Catena 140 (2016) 195–204

Contents lists available at ScienceDirect

Catena

j ourna l homepage: www.e lsev ie r .com/ locate /catena

Understanding the effect of three decades of land use change on soilquality and biomass productivity in a Mediterranean landscape in Chile

Ángela Hernández a, Eduardo C. Arellano a,b,⁎, David Morales-Moraga a, Marcelo D. Miranda a,b

a Departamento de Ecosistemas y Medio Ambiente, Pontificia Universidad Católica de Chile, Ave. Vicuña Mackenna 4860, Santiago, Chileb Center of Applied Ecology & Sustainability (CAPES), Pontificia Universidad Católica de Chile, Santiago, Chile

⁎ Corresponding author at: Ave. Vicuña Mackenna 486E-mail addresses: angelahernandez@uc.cl (Á. Hernánd

(E.C. Arellano), dlmorale@uc.cl, mmirands@uc.cl (D. Mora

http://dx.doi.org/10.1016/j.catena.2016.01.0290341-8162/© 2016 Elsevier B.V. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 15 October 2014Received in revised form 28 January 2016Accepted 29 January 2016Available online xxxx

Land-use and land-cover change (LUCC) can deeply alter soil quality (SQ), affecting important productivity func-tions like vegetation biomass. A further understanding of the history of LUCC is essential to explain natural or an-thropic landscape cover. The Mediterranean region of central Chile has historically been affected by LUCCimpacts on the vegetation gradient of the landscape, which ranges from natural cover to anthropic cover like graz-ing pastures and crop lands. The main objective of this study is to define the historical impact of LUCC on SQ in aMediterranean landscape in central Chile. To conduct the study, historical LUCC trends between 1975 and 2011were analyzed. A Soil Quality Index (SQI) was developed to comparatively assess six types of land use (AnnualCrops, Perennial Crops, Grazing, Espinal, Dense Espinal andNative Forest); andfinally, SQIwas interpolated at land-scape scale using the soil-adjusted vegetation index (SAVI) as an auxiliary variable. SAVI was selected to representindirect information on vegetation biomass productivity. The results indicate that most LUCC dynamics were ob-served in the Espinal woodland, where agriculture and grazing activities have been developed historically. TheSQI showed significant SQ deterioration when Native Forests (SQI = 0.82) degrade and are transformed by an-thropic interventions likeAnnual Crops (SQI=0.27), Perennial Crops (SQI=0.34) orGrassland (SQI=0.36). How-ever, a slight improvement in SQ compared to the other land useswas identified in the Dense Espinal (SQI=0.46).Finally, the quality model at landscape scale showed clear differences in SQI values for all landscape covers.

© 2016 Elsevier B.V. All rights reserved.

Keywords:Soil quality indexMediterranean agroecosystemsLand use and land cover changeSAVI

1. Introduction

LUCC is a representative consequence of the pressure of human de-velopment on natural landscapes occurring at different spatial and tem-poral scales (Conacher and Sala, 1998; Geist and Lambin, 2002; Lambinet al., 2001). Depending on the type and the intensity of LUCC, ecosys-tems end up reflecting different structures, functions, and dynamics,creating new and complex interactions among the elements vegetation,soil, and nutrients (Adeel et al., 2005). Current landscapes are the resultof natural and anthropogenic processes, thus a historical perspective isrequired to understand the dynamics that led to current conditions(Russell, 1997).

The resulting landscape incorporates physical and biological compo-nents (DeFries et al., 2004; Foley et al., 2005), reflecting changes in soilproperties and consequently in biodiversity and vegetation productivity(Matson et al., 1997; Tscharntke et al., 2005). Several studies havedemonstrated the close relationship between LUCC, soil propertiesand vegetation productivity in different systems (Dörner et al., 2010;

0, CP 7820436 Santiago, Chile.ez), eduardoarellano@uc.clles-Moraga).

Sharma et al., 2011; García-Orenes et al., 2013). In agroforestry systems,most of the land use conversion moves between grassland, new crops,forest substitution, and crop abandonment (Etter et al., 2006; Kastneret al., 2014). In many cases, the consequences of poor agricultural man-agement practices have led to severe degradation that impairs naturalfunctions, mainly regarding soil fertility, and productivity (Kang andJuo, 1986; Nael et al., 2004; Nardi et al., 1996; Zornoza et al., 2007).

The effects of LUCC on soil erosion and vegetation coverage are acommon problem in several agroecosystems in Mediterranean regions(Conacher and Sala, 1998; Geri et al., 2010). The ChileanMediterraneanregion is recognized as a critical biodiversity spot in the southernhemisphere (Myers et al., 2000). It is also themost important agricultur-al zone in the country, where most of the population is located (Schulzet al., 2011). This region has historically been subject to intensive landuse,mainly by overgrazing and overexploitation, resulting in high levelsof erosion and loss in agricultural productivity (Ovalle et al., 1999).Between 1975 and 2008 changes from forest to agricultural land, timberplantations and urban areaswere reported,with annual average growthrates of 1.1%, 2.7% and 3.2%, respectively (Schulz et al., 2010).

To understand historical LUCC trends and effects on ecosystems, it isnecessary to correlate the different land use (Otto et al., 2007;Symeonakis et al., 2007), and their possible impacts on biomass produc-tivity and SQ in agroforestry ecosystems (Andrews et al., 2002b; Glover

Fig. 1. Study location in the townof Catapilco located in theMediterranean regionof central Chile inValparaiso Region. Each dot indicates a soil sampling site in different landuse areas: AC:Annual Crop, PC: Perennial Crop, GR: Grassland, ES: Espinal, DE: Dense Espinal, and NF: Native Forest.

196 Á. Hernández et al. / Catena 140 (2016) 195–204

et al., 2000; Raiesi, 2007). Biomass productivity is an important variableto evaluate ecosystem functionality and health (Whittaker and Likens,1997). It reflects the amount of energy available to be transferredthrough vegetation throughout the ecosystem (Gaston, 2000; Erbet al., 2009; Kay et al., 1999). Land use change and soil degradation re-duce biomass productivity, influencing biodiversity and the generalstate of the ecosystem (Erb et al., 2009).

There is a close relationship between vegetation biomass and soilquality (Paz-Kagan et al., 2014). However, the perception of what consti-tutes good soil varies depending on prioritieswith respect to soil function(Armenise et al., 2013; Zornoza et al., 2008). Doran and Parkin (1994)defined soil quality as the capacity of specific soils to sustain biologicalproductivity and generate the bases for ecosystem health. One of themost commonly used techniques to quantitatively assess SQ is to trans-form and assign weights to soil indicators, and their combination in anindex (Andrews and Carroll, 2001; Bastida et al., 2006; Karlen et al.,1994; Zornoza et al., 2007). Several local studies of Mediterranean sys-tems show that poor soil quality due to poor management has resultedin degraded natural (Bastida et al., 2006; Trasar-Cepeda et al., 1998;Zornoza et al., 2007) or productive systems (Armenise et al., 2013; Imazet al., 2010).

Table 1Average values of the main soil properties in the studied sites.

