Global Ecology and Conservation 3 (2015) 776–788

Contents lists available at ScienceDirect

Global Ecology and Conservation

journal homepage: www.elsevier.com/locate/gecco

Original research article

Ecosystem process interactions between central ChileanhabitatsMeredith Root-Bernstein a,b,∗, Fabián M. Jaksic a,ca Department of Ecology, Pontificia Universidad Católica de Chile, Santiago, Chileb Bioscience Department, Aarhus University, Aarhus, Denmarkc Center of Applied Ecology & Sustainability (CAPES), Pontificia Universidad Católica de Chile, Santiago, Chile

a r t i c l e i n f o

Article history:Received 18 December 2014Received in revised form 13 April 2015Accepted 14 April 2015Available online 20 April 2015

Keywords:Acacia cavenEcosystem processEspinalFunctional traitMatorralSilvopastoral system

a b s t r a c t

Understanding ecosystem processes is vital for developing dynamic adaptivemanagementof human-dominated landscapes. We focus on conservation and management of thecentral Chilean silvopastoral savanna habitat called ‘‘espinal’’, which often occurs nearmatorral, a shrub habitat. Although matorral, espinal and native sclerophyllous forest arelinked successionally, they are not jointly managed and conserved. Management goalsin ‘‘espinal’’ include increasing woody cover, particularly of the dominant tree Acaciacaven, improving herbaceous forage quality, and increasing soil fertility.We askedwhetheradjacentmatorral areas contribute to espinal ecosystemprocesses related to the threemainespinal management goals. We examined input and outcome ecosystem processes relatedto these goals in matorral and espinal with and without shrub understory. We found thatmatorral had the largest sets of inputs to ecosystem processes, and espinal with shrubunderstory had the largest sets of outcomes.Moreover, we found that these outcomeswerebroadly in the directions preferred bymanagement goals. This supports our prediction thatmatorral acts as an ecosystem process bank for espinal. We recommend that managementplans for landscape resilience consider espinal and matorral as a single landscape coverclass that should be maintained as a dynamic mosaic. Joint management of espinal andmatorral could create new management and policy opportunities.© 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC

BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

Understanding ecosystem processes is essential to developing conservation and management interventions. Ecosystemprocesses are broadly controlled by the distribution of functional traits, habitat modification by species, and abiotic inputscontrolling primary productivity (Crain and Bertness, 2006; Fischer et al., 2006; Mouillot et al., 2011; Maestre et al., 2012).Anthropogenic transformations due to agriculture, forestry and climate change, as well as natural successional processes,affect ecosystem processes by creating heterogeneity (Loreau et al., 2001; Fischer et al., 2006). Within habitat mosaics, eachhabitat type may harbor components of biodiversity, abiotic inputs and physical substrates needed for different ecosystemprocesses (Loreau et al., 2003; Fischer et al., 2006). Ecosystemprocesses, and their inputs, also travel beyond their immediatespatial distributions via biotic and abiotic fluxes and interactions (e.g. Rand et al., 2006; Alongi, 2008). Thus landscape-scaleheterogeneity may contribute to providing more inputs for more ecosystem processes, yet many studies show that it canalso reduce overall functioning due to island and matrix effects (Loreau et al., 2001, 2003; Fischer et al., 2006).

∗ Corresponding author at: Department of Ecology, Pontificia Universidad Católica de Chile, Santiago, Chile.E-mail address:mrootbernstein@gmail.com (M. Root-Bernstein).

http://dx.doi.org/10.1016/j.gecco.2015.04.0072351-9894/© 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

http://dx.doi.org/10.1016/j.gecco.2015.04.007http://www.elsevier.com/locate/geccohttp://www.elsevier.com/locate/geccohttp://crossmark.crossref.org/dialog/?doi=10.1016/j.gecco.2015.04.007&domain=pdfhttp://creativecommons.org/licenses/by-nc-nd/4.0/http://creativecommons.org/licenses/by-nc-nd/4.0/http://creativecommons.org/licenses/by-nc-nd/4.0/http://creativecommons.org/licenses/by-nc-nd/4.0/http://creativecommons.org/licenses/by-nc-nd/4.0/http://creativecommons.org/licenses/by-nc-nd/4.0/http://creativecommons.org/licenses/by-nc-nd/4.0/mailto:mrootbernstein@gmail.comhttp://dx.doi.org/10.1016/j.gecco.2015.04.007http://creativecommons.org/licenses/by-nc-nd/4.0/http://creativecommons.org/licenses/by-nc-nd/4.0/http://creativecommons.org/licenses/by-nc-nd/4.0/http://creativecommons.org/licenses/by-nc-nd/4.0/http://creativecommons.org/licenses/by-nc-nd/4.0/http://creativecommons.org/licenses/by-nc-nd/4.0/http://creativecommons.org/licenses/by-nc-nd/4.0/

M. Root-Bernstein, F.M. Jaksic / Global Ecology and Conservation 3 (2015) 776–788 777

One possible explanation for these contrasting outcomes is that successionally linked mosaic patches may increaseoverall functioning within the landscape, while mosaics not linked by a successional pathway may more frequently showdecreased functioning as a result of fragmentation (Odum, 1969). By successional, we refer broadly to all endogenous habitatchanges, including shiftingmosaics, while by non-successional wemean land cover change created andmaintained throughhigh levels of disturbance and human niche construction, such as cropland or urban areas. Successional pathways betweenhabitat types imply spatiotemporal dispersal of nutrients, propagules, and/or ecosystem functions at a landscape scale. Thus,when habitat types are linked via successional processes, the functional ‘‘insurance’’ effect of diverse taxon dispersal acrosslandscapes (Loreau et al., 2003) may be facilitated, compared to habitat types not linked by succession. This is becausecropland, for example, acts as a successional sink. Ecosystem processes leading to succession, e.g. seed dispersal, may entercropland but the cropland is either physically inhospitable or maintained through anthropogenic disturbance in a state farfrom natural succession, until abandonment. While the flux of ecosystem functions across anthropogenically maintainedmosaics with cropland, plantations, cities, etc. is of particular interest to conservationists taking an ecosystem servicesperspective, many semi-natural production systems, such as rangelands, silvopastoralism and agroforestry, can bemanagedfor natural successional mosaics (Fischer et al., 2006).

Ecosystem functions, and the ecosystem services they provide, can be difficult to define and measure, due to theircomplexity. Underlying ecosystem functions are multiple functional traits belonging to many species with different inspatial and temporal distributions, which together contribute at different rates and in different quantities to ecosystemfunction dynamics (Benggtson, 1998; de Groot et al., 2010). Functional traits contribute to ecosystem processes that, inturn, have difficult-to-assess spatiotemporal dynamics. Functional traits and other inputs each have a different strength ofcontribution to the ecosystem process, a different range and frequency of mobility, and a different lag or residence time atthe destination. For example, if seed germination depends on seed production, exozoochory and nurse plant availability,then the spatiotemporal distribution of the seed germination process depends on the spatiotemporal dynamics of shrubseed production, animal movement, and nurse plant distribution. It is unlikely that ecosystem process distributions show alinear decay away from the area of origin (e.g. the seed producing shrubs). Rather, they are likely to distribute in complexand patchy ways. Although there is a great deal of literature on regional and continental-scale spatial mapping of ecosystemservices, practical difficulties limit attempts to trace the spatiotemporal dynamics of ecosystem processes at a landscape orpatch scale (for related approaches see Jordano et al. 2007; Root-Bernstein et al., 2013a).

In addition, ecosystem processes are often cyclical, such as reproduction, thewater cycle, or trophic energy transfer (FathandHalnes, 2007; Scanlon et al., 2005). Because a given functional trait can both cause and be affected by a cyclical ecosystemprocess, functional traits are not only inputs to processes, but also outcomes. The input–outcome relationship is similar to theresponse-and-effect framework (Suding et al. 2008; Laliberté et al., 2010), but here we focus on the endogenous ecosystemprocesses underlying succession (Odum, 1969), rather than the functional responses uniquely associated with exogenousfactors such as climate change. The functional traits that occur in a habitat as a result of successional processes (outcomes)are rarely studied in relation to functional trait inputs (but see Eldridge et al., 2011).

An additional complication, from a community ecological point of view, may be that counting functional traits aloneleaves out many important characteristics of ecosystems that interact closely with other traits, such as properties of thesoil. Methods that mix functional traits and other functional non-trait elements fit better with diverse perspectives (e.g.ecosystem engineering, Crain and Bertness, 2006), and can provide good ecological models (e.g. Maestre et al., 2012; Dantaset al., 2013).

