Integrating pattern-based modelling and political ecology in land-use change research: the case of Mexican dry tropics

ABSTRACT

Agricultural and extractive frontiers experiment rapid landscape transformation. Land-Use Sciences and Political Ecology are complementary approaches for analysing how landscape transformations are related to biophysical conditions, and socioeconomic, cultural and political processes developed at global, national and local scales. This study examines such relationships in a long-standing agrarian and resource frontier in southern Mexico for the 1986–2015 period. We combine insights from: (1) a quantitative land use/land cover change pattern-based model, involving a weights of evidence and cellular automata simulations, and (2) a qualitative content analysis of literature and of local actors’ perspectives. Two grand frontier processes have developed in this region: NAFTA-related agrarian transition toward intensification and the establishment of wind farms. Both were triggered by global forces and new forms of land and resources use, but mediated by national-to-local ecological, socioeconomic and political processes, producing particular landscape transformations.

KEYWORDS:

• Agricultural frontiers
• extractive frontiers
• deforestation
• territorialisation
• tropical dry forest
• Mexico

Introduction

Agricultural and extractive frontiers

Landscape transformations are not only among the most significant contemporary global changes, but they are also closely related to other equally worrisome processes like biodiversity loss, climate change and land degradation (Turner et al., 2007). Landscape and societal changes are mutually shaped in the construction of territories through ecological conditions and processes, modes of production, identities, and the configuration of socio-political and economic relations at global, regional and local scales (Bebbington & Batterbury, 2001; Hecht, 2010).

Landscape transformations in agricultural and extractive frontiers have long received scholarly attention (e.g. Bilsborrow & DeLargy, 1990; Carr, 2004; Hecht, 1985). Frontiers have traditionally been conceptualised as areas of rapid socioecological change, as forested regions become the destination of rural-poor population, resulting from State-sponsored or spontaneous migration, or the subject of market-based commodities extraction and production (Barbier, 1997, 2012; Carr, 2004; Pichon, 1997). Frontiers are described as spaces where rural and urban processes intertwine, while new opportunities for capital investment, novel production relations, livelihood diversification, socioecological conflicts, dispossession and complex global to local interactions emerge (Bebbington & Batterbury, 2001; Hirsch, 2009). In developing economies, frontiers have been regarded as having a dualistic nature, with modern and profitable commercial activities coexisting with more traditional and relatively poor agricultural activities (Barbier, 2012).

Instead of portraying frontiers as contact zones between the ‘wild and untamed, the savage and uncivilized’, and the ‘conquered, mastered and civilized space’, Rasmussen and Lund (2018) depict them as spaces where the dismantling of previous territorial orders takes place (property systems, political jurisdictions, rights, social contracts); as a consequence, new orders begin, reshaping patterns of access and control of land and resources, giving rise to processes of territorialisation. From this perspective, frontiers are conceptualised as processes occurring in certain places (Peluso, 2018; Rasmussen & Lund, 2018) and may coexist with territorialisation in space and time.

The emergence of frontiers is associated with changes in land control through enclosures, privatization, and dispossession. New property regimes and territorialisation, in turn, produce new ‘urban-agrarian-natured’ environments, which entail new labour and production processes, actors and subjects that seek hegemony or resist the new order, and networks connecting them (Peluso & Lund, 2011).

Contemporary globalizing political economies, patterns of investment, and movements of labour, capital, and ideas made possible new forms of land control worldwide. Thus, complex socioecological dynamics are shaped by articulations with large-scale processes including public and private investment, links to regional and global markets and the creation of new markets for commodities, migration and remittances, and conservation and sustainability initiatives (Hecht, 2010). However, particular trajectories are produced by situated histories and geographies, the biophysical traits of particular environments, production practices, and local responses (Peluso & Lund, 2011). Therefore, varying landscape configurations and transformations, such as urbanisation, agricultural expansion or primary vegetation recovery are produced.

The complex socioecological interactions involved in landscape change have been addressed from different fields of inquiry, each focusing on one or some of these dimensions. Yet, so far few studies have integrated the long-term entanglements of global, national, and local socioeconomic and political processes with biophysical conditions, and their influence on decision-making over lands at the local level. A more integrated understanding of these dynamics may arise from combining insights from multiple research perspectives, among which Land-Use Science and Political Ecology are prominent (Brannstrom & Vadjunec, 2013; Turner & Robbins, 2008). The goal of this study was to analyse the relationships of landscape transformations with biophysical conditions, and global-to-local socioeconomic and political processes in a long-standing agrarian and resource frontier in southern Mexico, for the 1986–2015 period. To this end, we combine insights from Land-Use Science and Political Ecology. The following section provides a succinct overview of research efforts integrating both approaches, and some of its most salient challenges and possibilities.

Integrating research approaches for studying socioecological systems

Both Land-Use Science and Political Ecology are interdisciplinary fields of research concerned with socioecological phenomena; yet they differ in the study focus, problem framing and explanations, and are linked to different scientific paradigms (Turner & Robbins, 2008). Land-Use Science examines patterns and changes in landscapes, as a basis for evaluating their impacts, and addresses complex multi-scalar dynamics, drivers, and feedbacks of such changes (Chowdhury, 2013; Lesterlin, Castella, & Fox, 2013). This approach is closely linked to global change and sustainability sciences, focussing on land cover dynamics and emphasising the ecological subsystem, the influence of biophysical and ecological conditions, and of socioeconomic phenomena (Turner & Robbins, 2008). In turn, Political Ecology is concerned with the links between social and ecological processes, emphasising how political economy and different interpretations of the environment shape social relations and environmental systems (Bebbington & Batterbury, 2001; Bryant, 2015; Durand, Figueroa, & Guzmán, 2011; Turner & Robbins, 2008). This field of research places environmental change within the context of wealth distribution, asymmetrical power relations and social complexity regarding how material and symbolic struggles shape access to resources and territory (Perz et al., 2013; Walker & Richards, 2013; Wilshusen, 2003). Additionally, Political Ecology progressively contextualises socio-political and ecological processes from local to global scales (Blaikie, 1983).

