Unraveling households’ natural resource management strategies: a case study in Jalisco, Mexico

ABSTRACT The analysis of resource management strategies is considered to be relevant to the conservation and sustainable use of ecosystems. In Mexico, ejidos are the most important of the land tenure institutions that grant access to resources inside communities. Although it is recognized that an ejido’s internal structure creates different social groups, few studies have explored their resource management strategies. The aim of this study was to characterize natural resource management strategies in two ejidos in a highly biodiverse region on the southern coast of Jalisco, Mexico. We sought to identify differences in strategies and the variables that explain these differences. We took the household as the unit of analysis, and conducted 55 structured interviews that tackled different aspects of households’ productive activities and natural resource management. We used cluster and ordination analyses to generate a typology of natural resources management strategies, and linear models to identify the variables that differed among groups. The results show that four different natural resources management strategies were strongly associated with differences in land tenure and the type of ecosystem that a household manages. This information can help us to enhance and diversify strategies for sustaining both ecosystems and community livelihoods.


Introduction
Most of the Earth's ecosystems have been consciously managed and modified by humans (Gómez-Pompa and Kaus 1992; Bush et al. 2015;Levis et al. 2017).The condition of most forests, even those that are considered to be 'pristine', is in fact the result of interactions between society and nature (Bush et al. 2015).According to Toledo et al. (2003), people do not interact with ecosystems spontaneously, but rather carry out internalized plans connected to actions for achieving specific goals.The natural resource management strategies that local people living in tropical forests and other ecosystems carry out denote a mode of appropriation of nature (Toledo et al. 2003).These management strategies, which involve interactions between people and ecosystems, are based on the resource users' knowledge and experience (Berkes and Folke 2000).Such strategies include the way in which families make management decisions under certain political, socioeconomic and ecological conditions, with specific responses to local conditions (van der Ploeg 1990;Gerritsen 2004).
According to Gasselin et al. (2012), a relevant aspect in understanding how households manage their natural resources is to move away from the traditional conception of rural people as mere agricultural producers.These authors consider that agricultural activities occur within a broader life project which often encompasses non-agricultural activities.Therefore, we use the notion of natural resource management strategy rather than the traditional livelihoods approach, under which resources are usually viewed simply as 'natural capital' with a utilitarian function (Binder et al. 2013).
According to Gerritsen (2004Gerritsen ( , 2010)), the natural resources management strategy approach incorporates the sets of notions, elements of knowledge and experiences that groups of farmers possess in a specific region as well as their interactions with multiple actors and institutions.A very important trait of the approach is an understanding of the heterogeneity of responses that farmers express (van der Ploeg 2010).The study of natural resource management strategies that have developed in particular contexts is important for the design of more comprehensive and sustainable development and conservation policies in rural areas (Thompson et al. 2007;Thompson and Scoones 2009;Chambers 2013).
The Chamela-Cuixmala region of Jalisco, Mexico, has great ecological value and a high degree of biodiversity (Ceballos et al. 1999;Noguera et al. 2002;Maass et al. 2005), which has led to the creation of the Chamela-Cuixmala Biosphere Reserve (CCBR) in 2006.Several studies have shown that the ecosystems in this region have been dramatically transformed over the last century as a result of the expansion of agriculture and the exploitation of forests (Castillo et al. 2009).At present, ecosystem preservation in the region remains in constant conflict with extensive farming and tourism development (Maass et al. 2005;Castillo et al. 2009;Riensche et al. 2015).Around 40% of the territory in the region is under the tenure of ejidos (Pérez-Escobedo 2011), a community-based land tenure system that formally emerged as a result of the land redistribution policies following the Mexican Revolution (Knowlton and Orensanz 1998;Perramond 2008).In the Mexican countryside, ejidos are important institutions in the management of natural resources (Alcorn and Toledo 1998;Barnes 2009).At present, approximately 54% of all land in Mexico, and 60% of all forest land, fall within ejidos or similar systems of land tenure (Skutsch et al. 2013(Skutsch et al. , 2015)).Analysis of the management strategies that are practiced in ejidos is therefore important in terms of conservation and sustainable development (Castillo et al. 2005;Cohen Salgado 2014).
Taking as our starting point the requirement that development and conservation policies be locally contextualized, in this study we characterize the natural resource management strategies of the households of two ejido communities located in the Chamela-Cuixmala region and identify the variables that differentiate these strategies.The objective is to understand what determines the different natural resource management strategies of these two communities.