We address the concept of ecosystem process interactions with a non-spatially explicit, non-temporally explicitapproach, through a case study of the central Chileanmediterranean-climate habitats espinal andmatorral (see Fig. 1; Root-Bernstein and Jaksic, 2013; Maestre et al., 2012). Espinal is a savanna dominated by Acacia caven, traditionally used as asilvopastoral system (Ovalle et al., 1990; Fuentes et al., 1989). Espinal can be found with or without a shrub understory, andoften occurs next to matorral, a dense shrub habitat typical of the foothills of the Andes and the coastal mountain range(Donoso, 1982). The successional relationships between espinal habitats and matorral have been largely ignored after a fewearly studies (Armesto andPickett, 1985; Fuentes et al., 1986) and are only recently attracting renewed interest as an elementof land cover change (e.g. Hernández et al., 2015; Fuentes-Castillo et al., 2012; Newton et al., 2011). Espinal successionmay be characterized as in Fig. 2. Both matorral and espinal have been described as degradations of native sclerophyllousforest (assumed to have been the dominant climax habitat in prehispanic Chile), implying a simple successional model inwhich forest is degraded, crosses a threshold and rarely recovers (see Ovalle et al., 1990; Aronson et al. 1993; Schulz et al.,2010; van deWouw et al., 2011). A. caven can also be described as a slow-reproducing pioneer species that establishes afteranthropogenic disturbance, suggesting that it is not a dead-end degraded state, but rather the initial stage of a successionalpathway (Fuentes et al., 1989; Baranelli et al., 1995; Torres et al., 2002; Root-Bernstein and Jaksic, 2013). Espinal can serveas nurse plant habitats, making them important for regrowth of sclerophyllous forests in central Chile (Hernández et al.,2015; Fuentes-Castillo et al., 2012; pers. comm. C. Peña). Within espinal it is often possible to find some sclerophyllous treespecies at low density (pers. obs. MR-B). However, anthropogenic disturbances of forest and matorral, and overgrazing ofespinal areas preventing tree and shrub recruitment may result in the last part of the cycle, sclerophyllous forest regrowth,rarely occurring.

The lack of a well-studied dynamic functional viewpoint on matorral–espinal–sclerophyllous forest relationships haseffects on their conservation and management. The three habitats are formally considered as separate and unrelated, withwidely differing protections and management regimes. From an institutional governance perspective, both matorral and

778 M. Root-Bernstein, F.M. Jaksic / Global Ecology and Conservation 3 (2015) 776–788

Fig. 1. Top left, espinal in winter with 3 individual sclerophyllous trees (middle ground, identifiable by dark green leaves). Top right, shrub espinal inwinter. Bottom left, shrub espinal in summer. Bottom right, matorral in winter (note individual sclerophyllous trees in the ravine, top left of photo). Topleft photo©Adrien Lindon, other photos©MR-B.

Fig. 2. Potential successional pathways between sclerophyllous forest, espinal and matorral. Arrows indicate exogenous or endogenous processes, asappropriate. We have marked with question marks explanations that we consider to be uncertain or speculative. In this study we only consider theecosystem processes underlying succession in espinal, shrub espinal and matorral.

espinal may be consideredmarginal ‘‘orphan’’ habitats (Buckingham et al., 2011) without clearly developed or implementedmanagement and protection goals (see Root-Bernstein and Jaksic, 2013). Matorral has no conservation protections perse. Espinal areas are managed for silvopastoral production from a traditional agronomical perspective (Olivares, 2006a,b;Ovalle et al., 1990). Acacia caven is sometimes considered to be a protected native species that cannot be cut down, butinterpretation in practice appears to be inconsistent (MR-B pers. obs.). By contrast, sclerophyllous forest is protected by theNative Forest law (N° 20.283). Here, we focus specifically on the functional links between matorral and espinal, the leastconserved and most heavily exploited of the three habitats.

We adopt an approach that allows us to combine functional traits and other functional elements, and to divide thembetween inputs to ecological processes and outcomes of those processes. We hypothesize that matorral and espinal withandwithout shrubs show exchanges of ecosystemprocesses.We look at the net effects of that exchange across habitat types,considering each habitat type as an abstract box. We do not attempt to account for how the net effects vary with habitatspatiotemporal distribution. Since matorral is usually considered to be the more natural and least degraded of the habitats,we predict that matorral will ‘‘subsidize’’ espinal (with and without shrubs) ecosystem processes, such that traits and other

M. Root-Bernstein, F.M. Jaksic / Global Ecology and Conservation 3 (2015) 776–788 779

inputs found in matorral can account for ecosystem process outcomes in adjacent espinal. This assumes that espinal is partof the successional pathway. If the successional pathway in Fig. 2 is correct, then the greatest subsidization effect should bebetween matorral and espinal with shrub understory. By contrast, if espinal habitats are degraded dead ends, they shouldact as ecosystem process sinks. Ecosystem process inputs from matorral will have no outcomes in espinal habitats.

We look for this ‘‘spillover’’ or subsidization effect with a measure (a convex hull, see Methods) that incorporatesabundance and divergence (range) for groups of functional traits and other variables. Each variable exists on an independentaxis, so that direct comparisons between groups of variables that are not all traits, or that have different units, is valid aslong as the set of axes is the same for both groups. Input variable sets can then be compared to outcome variable sets acrossadjacent successional matorral–espinal habitat transitions to understand, inductively, the direction of ecosystem processflow.

We focus on ecosystem processes that are targets of existing silvopastoral management in espinal, specifically (1) highquality herbaceous biomass production, (2)woody plant reproduction, and (3) soil fertility improvement.Wepredict that (1)matorral should have larger sets of input ecological variables than espinal, and (2) both espinal habitats (with and withoutshrubs) should show a disproportionately wide range of outcomes, relative to its inputs. We further examine whetheroutcomes are consistent with the goals of espinal conservation and management.

2. Materials and methods

2.1. Research site

The study took place in the Pajaritos area of the Estación Experimental Rinconada deMaipú (33°23′S, 70°31′ W, elevation495 m), a field station of Universidad de Chile, Santiago, Chile. This is located within the semi-arid mediterranean-climateregion of central Chile, characterized by a rainy winter season when most plant growth occurs, and a hot dry summer.Our study site encompassed a mosaic of espinal (Acacia caven savanna), with and without a shrub understory, and matorral(shrub habitat) remnants on and near hillsides. There is no sclerophyllous forest at the site, althoughwe observed individualsof the sclerophyllous trees Lithrea caustica and Quillaja saponaria (by chance none occurred in the plots). The dominanttree characterizing espinal is A. caven, a small thorny native South American acacia with an inverse phenology to non-evergreen plants in central Chile, such that its leaves appear in late spring and remain green during the summer. Previousresearch at this site indicates that some invasive herbs from other mediterranean-climate regions are highly abundant, butthat there are more than twice as many Chilean mediterranean-zone endemics as non-endemic species (Root-Bernsteinet al., 2014). Common vertebrate species at the site include sheep, European rabbits (Oryctolagus cuniculus), culpeo foxes(Lycalopex culpaeus), rodents such as the degu Octodon degus (semifossorial) and the cururo Spalacopus cyanus (fossorial),several raptors and a few dozen granivorous and insectivorous bird species, and several species of lizard and snake. ThePajaritos section, which is partially fenced, is 525 ha of a total 898 ha used as pasture for Suffulk Down and Merino sheep.The sheep are kept at a stocking rate of 0.92 sheep equivalents per ha over the entire 898 ha, yielding 18.3 kg of meat/haand 3.0 kg of wool/ha (C. Araneda pers. comm.). Although not part of the official management plan, cattle or their feces aresometimes observed in the area.

2.2. Plots

Ecological variable datawas collected at the plot scale. Eachplotwas a 2m×2msquare. Plotswere arranged into transectsto facilitate fieldwork, and were placed at 10 m intervals. Transects were placed about 100 m apart using stratified randomsampling, within a matorral–espinal transition area. 40 plots were placed within two matorral remnants in the foothills onthe north side of Pajaritos. 50 plots were placed within the continuous espinal habitat south of the hills, between 200 and300 m from the nearest matorral fragments. 70 plots were placed within espinal with a shrub understory (hereafter ‘shrubespinal’). 40 of these shrub espinal plotswere in an area adjacent to amatorral remnant (within 25mof thematorral–espinaltransition), while the other 30 were 300–500 m frommatorral, with a ∼100 m grassland gap between matorral and espinaledges.