ACa PC GR

pH 6.9 ± 0.8 6.2 ± 0.4 5.8TC (%) 1.59 ± 0.44 1.22 ± 0.21 5.8TN (%) 0.11 ± 0.03 0.08 ± 0.02 0.08AP (mg/kg) 68.18 ± 47.07 13.95 ± 5.74 11.60OM (%) 2.93 ± 0.45 2.51 ± 0.25 2.58SR (g h−1 m−2) 0.38 ± 0.20 0.31 ± 0.12 0.23BD (g cm−3) 1.20 ± 0.11 1.22 ± 0.12 1.39WSA (%) 0.17 ± 0.21 0.10 ± 0.07 0.13Clay (%) 25.5 ± 4.2 23.3 ± 3.7 21.8Silt (%) 33.8 ± 4.0 35.9 ± 4.1 39.8Sand (%) 40.8 ± 7.2 39.3 ± 6.9 38.5

a AC: Annual Crop, PC: Perennial Crop, GR: Grassland, DE: Dense Espinal, ES: Espinal, NF: Na

In a Mediterranean landscape, the current ecosystem function is aconsequence of historical connections of all the elements that are partof the landscape. These connections are rarely considered in soil qualitymodels that focus on specific soil parameters (Paz-Kagan et al., 2014).Nevertheless, the inclusion of additional ecosystem functions in modelscan be a useful way to improve soil quality interpretation (Herrick,2000). Some studies have attributed the degradation of the vegetationcover in the Mediterranean area of central Chile to LUCC (Balduzziet al., 1982; Fuentes et al., 1989), land use pressures (Ovalle et al.,1996b) and the changing dynamics in land cover (Schulz et al., 2010).However, to improve our understanding of the effects of land usechange, it is necessary to integrate specific SQ models with vegetationproductivity and the current landscape. The objective of this studywas to determine the impact of historical LUCC on SQ for typical Medi-terranean landscapes, using vegetation biomass productivity as a specif-ic function in a vegetation gradient with different levels of humandisturbance over 36 years. The study, which considered a Mediterra-nean landscape in the Valparaiso Region, Chile, was developed inthree steps: 1) definition of historical LUCC trends between 1975 and2011, 2) development and implementation of a SQI, and 3) proposalof a SQ model based on vegetation productivity at a landscape scale.

DE ES NF

± 0.1 5.9 ± 0.1 6.6 ± 0.8 5.7 ± 0.3± 0.1 5.9 ± 0.1 6.6 ± 0.8 5.7 ± 0.3± 0.02 0.10 ± 0.02 0.10 ± 0.06 0.19 ± 0.04± 3.14 22.73 ± 6.95 8.92 ± 2.22 8.75 ± 3.00± 0.38 3.08 ± 0.38 2.73 ± 0.91 5.18 ± 1.00± 0.06 0.43 ± 0.09 0.26 ± 0.05 0.61 ± 0.15± 0.08 1.35 ± 0.13 1.22 ± 0.10 0.95 ± 0.13± 0.07 0.11 ± 0.04 0.15 ± 0.07 0.45 ± 0.10± 2.7 26 ± 6.5 25 ± 5.9 27.5 ± 4.8± 2.5 43.5 ± 13.8 39.9 ± 7.1 39.3 ± 7.4± 3.3 30.5 ± 19.9 35.1 ± 12.7 33.3 ± 11.7

tive Forest.

197Á. Hernández et al. / Catena 140 (2016) 195–204

2. Materials and methods

2.1. Study area

The studywas conducted in theMediterranean area of central Chile,specifically in the town of Catapilco (32° 43′6″ S–71° 16′31″W), Valpa-raíso Region (Fig. 1), which is at altitudes between 80 and 650 m.a.s.l.and covers an area of approximately 10,000 ha. The average annualtemperature is 15.4 °C, the warmest month being January, with a max-imum average temperature of 27.6 °C, and the coldestmonth being July,with a minimum average temperature of 5.4 °C. Average annual precip-itation is 547.8mm, distributedmainly betweenMay andAugust, with aprolonged dry season of six months between October and March. Ac-cording to the national soil survey (CIREN, 2003) and the USDA soil tax-onomy system, the predominant soil orders in the area are Alfisols,followed by Mollisols and Entisols. In general soils in the study areaare highly degraded. Table 1 shows the main characteristics of theassessed soils.

The study area is characterized by the presence of Espinal Acaciacaven, the typical Mediterranean landscape of central Chile. This areahas historically been affected by farming and intensive livestock grazing(Ovalle et al., 1999). Currently, the main land uses are agriculture, witheither Annual Crops (i.e. vegetables and cereals) or Perennial Fruit (i.e.olive trees and vines), cattle and sheep grazing, native forest, scruband less common, forest plantations (i.e. Pinus radiata and Eucalyptusglobulus).

2.2. Land use and land cover change (LUCC)

In order to determine historical trends of the different land coversand land uses, along with establishing the degree of change over thelast 36 years (1975–2011), four Landsat images were used: one MSSimage from 1975 and three TM5 images from 1992, 2001 and 2011.All the images were geometrically, atmospherically and topographicallycorrected. Supervised classification was defined by the statistical deci-sion criterion of maximum likelihood (Chuvieco, 2002), using aerialphotographs from the same years and high-resolution Google Earth im-ages as references. Image classification details and accuracy assessmentmeasurements for these four images can be found in Hernández et al.(2015). Fourmapswere developed from this processwith the followingcover and land use types: Native Forest, Shrubland, Dense Espinal,Espinal, Grassland, Agricultural, Water, Bare Soil and Plantation(Table 2). All the sites were georeferenced with GPS and their locationsincluded on the maps. For each site, the current land use/cover wasregistered based on field observations on 2011 map. The land use

Table 2Description of map classes regarding land use and their relationship to sampling sites.

Map class Sampled cover Description

Native Forest Native Forest Reference soil. AdvancPeumus boldus, Quillajaaltered by human inte

Shrubland Not sampled Intermediate situationsuch as: Acacia caven,second-growth sclerop

Dense Espinal Dense Espinal Thick Acacia caven covit was used (N20 years

Espinal Espinal Cover showing high desporadically used for liagriculture. 26–50% w

Grassland Grassland Herbaceous cover withAgriculture Annual Crop Rainfed and irrigated a

Perennial crop Olive tree cultivation bBarren Land Not sampled Urban areas, rocks, barWater Not sampled Rivers, lakes, water resPlantation Not sampled Ornamental trees and

maps from 1975, 1992 and 2001 were made using the same selectedpoints (Table 3). The maps were prepared using ArcMap 10 software(ESRI, Redlands, Calif.).

2.3. Soil characterization based on land use and land cover

Soil sampling sites were selected based on prior knowledge of thearea and considering landscape variability based on topography, slope,elevation, soil series and texture. In order to evaluate and compare dif-ferent soil quality under a gradient of historical land uses we selectedthe following cover/uses for soil sampling: 1) Annual crops, 2) PerennialCrops, 3) Grassland, 4) Espinal 5) Dense Espinal, and 6) Native Forest(reference soil) (Table 2; Fig 1). Plantation, Shrubland, Water and BareSoil were excluded from the sampling given that: 1) Plantations andShrubland covered a small portion of the study area, and 2) Water andBare Soil were not appropriate for establishing a SQI. Reference soilsare normally required for the construction of soil quality models. NativeForest soil was used because it develops freely and reaches a balance inits properties, resulting in long-term stability in its natural ecosystems(Fedoroff, 1987; Gil-Sotres et al., 2005).

Eight replicates sites for each land use were selected to obtain goodlandscape representation, resulting in a total of 48 soil-sampling sites(Fig. 1). Soil sampling was conducted in June 2012, to a depth of20 cm at 15 random points at each site. Once in the laboratory, the 15samples were composited for each site, air dried, and passed througha 2-mm mesh sieve before soil analysis. The following selected chemi-cal, physical and biological properties, commonly used for soil qualitymodels, were determined: pH was measured in deionized water(1:2.5 soil/water) (Zornoza et al., 2007); total C (TC) and total N (TN)were determined by combustion in a Carlo Erba NA 2500 elementalanalyzer (Robertson et al., 1999); available P (AP) was extracted usingthe calcium-acetate-lactate method (Steubing and Fangmeier, 1992).Organic matter (OM) was determined by oxidation with a mixture ofdichromate and sulfuric acid, and measured colorimetrically (Schulte,1995). Bulk density (BD) was calculated by the core method (Blakeand Hartge, 1986), with four samples per plot. The water-stability ofaggregates (WSA) was quantified using the adapted method describedby John et al. (2005), and was expressed as the percentage of totalsoil in water-stable macro-aggregates N250 μm diameter. Soilrespiration (SR) was the only parameter measured in-situ with fourreplications, using a closed chamber system with an infrared gasanalyzer (SRC-1, PP systems, Hitchin, UK). The measurements werecarried out between 9:00 am and 3:00 pm. The organic horizon wasremoved and chambers were placed on the surface of the first mineralsoil horizon.