Study area
The Chamela-Cuixmala region has been delimited by researchers studying ecological watersheds (Figure 1).It comprises an area of 6400 km 2 and the San Nicolás, Cuixmala, and Purificación river watersheds (Maass et al. 2005).In the region, we find very different development patterns: the conditions along the coast (where tourism and intensive agriculture take place) are different from those in the uplands (where extensive livestock farming takes place).Also, there are very few indigenous communities; the two ejidos that were selected in this study are composed mainly of mestizo people 1 (Regalado 2000;Castillo et al. 2009).We selected our case studies to highlight contrasts in relevant characteristics (Yin 2003;Newing 2011), particularly their different ecological conditions, their locations in the upper and lower basins of the watersheds, and differences in the date of their foundation.

Data collection
Permission for the study was granted both by the ejidos and by the municipal authorities.We considered the household to be the basic unit of analysis (Eakin 2003).We performed semi-structured interviews with households, in which we addressed such topics as productive activities, family unit characteristics, and the management of natural resources.Interviews were carried out with ejidatarios, posesionarios, avencindados and private landowners.Ejidatarios are the landholding members of the ejido.The land right can be inherited by a single descendant or sold.However, within ejidos, there are other classes of people including those who have been allocated land (a plot) for cultivation during their lifetime, but cannot pass it on to the following generation; further, they do not hold rights over communal lands and cannot vote in the Assembly.These people are called posesionarios (those who possess) and they are often the non-inheriting descendants of ejidatarios.Furthermore, ejidos recognize people without land to cultivate and without rights over communal land, called avecindados (those who settle), who have only a house plot (solar) within the ejido.Finally, there are 'private landowners', who are technically a type of avecindado, because they live inside the ejido, but own private land outside the ejido.
We asked context-specific questions on the use of reciprocal labor, amount and type of land, plant resources utilized, and activities carried out.These variables have been found to be important to natural resource management strategies in other studies conducted in the region (Gerritsen 2004).We worked with two types of information about natural resource management strategies.The first type included all actions or elements of the household's strategy both inside and outside the plot, which may or may not involve natural resource management.The second included activities and elements related specifically to natural resource management (Table 2).We considered that these two types of information both represent parts of the household's overall integrated natural resources strategy.With regard to resource management, we considered the plot to be the space where many of the family's decisions materialize, since it represents the basis of their production system (Ugartechea-Salmerón 2015).In total, 26 interviews were carried out in Pabelo and 29 in Ranchitos.The interviews usually involved more than one household member.In most cases, the head of the family and the holder of land rights was a man, except for four cases, two in each ejido, where a woman was the head of the family.Interviewees in Pabelo were selected using stratified sampling, while in Ranchitos we interviewed all households.In addition, field observations of plots were performed with some interviewees to identify specific characteristics of ecosystems and management practices.

