Particularly, this research shows that institutional support has a large marginal effect compared to other variables, meaning that sectorial policies have the potential to change households' strategies. Therefore, developing CBE efficiently requires coherence and regional integration of tourism policies with environmental, social, and productive policies. In addition to government programs, other political and social contexts of the region are important. In that respect, one of the limitations of this paper is that contextual aspects are not taken into account in detail, and social network analysis and policy coherence research would be useful to include exogenous factors influencing households. Moreover, a deeper analysis could be done only with household members of the ecotourism cooperative, as well as a dynamic analysis of household assets to see if the determinants change over time.

Conclusions
This research highlights that the determinants of different forms of diversification are not equivalent. The capital asset approach allows us to consider different dimensions of households' capacities that define their livelihoods. Specifically, this paper analyses the determinants of community-based ecotourism as a strategy for reaching sustainable tourism.
CBE is determined by household assets, especially by age, government transfers, and land for cultivation. That is, CBE can be considered an alternative for young households that have land to grow food for their own consumption and receive government support that provides for their basic needs and other specific inputs for tourism. In contrast, households that already have commercial agriculture and irrigation are generally involved in other off-farm activities. For these households, social capital and woodfire use are important since they are used in traditional restaurants. CBE is possible if communities have their basic needs covered, a local working force, and a minimum of capital to invest.
Mexican tourism policy is actually promoting the creation of sustainable tourism zones, which are destinations where tourism development respects the environment and local culture and generates opportunities for development [68]. Nature tourism is considered one of the main tourism segments to be developed within the region studied, but policies need to be aware that tourism is part of a diversified strategy for local communities and that different capital assets should be considered to start a community-based project; therefore, social, economic, and environmental policies need to be coherent and programs for poverty alleviation, agriculture, and tourism aligned for the community's sustainable development.