ed succession stage of the sclerophyll forest, with species such as: Cryptocarya alba,saponaria, Lithrea caustica, among others. These natural soils have not been recently

rvention (N40 years). 80–98% woody cover.between the matorral and the sclerophyll forest; cover by arborescent speciesMaytenus boaria, Prosopis chilensis, Trevoa trinervis, Colliguaja odorífera andhyll species. 35–75% woody cover.er in arboreal stage, together with second-growth Quillaja saponaria. In the past,) for grazing and rainfed agriculture.51–80% woody cover.nsity of shrubby Acacia caven, Trevoa trinervis and herbaceous cover. It isvestock rotation and in the past (N10 years), it was used for rainfedoody cover.isolated Acacia caven and Trevoa trinervis shrubs. 0–10% woody covergricultureetween 6 and 8 years oldren lake beds, recently cleared lands, roads, highwayservoirsplantations of Pinus radiata and Eucalyptus globulus

Table 3Land cover and land use changes during 1975–1992, 1992–2001 and 2001–2011. Changesobserved in the sample plots; eight plots for each land use. For example, of the eight An-nual Crop plots selected in 2011, four (50%) were Grassland in 1975, two (25%) wereEspinal, one (12.5%) was Shrubland, and one (12.5%) Annual Crop.

1975 1992 2001 2011

GR (50%) AC (100%) AC (100%) AC (100%)ES (25%)AC (12.5%)SH (12.5%)GR (37.5%) GR (50%) GR (50%) PC (100%)ES (25%) SH (12.5%) ES (25%)AC (25%) ES (12.5%) FP (12.5%)SH (12.5%) BL (12.5%) AC (12.5%)

AC (12.5%)GR (75%) GR (50%) SH (50%) DE (100%)AC (25%) ES (37.5%) ES (25%)

AC (12.5%) GR (12.5%)FP (12.5%)

GR (37.5%) GR (37.5%) SH (75%) GR (100%)ES (37.5%) ES (25%) ES (25%)SH (12.5%) AC (25%)BL (12.5%) SH (12.5%)AC (75%) GR (37.5%) GR (87.5%) ES (100%)GR (25%) ES (37.5%) BL (12.5%)

AC (25%)SH (100%) SH (75%) NF (62.5%) NF (100%)

NF (25%) SH (37.5%)

AC: Annual Crop, SH: Shurbland, ES: Espinal, GR: Grassland, PC: Perennial crop, DE: DenseEspinal, NF: Native Forest, BL: Barren Land. FP: Forest Plantation.

198 Á. Hernández et al. / Catena 140 (2016) 195–204

2.4. Development of the Soil Quality Index (SQI)

The first step to evaluate SQ was to determine the ecosystem func-tion of interest (Armenise et al., 2013). The vegetation biomass produc-tivity function (natural and anthropogenic) was selected because it ismeasurable at the two levels of interest (site and landscape).

SQI was developed following the general approach of Andrews et al.(2002a), which consists of three main steps: 1) selection of indicatorsfor a minimum data set (MDS), 2) transformation and weighting ofMDS indicators, and 3) integration of indicator scores into an index.

The MDS met the conditions set by Burger and Kelting (1999);Doran and Parkin (1996) and Etchevers et al. (2009), which are asfollows: 1) is easy to measure, 2) measures changes in soil functions,3) includes physical, chemical and biological soil indicators 4) is acces-sible to evaluators and applicable in field conditions, and 5) is sensitiveto climatic and management variations. Selected MDS indicators weredefined considering 1) expert opinion, 2) consensus of regional

Table 4Rotated factor loadings and communalities for a three-factor model based on MDSindicators.

IndicatorFactor1

Factor2

Factor3

Weightedcommunality

Finalweight

WSA 0.49 0.01 0.87 0.11 0.08TC 0.87 −0.05 0.30 0.25 0.19BD −0.65 −0.11 −0.41 0.15 0.11OM 0.94 −0.11 0.18 0.30 0.23TN 0.97 0.05 0.17 0.32 0.24AP −0.01 0.61 0.09 0.02 0.01pH −0.03 0.99 −0.13 0.05 0.04SR 0.58 −0.02 0.36 0.12 0.09SS Loadingsa 3.58 1.38 1.23Proportionb 0.45 0.17 0.15Weighted proportionc 0.58 0.22 0.20

a Quadratic sum of the loads for each factor.b Proportion of total variance explained by each factor.c The weighted proportion corresponds to the value of the proportion explained by

each factor divided by the sum of all proportions. e.g. for factor 1: 0.45 ∗ (0.45 + 0.17 +0.15)−1 = 0.58.

researchers following Andrews et al. (2002a), and the protocol ofAndrews et al. (2003), and 3) recommendations from SQ studies inMediterranean ecosystems (Armenise et al., 2013; Bastida et al., 2006;Marzaioli et al., 2010; Zornoza et al., 2007). The final set of MDS indica-tors included soil pH, TC, TN, AP, BD, WSA, OM and SR. Pearson correla-tion coefficients were computed to understand the relationships amongthe selected indicators (Fig. 3).

Subsequently, a factorial analysis with rotated axeswas used to esti-mate the weights of the indicators of the MDS, using a correlation ma-trix to avoid the effect of the measurement units on the final weights(Bastida et al., 2006; James and McCulloch, 1990). To reinforce the use-fulness of the factors selected, the parallel test, a method based on gen-erating random variables by simulation, was applied to determine thenumber of selected factors (Glorfeld, 1995). The final weights appliedto the indicators were obtained from theMDSweighted communalitiesnormalized by dividing by their sum (Table 4) (as it is exemplify usingthe WSA indicator1). The weighted communalities correspond to thesum of the square of the product between each factor loading and thecorresponding weighted proportion (Andrews et al., 2002a) (Table 4).

In order to integrate the final weights to form the index, scoringfunctions of the MDS indicators were then generated. Due to the vari-ability of the absolute values of the indicators, the valueswere standard-ized between 0 and 1. The scoring functions were based on themethodology proposed by Andrews et al. (2003). The X axis for thesefunctions represented a specific site range based on the inherent prop-erties of soils or factors of soil formation (Andrews et al., 2003). The Yaxis varied from 0 to 1 and provided the transformed score. It is as-sumed that the expected range for each indicator varies according to aparticular site defined by factors such as weather or the inherent prop-erties of soil (Andrews et al., 2004). Consequently, these valueswere setbased on the soil reference values (Native Forest) and literature onMediterranean systems (Bustamante et al., 1995; Rodríguez et al.,2001). The general forms of the functions are: “more is better” (upperasymptotic sigmoid curve), “less is better” (lower asymptotic sigmoidcurve), and having a “midpoint optimum” (Gaussian function)(Andrews et al., 2002a; Karlen and Stott, 1994). The upper function,more is better, was used for soil OM, TC and WSA based on their rolein soil fertility, water distribution and structural stability (Andrewset al., 2003; Tiessen et al., 1994). It was also used for soil TN to determineappropriate nutritional levels and ecologically reasonable levels of ni-trogen in the soil (Glover et al., 2000), and for SR due to its key role inthe dynamics of soil OM (Marzaioli et al., 2010; Needleman et al.,1999). To normalize the tendency, an equation that defines a sigmoidalcurve (Eq. (1))was used, with an asymptote that tends to 1 and anotherthat tends to 0. The function of less is better (Eq. (1)) was used for BDbecause at high values it has an inhibitory effect on plant root growthdue to its effect on soil porosity (Andrews et al., 2003; Grossman et al.,2001). The optimum or Gaussian function (Eq. (2)) was used for pHbased on the effects on nutrient availability (Andrews et al., 2004;Smith and Doran, 1996). It was also used for soil AP based on crop re-sponse and environmental hazard (Andrews et al., 2003; Maynard,1997).

yij ¼ 1= 1þ xij−aj−bjk� �2� �

ð1Þ

yij ¼ 1= 1þ z j xij−aj� �2� �

ð2Þ

where: yij is the value of each observation i in the j indicator obtainedfrom the membership function; xij is the measured value (field/labora-tory) of each i observation of the j indicator; aj is themedian observation

1 Example of calculation of the final weight (Table 4): WSA = ((0.49 ∗ (0.45 ∗0.77−1))2 + (0.01 ∗ (0.17 ∗ 0.77−1))2 + (0.87 ∗ (0.15 ∗ 0.77−1))2) ∗ 1.3−1 = 0.08Where 0.77 is the sum of the proportions of the total variance explained by the selectedfactors and 1.3 is the sum of weighted communalities of all indicators.

199Á. Hernández et al. / Catena 140 (2016) 195–204

taken for each j indicator; zj is themaximum value among the j indicatorvalue; regarding k = 1 (more is better), bjk corresponds to zj − aj foreach j indicator; regarding k = 2 (lower is better), bjk corresponds tothe difference between aj and theminimum value of the field/laborato-ry measured data for each j indicator.

The normalized scores of the MDS indicators were obtained usingthe scoring functions, and the final weights derived from factor analysis.Theywere then integrated into a standardized weighted-additive indexbetween 0 and 1 (Eq. (3)):

SQI ¼ ∑k

j¼1wj � yj ð3Þ

where SQI is the soil quality index value, k is the number of indicatorsthatmake up theMDS, in this case, k=8;wj is thefinalweight assignedto each indicator, and yj is the score for each j indicator. Tukey's HSDmultiple comparison test (p b 0.05) was applied to identify differencesbetween SQI and different land uses.

Statistical analysis was performed in R software.

2.5. Evaluation of soil quality at the landscape scale

To validate the relationship of the SQI to the function of vegetationbiomass productivity, a remote sensing tool was applied to extrapolatesoil quality and biomass productivity results at the landscape scale. SAVIwas selected (Huete, 1988) to measure vegetation greenness (relativebiomass), which can be linked to either high or low SQ and productivityof vegetation biomass. This index uses the contrast of the characteristicsof two bands in a multispectral raster dataset: chlorophyll pigmentabsorptions in the red band (R) and the high reflection capability ofvegetation in the near-infrared band (NIR).

Fig. 2.Map showing one, two or three chang

SAVI was calculated from the Landsat-5TM images dated September5 2011. This date was chosen because it corresponded to the spring sea-son and vegetation showed fairly good representation. Subsequently,SAVI values were derived from the pixels, where the different landcover/use sampling points were georeferenced and on which SQI wasevaluated.

As SAVI is a soil-adjusted index, there is an L parameter thatattempts to minimize soil brightness using a correction factor. This isparticularly important in semi-arid regions such asMediterranean loca-tions, where vegetation cover is relatively sparse. This index is given byEq. (4) (Huete, 1988):

SAVI ¼ NIR � RNIR þ R þ L

� �� 1þ Lð Þ ð4Þ

where SAVI is the vegetation index value, which varies between−1 and1; L varies depending on vegetation cover (Table 2), assigning a value of0 to areaswith high vegetation cover and 1 to areaswithout vegetation.

To estimate L values, richness, abundance and density of vegetationobtained in the study area were used through vegetation characteriza-tion. L values were established as follows: Bare Soil and Water L = 1,Grassland L = 0.7, Agriculture L = 0.5, Espinal L = 0.3, ShrublandL = 0.2, Dense Espinal L = 0.1, and Plantation and Native Forest L =0. This correction factor was applied to all classes obtained from the2011 land usemap, as the goalwas to have a SQmonitoringmechanismat a landscape scale and not just for land uses evaluated with SQI. Withthis goal in mind, and considering that SAVI is a continuous representa-tion of the landscape vegetation, SQI values were correlated with SAVIvalues by a Pearson correlation coefficient, taking into considerationthat the points overlap in both cases (Fig. 5).

From this analysis, a SQ model was created for the landscape of in-terest (Fig. 7). The spatial data were interpolated using geostatistical

es in land use between 1975 and 2011.

200 Á. Hernández et al. / Catena 140 (2016) 195–204

analysis to estimate SQI values in unsampled locations. We used thecokriging technique from the cross-correlation between the SQI andSAVI. A cross-variogramwas determined to calculate the degree of spa-tial variability among neighboring observations for each variable. Thecross-variogram function was calculated as follows (Eq. (5)):

γ12 hð Þ ¼ 12N hð Þ

XN hð Þi¼1

z1 x1ið Þ−z1 xi1 þ hð Þ½ � z2 x2ið Þ−z2 x2i þ hð Þ½ � ð5Þ

where N (h) is the number of pairs of experimental observations and Z(xi), Z (xi+ h), separated by an h vector, and Z1 and Z2 are spatially cor-related variables. In fitting the theoretical model to experimentalvariograms, the coefficients, nugget effect (C0), sill (C0+C1), structuralvariance (C1) and range (a) were determined. The model tested forfitting was spherical. The model was chosen based on ordinary leastsquares.

Geostatistical analyses, aswell as the interpolations,were performedusing GS+ software for Windows, version 9.1 and the geostatisticsmodule of ArcMap 10 software (ESRI, Redlands, Ca.).

3. Results and discussion

3.1. Historical trends of land cover and land use change

As a result of the historical LUCC trends presented throughout thestudy period (1975–2011), the spatial distribution of different coverswas configured in the landscape and the persistence of changes present-ed on the various landscape covers was identified (Fig. 2). We foundthat only 4% of the area did not suffer any LUCCs, while 19% sufferedone change, 42% two changes, and 35% three changes (Fig. 2).

The most intense changes occurred in the Espinal, with rotation be-tween grazing and rainfed crops, mainly between 1975 and 1992(Table 3). Later (1992–2001) rainfed crops were abandoned, grazingremained in several areas, even to the present day, and in other areascattle raising had discontinued (Table 3). This dynamic has allowedfor the regrowth and increase of Espinal A. caven in some areas, forming

Fig. 3. Pearson correlation coefficients for different soil properties, with a level of significance αstatistical difference from zero.

a Dense Espinal cover in the last period (2001–2011; Table 3). In thesteepest areas, the main changes consisted of converting Native Forestto Shrubland (Table 3). Land use for Annual Crops was more stable,showing just one change throughout the study period (Table 3). Peren-nial Crops are introduced recently, so these soils have been subjected totwo and three changes, mainly from Grassland to Espinal (Table 3).Schulz et al. (2010) studied the entire Mediterranean region of centralChile, and found a similar tendency for changes among grassland,espinal forest, and agriculture land. Changes from degraded or aban-doned land to forest were less common.

3.2. Comparative assessment of SQ among different land uses/covers

Factor analysis showed that three factors explained 77% of the vari-ance of theMDS indicators (Table 4).With the first factor, chemical (TN,TC, OM), biological (SR) and physical (BD) indicators showed thehighest loadings due to their high correlation (Fig. 3). Factor 2 showeda clear separation of chemical indicators (pH and AP), and factor 3showed a clear separation of physical indicators (WSA) (Table 4).Additionally, the final weight assigned to the MDS indicators signifi-cantly pondered the indicators OM, TC and TN (Table 4). Furthermore,the remaining indicators made a low contribution to explaining thetotal variability in the factorial model due to a lower correlation withthe other indicators (Table 4, Fig. 3).

The final weights assigned to each MDS indicator were integratedinto a SQI ranging from 0 to 1. The results from the application of theSQI showed a clear separation of SQ in the different studied land uses.

As expected, the comparison of SQ among the different land usesshowed that the SQI in Native Forest soils presented the best SQ(SQI = 0.82), which is significantly different (p b 0.05, Tukey HSD)from those of the other uses (Fig. 4). This wasmainly due to the high in-dividual contribution of WSA, TC, OM, TN and pH (Fig. 4). Islam andWeil (2000) tested a degradation index in which farming soils showedmuch higher degrees of degradation than soils with natural covers.Similarly, Marzaioli et al. (2010) applied a SQI to different land uses inMediterranean Italy, and found that coniferous and broadleaf forests

= 0.05. Histograms and trends for each indicator are shown. The asterisk (*) indicates a

Fig. 4. SQI comparison among different land uses. The different shades of gray show thecontribution of each MDS indicator to SQI. Error bars indicate the standard deviation ofall values in the index. Significant differences among land uses are identified withdifferent letters after a Tukey HSD test (α = 0.05). AC: Annual Crop, PC: Perennial Crop,GR: Grassland, DE: Dense Espinal, ES: Espinal, and NF: Native Forest.

Fig. 5. Linear associations between SQI and SAVI for the different sampling points. Thecoefficient of determination (R2) regarding the relationship among the indicators isshown.

201Á. Hernández et al. / Catena 140 (2016) 195–204

have a superior quality than scrubland and grassland, the latter twohaving intermediate values, and than perennial crops, which presentedthe lowest SQ values. This can be attributed to the fact that Native Forestsoils had the least human intervention during the study period(Table 3).

Dense Espinal soils showed an intermediate SQ value (SQI = 0.46)relative to the other land uses (Fig. 4). This may be due to the LUCChistory of these soils (Table 3), which were used extensively between1975 and 1992, with rotation between grazing and rainfed crops, beforethe land was abandoned. Hernández et al. (2015) suggested that landabandonment resulted in increased shrubland and tree density, mainlyA. caven, along with an increase in herbaceous vegetation and recoloni-zation of some typical native forest species (e.g.Quillaja saponaria). Thisdynamic probably produced a gradual improvement in soil properties,suggesting a SQ recovery. Applying an SQI, Marzaioli et al. (2010)found that SQ recovered to a level comparable to the quality of undis-turbed soils in scrublands with the presence of a herbaceous layer andother wild plants. Using two distinct SQIs, Zornoza et al. (2008) foundthat abandoned agricultural soils in the Mediterranean area showedless variability in relation to soils in less disturbed forest than soilsunder almondproduction. Sánchez-Marañón et al. (2002) also observeda slow recovery of soil physical and chemical properties in abandonedcereal croplands in southern Spain. Similarly, studies in the Mediterra-nean basin have found that the regeneration of natural vegetation,often after land abandonment, decreases soil erosion (Grove andRackham, 2001). A similar situation may be occurring in the DenseEspinal soils, as vegetation cover is an important requirement to im-prove OM content, soil structure and the infiltration rate, protectingsoils from erosion (Kosmas et al., 2000; Trimble, 1990), and thereforeimproving SQ. Similarly, Caravaca et al. (2003) suggested that shrubsand perennial grasses improved SQ in anabandoned SpanishMediterra-nean agro-ecosystem by increasing soil OMand soil nitrogen content, aswell as favoring the formation of stable aggregates and developingpropagules. Contrarily, Pascual et al. (2000) found that the abandon-ment of agricultural soils in semi-arid Mediterranean Spain producedsevere soil degradation, with a decrease in soil OM, soil microbialbiomass, basal respiration and soil enzyme activities.

Perennial Crop and Grassland soils showed similar SQs, althoughweaker than Dense Espinal soils, with SQI values of 0.34 and 0.36,respectively (Fig. 4), which is explained by constant agroforestry

rotation of these sites, such as Espinal, grazing pastures and rainfedcrops (Table 3). Raiesi (2007) reported that LUCC (conversion ofpastures to almond orchards and then to alfalfa fields) negativelyinfluenced properties like bulk density, porosity and aggregate stability,decreasing SQ. Additionally, Perennial Crops, apart from olive trees, donot have cover that protects the soil (e.g. grasses or herbaceous plants),which probably decreases the SQ of these soils (Fig 4).

Finally, the SQI value of Annual Crop soils was not significantlydifferent from that of Espinal soils, presenting the lowest SQ values,0.27 and 0.29, respectively, because they have similar individual contri-butions from the different indicators (Fig. 4). The poor SQ of AnnualCrops (Fig. 4) is explained by the permanent use (N 20 years) of thesesoils for agriculture (Table 3). Applying an SQI, Zornoza et al. (2008)found in undisturbed, agricultural and abandoned agricultural soils,that agricultural practices disrupt the natural balance of the soil. In addi-tion, continuous tillage and fertilization lead to imbalance among soilindicators, accompanied by decreased OM, which has been reported asthe main contributor of cultivated soil degradation (Davidson andAckerman, 1993; Moscatelli et al., 2007; Nardi et al., 1996).

These SQI results show that SQ deteriorates significantly whennative forest systems are degraded and transformed to uses such asAnnual Crops, Perennial Crops or Grassland. It is also clear that SQ canimprove when anthropogenic uses like cultivation or grazing areabandoned re-vegetation subsequently occurs (Fig. 4).

3.3. SQ Monitoring at the landscape scale

To monitor SQ at a landscape scale, a model was generated from theinterpolation of SQI with SAVI.

Studies have shown the efficiency of SAVI tomeasure the productiv-ity of different vegetation covers (Farzadmehr et al., 2004; Hobbs,1995). This efficiency was related to SQI (Fig. 5), demonstrating thatthese two indices are suitable for interpolating SQ at landscape scale(Figs. 6, 7). From this relation we adjusted a cross-variogram betweenSQI and SAVI (Fig 6) with a high degree of spatial dependence (R2 =0.86) for the observed and the estimated semivariance (nugget =0.018, sill = 0.029, structural variance = 0.011, range = 772.81). Themodel showed that the spatial continuity of the interpolated variable(SQI) was enhanced by the inclusion of the continuous spatial variable

Fig. 6. Cross-variogram of SQI and SAVI.

202 Á. Hernández et al. / Catena 140 (2016) 195–204

(SAVI). The spatial distribution of sample points is not homogeneousthroughout the study area (Fig. 1). Consequently, the application ofkriging prediction without the use of an auxiliary variable presented aweak value (R2 = 0.56, exponential model) for the SQI estimation.Previous studies have demonstrated that when the auxiliary variablehas greater continuity than the main variable, the resulting cokrigingmap tends to have this characteristic, contributing to a higher represen-tation of the spatial variability (Silva et al., 2010; Silva and Lima, 2014).

The landscape model allowed us to identify the spatial distributionof the differences in SQ throughout the study area and helped identifya SQ gradient among different vegetation covers (Fig. 7). The highcorrelation between SAVI and SQI (Figs. 5, 6) helped to generate a

Fig. 7. SQ Model at the landscape scale as a result of interpolated

more accurate spatial extrapolation among the measurement pointsperformed on land (Cyr et al., 1995; Silva et al., 2010).

This made it possible to determine that the main SQ fluctuationsoccurred in the Espinal valleys (light blue to yellow in Fig. 7). Thisarea has the largest Espinal distribution at different densities (i.e.Espinal, Dense Espinal and Grassland). The variation in SQ is the resultof the historical land use that these Espinal have undergone (Fig. 2,Table 3). This trend is reflected in the whole Mediterranean area ofcentral Chile, where the intense use of land by overgrazing andover-harvesting has dramatically depleted soil nutrients and damagedprimary and secondary productivity, as well as biodiversity (Ovalleet al., 1999). This is because Espinals are compatible with most of theChileanMediterranean livestock industry and rainfed cereal productionareas (Ovalle et al., 1990; Ovalle et al., 1996a), which are normallypresent in small locations (2 to 20 ha) and within a heterogeneouslandscape dominated by different levels of coverage of A. caven(Muñoz et al., 2007). Espinal ecosystems are more degraded on slopeswhere pasture-crop rotation has been intense, (blue tones in Fig. 7)and with less A. caven coverage. Therefore, more erosion and soilnutrient depletion are observed (Muñoz et al., 2007). In contrast, onplains where land use has been less intense and where rainfed cropshave been discontinued, A. caven cover ismore intense and some typicalspecies of native forest have established themselves (e.g. Q. saponaria),allowing Dense Espinal development in the last 10 years (Table 3). Thisdynamic resulted in an SQ improvement in the landscapemodel (in yel-low in Fig. 7). Ovalle et al. (1996a) found that discontinuation of grazingand crop production resulted in further development of A. caven cover-age, resulting in better soil conservation. This evidence demonstratesthat SQ can gradually improve even after the soil has been subjected

SQI using SAVI as a covariate in cokriging for the study area.

203Á. Hernández et al. / Catena 140 (2016) 195–204

to intensive use for several years. Ovalle and Godron (1989) stated thatA. caven in espinal cover seems to favor patch growth as a consequenceof more favorable conditions under tree canopies, where productiveherbaceous species are concentrated. This improvement in vegetationalso boosts soil properties, leading to improved SQ. Finally, better SQwas observed in the higher elevations of the landscape, wheremost Na-tive Forest is distributed (brown in Fig. 7), which is because there is lesshuman intervention and better plant succession (scrubland to forest)during the period under study (Table 3).

SQ is essential to maintain good vegetation cover and vice versa.These elements must be carefully considered in the landscape mosaicwhen planning to maintain and regulate habitat and landscapefunctions.

4. Conclusions

The results of this study demonstrate the efficiency of SQI inassessing SQ for different land uses. In addition, they provided impor-tant evidence about the effect of the LUCC dynamic on SQ. The SQIresults showed that there was a significant SQ deterioration whenNative Forest systems degrade and are transformed for anthropic useslike Annual Crops, Perennial Crops or Grassland.

In addition, we found that SQ gradually recovers in areas that wereused for grassland and crops and then abandoned (e.g. Dense Espinal).Based on this, we can say that: 1) except for Native Forest, the evaluatedcovers decrease the capacity for vegetation productivity, and 2) soil ca-pacity for vegetation productivity improves when land is abandonedand subsequent revegetation takes place.

The results also demonstrate the advantages of using spectral dataalong with modeling field data to identify and evaluate SQ in differentland uses at the landscape scale, by identifying a SQ gradient of differentlandscape covers. The application of spectral data is a good tool toimprove the interpolation of SQI. SAVI was a good auxiliary variable toimprove the spatial interpretation of SQ at the landscape scale present-ed in this case. However, because the landscape assessment was notconducted on an extensive area, further studies are necessary to deter-mine the sensitivity of the proposed models to other areas.

The landscape model can be a valuable tool for decision-making,helping to detect potential areas to be protected and helping todetermine which land uses are most suitable for a given area. Theparameters selected in this study to demonstrate the efficiency of SQIcan be analyzed andmodified according to the function to be evaluated.

In general, by understanding SQ at a site level we can associate basicecosystem functions like total vegetation biomass. However, maintain-ing a good soil quality based on total vegetation biomasswithout recog-nizing the reality of managing land for other uses is too simplistic.

Acknowledgments

This study was supported by MECESUP-UC0707, and Minera AngloAmerica Chile. The authors thank the of Center of Applied Ecology &Sustainability (CONICYT FB 0002/2014). AH thanks the community ofCatapilco for providing valuable support to this study.

Appendix A. Supplementary data

Supplementary data associated with this article can be found in theonline version, at http://dx.doi.org/10.1016/j.catena.2016.01.029.These data include the Google map of the most important areasdescribed in this article.

References

Adeel, Z., Safriel, U., Niemeijer, D., White, R., 2005. Millennium Ecosystem Assessment,Ecosystems and HumanWell-being: Desertification Synthesis. World Resources Insti-tute, Washington, DC.

Andrews, S.S., Carroll, C.R., 2001. Designing a decision tool for sustainable agroecosystemmanagement: soil quality assessment of a poultry litter management case study. Ecol.Appl. 11, 1573–1585.

Andrews, S.S., Karlen, D.L., Mitchell, J.P., 2002a. A comparison of soil quality indexingmethods for vegetable production systems in Northern California. Agric. Ecosyst. En-viron. 90, 25–45.

Andrews, S.S., Mitchell, J.P., Mancinelli, R., Karlen, D.L., Hartz, T.K., Horwath, W.R.,Pettygrove, G.S., Scow, K.M., Munk, D.S., 2002b. On-farm assessment of soil qualityin California's Central Valley. Agron. J. 94, 12–23.

Andrews, S.S., Flora, C.B., Mitchell, J.P., Karlen, D.L., 2003. Grower's perceptions and accep-tance of soil quality indices. Geoderma 114, 187–213.

Andrews, S.S., Karlen, D.L., Cambardella, C.A., 2004. The soil management assessmentframework: a quantitative soil quality evaluation method. Sci. Soc. Am. J. 68,1945–1962.

Armenise, E., Redmile-Gordon, M.A., Stellacci, A.M., Ciccarese, A., Rubino, P., 2013.Developing a soil quality index to compare soil fitness for agricultural use under dif-ferent managements in the Mediterranean environment. Soil Tillage Res. 130, 91–98.

Balduzzi, A., Tomaselli, R., Serey, I., Villaseñor, R., 1982. Degradation of the Mediterraneantype of vegetation in Central Chile. Ecol. Mediterranea 7, 223–240.

Bastida, F., Moreno, J.L., Hernández, T., García, C., 2006. Microbiological degradation indexof soil in a semiarid climate. Soil Biol. Biochem. 38, 3463–3473.

Blake, G.R., Hartge, K.H., 1986. Bulk Density. In: Klute, A., Campbell, G.S., Jackson, R.D.,Mortland, M.M., Nielsen, D.R. (Eds.), Methods of Soil Analysis. Part I. ASA and SSSA,Madison, WI, USA, pp. 363–375.

Burger, J.A., Kelting, D.L., 1999. Using soil quality indicators to assess forest standmanage-ment. For. Ecol. Manag. 122, 155–166.

Bustamante, J.A., Mery, M.L., Sanhueza, E., Zárate, E., Pinto, H., 1995. Requerimientos declima y suelo, chacras y hortalizas. CIREN. (Santiago, Chile).

Caravaca, F., Figueroa, D., Barea, J.M., Azcón-Aguilar, C., Palenzuela, J., Roldán, A., 2003. Therole of relict vegetation inmaintaining physical, chemical, and biological properties inan abandoned Stipa-grass agroecosystem. Arid Land Res. Manag. 17, 103–111.

Chuvieco, E., 2002. Teledetección ambiental: la observación de la tierra desde el espacio.Ariel Ciencia. Catena, Barcelona, España.

CIREN, 2003. Descripciones de suelos materiales y siembras. Estudio Agrologico, V Región(Tomo 1, Publicación No. 116).

Conacher, A.J., Sala, M., 1998. Land Degradation in Mediterranean Environments of theWorld: Nature and Extent, Causes and Solutions. JohnWiley and Sons Ltd., Chichester.

Cyr, L., Bonn, F., Pesant, A., 1995. Vegetation indices derived from remote sensing for anestimation of soil protection against water erosion. Ecol. Model. 79, 277–285.

Davidson, E.A., Ackerman, I.L., 1993. Changes in soil carbon inventories following cultiva-tion of previously untilled soils. Biogeochemistry 20, 161–193.

DeFries, R., Asner, G., Houghoton, R., 2004. Ecosystem and Land Use Change. AmericanGeophysical Union, Geophysical Monograph Series, Washington, DC.

Doran, J.W., Parkin, T.B., 1994. Defining and assessing soil quality. In: Doran, J.W.,Coleman, D.C., Bezdicek, D.F., Stewart, B.A. (Eds.), Defining Soil Quality for a Sustain-able Environment. SSSA, Inc., Madison, Wisconsin, USA, pp. 3–21.

Doran, J.W., Parkin, T.B., 1996. Quantitative indicators of soil quality: a minimum data set.In: Doran, J.W., Jones, A.J. (Eds.), Methods for Assessing Soil Quality. Special Publica-tion 49. Soil Science Society of America, Madison, WI, pp. 25–37.

Dörner, J., Dec, D., Peng, X., Horn, R., 2010. Effect of land use change on the dynamic be-haviour of structural properties of an Andisol in southern Chile under saturatedand unsaturated hydraulic conditions. Geoderma 159, 189–197.

Erb, K.H., Krausmann, F., Gaube, V., Gingrich, S., Bondeau, A., Fischer-Kowalski, M., Haberl,H., 2009. Analyzing the global human appropriation of net primary production pro-cesses, trajectories, implications. An introduction. Ecol. Econ. 69, 250–259.

Etchevers, J., Hidalgo, C., Vergara, M., Bautista, M., Padilla, J., 2009. Calidad de suelo:conceptos, indicadores y aplicación en agricultura. En: López-Blanco, J., Rodríguez-Gamiño, M. (Eds.), Desarrollo de indicadores ambientales y de sustentabilidad enMéxico. Universidad Nacional Autónoma de México, pp 196.

Etter, A., McAlpine, C., Wilson, K., Phinn, S., Possingham, H., 2006. Regional patterns ofagricultural land use and deforestation in Colombia. Agric. Ecosyst. Environ. 114,369–386.

Farzadmehr, J., Arzani, H., Darvish Sefat, A.A., Jafari, M., 2004. Investigation in estimatingvegetation cover and phytomass production, using enhanced landsat data in a semiarid region. Iran. J. Nat. Resour. 57 (2), 339–351.

Fedoroff, N., 1987. The production potential of soils. Part 1. Sensitivity of principal soil typesto the intense agriculture of north-western Europe. In: Barth, E., L'Hermite, P. (Eds.), Sci-entific Basis for Soil Protection in the European Community. Elsevier, London, pp. 65–86.

Foley, J.A., DeFries, R., Asner, G.P., Barford, C., Bonan, G., Stephen, R.C.F., Chapin, S., Coe,M.T., Daily, G.C., Gibbs, H.K., Helkowski, J.H., Holloway, T., Howard, E.A., Kucharik,C.J., Monfreda, C., Patz, J.A., Prentice, I.C., Ramankutty, N., Snyderl, P.K., 2005. Globalconsequences of land use. Science 309, 570–574.

Fuentes, E.R., Avilés, R., Segura, A., 1989. Landscape change under indirect effects ofhuman use: the savanna of Central Chile. Landsc. Ecol. 2 (2), 73–80.

García-Orenes, F., Morugán-Coronado, A., Zornoza, R., Scow, K., 2013. Changes in soil mi-crobial community structure influenced by agricultural management practices in aMediterranean agro-ecosystem. PLoS One 8 (11).

Gaston, K.J., 2000. Global patterns in biodiversity. Nature 405, 220–227.Geist, H.J., Lambin, E.F., 2002. Proximate causes and underlying driving forces of tropical

deforestation. Bioscience 52, 143–150.Geri, F., Amici, V., Rocchini, D., 2010. Human activity impact on the heterogeneity of a

Mediterranean landscape. Appl. Geogr. 30, 370–379.Gil-Sotres, F., Trasar-Cepeda, C., Leirós, M.C., Seoane, S., 2005. Different approaches to

evaluating soil quality using biochemical properties. Soil Biol. Biochem. 37, 877–887.Glorfeld, L.W., 1995. An improvement on Horn's parallel analysis methodology for

sleceting the correct number of factors to retain. Educ. Psychol. Meas. 55, 377–393.

204 Á. Hernández et al. / Catena 140 (2016) 195–204

Glover, J.D., Reganold, J.P., Andrews, P.K., 2000. Systematic method for rating soil qualityof conventional, organic, and integrated Apple orchards in Washington State. Agric.Ecosyst. Environ. 80, 29–45.

Grossman, R.B., Harms, D.S., Seybold, C.A., Herrick, J.E., 2001. Coupling use-dependent anduse-invariant data for soil quality evaluation in the United States. J. Soil WaterConserv. 56, 63–68.

Grove, A.O., Rackham, O., 2001. The Nature of Mediterranean Europe: an Ecological Histo-ry, An Ecological History. Yale University Press.

Hernández, A., Miranda, M., Arellano, E.C., Saura, S., Ovalle, C., 2015. Landscape dynamicsand their effect on the functional connectivity of a Mediterranean landscape in Chile.Ecol. Indic. 48, 198–206.

Herrick, J.E., 2000. Soil quality: an indicator of sustainable land management? Appl. SoilEcol. 15, 75–83.

Hobbs, T., 1995. The use of NOAA-AVHRR NDVI data to assess herbage production in thearid rangelands of Central Australia. Int. J. Remote Sens. 16 (7), 1289–1302.

Huete, A.R., 1988. A soil adjusted vegetation index (SAVI). Remote Sens. Environ. 25, 295–309.Imaz, M.J., Virto, I., Bescansa, P., Enrique, A., Fernandez-Ugalde, O., Karlen, D.L., 2010. Soil

quality indicator response to tillage and residue management on semi-arid Mediter-ranean cropland. Soil Tillage Res. 107, 17–25.

Islam, K.R., Weil, R.R., 2000. Land use effects on soil quality in a tropical forest ecosystemof Bangladesh. Agric. Ecosyst. Environ. 79, 9–16.

James, F.C., McCulloch, C.E., 1990. Multivariate analysis in ecology and systematics: pana-cea or pandora's box? Annu. Rev. Ecol. Syst. 21, 129–166.

John, B., Yamashita, T., Ludwig, B., Flessa, H., 2005. Storage of organic carbon in aggregate anddensity fractions of silty soils under different types of land use. Geoderma 128, 63–79.

Kang, B.T., Juo, A.S.R., 1986. Effect of forest clearing on soil chemical properties and cropperformance. In: Lal, R., Sanchez, P.A., Cummings Jr., R.W. (Eds.), Land Clearing andDevelopment in the Tropics. Belkema, Rotterdam, pp. 383–394.

Karlen, D.L., Stott, D.E., 1994. A framework for evaluating physical and chemical indicatorsof soil quality. In: Doran, J.W., Coleman, D.C., Bezdicek, D.F., Stewart, B.A. (Eds.),Defining Soil Quality for a Sustainable Environment. SSSA Special Publ. 34. SoilScience Society of America, Madison, Wisconsin, pp. 53–72

Karlen, D.L., Wollenhaupt, N.C., Erbach, D.C., Berry, E.C., Swan, J.B., Eash, N.S., Jordahl, J.L.,1994. Long-term tillage effects on soil quality. Soil Tillage Res. 32, 313–327.

Kastner, T., Karl-Heinz, E., Helmut, H., 2014. Rapid growth in agricultural trade: effects onglobal area efficiency and the role of management. Environ. Res. Lett. 9 (3), 1–10.

Kay, J.J., Regier, H.A., Boyle, M., Francis, G., 1999. An ecosystem approach forsus-tainability: addressing the challenge of complexity. Futures 31, 721–742.

Kosmas, C., Danalatos, N.G., Gerontidis, S.T., 2000. The effect of landparameters onvegetationperformance and degree of erosion under Mediterranean conditions. Catena 40, 3–17.

Lambin, E.F., Turner, B.L., Geist, H.J., Agbola, S.B., Angelsen, A., Bruce, J.W., Coomes, O.T., Dirzo,R., Fischer, G., Folke, C., George, P.S., Homewood, K., Imbernon, J., Leemans, R., Li, X.,Moran, E.F., Mortimore, M., Ramakrishnan, P.S., Richards, J.F., Skanes, H., Steffen, W.,Stone, G.D., Svedin, U., Veldkamp, T.A., Vogel, C., Xu, J., 2001. The causes of land-useand land-cover change: moving beyond myths. Glob. Environ. Chang. 11, 261–269.

Marzaioli, R., Ascoli, R.D., De Pascale, R.A., Rutigliano, F.A., 2010. Soil quality in a Mediter-ranean area of Southern Italy as related to different land use types. Appl. Soil Ecol. 44,205–212.

Matson, P.A., Parton, W.J., Power, A.G., Swift, M.J., 1997. Agricultural intensificationandecosystem properties. Science 277, 504–509.

Maynard, D.N., 1997. Knott's Handbook for Vegetable Production. fourth ed. JohnWiley &Sons, New York.

Moscatelli, M.C., Di Tizio, A., Marinari, S., Grego, S., 2007. Microbial indicators related tosoil carbon in Mediterranean land use systems. Soil Tillage Res. 97, 51–59.

Muñoz, C., Zagal, E., Ovalle, C., 2007. Influence of trees on soil organic matter in Mediter-ranean agroforestry systems: an example from the ‘Espinal’ of central Chile. Eur. J. SoilSci. 58, 728–735.

Myers, N., Mittermeier, R.A., Mittermeier, C.G., Da Fonseca, G.A.B., Kent, J., 2000. Biodiver-sity hotspots for conservation priorities. Nature 403, 853–858.

Nael, M., Khademi, H., Hajabbasi, M.A., 2004. Response of soil quality indicators and theirspatial variability to land degradation in central Iran. Appl. Soil Ecol. 27, 221–232.

Nardi, S., Cocheri, G., Dell'Agnola, G., 1996. Biological activity of humus. In: Piccolo, A.(Ed.), Humic Substances in Terrestrial Ecosystems. Elsevier, Amsterdam, pp. 361–406.

Needleman, B.A., Wander, M.M., Bollero, G.A., Boast, C.W., Sims, G.K., Bullock, D.G., 1999.Interaction of tillage and soil texture: biologically active soil organic matter in Illinois.Soil Sci. Soc. Am. J. 63, 1326–1334.

Otto, R., Krusi, B.O., Kienast, F., 2007. Degradation of an arid coastal landscape in relation toland use changes in southern Tenerife (Canary Islands). J. Arid Environ. 70, 527–539.

Ovalle, C., Godron, M., 1989. Influencia del árbol sobre la vegetación herbácea enmatorrales de Acacia caven. Rol del animal (ovinos) en las interacciones árbol-pasto. Stud. Oecol. 6, 225–243.

Ovalle, C., Aronson, J., Del Pozo, A., Avendaño, J., 1990. The Espinal: agroforestry systemsof the Mediterranean-type climate región of Chile. Agrofor. Syst. 10, 213–239.

Ovalle, C., Avendaño, J., Aronson, J., Del Pozo, A., 1996a. Land occupation patterns and veg-etation structure in the anthropogenic savannas (espinales) of Central Chile. For. Ecol.Manag. 86, 129–139.

Ovalle, C., Longeri, L., Aronson, J., Herrera, A., Avendaño, J., 1996b. N2-fixation, nodule ef-ficiency and biomass accumulation after two years in three Chilean legume trees andTagasaste (Chamaecytisus proliferus subsp. palmensis). Plant Soil 179, 131–140.

Ovalle, C., Aronson, J., Del Pozo, A., Avendaño, J., 1999. Restoration and rehabilitation ofmixed espinales of central Chile. Arid Soil Res. Rehabil. 13, 369–381.

Pascual, J.A., García, C., Hernández, T., Moreno, J.L., Ros, M., 2000. Soil microbial activity as abiomarker of degradation and remediation processes. Soil Biol. Biochem. 32, 1877–1883.

Paz-Kagan, T., Shachak, M., Zaady, E., Karnieli, A., 2014. Evaluation of ecosystem responsesto land-use change using soilquality and primary productivity in a semi-arid area,Israel. Agric. Ecosyst. Environ. 193, 9–24.

Raiesi, F., 2007. The conversion of overgrazed pastures to almond orchards and alfalfacropping systems may favor microbial indicators of soil quality in Central Iran.Agric. Ecosyst. Environ. 121, 309–318.

Robertson, G.P., Coleman, D.C., Bledsoe, C., Sollins, P., 1999. Standard Soil Methods forLong-Term Ecological Research. Oxford University Press, USA.

Rodríguez, J., Pinochet, D.E., Matus, F.J., 2001. La fertilización de los cultivos: un métodoracional. Impr. LOM (Chile).

Russell, E.W.B., 1997. People and the Land through Time: Linking Ecology and History.Yale University Press, New Haven, USA.

Sánchez-Marañón, M., Soriano, M., Delgado, G., Delgado, R., 2002. Soil quality in Mediter-ranean environments: effects of land use change. Soil Sci. Soc. Am. J. 66, 948–958.

Schulte, E.E., 1995. Recommended soil organic matter test (Chapter 8) In: Horton, M.L.(Ed.), Recommended Soil Testing Procedures for the Northeastern United States, sec-ond ed. Northeastern Regional Publication No 493.

Schulz, J.J., Cayuela, L., Echeverría, C., Salas, J., Rey-Benayas, J.M., 2010. Monitoring landcover change of the dryland forest landscape of Central Chile (1975–2008). Appl.Geogr. 20, 436–447.

Schulz, J., Cayuela, L., Rey-Benayas, J.M., Schröder, B., 2011. Factors influencing vegetationcover change in Mediterranean Central Chile (1975–2008). Appl. Veg. Sci. 14, 571–582.

Sharma, A., Tiwari, K.N., Bhadoria, P.B.S., 2011. Effect of land use land cover change on soilerosion potential in an agricultural watershed. Environ. Monit. Assess. 173, 789–801.

Silva, S.A., Lima, J.S.S., 2014. Spatial estimation of foliar phosphorus in different species ofthe genus Coffea based on soil properties. Rev. Bras. Ciênc. Solo 38, 1439–1447.

Silva, S.A., Lima, J.S.S., Sousa, G.S., Oliveira, R.B., Silva, A.F., 2010. Variabilidade espacial dofósforo e das frações granulométricas de um Latossolo Vermelho Amarelo. Rev.Ciencia Agron. 41, 1–8.

Smith, J.L., Doran, J.W., 1996. Measurement and use of pH and electrical conductivity forsoil quality analysis. In: Doran, J.W., Jones, A.J. (Eds.), Methods for Assessing SoilQualitySSSA Spec. Publ. No. 49. SSSA, Madison, WI, pp. 169–186.

Steubing, L., Fangmeier, L., 1992. Pflanzen-Ökologisches Praktikum. Ulmer Verlag, Germany.Symeonakis, E., Calvo-Cases, A., Arnau-Rosalen, E., 2007. Land use change and land degra-

dation in southeastern Mediterranean Spain. Environ. Manag. 40, 80–94.Tiessen, H., Cuevas, E., Chacon, P., 1994. The role of organic matter in sustaining soil fer-

tility. Nature 371, 783–785.Trasar-Cepeda, C., Leirós, C., Gil-Sotres, F., Seoane, S., 1998. Towards a biochemical quality

index for soils: an expression relating several biological and biochemical properties.Biol. Fertil. Soils 26, 100–106.

Trimble, S.W., 1990. Geomorphic effects of vegetation cover andmanagement: some timeand space considerations in prediction of erosion and sediment yield. In: Thornes, J.(Ed.), Vegetation and Erosion, Processes and Environments.Wiley, London, pp. 55–65.

Tscharntke, T., Klein, A.M., Kruess, A., Steffan-Dewenter, I., Thies, C., 2005. Landscapeperspectives on agricultural intensification and biodiversity ecosystem servicemanagement. Ecol. Lett. 8, 857–874.

Whittaker, R.H., Likens, G.E., 1997. Primary production: the biosphere and man. Hum.Ecol. 1, 357–369.

Zornoza, R., Mataiz-Solera, J., Guerrero, C., Arcenegui, V., Mayoral, A.M., Morales, J.,Mataix-Beneyto, J., 2007. Soil properties under natural forest in the Alicante Provinceof Spain. Geoderma 142, 334–341.

Zornoza, R., Mataix-Solera, J., Guerrero, C., Arcenegui, V., Mataix-Beneyto, J., Gómez, I.,2008. Validating the effectiveness and sensitivity of two soil quality indices base onnatural forest soil underMediterranean conditions. Soil Biol. Biochem. 40, 2079–2087.