Sincewe are considering a range of ecosystem functions and variables (see Variables section below), in order tominimizepseudoreplication andmaximize accuracy across variables,wehave chosen small plots relatively far apart. Soil fertility variesat a fine scale due to factors such as slope, plant structure, and local nutrient inputs such as feces or decomposing matter, so4 m2 plots spaced 1000 m2 blocks apart (10 m × 100 m) are likely to be independently affected for soil fertility inputs andoutcomes. For forage provision inputs, at our site factors such as the combination of small mammal disturbances andwoodyvegetation structure also vary at a scale smaller than 1000 m2 blocks. For woody shrub establishment functions, birds andrabbits are both habitat selective during foraging at less than 1000 m2 scales (Root-Bernstein et al., 2013a,b; Jaksic et al.,1979).

2.3. Variables

We collected two sets of ecological variables, inputs and outcomes. Inputs were chosen as those variables that we expectto move between habitats due to successional processes. Outcome variables were chosen as variables characterizing the

780 M. Root-Bernstein, F.M. Jaksic / Global Ecology and Conservation 3 (2015) 776–788

Table 1Variables used and how they were measured. Filled-in boxes indicate input or outcome variables.

Ecological function Variable Data from literature Data from field Scale of data collected in field

Forage Input Octodon degus disturbances x plot

provision Rabbit disturbances x plot

Bird disturbances x plot

Tree canopy cover x plotOutcome Secondary metabolites x

Rosette diameter x One site, exemplarsamples

Hairy, thorny, stinging x One site, exemplarsamples

Herbaceous biomass x One site, exemplarsamples

Leaf area x One site, exemplarsamples

Soil fertility Input Octodon degus disturbances x plot

Rabbit disturbances x plot

Bird disturbances x plot

Bird feces x plot

Shrub canopy cover x plot

Hairy, thorny, stinging x One site, exemplar

samples

Microarthropod detritivore x plot

biomass

Tree canopy cover x plotOutcome Nitrate x Plot; but analyzed

Ammonium x by grouping plotsPhosphorous x into 11 habitat-Percent organic matter x cover categories

(Table 2).Woody plant Input Granivorous birds

establishment Rabbit pits

Bird pits

Recent fireOutcome Tree canopy cover x plot

Shrub canopy cover x plotReproduction type xDispersal type xShade tolerance xRoot:shoot ratios x1-year sapling or shoot heights xSeed diameters of woody species x

quality and quantity of the ecosystem process provided, with relevance to silvopastoral management. Input and outcomevariables included functional traits, modified habitat features and abiotic variables (Maestre et al., 2012). We definedinput variables as factors identified in the literature as initiating, facilitating, or otherwise making possible the ecosystemprocesses of interest. We defined outcomes as characterizations of the set of species life-history strategies and ecologicalconditions that result from the ecosystem processes identified as inputs. Since ecosystem processes can be cyclical orhave multiple feedback mechanisms, we took a restoration perspective and defined outcomes as life-history strategiesand ecological conditions expected to result at least a year after a set of ecological processes was initiated or restored toa habitat. We considered each variable to be either an input or an output of a given process, but not both. Here we give asuccinct justification for each ecological variable used (see Table 1).

For forage provision, we argue that ecosystem process movement primarily takes the form of plant species dispersal.Higher densities of a variety of microclimates and microsites suitable for plant establishment in a given habitat provide alarger species pool for dispersal to adjacent habitats. At the same time, diffusion of animals producing microsites throughdisturbance from a high-density habitat to a low-density habitat can contribute to some ecosystem process movement.

Inputs included disturbances to the soil by the social rodent Octodon degus, pits dug by rabbits (O. cuniculus), cururos (S.cyanus) and birds (mainly Sturnella loyca), and tree canopy cover. The soil disturbances by smallmammals and birds increaseherbaceous species diversity in this region (Root-Bernstein et al., 2014; Root-Bernstein and Ebensperger 2012; Kelt, 2011).Higher herbaceous species diversity is likely to mean a broader range of functional traits (Loreau et al., 2001). Tree canopy

M. Root-Bernstein, F.M. Jaksic / Global Ecology and Conservation 3 (2015) 776–788 781

Table 2Number of plot-samples homogenized to form a combined soil sample for each woody cover class. The total volume of eachcombined soil sample is equal to approximately 100 g × the number of plot samples. From each combined sample, two 100g subsamples were taken from the homogenized combined soil sample for lab analysis.

Woody cover class Matorral Shrub espinal Espinal

First quartile (0%) 16 40 46Second quartile (1%–10%) 12 17 3Third quartile (11%–45%) 6 7 1Fourth quartile (46%–100%) 6 6 0

cover increases herbaceous biomass in this semi-arid region (Olivares, 2006a). We did not consider abiotic variables suchas rainfall or slope that, though relevant to overall biomass production, cannot be considered independent between plots.The tree canopy effect appears to have a larger local effect on available soil moisture than these other variables in any case(Olivares, 2006a; pers. obs. MR-B). Large-scale non-independent factors add to the background variance to which speciesadapt in each habitat, and will be reflected in other variable values.

Forage provision outcomes sought to characterize the overall condition of the herbaceous community and its suitabilityas forage. Herbaceous communities that best support livestock in the long termwill have amix of palatable and unpalatablespecies, along with other adaptations to herbivory, and will have high biomass and large leaves. We obtained partialdata from the literature on secondary metabolite presence in the herbaceous plant community (Bustamante et al., 2006;Niemeyer, 1995), and directly measured rosette diameter and presence of hairy, thorny or stinging traits. These variableswere intended to indicate anti-herbivory adaptations thatwould generally reduce the value of the plant community as forage(although eating some plants with secondary metabolites can increase livestock weight gain Provenza and Papachristou,2009). Herbaceous biomass and leaf area were intended as positive indicators of forage quality. The period during whichforage biomass was available did not vary significantly within the annual species and thus was not considered as a variable.

For soil fertility, we argue that an increase in soil turnover, soil moisture, and decompositional processes can leave alegacy of higher soil fertility. This is likely to affect other habitats partly through diffusion of these functions beyond thehabitat of high density due to animal movements, and over longer time periods due to shifting mosaics of habitat typesexchanging positions.

Soil fertility inputs included several classes of variables. Soil disturbances and areas of activity of O. degus, rabbits andbirds, and bird feces, are variables which represent the contribution to nutrient inputs from small mammal and bird fecesand the trapping of moisture and litter in disturbed microsites (Dean and Milton, 1991). Shrub canopy and spines wouldalso contribute to trapping litter prior to its decomposition (Padilla and Pugnaire, 2006; Gómez-Aparicio et al., 2004).Macroarthropod detritivore biomasswas included because arthropod detritivoresmay affect the decomposition rate (cf. Voset al., 2011; Coulis et al., 2013). Finally, A. caven cover was used as a proxy for nitrogen fixing by this leguminous tree (Sitterset al., 2013).

Soil fertility outcomes were measured with four common variables of soil quality: nitrate concentration, ammoniumconcentration, phosphorous concentration, and percent organic matter. As these increase, soil fertility is considered toincrease.

For the movement of woody plant establishment processes we make a similar argument to soil fertility processes. Inthe short term, a high density of animals producing germination microsites and transporting seeds in one habitat type candiffuse out into adjacent ones. Fire can also move between habitats. In the longer term, the outcomes of these functionaltraits appear to have moved into another habitat because the other habitat has actually moved into the space where theywere previously more abundant, due to succession.

Woody plant establishment input variables included a measure of granivorous avian presence, abundance of pits madeby rabbits and birds, and recent fire presence. Birds are the most important granivores and endozoochorous seed dispersersin central Chile, with ants collecting up to as many seeds, but during a shorter period of the year (late summer only) (Keltet al., 2004; Vasquez et al. 1995; Jiménez and Armesto, 1992). Ants were not present in significant numbers during our fieldseason. Pits made by rabbits and birds can act as seed traps that facilitate germination (Dean and Milton, 1991). Althoughnaturally occurring fire is thought to be rare in central Chile, some shrubs and trees show vigorous vegetative resproutingafter fire, suggesting that they are to some degree fire-adapted (see also Gómez-González et al., 2008). Anthropogenic firesare now common in central Chile. Consequently, recent fire can favor the establishment or regrowth of particular shrubspecies more than others.