Efforts towards interdisciplinarity to address socioecological change face numerous challenges given ontological, epistemological and axiological differences between social and natural sciences (Rist, Zimmermann, & Wiesmann, 2004). Moreover, some criticism directed at both approaches come into play. Land-Use Science has been questioned for relying on neoclassical economic theory to interpret individual behaviour, and on the concept of ‘driving forces’ (Brannstrom & Vadjunec, 2013; Robbins, 2012). Also, some Political Ecology scholars criticise Land-Use Science’s superficial understanding of socio-political processes given its insufficient incorporation of social theory, and its tendency to reproduce the nature-society divide when analysing socioecological dynamics (Brannstrom & Vadjunec, 2013; Walker & Richards, 2013). Political Ecology, in turn, has been criticised for the marginal role it gives to ecology and its weak incorporation of insights and methods from the natural sciences. Of particular concern are the lack of ecological theory to conceptualise and measure the specific ecological conditions that shape landscapes, and the ecological consequences of environmental change (Nygren & Rikoon, 2008; Zimmerer, 2015).

Land-Use Science strives to generalise, and thus, by necessity, it simplifies. For this reason, it leads to the identification of general patterns but usually lacks a well-developed understanding of the processes behind those patterns. In turn, Political Ecology often struggles to move from the particular to the generalizable because of its inherent reluctance to simplify and its focus on local-scale and contextualised research (Radel, Schmook, & Méndez, 2013). Weaknesses highlighted for these fields open arenas of complementarity, that in turn avoid the excessive simplification that may be associated with some new multi- and transdisciplinary research agendas (Robbins & Turner 2013; Walker & Richards, 2013). For example, Castree et al. (2014) question the explanation of socioecological dynamics from some global change sciences as they fail to recognise the diversity of values, means, ends, and desired socioecosystemic regimes among different social groups, and hardly incorporate power, violence and inequality in their analyses.

Materials and methods

The seasonally dry tropics and the isthmus of tehuantepec

The Isthmus of Tehuantepec region is dominated by seasonally dry tropical ecosystems (SDTE), which play a crucial role in biodiversity conservation and the continuity of life-support functions (i.e. flood control, climate regulation, freshwater provision, and the maintenance of soil fertility; Maass et al., 2005). These are also fragile ecosystems undergoing high transformation rates (Espírito-Santo et al., 2009; Stoner & Arturo Sánchez-Azofeifa, 2009), particularly resulting from agricultural expansion, and their considerably lower recovery rates, as compared to their humid counterparts (Poorter et al., 2016).

In Mexico, SDTE comprise various vegetation types (Pérez-García, Meave, & Cevallos-Ferriz, 2012), which combined occupy ca. 3.4% of the country’s total land area. It spans in the coastal plains from ca. 28°N to the Guatemala border. On the Pacific watershed, it is distributed almost continuously, whereas on the Gulf of Mexico’s coast, it is more discontinuous (Miranda & Hernández-X, 2014; Trejo & Dirzo, 2000). These ecosystems are characterised by high plant species richness, partly as a result of high beta-diversity and high endemism (Cué Bär, Villaseñor, Monrroe, & Ibarra-Manríquez, 2006; Trejo & Dirzo, 2002)

The geographical setting for this study was the Oaxacan portion of the Isthmus of Tehuantepec, Southern Mexico. The analysis focused on two municipios [administrative units in which Federal states are divided]: Ciudad Ixtepec and Asunción Ixtaltepec (Figure 1). These municipios comprise a total area of 95,194 ha, with altitudes ranging from sea level to 907 m a. s. l. The regional climate is strongly seasonal, with an average total precipitation of 902 mm that is mostly concentrated in a short rainy season spanning from May to October (SMN n.d.). The region is exposed to strong Trade Winds that speed up as they cross the Isthmus due to a topography-related funnel effect (Jaramillo & Borja, 2004).

Figure 1. Study site, a) location of the Oaxacan portion of the Isthmus of Tehuantepec, Mexico, b) location of the municipios included in this study in the state of Oaxaca, c) municipios of Asunción Ixtaltepec and Ciudad Ixtepec.

The two municipios comprise 13 ejidos and one agrarian community (hereafter collectively referred to as agrarian units). Agrarian units are collectively managed. Ejidos are State land cessions derived from the post-Mexican Revolution Agrarian Reform, while agrarian communities are land restitutions to indigenous communities by the State, based on the recognition of their historical rights over the land; yet ejido right-holders may also be indigenous people. The population in the study area belongs mainly to the Zapotec ethnic group. The household economy is highly diversified, encompassing the combination of multiple activities, such as agriculture, hunting, cattle raising, wage labour and commerce (Coronado Malagón, 2009; Rodríguez, 2014).

Vegetation in the two municipios is a complex mosaic that encompasses tropical dry forests, savannas, pine and oak tropical woodlands, thorn forests, and semi-evergreen seasonal forests (Calzada, Meave, Bonfil, & Figueroa, 2018; Pérez-García et al., 2012). Ciudad Ixtepec has a large settlement of ca. 27,000 inhabitants (INEGI, 2011), while Asunción Ixtaltepec is less populated and encompasses scattered settlements and large areas dominated by induced pastures, often established at the expense of savannas (Figure 1(c)). Conversion of primary and secondary vegetation to crops and pastures has increased between 1986 and 2015; however, a portion of the latter returned to secondary or primary vegetation after being used for agriculture (at least 5%, in each 5-year period of analysis, Calzada et al., 2018). This pattern results from local agricultural practices, common in other tropical regions, characterised by agriculture-fallow periods that create cyclic pathways in land-use dynamics (Wright, 2005; Burgos & Maass, 2004). Such pathways favour the partial or total recovery of vegetation through ecological succession (Lebrija-Trejos, Bongers, Pérez-García, & Meave, 2008; Lebrija-Trejos, Meave, Poorter, Pérez-García, & Bongers, 2010), where fallow lands play a key role in soil and biodiversity conservation, and in households’ economy (Toledo et al. 2008; Chazdon et al. 2016).