Data analysis
We coded the information in a database with the 53 socio-ecological variables that were used to characterize each household's natural resources management strategy (Table 1).Qualitative information was grouped by theme and used to interpret the results and to ground aspects of the discussion.To characterize different types of household strategy, we performed a cluster analysis on the socio-ecological variables, using the average or UPGMA method on a Gower dissimilarity matrix that was calculated using information from the database.The optimal number of clusters was defined, using a Table 1.Variables on which data was coded and the type of variable.silhouette plot; the optimal is thus the number that minimized the negative silhouette widths (Borcard et al. 2011).
To visualize the clustering and to identify the most important socio-ecological variables that defined the groups, we performed a Principal Coordinate Analysis (PCoA) based on the same Gower dissimilarity matrix.This ordination analysis illustrates the basic structure of a multivariate dataset in the form of continuous axes that highlight the variables with the greatest influence on the structure's definition as well as the relative position of the entities (Borcard et al. 2011).The first two ordination axes were extracted and plotted, and the clusters were projected by identifying them with different colors.
Finally, to identify which socio-ecological variables differed between social groups, we fitted generalized linear models (GLMs) using the group as the predictive variable.GLM was performed on pooled data from both ejidos.This allowed us to compare and contrast all groups.However, as the two ejidos did not have equal representation across the resulting groups (two of the groups had no representatives in Ranchitos, see Table 3), a potential confounding effect could arise.Therefore, we fitted a second set of GLMs using only information from the two groups that were represented in both ejidos and including ejido and its interaction with a group as additional predictors, to control for their effects.Type I ANOVA F tests (for continuous variables) and deviance tests (Chi-square tests) were performed, with ejido being tested first in the second set of models.All analyses were performed using basic routines in the R language (R Core Team 2016).The Gower distance was calculated using the function 'gowdis' in the 'FD' library (Laliberté et al. 2014).Silhouette plots were obtained using the 'silhouette' function in the 'cluster' library (Maechler et al. 2016).The PCoA was performed using 'dudi.mix' in the 'ade4ʹ library (Dray and Dufour 2007).

The ejido Pabelo
In  2).In addition, there are some cropping areas and some secondary vegetation (natural forest that is undergoing a regeneration process).This heterogeneity in vegetation types is due, in part, to variations in elevation within the ejido, which ranges from 600-2,400 meters above sea level.The annual average precipitation is approximately 1,500 mm, and permanent rivers and streams flow through many of the plots.The communal area has been divided up, in an informal manner, between some of the ejidatarios, who use it as agostadero (sites where livestock can feed).The ejidatarios handle the distribution of benefits and the administration of forestry activities through the Assembly.Other activities performed in this communal area include maintenance of firebreaks, fishing, recreation, hunting, the monitoring of an area under a scheme for payment for environmental services, and the extraction of timber and non-timber resources, such as mushrooms, soil, fruits, and fodder.

The ejido Ranchitos
Ranchitos has an area of 3,350 ha, of which

Landscape and plot configuration in Pabelo and Ranchitos
We schematized a typical plot in Pabelo and in Ranchitos according to the main characteristics and elements that are found in plots in the two ejidos (Figure 2).The typical plots have in common the following elements: vegetation patches, sources of water, pasture and forage, and scatterings of trees.In addition, plots are divided into paddocks in which farmers keep cattle for a certain amount of time (between 15 days and 1 month) depending on the area and the season.A typical plot in Pabelo (Figure 2(a)) contains perennial rivers or streams with riparian forest alongside to protect the water flow.Patches of oak forest are also common because oak species are very useful for fences and firewood.The 'typical plot' in Pabelo varies in topographic conditions and altitude, which determine the kinds of vegetation and species that are present.The landscape in Pabelo is made up of productive areas for cattle ranching on large expanses of pastures that dominate the lower elevations and less steep terrain, with forested areas at the high elevations and on steeper terrain.
A typical plot in Ranchitos (Figure 2(b)) contains large patches of tropical deciduous forest that are in various stages of succession due to the rapid regrowth of this forest type.These plots tend to have watering holes, dug by farmers because of the low availability of natural water sources; despite the presence of these, most farmers in Ranchitos need to bring water to their animals by vehicle in the dry season (from April to June).Another characteristic is the presence of areas for maize cultivation (maize is used for cattle feed).The landscape in Ranchitos represents a mosaic in which the tropical deciduous forest is dominant but actively managed and the cleared pastures are smaller than in Pabelo, because of the amount of work that is involved in clearance due to the rapid natural regeneration of this type of dry tropical forest.These different landscape and plot configurations also imply that farmers need to perform different activities during the year.In both ejidos, maintenance activities for pastures, such as 'clearing', applying herbicides, and sowing, are performed.The productivity of the pastures in Pabelo is sufficient to feed the cattle throughout the year.In Ranchitos, farmers use the natural vegetation that grows during the rainy season to feed the cattle and reserve the pastures for the dry season.