Woody plant establishment forms a species pool for future succession and ongoing ecosystem processes, so we soughtto measure variables related to the reproductive strategies and capacities of established woody plants. Outcomes obviouslyincluded canopy cover of shrubs and trees.We also sought to understandwhether woody plants were favoring vegetative orsexual reproduction, wind or endo- and ecto-zoochory dispersal, and whether they showed rapid or slow growth and wereprimarily shade or drought tolerant. These variables can help to characterize how shrub understories grow and spread frommatorral, and how trees reproduce (Cornelissen et al., 2003). These variables included the abundance of phanerophytes andendozoochory-dispersed woody species (Arroyo and Uslar, 1993) and shade tolerant species (Badano et al., 2005), as wellas the root:shoot ratios, 1-year sapling or shoot heights, and seed diameters of woody species (Hoffman and Kummerow1978; Montenegro et al., 1979; Hoffmann et al., 1989; Aronson, 1992; Ginocchio andMontenegro, 1992; Torres et al., 2002).

782 M. Root-Bernstein, F.M. Jaksic / Global Ecology and Conservation 3 (2015) 776–788

Although seed mass is considered a better functional trait than seed diameter for understanding plant adaptive strategiesbecause it provides information on the investment in nutrients per seed by the parent plant, which may interact withdispersal strategy (Cornelissen et al., 2003), these data were not available in the literature and collection was not feasible.The root:shoot ratio is a rough measure of investment in water uptake vs. photosynthesis, and reflects a growth strategy tooptimize across experienced abiotic limitations (Cornelissen et al., 2003). The 1-year sapling or shoot heights were used as aproxy for the ability to recover after a disturbance such as a fire, and/or establish from an independent seedling (Cornelissenet al., 2003).

Fivemain types of datawere collected in the field: (1) Abundances of herbaceous species andwoody species canopy coverfor each plot; (2) Invertebrate detritivore abundances; (3) Soil samples; (4) Measures of small mammal and bird presenceor activity; (5) Plant functional traits (see Table 1). In designing data collection, we preferred methods that were efficient aswell as accurate and that could potentially be developed into ecological indicators for adaptivemanagement. This facilitatestransfer to practice.

(1) Herbaceous and canopy cover data were collected at the period of maximum diversity, abundance and biomass,throughout October 2012. Abundances of herbaceous species were estimated using an analogue of the pin-grid methodadapted for use at our site (Jonasson, 1983). Herbaceous vegetation sampling consisted of placing a 20 cm × 20 cmplastic frame at 2 locations within each 2 m × 2 m plot. The plants growing within each frame were recorded bytaking photographs with a digital camera in autofocus mode, from a distance of approximately 1 m above the ground.Plants were identified to species level in SamplePoint (R), which records the classification into user-defined categoriesat crosshairs (‘pins’) arranged in a regular grid over a digital photograph (Booth and Cox, 2011). We set our grid to 25crosshairs per photo, the equivalent of one pin per 16 cm2 in a pin-point sampling protocol. Plants rarely completelyoverlapped due to an absolute low abundance in all plots and it was possible to identify each species from thephotographs by the shape of cotyledons, leaves, and flowers. Overlapwould lead to an overestimation of the abundancesof large plants and an underestimate of the abundances of small rare plants, but previous studies at this site using thismethodology found that overlap occurred at less than 10% of ‘pins’ (Root-Bernstein et al., 2014). Due to this possible bias,this method does not yield a true measure of herbaceous populations. We measured a mean of 5.43 ± 0.01 (standarderror) herbaceous species in each grid, with 32 species observed in total. Woody species were identified in the field andcanopy cover was estimated as the area of its projection onto the ground.

(2) Invertebrate detritivore abundances were measured via pitfall traps. One pitfall trap was set in the center of each plot.Pitfall traps consisted of two sturdy plastic disposable coffee cups, one inside the other, set in a hole so that the rim waslevel with the soil surface. Digging-in effects were avoided by placing the traps between 12–19 October, but only fillingthemwith liquid on the 25th and 26th of October 2012. The liquid consisted of a 50:50mixture of glycerine and ethanol.Invertebrates were collected from the traps on the 7–9 November, after having been left open for approximately 10days. Detritivores were identified to family level and separated from other invertebrates on the basis of a key developedduring past research at the same site (Root-Bernstein et al., 2013a,b). Detritivores were then dried for 48 h at 60° C andimmediately weighed to determine the dry biomass per plot.

(3) Soil sampleswere taken from the first 10 cmdepth,with one sample of around100 g per plot (the resultant holewas usedto install the pitfall traps). The cost of analyzing each sample individually was prohibitive. Consequently we assignedeach plot to a habitat type (espinal, shrub espinal, matorral) and four woody cover classes divided on the quartile valuesof percent woody cover, yielding 12 habitat-cover categories (Table 2). There were no samples in the category espinal-highest cover class, so only 11 categories were used. Feces, fresh plantmaterial, litter and stoneswere excluded from thesamples. Soil samples were stored in plastic bags at 5–8° C for a month during collection and mixing. The soil sampleswere mixed together according to the habitat-cover category to which the plot belonged (Robertson et al. 1999). Two100 g subsamples were taken from each soil mixture, except for one habitat-cover category in which the single 100 gsample from one plot was the only sample. This yielded 21 soil samples. These samples were sent to the Soil Laboratoryat the Universidad de Concepción, Chile, for analysis. The samples were analyzed for percent organic material, mg/Kgnitrate, mg/Kg ammonium, and mg/Kg Olsen phosphorous. A mean value per category was calculated for each measureand then attributed to each plot in that category.

(4) Degu (Octodon degus) herbivory and biopedturbation occur primarily on and adjacent to their runways, and is associatedwith changes in the plant community (Root-Bernstein et al., 2014). Degu runways connect degu burrows. To estimatedegu activity levels we thus counted the number of runways (identified as linear 8-cm wide paths) crossing each plotand the number of burrow entrances in each plot. Number of cururo (S. cyanus) burrow entrances was counted as ameasure of cururo disturbance activity (Contreras and Gutiérrez, 1991; Contreras et al., 1993). We also counted thenumber of pits dug by rabbits and birds in each plot. Pits are funnel-shaped and between 2 and 15 cm deep. We did notdistinguish between rabbit and bird pits during counting. As most bird pits are made during winter by Sturnella loyca,we assayed all other recent bird activity in each plot by counting bird feces. To estimate howmuch of this activity couldbe attributed to granivorous birds, we conducted two bird surveys on the 13th of October and the 11th of November.Surveys took place along six transects, two in each habitat type. Surveys took place between dawn and 10:45. Birdswere identified to species level visually, which was adequate due to the open habitat. Only birds within 100 m of thetransect were included; a previous bird survey in this study site showed no significant effect of habitat on detectability(Root-Bernstein et al., 2013a,b). The percent of granivorous birds by abundance was calculated for each transect, and

M. Root-Bernstein, F.M. Jaksic / Global Ecology and Conservation 3 (2015) 776–788 783

Fig. 3. Example of convex hulls using sites as the data points. The three axes are the same between habitats A and B, making the convex hulls comparable.Each sampled site for each habitat is plotted along the axes representing the variables collected. The n-dimensional convex hull (here, 3 dimensionalbecause there are 3 axes) for each habitat is a surface that contains the data points, here all the sites. The volume of each hull can be calculated and directlycompared across habitats. This gives a measure of the amount of variable space needed to describe all the sites of each habitat. The centroid, here shown asa white dot, lies at the n-dimensional center of each hull. The linear distance in n dimensions between each centroid can be calculated comparing habitatsA and B, e.g. (xi, yi, zi) – (xj, yj, zj). This gives an indication of the tendency towards certain variable values, or more generally a difference in variabledistribution between each habitat. An imprecise one-dimensional analogy would be to comparing the means and ranges of two box and whisker plots.

the mean was calculated for each habitat type. We multiplied the percent granivorous birds in each habitat type by thenumber of bird feces in each plot to get an estimate of granivorous bird activity in each plot.

(5) Plant functional traits were estimated by selecting well-grown but not extreme individuals (Weiher et al. 1999). Weselected specimens from a transitional area between matorral and espinal that was not within a plot. Rosette diameterand spine length were measured on 10–15 individuals of all species displaying these traits. Leaf area was measuredby collecting 5 adult leaves of each species, each from a different individual. Leaves were photographed on grid paperand their area was estimated. Mean leaf area was calculated for each species. Biomass was estimated for each speciesby collecting 5 individual plants, excluding roots, which were dried at 60oC for 72 hr and then weighed to 0.01 g todetermine the dry mass. A mean was calculated for each species. These data were then attributed to plots by calculatingthe abundance weighted means.

At the plot scale, all data were presented either as an abundance weighted mean, or as an abundance, as appropriate. Foranalysis, each plot was treated as an independent data point.