Pattern-based modelling strategy

We used land cover maps produced through the supervised classification of Landsat images (for detailed procedures, see Calzada et al., 2018), corresponding to 1986, 1997, 2004, 2011 and 2015, in which 12 classes were distinguished. We developed a pattern-based model in the DINAMICA EGO software (v. 4.0) and used a cellular automata algorithm to calculate the weights of evidence for independent variables. The model was evaluated by comparing the resulting transition probability maps with the supervised classification (Soares-Filho et al., 2004; Soares-Filho, Cerqueira, & Pennachin, 2002; Weed, 2005). We calculated weights based on the transition frequency from one land use to another, for the distribution range of each independent variable. Selected variables were: soil and bedrock type; topography (slope, elevation, and slope orientation); climatic variables (mean total annual precipitation, mean annual temperature, and mean total annual potential evapotranspiration); agro-productive variables (access to training, credit, and insurance, use of agrochemicals, and market/subsistence production); agrarian unit identity, intra-community zoning (community agreements dividing their territory into common- and individual-use areas), distance to pre-existing agriculture and cattle raising areas, and distance to roads and water bodies. We integrated wind farms and Voluntary Conservation Areas to the model only for the 2011–2015 period, given the year of their establishment. Variable selection depended on previous research regarding drivers of agriculture expansion, but data availability was a significant constraint in this process. A description of analysed variables, evaluation intervals for each variable, previous geoprocessing and data sources are provided as supplementary material (Table S1).

The importance of independent variables for explaining land use/land cover change (LULCC)¹ is based on the value of weight of evidence in the first and last quartile, considering all values (Supplementary material, Table S2). Thus, if a variable showed weights of evidence near the median, it was considered as non-important. In turn, if values appeared only in either one of both quartiles, they were considered important. Finally, if a variable had weights of evidence in both quartiles, it was considered as being important, but with an ambivalent influence on LULCC; that is, it may foster or limit change depending on its own range of values. For example, high elevations may limit agricultural expansion, while low elevations may favour it. Annual transition rates (i.e. the proportion of surface that changed from one year to another) were calculated, based on areas derived from land cover simulated maps (Supplementary material, Table S3). The cellular automata model considered all transitions among the original 12 cover classes, and all variables, but we only analysed here those transitions representing the most important processes: agricultural expansion, the transition of primary to secondary vegetation and vice versa, abandonment of fields, and urbanisation.

Analysis of socio-political processes

Two of the main strains of analysis within Political Ecology are particularly useful for framing our study: Marxist political economy (structural analysis), and actor-centred political analysis (Durand et al., 2011; Robbins, 2012; Wilshusen, 2003). For structural analysis, environmental change is determined by political-economic structures. Power relations are incorporated through these structures in the form of productive and labour relations, commodity markets, legal frameworks or public policies. For the actor-centred approach, environmental change is shaped by the negotiation and confrontation of interests, needs and views of diverse social actors (e.g. peasants, social organizations, government institutions, business) struggling to control the environment and its resources (Peet, Robbins, & Watts, 2011; Robbins, 2012). According to Bryant (2015), in these ‘politicised environments’, asymmetric power relationships give rise to a differential distribution of costs and benefits derived from decision-making processes. However, agency is central to this approach, as subordinated actors are not passive recipients of environment-development interventions and decisions; instead, they reinterpret and modify them in response to their own needs and perspectives (Long, 2007).

Investigating the processes influencing peasants’ decisions involved literature research on the regional history and recent socioeconomic and political processes. This step was supplemented by governmental statistical data, which allowed constructing the historical context, detecting the most salient socioecological events for the period 1986–2015, and establishing the links of local to global processes.

We analysed peasants’ perspectives on landscape change and management, and those events perceived by them as being relevant for their present and past decisions. Also, we examined the perspectives of government officials regarding the implementation of environmental and agricultural public policies. We carried out 48 semi-structured interviews to peasants and 13 to government officials during five visits to the field conducted between 2015 and 2017. In addition, we held discussions with two focus groups in the agrarian units of Nizanda and Ciudad Ixtepec, in order to understand local perspectives on the history of production and events that were relevant for peasants’ decisions. Focus groups included 11 people in Nizanda and five in Ciudad Ixtepec. They were all male land right-holders from agrarian units. All interviews were recorded and transcribed textually. We carried out content analysis using the Atlas.ti software (v. 7.5.4). Quotations were labelled with five different codes: environmental transformations, economic activities, socio-political relations, public policies, and perspectives about forest conservation. These quotations were analysed in search for patterns and diversity of perspectives (Sautu, 2005).

Results

This section opens with a brief historical context of the region, after which a description of the general patterns of LULCC is presented along with the factors that, according to the model and the interviews and focus groups, were important throughout the study period. We then offer an integrated analysis of factors and processes influencing decision-making and LULCC, divided in four chronological periods, derived from all sources of information (relevant variables for that particular lapse derived from the model, and from interviews, focus groups, and literature). Figure 2 depicts a timeline showing the annual simulated rates of change for each cover class and the most salient factors and processes involved in landscape transformation.

Figure 2. Timeline integration of simulated annual change rates for each land cover class. Primary vegetation (dark green), secondary vegetation (light green), agriculture (yellow) and human settlements (grey). Analysed factors and their occurrence in relation to land cover transitions through time are shown. For factors analysed with the quantitative model: (+) favour transitions, (-) limit transitions, (±) limit or favour transitions depending on their values, (0) no influence. ADVC: Voluntary conservation areas.