Typology of natural resource management strategies in the Chamela-Cuixmala region
The results of the cluster analysis indicate that four groups of natural resource management strategies exist (numbered 1 to 4). Figure 3 shows the main variables that divide these groups: Groups 1 and 2 with small or no plots and no agriculture and crops, and Groups 3 and 4 with medium to large plots and agriculture and crops.Of the 53 variables analyzed across the groups, 35 proved to be significantly different between at least two of the four groups.Most of these 35 variables were resource management variables.For the GLMs between Groups 2 and 4, there was a significant group effect in 31 variables after controlling for the ejido effect.Five variables showed an ejido effect but no group effect, and only one variable (number of activities) showed an interaction effect between ejido and group.On the other hand, certain elements of the strategies were found almost without exception across households and groups: the work of housewives, productive activities, and the presence of family members in the United States.Overall, the main use of plots was for pasture (67.7%), followed by forest (28.1%) and finally crops (4.1%).We provide a description of each group below.Table 3 summarizes the information on each group and on the whole household sample.
Group 1 'Posesionarios with a cattle specialization': This group consists mostly of posesionario households.The central productive strategy and principal activity of this group is the raising and commercialization of cattle.The households in this group generally do not grow crops.People in this group recognizes the greatest number of useful wild plant  species and they include these within their household strategies.The plots are distinguished by large expanses of pastures and forests.Many households in this strategy group require extra labor and pay wages to non-family members to perform various activities.Furthermore, reciprocal unpaid work between members of this group is common.As posesionarios, the households in this group do not receive the benefits that come with being a full member of the ejido, such as the possibility of using the communal area and gaining financial benefits from timber and non-timber forest products.
Group 2 'Day laborers or off-farm workers': This group mostly comprises households of avecindados.People in this group do not possess land, or have very little, which implies that they do not have cattle, crops, provisioning services, or milk or cheese production, among other things.The strategy depends largely on daily wage work and the performance of activities outside the family farm, such as labor on other farms or commercial activities.This strategy is the only option for households that do not have their own farmland.On average, households in this group are involved in fewer types of activities per household (3.7) than households in the other groups (3.9-4.7).Households in this group receive less economic support from the government than households in the other groups.Furthermore, they do not receive the benefits from the management of communal areas, although some of them are paid to work on forest activities (33.3%).However, with permission from ejidatarios or posesionarios, they may access some subsistence resources from the forest, such as firewood.
Group 3 'Private landowners': This group mainly comprises families that live in the ejidos as avecindados but possess private property outside the ejidos, although some ejidatarios and posesionarios can also be found in this group.One of the main characteristics of this group is the large expanses of land, with both forest and pasture for cattle.Some areas are also dedicated to crop production.Of the four groups, this one receives the most government support of all types.The major part of these households' cash income comes from the sale of large quantities of cattle each year.The households in this group are important in terms of overall resource management.On average, the size of the plot per household is by far the largest (131 ha vs. 33 ha (Group 1), 10 ha (Group 2), and 63 ha (Group 4)).An important feature of the plots that households in this group possess is that 50% of the area is covered by forest, which is used for cattle ranching.In addition, these households can engage in commercial forestry or programs in which the government pays for the provision of environmental services.
Group 4 'Diversified ejidatarios': This group is mostly composed of families of ejidatarios, and includes close to 50% of the entire sample.This group uses a more diversified strategy that involves various activities (cattle ranching, agriculture, forest activities, and day labor).The forest activities include the extraction of wild timber species from communal areas under a legally enforceable management plan.The plots the households possess have both forests and pastures and are also used to cultivate crops of diverse types.There is heterogeneity in this group, as it includes households with moderate quantities of land as well as households with land areas that rival those in Group 3. It is worth mentioning that the households in this group complement their strategy with the commercialization of products derived from cattle ranching, such as milk and cheese.In addition, they receive benefits from the ejido related to the management of the communal areas.
The results of the ordination analysis show that, along the continuous axes that summarize the socio-ecological variation in household characteristics (Figure 4), individuals from the same group are closest to each other.There are, however, no very clear-cut divisions; rather, households are located along a continuum.The first PCoA axis differentiates between two strategies: households without cattle, without provisioning services, and commonly having small plots or no land at all (on the right in Figure 4, mostly from Group 2); and households that practice cattle ranching, use provisioning services, and have much larger amounts of land and pasture (on the left in Figure 4, mostly from Groups 3 and 4).The second PCoA axis reveals a striking partition between the two ejidos, with most households from Ranchitos taking positive values along this axis, and most households from Pabelo taking negative values.Other variables that are relevant, but that have less discriminating power, include: the presence of crops in plots, crops that are designated for self-consumption, milk and cheese production, and the payment of daily wages to laborers.Overall, this axis, or the principal components 1 and 2, explains 30.7% of the variation in the data.However, the first five axes, which account for non-random variation in the data, explain 51.1% of the variation (see Appendix 1).Taken together, the results of the analyses suggest that the grouping is strongly associated with the land tenure classes ejidatario, posesionario, avecindado, and private landowner.

