Journal of Environmental Management 80 (2006) 248–265 www.elsevier.com/locate/jenvman

Assessment of the rangelands of southwestern Santiago del Estero, Argentina, for grazing management and research

Carlos Kunst *, Eliseo Monti, He´ctor Pe´rez, Jose´ Godoy

Instituto Nacional de Tecnologı´a Agropecuaria (INTA), Estacio´n Experimental Agropecuaria Santiago del Estero, Jujuy 850, 4200 Santiago del Estero, Argentina

Received 30 October 2004; received in revised form 17 September 2005; accepted 10 October 2005 Available online 20 December 2005

Abstract

Native rangelands of the southwest part of the province of Santiago del Estero, Argentina, are a key source of forage for cow–calf operations. The objectives of this study were to delineate the ecosystem units of the area, to describe the associated plant communities and to interpret the role that physical factors and disturbances such as fire and grazing have had in the changes of the structure of these plant communities. This information is needed for developing recommendations for grazing management, for prescribing appropriate improvement practices (e.g. shrub control, prescribed fire) and as guidelines for future research. The ecosystem was divided into smaller units using a hierarchical method, the categories of practical importance being ‘range unit’ and ‘range site’. They represent the catchment and hillslope scale of the water runoff–runon phenomenon, respectively. Vegetation was sampled using a block and cluster sampling design, registering tree, shrub, forb and grass species frequency, and the standing aerial biomass of the herbaceous layer in a sampling unitZ1 ha. Environmental data (topographic position, fire frequency, current and past use, and tree and shrub cover) were also registered for each sampling unit. Indirect ordination of sampling units classified according to range units and range sites, and correlation with environmental variables were performed using multidimensional scaling (MDS) as well as the vector fitting technique. Standing forage and stocking rate were estimated from biomass data. Results indicate that ‘range site’ is the ecosystem unit that should be considered for management purposes since it correlates well with plant communities: tall, hardwood forests are located on upland sites, woodlands are located on midland sites and savannas are located on lowland sites. Dense shrub thickets dominate in areas rated in poor condition, irrespective of range site. Disturbances such as fire and current and past use have a significant positive and negative correlation with range condition, respectively, suggesting that a state and transition model would explain vegetation dynamics better than the succession model. The estimated stocking rate in lowland sites in good condition was 2 ha UGK1, while in upland sites in poor condition the stocking rate was 90 ha UGK1. Active (fire, mechanical treatments) rather than passive (grazing management) methods should be used for range improvement in order to achieve the full potential of the ecosystem. q 2005 Elsevier Ltd. All rights reserved.

Keywords: Chaco region; Range management; Range site

1. Introduction types, including woodlands, shrublands, savannas and forests (Bordo´n, 1983; Fumagalli et al., 1997). Sustainable grazing The province of Santiago del Estero is located in the management of these ‘rangelands’ requires both the generation ‘Chaco’ region, a vast plain that extends into northwestern of information leading to the proper management of animal Argentina and surrounding countries. Its climate ranges from numbers, such as forage yields, species composition, etc. and subhumid to arid and it is highly suitable for cow–calf the incorporation of new areas of complexity, such as spatial operations. In fact, the latter represent a significant source of variation and dynamics, into the management scheme (Walker, income for ranchers and farmers (Fumagalli et al., 1997). 1993). Ecological inventories should provide information on A key source of forage for cow–calf operations is the native physical landscape features as well as disturbance events and vegetation of the Chaco, that comprises several vegetation biotic processes (Maxwell et al., 1995). Several authors have described the landscape, plant communities, temporal dynamics and spatial variation of * Corresponding author. Tel.: C54 385 4224430. the native vegetation of the argentinian Chaco (Frenguelli, E-mail address: [email protected] (C. Kunst). 1940; Morello and Saravia Toledo, 1959; Sarmiento, 1963; 0301-4797/$ - see front matter q 2005 Elsevier Ltd. All rights reserved. Morello and Ada´moli, 1974; Ada´moli et al., 1972; Ada´moli doi:10.1016/j.jenvman.2005.10.001 et al., 1990). These surveys present valuable information on C. Kunst et al. / Journal of Environmental Management 80 (2006) 248–265 249 the physical (climate, geomorphology, soils and fire regime) and biological (botanical composition) features of the Chaco ecosystems. However, differences in concept, objectives, level of perception, sampling methods of vegetation and interpretation sometimes preclude the application of this information in practical rangeland management. More recent information related to range condition and stocking rates is applicable only to a specific ranch or paddock (Kunst et al., 1987). The objectives of this study were: (a) to delineate and describe the ecosystem units of the southwestern part of the Santiago del Estero Province, Argentina; (b) to describe the plant communities (y rangelands) associated with the ecosystem units; (c) to interpret the relative role that physical site factors and disturbances such as fire and grazing have had in the changes of the structure of these plant communities; and (d) to estimate relative stocking rates from the peak forage produced by the herbaceous component of the rangelands. Inferences were also made about which of the current models of vegetation dynamics (e.g. succession and state-and- transition) (Briske et al., 2005) best explain the response of the plant communities to disturbances. Fig. 1. Range units of southwestern Santiago del Estero, with boundaries and major features. Adapted and modified from Angueira and Vargas Gil (1993). Detailed information on native rangeland is needed for developing recommendations for grazing management and prescribing appropriate improvement practices (e.g. shrub water infiltration, thus influencing local soil formation, control, prescribed fire) (Allen-Diaz and Bartolome, 1998; fertility, and growth conditions, as well as species composition Stringham et al., 2003); as guidelines for future research, and and yield of the plant communities (Teague and Smit, 1992; for appraisals of potential returns from economic investments Walker, 1993; Mauchamp et al., 1994; Ludwig and Tongway, (Danckwerts and Teague, 1989). 1995; Wondzell et al., 1996). We modified the Gasto´ method (Gasto´ et al., 1990, 1993) by introducing two ecosystem 2. Material and methods categories that represent two spatial scales of the water runoff– runon phenomenon: 2.1. Study area † the ‘range unit’ category (Schwartz and Walsh, 1991). It It is located in the southwestern part of the province of represents the ‘catchment’ scale of perception of the Santiago del Estero, Argentina, between 278 450–288 450 S process of the water runoff–runon phenomenon (Blo¨schl and 648 450–658 300 W. The study area is a square of and Sivapalan, 1995). A ‘range unit’ was outlined taking 50 km!50 km and comprises the Departments of Choya and into account landform at an approximate scale Guasaya´n of the same province, comprising approximately 1:100,000, soil series, soil associations, and soil texture 250,000 ha (Fig. 1). and depth. The distances involved could be expressed in units of 10 km each. 2.2. Ecosystem delineation and mapping † ‘range site’. It represents the ‘hillslope’ scale of perception of the process (Blo¨schl and Sivapalan, 1995). Range site We divided the ecosystem into homogeneous units using the was the basic ecosystem unit in this survey and was defined progressive and hierarchical disaggregation method developed by the relative position of a sampling unit (SU) in the by Gasto´ et al. (1990, 1993), which is similar to the concept of landscape. Three topographic positions were considered: ‘controlling factors’ described by Bailey (1996).Inthis lowland, intermediate or midland, and upland (Table 2): approach, the first controlling factor of the ecosystem is climate, they either receive extra water or lose water, according to acting at a broader scale, at the top of the hierarchy. On the other their position in the landscape. Distances involved are less hand, soil type and current use of the land act close to the ground, than 10 km, usually 1000–2000 m. at a greater scale. A detailed description of the hierarchy of controlling factors, the units into which the ecosystem was A map of range units and range sites with average scale divided, their conceptual definition and the scale of resolution y1:50,000 was produced by photointerpretation and photo- are presented in Table 1, which also has a list of the selected lecture of remote images (Landsat, aerial photographs and sources consulted. mosaics scale 1:50,000) and by gathering available geomor- In semiarid and arid regions of the world, the process of phological, climate and relief information, soil reports and water runoff–runon from source to sink areas controls soil maps at different scales (Table 1). 250 C. Kunst et al. / Journal of Environmental Management 80 (2006) 248–265