The historical context of a frontier

The Isthmus of Tehuantepec region has long been recognised as an agricultural and extractive frontier. Since the mid-19th century, it has been a geostrategic region for economic growth and the consolidation of the Mexican State (Hernández, Leonard, Hoffman, & Prévôt-Schapira, 2009). It has been subjected to multiple development interventions, largely due to its condition as the narrowest land connecting the Gulf of Mexico and the Pacific Ocean in the Mexican territory, with the consequent potential for commerce, transportation, and economic growth (Rodríguez, 2014).

The construction of the Trans-Isthmic Railroad marked the entrance of the region into the realm of global interests, including land speculation of areas near the railroad, which gave way to dispossession. Nonetheless, the railroad also articulated the production of local communities with larger-scale markets (Hernández et al., 2009). This economic dynamism, also fostered by governmental incentives, encouraged the development of many economic activities, such as agriculture and cattle raising, particularly since the late 1960s. Public investment in agriculture and livestock during the 1950–1985 period consolidated the client-based relationship of peasants and productive organisations with the Partido Revolucionario Institucional, the political party that ruled Mexico uninterruptedly until the year 2000, increasing its regional political control (Léonard, 2009).

Regional socioecological dynamics have also been shaped by the development of the petrochemical industry. Since the 1970s, with the construction of an oil refinery in Salina Cruz, the development of the petrochemical industry began to influence the region. Projects multiplied in other municipios and, as a result, wage labour became available, so that temporary emigration to those sites became an economic alternative (Prévôt-Schapira, 2009). More recently, the establishment and expansion of wind farms inaugurated a new cycle of transformations on resources use, capital investment and social change (Jara Castillo, 2012; Martínez-Laguna, Sánchez-Salazar, & Casado Izquierdo, 2002).

General patterns of land use and land cover change (1986-2015)

Landsat imagery showed that the most important LULCC observed in every period was the replacement of primary and secondary vegetation by agriculture and induced pastures. Agricultural terrains accounted for 15.9% (15,146.1 ha) of the total combined area of Ciudad Ixtepec and Asunción Ixtaltepec in 1986, but by 2015 they covered 24.1% (22,932 ha). Induced pastures increased from 3.3% (3,159.4 ha) to 3.6% (3,472.3 ha) of the total area, with the most important growth in the 2011–2015 period (1,755 ha). Primary tropical dry forests maintained the largest cover throughout the study period but decreased from 34.2% (32,516.7 ha) to 28.9% (27,513.4 ha). Vegetation types with the highest cover loss were tropical dry forests (−5,003.3 ha) and thorn forests (−3,301.5 ha). Savanna was the only vegetation cover that increased from 12.2% (11,654.6 ha) to 13.4% (12,711.7 ha). Secondary vegetation decreased from 5.3% (5,037.9 ha) to 3.5% (3,341.9 ha), although it increased slightly between 1997 and 2004, when its percent cover reached 5.4% (5,099.1 ha). Finally, human settlements almost doubled in extent, covering 1.7% (1,577.1 ha) in 1986, and 3.1% (2,950.8 ha) in 2015 (Table 1). Gains and losses for each period and cover class are depicted in Table 2. For a detailed LULCC pathways description, see Calzada et al. (2018).

Table 1. Area and percentage of total area for each cover class in the municipios of Ciudad Ixtepec and Asunción Ixtaltepec, Oaxaca (1986–2015).

	Surface
	1986		1997		2004		2011		2015
Cover class	ha	%	ha	%	ha	%	ha	%	ha	%
Tropical dry forest	32,516.7	34.2	31,386.8	33.0	30,301.3	31.8	28,939.5	30.4	27,513.4	28.9
Thorn forest	9,578.6	10.1	9,080.1	9.5	8,763.1	9.2	7,376.8	7.7	6,277.1	6.6
Semi-evergreen seasonal forest	4,697.6	4.9	4,691.4	4.9	4,662.8	4.9	4,664.5	4.9	4,518.8	4.7
Evergreen seasonal forest	1,723.0	1.8	1,723.4	1.8	1,720.7	1.8	1,716.5	1.8	1,716.5	1.8
Savanna	11,654.6	12.2	11,630.7	12.2	12,750.4	13.4	12,754.6	13.4	12,711.7	13.4
Oak woodland	7,676.0	8.1	7,252.5	7.6	7,276.6	7.6	7,588.5	8.0	7,314.0	7.7
Pine woodland	1,189.8	1.2	1,189.8	1.2	1,189.8	1.2	1,189.8	1.2	1,189.8	1.2
Secondary vegetation	5,037.9	5.3	4,656.9	4.9	5,099.1	5.4	3,639.7	3.8	3,341.9	3.5
Agriculture	15,146.1	15.9	17,080.3	17.9	18,095.7	19.0	21,531.2	22.6	22,932.0	24.1
Induced pastures	3,159.4	3.3	3,384.1	3.6	1,480.8	1.6	1,932.9	2.0	3,472.3	3.6
Bare ground	1,237.5	1.3	1,316.3	1.4	1,525.2	1.6	1,337.7	1.4	1,256.0	1.3
Human settlements	1,577.1	1.7	1,802.1	1.9	2,328.9	2.4	2,522.6	2.6	2,950.8	3.1
Total	95,194.3	100.0	95,194.3	100.0	95,194.3	100.0	95,194.3	100.0	95,194.3	100.0

Table 2. Surface gains and losses for each cover class, study period, and overall and percent change, in the municipios of Ciudad Ixtepec and Asunción Ixtaltepec, Oaxaca (1986–2015).