Discussion
The results of this study show that households from Pabelo and Ranchitos can be differentiated into groups according to the specific characteristics of their natural resource management strategies despite the heterogeneity in the socio-ecological contexts.We were able to establish a typology of natural resources management strategies composed of four distinct groups: 1) posesionarios with a specialization in cattle raising; 2) day laborers or outside farm workers; 3) private landowners; and 4) diversified ejidatarios.We now discuss each of these groups in light of three factors that we consider to be influential in the shaping of management strategies: 1) land tenure; 2) the role of ecosystems, and 3) the diversification of strategies.

Heterogeneous strategies through land tenure and access to resources
One interesting result of our study is that the household groups that were identified through the cluster analysis match almost exactly with the household land tenure characteristics, suggesting that land tenure constitutes the main driver of the natural resources management strategies observed.This does not mean that having land is the only way for households to develop a strategy, but that heterogeneous strategies emerge as a result of differences in access to land.We found that households in Group 1 are mostly posesionarios without ejidal rights who are excluded from participating in the Assembly and from the use of resources and other benefits of the communal areas.We suggest that, in response to this situation, they have developed alternate mechanisms of interchange, such as reciprocal labor, to complement their strategies.These mechanisms have been documented in other studies in the region (Gerritsen 2001).Avecindados (mostly households in Group 2) have little formal power to manage resources as a consequence of the institutional structure of the ejido and can experience social exclusion (Appendini 2010).Additionally, they do not participate in collective strategies and receive little benefit from communal resources, either directly or from government support programs (payment for environmental services).It has been pointed out that the resilience and adaptability of an ejido institution can be hindered by the inflexibility and often non-participatory governance structure that it maintains (Barnes 2009).However, while ejidatarios have formal authority to make decisions about land resources, non-ejidatarios can form a pressure group and modify decision-making dynamics, especially where they outnumber the ejidatarios (Braña and Martínez 2005;Moya 2012).This ability to modify decision-making dynamics is meaningful in some places in Mexico, particularly in those ejidos where around one-third to one-half of the households are non-ejidatarios (Skutsch et al. 2015).
Observing the four types of strategies within the ordination analysis (Figure 2), we see a strong social differentiation gradient, where towards one extreme we have the strategy of households without land (Group 2) and towards the other extreme, owners of large parcels of land (Group 3).In between these extremes, most households have some land but are either specialized or diversified (Groups 1 and 4).This social differentiation is clearly related to the rights and the access of households to different resources as determined by local institutions (Ellis 1998;DFID 2001).The two ejidos present some interesting differences.In Ranchitos, almost all of the households are ejidatarios and fall into Group 4, and there is a much lower degree of social differentiation than in Pabelo, where there are many families without rights or land, and there are also owners of large parcels of land.In Ranchitos, the younger sons of ejidatarios, and almost half of the ejidatarios, live outside the ejido; this may be due to the low economic returns from farming.A production system characterized by high levels of water scarcity and low productivity has resulted in the younger generation looking for employment options other than landbased labor, outside the ejido.This in turn means that there are fewer posesionarios, avencidados and private landowners.Schroeder and Castillo (2012) have pointed out that the more homogeneous social conditions in Ranchitos could have favored the development of what appears to be a more cooperative institutional structure among ejidatarios in this ejido than in Pabelo.Although it is not the focus of this study, we also found a huge difference between the rights granted to women and the rights granted to men in our study.This is not surprising, since it is the case in most ejidos in Mexico, where women are acknowledged to have very limited rights over land and resources (Barnes 2009;Ruiz Meza 2009).