Table 1 Categories of the hierarchical scheme used for ecosystem delineation

Hierarchy level Mapping scale/ Main determinants Attributes used Source of information level of perception for delineation Kingdom 1:5,000,000 Climate Rainfall, air temperatures, Cabrera and Willink (1973) and Walter (1977) climatic indexes World vegetation types.l Dominion 1:10,000,000 Climate Rainfall, air temperatures, Cabrera and Willink (1973) climatic indexes Continental vegetation types Province 1:1,000,000a Climate Rainfall, air temperatures, Morello (1968), Cabrera and Willink (1973), Pen˜a Zubiate climatic indexes et al. (1978), Bianchi and Yan˜ez (1992) and Angueira and Vargas Gil (1993) 1:500,000 Regional vegetation types Range unit 1:500,000 to Geomorphology Landform Zuccardi and Fadda (1971) and Angueira and Vargas Gil 1:125,000 (1993) Slope texture, drainage, soil depth Site 1:20,000 and Local physiogra- Topographic position Field work, aerial photographs higher phy Contour lines EA–IGM topographic sheets, 1970–1980 Range condition – Past and current Botanical composition Field sampling use, distance to watering points Shrub and tree cover, etc.

Based on Gasto´ et al. (1993).

Table 2 Environmental attributes registered at each sampling unit in the range survey of the southwestern Santiago del Estero Province, Argentina

Attribute Scale Units Classes/magnitudes Interpretation Topographic position (tp) Nominal – Upland (K1) Soil depth Intermediate (0.5) Water runoff Lowland (1) Fire frequency (ff) Ordinal – High (3) Frequency of disturbance Medium (2) Low (1) Nil (0) Time since last fire (af) Ordinal – Old (3) Time since last disturbance Medium (2) Recent (1) Nil (0) Shrub and tree cover, height !3 m (cv1)- Interval Number of points haK1 – Light availability Forage accessibility Shrub and tree cover, height O3 m (cv2)- Interval Number of points haK1 – Light availability Soil depth Soil erosion (er) Ordinal – Severe (3) Degree of disturbance Moderate (2) Low (1) Nil (0) Past use (pu) Ordinal – Severe (3) Degree of disturbance Moderate (2) Management Style Nil (1) Current use (cu) Ordinal – Severe (3) Degree of disturbance Moderate (2) Management style Low (1) Range condition (rc) Ordinal – Good (3) Current suitability for grazing Fair (2) Poor (1)

The assumptions of this hierarchical method of ecosystem related to the corresponding ecosystem units at all levels of the partition are: (a) the units into which the ecosystem is divided have hierarchy, both real (stand, patches) and abstract (types) due to rather homogeneous features at a specific scale or level of ‘homogeneity’ of certain features within compartments (Koma´r- perception (Maxwelletal.,1995); and (b) vegetation types are kova, 1993; Klijn and Udo de Haes, 1994; Zogg and Barnes, 1995). C. Kunst et al. / Journal of Environmental Management 80 (2006) 248–265 251