	1986-1997		1997-2004		2004-2011		2011-2015		Overall change
Cover class	Losses (ha)	Gains (ha)	Losses (ha)	Gains (ha)	Losses (ha)	Gains (ha)	Losses (ha)	Gains (ha)	ha	%
Tropical dry forest	−1,489.0	359.1	−1,676.9	591.4	−1,861.2	499.5	−1,612.6	186.6	−5,003.3	−5.3
Thorn forest	−860.0	361.5	−740.2	423.2	−1,668.9	282.7	−1,136.5	36.8	−3,301.5	−3.5
Semi-evergreen seasonal forest	−6.3	0.1	−32.0	3.3	−3.1	4.8	−153.0	7.4	−178.8	−0.2
Evergreen seasonal forest	0.0	0.5	−2.8	0.0	−4.2	0.0	0.0	0.0	−6.5	0.0
Savanna	−29.7	5.8	0.0	1,119.7	−15.7	19.8	−59.5	16.5	1,057.1	1.2
Oak woodland	−424.4	0.9	−134.2	158.3	−22.9	334.8	−274.5	0.0	−362.0	−0.4
Pine woodland	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
Secondary vegetation	−2,295.7	1,914.7	−2,418.5	2,860.7	−3,424.6	1,965.3	−1,610.7	1,313.0	−1,696.0	−1.8
Agriculture	−598.5	2,532.7	−1,802.4	2,817.8	−462.1	3,897.4	−1,402.3	2,803.1	7,785.9	8.2
Induced pastures	−1,273.1	1,497.8	−2,378.1	474.9	−365.6	817.7	−216.5	1,755.8	312.9	0.3
Bare ground	−95.6	174.4	0.0	208.9	−187.5	0.0	−176.5	94.8	18.5	0.0
Human settlements	0.0	225.0	0.0	526.8	0.0	193.7	0.0	428.2	1,373.7	1.4
Total	−7,072.4	7,072.4	−9,185.1	9,185.1	−8,015.7	8,015.7	−6,642.1	6,642.1	–	–

According to the model, the variables ‘soil type’, ‘distance to agriculture’, ‘distance to pastures’, and ‘agrarian unit ID’, were relevant throughout the 29-y lapse. Also, ‘rock type’ ‘mean temperature’, ‘mean precipitation’, ‘slope’, and ‘distance to settlements’ were important in three, out of the four time-lapses in which we divided the study period (Figure 2). Some of these results were confirmed by most of our interviewees, who expressed that the main problem for their communities is the lack of water, and that agrarian units contrast in environmental conditions, economic activities, and decisions taken over the use lands.

Structural adjustments and the growth of agro-industrial production (1986-1997)

For most of this period, there was a low and stable increase in agriculture and secondary vegetation at the expense of primary vegetation; however, between 1996 and 1997 the rate urbanisation and secondary vegetation recovery grew substantially, while there was a decrease of agricultural surfaces (this trend lasted until 2000). Our model shows that landscape dynamics were associated with ‘access to credits’ and with traditional land management in the form of ‘intracommunity zoning’ (i.e., the way in which each community and ejido establishes areas for collective use, productive plots and conservation areas).

According to our interviews and focus groups, at the beginning of the period, traditional subsistence agriculture and firewood commercialisation by peasants were predominant activities in these communities, whilst livestock and agro-industrial production were still incipient. Communities were connected by dirt roads, which were difficult to drive on in the rainy season, and people depended heavily on the train for transportation. However, wage labour was available in the region and enabled the improvement of peoples’ living standards; they could, for example, afford education for their children, and invest in agriculture intensification, although by the end of the period, their livelihoods depended less on the product of their lands.

The Pacific Oil Project (Proyecto Petrolero del Pacífico) constructed oil ducts and pipes to transport crude oil, distilled oil products, ammonia, and natural gas in the region. By the early 1990s, the State-owned Mexican Oil Company (PEMEX; Petróleos Mexicanos) became the third-largest petrochemical industry worldwide. Its growth attracted large-scale investments to the region, like the construction of an isomerizing plant in 1995, and the expansion of the refinery in 1997, both of them in Salina Cruz (Rodríguez, 2014). Thus, the Isthmus of Tehuantepec became strategic for satisfying national and international energy demands (Prévôt-Schapira, 2009). This wave of investments also induced the expansion of paved roads, development of other industries, construction of ports, and extractive development, such as metallic and non-metallic mining (Rodríguez, 2014). Although our interviewees believed that roads accelerated forest clearing, our model did not detect the influence of roads on landscape change during this period.

Industrial and extractive development transformed the region in various forms. According to our interviewees, the duct network was constructed underground across agrarian units and individual plots within them. This caused the temporary clearing of forest along these paths. Although this network is underground and agricultural production is still possible, PEMEX pays a due for crossing the plots since then. These payments stimulated forest clearing further, as higher financial compensation was assigned to plots already under use, in comparison to those covered by forests. The same phenomenon was associated with the Federal Energy Company (CFE; Compañía Federal de Electricidad) spanning transmission lines, as declared by an interviewee: ‘ … if it is a cultivated plot, yes [you receive the payment], you let them see that you are working it daily’. Moreover, the way peasants valued and controlled their lands changed, as one of them stated: ‘Long ago, nobody would let anyone place anything in their plots; today only money matters, not like before.’ Besides the effect of the duct network, petrochemical projects affected access to water through stocking large volumes of water, thereby restricting irrigated agriculture.

In the early 1990s, structural adjustments associated with the North American Free Trade Agreement (NAFTA) were put into place, affecting agriculture and land tenure nationwide. Resulting Constitutional legal changes permitted agrarian units to grant concessions for extractive projects. Also, the selling of ejido lands became possible, which initiated substantial transformations of common property and natural resource management (de Appendini, Calva, Quintana, García, & Gómez Cruz, 1993). Interestingly, our model showed that intracommunity zoning, related to decisions over lands, ceased to be important after 1997. Industrial agriculture and cattle production for export were favoured, whilst State support for subsistence and small-scale production diminished importantly, with the disappearance of credits, training, and commercialisation support. Producers faced market liberalisation without adequate conditions for competing and became vulnerable to shifting international commodity prices. For many, migration or agriculture abandonment were necessary, while for others extensification was the alternative. For those with State support or assets, intensification using agrochemicals, mechanization, and the establishment of monocultures was the solution (Calva, 1996; Rodríguez, 2014).