Strategy diversification as a socio-ecological system outcome
Another interesting result is how the diversification of management strategies in the two ejidos occurs in different ways.In Ranchitos, diversification largely occurs within the household strategy (i.e., most households carry out quite diversified strategies).In Pabelo, a more marked differentiation of strategies and productive specialization exists.In addition, the diversification within each strategy as well as between strategies takes place on different scales and comes from productive activities, sources of income, the landscape, and natural resource uses.Even at the individual and the plot levels, productive activities and conservation practices for vegetation and natural resources are spatially mixed, for example in the riparian forest in Pabelo.This is especially evident in the strategies of the most diversified ejidatarios in Group 4, where there is a balance between marketoriented activities and subsistence activities, together with temporary migration of family members.Livestock, for its part, represents savings and a way of maintaining monetary liquidity (Ugartechea-Salmerón 2015).This diversification at different scales is directly related to the resilience that households have in the face of uncertainty and changes in the socio-ecological system (see Toledo et al. 2003;Barrera-Bassols and Toledo 2005;García-Frapolli et al. 2008;Gerritsen 2010;Garcia-Frapolli et al. 2013), since it provides a range of solutions and resources to deal with different situations (Chapin et al. 2009;Speranza et al. 2014).
Understood in an integrated manner, the different strategies are complementary or co-strategies, which help to mobilize the factors of production or capital (land, work, and capital) within the socio-ecological system (Gerritsen 2004).For example, most of the land is held by ejidatarios or private landowners (mainly Groups 3 and 4), who, in turn, need additional labor from the avecindados (mainly Group 2) to implement their strategy.Furthermore, work outside the community and the migration of family members to the United States are, in all of the strategies, important means to raise capital.However, this migration can have important consequences for natural resource strategies; for example, it results in committing less to agricultural activities, such as maize crops, and increasing the amount of land for pasture (Schmook and Radel 2008;Radel et al. 2010).We can say that, in any specific place, the different strategies are related to each other and the ecosystems that sustain them (Chapin et al. 2009).In this sense, each ejido institution has different spatial, historic, and social relations; it is a unique setting that shapes the social-ecological system.This is an important finding, since most government programs are designed with the logic of 'one-size-fits-all', which ignores social and ecological differences (Brondizio and Le Tourneau 2016;Elinor Ostrom 2009).Enhancing the diversity of institutional settings is considered to be an important way to improve a social system's adaptation to ecological contexts and biodiversity (Ostrom et al. 1999;Janssen et al. 2007;Monroy-Sais et al. 2016).