2.3. Characterization of the plant communities and the Soil erosion (er). It refers to litter and soil movement caused environment by water runoff. It was assessed by observing the accumulation of litter and the presence of pedestals, etc. in the soil surface. The objective of this step was to gather field information on Past and current land use (pu/cu). This information was the plant communities located in the ecosystem units gathered from landowners and refers to fire history, stocking delineated in the previous step and their environment. rates, timber operations, grazing times, etc. Cover of shrubs and trees (cv1 and cv2). As defined in Section 2.3. It refers tosunlight availability tothe lower vegetation strata. 2.3.1. Vegetation sampling Range condition (rc). We assessed the ‘value’ of some Within the ecosystem category ‘range unit’, sampling attributes of a SU for the livestock industry (Rightmeyr and locations were selected according to the distance from watering Keys, 1991). SUs were classified into three ‘condition’ classes: points, past and current use, shrub and tree cover, slope aspect, poor, fair and good, based on the following attributes: mean degree of fire disturbance, etc. In the selected sampling standing forage (Section 2.5), botanical composition, shrub and locations, at least two sampling units (SU), areaZ1 ha each, tree cover (Section 2.3), and potential hindrances for livestock selected at random, were marked with flag poles. The following grazing and management. The concept of range condition was attributes were sampled: used in this study as a classification tool rather than a vegetation theory (Pieper and Beck, 1990; Scarnecchia, 1995). † Species relative frequency (absence–presence, %) and Field work was conducted between 1993 and 1997. standing biomass (kg DM haK1) of the herbage stratum 2 using the method BOTANAL, using a 0.25 m quadrat as 2.4. Assessment of the role of physical factors and disturbances subsample, (nZ52) (Tothill et al., 1978). K1 † Total cover (number of strikes ha ) and species relative The objectives of this step were: (a) to reduce complexity of frequency (%) of tree and shrub stratum by the Levy– the information; (b) to identify groups of SUs and key species Madden method, modified as suggested by Passera et al. by means of an objective method; and (c) to assess the links (1983), nZ52. Two heights were considered in the between the configuration of these groups and species in the estimation of total cover: less than 3 m (cv1) and above ordination space with environmental attributes (Coxon, 1982; 3 m (cv2). Height of mature trees in the Chaco region is Minchin, 1987; Clarke and Warwick, 1994). We used non- above 3 m (Morello, 1968); so cv2 estimates mainly tree parametric ordination techniques, the database being the matrix cover, while cv1 estimates shrub cover. of the species relative frequency classified by SU (columns) and species (lines). The sampling design was block and cluster (Cook and Stubbendieck, 1986), a sampling location being a block and the 2.4.1. Sample and species ordination SUs within a block were considered clusters. The ordination method used was multidimensional scaling Physiognomy of the current vegetation was assessed (MDS) (Minchin, 1987, 1998). Previous to a MDS analysis, the visually, using Morello (1968) standards. frequency of the species observed in each SU was transformed Herbage samples taken in the field were oven-dried at 60 8C using log(1Cxi), where xiZrelative frequency of species i. Only for 48 h to obtain dry weight. Plant species nomenclature species with relative frequency O1% were included. The Bray follows Ragonese (1951) and Burkart (1969). Curtis coefficient of similarity (Bray and Curtis, 1957) was used as the basis for the MDS analysis. Each ordination involved 10 random starting configurations and the ordination with the least 2.3.2. Environmental characterization stress was selected as the best. A stress value z0.10 was used as In each SU, the following environmental and conceptual a reference to assess the quality of an ordination (Clarke and characteristics were also registered using nominal and ordinal Warwick, 1994). An ordination axis is assumed to represent an scales (Table 2). Definition of those characteristics follows: underlying gradient controlling species composition, as well as Topographic position (tp). It is the relative position of the community similarity, and can be used for data interpretation sampling unit in the landscape and it is related to the water (Clarke and Warwick, 1994). Ordination of SUs was performed runon–runoff phenomenon (McNab, 1989, 1993). It was at two levels of perception: (a) within the entire study area; and measured visually, observing if the water tended to go away (b) within a specific range unit. from the observer or toward the observer, and also by litter accumulation and movement. 2.4.2. Links of ordination axes with environmental attributes Time since last fire (af). It refers to the period of time passed since the occurrence of a fire in the sampling unit. It was 2.4.2.1. Within the entire study area. The mean relative estimated by interviewing ranchers. frequency of each species was calculated for each range unit. Fire frequency (ff). It was defined by counting the fires that The assessment of links between ordination axes and occurred in a SU in the last 10 years. It was estimated by environmental attributes was done subjectively, observing the interviewing ranchers and observing the number and extent of magnitude of the mean relative frequency, the position of charred branches and trunks on trees and shrubs. the species and SU identified by range units in the ordination 252 C. Kunst et al. / Journal of Environmental Management 80 (2006) 248–265 plots, and using descriptions and agronomic assessments of soil similarities between SUs grouped using hierarchical categories series given by Zuccardi and Fadda (1971), Angueira and of classification (e.g. range unit, range condition and range site). Vargas Gil (1993), and Angueira (1994). In this case, SUs were The null hypothesis is no difference in botanical composition considered ‘replications’ of observations of a range unit. among groups of SUs. The magnitude of R as well as its statistical significance were used for interpretation: there will be 2.4.2.2. Within a specific range unit. Environmental gradients small negative values or positive values near zero when H0 is are not necessarily parallel to an ordination axis (Minchin, true, while Rz1 if groups are completely dissimilar. Usually R 1987). Thus, we searched in all directions in the space for falls between 0 and 1, indicating a degree of discrimination vectors of maximum correlation with any environmental between a set of samples (Clarke and Warwick, 1994; Chapman variable, using the vector fitting technique (Minchin, 1987), and Underwood, 1999). Minimum permutations were 5000, or using the environmental attributes registered for each SU the possible permutations according to the number of samples during their environmental characterization. We calculated actually compared. The package PRIMER (Clarke and maximum Pearson product–moment correlation coefficients Warwick, 1994) was used for calculations. between the environmental attributes and the score in an ordination axes obtained above for each SU (King et al., 2000). 2.5. Assessment of stocking rates of ecosystem units The statistical significance of environmental correlations was for cow–calf operations tested using a Monte-Carlo permutation test rather than a parametric test since samples in ordination space are not The standing dry biomass of grass and broadleaf plants independent (Minchin, 1998; King et al., 2000). We computed estimated in previous steps in each SU was used as an 1000 random permutations and considered correlations estimator of herbage biomass yield (Mitchell, 1983). This significant if p!0.05 (King et al., 2000). Angles between the assumption implies that no specific period of time was fitted vectors and the ordination axes and between fitted vectors considered for the estimation of yield. In order to partially were used to interpret the ordinations as follows: a right angle overcome this limitation, we sampled enclosures of minimum represents independence (Zzero correlation); 1808 represents size w2 ha and annual deferred paddocks, concentrating field perfect negative association; and a 458 angle represents a linear sampling at the end of fall, since the study area has a summer correlation of 0.707 (Coxon, 1982). rainfall pattern. It was assumed that bovines primarily The computer package DECODA (Minchin, 1998) was used consume grass and broadleaf plants, browsing shrubs and for MDS ordinations, vector fitting and calculations of angles trees in extreme situations. Standing forage was estimated by between vectors and of coefficients of correlation. correcting herbaceous standing biomass by indexes consider- ing species preference based on information given by Kunst 2.4.3. Association between SU groups, species and et al. (1986). Means of standing biomass and forage were disturbances estimated for range units, range sites and range conditions Within a range unit, the SUs could be grouped according to across sites within range units and separated using the Duncan range site and range condition: we identified these groups test with an aZ0.05. To protect the mean comparisons from (clusters) in the ordination space. We hypothesized, following unwanted errors we performed an ANOVA using standing Rowe (1991), that ‘the landform, the slowest changing land- biomass and forage as dependent variables and range unit, scape element, is the stage on which more changeable players act range site and range condition as classification factors, out their successional roles’. So, by identifying the topographic considering SU as samples (blocks) and replications as position of a SU, we were able to maintain the position in the subsamples (clusters). All the biomass data estimated during landscape (tp, range site) ‘fixed’, and vegetation changes the 4 years of the study were included in the analysis: thus this attributable to factors such as grazing pressure, fire history, design identifies only the variation that could be attributed to etc. could be promptly assessed by differences in botanical topographic position and soils since no adjustment for total composition and community physiognomy. The ‘membership’ rainfall variation among years was applied. Results should be of a SUs to a specific group was checked in the ordination plots interpreted considering this restriction. The PROC GLM of and changed accordingly, if necessary. the SAS Package (SAS, 1998) was used for statistical Plant species can be grouped according to their response to calculations. biotic and abiotic factors and various perturbations (Friedel Stocking rate was defined as the number of hectares needed et al., 1988; Kashian et al., 2003). The position of the species in by an ‘animal-unit’ (AU) grazing in a specific SU on a the ordination space was estimated as the average of the sustainable basis during a year. It was estimated with the coordinates of the samples where they occur, using the mean following formula: relative frequency in each SU as a weight (Minchin, 1998). The relative position of a species in the ordination plots helped to K1 identify indicator species related to one or more environmental Stocking rate ðha AU Þ factors. The computer package DECODA (Minchin, 1998) was Z 1 used for calculations of species weighted averages. ½Standing forage ðkg DM ha Þ The one- or two-way ANOSIM procedure, based on the R ! = 1; statistic (Clarke and Warwick, 1994) was used to test vegetation UF 3650 kg AU (1) C. Kunst et al. / Journal of Environmental Management 80 (2006) 248–265 253 where AUZanimal-unit, a cow weighing 350 kg liveweight, than the Rolling Plains Range Unit (Table 3). Boundaries weaning one calf a year and consuming 10.5 kg offorage dayK1 between range units are not sharp, but ecotonal. (Z3650 kg yearK1)(Anderson et al., 1980)andUFZuse A total of 110 SUs were sampled and used in this study: 38 factorZ0.5. in the Southern Plain Range Unit, 26 in the Sierra and 46 in the The rain use efficiency (RUE) index (Le Houe´rou, 1984) Rolling Plains Range Units. was used to assess the effect of the runon–runoff on biomass and forage yield. The RUE was calculated using the mean 3.2. Plant communities, their structure and the role of physical standing aerial herbage biomass and standing forage observed factors and disturbances in each range site and range condition across range units, using 600 mm as the average annual rainfall for the study area in 3.2.1. Study area order to compare data of forage yields. The dominant vegetation types in the Southern Plain and Sierra Range Units were forests and shrub thickets while in the Rolling Plains Range Unit they were forest ‘islands’ 3. Results surrounded by savanna. The same plant species were present throughout the study 3.1. Ecosystem classification and mapping area, although their mean relative frequency varied among range units (Table 4). The forest canopy of the Chaco is According to the Ko¨ppen’s classification system, the study composed of two tall (height O15 m when mature) hardwood area is located in the Dry Kingdom, Steppe Domain of South species: Schinopsis quebracho colorado and Aspidosperma America (Bailey, 1996). Its climatic type is tropical/subtropical quebracho blanco (Frenguelli, 1940; Lopez de Casenave et al., semiarid, with evaporation exceeding rainfall and all months 1995). The first species was present in the three range units with air temperatures above 0 8C(Morello and Ada´moli, 1974; with a mean frequency of 1–1.30%, while the second increased Bailey, 1996). The study area has two well-defined climatic its mean frequency from 1.7 to 1.8% in the Southern Plain and seasons: the rainy and hot, from summer to fall (October to Sierra Range Units and up to 5% in the Rolling Plains Range April), and the dry and cold from winter to early spring (May to Unit (Table 4). Prosopis nigra, a medium size tree that is September) (Boletta, 1988). Rainfall in Frias (the main city of common in the forest edge, increasing its abundance in the study area) averages 600 mm annually (extremes 300– degraded areas (Morello and Saravia Toledo, 1959; Lopez de 1000 mm) (Bianchi and Yan˜ez, 1992). Casenave et al., 1995), was present in all range units with a The method used to divide the ecosystem isolated four range mean frequency of 3–5% (Table 4). units, called ‘Sierra’, ‘Rolling Plains’, ‘Southern Plain’ and In the Southern Plain and Sierra Range Units, approxi- ‘Depressional’ (Fig. 1). The latter was not considered in this mately 15–20% of the total mean species frequency study. The ‘Southern Plain’ Range Unit is a plain with a S–SE corresponded to shrub species such as Celtis pallida and aspect, and slope !1%, comprising almost 60% of the study Mimosa detinens (height !2 m and thorny), Lippia turbinata area. The original material of the soils is pampean loess and Larrea divaricata, usually abundant in thickets and (Zuccardi and Fadda, 1971). The ‘Rolling Plains’ Range Unit is degraded areas (Morello and Saravia Toledo, 1959); and a mix of gentle hills alternating with drainage channels with a Aloyssia gratissima, a shrub frequent in fire-prone areas predominant N–NW aspect, and gypsum subsoil (Zuccardi and (Kunst, personal observation) (Table 4). In the Rolling Plains Fadda, 1971). Slope varies between 1 and 5%. The boundary Range Unit, the grasses Pappophorum pappipherum and between the Southern Plain and Rolling Plains units is the Setaria leiantha summed up 18% of the total mean species ‘Ban˜ado de Ovanta-Lavalle’ fault (Fig. 1). The ‘Sierra’ Range frequency (Table 4). Elionorus muticus, a bunchgrass species Unit is a series of hills (sierras) and valleys located west of the dominant in the Chaco savannas (Kunst et al., 2003), showed a Southern Plain Range Unit, related to the Guasaya´n Hills mean frequency of 2% in the Rolling Plains, less than 0.05% in complex (Fig. 1). Soil surveys indicate that the Southern Plain the Southern Plain, and was absent in the Sierra Range Unit. and Sierra Range Units have deeper soils, with textures finer Broadleaf species had a mean frequency of 3–4% in the three