Structural reforms, agro-industrial consolidation and urbanisation (1997-2005)

In this period, our model showed that LULCC trends changed. Between 1997 and 2000, agriculture decreased, while secondary vegetation grew. This trend was reversed by 2000, when agriculture slowly began to recover, mainly at the expense of secondary vegetation. In 2003, there was a strong increase in agriculture expansion rates, at the expense of both primary and secondary vegetation. During all these years, urbanisation grew rapidly. For this period, relevant variables explaining landscape transformation were ‘access to training’, ‘distance to water bodies’ and ‘mean potential evaporation’. Interestingly, interviewees stated that between 1995 and 1997, there was a strong drought.

These changes coincided with a nation-wide rural crisis derived from the structural changes implemented in the mid-1990s, affecting rural subsistence production (Calva, 1996). Accordingly, our model showed that, for this period, the variable ‘access to credits’ ceased to be important. Nationwide, the PROCAMPO program (Programa de Apoyos Directos al Campo, rebranded later as PROAGRO) was implemented. The program was originally intended to support peasants in transition and help them become competitive producers in a free market. Yet, for most of them this support was insufficient (Calva, 1996). According to our interviews, this program began in the region in 1997. Also, the PROCEDE programme was put into place to demarcate individual plots, so that ejido lands could be managed independently by individual owners instead of collectively (Rodríguez, 2014), although many agrarian units retained a portion of their lands as collective.

Interviews and focus groups indicated that the interruption of local commodities’ commercialisation through the railroad aggravated the rural crisis. Even though roads were paved, the former commercial networks were disarticulated. The PROCEDE programme was implemented in the region in 2003, which coincides with an abrupt increase in agriculture expansion, observed from the Landsat images. According to interviews and focus groups, by the end of this period agro-industrial production had gained importance, particularly the production of sorghum, as a result of public policies for agricultural intensification. Also, the first tractors appeared in the study site. The agricultural crisis fuelled migration to cities, which offered wage labour, explaining partly urban growth. Migration, urban-based wage labour, and financial flows back to rural communities also allowed peasants to invest in agrochemicals and mechanisation, furthering agricultural intensification.

Agriculture intensification and the crisis of subsistence production 2004-2011

Analysis of Landsat images shows that, after agricultural expansion peaked in 2003–2004, there was a slight but constant decline of this activity, mirrored by a recovery of secondary vegetation. Also, urbanisation rate decreased substantially (Figure 2). Our model showed that biophysical variables were important (‘soil type’, ‘rock type’ ‘mean precipitation’, ‘slope’ and ‘elevation’). Additionally, ‘distance to agriculture’ and ‘distance to pastures’ were relevant. For the first time, ‘access to markets’, ‘use of agrochemicals’ and ‘distance to roads’ influenced landscape transformation. Precisely during this period, according to our interviews and focus groups, intensive agriculture strongly developed with the introduction of sorghum, tractors and technological packages. Even though PROCAMPO and PROCEDE programmes were still in operation in this period, the growing dependency on global markets furthered peasants’ vulnerability to changes in global market prices of agricultural commodities.

Another salient transformation regionally was that, in 2003, Voluntary Conservation Areas (Áreas Destinadas Voluntariamente a la Conservación-ADVC) appeared in agrarian units. These areas became part of community territorial management, protecting lands from grazing and agricultural expansion. These are small-sized areas traditionally conserved and managed by local communities (See Kelly, this issue), that were recognised as a new conservation category by the National Commission for Protected Areas (CONANP; Comisión Nacional de Áreas Naturales Protegidas; www.conanp.gob.mx). This designation allows communities to access federal funds for management activities, provided they develop a management plan. However, ADVC management remains in the hands of communities. According to our interviews and focus groups, community assemblies approve the conservation and management of these tracts of primary vegetation, and implement rules restricting their use (i.e. only extraction for subsistence purposes, with previous assembly permission). These rules are largely (but not always) respected. These areas enable the conservation of resources that are important for rural households. Moreover, most interviewees shared the conviction that the remaining primary vegetation protects water availability, one of the factors most limiting of their economic activities.

Finally, agrarian units previously characterized by the predominance of agriculture, began to diverge into various economic activities. For example, Ciudad Ixtepec and Nizanda have currently remained largely as agriculture-based economies, whereas Mazahua is involved in non-metallic mining, and Ciudad Ixtepec, Asunción Ixtaltepec and La Mata, in wind-energy production.

Wind rights and community conservation (2011-2015)

Landsat imagery revealed an abrupt agricultural and urban expansion at the beginning of this period. Contrasting with previous years, throughout this time span, the highest rate of urban growth took place; both agriculture and settlements expanded at the expense of primary vegetation. Our model showed that biophysical variables (‘soil type’, ‘slope’, ‘mean precipitation’, ‘mean temperature’, ‘mean potential evaporation’) were relevant. Also, ‘distance to settlements’ ‘distance to agriculture’, ‘distance to pastures’ and ‘distance to wind turbines’ were important. Finally, the use of agrochemicals, access to markets, agrarian unit ID, presence of ‘ADVC’ and cattle were significant for explaining landscape transformation.

According to our interviews and focus groups, agro-industrial production became dominant, yet some subsistence agriculture persists until today, like milpa (traditional subsistence production of maize, beans, squash, among other products). Most producers are involved in commercial intensive crops, such as sorghum. Livestock has been fostered by earnings derived from wage-labour. Free-grazing takes place on primary or secondary vegetation or in induced pastures; however, in bad years when the harvest is lost, cattle are raised on leased land, formerly used as agricultural fields. The majority of the population is involved in these primary activities in combination with wage-labour and commerce. PROCAMPO is still operating, and it stimulates sowing even when peasants suspect that climatic conditions in a given year may lead to a failed harvests: ‘The problem for peasants is that here, in this region, there is very little rain and in many years the harvest is lost (…) we are impelled to sow because they give us the support of PROCAMPO’.