The extent to which the ecosystem can condition strategies
An important aspect of the strategy classification is the particular ejido to which each household belongs.We suggest that this differentiation is at least partially due to differences in the biophysical and ecological conditions between the two ejidos, which may condition the productive activities that households can implement.Some authors have termed this the 'system of strategies' or the 'farming style', referring to the relationship between the human groups in a specific region and their surroundings, which creates a spatial identity (van der Ploeg 1990;de Haan 2000;Gerritsen 2004;Cochet 2015).The comparison of plots in the ejidos shows that plots in Pabelo have water available throughout the year in permanent streams.Hence, we see that in Pabelo there is a greater tendency towards cattle ranching and a larger area for pasture all year round (Groups 1 and 3).Conversely, the ejido of Ranchitos is found in an ecosystem with high climatic variability, where cattle ranching requires much more effort and investment (financial and human).In fact, livestock farming in many of the Ranchitos households, as Ugartechea-Salmerón (2015) shows, may result in financial losses to those households that have only a few hectares of land and small herds.In addition to this, the forest in Ranchitos has a higher rate of secondary succession than forest in Pabelo (Mora et al. 2016), which means that this growth must be cleared from the pastures every year, implying a larger amount of work.These differences also have important consequences for maintaining a specific amount of biodiversity in the managed sites (Schmitzberger et al. 2005;Casas et al. 2007;Moreno-Calles et al. 2011).
In relation to the use of, and rights over, forests and their resources, each strategy has unique features.For instance, private landowners (Group 3) possess the largest quantity of forest per household (up to 77 hectares), and have more independence in their forest management decisions.On the other hand, ejidatarios (mostly Group 4) have rights over communal forest areas, with collective strategies that can give rise to a large number of benefits and products (Knox and Meinzen-Dick 1999).These are important, since each ejidatario has the equivalent of 38 ha of forest in Pabelo and 26 ha in Ranchitos.Posesionarios (mostly Group 1) only possess the forest remnants that are found on their plots.However, interestingly, they were found to be able to recognize, on average, the largest number of useful plant species, apparently because their restricted access to the communal forests forces them to optimize the management of useful plant species within their own plots.Avecindados (mainly in Group 2) have the least access to forest resources, such as provisioning services.
Here, we see that benefits (economic or of specific resources) derived from forests are not equally distributed among the different strategies and groups, as has been shown in other studies (Morton et al. 2016).It is interesting to see that, in our study, the better-off groups are the ones that benefit the most from forests.This point was also demonstrated in a global study of 8000 households, which showed that the income stemming from natural resources and forests was up to 5 times higher in households with greater resources (Angelsen et al., 2014).

Conclusions
According to our research, the biophysical and ecological conditions, the degree of access to land, and the social position of the head of household regarding rights within the ejido institution are the main variables that determine individual household natural resource management strategies in Pabelo and Ranchitos.Our results show that land tenure difference within communities is one of the main drivers of differentiation in natural resource management strategies, with important consequences for biodiversity and conservation as well as for the economic returns to farmers.This study represents a first attempt to highlight how differences in land tenure inside ejidos may condition natural resource management strategies, but to probe this issue more deeply, further studies with different ejidos would be required.Our two ejidos case studies indicate that ecosystems are strong determinants of strategies, but we are aware that this variable has been found to have little influence in some other studies (Carney 2002;Small 2007;Binder et al. 2013).It is important to recognize that public policies can play a very important, but not necessarily determining role in pushing farmers towards a more diversified strategy or towards a more specialized one.If integrated natural resource management is sought, new socially and environmentally coherent policies and institutions need to be created so that the situation of farmers of all types within the ejidos may be taken into consideration.
ejido Pabelo is located in the municipality of Villa Purificación in the upper part of the Cuixmala and San Nicolas River basins.It was one of the first ejidos founded in the region (in 1938), where previously there had been large private ranches or haciendas dedicated to cattle ranching.It has a population of 1,073 inhabitants and 272 households (INEGI 2010).The ejido Los Ranchitos is located in the municipality of La Huerta in the Chamela river basin and falls within the area of influence of the CCBR.This ejido was founded in 1968 by people who came from other places, supported by a government program for the establishment of new

