Navigating challenges and opportunities of land degradation and sustainable livelihood development in dryland social–ecological systems: a case study from Mexico

Abstract

Drylands are one of the most diverse yet highly vulnerable social–ecological systems on Earth. Water scarcity has contributed to high levels of heterogeneity, variability and unpredictability, which together have shaped the long coadaptative process of coupling humans and nature. Land degradation and desertification in drylands are some of the largest and most far-reaching global environmental and social change problems, and thus are a daunting challenge for science and society. In this study, we merged the Drylands Development Paradigm, Holling's adaptive cycle metaphor and resilience theory to assess the challenges and opportunities for livelihood development in the Amapola dryland social–ecological system (DSES), a small isolated village in the semi-arid region of Mexico. After 450 years of local social–ecological evolution, external drivers (neoliberal policies, change in land reform legislation) have become the most dominant force in livelihood development, at the cost of loss of natural and cultural capital and an increasingly dysfunctional landscape. Local DSESs have become increasingly coupled to dynamic larger-scale drivers. Hence, cross-scale connectedness feeds back on and transforms local self-sustaining subsistence farming conditions, causing loss of livelihood resilience and diversification in a globally changing world. Effective efforts to combat desertification and improve livelihood security in DSESs need to consider their cyclical rhythms. Hence, we advocate novel dryland stewardship strategies, which foster adaptive capacity, and continuous evaluation and social learning at all levels. Finally, we call for an effective, flexible and viable policy framework that enhances local biotic and cultural diversity of drylands to transform global drylands into a resilient biome in the context of global environmental and social change.

1. Introduction

Drylands form the largest biome complex on Earth, covering about 41 per cent of the terrestrial surface [1,2]. Life there has evolved under the most variable, unpredictable and extreme of all global climates considering the size, frequency and spatio-temporal distribution of precipitation events [3,4], and the high diurnal temperature fluctuations and incoming solar radiation intensities [5]. This large biome has given rise to unique ecosystem types with striking ecological diversity, offering all fundamental ecosystem goods and services for mankind and the many early cultures originating in arid regions [6]. Drylands in Africa are the cradle of humankind. High reciprocal adaptability of drylands and their livelihoods to water scarcity has shaped and strengthened land–human connections for many millennia [7–9]. Hence, drylands carry one of the longest legacies of any social–ecological system (SES) on Earth.

Global drylands lie mostly (72%) in developing countries and are home to over two billion people [10] of whom 90 per cent depend on rural livelihoods [9]. Global drylands are often common-pool resources owned by national, regional or local governments, communal groups or private individuals [11]. Because of their comparatively low productivity, global drylands have collective property rights regimes. This tenure system remains hotly debated [12–14] in the context of resource exploitation and sustainable development [15]. While privatization has been proposed as a viable alternative to collective property, common-pool resources in drylands essentially require local adaptive governance institutions and innovative public policy [16,17]. Over the last four to five decades, drylands have undergone loss of productivity and biodiversity, and increasing drought frequency, food insecurity, poverty, migration and social disintegration. These are key indicators of land degradation [7,9], which may lead to desertification [18–21]. Desertification is an advanced stage of land degradation (the reduction or loss of biological and economic productivity) in arid, semi-arid and dry subhumid areas, resulting from climatic variations and human activities, among others [18] (Article 1 of United Nations Convention to Combat Desertification, UNCCD [21]). Fundamental changes in demographic, economic, technological, climatic, policy, institutional and cultural drivers influence human and biophysical actions (e.g. livestock production, irrigation, deforestation, land-cover reduction), ultimately leading to different syndromes of desertification [22–25]. Long-term mitigation efforts to combat desertification and to rehabilitate or restore degraded or desertified land [21] have rarely been successful. For instance, government help programmes are economy- and/or productivity-driven, not system-based [26].

We are charged with the daunting task of offering timely solutions. We propose three pathways: (i) bring new perspectives and efficient knowledge to strengthen continuous science-policy dialogue; (ii) unfold translational models to frame policy-relevant research, policy development and monitoring and assessment schemes; and (iii) transform desertification into sustainable livelihood solutions. This quest requires an overarching, multi-level approach to broaden and deepen our understanding of SES dynamics. We adopted existing frameworks that were developed to increase our understanding of the unpredictable nature of complex dynamic systems.

In the following sections, we first explore why global drylands, after millennia of existence and adaptation to strong climatic, socio-political and cultural changes, have suddenly become vulnerable to land degradation and, as a global biome, may be reaching the tipping point to desertification (UNCCD [18]). We then take the resilience perspective and examine how the adaptive capacity of dryland social–ecological systems (DSESs) may have changed to recover their structure, function and identity after disturbance events (adaptive renewal cycle [27,28]) and in the light of rapid directional global environmental and social change [29]. With the Drylands Development Paradigm (DDP; for details, see the electronic supplementary material, appendix S1; [19]), we then identify potential underlying causes of dryland degradation with the goal to halt, prevent and mitigate land degradation and to effectively restore affected systems. We then apply the ecosystem services framework [30], and highlight why ecosystem goods and services of drylands are essential for building natural, social and cultural capital as fundamental assets for the development of dryland livelihoods. We took a case study approach that centres on the isolated rural community of Amapola, located in the drylands of Mexico. Its current livelihoods and degrading landscape have coevolved in a greatly fluctuating social–ecological environment over the past 450 years, representing dynamics found at regional and continental scales. We addressed the following questions: (i) What biophysical and socio-economic/cultural/political variables, processes and drivers have shaped the vulnerability and resilience of Amapola and its livelihoods over the past 450 years? (ii) What critical issues and key emergent properties of the Amapola SES need to be considered to determine its adaptive capacity and longevity? (iii) Based on what we have learned, how transferable is this information to other DSESs regionally and globally? We discuss how our approach may serve for the development of new adaptive models required to effectively integrate research, monitoring, assessment, policy development and capacity building [31].

(a). Drylands navigating between vulnerability and resilience

DSESs have undergone innumerable transformations for many millennia without losing their resilience (i.e. the capacity to absorb shocks, readapt and sustain overall structure, function and feedbacks within favourable stability domains) [32–34]. However, over the past half-century, DSESs have become astonishingly vulnerable to climate change [35–38], land degradation [7] and desertification [7,39]. Some DSESs have lost their resilience and crossed critical thresholds beyond the possibility of returning to favourable states [10,25]. This has put dryland systems on the global map as food, water, health and environmental insecurity hot spots [40–42].

Increasing dryland vulnerability to biophysical change (i.e. susceptibility to harm [43,44]) is attributable to the following four main drivers. (i) Recent anthropogenic climate changes are beyond the natural rate and range of the broad inherent variability of dryland systems [45]. (ii) Humans have been eager to eradicate high variability and unpredictability, two of drylands' unique inherent characteristics and sources of resilience, in order to increase the overall low dryland productivity [9], to manage for more continuous water supply [46] and to increase human well-being. (iii) Globalization and the expansion of neoliberal politics, particularly in developing countries of Latin America [47], have significantly changed environmental and production policies, and thus the biophysical properties and dynamics of the world's drylands. (iv) Furthermore, trade liberalization and global market-driven economies [47] have called for regularizations in land rights and agrarian reforms [48], and for replacements of local governance structures and knowledge systems with market-based mechanisms [9,47]. Resulting losses of land productivity and biodiversity as well as increasing vulnerability to poverty, migration and social disintegration have undermined the natural, social and human capital of dryland rural livelihoods. Loss of these assets and of adaptability to change eventually leads to desertification [33].