Interviews also show that, today, many peasants are pensioned either by Ferrocarriles Nacionales (the railroad State-owned company) or by PEMEX but continue involved in their productive activities. As stated by one interviewee: ‘Now we do not need support [governmental or from remittances], I am pensioned by PEMEX, and from there [I obtain] the money’. Most young people are not involved in agriculture and cattle raising, as they find opportunities to continue studying and working either in the region or elsewhere, mostly in oil and petrochemical production, construction of infrastructure, and more recently in wind farms.

The establishment of wind farms was fostered by the international demand for renewable energy resources, combined with the particularly favourable atmospheric conditions in the Isthmic region (Martínez & Gómez, 2017 – see Lehmann & Backhouse, this issue). For most interviewees, the arrival of wind energy production turned out to be one of the most salient transformations in the last 20 years. These private projects offered job opportunities but also attracted a new population. Conflicts derived from wind farms were associated with the loss of control over territories, the lack of generalised benefits (such as local access to cheaper energy), and the growing social stratification of local society, since the owners of plots where wind turbines are placed receive considerable high payments, considering peasants’ economy.

According to one interview, acceptance of these projects has involved coercion, as the opportunities to work are closed for those who show opposition: ‘(…) if you are leftist and do not comply with how they are working, you cannot work anymore (…) at the beginning (…) I wanted things to be done rightly [but] you have to adapt, so that you can work; otherwise, you are just left behind.’ In contrast, for those who receive these payments, wind turbines are equivalent to ‘winning the lottery’ and represent a financial insurance for the rest of their lives. In some cases, there is a payment for the whole agrarian unit (derecho de viento or wind rights), but this is not always the case. For example, in Asunción Ixtaltepec, there are no wind rights, because intracommunity conflicts and disorganisation prevented collective bargaining with wind-energy companies. In contrast, in La Mata, collective organisation achieved individual payments and wind rights for the whole agrarian unit. Sometimes wind-energy companies invest in infrastructure, as one interviewee commented: ‘(…) they have paved streets, constructed a sports field (…) and a library for the school with many computers’, and their presence has also represented a more local source of employment for a younger generation. Therefore, different perspectives on wind turbines exist, confronting the individual and collective values.

According to interviews, the capitalisation of peasants resulted sometimes in the abandonment of agriculture, because peasants can now rely on a steady income source; yet, others have invested in the intensification of production, particularly of sorghum. Additionally, wind farms are transforming the control of lands, that was previously rooted in the assemblies of agrarian units, where all right holders were involved in making decisions over collective lands. In La Mata and Asunción Ixtaltepc assemblies are called on only when there are conflicts to be solved, whereas in other agrarian units, such as Nizanda, where wind farms are absent, assemblies are carried out on a monthly basis. The relatively high payments derived from wind turbines have favoured individual control over lands, and negotiations regarding wind-energy turbines and payments are being held directly between individual peasants and transnational companies in some agrarian units, where assemblies no longer influence these decisions. The owners of plots where wind turbines are placed, in turn, decide who is employed to work in wind farms, creating a new broker elite supported by client-based relationships.

Discussion

Frontier and territorialisation processes are shaped by historic, geographic and environmental contexts (Peluso & Lund, 2011). They are produced by the articulation of local socio-political and ecological dynamics with national and global socioeconomic and political forces (Hecht, 2010). We detected two salient frontier episodes in the Isthmus of Tehuantepec. The first was triggered by the signing of the NAFTA agreement, and the consequent structural adjustments (legal and institutional changes), along with global energy demands and a strong wave of investments. These forces provoked simultaneously the growth of industrial development and the availability of wage labour, which contained population pressures over land, and pushed a transition towards the intensive production of commercial crops. Structural adjustments included the flexibilization of land tenure for ejido lands, eroding collective control. This change would later allow for the establishment of mines and, more recently, of wind farms. This is an example of how dismantling of previous territorial orders and the emergence of new territorialisation processes evolve within frontiers (Rasmussen & Lund, 2018).

The consequences of NAFTA were felt in the region long after the 1990s. Aggravated by a severe drought, the lack of access to credits and other forms of State support compounded the crisis of traditional subsistence agriculture. Consequently, agricultural cover diminished until the year 2000, and emigration increased. Technological packages and access to markets propelled intensive industrialised agriculture, hence fostering the consolidation of an agricultural elite. However, industrialised agriculture was confined to lands with certain biophysical characteristics (flatlands close to roads and settlements). The implementation of PROCEDE incited the individual management of lands and allowed for concessions, which in turn are related to post-2000 agricultural expansion and to the economic diversification of agrarian units. Finally, the emergence of ADVC, which have been effective to deter LULCC, was possible given the existence of large tracts of well-preserved forests. The availability of wage labour, which reduced pressure on lands, biophysical constraints that limit the expansion of agriculture and cattle, and culturally driven conservation are all part of this possibility.

The second frontier process is associated with global-national energy demands. Previously met by PEMEX and CFE, this demand now is centred on private renewable energy production. The arrival of wind farms transformed the region in various forms. Social conflicts, immigration and the associated urban growth took place, and a new elite emerged. Also, agricultural intensification strengthened, as propelled by access to markets and agrochemical use. Social asymmetries grew and the collective control of lands was further eroded.

The landscape in the Isthmus of Tehuantepec possesses many features that reflect past and present frontier and territorialisation dynamics. Landmarks are visible: the railroad, PEMEX ducts, CFE towers, highways, and wind-turbines. Despite these large-scale socioeconomic transformations, the landscape is still characterised by large tracts of well-conserved STDE. Deforestation rates in tropical dry forests vary considerably across regions (Sánchez-Azofeifa, Arturo Sánchez-Azofeifa, & Portillo-Quintero, 2011). However, the rates we observed are lower than most of those reported for other frontier regions (Bianchi & Haig, 2013; Gasparri & Grau, 2009; Killeen et al., 2007; Trejo & Dirzo, 2000). The lower rates of change result from biophysical features that impose limits to agriculture and livestock, given the low precipitation and high evaporation rates, and from regional economic diversification and cultural local characteristics.