3. Analysis and statistics

We used convex hulls to assess the size of the variable ranges for inputs and outcomes. Rather than treating each speciesas a data point, as in studies where only functional traits are considered (e.g. Lin et al., 2011), we treated each plot as a datapoint. This forms amultidimensional space, where the dimensions in the space represent the range of values of each variableincluded in an ‘input’ set or an ‘output’ set. The convex hull is a multidimensional shape that fits the outermost points alongall axes. The relative volumes and centroid positions of convex hulls with the same variable axes can be compared to assessthe position in variable space and amount of variation present across habitat types (Fig. 3). We used the qhull package inR version 2.15.2 to calculate convex hulls. The centroids and the distance between centroids were calculated using codeprovided by one of the authors of Lin et al. (2011), D. Flynn (pers. comm. 2013). Under the null hypothesis that no ecosystemprocesses are exchanged between habitats, we expect outcome volumes to be proportional to input volumes across habitats.We compared null hypothesis expected outcomes to observed outcome volume distributions across habitats using the χ2test in Graphpad (http://graphpad.com/quickcalcs/chisquared2/, used 3 April 2014).

4. Results

4.1. Convex hull volumes and centroids

The convex hull relative volumes and centroid distances are summarized in Fig. 4. Larger convex hulls represent largerranges across all variables. Centroids measure the median point across all variables, so centroid distances indicate thesimilarity, between convex hulls, of the ranges of all variables together. Shorter centroid distances indicate more overlapbetween all variable values.

For forage quality, input variables showed the greatest centroid distance between shrub espinal and espinal. Shrubespinal had the largest range of variable values, with a hull volume that was 6 times larger than espinal and 11 timeslarger than matorral. This general pattern was reflected in the outcomes, although the volume of outcome pools wassignificantly biased towards matorral and away from espinal compared to the expected null distribution from inputs(χ2 = 11763, n = 32164 (sum of output volumes), p < 0.0001). Shrub espinal retained the largest range of variable

http://graphpad.com/quickcalcs/chisquared2/

784 M. Root-Bernstein, F.M. Jaksic / Global Ecology and Conservation 3 (2015) 776–788

Fig. 4. Convex hull volumes and centroid distances for input and outcome variables of forage quality, soil quality and woody plant establishment. Theconvex hull volumes are represented by the circles. Volumes are shown as relative, not absolute. The centroid distances are represented by the lines, withthe distances shown.

values, with a hull volume 27 times larger than espinal and 3.5 times larger than matorral. Matorral and espinal outcomecentroids became more distanced, compared to input centroids.

For soil quality, the matorral centroid was closer to espinal than shrub espinal, and shrub espinal and espinal were morethan twice as distanced as matorral and espinal. The pool of input variables was largest for matorral, which had a convexhull about 2 times larger than the convex hull of shrub espinal and nearly 500% larger than that of espinal. This pattern wasnot reflected in outcome variables (χ2 = 211.5, n = 140 (sum of output volumes), p < 0.0001). The matorral outcomeconvex hull volume was the smallest, with the espinal convex hull 1.3 times larger, and the shrub espinal convex hull 2.5times larger. At the same time, the matorral centroid became relatively much closer to shrub espinal and farther away fromespinal.

For woody plant establishment, the centroids of shrub espinal and espinal were closest together, with the matorralcentroid nearly twice as far away from each. Matorral had the largest pool of input variables, with a convex hull volumeabout 1.4 times larger than the convex hull of shrub espinal, and about 4.7 times larger than the convex hull of espinal.As for soil and forage quality, this pattern was not reflected in outcome variables (χ2 = 90985.6, n = 214756 (sum ofoutput volumes), p < 0.0001). Rather, as for soil and forage quality, shrub espinal had the largest range of variable values,with a convex hull volume 2 times larger than for matorral, and about 96 times larger than for espinal. In addition to beingrelatively extremely small in volume, the espinal convex hull showed a large shift in its centroid away from shrub espinal.

4.2. Ecological variable values

For forage quality in espinal, quality decreased relative to matorral on some measures but increased on others.Herbaceous plant defenses against herbivory were most abundant in espinal. Plants with secondary metabolites were moreabundant in espinal than in shrub espinal, and least abundant in matorral (F = 29.1, df = 2, p < 0.001, ANOVA). Rosettegrowth showed a similar pattern (F = 17.7, df = 2, p < 0.001, ANOVA). Plants with physical defenses (hairy, thornyor stinging) were of equal and low abundance across habitat types (F = 0.8, df = 2, p = 0.437, ANOVA). On the otherhand, the abundance weighted means of herbaceous plant biomass and leaf area were slightly but significantly higherin espinal, and not different between shrub espinal and matorral (biomass: F = 10.1, df = 2, p < 0.001; leaf area:F = 3.7, df = 2, p = 0.026, ANOVA).

For soil, shrub espinal had the highest quality onmostmeasures, compared to espinal andmatorral. Nitrate concentrationwas highest in espinal (F = 18.0, df = 2, p < 0.001, ANOVA). However, ammonium concentration was highest in shrubespinal (F = 35.0, df = 2, p < 0.001, ANOVA), as was phosphorous concentration (F = 107, df = 2, p < 0.001,ANOVA). Percent organic material was also highest in shrub espinal, and not different between espinal and matorral(F = 3.3, df = 2, p = 0.038, ANOVA).

For woody plant establishment in espinals, shrub espinal appeared to have more variables that together favored rapidshrub or tree dispersal, establishment and growth within that habitat compared to espinal and matorral. The abundanceweighted mean of seed size was highest in shrub espinal and lowest in espinal (F = 6.5, df = 2, p = 0.002, ANOVA).Correspondingly, endozoochory dispersed species were more abundant in shrub espinal and least abundant in espinal,with no difference in wind dispersed species between habitats (endozoochory: F = 15.3, p < 0.001, df = 2; wind:F = 1.5, p = 0.227, df = 2, ANOVA). The tree canopy cover did not differ between plots in each habitat, reflecting the

M. Root-Bernstein, F.M. Jaksic / Global Ecology and Conservation 3 (2015) 776–788 785

wide spacing between trees in the espinal (F = 0.7, p = 0.506, df = 2, ANOVA). As expected, shrub canopy was highestin matorral and almost nonexistent in espinal (F = 20.7, p < 0.001, df = 2, ANOVA). Correspondingly, abundance ofshade tolerant woody species was highest in shrub espinal, which had both trees and shrubs (F = 13.7, p < 0.001, df = 2,ANOVA). The abundance weighted mean of the one-year height of saplings or shoots, and the abundance of phanerophytes,were also highest in shrub espinal and lowest in espinal, but there was no difference in root:shoot ratios between habitats(sapling or shoot height: F = 30.7, p < 0.001, df = 2; phanerophytes: F = 16.1, p < 0.001, df = 2; root/shoot ratio:F = 2.4, p = 0.091, df = 2, ANOVA).

5. Discussion

The relative volumes and centroid positions of matorral input and outcome ecological variables were consistent withthe hypothesis that matorral acts as an ecological process subsidizer or source from which espinal habitats gain ecologicalfunction outputs. Shrub espinal was overall the main beneficiary of the three ecosystem functions measured. Shrub espinalwas the main provider of inputs to forage quality, while inputs for soil quality and woody plant establishment functionsoriginated primarily in matorral. The main flow of these ecosystem functions was thus from matorral to shrub espinaland within shrub espinal. This is consistent with our prediction that matorral interacts most closely with shrub espinal.The results also suggest that at this site, succession is primarily moving from matorral to shrub espinal, one of the leastunderstood successional transitions (Fig. 2). Espinal without shrub understory appeared to gain little spillover effect fromespinal with shrubs or matorral. This may be consistent with espinal without shrubs being an ecosystem process sink atleast for some processes. For example, both forage quality and woody plant establishment outcomes may be reduced dueto intensive grazing by livestock. However, we are not aware of any examples of savanna, wood pasture, or open woodlandwith and without shrub understories where habitat with and without shrub understory arise via unconnected degradationtrajectories (e.g. Olff et al., 1999; Barnes & Archer 1999; Peterson & Reich 2001; Pinto-Correia et al. 2010). In light of this,our results strongly suggest that all three habitats are successionally linked.

An alternate explanation is that no ecosystem process movement is taking place. We did not directly measure ecosystemprocess movement, so each habitat may experience outputs caused only by its own inputs. In this scenario, inputs haveoutputs disproportionate to input values across habitats due to other, unmeasured factors that differ between habitats.Similarly, variance in one input might have a disproportionately large effect on outcomes. In this case, the results simply tellus that shrub espinal is the best converter of inputs to outputs, and management should focus on the production of shrubespinal. Under this scenario, as there is no interaction between habitats, there is no succession between them, and so theconditions under which shrub espinal is produced should be studied further.

Functional traits and other variables were broadly distributed in a way favorable to management of espinals forsilvopastoralism. Although the largest set (convex hull volume) of herbaceous forage quality variables was in shrub espinal,biomass and leaf area were highest in espinal, which is good for silvopastoral forage provision. On the other hand, therewere more plants with anti-herbivory adaptations in espinal. Relative increases in the set of soil outcome variables wereassociated with higher values of nutrients in the soil and organic material. Although nitrogen fixation of A. caven may havemade a large contribution to soil fertility outcomes in the espinal habitats, shrub espinal soil variables were more similar to(had a shorter centroid distance to) matorral. Shrub espinal showed the largest set of ecological variables for woody plantestablishment, and also a wide range of dispersal, establishment and growth strategies likely to favor A. caven and nurseplant reproduction. Shrub espinal showed the largest abundance of plants investing in larger seeds dispersed by birds ormammals, as well as vegetatively reproducing plants. Consequently shrub espinal had a higher abundance weighted meanof sapling and shoot growth, a larger canopy, andmore shade-tolerant species. In this case the greater variance in ecologicalvariables corresponds to awider range of plant reproductive strategies, and greater cover of establishedwoody plants.Whilewe do not show that awider range of strategies has led to the observed higher woody plant cover, a greater range of possiblereproductive and growth strategies should help shrub espinal to recover more quickly from a wider range of disturbances(e.g. livestock grazing, fire, wood collection).

Recent work has suggested that at least one kind of successional process driven by natural disturbances is characterizedby mosaic patches with non-overlapping functional traits and soil properties (Dantas et al., 2013). Breakpoints are createdalong a gradient of habitat transition via feedback from a filter such as fire frequency, and distinguish between communities(Dantas et al., 2013). What remains unclear is whether ecosystem processes move across breakpoints. One possibility is thatthey do, but they have no outcomes, generating the lack of functional overlap: rather than allowing exchange of ecosystemprocesses along a successional gradient, the disturbance regime creates a partial or total ecosystem process sink at thebreakpoint (either directionally or mutually). This should have implications for the ecological systems in which functional‘‘insurance’’ can operate. Landscape heterogeneity is expected to contribute to insurance against catastrophic regime shiftsto degraded states (van Nes & Scheffer 2005). However, we predict that only mosaics with low levels of disturbance favorconditions that buffer against disturbance events.

The landscape-scale management recommendation that emerges from these results is to maintain espinal in a mosaicwith and without shrub understory, near patches of matorral. This contrasts with the current focus in Chile on protectingnon-degraded native forests, excludingmatorral and espinal. This research could be extended by examining temporal trendsin anthropogenic uses, as well as climate change and ENSO events, to determine whether matorral also offers espinalecosystem functioning ‘‘insurance’’ andwhethermatorral and espinal together buffer ecosystem functioning in other habitat

786 M. Root-Bernstein, F.M. Jaksic / Global Ecology and Conservation 3 (2015) 776–788

types (Folke, 2006; Loreau et al. 2007; Maestre et al., 2012). Newton et al. (2011) modeled land-use scenarios suggest thatmoderate anthropogenic disturbances maintaining open habitats with A. caven, among other species, increase landscape-level tree diversity. Spatially explicit modeling to assess the potential ranges of ecosystem process outcomes under thesescenarios could assist in landscape-scale land use planning.

Many landscapes are recommended for management as mosaics, due to benefits to livestock grazing, cultural landscapevalues, resilience of ecosystem processes, or habitat needs over the life histories of species (Law and Dickman, 1998;Fuhlendorf and Engle, 2001; Fischer et al., 2006; Brady et al., 2009; Newton et al., 2012; Kumaraswamy and Kunte, 2013).The costs and benefits of prioritizing the conservation of mosaics versus continuous habitats are specific to the habitattypes, species present, and the management context (Bennett et al., 2006; Estrada and Coates-Estrada, 2002a,b; Ghazoul,1996). In particular, it should be important to establish that where natural successional processes take place, ecosystemprocesses are shared between successional stages rather than forming breakpoints (e.g. Dantas et al., 2013). It should also beimportant to understand the direction of ecosystem processes of interest from input to outcome at the site in question. Herewe have argued that espinal and matorral form a successional mosaic exchanging key ecosystem processes, of which shrubespinal is the primary beneficiary at our site. How shrub espinal in particular is successionally linked to sclerophyllous forestestablishment is not clear and requires further research (Fig. 2). However, maintaining large mosaics of espinal–matorralcould safeguard future sclerophyllous forest recovery, ecological processes and landscape resilience (Newton et al., 2011).

Implementing conservation and management non-forest and anthropogenic habitats in contexts where environmentalpolicy is weak presents multiple challenges (e.g. Root-Bernstein et al., 2013a,b; Sodi et al. 2010; Gockowski and Sonwa,2011). Private and public actors need to be convinced why habitats in mutually exclusive cultural and policy categories(e.g. agriculture [espinal] vs. wilderness [matorral, sclerophyllous forest]) should be managed jointly. Inter-agencycommunication about policy could be argued to be necessary where there is inter-habitat exchange of key ecosystemprocesses underlying the flow of ecosystem services to agriculture (Villamagna et al., 2013). Our modified convex hullapproach is a suitable tool for showing the flows and dependencies of ecosystem processes between habitat classes. As weshow in this case study, this approach is sensitive to smaller-scale ecological processes,making it suitable for communicationwith private landowners as well as national planners.

Quantifying ecosystem process interactions between orphan habitats can have management and policy implications.Within the central Chilean context, these results make an argument for considering espinal and matorral as a single landcover class for official conservation planning. Such a class would have the largest land cover of any forest or agriculturalcategory (Schulz 2010; CONAF, 2011). A joint espinal–matorral landcover class could further provide an argument forcreating a new ‘‘semi-natural habitat’’ or ‘‘cultural landscape’’ planning category, whose creation could bring numerouspolicy opportunities (Aronson et al., 1998; Root-Bernstein and Jaksic, 2013), including better protection of ecosystemservices (Durán et al., 2013). Tying the management of espinal to the protection of matorral could also help to broadenthe scope of silvopastoral management in central Chile to include considerations of ecosystem service resilience.

There are many orphan habitats around the world e.g. Caatinga in Brazil, (Paes Marangon et al., 2013); Mediterraneanforests of North Africa, Eastern Guinean forests and Nigerian lowland forests (Burgess et al., 2004); Indonesian beechforests (IUCN, 2008); and bamboo forests (Buckingham et al., 2011). Clearly demonstrating the magnitude of the ecologicalinterlinkages across habitat types may help to raise the profile of many underprotected natural habitats. In addition,harmonizing agricultural and conservation policies is a challenge everywhere. Approaches that help to measure whenand where landscape mosaic management can contribute to agronomical production resilience as well as to biodiversityprotection can help to improve applications of the mosaic concept (Fischer et al., 2006; Kumaraswamy and Kunte, 2013).

Acknowledgments

Many thanks to Sandra Díaz for comments on an early stage of the research, and to Paul Jepson for insights about policyorphans. MR-B is funded by FONDECYT Post Doctoral Fellowship No. 3130336 and Danish National Research FoundationNiels Bohr professorship project Aarhus University Research on the Anthropocene (AURA). FMJ is funded by FB 0002-2014.

References

Alongi, D.M., 2008. Mangrove forests: Resilience protection from tsunamis and responses to global climate change. Estuarine Coastal and Shelf Sci. 78,1–13.

Armesto, J.J., Pickett, S.T.A., 1985. A mechanistic approach to the study of succession in the Chilean matorral. Revista Chilena de Historia Nat. 58, 9–17.Aronson, J., 1992. Evolutionary biology of Acacia caven LeguminosaeMimosidae: Infraspecific variation in fruit and seed characters. Ann.Missouri Botanical

Garden 79 (4), 956–968.Aronson, J., del Pozo, A., Ovalle, C., Avendaño, J., Lavin, A., Etienne, M., 1998. Land use changes and conflicts in central Chile. In: Landscape disturbance and

biodiversity in Mediterranean-type ecosystems. Springer, Berlag, Heidelberg, pp. 155–168.Arroyo, M.T.K., Uslar, P., 1993. Breeding systems in a temperate mediterranean-type climate montane sclerophyllous forest in central Chile. Botanical J.

Linnean Soc. 111, 83–102.Badano, E.I., Cavieres, L.A., Molina-Montenegro, M.A., Quiroz, C.L., 2005. Slope aspect influences plant association patterns in the Mediterranean matorral

of central Chile. J. Arid Environ. 62 (1), 93–108.Baranelli, J.L., Cucucci, A.A., Anton, A.M., 1995. Reproductive biology in Acacia caven Mol Mol Leguminosae in the central region of Argentina. Botanical J.