Current efforts to tackle and mitigate desertification are mainly globally coordinated by the UNCCD. However, effective bottom-up governance and management approaches are needed to achieve sustainable livelihood development, which builds on the natural and social capital as the essential pillar of dryland life-support systems [17,33]. Hence, knowledge of the dynamics of natural and social capital constituents is needed to understand their adaptive capacities and vulnerabilities to change.

(b). Complex system characteristics of drylands

Drylands are characterized by nonlinear dynamics with inherently high levels of uncertainty and unpredictability, high responsiveness to cross-scale interactions, resilience, self-organization and emergent properties. This behaviour classifies DSESs as complex [26,49–52] with a range of emerging functions [53]. Complex systems by definition never reach a long-term stable equilibrium state, but go through Holling's adaptive renewal cycles [27,54] after disturbance events. For instance, for a system to move from state S₂ to S₁ (figure 1a), it will first go through an adaptive cycle (see the electronic supplementary material, appendix S2 and figure S2.1). It is important to acknowledge that system breakdown does not imply irreversible destruction but rather renewal for continuous development [55]. (For more details, see the electronic supplementary material, appendix S2.)

Figure 1.

Coupled Dryland Social-Ecological Systems (DSESs). (a) Resilience of DSESs. The social (left) and biophysical (right) subsystems are coupled through a mutual basin of attraction, which represents the social–ecological landscape of opportunities and constraints as a function of the social, cultural, physical, financial and natural capital. In response to unpredictable disturbance events between time t₁ → t_n, the DSES has gone through three adaptive cycles of change and reorganized in three different stable states (S₁, S₂ and S₃) without having changed its fundamental structure, function and feedback mechanisms and without having crossed a biophysical (TB) or socio-economic (TS) threshold. (b) Loss of resilience in response to past disturbance events (triggered by external or internal drivers) and as a consequence of decline in socio-economic and natural capital (slow variables); system states S₂ and S₃ in regime X are better adapted to the changing social–ecological conditions than S₁. However, S₂ and S₃ are increasingly vulnerable to disturbance, and the system approaches and finally reaches the critical TB_X/Y (e.g. shrub encroachment, a slow variable, could be inhibiting grass recovery), which separates regime X from Y. In this case, the DSES can temporarily recover in regime X, but continuing drought (slow variable) and a lack of rapid recovery of the financial and social capital (slow variable; shown by the difference in height of vertical dotted lines at t₁ and t_n) ultimately lead to massive death of livestock; by t_n, the DSES finally crosses TS_Z/X and enters a new regime Z (regime shift from X → Z). In regime Z, the DSES either continues as a highly degraded system, or it is intentionally transformed by human intervention to generate new livelihood options (e.g. diversify by introducing rain-fed agriculture; ecotourism).

After a collapse, the system enters an unspecified state and releases all accumulated resources, energy and/or information (see the electronic supplementary material, figure S2.1: omega phase = release phase) before it readjusts. The process of reorganization (see the electronic supplementary material, figure S2.1: alpha phase = renewal phase) is determined by the system's memory (i.e. its historic path dependence), overall system condition and new social and/or ecological opportunities. Once reorganized, the system rapidly exploits all available resources (see the electronic supplementary material, figure S2.1: r-phase = growth phase) and finally transits to the slow accumulation and storage of certain resources and information (see the electronic supplementary material, figure S2.1: K-phase = conservation phase) [27]. The longer a system remains in the conservation phase, the more internally connected and responsive it becomes to small external shocks bringing the K-phase to an end; a new adaptive cycle will initiate with the release phase.

The higher the diversity of options to respond to, the higher a system's capacity and potential to reorganize under new social–ecological conditions (i.e. the higher a system's resilience) [56]. In DSESs, the diversity constituting the natural, social and cultural capital (figure 2) together enhances the system's capacity to reorganize in the same state (figure 1a). For instance, in figure 1a, the system reorganizes in state S₂, though under changing environmental and socio-economic conditions (figure 1a: moving along the time axis, t₁ − t_n). Alternatively, the system may shift to a new state (figure 1a: from S₂ to S₁) in the same regime (figure 1: system states S₁ and S₂ within regime X). In case of adaptive capacity loss, the system is pushed towards unstable conditions (figure 1b: the system in state S₁ is pushed towards a biophysical threshold, TB_X/Y) before it crosses a threshold (figure 1b: socio-economic threshold, TS_Z/X) and shifts into a new regime (figure 1b: after crossing TS_Z/W, the system enters regime Z). The capacity of a complex system to navigate between a number of alternate stable states within the same stability domain/regime determines the resilience of a DSES (figure 1a: S₁ and S₂ from time t₀ to t₁). Regime shifts may occur because interactions of variables across scales within a single domain (biophysical, social or economic) lead to drastic changes in the properties and thus the responsiveness of a system [57]. Alternatively, such shifts may be induced by the interaction of social, ecological and economic drivers that may lead to the crossing of a series of thresholds (cascading thresholds), and thus cause a series of regime shifts [58]. Humans often interfere in the adaptive cycle by adopting short-sighted management practices disregarding natural system dynamics (for more details, see the electronic supplementary material, appendix S2).

Figure 2.

Dryland social (S)–ecological (E) systems (DSES) are dynamically coupled through the quantity and quality of assets (productive base) of social, cultural and natural capital, which consist of or are indirectly contributed by the supporting, regulating and cultural/social services. Together and in response to internal and external socio-economic and biophysical drivers, these services determine the dynamics in the quantity and quality of the provisioning services. Hence, slow variables (labelled V_S) determine the variability in quantity and quality of natural resources (fast variables, labelled V_F), which deliver the provisioning ecosystem services to sustain livelihoods. The strength of coupling between social and ecological provisioning services (the length of the white horizontal arrow) contributes to DSES resilience for the set of desirable stable states within a certain regime (see figure 1).

(c). The Drylands Development Paradigm

The DDP [9,19,25,59] is a conceptual framework based on five principles (for more details, see the electronic supplementary material, appendix S1) that best responds to the research, monitoring and assessment needs explained above. It allows the simultaneous analysis of the biophysical (ecological and climatic aspects, natural capital), socio-economic/socio-cultural (social, human, financial, infrastructure) and cross-cutting (ecosystem goods and services) domains (figure 2) of DSESs vulnerable to land degradation, drought and desertification [18,42]. In particular, and unlike other frameworks, the DDP permits an integrated analysis of complex adaptive SESs considering past, present and future policy, governance and management. Since the DDP builds on complex systems and resilience theory, this framework guides the search for key interrelated ecological and socio-economic variables and processes that characterize system structure and function. Unlike other frameworks and paradigms developed to understand dryland degradation and desertification, the DDP strongly emphasizes key biophysical and socio-economic sources and the roles of change, cross-scale interactions and overall system dynamics. Also, it stresses the importance of system memory (e.g. in the sense of traditional local knowledge and social learning) and legacy (e.g. historic development of land-use change; path dependence) and their implications for SES stewardship [9,19].

(d). Dryland ecosystem services couple social–ecological systems

Dryland ecosystems are highly diverse, and this diversity contributes to sustain their multi-functionality and the delivery of ecosystem goods and services to over two billion people [7,60]. Long-term multi-functionality increases the resilience and resistance of drylands to climate change and desertification. Most dryland ecosystems' diversity and their goods and services, however, are not protected in reserves or national parks but are directly appropriated by people who are an integral element of dryland landscapes [9] (figure 2). Many external and internal biophysical and socio-economic drivers will put high pressure on the natural capital of DSESs and thereby impair dryland livelihoods (figure 2). Increasing water scarcity associated with climate change and excessive extraction will be one of the greatest threats for the long-term provisioning of ecosystem goods and services in drylands (figure 2; see the electronic supplementary material, appendix S3, for a detailed description of important ecosystem services in drylands).

Redistribution of precipitation water in dryland landscapes is controlled by topography, vegetation cover, vegetation composition, rooting depth, soil depth, soil texture and structure and human interventions. Hence, any alteration of these fundamental dryland ecosystem structures and functions will influence dryland capacity to deliver ecosystem goods and services and to contribute to human well-being [30]. Rural dryland populations depend mainly on livestock and crop production, and occasionally on wild terrestrial foods. Thus, rural livelihoods draw on the many products and assets that rangelands (food, fibre, freshwater, construction material) and rain-fed agriculture (staple foods, cereals, legumes) provide (figure 2). Drylands have high cultural diversity, suggesting that peoples living in different drylands around the world have established strong identity, connections to and knowledge of their lands and resources (figure 2; natural capital is tightly coupled to cultural capital). The maintenance of high biodiversity and the multi-functionality of ecosystems foster cultural identity and are the basis for livelihood diversification.

For rural dryland systems to remain long-term, multi-functional, sustainable life-support systems, their landscapes must continuously and adaptively provide multiple ecosystem goods and services. This permits livelihood adjustment and development compatible with the inherent natural, cultural and social capital (figure 2). Historic landscape transformations and water management affect the water cycle and thereby all fundamental ecosystem services directly by diverting and retaining runoff water in dams for irrigation or extraction, and/or indirectly by changing land-use and cover type. One of the greatest challenges drylands will be facing is to balance water for humans and nature [40,61,62], in order to balance the provisioning and cultural services with the supplying and regulating ecosystem services in the future [30] (for more details, see the electronic supplementary material, appendix S3).

2. Mexican case study of a dryland social–ecological system

Land degradation and desertification are extensive and growing problems in Mexico, the fifth largest and third most populated country in America. Close to 50 per cent of Mexico's land area is characterized by arid and semi-arid climates [63], which host nearly 30 per cent of the national population [64]. Rural livelihoods depend on livestock production and rain-fed agriculture [65]. Land degradation is causing approximately 2250 km² of formerly productive farmland to become abandoned annually, directly impacting two-thirds of the country's poor population [66].

In a recent assessment of soil degradation [64], agriculture (39%) and overgrazing by livestock (39%) were identified as the main drivers of land degradation [67]. Rain-fed agriculture occurs on former semi-arid grasslands. These are low-input traditional farming systems, which include a diversity of domesticated plant species (corn, bean, squash) adapted to highly dynamic local environmental conditions and traditional cropping practices dating back to 5000 BC [68,69]. Episodic droughts in the last century (in decadal cycles, [70]) have greatly impacted livestock production and agriculture, thereby affecting food security of rural livelihoods. In colonial times, climate variability, extreme droughts, agrarian crisis and famine have influenced socio-political events such as the Independence War and the Mexican Revolution [65,71–73]. Mexican society's responses to these crisis events illustrate the resilience and adaptive capacity of SESs in Mexico's history [74]. Since the agrarian land reforms (Articles 27) in 1917, Mexico's drylands have been common-pool resources (ejido) under a collective property rights regime [75]. However, recent changes (1992) in national agrarian policies introduced new land regularizations permitting individual ejido members to sell or rent their land [48].

In the following sections, we present the case study of Amapola, a small, isolated rural community whose social–ecological legacy dates back to pre-colonial hunter-and-gatherer societies. We will roughly reconstruct how changes in land property rights, land-use policies, migration, droughts, urbanization, fluctuations in global markets and government interventions have led to the adjustment, collapse and renewal of the current Amapola DSES.

Amapola is the most remote community within the Sierra San Miguelito, 35 km southwest of San Luis Potosí (latitude 21°56′60″ N; longitude 101°10′0″ W; altitude 2105 (m)a.s.l.), an area of Mexico covered by an extensive, semi-arid pine oak forest and open rangeland [62] (see the electronic supplementary material, figure S4.1). Founded in 1924 [62], Amapola is situated near the hydrological divide of the Sierra San Miguelito. Exposed rocks in the area are volcanic rhyolites; there are numerous lava domes that contain localized tin. Soils are lithosols and xerosols in the hills and valleys, respectively, and are shallow overall. Rainfall averages 480 mm yr^–1, dominated by torrential storms during July and September, which create large volumes of runoff. The Amapola region is characterized by a heterogeneous landscape consisting of rocky hills, alluvial fans and valley floor. (For detailed description and illustration, please see the electronic supplementary material, appendix S4).

(a). Methods

The five principles (P1–P5; see the electronic supplementary material, appendix S1) of the DDP [19] guided our spatio-temporal analysis (DDP-P1; see the electronic supplementary material, appendix S1) of the Amapola DSES in three steps. In step 1, we identified and characterized specific system states (figure 1), for which we estimated the relative size of the natural, social, cultural, human and built capital (figures 2 and 3; DDP-P1, -P2 and -P5; see the electronic supplementary material, appendix S1). In addition, we identified key exogenous and endogenous drivers that shaped DSES states and livelihood development (figures 2 and 3; DDP-P1 and -P4; see the electronic supplementary material, appendix S1). Finally, we explored possible biophysical, socio-political and socio-economic crisis events at different scales that may have caused a change in system state or regime shift (figure 1; DDP-P1–P4; see the electronic supplementary material, appendix S1) considering the past 450 years. Based on historical, political, economic, social and cultural contexts, we determined specific periods corresponding to specific states or regimes of the Amapola DSES (figure 3) with their respective livelihoods considering the natural, social, cultural (figures 2 and 3; DDP-P1,- P2 and -P5; see the electronic supplementary material, appendix S1), human and built capital. In step 2, we zoomed in on the current state of the Amapola DSES. We analysed the role of key exogenous and endogenous drivers and how they interact with land-use/cover change and current states of ecosystem goods and services (natural capital) and livelihoods (social and cultural capital; DDP-P5; see the electronic supplementary material, appendix S1) in the Amapola DSES. Also, we demonstrated how local-scale land-use dynamics relate to regional-level land-use dynamics (DDP-P4; electronic supplementary material, appendix S1) and their consequences on regional hydrological function considering an SES scenario of interacting public policies and climate change. In step 3, we integrated the findings of steps 1 and 2 and describe emergent properties that help explain the socio-ecological functioning and state of resilience of the Amapola DSES.

Figure 3.

Developmental trajectory of the Amapola DSES over the past 450 years considering four periods corresponding to four adaptive cycles [27]. Each period is characterized by a certain combination and size of capital (horizontal bar diagrams), local, regional, national, international drivers (box around text) and livelihoods (circle around text).

For the analysis of key external international, national, regional socio-economic and socio-political and socio-cultural drivers, we consulted archived historical information and databases on indigenous tribes/cultures, land property rights, demography, national, state and municipal institutions and associated government programmes for sustainable rural development corresponding to Central Mexico [64,76–83]. From these sources, we extracted dominant drivers pertinent to the region and certain periods.

To identify historic internal/local drivers explaining changes in land-use and economic practices, we consulted archived material, databases and aerial photos (Registro Agrario Nacional, Instituto Nacional de Estatistica, Geografia y Informatica). To explore current local drivers, we consulted government programmes, agrarian and environmental laws, and conducted semi-structured interviews of all Amapola households (n = 13). We collected information on family structure, population age structure, migration, non-agricultural labour, access to subsidies and government programmes, cropping systems, water-use types, social organization, local environmental knowledge, cultural ties, land-use history, local economy and land-use and management decisions (DDP-P2; see the electronic supplementary material, appendix S1: we evaluated the decision-making process with respect to differential responsiveness to fast and slow variables). For most variables associated with Amapola community characteristics, we examined the frequency distributions among households and community members.

From this information, we extracted three groups of variables: (i) those that form a key set of socio-economical and biophysical drivers at multiple spatio-temporal scales (DDP-P1 and -P4; see the electronic supplementary material, appendix S1); (ii) those that explain slow and fast changes in the structure and functioning of the Amapola DSES (DDP-P2; see the electronic supplementary material, appendix S1); and (iii) those that resulted in livelihood diversification and associated land-use change.

(i). Spatio-temporal transformations of Amapola dryland social–ecological system and the landscape-function scenario

We assessed land-use change dynamics at the Amapola landscape scale and compared those with larger scale land-use change dynamics in the Sierra San Miguelito, San Luis Potosí (see the electronic supplementary material, appendices S4 and S5). With this upscaling exercise, we demonstrate how transformations of rural livelihoods are coupled to land-use change dynamics and how they may feed back on landscape function.

For the Amapola landscape (for additional information, see the electronic supplementary material, appendix S4), we used aerial photographs from 1969 and 2004 (the only aerial photographs available for the area) to explore net legacy effects of external and internal drivers on land-cover change and spatial reconfiguration of land-use types in the Amapola region considering different spatial and temporal scales (DDP-P1 and -P4; see the electronic supplementary material, appendix S1). We digitized georeferenced orthophotos and with visual classification created polygons with reference points (with a Geographical Positioning System) taken in the field for each land-use category: pine–oak forest, grassland/rangeland and cropland and abandoned cropland. This analysis was performed with GIS IDRISI v. 3.2 (Clark Labs, Worcester, USA). For the Sierra San Miguelito region (for additional information, see the electronic supplementary material, appendix S5), we generated thematic maps of the same land-cover types as for Amapola for the years 1979, 1989, 1999 and 2009 using a combination of supervised and non-supervised satellite images (Landsat) with the GRASS-GIS software (Geographic Resources Analysis Support System, v. 6.4.2 RC3). To scale runoff from our experimental plots to the Amapola landscape level, and to account for differential effects of land-use type on hydrological processes, we multiplied the total area (determined with IDRISI; figure 4) with the respective variable (runoff volume, C and N content) for each land-use type and year.

Figure 4.

Land-use/cover types of the Amapola watershed in the Sierra San Miguelito, San Luis Potosí, Mexico, in (a) 1969 and (b) 2004, derived from aerial photography and visually classified polygons.

We simulated how the dominance of livestock-based livelihoods at the scale of Sierra San Miguelito (stimulated by neoliberal agrarian policies and global markets, and prolonged droughts and tree mortality) [84] may cause irreversible regime shifts in DSES function as a consequence of crossing the threshold of critical soil depth (slow variable; figure 2b; DDP-P2 and -P3; see the electronic supplementary material, appendix S1). We parametrized the biogeochemical model BIOME-BGC [85] and incorporated model outputs into thematic maps to determine regional hydrological function (for details, see the electronic supplementary material, appendix S4).

(ii). Current status of key dryland ecosystem services in the Amapola social–ecological system

To analyse how farmers' decision-making, past climate conditions and historical regional socio-environmental transformations (see previous section) have influenced the natural capital of the Amapola SES, we examined key supporting and regulating ecosystem services (figure 2; DDP-P2; see the electronic supplementary material, appendices S1 and S4) of the three land-use types characterizing the Amapola landscape. To examine whether a government agrarian subsidy programme (PROCAMPO; see the electronic supplementary material, appendix S6) influenced the provision of different supporting and regulating ecosystem services, we compared crop management type (traditional—low-impact tillage; technified—high-impact tillage) applied to different crop types (corn, oat, abandoned).

Supporting services

In drylands, we consider soil characteristics to be fundamental supporting services (figure 2). We determined soil fertility, soil organic matter content (SOM, t ha^–1), soil organic carbon (SOC, t ha^–1), soil total nitrogen (t ha^–1) and soil compaction (soil bulk density). We randomly selected five plots per land-use type and collected within each plot a composite sample (n = 5) at depths of 0–15 cm and 15–30 cm. To determine potential nitrogen (N) mineralization rates as an indicator of soil fertility for each land-use and crop management type, we collected three randomly distributed soil samples at 10 cm following the incubation and extraction protocol described by Robertson et al. [86]. To determine soil depth, we measured the ‘A’ horizon in 50 gullies of the watershed and determined average soil depth for each land-use type.

Regulating services

We consider perennial basal vegetation cover (%) in grassland and forest sites fundamental in regulating water redistribution (figure 2). We used five 50 m transects running parallel to the slope [87]. We also determined the redistribution of each rainfall event from August 2006 to July 2007 comparing the three land-use types. Rainfall was collected in five pluviometers spread across land-use types. To determine runoff, we installed five surface runoff plots (1.6×0.5 m×35 cm) connected to 50-litre collector tanks [88] in each land-use type. Runoff was recorded after each rain event during a whole year. Within runoff plots, we determined water infiltration rates during the dry season with a single-ring infiltrometer [76]. To determine the impact of sediment erosion on the loss of SOC and nitrogen, we collected one composite (n = 5) sample (4 cm diameter × 15 cm depth) of the deposited sediment in eight earthen dams of the Amapola watershed in March 2007 and in five samples of water of each dam (total dissolved organic carbon).

Cultural services

Milpa is an ancient indigenous polyculture cropping system which offers a wide variety of annual subsistence crops (corn, beans, squash, corn fungus/huitlacoche, chile) and thus contributes fundamentally to farmers' food security (figure 2). Milpa is a traditional institution that provides important cultural services such as cultural identity, local environment knowledge and ornaments for religious purposes [89] (DDP-P5; see the electronic supplementary material, appendix S1).

(b). Laboratory analyses

Composite soil samples were oven-dried at 70°C during 72 h and sieved (less than 2 mm for all the analyses). SOM was determined with the calcination method (600°C for 2 h; [90]). For soil bulk density (g cm^–3), we determined soil dry weight and core soil volume [91]. Soil and sediment (deposited in dams) organic carbon and total nitrogen were analysed with an elemental analyser (Costech, modelo 1016), and total dissolved carbon and nitrogen of dam water (TOC, TON) were determined using a total organic carbon analyser (TOC-VCSN, SHIMADZU). Potential nitrogen mineralization rates were determined following the laboratory incubation and extraction methods described by Robertson et al. [86].

(c). Statistical analysis

To identify past livelihoods, we discriminated between different groups of local, regional, national and international drivers, and assigned a period-specific descriptor to each livelihood (figure 3). To identify current livelihoods, we matched qualitative and quantitative information of land-use type, management practices, access to government subsidy programmes, household age structure, income sources and level of education for each household to information corresponding to key local, regional, national and international drivers (figure 3). Based on frequency distribution analysis of all categorical variables, we grouped households and farmers, which resulted in two livelihood categories currently coexisting in Amapola.

Soil, vegetation and water variables were tested for normality using Shapiro–Wilk's test. For water variables, such as runoff that included a time component, we applied a repeated-measures analysis of variance with the mixed model (SAS). Land-use type (forest, grassland, cropland) was the fixed effect, while time (56 sampling dates) was the random effect. Soil variables, including soil bulk density, soil total N, SOM, SOC and soil depth, were obtained once and analysed using a nested ANOVA with the factor soil depth (0–15 cm, 15–30 cm) nested within the main factor land-use type (forest, grassland, cropland, abandoned cropland). To examine the effect of PROCAMPO on soil quality (fertility, structure, stability) characteristics, we applied a two-way nested ANOVA with soil depth (0–15 cm, 15–30 cm) nested within cropping type (with subsidy, without subsidy) as the main factor. Analyses were performed with SAS v. 8 (SAS Institute Inc., Cary, NC, USA). We compared the condition of current supporting and regulating ecosystem service variables (see the electronic supplementary material, appendix S3, tables S3.1–S3.3; figure 5) for the three/four land-use types by using the values of the ecosystem service variables for the forest ecosystem as reference values (100%). For databases used in this study, please consult doi:10.5061/dryad.1c2b7.

Figure 5.

(a) Effect of land-use change on key regulating and supporting ecosystem services for forest, grassland and cropland systems. Spider diagram depicting relative changes in condition of key slow variables compared with the reference (100%) forest ecosystem in the Amapola landscape, San Luis Potosí, Mexico. (b) Effect of government subsidies on key regulating and supporting ecosystem services of croplands. Spider diagram depicting relative changes in condition of key slow variables compared with the reference (100%) forest ecosystem in the La Amapola landscape, San Luis Potosí, Mexico.

3. Results

Considering the past 450 years, the developmental trajectory of the Amapola DSES can be divided into pre-Hispanic, colonial, post-revolution and neoliberal periods (figure 3). In each period, a distinct set of dynamic external and internal socio-political, socio-economic, socio-cultural and environmental drivers acted and interacted across different spatial and temporal scales leading to distinct livelihood developments. (For a detailed description of each of the four periods, see the electronic supplementary material, appendix S6.)

Over the past 24 years, the inhabitants of Amapola have been exposed to a wide variety of external drivers, and depending on their adaptive capacity and willingness to change, they have developed one of two livelihood types that currently coexist in the community (figure 3).

Livestock producers base their living on farming and livestock; they represent 47 per cent of the households; they are ejidatarios with land ownership rights (for a more detailed description, see the electronic supplementary material, appendix S6). Herd sizes fluctuate interannually between 40 and 80 animals per household depending on annual precipitation and rangeland productivity. To reduce feed shortages, multi-cropping systems have been converted to corn and oat monocultures, thereby sacrificing their essential subsistence crops. Government subsidies were used to buy livestock, rent tractors for ploughing and occasionally acquire vehicles to transport livestock and crops (feed sale) to local markets. This livelihood depends directly on the fluctuation of local livestock prices and thus global meat markets, while putting food security at risk.

Semi-proletarian farmers represent 53 per cent of the farmers' households; these are young farmers between 30 and 40 years who make their living growing staple foods (rain-fed) and herding livestock (three to 25 goats per household). Compared with livestock producers, this group has little access to government subsidies and cannot buy livestock. This group favours the traditional milpa cropping system mostly for self-consumption. They complement income with non-farming activities outside the community in nearby cities. Temporary migration has turned into a highly attractive alternative, which allows income to be diversified and new consumer-oriented lifestyles to be adopted (as learned from city jobs; figure 3).

Migrants. This is not a livelihood, but considering 66 people migrated to the USA or nearby cities, their influence on Amapola landscape function (natural capital) and social cohesion (social capital) is considerable. With PROCEDE, as ejidatarios they keep their land as heritage, to conserve land, community and family ties and for economic security. Since migrants do not sell or rent their land, these agricultural plots are in all cases abandoned.

Livelihood development is tightly coupled to Amapola landscape (3300 ha) transformation (figure 4). The forested area is the dominant land-cover type (60%), followed by secondary grasslands (30%) and agricultural land (10%) (see the electronic supplementary material, appendix S4 and figure S4.1). Between 1969 and 2004, 4 per cent of the forested area was converted to grassland, and 5 per cent of the croplands was abandoned (figure 4; see the electronic supplementary material, appendix S4 and table S4.1). At the regional scale, land-use change between 1979 and 2009 led to 6 per cent deforestation (see the electronic supplementary material, appendix S5 and figure S5.1).

After 450 years of co-adaptation between land and people, the four land-use types currently shaping the Amapola landscape differ in quantity and quality of the ecosystem goods and services they provide. As in other dryland regions, water scarcity has limited the production of crops, forage, fuel wood and other provisioning services. However, despite these limitations, several rural livelihoods have developed around the fluctuating supply of ecosystem goods and in response to newly emerging socio-environmental conditions, restrictions and opportunities. A detailed description of the four ecosystem service types of the four land-use types is presented in electronic supplementary material, appendix S3.

Exploitative land use for several decades has severely altered the supporting and regulating services of the Amapola landscape. Livestock, drought and overall shallow soils induced loss of vegetation cover and soil erosion. When comparing key supporting and regulating services and using the forest ecosystems as baseline, we found that SOM dropped in grasslands by 26 per cent and 43 per cent, and in cropland by 34 per cent and 37 per cent at 0–15 and 15–30 cm depth, respectively (figure 5a; see the electronic supplementary material, appendix S3 and table S3.1). Grasslands suffered the greatest loss of SOC (74% and 87%, respectively, for the two soil depths), while both croplands and abandoned croplands showed similar declines of around 60 per cent at both soil depths (figure 5a; see the electronic supplementary material, appendix S3 and table S3.1). Similarly, total soil N suffered twice as much loss in grasslands than in both cropland types. The potential nitrogen mineralization rate was four times lower in grasslands and croplands than in pine–oak forests (figure 5a; see the electronic supplementary material, appendix S3 and table S3.1). For more details on supporting and regulating services, see electronic supplementary material, appendix S3.

To evaluate the influence of external drivers on supporting and regulating ecosystem services, we examined the effect of agriculture subsidy programmes (PROCAMPO) on slow variables (figure 2). ‘Non-subsidized’ agriculture plots exhibited 27 per cent, 30 per cent and 33 per cent greater SOM, SOC and total soil N, respectively, than ‘subsidy’ plots (figure 5b; see the electronic supplementary material, appendix S3 and table S3.3). Also, potential N mineralization rates were substantially higher and soil compaction overall lower in ‘non-subsidy’ than in ‘subsidy’ plots, highlighting the enormous benefits of low-impact agricultural practices as well as a relatively rapid change in slow variables (figure 5b; see the electronic supplementary material, appendix S3 and table S3.3). (For additional information, see the electronic supplementary material, appendix S3.)

Some of the changes in supporting and most regulating services have to be seen in a landscape context. Earth dams are sinks for runoff water and laterally transported soil. The recent installation of 16 dams in the Amapola landscape has led to a substantial redistribution of runoff water and thus landscape hydrological function. Hence, SOC and N transported by water erosion from grasslands and forests resulted in relatively high concentrations of total dissolved organic C and N (9.18 and 1.43 mg l⁻¹, respectively) in dam water. In dam sediments, organic C sinks corresponded to 14.7 t ha⁻¹, greater densities than those observed in grasslands. In the case of N, we estimated an average of 1.06 t ha⁻¹ corresponding to N densities found in grassland soils.

At the regional scale of the Sierra San Miguelito, deforestation caused a 6.5 per cent decline in pine–oak forest between 1979 and 2009 (see the electronic supplementary material, appendix S5 and figure S5.1). An increase in annual precipitation by 100 mm between 1979 and 2009 compensated for the negative effects of deforestation on evapotranspiration and runoff (see the electronic supplementary material, appendix S5 and figure S5.2). Considering a decline in soil depth by 10 cm simulating prevalent laminar soil erosion at the regional scale, and a 6.5 per cent land-use change by deforestation, evapotranspiration and thus primary productivity declined significantly and runoff increased (see the electronic supplementary material, appendix S5 and figures S5.2 and S5.3).

4. Discussion

Land degradation and sustainable livelihood development in global drylands are currently the most daunting challenges the global policy and science communities are facing (UNCCD; Millennium Development Goals; United Nations Department of Economic and Social Affairs, Division for Sustainable Development). Global drylands are intrinsically strongly coupled SESs whose complexity in the context of desertification and dryland development is still poorly understood [19]. It has been argued that our scientific understanding of the interactions of global environmental change with the Earth system is much more advanced than that related to complex social/cultural responses to global environmental changes [92]. We have increasing evidence that environmental subsystems have reached or are close to reaching tipping points in the near future [93]. However, we have no comparative assessment of equivalent social tipping points considering environmental, economic or governance stressors. Nevertheless, this knowledge gap is of considerable concern as societal transformations in response to global environmental or social change shift the context of adaptation, resilience and future transformability [94,95]. Hence, suitable frameworks that include long-term integrative perspectives of SES dynamics have recently been advocated to deepen our knowledge of past breakdowns and historical collapses in SESs [96], complex land system dynamics [94] and land function [97] and resilience of past landscapes [98].

We used three complementary frameworks in an empirically grounded, heuristic case study that spanned 450 years. We explored the historical context of the coevolution of the cultural, social and natural system components and livelihood diversification of the Amapola DSES. We identified key slow and fast multi-scalar biophysical, socio-economic, socio-political and socio-cultural drivers that caused individually or interactively sudden or continuous changes in the internal Amapola DSES dynamics. These internal dynamics included coupled transformations of livelihoods, land-use types and landscape function of the Amapola DSES. In the following sections, we will first synthesize key characteristics of Amapola's complex adaptive system applying the three frameworks: (i) the DDP [19] (see the electronic supplementary material, appendix S1); (ii) Holling's adaptive renewal cycle [27] (see the electronic supplementary material, appendix S2 and figure S2.1) coupled to the resilience/vulnerability landscape with multiple basins of attraction and critical transitions between regimes/domains [34] (figure 1; [54,99,100]); and (iii) an integrative ecosystem services model highlighting the strong interdependence between the social, cultural and natural subsystems supporting adaptive livelihood development in dryland regions (figure 2). We will then propose a shift towards novel ‘next generation’ research, monitoring, assessment and policy development to advance our knowledge for effective DSES stewardship [9] in a rapidly changing world, and to respond to a recent call for an effective science-policy dialogue by the Intergovernmental Platform on Biodiversity and Ecosystem Services [101] in the context of interactive global environmental and social changes [29].

(a). Linking metaphors and frameworks towards understanding complex dryland social–ecological systems

We reconstructed the dynamic trajectory of the Amapola DSES for the last 450 years focusing on the coupled dynamics of the biophysical and social/cultural/economic/political dimensions of this system (DDP-P1; see the electronic supplementary material, appendix S1). In particular, we looked at the composition, size and quality of the total capital (DDP-P2 and -P5; see the electronic supplementary material, appendix S1; figure 2) of the local Amapola DSES (figure 3) as an integrated measure of system state including livelihood development (figure 1a: S₁ or S₂ or S₃) and its adaptive capacity to external or internal drivers of change.

The Amapola DSES underwent four adaptive cycles corresponding to the pre-Hispanic, colonial, post-revolution and neoliberal periods. Transitions between periods were either triggered by induced system transformation [95], regime shift [34,95] or by a shift between alternate states between or within periods (figure 6). With the invasion of the Spaniards and imposition of new ideologies (religion, culture, governance), the native Guachichiles lost their cultural identity (DDP-P2 and -P5; see the electronic supplementary material, appendix S1) and strong ties to their territory (figure 2), triggering long-term loss of social resilience and resistance and ultimately the collapse and end of the indigenous culture (Chichimeca War; omega phase). A new social, political, economic and land property rights context shifted the Amapola DSES into a new regime (colonial period; figures 1b and 3). Strong institutional, top-down governance structures maintained the Amapola DSES for 350 years within this regime (figures 3 and 6).

Figure 6.

Hierarchical adaptive cycles (panarchy; see also electronic supplementary material, appendix S2) at different organizational scales ranging from local dryland livelihoods, national Mexican governance/economy, to global economy and Earth system dynamics (y-axis) for the pre-Hispanic, colonial, post-revolution and neoliberal periods (x-axis). External drivers of local livelihood development in the Amapola DSES correspond to system changes at higher organizational (e.g. national or global economy in response to human population growth) and spatial (biome function in response to climate change) scales. System changes at certain scales (y-axis) may be state changes (alternative state; AS) within a regime, or regime shifts (RS; see also figure 1). A change of phase in the adaptive cycle at one scale (e.g. r → K-phase: expansion of New Spain into North America) may trigger a phase change (e.g. K → Ω phase: annihilation of Guachichiles and hunter/gatherer livelihood) and lead to a regime shift (e.g. shift to labourers livelihood) at a lower scale. Alternatively, 350 years of poor labourer and peasant livelihoods (e.g. a long K-phase in colonial times) and 30 additional years of dictatorship and suppression under Porfirio Díaz (1876–1910) stirred up Mexican peasant and worker groups. The Mexican Revolution (1910–1920) pushed the national governance system across a threshold resulting in the Mexican Republic (change at the higher scale is triggered through ‘revolt’ mechanism at the lower scale; see also electronic supplementary material, appendix S2). Since the post-revolution period, local adaptive cycles of the Amapola DSES have become increasingly coupled to the adaptive cycles in the global economy (from capitalism to neoliberalism) triggering regime shifts at the national level (from a welfare state to a neoliberal state). By intensifying cross-scale interactions, local systems have become increasingly coupled to higher hierarchical scales and increasingly de-coupled from local conditions (see thin arrows connecting adaptive cycles across scales). Earth system changes associated with global environmental and social changes following the industrial revolution have caused the crossing of global biophysical thresholds [93], leading to climate change and biodiversity loss. These global changes are directional and rapid [102], and feed back on nonlinear system dynamics at the local scale.

Long-term exploitative mining and wood extraction caused a net loss of the natural capital (DDP-P2; see the electronic supplementary material, appendix S1) of the Amapola DSES. The social capital was reorganized under the governance of Hacienda patrons specializing workers as miners, charcoal producers and wood extractors. The Hacienda SES experienced rapid growth (r-phase) and accumulated wealth and resources thanks to the abundant local human and natural capital (DDP-P2; see the electronic supplementary material, appendix S1). The Hacienda governance system maintained itself for centuries in the conservation phase (K-phase) by maximizing silver, tin and wood extraction (all fast variables) at the cost of trust building, social and transformative learning and allowing bottom-up adaptive governance structures (all slow variables). Cultural capital never recovered. In contrast, the ‘cosmic’ world view of the indigenous people changed. As haciendas introduced cognition, information, technology [103] (these are all slow variables leading to shifts in fundamental system characteristics) and infrastructure (a fast variable that enhances shift), this imposed changes in values (for instance, change in perception of natural resources), attitudes (loss of identity) and decision-making (shift from local to top-down governance; DDP-P2, decline in key slow variables; see the electronic supplementary material, appendix S1). Original local traditional knowledge was enriched with newly acquired knowledge (DDP-P5; see the electronic supplementary material, appendix S1).

Increasing social bottom-up organization among workers, strengthened by frequent droughts and famine together triggered the Mexican Independence (1821) and Mexican Revolution (1917), and ultimately the collapse of the higher hierarchical socio-political Hacienda system (DDP-P2; see the electronic supplementary material, appendix S1; figure 3). With the fall of the Hacienda land tenure system, a critical threshold was crossed (DDP-P3; see the electronic supplementary material, appendix S1; see also see the electronic supplementary material, appendix S2 and figure S2.1—‘revolt’). This had cascading effects [58] in the socio-political arena of Mexico in the post-revolution period (strong connectedness among scales, DDP-P4; electronic supplementary material, appendix S1).

The Amapola DSES shifted state within the same stability domain (figure 1; e.g. shift from S₂ to S₁) and entered a new adaptive cycle at a local scale (DDP-P4, see the electronic supplementary material, appendix S1; figure 6) as it had maintained sufficient social–ecological resilience. The phase of reorganization (alpha) took time. It was controlled by new national land-reform legislation (DDP-P4; electronic supplementary material, appendix S1), which resulted in a new scheme of land property rights (DDP-P2; see the electronic supplementary material, appendix S1). This opened new options for local governance, land use and local decision-making, etc. (build-up of resilience, DDP-P2; see the electronic supplementary material, appendix S1). Change in local social–political governance structures during the post-revolution period triggered livelihood diversification. Abundant pine-oak forests were attractive common-pool resources that triggered rapid in-migration by a mixture of people from different communities and the foundation of the Amapola community (see the electronic supplementary material, appendix S6; figure 3). After several decades of adaptation with slow accumulation of capital (K-phase), a regime shift in the global economic system (figure 6) triggered a regime shift at the Amapola DSES in 1982.

When in the late 1980s international socio-political decision-makers called for a shift in the global economy paradigm from a welfare to a neoliberal regime resulting in globalization, free-trade agreements (NAFTA), and so on, the critical slow-variable global economy crossed a threshold with socio-political, socio-economic and socio-cultural and environmental repercussions at all geographical scales (figure 6). This regime shift was induced at a late stage of the K-phase of the global adaptive cycle of the economy. Rather than waiting for a global economic crisis (as in 1929), the regime shift was induced and the global economy transformed to the neoliberal model, which has proved to be highly vulnerable considering recent years' food (2007) and economic crises (2010).

Globalization, trade liberalization, change in the national land tenure systems, elimination of producer price support of basic crops and the increasing import of staple foods [76,104–107] led to the shift from a welfare to a neoliberal state [108]. Hence, local transformations of the Amapola DSES towards a new regime were induced by new international and national socio-economic policies and supported by national government help and subsidy programmes. Policy shifts from price subsidies to cash transfer programmes (PROCAMPO and OPORTUNIDADES; see the electronic supplementary material, appendix S6) were rural development and poverty alleviation programmes (intended to avoid local collapse; extension of the K-phase). However, by replacing subsistence crops with forage crops, redundancy (and resilience) in the production system was removed, and livestock production fully controlled and maximized. New agrarian policies strengthened livestock producer livelihoods (see the electronic supplementary material, appendix S6), which specialized in a single commodity maintaining the Amapola SES in the K-phase at the cost of its natural capital and thus ecological resilience, further diminishing potential future options (i.e. desirable alternative stable states within the current regime). This livelihood is highly vulnerable to surprise events (animal disease, massive soil loss, change in markets, shifts in economic models, etc.).

Livestock farmers are relatively well organized socially; however this organization is based on fast external drivers (markets, government programmes, etc.). Local environmental knowledge (DDP-P5; see the electronic supplementary material, appendix S1) and socio-cultural ties (slow variable) to the traditional milpa production system were increasingly neglected with severe consequences on long-term food security. Overall, livestock farmer livelihood had negative impacts on key dryland supporting and regulating ecosystem services (figure 2; see the electronic supplementary material, tables S3.1–S3.3; land degradation as an emergent landscape-scale property).

The semi-proletarized livelihood seems more resilient and resistant to environmental or economic surprises, as it diversifies between livestock production, low-input agriculture and wage labour. However, this livelihood is ‘young’ and in the early-to mid-K-phase of the adaptive cycle, as it is still exploring alternatives and opportunities. Overall, the combination of exogenous drivers and local biophysical system conditions led to a general decline in the diversity (decline in resilience) of productive activities, and thus to livelihood diversification.

A thorough analysis of the supporting, regulating (DDP-P2; see the electronic supplementary material, appendix S1) and provisioning (fast variables) ecosystem services of the Amapola landscape suggests the command-and-control approach for livestock production is in the process of moving the Amapola DSES towards a biophysical threshold (figure 1b), as the natural capital is increasingly deteriorating thereby putting the Amapola life-support system at risk (approaching the biophysical threshold feeds back on the socio-economic outcomes; figure 1b). Grasslands in the Amapola landscape are the land-use type producing the most economically valuable provisioning services (forage) at the enormous cost of most of the regulating and supporting services (figure 5; see the electronic supplementary material, appendix S3, tables S3.1–S3.3). If current livelihoods persist in the future, it is expected that the semi-proletarian farmers will eventually convert into a migrant group (nonlinear unpredictable trajectory of system development). This phenomenon has become rather frequent in many rural areas of the world [109] (emergent property at a large spatial scale). This will allow livestock farmers to increase stocking rates and to expand the rangelands into the forest ecosystem with obvious environmental consequences (pushing the system at larger spatial scales into an increasingly vulnerable K-phase) risking desertification [21], and the loss of the life-support system of Amapola. This trend of extending rangeland into forest can be observed in the Sierra San Miguelito, where an increase in shrubland vegetation in the former grassland suggests shrub encroachment and forest deterioration as a consequence of overgrazing (see the electronic supplementary material, appendix S5, figure S5.1).

Many of the slow variables (loss of soil, soil fertility, soil cover) are close to reaching biophysical thresholds (DDP-P3; see the electronic supplementary material, appendix S1; livestock producer livelihood triggers cascading effects on thresholds [58]). However, loss of fundamental supporting and regulating services will trigger land degradation at the larger scale (DDP-P4; see the electronic supplementary material, appendix S1) and loss of the multi-functionality of the landscape (landscape trap, [110]). The high cost associated with a command-and-control approach (focusing only on provisioning services) can be demonstrated for the Amapola landscape (electronic supplementary material, table S3.1, S3.2). Based on observed local annual runoff for forest and grassland sites, we estimated that a 4 per cent conversion of forest to grassland corresponds to 86 300 m³ annual loss of water by surface runoff. Considering soil C and N pools, a 4 per cent deforestation at the landscape level is equivalent to losing 11 855 and 435 t ha⁻¹ of C and N, respectively. This is alarming, as Amapola represents one of a universe of similar communities in Mexico's drylands. When applying the previous exercise to the regional scale of Sierra San Miguelito (1240 km²) using the biogeochemical model BIOME-BGC [85] (see the electronic supplementary material, appendix S5 for a detailed description), we showed that 3 per cent deforestation (electronic supplementary material, figure S5.2) of pine–oak forest between 1989 and 2009 triggered almost one million cubic metres increase in annual runoff (electronic supplementary material, figures S5.2 and S5.3). When simulating soil loss (one of the most critical slow variables in drylands) for 30 years, total evapotranspiration and net primary productivity decreased on average by 5.4 per cent, while runoff increased by 12.5 per cent (electronic supplementary material, figure S5.3). Hence, an increase in the establishment of dams and the massive redistribution of soil and water caused by erosion have fundamental implications for the long-term multi-functionality of landscapes as providers of a diversity of ecosystem goods and services [110].

With Holling's adaptive cycle metaphor and the DDP analytical framework, we were able to identify key processes that explain the resilience and vulnerability of the Amapola SES dynamics. Having gone through four adaptive cycles in its 450 year history, the Amapola DSES has a long and complex socio-cultural and biophysical legacy [94] having crossed socio-cultural, socio-political and socio-economic thresholds [19,94]. The Amapola SES has maintained itself because of the surprisingly high resilience of its biophysical and social systems. However, during its recent history, the Amapola SES has become increasingly coupled to external national (policies), international (policies) and global (climate) drivers, at the cost of its natural, social, human and cultural capital (cascading thresholds, [58]; figures 2 and 6).

(b). Navigating sustainable coupled dryland social–ecological systems for research, monitoring, assessment and policy development—insights from emergent system properties

It has been acknowledged that only a small set of key biophysical and socio-economic variables (slow variables) determines the underlying dynamics, resilience or vulnerability of SESs to disturbance events [26,111]. While it is important to identify, examine, monitor and assess the changing conditions [25,112] of these variables, it is also important to ask which fundamental emergent properties of SESs besides these variables can provide conclusive integrative information and insight into the long-term dynamics and functioning of SESs for a new, integrated science-policy process [31]. With our multi-framework analysis of a complex DSES, we identified the following emergent properties and processes as fundamental to enhancing dryland system resilience, adaptability and transformability [95]. (i) Managing ecosystem service packages instead of single provisioning (livestock production) or supporting (C sequestration) services enhances the multi-functionality of landscapes and thus biophysical resilience and adaptive capacity under future environmental change conditions [7,113]. (ii) Human appropriation of water interferes strongly with natural hydrological fluxes at the cost of ecosystem function. Since water is becoming increasingly scarce, a shift in water resource management from a pure runoff-based to an integrated blue (constituting runoff) / green water (constituting soil moisture, transpiration—sustaining all ecosystem services) flow-based paradigm has been advocated as a resilient approach [40,61] to better balance water between nature and people and to increase food security in the face of global environmental and social change [114]. (iii) Household livelihoods depend on balanced natural, social and human capital (figure 2). Livelihood development at the community and landscape scale needs to be able to diversify in response to cultural, biotic and policy diversity to enhance social resilience and stimulate social innovation [115] coupled to multi-functional and biophysically resilient landscapes [116]. (iv) Food security in drylands is a matter of local adaptive governance strategies [117] considering (i)–(iii) and the fostering of traditional knowledge including cognition of legacy. (v) Dryland dynamics need to be seen as a cyclical rhythm of coupled systems that are hierarchically connected across scales (panarchy). Some of the complex system dynamics result from cross-scale interactions and hierarchical nesting of subsystems both in the biophysical (plot–landscape–watershed) and social (local governance, community organization, land property rights systems, institutional) contexts [118], but also from going through hierarchically linked adaptive cycles (panarchy) operating at different scales in time and space [27]. Drivers are variables (slow or fast) of other (coupled) systems that have undergone state changes or regime shifts themselves, by going through higher or lower level adaptive cycles, and thereby couple system dynamics at all scales. Managing systems (including the development of policies) requires the understanding of slow and fast variables and processes and their inter-connectedness (figure 6 and electronic supplementary material, S2.1). These are system-specific properties of SESs that can be identified when adopting the adaptive cycle metaphor coupled to the basin of attractions (figure 1) and the DDP. These emergent properties are outcomes of this analysis, and when developing management or new policies, these system properties should be targeted. This requires long-term monitoring of interacting variables and coupled processes across scales—an integration of different knowledge systems considering observational, modelling and empirical approaches.

(c). Challenges and opportunities

Complex systems research is needed that considers the dynamic interrelations and feedback that couple the biophysical and social subsystems, key slow variables and processes providing resilience and how they interact with fast variables on which humans most frequently base management and governance decisions. New research is needed on emergent properties that addresses the following questions: Can we identify a priori emergent properties of the DSES and consequently examine, monitor and assess its changing condition as a central objective of complex system analysis (rather than a result of complex system analysis)? Are emergent properties more tangible assets and communication tools to more efficiently inform policy and decision-makers? Can we identify key emergent properties for DSESs and other SESs (coastal areas, forest ecosystem) that are fundamental for novel adaptive governance models, stewardship strategies and/or institutional capacity?

The Amapola case study is widely representative for rural SESs not only in Mexico, but also in Latin America and globally. SESs similar to the Amapola community exist worldwide as they share many characteristics including highly ecologically diverse ecosystems extremely vulnerable to livestock production controlled by global markets, sharing of common-pool resources, extensive land degradation coupled to desertification and drought, poverty and migration. In recent decades, drivers influencing the decision-making of these communities have shifted directionally from local to global by the adoption of neoliberal economies which feed back on fundamental assets of local SESs [116]. This trend is widespread in rural regions of Latin America as they increasingly depend on and respond to global market dynamics, and increasingly ignore the production limitations of local small household farms [118,119]. This is of great concern as rural environments are the long-term basis for the development of farming-based livelihoods [120], which perhaps are the most frequent livelihoods in developing countries. A trend towards proletarianization has been explained by the lack or scarcity of private land [121] and neoliberal policies that focus only on economic growth [122].

We took a heuristic case study approach where we linked an empirical multi-scalar, interdisciplinary study with several frameworks related to resilience, complex systems and sustainable development theory. The DDP functioned as an umbrella framework that defined the social–environmental scope and spatio-temporal dimensions of this study. While the DDP is highly flexible and adaptable to a wide variety of SESs, it also provides clear-cut guidelines for the analysis of SESs targeted towards effective pathways to transform desertification into sustainable development [19]. To our knowledge, this is one of the first studies to explore path-dependent relationships of a local coupled DSES. We focused our analysis on the behaviour and responsiveness of livelihoods as key indicators of the vulnerability and resilience of DSESs in the context of cross-scale demographic, socio-cultural, socio-political, socio-economic and environmental change. From path-dependent analysis, we learned livelihoods need to maintain local options for diversification. Local environmental knowledge, cultural heritage and high biotic and cultural diversity are fundamental elements to increase livelihood security and diversification. Only by considering the cyclical rhythm of past and future SES trajectories, can we learn about the adaptability, transformability and resilience of DSESs [95]. We will have to focus on these emergent phenomena of complex systems when developing policies, management, governance or stewardship strategies for SESs [33]. Fostering resilience thinking [95] increases capacity for innovation, creativity and cross-lateral multi-stakeholder learning and opens new windows of opportunity in times of crisis and uncertainty. In the new area of transition towards a resilient planet [123], rather than operating within the safety margins of control and aiming at maximizing productivity of market-driven commodities, we have to transform our mindsets and start learning to induce shift and transformation and thereby increase options and longevity of sustainable livelihoods in times of foreseeable crises and breakdown.

Global environmental and social changes have made dryland livelihoods highly vulnerable. Climate change, land-use change, loss of biodiversity and degradation of ecosystem services as well as globalization, neoliberal politics, monetary crisis, global markets, industrialization and urbanization are all global change drivers directly or indirectly affecting DSESs. It is the biophysical or social subsystem that is responding to external drivers of change; however, since SESs are coupled, the response to multiple drivers may be amplified [94]. Thus, it is important to understand the nature of SES responses and feedback to multi-causal continuous or abrupt changes in system dynamics that may lead systems to local thresholds or global tipping points. Global change drivers are systemic, in that they affect the functioning of the whole Earth system (global warming), or accumulative, in that they occur first at local or regional scales (land-use change), yet when increasing in impact potentially affect the whole Earth system [29]. In recent decades, local and regional land degradation in drylands has emerged as a global phenomenon, strongly interacting with global climate change, globalization, neoliberal policies and global markets, together exerting a continuous impediment for dryland development. Desertification is the most advanced stage of land degradation, when SESs have lost the elements and interconnections to sustain an adaptive life-support system for livelihood development in a world increasingly affected by global environmental and social change. Actively inducing the transformation of a degraded or desertified system in a new regime could offer windows of opportunities for system development. However, this intervention requires sound scientific understanding, great skill in leadership, novel governance structures and continuous dialogue between scientists and policy-maker representatives of the two billion people living in the largest complex SES on Earth. Future research and policy development need to catalyse new and continuous dialogues and effectively engage local to global stakeholders in a new science-policy process to jointly tackle the daunting challenge to transform the process of desertification into sustainable development [31].