Frontiers are produced through new forms of resource use and exploitation (Peluso & Lund, 2011) associated with capital accumulation and commodity consumption at various scales. Frontier dynamics and territorialisation have transformed land use and attracted diverse social actors (see Lehmann & Backhouse, this issue): from land speculators when the railroad was constructed, to governmental institutions, such as CFE and PEMEX, or more recently the CONANP and wind-energy companies, immigrants and workers.

Poverty in the region has decreased in the last 20 years; yet, social stratification has worsened due to the asymmetrical distribution of the benefits derived from these transformations; this has fuelled social conflicts and transformed social relations. Moreover, the ongoing erosion of collective land control may lead to land privatization and higher rates of deforestation (already observed, particularly since the PROCEDE programme granted individual formal rights to plots in ejidos). Most benefits derived from new forms of land and resources control flow out of the region and concentrate elsewhere, whereas social and ecological consequences are assumed locally.

Local actors, however, are not passive recipients of policies, but rather they re-interpret and shape, to some degree, the implementation of policies through opposition and negotiation (Long, 2007; McLaughlin & Dietz, 2008). In the Oaxacan Isthmus, social movements have contested wind farms and mining projects. Also, some agrarian units have maintained social networks, voluntary community work, solidarity relations, and community control over their territory, including the protection of conserved forests. The societal transformation has taken place, but some of these communities maintain their identities and the possibility of collective action.

Integration of research perspectives

There is ample evidence in the literature of the enormous potential of engaging Land-Use Science and Political Ecology. Yet, many scholars move between these fields without reflecting on the epistemological implications and are led by the general sense that all approaches are partial and no single one will ever produce a fully satisfactory explanation of LULCC and the underlying human-environmental relations. It is difficult to envision a full fusion between both approaches, but this should not restrict traffic across ‘friendly, porous, intellectual borders’, as mutual learning and conversation are possible (Robbins & Turner 2013). Triangulating between approaches, recognising the strengths and weaknesses of each one, and drawing upon them, derive in a better understanding of environmental phenomena (Chowdhury, 2013; Radel et al., 2013).

Integrated Land-Use Science and Political Ecology efforts have advanced explanations about some of these processes (e.g., Reid et al., 2000; Vásquez-León & Liverman, 2004; Meyfroidt et al., 2010; Radel et al., 2013), but few studies integrate scales from the pixel to global processes, and from the biophysical and ecological realm, to social, cultural and economic dimensions. In this study, we were able to disentangle LULCC complex articulations in a frontier context, by integrating ecological, social, economic and cultural processes developing at local (from the pixel to the community), national, and global scales. We were also able to detect long-term effects of global and national processes in the region, through simultaneous synchronic and diachronic analyses.

Conclusion

As shown here, frontiers are characterized by the simultaneous dilution and creation of forms of control of people and resources (Rasmussen & Lund, 2018). However, how exactly these phenomena take place along with their trajectories and expressions in the landscape (i.e. land cover changes) is shaped by productive practices (i.e., the transition from subsistence to intensive agriculture), biophysical and ecological conditions and processes (i.e., climatic conditions or the ecological structure and function of SDTE), and the evolving articulation of local social, political, cultural processes, as well as institutions with national (i e. structural adjustments) and global political–economic structures and forces (i.e. global energy demands or international trade agreements).

Frontiers and territories in other regions may have their own particular trajectories, given local and regional-specific conditions and processes. However, similarities may be found given the shared global context. Numerous research questions arise from the possibilities of examining similarities and dissimilarities among different regions in the world. Moreover, a more comprehensive integration is pending, as ecological processes are not usually fully accounted for in Political Ecology-Land Use Science studies. The dynamics of ecosystems, particularly, vegetation dynamics, productivity, resilience, or recovery trajectories, and their links to productive processes and uses of lands should, in the future, should be incorporated as part of these more complete and in-depth explanations. We embrace Radel et al. (2013)’s plea for an approach to knowledge of ‘humble partiality’, where ‘we can embrace the imperfect and partial nature of all approaches and knowledge positions’; indeed, the potential collaboration among fields of research is thrilling.

Acknowledgments

This study was funded by DGAPA, UNAM through PAPIIT, grant IA205216. LC was the recipient of a graduate scholarship (CONACYT no. 607891). We are grateful to Carolina García, Víctor Rangel, María Eugenia Romero, Daffne Jasso, María Cecilia Sosa for their support in the fieldwork and in the transcription of interviews. We are also in debt with José Paruelo for the seminal idea for this manuscript. Also, we thank all the people in the communities of Ciudad Ixtepec and Asunción Ixtaltepec who shared with us their perspectives and thoughts with us. Finally, we are grateful for the comments made to the earlier versions of our manuscript by the reviewers and editors of this Special Issue, which substantially improved it. This article belongs to the forthcoming Special Issue on “Transdisciplinary perspectives on current transformations at the extractive and agrarian frontiers in Latin America”, edited by Anne Cristina de la Vega-Leinert and Regine Schönenberg.

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary material

Supplemental data for this article can be accessed here.

Additional information

Funding

This work was supported by the Dirección General de Asuntos del Personal Académico, Universidad Nacional Autónoma de México [PAPIIT / IA205216].

Notes

1. We understand land use and land cover change as all human modifications of Earth’s terrestrial surface. Land cover refers to all physical and vegetation cover present over the Earth’s terrestrial surface, including water bodies, forests, grasslands, agriculture, bare ground or human-made structures, among others. Land use includes land cover and its social and economic purposes and contexts, for example, different types of agriculture -subsistence vs. commercial or of various tenure regimes-, forestry, types of settlement -urban versus rural-, or tourism (Meyer & Turner, 1994; Turner et al., 1990).