Linnean Soc. 119, 65–76.Benggtson, J., 1998. Which species? What kind of diversity? Which ecosystem function? Some problems in studies of relations between biodiversity and

ecosystem function. Applied. Soil. Ecology. 10, 191–199.

http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref1http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref2http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref3http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref4http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref5http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref6http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref7http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref8

M. Root-Bernstein, F.M. Jaksic / Global Ecology and Conservation 3 (2015) 776–788 787

Bennett, A.F., Radford, J.Q., Haslem, A., 2006. Properties of land mosaics: Implications for conservation in agricultural environments. Biol. Cons. 13 (3),250–264.

Booth, D.T., Cox, S.E., 2011. Art to science: tools for greater objectivity in resource monitoring. Rangelands 33 (4), 27–34.Brady, M., Kellermann, K., Sahrbacher, C., Jelinek, L., 2009. Impacts of decoupled agricultural support on farm structure biodiversity and landscape mosaic:

Some EU results. J. Agricul. Econ. 60 (3), 563–585.Buckingham, K., Jepson, P., Wu, L., Ramanuja Rao, I.V., Jiang, S., Liese, W., Lou, Y., Fu, M., 2011. The potential of bamboo is constrained by outmoded policy

frames. Ambio 40, 544–548.Burgess, N., et al., 2004. Terrestrial ecoregions of Africa and Madagascar: A conservation assessment. Island Press, London.Bustamante, R.O., Chacón, P., Niemeyer, H.M., 2006. Patterns of chemical defenses in plants: an analysis of the vascular flora of Chile. Chemoecology 00,

1–7.CONAF 2011. Catastro de los recursos vegetacionales nativos de Chile: Monitoreo de cambios y actualizaciones período 1997–2011. Santiago: LOM

Ediciones.Contreras, L.C., Gutiérrez, J.R., 1991. Effects of the subterranean herbivorous rodent Spalacopus cyanus on herbaceous vegetation in arid coastal Chile.

Oecologia 87, 106–109.Contreras, L.C., Gutiérrez, J.R., Valverde, V., Cox, G.W., 1993. Ecological relevance of subterranean herbivorous rodents in semiarid coastal Chile. Revista

Chilena de Historia Nat. 66, 357–368.Coulis, M., Hättenschwiler, S., Fromin, N., David, J.F., 2013. Macroarthropod-microorganism interactions during the decomposition of Mediterranean shrub

litter at different moisture levels. Soil Biol. Biochem. 1–8.Cornelissen, J.H.C., Lavorel, S., Garnier, E., Díaz, S., Buchmann, N., Gurvich, D.E., Reich, P.B., ter Steege, H., Morgan, H.D., van der Heijden, M.G.A., Pausas, J.G.,

Pooter, H., 2003. A handbook of protocols for standardised and easy measurement of plant functional traits worldwide. Aust. J. Bot. 51, 335–380.Crain, C.M., Bertness, M.D., 2006. Ecosystem engineering across environmental gradients: implications for conservation and management. BioScience 56

(3), 211–218.Dantas, V., de, L., Batalha, M.A., Pausas, J.G., 2013. Fire drives functional thresholds on the savanna-forest transition. Ecology 94 (11), 2454–2463.Dean, W.R.J., Milton, S.J., 1991. Disturbances in semi-arid shrubland and arid grassland in the Karoo South Africa: mammal diggings as germination sites.

Afr. J. Ecol. 29, 11–16.de Groot, R.S., Alkemade, R., Braat, L., Hein, L., Willemen, L., 2010. Challenges in integrating the concept of ecosystem services and values in landscape

planning, management and decision making. Ecol. Complexity 7, 260–272.Donoso, Z.C., 1982. Reseña ecologica de los bosques mediterraneos de Chile. Bosque 42, 117–146.Durán, A.P., Casalegno, S., Marquet, P.A., Gaston, K,J., 2013. Representation of ecosystem services by terrestrial protected areas: Chile as a case study. PLoS

one 8 (12), e82643.Eldridge, D.J., Bowker, M.A., Maestre, F.T., Roger, E., Reynolds, J.F., Whitford, W.G., 2011. Impacts of shrub encroachment on ecosystem structure and

functioning: towards a global sythesis. Ecol. Lett. 14 (7), 709–722.Estrada, A., Coates-Estrada, R., 2002a. Dung beetles in a continuous forest forest fragments and in an agricultural mosaic habitat island at Los Tuxtlas,

Mexico. Biodivers. Conserv. 11, 1903–1918.Estrada, A., Coates-Estrada, R., 2002b. Bats in continuous forest forest fragments and in an agricultural mosaic habitat-island at Los Tuxtlas, Mexico. Biol.

Cons. 10 (3), 237–245.Fath, B.D., Halnes, G., 2007. Cyclic pathways in ecological food webs. Ecol. Model. 208, 17–24.Fischer, J., Lindenmayer, D.B., Manning, A.D., 2006. Biodiversity ecosystem function and resilience: Ten guiding principles for commodity production

landscapes. Front. Ecol. Environ. 42, 80–86.Folke, C., 2006. Resilience: the emergence of a perspective for social-ecological systems analyses. Global Environ. Change 16, 253–267.Fuentes, E.R., Avilés, R., Segura, A., 1989. Landscape change under indirect effects of human use: the Savanna of Central Chile. Landscape Ecol. 2, 73–80.Fuentes, E.R., Hoffmann, A.J., Poiani, A., Allende, M.C., 1986. Vegetation change in large clearings: patterns in the Chilean matorral. Oecologia 68, 358–366.Fuentes-Castillo, T., Miranda, A., Rivera-Hutinel, A., Smith-Ramírez, C., Holmgren, M., 2012. Nucleated regeneration of semi-arid sclerophyllous forests

close to remnant vegetation. Forest Ecol. Manag. 27 (4), 38–47.Fuhlendorf, S.D., Engle, D.M., 2001. Restoring heterogeneity on rangelands: Ecosystem management based on evolutionary grazing patterns. BioScience

51 (8), 632–635.Ghazoul, J., 1996. Are fragments worth conserving? Trends Ecol. Evol. 11 (12), 507.Ginocchio, R., Montenegro, G., 1992. Interpretation of metameric architecture in dominant shrubs of the Chilean matorral. Oecologia 90, 451–456.Gockowski, J., Sonwa, D., 2011. Cocoa intensification scenarios and their predicted impact on CO2 emissions biodiversity conservation and rural livelihoods

in the Guinea Rain Forest of West Africa. Environ. Manag. 48, 307–321.Gómez-Aparicio, L., Zamora, R., Gómez, J.M., Hódar, J.A., Castro, J., Bazara, E., 2004. Applying plant facilitation to forest restoration: a meta-analysis of the

use of shrubs as nurse plants. Ecol. Appl. 14 (4), 1128–1138.Gómez-González, S., Sierra-Almeida, A., Cavieres, L.A., 2008. Does plant-derived smoke affect seed germination in dominant species of the Mediterranean

matorral of central Chile? Forest Ecol. Manag. 25 (5), 1510–1515.Hernández, A., Miranda, M., Arellano, E.C., Saura, S., Ovalle, C., 2015. Landscape dynamics and their effect on the functional connectivity of a Mediterranean

landscape in Chile. Ecological Indicators 48, 198–206.Hoffmann, A.J., Teillier, S., Fuentes, E.R., 1989. Fruit and seed characteristics of woody species inmediterranean-type regions of Chile and California. Revista

Chilena de Historia Nat. 62, 43–60.IUCN 2008. Papua Indonesia: Securing sustainable progress for local benefit http://cmsdataiucnorg/downloads/indonesiapdf.Jaksic, F.M., Fuentes, E.R., Yañez, J.L., 1979. Spatial distribution of the OldWorld Rabbit (Oryctolagus cuniculus) in Central Chile. J. Mammal. 60 (1), 207–209.Jiménez, H.E., Armesto, J.J., 1992. Importance of the soil seed bank of disturbed sites in Chilean matorral in early secondary succession. J. Vegetation Sci. 3,

579–586.Jonasson, S., 1983. The point intercept method for non-destructive estimation of biomass. Phytocoenologia 11, 385–388.Kelt, D.A., 2011. Comparative ecology of desert small animals: a selective review of the past 30 years. J. Mammal. 92 (6), 1158–1178.Kelt, D.A., Meserve, P.L., Gutiérrez, J.R., 2004. Seed removal by small mammals birds and ants in semi-arid Chile and comparison with other systems.

J. Biogeogr. 31, 931–942.Kumaraswamy, S., Kunte, K., 2013. Integrating biodiversity and conservation with modern agricultural landscapes. Biodivers. Conserv. 22, 2735–2750.Laliberté, E., Wells, J.A., DeClerck, F., Metcalfe, D.J., Catterall, C.P., Queiroz, C., Aubin, I., Bonser, S.P., Ding, Yi, Fraterrigo, J.M., McNamara, S., Morgan,

J.W., Sánchez Merlos, D., Vesk, P.A., Mayfield, M.M., 2010. Land-use intensificiation reduces functional redundancy and response diversity in plantcommunities. Ecol. Lett. 13, 76–86.

Law, B.S., Dickman, C.R., 1998. Theuse of habitatmosaics by terrestrial vertebrate fauna: implications for conservation andmanagement. Biodivers. Conserv.7, 323–333.

Lin, B.B., Flynn, D.F.B., Bunker, D.E., Uriarte, M., Naeem, S., 2011. The effect of agricultural diversity and crop choice on functional capacity change ingrassland conversions. J. Appl. Ecol. 48, 609–618.

Loreau, M., Naeem, S., Inchausti, P., Bengsston, J., Grime, J.P., Hector, A., Hooper, D.U., Huston, M.A., Raffaelli, D., Schmid, B., Tilman, D., Wardle, D.A., 2001.Biodiversity and ecosystem functioning: Current knowledge and future challenges. Science 294, 804–808.

Loreau, M., Mouquet, N., Gonzalez, A., 2003. Biodiversity as spatial insurance in heterogeneous landscapes. Proc. US Natl. Acad. Sci. 100 (22), 12765–12770.Maestre, F.T., et al., 2012. Plant species richness and ecosystem multifunctionality in global drylands. Science 335 (6065), 214–218.Montenegro, G., Aljaro, M.E., Kummerow, J., 1979. Growth dynamics of Chilean matorral shrubs. Botanical Gazette 1401, 114–119.Mouillot, D., Villéger, S., Scherer-Lorenzen,M., Mason, N.W.H., 2011. Functional structure of biological communities predicts ecosystemmultifunctionality.

PLoS one 63, e17476.

http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref9http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref10http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref11http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref12http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref13http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref14http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref16http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref17http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref18http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref19http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref20http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref21http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref22http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref23http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref24http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref25http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref26http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref27http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref28http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref29http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref30http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref31http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref32http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref33http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref34http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref35http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref36http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref37http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref38http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref39http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref40http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref41http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref42http://cmsdataiucnorg/downloads/indonesiapdfhttp://refhub.elsevier.com/S2351-9894(15)00043-8/sbref44http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref45http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref46http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref47http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref48http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref49http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref50http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref51http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref52http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref53http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref54http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref55http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref56http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref57

788 M. Root-Bernstein, F.M. Jaksic / Global Ecology and Conservation 3 (2015) 776–788

Newton, A., del Castillo, R.F., Echeverría, C., Geneletti, D., González-Espinoza, M., Malizia, L.R., Premoli, A.C., Rey Benayas, J.M., Smith-Ramírez, C., Williams-Linera, G., 2012. Forest landscape restoration in the drylands of Latin America. Ecol. Soc. 17 (1), 21.

Newton, A., Echeverría, C., Cantarello, E., Bolados, G., 2011. Projecting impacts of human disturbances to inform conservation planning and managementin a dryland forest landscape. Biol. Cons. 14 (4), 1949–1960.

Niemeyer, H.M., 1995. Biologically active compounds from Chilean medicinal plants. In: Amason, J.T., et al. (Eds.), Phytochemistry of Medicinal Plants.Plenum Press, New York.

Odum, E.P., 1969. The strategy of ecosystem development. Science 164 (3877), 262–270.Olivares, A.E., 2006a. Relaciones entre el estrato arbóreo el estrato herbáceo y la conducta animal en el matorral de Acacia caven espinal. Sécheresse 17

(1–2), 333–339.Olivares, A.E., 2006b. Pastizales y producción animal en zonas áridas de Chile. Sécheresse 17 (1–2), 257–264.Olff, H., Vera, F.W.M., Bokdam, J., Bakker, E.S., Gleichman, J.M., deMaeyer, K., Smit, R., 1999. Shiftingmosaics in grazed woodlands driven by the alternation

of plant facilitation and competition. Plant Biol. 1, 127–137.Ovalle, C., Aronson, J., del Pozo, A., Avendaño, J., 1990. The espinal: agroforestry systems of the mediterranean-type climate region of Chile. Agroforestry

Systems 10, 213–239.Padilla, F.M., Pugnaire, F.I., 2006. The role of nurse plants in the restoration of degraded environments. Front. Ecol. Environ. 44, 196–202.Paes Marangon, G., Caraciollo Ferreira, R.L., Aleixo da Silva, J.A., D, Fagner de Souza e Lira, Araújo, Silva E., Hambrecht Loureiro, G., 2013. Estrutura e padrão

espacial da vegetação em uma área de caatinga. Floresta 43 (1), 83–92.Provenza, F.D., Papachristou, T.G., 2009. Behavior-based management of ecosystems. Opt. Méditerraneéennes Ser. A 85, 13–28.Rand, T.A., Tylianakis, J.M., Tscharntke, T., 2006. Spillover edge effects: the dispersal of agriculturally subsidized insect natural enemies into adjacent natural

habitats. Ecol. Lett. 9, 603–614.Root-Bernstein, M., Bennett, M., Armesto, J., Ebensperger, L., 2014. Small mammals as indicators of cryptic species diversity in the central Chilean plant

endemicity hotspot. Global Ecol. Conserv. (GECCO) 2, 277–288.Root-Bernstein, M., Fierro, A., Armesto, J., Ebensperger, L.A., 2013a. Avian ecosystem functions are influenced by small mammal ecosystem engineering.

BMC Res. Notes 6, 549.Root-Bernstein, M., Jaksic, F., 2013. The Chilean espinal: Restoration for a sustainable silvopastoral system. Restoration Ecol. 21 (4), 409–414.Root-Bernstein, M., Montecinos Carvajal, Y., Ladle, R., Jepson, P., Jaksic, F., 2013b. Conservation easements and mining: the case of Chile. Earth’s Future 1

(1), 33–38.Scanlon, B.R., Levitt, D.G., Reedy, R.C., Keese, K.E., Scully, M.J., 2005. Ecological controls on water-cycle response to climate variability in deserts. Proc. Nat.

Acad. Sci. 102 (17), 6033–6038.Schulz, J.J., Cayuela, L., Echeverria, C., Salas, J., Rey Benayas, J.M., 2010. Monitoring land cover change of the dryland forest landscape of central Chile

1975–2008. Appl. Geogr. 30, 436–447.Sitters, J., Edwards, P.J., Venterink, H.O., 2013. Increases of soil C N and P pools along an Acacia tree density gradient and their effects on trees and grasses.

Ecosystems 16, 347–357.Torres, C., Eynard, M.C., Aizen, M.A., Galetto, L., 2002. Selective fruit maturation and seedling performance in Acacia caven Fabaceae. Int. J. Plant Sci. 163

(5), 809–813.van de Wouw, P, Echeverría, C., Rey-Benayas, J.M., Holmgren, M., 2011. Persistent Acacia savannas replace Mediterranean sclerophyllous forests in South

America. Forest Ecol. Manag. 26 (2), 1100–1108.Villamagna, A.M., Angermeier, P.L., Bennett, E.M., 2013. Capacity pressure demand and flow: A conceptual framework for analyzing ecosystem service

provision and delivery. Ecol. Complexity 15, 114–121.Vos, V.C.A., van Ruijven, J., Berg, M.P., Peeters, E.T.H.M., Berendse, F., 2011. Macro-detritivore identity drives leaf litter diversity effects. Oikos 120,

1092–1098.

http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref58http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref59http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref60http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref61http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref62http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref63http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref64http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref65http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref66http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref67http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref68http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref69http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref70http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref71http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref72http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref73http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref74http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref75http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref76http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref77http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref78http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref79http://refhub.elsevier.com/S2351-9894(15)00043-8/sbref80

Ecosystem process interactions between central Chilean habitatsIntroductionMaterials and methodsResearch sitePlotsVariables

Analysis and statisticsResultsConvex hull volumes and centroidsEcological variable values

DiscussionAcknowledgmentsReferences