Figure 1 .
Figure 1.Map of study area, the Chamela-Cuixmala region, the three watersheds and the biosphere reserve.Location of the study sites: the ejido Pabelo and ejido Ranchitos.
crops (b) Hectares of crops (cont) Perform specific season cattle sale (b) Market crops (b) Hectares of forest (cont) Minimum annual sale of cattle (d) Orchard cultivation (b) Pasture ratio (cont) Maximum annual sale of cattle (d) Cultivation of useful plants (b) Crops ratio (cont) Produce milk and cheese (b) Number of useful plants (d) Forest ratio (cont) Sale milk and cheese (b) Use firewood and poles (b) Provision services (b) Perform clearcutting (b) Regulation services (b) Backyard animals (b) Cultural services (b) (b) binary variable, (c) categorical variable, (cont) continuous variable, (d) discrete variable.

Figure 3 .
Figure3.Four groups of the cluster analysis and the main variables that define the divisions between groups.

Figure 4 .
Figure 4. Ordination analysis, differentiating the groups from the cluster analysis.The main variables are also observed on the two axes, these are: Fig A: avecindados; Fig E: ejidatarios; Fig PL: private landowners; Fig P: posesionarios; No Cattle: without cattle; Cattle: with cattle; No Prov: no provision services from plots; Prov: Provision services from plots; Plot Area: total hectares of plot; Grass Area: total hectares of pasture in the plot; No Milk: no milk or cheese production; Milk: milk and cheese production.

Table 2 .
Comparative characteristics of population and territorial extension between the two ejidos.
*Percentages representing the input to this vegetation type from the CA: common area, and PA: plot area.a PF: pine forest, OF: oak forest, POF: pine and oak forest, CF: cloud forest, RF: riparian forest, TSF: tropical semi-deciduous forest, TDF: tropical dry forest.

Table 3 .
Summary of strategies' characteristics of the different groups and the total of the households of the sample from the interviews.
(%) Percentage of households of the sample or group.***, **, * Variable significantly different between at least two groups with a p <.005,.01and.05, respectively.The data between parentheses shows the standard deviations.
the ejido Pabelo there are 164 ejidatarios, 151 posesionarios, and 20-40 avecindados, some of whom are private landowners holding property out- side the ejido.The principal agricultural activity is cattle ranching.There are high levels of migration (mostly to the United States), involving ejidatarios and others, caused by a lack of alternative sources of employment.The ejido has an area of 14,347 ha, of which 8,047 ha are in the form of individual plots and 6,300 ha are communal land.In terms of land use, 6,841 ha are covered by pasture for cattle ranching but were originally forested and 7,506 ha are covered by pine forest, oak forest, mixed pine and oak forest, cloud forest, riparian forest and tropical semideciduous forest, most of which is located in the communal area (Table Ranchitos has a total of 29 households, with a population of 112 people.The original allocation was for 54 rights holders, but currently some ejidatarios have more than one ejidal right, having bought rights from others.Many of the original ejidatarios (20), however, live outside the ejido's territory, either in nearby population centers or the United States.Of the 29 households, three households are of posesionarios, and five are of avecindados (Ugartechea-Salmerón 2015).The elevation is approximately 160 meters above sea level.The annual average precipitation is nearly 750 mm and falls in a very random pattern.Forest areas, mainly tropical deciduous forest, represent approximately 2,580 ha.Approximately 770 ha have been changed from forest to pasture in addition to other uses, including agricultural and urban zones (Table2); the plotted areas retain important forest remnants.The communal area has been divided, in an informal manner, between individual ejidatarios, and is used for cattle ranching and commercial forestry.The earnings from commercial forestry are divided up among the individual ejidatarios, and, in addition, each ejidatario benefits from timber exploitation in his/her forest area.The main activities in the ejido are cattle ranching, agriculture, and forest timber extraction.In addition, local people carry out activities such as fishing, brick-making, hunting, and the extraction of timber and non-timber resources as well as other non-farm activities, such as commerce and construction (Cohen Salgado 2014; Ugartechea-Salmerón 2015).
1,900 ha consist of individual plots and the urban settlement.The communal area is approximately 1,450 ha.