Table 3 Soil characteristics of the range units delineated in southwestern Santiago del Estero, Argentina

Range units Sierra Southern plain Rolling plain Representative soil series Not defined Tapso Frı´as El Abra Guasayan Soil features Total soil depth (cm) 100 100 100 90 30 A Horizon Depth (cm) 23 20 25 28 8 Organic matter (%) 1.84 0.19 4.1 1.95 4.02 Clay (%) – 9 11 10 1 Sand (%) – 32 33 37 43 Silt (%) – 59 56 53 56

Based on Zuccardi and Fadda (1971); Angueira and Vargas Gil (1993). 254 C. Kunst et al. / Journal of Environmental Management 80 (2006) 248–265

Table 4 Average relative frequency of plant species by range unit

Southern Plain, nZ38 Sierra, nZ46 Rolling Plain, nZ46 Species Freq. Species Freq. Species Freq. Lippia turbinata 7.56 Lippia turbinata 6.54 Pappophorum pappi- 12.59 pherum Celtis pallida 7.33 Larrea divaricata 6.09 Setaria leiantha 6.02 Trichloris pluriflora 5.04 Aloyssia gratissima 5.92 Aspidosperma quebra- 5.32 cho blanco Prosopis nigra 4.95 Broadleaf species 3.90 Atamisquea emarginata 4.74 Setaria leiantha 3.86 Setaria leiantha 3.79 Aloyssia gratissima 4.24 Broadleaf species 3.76 Acacia furcaptispina 3.59 Prosopis nigra 4.20 Acacia aroma 3.67 Celtis pallida 3.48 Geoffroea decorticans 3.58 Solanum spp. 3.55 Trichloris pluriflora 3.20 Schinus spp. 3.49 Larrea divaricata 3.31 Setaria spp. 3.05 Lippia turbinata 3.49 Schinus spp. 3.20 Acacia praecox 3.01 Celtis pallida 3.30 Pytecoctenium cynan- 3.03 Digitaria insularis 2.91 Broadleaf species 3.06 choides Digitaria insularis 2.91 Prosopis nigra 2.69 Trichloris pluriflora 2.64 Mimosa detinens 2.89 Stipa spp. 2.67 Acacia aroma 2.53 Aloyssia gratissima 2.80 Pappophorum mucro- 2.30 Digitaria insularis 2.43 nulatum Gouinia latifolia 2.40 Trichloris crinita 2.22 Elionurus muticus 2.29 Paspalum spp. 2.39 Justicia spp. 2.05 Cercicium australe 2.15 Sida spp. 1.97 Zizyphus mistol 1.91 Gouinia latifolia 1.93 Pappophorum pappipherum 1.89 Paspalum spp. 1.81 Larrea divaricata 1.91 Aspidosperma quebracho 1.81 Acacia aroma 1.74 Wissadulla densiflora 1.67 blanco Ruellia ciliatiflora 1.68 Aspidosperma quebra- 1.72 Wedelia glauca 1.39 cho blanco Cercicium australe 1.67 Mimozyganthus carina- 1.65 Setaria spp. 1.36 tus Acacia praecox 1.57 Clematis montevidensis 1.61 Justicia spp. 1.33 Digitaria californica 1.55 Sida spp. 1.52 Castela coccinea 1.29 Malvaceae 1.53 Gouinia latifolia 1.49 Condalia microphylla 1.28 Zizyphus mistol 1.50 Aristida spp. 1.28 Schinopsis quebracho 1.28 colorado Clematis montevidensis 1.39 Digitaria californica 1.26 Zizyphus mistol 1.17 Schinopsis quebracho col- 1.14 Schinopsis quebracho 1.24 Pytecoctenium cynan- 1.07 orado colorado choides Atamisquea emarginata 1.13 Gouinia paraguariensis 1.19 Eragrostis orthoclada 1.06 Setaria spp. 1.13 Castela coccinea 1.14 Acacia furca 0.97 Helitropium spp. 1.12 Gomphrena martiana 1.13 Trichloris crinita 0.95 Chloris ciliata 1.06 Digitaria villossissima 1.13 Crotalaria spp. 0.88 Pappophorum mucronula- 1.03 Schinus spp. 1.08 Acacia praecox 0.82 tum Wissadulla densiflora 0.96 Prosopis torquata 1.04 Botriochloa spp. 0.76 Condalia microphylla 0.87 Eragrostis orthoclada 0.88 Malvaceae spp. 0.74 Geoffroea decorticans 0.80 Cercicium australes 0.77 Setaria can˜uda 0.68 Sphaeralcea bonariensis 0.75 Prosopis alba 0.70 Cassia aphilla 0.66 Trichloris crinita 0.62 Geoffroea decorticans 0.63 Digitaria californica 0.64 Justicia spp. 0.59 Mimosa detinens 0.62 Gouinia paraguariensis 0.59 Atamisquea emarginata 0.51 Setaria globulifera 0.57

Species with average relative frequency O5% shown in bold. range units. Shade adapted grass genera such as Setaria, and Sierra Range Units superimposed each other, while those Trichloris and Digitaria, common in forest gaps (Morello and of the Rolling Plains Range Unit were located towards the right Saravia Toledo, 1959; Kunst et al., 1986, 1987) were also (Fig. 2A). In the ordination space, trees, shrub species of the present with high frequency in the three range units (Table 4). forest and shade adapted grasses were located toward the The MDS ordination of the 110 SUs yielded four axes, with center-left of the plot, while savanna grasses and shrub species a minimum stress of 0.14. In the space defined by the first two such as Acacia aroma were located at the right (Fig. 2B). The ordination axes, the sets of SUs belonging to the Southern Plain shortgrass Neobouteloua lophostachya and the broadleaf C. Kunst et al. / Journal of Environmental Management 80 (2006) 248–265 255

A 1 Table 5 ANOSIM analysis of plant communities at the range unit level of perception Coarse soil texture Differences among Global R Numbera p!R Range units 0.31 0 0.0 Differences between 0 Southern Plain vs Sierra 0.258 0 0.0 Southern Plain vs Rolling 0.303 0 0.0 MDS Axis 2 Increasing light Plains Sierra vs Rolling Plains 0.378 0 0.0 Increasing aridity a Number of permuted statistics greater than or equal to global R. Fine soil texture -1 -1.5 0 1.5 However, the ANOSIM test rejected the null hypothesis of MDS Axis 1 equality of species composition of the plant communities Southern Plain Sierra Rolling Plains (Table 5).

B 1 3.2.2. Within Range Units The identification of the topographic position (range site) of a SU was sometimes difficult in the field. This fact could be Gal lat Des vir Asp que Jus spp Ata ema Zex spp attributed to the gentle topography and the ever present shrub Pap pap Lar div Sch que Eli mut 0 Neo lop Cel pal Set lei Ryc spp thickets of Acacia, Prosopis and Larrea that originated from Aca fur Bot alt Pro tor Pro nigt Alo gra Zin per Gou lat Tri pluAca aroBot spp improper range and forest management. They appeared as a MDS Axis 2 Aca pra Pas spp Tab nod Lip tur homogeneous grayish mantle in black and white aerial Het con photographs and impeded visual assessment in the field. However, sufficiently accurate range site identification of these -1 mismanaged areas could be made from observations of relic -1 -0.5 0 0.5 1 1.5 vegetation found during the survey. MDS Axis 1 Ordinations in four dimensions were found necessary to reduce stresses to the target level in the ordination of the SUs of Fig. 2. MDS ordination of SU by range units: (A) sampling units belonging to the three range units; (B) location of selected species in the ordination space the three range units studied (Table 6). defined by the two first MDS axes. References: Grass species in black, woody species in plain text. Abbreviations: Zex spp.: Zexmenia spp.; Elio mut: 3.2.2.1. Southern Plain Range Unit. The first two MDS Elionurus muticus; Het con: Heteropogon contortus; Pro alb: Prosopis alba; axes separated groups of SUs adequately when classified Set lei: Setaria leiantha; Pap pap: Pappophorum pappipherum; Geo dec: using the attributes ‘range condition’ and ‘topographic Geoffroea decorticans; Tri plu: Trichloris pluriflora; Pasp spp.: Paspalum spp.; z Alo gra: Aloyssia gratissima;Digins:Digitaria insularis;Papmuc: position’ ( range site) (Fig. 3A). The vectors representing Pappophorum mucrunatum; Tab nod: Tabebuia nodosa; Sti spp.: Stipa spp.; fire frequency (ff), soil erosion (er), topographic position (tp) Mim car: Mimozyganthis carinatus; Pro tor: Prosopis torquata; Ata ema: and shrub cover (cv1) presented highly significant coefficients Atamisquea emarginata; Pro nig: Prosopis nigra; Cel pal: Celtis pallida; Lar of correlation with the ordination axes (Table 6). Range div: Larrea divaricata; Aca fur: Acacia furcatispina; Aca aro: Acacia aroma; condition (rc) was positively associated with fire frequency Lip tur: Lippia turbinata; Sch col: Schinopsis quebracho colorado; Asp que: Aspidosperma quebracho blanco. (ff), time since last fire (af) and topographic position (tp); and negatively with soil erosion (er) and shrub cover (cv1). The latter two attributes were highly correlated between them- species Zinnia peruviana, both indicators of degradation selves (Fig. 3A and Table 7). The vectors representing shrub (Morello and Saravia Toledo, 1959), were located to the left, cover (cv1) and tree cover (cv2) were not highly correlated a position suggesting an increase in aridity and overgrazing. (Table 7 and Fig. 3). Tree cover (cv2) was negatively Tabebuia nodosa, a tree species frequent in saline areas associated to topographic position (tp) (Fig. 3). (Ragonese, 1951), was located to the center-left as well. E. A savanna was the dominant physiognomy in the lowland muticus and other species common in depressions within the sites, while forests and thickets dominated in the intermediate savannas, such as Heteropogon contortus and Paspalum spp., and upland sites. Plant communities associated to sites and to were located to the center-right, a position suggesting an condition groups across sites were declared different by the increase in fine soil texture. The first axis was interpreted as a ANOSIM analysis (Table 8). Differences in plant communities gradient of degradation (Fig. 2A and B) and the second MDS were significant along the condition gradient only in the axis as a gradient of sunlight and soil texture (Fig. 2A and B). lowland site (Table 8). In the midland and upland sites only Vegetation similarity between the Southern Plain and Sierra the plant communities of the SUs in poor condition were Range Units was 70%. Similarity between the Rolling Plains significantly different from the others. E. muticus and other and the Sierra Range Units was 68.5%; and between the grass species frequent in full sunlight and savannas, such as Rolling Plains and the Southern Plain Range Units was 63%. Botriochloa alta and Paspalum spp. (Kunst et al., 2003), were 256 C. Kunst et al. / Journal of Environmental Management 80 (2006) 248–265

Table 6 Relationships between environmental attributes and ordination axes of plant communities in range units of southwestern Santiago del Estero: maximum correlations (R) of significant vectors

Range unit Ordination results Vector fitting Angles with ordination axes Stress Axes Attribute Max Rn Prob 1 2 3 4 Southern 0.11 4 rc 0.89 38 0.000 160.7 108.6 85.1 89.5 tp 0.74 38 0.000 112.9 154.0 97.4 98.9 ff 0.86 38 0.000 149.2 93.6 72.7 65.6 af 0.65 38 0.001 156.0 79.2 87.4 111.0 er 0.75 38 0.000 43.5 116.7 75.9 117.5 pu 0.62 38 0.001 93.3 142 87.2 127.6 cu 0.62 38 0.004 81.1 150.3 85.9 117.7 cv1 0.70 38 0.001 25.2 93.6 94.8 114.3 cv2 0.56 38 0.01 66.2 48.0 73.1 123.5 Sierra 0.08 4 rc 0.92 26 0.000 156.5 104.3 72.6 95.2 ff 0.71 20 0.019 128.3 133.6 82.8 69.3 af 0.68 20 0.052 132.8 129.2 82.4 69.6 er 0.86 20 0.000 38.4 73.9 57.4 82.1 pu 0.73 20 0.015 83.0 63.8 82.8 151.7 cu 0.64 20 0.068 50.1 79.8 42.1 94.3 cv2 0.68 26 0.007 85.0 64.1 101.2 151.0 Rolling 0.09 4 rc 0.49 46 0.015 67.2 93.9 76.1 27.4 Plains tp 0.78 46 0.000 39.0 72.8 89.3 56.2 ff 0.56 46 0.008 60.6 67.4 45.3 70.1 af 0.56 42 0.007 60.6 67.4 45.3 70.1 cv1 0.79 46 0.000 109.9 40.7 84.6 123.3 cv2 0.74 46 0.000 128.9 127.0 101.6 116.8

For abbreviations see Table 2. located at the left of the ordination space (Fig. 3B), suggesting a positive association with range condition (rc). On the other hand, smaller shrub species such as C. pallida and Acacia furcaptispina, quite abundant in thickets, were located toward the right (Fig. 3B). Shade adapted grass species such as Trichloris pluriflora and Trichloris crinita had a location similar to those shrub species. Tall tree species forming the Chaco forest canopy, such as A. quebracho blanco and S. quebracho colorado, were located in the upper-right center of the plot.

3.2.2.2. Sierra Range Unit. Range condition (rc), soil erosion (er) and current use (cu) presented significant coefficients of correlationwiththefirstandthirdMDSordinationaxes (Table 6). Neither topographic position (tp) nor tree cover (cv2) showed a significant correlation with any ordination axes. As was the case in the previous range unit, shrub cover (cv1) was negatively correlated with range condition (rc) (Table 7). Two large groups of SUs, A and B, were formed along the axes representing range condition (rc) and shrub cover (cv1) (Fig. 4A). The plant composition of these two large groups was declared significantly different; however, the R statistic did not present a large magnitude (Table 9). Within the group A, the SUs rated in good and fair condition were significantly different from those in poor condition. Within group B, differences of SUs in good condition were significantly different from those of SUs in fair and poor condition. A. quebracho blanco and S. quebracho colorado were located in the center of the plot, while L. turbinata Fig. 3. Southern Plain Range Unit: (A) MDS ordination of sampling units and and L. divaricata were located in the center-right. Grasses and fitted vectors of environmental attributes; (B) species ordination. References in shrub species associated with savannas or full sunlight, such as Fig. 2B. C. Kunst et al. / Journal of Environmental Management 80 (2006) 248–265 257

Table 7 Relationships between environmental attributes and ordination axes of plant communities in range units of southwestern Santiago del Estero: angles between fitted vectors

Range unit rc tp ff af er Pu cu cv1 Southern tp 50.0 plain ff 30.8 73.2 af 36.5 76.3 50.3 er 121.6 80.7 135.4 124.8 pu 72.3 35 98 82.8 52.6 cu 82.4 38.4 104.4 97.8 42.7 15.6 cv1 147.7 102.8 165.5 134 31.4 75.8 67.6 cv2 125.1 135.3 122 91.9 71.5 101.5 104 58.2 Sierra ff 41.3 af 38.0 4.8 er 129.7 124.2 126 pu 95.9 132.5 130.7 80.6 cu 113.8 117.0 118.4 19.7 71.7 cv2 99.6 133.5 131.7 92 18.4 86.8 Rolling tp 39.0 Plains ff 50.6 46.2 af 50.6 46.2 0.0 cv1 130.4 110.1 89.7 89.7 cv2 130.5 156.7 146.6 146.6 90.8

For abbreviations see Table 2.

Paspalum spp. and A. aroma, increased toward the left of the position (tp), and shrub (cv1) and tree cover (cv2) (Table 6). plot. Stipa spp., a common bunchgrass species in hill terrain, was Negative associations existed among range condition (rc), located in the center-below. topographic position (tp) and shrub (cv1) and tree cover (cv2); and positive associations between the former and fire frequency 3.2.2.3. Rolling Plains Range Unit. As in the Southern Plain (ff) and time since last fire (af) (Table 7). The shrub cover (cv1) Range Unit, a savanna was the dominant physiognomy in the was independent of tree cover (cv2) (Table 7 and Fig. 5A). lowland sites, while forests and thickets dominated in the Savanna species such as E. muticus, Paspalum spp., A. intermediate and upland sites. The groups of SUs classified by aroma and Eragrostis spp. were located in the right of the plot, topographic position (tp) (upland, midland and lowland) could while Justicia spp., a shade adapted broadleaf species be readily identified from left to right in the ordination defined by was located in the center-left, sharing the same location as the MDS axis 1 (Fig. 5A). The first two MDS axes showed the A. quebracho blanco. Again, tree cover (cv2) was negatively highest correlation coefficients with the attributes topographic associated to topographic position (tp) (Fig. 5A). The

Table 8 ANOSIM analysis of plant communities of the Southern Plain Range Unit

(a) Two-way analysis, global test Differences among Global R Numbera p!R Site groups (averaged across condition groups) 0.538 0 0.0 Condition groups (averaged across site groups) 0.637 0 0.0 (b) One-way analysis, among range conditions within sites Range site Global R Numbera p!R Pairwise tests R Numberb p!R (among range conditions) Lowland 0.907 0 0.0 Poor vs fair 0.93 1/330 0.3 Poor vs good 1.00 1/36 2.8 Good vs fair 0.64 15/2 13.3 Midland 0.322 227 2.3 Poor vs fair K0.12 1086/1287 84.4 Poor vs good 0.96 1/45 2.2 Good vs fair 1.00 1/21 4.8 Upland 0.874 2 0.3 Poor vs fair 0.25 2/3 66.7 Poor vs good 0.97 1/28 3.6 Good vs fair 1.00 1/28 3.6

a Number of permuted statistics greater than or equal to global R. b Number of significant statistics/possible permutations. 258 C. Kunst et al. / Journal of Environmental Management 80 (2006) 248–265

The mean standing biomass varied between 4900 and 182kgDMhaK1 and showed large standard deviations, making some comparisons not statistically significant. This variation was probably caused by the rainfall differences among the years of study. However, despite these difficulties and irrespective of the range unit, the largest standing biomass magnitudes were observed in SUs located in lowland sites in ‘good’ range condition (Table 11). The proportion of standing forage represented between 20 and 80% of the standing herbage biomass. The difference was due to species rated as ‘non-preferred’ grass species and broadleaf species considered as weeds. There was an increasing trend of non-preferred species as the condition rate decreased (Table 11). Means of standing forage of SUs in fair condition were usually closer to means of standing forage of SUs in good condition. Statistical effects of range unit, range site and range condition on the mean standing forage and on the mean standing biomass were similar. Calculated from the mean standing forage, estimated stocking rates varied between 95 ha AUK1 in upland sites in poor condition and 2 ha AUK1 in lowland sites in good range condition. In range sites rated in good condition, the RUE index calculated using the mean standing herbage biomass varied Fig. 4. Sierra Range Range Unit: (A) MDS ordination of sampling units of unit between 1.4 and 8. However, when calculated from the mean using the topographic axis and the tree and shrub cover axis; (B) species standing forage, the RUE varied between 1 and 5 for both the ordination. References in Fig. 2B. upland and lowland sites (Table 12). The lower RUE magnitudes occurred in the upland sites, while in the lowland ANOSIM analysis indicated that the plant communities sites magnitudes were the higher. The RUE index decreased corresponding to range sites were significantly different, when range condition decreased (Table 12). these differences being significant between the lowland and upland sites (Table 10). Differences across range condition 4. Discussion groups within range sites were not significant. Determining which is the ‘proper’ vegetation of an 3.3. Assessment of stocking rates ecosystem unit is the most difficult question to address in rangeland assessment (O’Brien et al., 2003). So, once the Range unit was not statistically significant as a classification ecosystem units have been defined conceptually and spatially, factor in the ANOVA. Within a specific range unit, range site the next question arising is if their botanical and structural and range condition were both declared significant (p!0.05). differences are due to site (physical) factors or caused by

Table 9 ANOSIM analysis of plant communities of the Sierra Range Unit

(a) Two-way analysis, global test Differences among Global R Numbera p!R Groups A & B (averaged across condition groups) 0.277 3 0.1 (b) One-way analysis, among range conditions within groups Group Global R Numbera p!R Pairwise test R Numberb p!R (among range conditions) A 0.54 1 0.0 Good vs fair 0.56 1/84 1.2 Good vs poor 0.74 1/84 1.2 Fair vs poor 0.52 1/462 0.2 B 0.68 0 0.0 Good vs fair 0.87 1/35 2.9 Good vs poor 0.77 1/35 2.9 Fair vs poor 0.61 1/35 2.9

a Number of permuted statistics greater than or equal to global R. b Number of significant statistics/possible permutations. C. Kunst et al. / Journal of Environmental Management 80 (2006) 248–265 259

abundance may be the reflection of the lower clay content of the soil A horizon of the dominant soil series in the Rolling Plains Range Unit, an abiotic feature that actually influences botanical composition. Therefore, the hypothesis that a specific vegetation type is associated to a range unit could be accepted at first. However, although the ‘woody’ component is normal in the Chaco vegetation, an increase in the shrub abundance is considered a retrogression being caused by mismanagement, a process called ‘homogeneization’ of the plant communities by Ada´moli et al. (1972). Zuccardi and Fadda (1971) indicated that soils of the study area evolved under conditions in which grasses were a more important vegetation type. Reports and vegetation sketches of the study area circa 1860–1870 (Moussy, 1873) indicate a more balanced relationship between the woody and the grass components of the vegetation. One hundred years later, Sarmiento (1963) reported a decrease in size of E. muticus savannas in the area. Monti (1998, personal communication) indicated that a lack of subterranean water in the Rolling Plains Range Unit impeded the sustainable development of ranching operations in the past, thus making this unit the most ‘pristine’ of the three. These reports confirm that the woody component, mostly shrubs, has dominated in recent times in the same parts of the study area and that Fig. 5. Rolling Plains Range Unit: (A) MDS ordination of SU using the differences in vegetation physiognomy and species compo- topographic axis and the range condition axis; (B) species ordination. sition among range units could be partially caused by References in Fig. 2B. management practices. Trunks of A. quebracho blanco and S. quebracho colorado human and/or natural disturbances, since managers only see the are the source of firewood, charcoal and railway tie production present vegetation (Pamo et al., 1991). The native vegetation of (Morello and Saravia Toledo, 1959). These timber operations the study area (southwestern corner of the Chaco region) is have reduced the mean frequency of these species in the forests considered to be rather homogeneous and dominated by tree of the region (Morello and Saravia Toledo, 1959). Although and shrub species (Cabrera and Willink, 1973). Hence, the there is no information in the local literature indicating which expected dominant physiognomy would be forests and thickets. frequency of S. quebracho colorado could be considered Our results indicate that while the woody component prevailed ‘normal’ in a Chaco forest, the frequency of this species in the in the Sierra and Southern Plain Range Units, ‘open areas’ such study area could be considered low. The slightly greater mean as savannas dominated in the Rolling Plains Range Unit. The frequency of A. quebracho blanco in comparison with the latter statistical significance of the ANOSIM test suggests that species in both the Southern Plain and Sierra Range Units differences in species composition among range units actually could be attributed to its wind dispersed seed, high fire exist. These differences in vegetation physiognomy and species resistance and low preference by domestic livestock, features that make it very aggressive (Morello and Saravia Toledo, 1959; Bravo et al., 2001). The high mean frequency of A. Table 10 ANOSIM analysis of plant communities of the Rolling Plains Range Unit quebracho blanco in the Rolling Plains could be attributed to the better adaptation of this species to coarser soil textures. The (a) One-way analysis, global test high mean frequency of P. pappipherum in the Rolling Plains a Differences Global R Number p!R Range Unit, a species common in woodlands and basic soils, among (Burkart, 1969; Anderson et al., 1980) also suggests a Site groups 0.52 0 0.0 difference in biotic factors among range units. (b) One-way analysis, between sites Nevertheless, the magnitude of similarity indexes among Range site Global R Numbera p!R range units and a close inspection of the list of species indicates Lowland vs 0.77 0 0.0 that the differences in botanical composition among range units midland are not large and that the test results could be a mathematical Lowland vs 0.99 0 0.0 artifact due to the large number of SUs involved in the upland calculations. The high similarity in botanical composition Midland vs 0.33 1 0.0 among range units could be attributed to three factors: the upland lumping of different scales of variation into the level of a Number of permuted statistics greater than or equal to R. perception of the ordination and ANOSIM analysis, the 260 C. Kunst et al. / Journal of Environmental Management 80 (2006) 248–265

Table 11 Mean standing aboveground biomass, (a); forage, (b); and estimated stocking rate (c), in range units, sites and conditions in southwestern Santiago del Estero

Range unit Range site Range condition Poor Fair Good (a) Aboveground standing biomass (kg DM haK1) Southern Plain Upland 182 (57) b 895 (25) ab 2283 (680) a Midland – 3890 (3768) a 4092 (1551) a Lowland 391 (53) b 3233 (1137) ab 4900 (2447) a Sierra Upland 425 (83) a 1343 (310) a 2438 (368) a Midland – – 3242 (2263) Lowland 909 (1516) a – 3777 (1564) a Rolling Plains Upland 998 (1054) a 506 (125) a 863 (160) a Midland 548 (296) c 2992 (921) b 4890 (1115) a Lowland 1841 (515) a 4674 (900) a 3173 (903) a (b) Aboveground standing forage (kg DM haK1) Southern Plain Upland 82 (44) b 683 (899) ab 1375 (387) a Midland – 1923 (1863) a 2890 (1203) a Lowland 77 (229) a 2144 (1117) a 3071 (1700) a Sierra Upland 241 (18) a 1107 (272) a 1808 (248) a Midland – – 2031 (1162) Lowland 174 (260) b – 2738 (1170) a Rolling plains Upland 630 (653) a 413 (100) a 626 (137) a Midland 288 (180) c 1589 (494) b 2487 (542) a Lowland 1185 (491) a 3362 (1358) a 1789 (450) a (c) Estimated stocking rate (ha UGK1) Southern Plain Upland 89 11 5 Midland – 4 3 Lowland 95 3 2 Sierra Upland 30 7 4 Midland – – 4 Lowland 42 – 3 Rolling Plains Upland 12 18 12 Midland 25 5 3 Lowland 6 2 4

Means among range conditions followed by different letters indicate significant differences, Duncan test, aZ0.05. adaptation of native species to a variety of environments, and Table 12 the already mentioned homogeneization due to management. Rain use efficiency index (RUE, Le Houe´rou, 1984) calculated for range units, So, there is no clear relationship between range unit, vegetation sites and conditions, using (a) mean standing aerial biomass and (b) mean type and species composition, indicating that some of the basic standing forage (Table 11) and 600 mm as the average annual rainfall assumptions of ecosystem classification and mapping (Koma´r- Range unit Range site Range condition kova, 1993; Zogg and Barnes, 1995) may not hold at certain Poor Fair Good levels of perception of the ecosystem. The weakness of natural (a) RUE with mean standing aerial biomass vegetation as an indicator of ecosystem units of similar Southern Upland 0.30 1.49 3.81 perception level as ‘range unit’ has previously been reported in Plain Midland – 6.48 6.82 the literature (Wright et al., 1998; McNab et al., 1999). Lowland 0.65 5.39 8.17 A more precise correspondence between physiography, Sierra Upland 0.71 2.24 4.06 Midland – – 5.40 vegetation physiognomy and plant species composition allowed Lowland 1.52 – 6.30 to objectively identify and outline the range site, the next level of Rolling Upland 1.66 0.84 1.44 perception of the ecosystem. In both the Rolling Plains and Plains Midland 0.91 4.99 8.15 Southern Plain Range Units topographic position (tp) signifi- Lowland 3.07 7.79 5.29 (b) RUE with mean standing forage cantly affected the species composition of the plant commu- Southern Upland 0.14 1.14 2.29 nities, a result strongly supported by the ANOSIM analysis. The Plain Midland – 3.21 4.82 location of some key species and species groups in the ordination Lowland 0.13 3.57 5.12 space and the correlation between tree cover (cv2) and Sierra Upland 0.40 1.85 3.01 Midland – – 3.39 topographic position (tp) suggest that in both range units, grass Lowland 0.29 4.56 species such as E. muticus, Paspalum spp. and P. pappipherum Rolling Upland 1.05 0.69 1.04 are abundant in lowland sites while trees increase in midland and Plains Midland 0.48 2.65 4.15 upland sites. These results indicate that in most of the study area Lowland 1.98 5.60 2.98 forests and savannas are distinctive communities (‘patches’) Southwest Santiago del Estero province, Chaco region, Argentina. C. Kunst et al. / Journal of Environmental Management 80 (2006) 248–265 261 associated with physical features (e.g. soil type, runon–runoff partition is needed to ascertain whether botanical differences phenomenon). The intermediate site could be interpreted as an among plant communities actually exist. ecotone between the two. Detailed soil surveys of the Chaco In both the Southern Plain and Rolling Plains Range Units region (!1:50,000, Pen˜a Zubiate et al., 1978) also reported shrub cover (cv1) was significantly correlated with erosion (er), differences in soil development and texture along the catena not correlated with tree cover (cv2), and negatively correlated from the upland to the lowland sites. In addition, other ecological with range condition (rc) and fire frequency (ff). When a surveys and range inventories indicate that the plant commu- specific range site was considered in the analysis, the nities in the Chaco are organized as follows: forests occupy the mean frequency of shrub species increased and that of grass upland sites, while the woodlands and savannas are located in the species diminished with increasing intensity of current and past intermediate and lowland sites (Morello and Saravia Toledo, use (y range condition). Shrub species decreased and grass 1959; Pen˜a Zubiate et al., 1978; Bucher, 1982; Morello and species increased when fire frequency increased. The structure Ada´moli, 1974; Kunst et al., 1987). This model of distribution of of the plant community changed and tended to be more ‘woody’ plant communities along a soil catena is quite similar to those with decreasing range condition, irrespective of the range site reported for Africa (Teague and Smit, 1992) and Australia considered. Patches in the landscape differentially capture and (Ludwig and Tongway, 1995). This result confirms our retain resources such as water and nutrients (Ludwig and hypothesis that the basic ecosystem unit in the southwestern Tongway, 1995), causing a diversity of plant communities to part of Santiago del Estero, called ‘range site’ in this study, coexist in the landscape. Replacement of the original plant should be delineated based on the local physiography rather than community patches with shrub communities combined with soil by differences in the plant species composition, as proposed by loss from erosion cause changes in availability of sunlight, fire the classical range methodology (SCS, Soil Conservation frequency, and the primary ecosystem processes (e.g. water Service, 1976). retention, nitrogen cycle), resulting in habitat conditions that Notwithstanding the importance of physiography, fire seems are no longer suitable for some of the original plant organisms. to be also a requirement for the existence of savanna vegetation Therefore, an increase in shrub abundance implies not only a types, confirming earlier ecological studies of the Chaco region decrease in range condition but also a degradation of the Chaco (Morello and Saravia Toledo, 1959; Morello and Ada´moli, ecosystems. 1974; Bucher, 1982). The magnitude of the RUE index of range sites fell within In the Sierra Range Unit, the mean species frequency was the range reported by Le Houe´rou (1984) for semiarid–arid not influenced by topographic position (tp), a fact attributed to areas of the world. When the aboveground herbage biomass is the broken terrain of the unit and the presence of rocks, gravel, considered, the magnitude of RUE of the lowland sites reflects or shallow soils, suggesting that a more detailed ecosystem an average rainfall of 1000–1200 mm, while the RUE of the

Fig. 6. Tridimensional model of range sites, vegetation physiognomy, species composition, forage yields (kg DM haK1) and stocking rates (ha AUK1) for the three range sites in good condition (left) and poor condition (right). Southern Plain and Rolling Plains Range Units, southwest Santiago del Estero Province, western Chaco region, Argentina. Block diagram adapted from Popolizio (1983). 262 C. Kunst et al. / Journal of Environmental Management 80 (2006) 248–265 upland sites would represent a 100–200 mm average rainfall. this study could be classified in two categories: those that are Differences in RUE between sites could be attributed to the site ‘dependent’, such as topographic position (tp), and those competition for sunlight by the woody component of that are site ‘independent’, such as shrub cover (cv1), fire the vegetation as well to intrinsic soil features. Management frequency (ff) and past (pu) and current use (cu), i.e. give a hint (z range condition) produces large differences in RUE, as of the disturbance and management regime of a specific SU. reported by Le Houe´rou (1984). The results of the MDS ordination and vector fitting suggest The high magnitude of the mean standing forage observed in that the presence of some vegetation types (e.g. savannas) and the lowland site could be attributed to water accumulation species is closely associated with the disturbance regime and caused by the runon–runoff phenomenon, a fact that should be management rather than with the site features, almost constant taken into account for paddock design and animal allocation. throughout the time scale considered. In our study, vegetation The magnitude of the stocking rates estimated for sites in good types and the magnitudes of the values of some of their condition is similar to stocking rates reported by Morello ‘attributes’ related to the cattle industry (w range condition) and Saravia Toledo (1959) for the northern Chaco region are a ‘state’ of the ecosystem brought about by disturbance, (5 ha UGK1) but higher than standing forage reported for San especially fire and grazing history, rather than by autogenic Luis (500 kg DM haK1) and stocking rates reported for La Rioja succession. Both factors have been cited as causes of change in (10 ha UGK1), argentinian provinces located in Chaco sub- state and transition models of vegetation dynamics (Rodriguez regions more arid than the southwestern Santiago del Estero Iglesias and Kothmann, 1997). Third, once established, shrub (Anderson et al., 1980; GSL—Gobierno de San Luis, 1995). thickets composed of C. pallida, A. aroma, Acacia furcatipsina This study also confirms that there are two groups of grass and L. divaricata are quite resilient and stable when ‘passive’ species in the southwest part of Santiago del Estero: the first management tools such as removing the degrading agent(s) are group, which comprises E. muticus, Paspalum spp., H. applied (McIver and Starr, 2001), a basic tenet of the contortus, etc. shows a marked preference for open, full succession model (Briske et al., 2005). In these cases, range sunlight areas and is frequent in savannas. The second group, improvement or restoration would mean, for example, the which comprises T. pluriflora, T. crinita and Digitaria spp., is removal of the shrub canopy above a certain magnitude in associated with tree and shrub species, and shows an adaptation order for the sunlight to be available, i.e. that a ‘threshold’ to upland sites with less sunlight availability. In fact, Morello should be passed and that energy will be spent. ‘Active’ rather and Saravia Toledo (1959) report that the latter species than ‘passive’ management tools should be used for range dominate in fire-induced grasslands occurring after fires in improvement or restoration: our field observations suggest that the Chaco forest. mechanical treatments followed with fire should be used in As a summary, a tridimensional model of range sites and the order to achieve the full potential of the native rangelands. mean frequency of species, stocking rates, etc. for good and poor range conditions is presented in Fig. 6. 5. Conclusions and management implications Vegetation classifications and ordination techniques can be used to assess which of the current models of rangeland Ecosystems are a mix of vegetation patches produced by dynamics may correctly explain the field observations and may climate, landform, soil and disturbance regimes. Although the be used to illustrate ecological relationships among community definition of an ecosystem is often arbitrary and not restricted types (Bork et al., 1997). Currently, two models of vegetation to any particular spatial unit or scale, the grouping of its dynamics could be applied in range management: plant biological and physical components in homogenous ecological succession and state-and-transitions, with thresholds separating portions (e.g. ‘ecoregions’, ‘range sites’, ‘ecological sites’) in different states (Briske et al., 2003, 2005). The first theory order to separate the ecosystem into ecologically meaningful assumes that the plant community evolves in a linear fashion, and more tractable units is a practice well established in range proceeding under an autogenic process. Some authors consider science for several purposes: the sampling of vegetation and the hardwood forest of the Chaco as the ‘climax’ community of environment, the incorporation of spatial variation into the region (Cabrera and Willink, 1973). Savannas and management programs, to know the extent over which a grasslands were considered either a successional sere or a particular model of rangeland dynamics would apply, etc. ‘disclimax’ community originated by fire or soil features (Pamo et al., 1991; Walker, 1993; Bailey, 1996; Jax et al., (Sarmiento, 1963; Cabrera and Willink, 1973). However, the 1998; Ludwig et al., 1999; Creque et al., 1999; Bestelmeyer level of perception is ‘regional’ and there is no indication in the et al., 2003). literature about how plant dynamics in the Chaco region would Therefore, the approach and the choice of method of proceed under the successional model. ecosystem mapping and classification, and the level of In the study area a state-and-transition model plus perception at which the spatial variation should be perceived, thresholds seems to be best suited for the data gathered. The are key questions in practical rangeland management. The evidence comes from several facts: first, the latter model in its method of Gasto´ (Gasto´ et al., 1990, 1993) used in this study newest version would accept that the vegetation of a range site for ecosystem stratification or ‘partition’ was multifactor and could be the result of several factors, such fire, use and hierarchical, providing a logical and repeatable ‘disaggrega- physiography (Briske et al., 2005) rather than only an tion’ of ecosystem units of southwestern Santiago del Estero, autogenic effect. Second, the environmental variables used in and agrees with the current approach for ecosystem C. Kunst et al. / Journal of Environmental Management 80 (2006) 248–265 263 partitioning proposed for range management (Stringham et al., References 2001; Pyke et al., 2002; Briske et al., 2005). It could be considered an extension of the approach used by Morello Ada´moli, J., Neumann, R., Colina, A., Morello, J., 1972. El Chaco aluvional (1968) and Ada´moli et al. (1972) in the Chaco region, updated salten˜o. INTA. Revistade Investigaciones Agropecuarias Serie 3 9, 165–237. with recent concepts (hierarchy of controlling factors and Ada´moli, J., Sennhauser, E., Acebo, J., Rescia, A., 1990. Stress and disturbance: vegetation dynamics in the dry Chaco region of Argentina. runoff–runon phenomenon) and more detailed information on Journal of Biogeography 17, 491–500. climate and soils. Allen-Diaz, B., Bartolome, J., 1998. Sagebrush-grass vegetation dynamics: At the range unit level, although differences in vegetation comparing classical and state-transition models. Ecological Applications 8, physiognomy and species composition may exist, they were 795–804. blurred by past management, suggesting that in the southwestern Anderson, D., del Aguila, J., