Rewilding-inspired transhumance for the restoration of semiarid silvopastoral systems in Chile
Nomadic pastoralism and transhumance are ancient human adaptations to the movements of large herbivores, which themselves migrate to follow favorable environmental conditions. Free-ranging livestock production has been criticized as less water efficient than factory farming and crop production. This fails to take into account both the additional ecosystem services made possible by rainfall over rangelands, and the ability of free-ranging animals to track water availability across environmental gradients. By analogy to transhumance, we propose a model of “transhumant rewilding,” or species reintroduction with managed herding of wild ungulates for the ecological restoration and sustainability of food production in (silvo)pastoral systems. We consider preliminary evidence for the feasibility of this model with a case study from central Chile in which guanacos (Lama guanicoe) could be used to help restore a silvopastoral savanna (“espinal”) via browsing and endozoochory. First, we present preliminary data on guanaco foraging in espinal. Second, we use a GIS analysis to identify least-cost paths between areas of high and low espinal condition in central Chile and assess the feasibility of using them as migratory pathways. Finally, we consider the relative ecosystem service advantages and costs of the transhumant rewilding scenario compared to other restoration and agricultural development scenarios for central Chile. We conclude that transhumant rewilding has the potential to be a useful model for rewilding-inspired land management in cultural landscapes and can contribute to food security and sustainable agricultural production.
KeywordsRewilding Transhumance Lama guanicoe Ecosystem services Food security Silvopastoral
Traditional pastoralist systems, especially in drylands, are adapted to geographic and temporal variation in rainfall, and thus availability of resources. Nomadism and seasonal transhumance allow livestock to track environmental variability (FAO 2001; Kratli et al. 2013). Traditional pastoralist systems thus represent a solution to sustainable, extensive, biodiversity-friendly food production under variable, especially arid and drought-risk, conditions (FAO 2001). Pastoralism and various forms of extensive rangeland livestock production are gaining traction in some areas, e.g., novel silvopastoral systems funded by payment for ecosystem service (PES) schemes (Giraldo et al. 2011; Montagnini and Finney 2011), while losing ground elsewhere due to criticisms of inefficiency and poor management, particularly in drylands and rangelands (FAO 2001; Mekonnen and Hoekstra 2012). Here, we examine pastoral systems as a potential component in sustainable and biodiversity-friendly food production systems of the future (Oteros-Rozas et al. 2013).
Pastoralism encompasses a range of strategies, both over historical time and over space (FAO, 2001). Pastoralism broadly evolved from hunting large herbivore species as they track the environment on their own, to managing the take of these species, to domesticating them into a nomadic or transhumance system (Alvard and Kuznar 2001; Manzano Baena and Casas 2010; Bar-Oz et al. 2011; Niven et al. 2012), but arguably the greatest difference in the human–herbivore–environment relationship emerges with sedentarization of pastoralists. Sedentarization is often related to attempts to govern human populations rather than environmental imperatives, and often has negative effects on the socioecological system (Sayre et al. 2013). Transhumant and nomadic pastoralists often play an important role in the exchange of edible species, goods, and social relations within large and environmentally challenging regions (Scheele 2010; Manzano Baena and Casas 2010). If livestock cannot move significant distances, their raising becomes decoupled from environmental variability (not only climatic, but also related to any stochastic ecological processes). Livestock are thus both more exposed to local resource limitations and more coupled to external production systems that must be accessed to buffer local climatic variability (FAO 2001; Pedersen and Benjaminsen 2008; Kratli et al. 2013). Consequently, we must consider the design of an effective (silvo)pastoral system in the context of the challenges imposed by sedentary production, which decouples management from the local environment (Kratli et al. 2013). One kind of solution has been to disconnect livestock raising inputs from the local environment through intensive factory farming. Reacting against factory farming, traditional extensive pastoralism is now trying to make a case for its sustainability in terms of ecosystem service values (Hoffman and Boerma 2014; FAO 2001). Further along the continuum of human–large herbivore relations, is there another kind of solution in which pastoralism can move even more toward “wild” systems?
Rewilding is an emerging trend in conservation, primarily in developed countries (Svenning et al. 2016; Jepson 2016). Rewilding has been described, inter alia, as the reintroduction of one or more usually large animal species acting as a keystone species, ecosystem engineer or top–down trophic influence, to restore natural ecosystem functioning, resulting in ecological restoration usually via subsequent passive management, and potentially changing human relationships to the landscape (Sandom et al. 2012; Lorimer and Driessen 2014; Lindon and Root-Bernstein 2015; Svenning et al. 2016). One strand of rewilding focuses on the use of large herbivores to transform habitats through herbivory (Olff et al. 1999; Vera 2000). This approach has roots in managed grazing to maintain certain types of grasslands and cultural landscapes (Gordon et al. 2004; Rook et al. 2004; Papanastasis 2009). Another important function is seed dispersal (Poschlod and Bonn 1998). Although the ecological principles are one and the same (Asner et al. 2004), rewilding is considered to be radically different from managed grazing due to its focus on the spontaneous activities of wild animals and, linked to this, by being process-oriented rather than end-state oriented (Hughes et al. 2011, 2012).
Proponents of rewilding often explicitly wish to create or preserve areas of “wilderness.” Wilderness is a subjective cultural category (Cronon 1996) and herein lies the effect on human relationships to the environment. Rewilding as currently practiced takes advantage of the poor economy and rural land abandonment to create something that, in EU policy terms at least, is new (Navarro and Pereira 2012; Lorimer and Driessen 2014; Jepson 2016). There are, however, other possible models of rewilding that draw much less on classical European and North American visions of wilderness while addressing different challenges and policy gaps. Here, we focus on rewilding-inspired approaches to conservation and management of dryland silvopastoral systems.
Rewilding, (silvo)pastoralism, and food security
Food security refers to access to sufficient, safe, and nutritious food meeting dietary needs and food preferences (Declaration of the World Summit on Food Security, FAO 1996). A number of factors can threaten food security throughout the food chain, from crop failure, through to post-retailing waste (Eriksen et al. 2009; Parfitt et al. 2010; Ingram 2011). One aspect of maintaining food security involves diversifying food production and buffering it against environmental variability (Vermeulen et al. 2012). This can be particularly important in drylands with a history of soil degradation and variable rainfall (Budds 2004; Solh and van Ginkel 2014).
The production of animal products is less water and energy efficient than the production of nutritionally equivalent plant products, due to the inefficient conversion of plant material fed to animals (Pimentel et al. 1997; Capone et al. 2013; Mekonnen and Hoekstra 2012). Grazing systems are also many times less efficient in converting water to food than are industrial livestock production systems (Mekonnen and Hoekstra 2012; Gerbens-Leenes et al. 2013). A larger amount of rainwater goes into producing less-digestible forage and non-forage plants over wide areas in pastoral habitats, compared to the relatively small amount of rainwater used to produce nutritious feed crops fed to fast-growing livestock in industrial systems (Mekonnen and Hoekstra 2012). According to these analyses, if the water footprint were the only variable in consideration, and assuming that food distribution problems were resolved, then global food production should shift away from animal production and toward industrially farmed plant products including vegetables and cereals in suitable environments (Vanham et al. 2013; Smith 2013; Stehfest et al. 2009), or at least switch to more efficient fodder crops (Bosire et al. 2015). These analyses are not without critique. Other approaches have claimed that low-input pastured cattle in non-irrigated systems have a water footprint similar to cereal crops (Ridoutt et al. 2012). Scholz et al. (2013) argue convincingly that water that soaks into the soil from rainfall in drylands has no other possible use than extensive grazing; the assumption underlying the analyses cited above, that it could be diverted to growing crops, is incorrect. Scholz et al. (2013) further argue that achieving increases in efficiency in extensive dryland pasture systems is feasible and makes this an attractive option for increasing food security in drylands.
An additional critique comes from a multifunctional perspective on extensive dryland livestock pasturing. Food security also considers access to traditional diets (including meat), provision of ecosystem services, and sustainability of the agricultural socioecological system (Vermeulen et al. 2012; Eriksen et al. 2009). Livestock production in semiarid rangelands can be compatible with wildlife conservation and associated ecosystem services (Tilman et al. 2002; Tscharntke et al. 2012; Bosire et al. 2015). If we take into account the contribution of rainwater over extensive (silvo)pastoral habitats to the production of provisioning, regulating, and cultural services, then extensive grazing systems have a much higher and more diverse “yield” than intensive farming (Hoffman and Boerma 2014).
Elements of silvopastoralism that could be rewilded
Potential rewilding example
Movement of large herbivores over the landscape
Keep some or all animals in a semi-wild state and allow to track environmental variability across large landscapes
Grazing and browsing with little supplementation from agriculture. Subject to natural rainfall and growing season patterns. Can include unpalatable as well as palatable plants, which can improve digestion and weight gain
Species identity and function
Non-domestic, semi-domestic or “dedomesticated” species and breeds that have needed ecological restoration functions as well as marketable products, e.g., camelids (wool), reindeer (fur). Mixed herds of compatible grazers and browsers, including guard animals such as llamas and donkeys
Tree and shrub cover mosaics
Allow to evolve through herbivory and disturbance. Overstocking should be regulated to prevent degradation where this is a potential outcome, but below the degradation threshold, temporal variation in cover should be expected
Habitat for biodiversity
Allow to evolve through herbivory and disturbance. Ground-nesting birds that are sensitive to trampling may decrease, as should rodents, but other species will move into newly created habitats
Human cultural adaptations to variability
Shepherding, transhumance. Foraging of, e.g., mushrooms, herbs, fruits, honey, game, reeds. Artesanal/“paleotechnics” production from wild, semi-wild, and domestic herbivores, such as wool, fur, skins, milk, cheese, meat, horn, bone. Management that is responsive to (socio)environmental change via LEK and social learning, e.g., taking advantage of modern participative and adaptive management approaches and technologies
Case study: restoration of the espinal with guanacos in central Chile
Central Chile is a semiarid region with a Mediterranean climate and high rainfall variability due to ENSO (van Leeuwen et al. 2013). Little is known about potential pastoralist practices of the now extinct indigenous people of central Chile, the Picunche. They had a domestic camelid called the chilihueque, which went extinct around the 1600s (Miller 1980). Like nomadic indigenous groups across the Andes, the Picunche might also have hunted or herded territorial or migratory guanacos (Lama guanicoe) (Medina and Rivero 2007). Guanacos were originally native to a large part of the southern cone of South America (Baldi et al. 2008), including central Chile, until about 500 years ago when they were extirpated from this region by Spanish colonists (Miller 1980). Guanacos are generalists that can live in grasslands, savannas, woodlands, and forest (González et al. 2006). Guanacos are hypothesized to have brought Acacia caven, the dominant tree in modern central Chilean silvopastoral savannas (“espinal”), from the Argentinian chaco via endozoochory, between 10,000 and 2000 years ago (Ovalle et al. 1990). Despite this potentially late arrival date, this does not mean that savanna must also be relatively novel, as A. caven is proposed, on inductive grounds, to have replaced Prosopis chilensis, another Fabaceae, relatively recently in a savanna or open woodland habitat (Fuentes et al. 1989).
Historical trends in espinal conditions are largely unknown, but it appears that degradation pressures such as harvesting A. caven for firewood and charcoal production may have diminished somewhat since the 1990s (M. Root-Bernstein, pers. obs.). Despite estimates that A. caven espinal covers up to 2 million ha (Peri et al. 2016), espinal, as a silvopastoral system, or as an Acacia-dominated savanna, is not an official land cover type considered in maps created by the Chilean Forestry Service (in charge of conservation), and consequently data on its dynamics are incomplete. A recent pair of book chapters giving an overview of both traditional and novel silvopastoral systems in Chile spend exactly three sentences on espinal silvopastoralism, with more attention given to systems based on non-native plantation trees (Dube et al. 2016; Rojas et al. 2016). We believe that the espinal should be conserved because there is no evidence that it is an “unnatural” degradation (Root-Bernstein and Jaksic 2013). Rather, it appears to be part of the successional network linking matorral (scrub) habitat and sclerophyllous forest, and thus should be protected along with these other habitat types to ensure area for future forest regeneration and successional dynamics (Root-Bernstein and Jaksic 2015). All of these habitats, including espinal, are home to many central Chilean endemic plants and other species, and together they represent the only Mediterranean-climate habitats of South America (Simonetti 1999; Myers et al. 2000). We also believe (and examine, in this paper) that it is broadly preferable to the other land cover types into which it is most likely to be converted without conservation measures.
As in other drylands, water security is an important issue in central Chile (Budds 2004). As in other silvopastoral systems, a key factor in reducing water use and improving meat yield in espinal savanna is provision of shade (Olivares 2006; Ovalle et al. 2006; Campos Paciullo et al. 2011). Shade cover in espinals increases as A. caven canopies increase with tree age, and growth can be stimulated via pruning as a management intervention (Vita et al. 1995; Navarro Gutiérrez 1995). Specifically, like other acacias, A. caven shows compensatory growth, whereby pruning, simulating browsing, causes increased regrowth (Gadd et al. 2001; Fornara and du Toit 2007; Dangerfield and Modukanele 1996; Gowda 1997). Compensatory growth after pruning suggests that A. caven is adapted to browsing by missing herbivores (compare Bond et al. 2004; Doughty et al. 2016).
We hypothesize the guanaco (L. guanicoe) is likely to be one of the missing browsing herbivores to which A. caven is adapted, and the only one that is not extinct. Domestic species common in espinal (cattle, horses) are grazers, not browsers. After the Early Holocene megafaunal extinctions, guanacos spread and increased in abundance throughout South America (González et al. 2006). In addition to showing adaptations to browsing, A. caven seeds germinate at a higher rate after large herbivore endozoochory (Gutiérrez and Armesto 1981). Although there are reports of guanacos browsing on trees or shrubs when available during winter (Puig et al. 1996, 1997; Cavieres and Fajardo 2005; González et al. 2006), we have encountered uncertainty about guanacos as potential A. caven browsers under non-starvation conditions. One of our goals is thus to test whether guanacos voluntarily browse these spiny trees. We predict that guanacos can stimulate growth and help seed germination of A. caven in degraded espinals. We are currently running a multiyear experiment, called Proyecto REGenera (Restoration of Espinal with Guanacos) at the private nature reserve Altos de Cantillana in central Chile to test these predictions about browsing, growth, and germination. Proyecto REGenera, which uses penned guanacos, is a first approach to assess the feasibility of transhumance between espinals. Preliminary results are encouraging and allow us to address the challenges of extrapolating from the penning study to transhumance, including issues such as spatial scaling, temporal periodicity of browsing, and in a final step, social feasibility.
Here, we test the feasibility of a model for restoration in which shepherds herd guanacos across the landscape, along a network of suitable movement corridors in and out of fenced espinals, in a seasonal and cyclical manner in order to target areas that can be restored by guanaco browsing. To substantiate the feasibility of this model inspired by both transhumance and rewilding, we first present preliminary data from Proyecto REGenera, focusing on guanaco foraging behavior. Secondly, we use GIS analysis to ask whether guanacos could be used to deliver pulses of restoration browsing in the most degraded espinals (probably in winter), with shepherded movement along suitable corridors to more-productive espinals during the rest of the year (probably summer), following the logic of either the seasonal movements of the wild guanacos, or the transhumance potentially practiced by the Picunche people with herded guanacos (hereafter “transhumant rewilding”). We see such a system as more feasible than re-establishing fully wild, freely migrating guanacos in central Chile due to the high density of fences, roads, and agricultural land, and feral dog packs. The likelihood of human–wildlife conflict and high guanaco mortality makes this scenario likely suboptimal, and we do not consider it further here. Guanacos will have to be owned (providing a monetary and regulatory stake in their welfare) and shepherded by shepherds for protection, along suitable corridors linking enclosed espinal habitats across the landscape. We assume that restoration of espinal by rewilding is a multidecade process and that transhumant rewilding will be cyclical. To address the feasibility of transhumant rewilding, we ask whether high- and low-quality espinals are near each other, compatible with a set of small-scale community-based networks of corridors linking high- and low-quality sites, or whether high- and low-quality espinals are dispersed at a regional scale, more compatible with a regionally organized, long-distance transhumant rewilding network. The least-cost paths forming the links in the networks we identify with this GIS analysis correspond to suitable movement corridors along which guanacos can be herded by shepherds between espinals. We also examine whether the networks show variation in length, connectivity, and position under fluctuations in climate associated with ENSO. Finally, we consider the potential ecosystem service benefits from guanaco restoration of the espinal silvopastoral system. We ask what the key benefits and costs are under four different scenarios of land-use change. Under the status quo, espinals are not restored and economic incentives continue to favor gradual conversion of espinal and other habitats to high-investment, high-yield fruit and wine production for export (Armesto et al. 2009; Schulz et al. 2010). We compare this to a scenario in which all espinal has already been rapidly converted either to factory farms or fruit and wine production, and a third scenario in which all espinal has been converted either to factory farms or protected sclerophyllous forest. Our fourth scenario is restoration of espinal with guanacos. Although we lack sufficient data to monetize or otherwise value these services and costs, we highlight key services and costs that distinguish between the cases.
Site and animals
Experimental conditions and procedure
The site is divided into four sections of 0.125 ha each. In winter and spring (the wet season), the guanacos spend between 4 and 2 weeks in each section (the shorter time interval prevents over-grazing in late spring as aridity increases) and are rotated from one section to the next clockwise (the “pulse” treatment). In summer and fall (December–May), they are allowed to move freely throughout the enclosure, in order to reduce the intensity of grazing on the herbaceous understory. Although Oba et al. (2000) report that Indigofera spinosa shows the largest compensation effect when pruned in the dry season, we decided to implement the browsing pulse in the wet season due to the reverse phenology of A. caven compared to other deciduous plants in Chile (it has leaves during the dry season), which may also reverse its growth response relative to the season. In other words, we follow Oba et al. (2000) in implementing more-intense browsing when the trees have no leaves, in the expectation that this will stimulate increased growth during their next growth season. Cromsigt and Kuijper (2011) also report that early-growth-season pulses of herbivory are associated with trees adapted to herbivory. Whether or not the rotational plan replicates the natural migratory patterns of guanaco in central Chile is unknown as we are not aware of any historical data on their migration in that area. Guanacos may have avoided colder temperatures at higher altitude or further south by moving into the central valley of central Chile in winter (the rainy season), where they could have eaten the fresh herbaceous substrate and leafless A. caven. However, they may also have eaten A. caven leaves and shoots in early summer, before moving south or to higher elevations. Our rotational plan is based on the expectation that winter herbivory will give the largest compensatory growth response. The rotational phase of the experiment also allows us to track the temporal lag of growth following known periods of herbivory. These data will help us to assess how long guanacos should remain on a site to induce growth but not damage the trees.
The rotational plan is not transhumance. However, it is intended to provide basic data on the guanaco–A. caven interaction, including information on the conditions under which guanaco browse A. caven, which parts of the tree are browsed, the size of the growth response of A. caven to guanaco browsing, and the effects of season and browsing intensity on that response, that could be applied to designing a transhumant rewilding system.
Characteristics of each experimental sector and the comparison site are shown in the Supplementary Data.
The guanacos were introduced to Section A on June 18, 2014. During the first month, we monitored their behavior in order to determine an appropriate supplementary feeding amount. We currently lack the necessary data to calculate the appropriate stocking rate for guanacos in espinal; these data will be obtained over the course of the experiment. The time required for the espinal to recover from intensive herbivory is also unknown, so, given the restricted space available for the experiment, attenuating the herbivory pressure with supplementary alfalfa is necessary. The amount of supplementary alfalfa is adjusted according to monitoring of the guanacos’ body conditions.
To determine how guanacos forage and behave in A. caven savanna habitat, we record behaviors using a focal animal method. Each animal was observed by a single individual for four sessions of 10 min at a time, in a random order, over a period of 4 h during the day, 3 days per week. Here, we report on observations in September, October, and November of 2014. Behaviors were recorded in JWatcher.
In addition, we downloaded photographs from two camera traps that were originally set up to monitor the site rather than guanaco behavior per se, but which captured many images of the guanacos. These were in sections A (October 2014, January, and April 2015), and B (November 2014) of the enclosure.
There is no regional map of espinal land cover, so we obtained a map of A. caven distribution derived from official land cover maps produced for Chile by CONAF, the Chilean Forestry Service (M. Bennett, unpublished data). The vast majority of the obtained areas with A. caven were classed as matorral which is defined as having <10 % tree cover and 10–75 % shrub cover, or as “matorral arborescente” or “tree matorral” which is defined as having between 10 and 25 % tree cover. While we do not focus on the contribution or role of shrubs in espinal in this paper, they are associated with increased ecosystem processes in espinal and thus contribute positively to espinal condition (Root-Bernstein and Jaksic 2015). We mapped an approximation of espinal condition using a multicriteria analysis. We combined available measures of NDVI from AVHRR (http://noaasis.noaa.gov/NOAASIS/ml/avhrr.html), and NPP and evapotranspiration (ET) from MODIS. NDVI was split between summer (October–March) and winter (April–September). Due to the reverse phenology of A. caven, summer NDVI is dominated by the contribution of A. caven in espinal, while winter NDVI is dominated by the contribution of the understory in espinal (Gerstmann et al. 2010). This allows us to split the two vegetation components. Our measure of NPP and ET was available at 1 km resolution, while the NDVI data were only available at 8 km resolution. We therefore took the mean of NPP and ET data at 8 km resolution. The ET represents a combination of soil and plant evapotranspiration. We combined these factors to produce an equally weighted index from 0 to 4, with 4 being an espinal with the relatively highest carbon production, photosynthetic activity in both the tree and understory layers, and movement of water into the atmosphere from both the soil and vegetation. These represent interrelated factors that espinal restoration aims to increase: tree growth, shade provision, forage provision, and soil moisture. We refer to this index as “espinal condition.” We do not explicitly consider biodiversity values in this index, as there are no available data, but we assume that biodiversity also benefits from less degraded and desertified conditions and greater shade, plant biomass, and available water (M. Root-Bernstein pers. obs.). Thus, 4 represents the best available espinal condition in a given year, and 0 the worst. We calculated the lowest cost paths from areas of high espinal condition (top two standard deviations) to lowest (bottom two standard deviations) and from lowest to highest (to ensure complete connectivity), to create a regional transhumance or transhumant rewilding network avoiding high elevation, urban areas, and roads. Lowest cost paths are representations of suitable movement corridors across the landscape, linking the identified espinal areas. We had complete data only between 1999 and 2006, and we focused on ENSO years, comparing El Niño (wet) years to La Niña (dry) years within this time period to assess the variability in paths across wet and dry years.
For each scenario (factory farms and orchards/vineyards; factory farms and sclerophyllous forests; Status quo; Restored espinals with guanacos), we followed and adapted the table of services in de Groot et al. (2010) to list the services provided via water input (rain, groundwater) in the landscape elements in question. We indicated those that are expected to increase relative to the status quo, and identified major costs and losses associated with establishing and maintaining the scenario. We further identified services that we expect to vary with or be independent of water input (e.g., dry and wet ENSO years), since water efficiency has been proposed as a key variable in evaluating livestock production models.
Preliminary experimental results
From the camera traps, we obtained 52 trap-days with images of guanaco. The data from 2014 correspond to pulse treatments, where all guanacos were in one section, so that the number of captured images was higher. We were able to observe the hours of greatest activity (hours with greatest percent photographs per day), group behavior, and foraging behavior (see Table 3, Supplementary Data). There were several peaks of activity during the morning, afternoon, and night. During the end of October–November 2014, the guanacos were photographically captured in a group, rather than alone, 59.6 % of the time (sections A and B combined), which fell to 32.2 % of the time in January (section A) and 5.8 % of the time in April (section A). At the end of October–November, grazing was observed 9/10 days with data in section A and 9/9 days in section B, while browsing was observed 7/10 days in A and 7/9 days in B. In January (section A), grazing was captured 7/20 days, while browsing was captured 5/20 days. In April (section A), grazing was captured only once and browsing only twice out of 13 days.
Our map of potential espinal distribution included 43,189 polygons or individual espinals, covering 1,174,206 ha. The mean size of espinals was 27 ha ± 66 (SD). The area-to-perimeter ratio, providing a measure of fragmentation, was mean 0.07 ± 0.04 (SD).
Fragment number, area, area-to-perimeter ratio, and mean espinal condition of high condition areas (top two SDs) and low condition areas (bottom two SDs)
Year espinal areas
Number of fragments
Mean area (km2)
Mean area-to-perimeter ratio
Mean espinal condition index
3.50 ± 3.11
0.13 ± 0.04
2.28 ± 0.40
7.50 ± 7.63
0.24 ± 0.21
0.47 ± 0.13
3.83 ± 3.01
0.12 ± 0.04
2.08 ± 0.23
10.57 ± 11.22
0.25 ± 0.21
0.68 ± 0.21
10.98 ± 9.62
0.20 ± 0.09
1.62 ± 0.20
7.77 ± 7.13
0.23 ± 0.16
0.50 ± 0.12
53.6 ± 0.39
1.82 ± 0.01
2.42 ± 0.38
11.6 ± 11.70
0.25 ± 0.21
0.72 ± 0.24
4.19 ± 3.47
0.12 ± 0.04
2.48 ± 0.37
11.43 ± 11.56
0.25 ± 0.21
0.74 ± 0.25
Ecosystem services produced by water input in the form of rain or groundwater under four land-use scenarios, and associated land-use costs and losses
Services produced by water inputs (per mm/year)
Factory farms and orchards/vineyards
Factory farms and sclerophyllous forest restoration with strict protection
Restored silvopastoral system with guanacos
Forest: air quality regulation
Inspiration for culture, art, design
Cultural heritage and identity
Spiritual and religious inspiration
Education and science
Wine and fruit production may come to acquire greater local specialty. Agronomical science opportunities
We assume that sclerophyllous forest has highest biodiversity and ecosystem services . We assume no provisioning value of forest under strict protection . Sclerophyllous forest is valued more highly than other landscapes/habitats 
Although there are no comprehensive estimates of espinal (or similar systems’) ecosystem services, we assume that their provision is suboptimal both in the short term and in terms of sustainability 
High-quality guanaco wool can contribute to raw materials . Urban Chileans positively rate areas with guanacos aesthetically and for tourism . Opportunities for socioecological system research
Polluted effluent water
Traditional landscape and culture
Polluted effluent water
Funding restoration of sclerophyllous forest
Opportunity cost to private landowners
Arriero land-use rights and forest protection conflicts
Low efficiency of livestock production
Ongoing conversion to fruit plantations/vineyards
Guanaco breeding or relocation facility
Guanaco care and management; CITES certification of wool
Feral dog control and associated social conflicts
Guanacos browse A. caven leaves and branches, voluntarily including both structures in their diet despite the presence of thorns. A priori this demonstrates the feasibility of using guanacos as browsers for espinal restoration. While we do not yet have data related to how effects on individual trees will scale up to landscape patterns, such scaling-up is characteristic of herbivory by large mobile mammals (Shipley 2007; Olff et al. 1999). We are unable to predict whether the level of effect on individual trees will show a linear scaling-up to a landscape level due to factors such as the distribution of browsing by guanacos at larger scales, variation in soil water availability, historicity effects due to past disturbances, or competition between A. caven and other tree species for resources in mixed habitats. We emphasize that at this stage, we are only demonstrating the feasibility of the potential intervention, and we are not in a position to estimate its effectiveness or effect size at any scale.
The lowest quality espinals are found predominantly in the north of the central zone of Chile, although we also observe values below the mean throughout central Chile. These patterns did not change dramatically across wet and dry ENSO years. The least-cost paths between highest and lowest quality espinals are within a feasible length range, by comparison with guanacos’ natural range sizes (see below) for a transhumant rewilding scenario using guanacos. Finally, the ecosystem service costs and benefits associated with guanaco transhumant rewilding are comparable to other scenarios, with potentially more institutionally achievable benefits and lower monetary costs.
Our observations of guanaco foraging on A. caven support the interpretation that leaves and branches are a normal element of their diet rather than a starvation food. Although our opportunistic camera trap data are probably biased due to the conflation of section, space availability (pulse treatments in 2014), and month, it suggests that group cohesion as well as browsing effort could be reduced in summer/fall. It remains to be seen whether the observed level of browsing during spring is adequate to stimulate a significant change in A. caven growth and with what lag time. Analysis of guanaco stocking density in espinal will allow us to determine a tree-to-guanaco density ratio where the amount of supplemental alfalfa provided could be reduced, which would substantially reduce guanaco maintenance costs as well as water input to the system.
The GIS analysis found very few putative espinals in the top range of condition for any year. This suggests that factors other than water availability are constraining espinal condition. The observed distribution of espinal condition values may be explained partially by the altitudinal gradient of the mountain ranges, the greater intensity of farming in the southern central valley, local socioeconomic factors affecting espinal exploitation history, or the interaction of these factors. The area of the selected espinals was larger in higher condition espinals and in wet years, which suggests that continuos espinals, regardless of the amount of edge, are in better condition and may capture available moisture more efficiently.
The obtained scale of transhumant rewilding also appears to be feasible. Guanacos can be either migratory, moving in family groups to winter ranges, or sedentary (Marino and Baldi 2014). The average range of three wild male guanacos, two of them solitary and one of them in a family group, tracked via GPS collar for a year in areas of savanna and shrub habitat, was 34 km2, varying between 8 and 98.4 km2 between seasons (Bonacic et al. unpublished data). This mean range size is close to the mean size of espinal fragments in our distribution map. The distances that guanacos migrate also appear to vary considerably, from 12 km (Ortega and Franklin 1995) to movements from the east to west within a 1700 km2 area [(actual distances moved are not reported) Puig et al. 2011]. This is comparable to the core overlap of paths in the identified transhumant rewilding network. The trade-offs in terms of body condition of guanacos for long-range movements should be studied further.
A finer-scale model informed by experimental results is necessary for the assessment of how a transhumant rewilding network can be implemented. Our analysis uses a coarse temporal scale (yearly) and a relatively coarse spatial scale (8 km resolution), and was only able to consider a small span of years for which all data were available. However, this is prudent given the clear limitations of our map of espinal distribution, which has not been ground-truthed, is outside the temporal range of our other data, and does not represent changes in espinal cover (estimates of which vary considerably, see Supplementary Data Table 5). The extent of espinal that we find, just over one million ha, is also only 25–50 % of published estimates (Serra 1997; Ovalle et al. 1990), i.e., less than can be accounted for by estimated land-cover change rates. In addition, the possibility of legacy effects from ENSO years was not examined here and should be considered (Monger et al.2015).
Our predictions for the relative water efficiency of potential ecosystem services under different land-use scenarios illustrate the complexity of trade-offs and valuation exercises that would be required in order to fully assess the water-use rationality of transhumant rewilding. The “factory farms and sclerophyllous forest restoration” scenario is the least realistic, since there is currently no governance mechanism by which a majority of espinal landowners would be induced to restore a protected forest type on their land (Root-Bernstein et al. 2013). The “factory farms and orchards/vineyards” scenario is similar to the current condition of California, where despite increasing droughts affecting natural habitats and ecosystem services, industrial agriculture monopolizes water resources (Diffenbaugh et al. 2015). The “restored silvopastoral system with guanacos” is predicted to increase the water efficiency of ecosystem services, not only due to restored espinals losing less water to evaporation, but also due to the potential for a transhumant rewilding network that would adjust restoration and management to optimize production over dry and wet years. The costs of guanaco breeding, care, management, and protection within fenced espinals appear more complex than expensive. However, comparing costs and benefits across scenarios is beyond our capacity and points to important data shortfalls associated with nearly all services, and their valuation.
Finally, we believe that developing a transhumant rewilding system can be compatible with the local socioeconomic and cultural context, although this should be the focus of further studies. Livestock transhumance, called “veranada,” is still practiced between the central valley of Chile and the Andes (Westreicher et al. 2007). Arrieros, or muleteers, also practice small-scale herding or pasturing throughout central Chile. Both smallholders and owners of large estates use espinal for extensive livestock, charcoal, and honey production, which we believe are compatible with guanaco browsing, either through rotation or coexistence. Reinforcement of fencing, adoption of guard animals, or active shepherding are possible strategies to reduce threats. However, uptake of guanaco management will clearly require regional or national programs for training and coordination, and a willingness among landholders to learn and adopt new skills.
In conclusion, our preliminary analysis and predictions associated with the Chilean case study suggest that traditions of transhumant or nomadic pastoralism can inspire feasible and flexible solutions to climate variation in drylands and silvopastoral systems. Notably, transhumant rewilding allows flexibility at regional and subregional scales of ecosystem processes and associated services in reacting to variance in water availability. (Silvo)pastoralism and transhumance here could serve as the human cultural interface, shepherding rewilded processes through complex anthropogenic landscapes.
Thanks to Joaquín Solo de Zaldivar, Fernanda Romero Gárate, and Francisco Novoa de la Maza for their cooperation with Proyecto REGenera; Jorge Ramos for guanaco behavior data collection; Samira Kolyaie for valuable GIS assistance; and Magdalena Bennett and Brody Sandel for GIS data preparation. MR-B was funded by Fondo Nacional de Desarrollo Cientifico y Tecnológico Post Doctoral Fellowship No. 3130336 and Danish National Research Foundation Niels Bohr professorship project Aarhus University Research on the Anthropocene (AURA). FMJ was funded by FB 0002-2014.
- Armesto JJ, Manuschevich D, Mora A, Smith-Ramirez C, Rozzi R, Abarzúa AR, Marquet PA (2009) From the Holocene to the Anthropocene: a historical framework for land cover change in southwestern South America in the past 15,000 years. Land Use Policy 27:148–160. doi:10.1016/j.landusepol.2009.07.006 CrossRefGoogle Scholar
- Baldi B, Lichtenstein G, González B, Funes M, Cuéllar E, Villalba L, Hoces D, Puig S (2008) Lama guanicoe. The IUCN red list of threatened species. Version 2015.2. www.iucnredlist.org. Accessed 23 July 2015
- Capone R, Iannetta M, El Bilali H, Colonna N, Debs P, Dernini S, Maiani G, Intorre F, Polito A, Turrini A, Cardone G, Lorusso F, Belsanti V (2013) A preliminary assessment of the environmental sustainability of the current Italian dietary pattern: water footprint related to food consumption. J Food Nutr Res 1(4):59–67. doi:10.12691/jfnr-1-4-5 Google Scholar
- Cronon W (1996) The trouble with wilderness: or, getting back to the wrong nature. Environ Hist 1:7–28. http://www.jstor.org/stable/3985059
- Dube F, Sotomayor A, Loewe V, Müller-Using B, Stolpe N, Zagal E, Doussoulin M (2016) Silvopastoral systems in temperate zones of Chile. In: Peri PL et al (eds) Silvopastoral systems in Southern South America, Adv Agrofor 11. doi:10.1007/978-3-319-24109-8_8
- FAO (1996) Declaration of the world summit on food security. Downloaded from http://www.fao.org/fileadmin/templates/wsfs/Summit/Docs/Final_Declaration/WSFS09_Declaration.pdf. Accessed 2015
- FAO (Food and Agriculture Organization of the United Nations) (2001) Pastoralism in the new millennium. FAO animal production and health paper 150, Rome, Italy. http://www.fao.org/docrep/005/y2647e/y2647e00.HTM Accessed 2015
- Fornara DA, du Toit JT (2007) Browsing lawns? Responses of Acacia nigrescens to ungulate browsing in an African savanna. Ecology 88(1):200–209. doi:10.1890/0012-9658(2007)88[200:BLROAN]2.0.CO;2Google Scholar
- Gordon IJ, Hester AJ, Festa-Bianchet M (2004) The management of wild large herbivores to meet economic, conservation and environmental objectives. J App Ecol 41:1021–1031. http://www.jstor.org/stable/3505778
- Gutiérrez JR, Armesto JJ (1981) El rol del ganado en la dispersión de las semillas de Acacia caven (Leguminosae). Cien Invest Agrar 8:3–8Google Scholar
- Hoffman T, Boerma D (2014) Ecosystem services provided by livestock species and breeds, with special consideration to the contributions of small-scale livestock keepers and pastoralists. FAO Commission on genetic resources for food and agriculture. Background study paper 66(1). http://www.fao.org/3/a-at598e.pdf
- Medina ME, Rivero DE (2007) Zooarqueología, Lama guanicoe y dinamica evolutiva del Chaco Serrano. Mundo de antes 5:211–234Google Scholar
- Miller S (1980) Human influence on the distribution and abundance of wild Chilean mammals: prehistoric-present. PhD thesis, University of WashingtonGoogle Scholar
- Navarro Gutiérrez RA (1995) Efecto de Intervenciones Silviculturales Sobre El Crecimiento y La Producción de Fitomasa de Acacia Caven En Milipilla Region Metropolitana. Dissertation, Universidad de ChileGoogle Scholar
- Oba G, Mengistu Z, Stenseth NC (2000) Compensatory growth of the African dwarf shrub Indigofera spinosa following simulated herbivory. Ecol Appl 10(4):1133–1146. doi:10.1890/1051-0761(2000)010[1133:CGOTAD]2.0.CO;2Google Scholar
- Olivares AE (2006) Relaciones entre el estrato arbóreo, el estrato herbáceo y la conducta animal en el matorral de Acacia caven (espinal). Sécheresse 17(1–2):333–339Google Scholar
- Ortega IM, Franklin WL (1995) Social organization, distribution and movements of a migratory guanaco population in the Chilean Patagonia. Rev Chil Hist Nat 68:489–500Google Scholar
- Peri PL, Dube F, Varella AC (2016) Silvopastoral systems in the subtropical and temperate zones of South America: an overview. In: Peri PL et al (eds) Silvopastoral systems in Southern South America, Adv Agrofor 11. doi:10.1007/978-3-319-24109-8_8
- Pimentel D, Houser J, Preiss E, White O, Fang H, Mesnick L, Barsky T, Tariche S, Schrek J, Alpert S (1997) Water resources: agriculture, the environment and society. BioScience 47(2):97–106. http://www.jstor.org/stable/1313020
- Poschlod P, Bonn S (1998) Changing dispersal processes in the central European landscape since the last ice age: an explanation for the actual decrease of plant species richness in different habitats? Acta Bot Neerl 47(1):27–44Google Scholar
- Rojas P, Gonzálex M, Benedetti S, Yates P, Sotomayor A, Dube F (2016) Silvopastoral systems in arid and semiarid zones of Chile. In Peri PL et al (eds) Silvopastoral systems in Southern South America, Adv Agrofor 11. doi:10.1007/978-3-319-24109-8_7
- Serra MT (1997) Acacia caven (Mol.) Mol. En: FAO: Especies arbóreas y arbustivas para las zonas áridas y semi áridas de América Latina. Serie Zonas Áridas y Semiáridas N° 12(1):57–168Google Scholar
- Simonetti JA (1999) Diversity and conservation of terrestrial vertebrates in mediterranean Chile. Rev Chil Hist Nat 72:493–500Google Scholar
- Svenning J-C, Pedersen PBM, Donlan CJ, Ejnæs R, Faurby S, Galetti M, Hansen DM, Sandel B, Sandom CJ, Terborgh JW, Vera FWM (2016) Science for a wilder anthropocene: synthesis and future directions for trophic rewilding research. PNAS 113(4):898–906. doi:10.1073/pnas.1502556112 CrossRefGoogle Scholar
- Vermeulen SJ, Aggarwal PK, Ainslie A, Angelone C, Campbell BM, Challinor AJ, Hansen JW, Ingram JSI, Jarvis A, Kristjanson P, Lau C, Nelson GC, Thornton PK, Wollenberg E (2012) Options for support to agriculture and food security under climate change. Environ Sci Policy 15:136–144. doi:10.1016/j.envsci.2011.09.003 CrossRefGoogle Scholar
- Vita A, Serra MT, Grez I, González M, Olivares DA (1995) Intervenciones silviculturales en espino (Acacia caven (Mol.)) en la zona árida de Chile. Rev Cien Forest (Chile) 10:51–62Google Scholar
- Westreicher CA, Mérega JL, Palmili G (2007) The economics of pastoralism: study on current practices in South America. Nomad People 11(2):87–105. http://www.jstor.org/stable/43123802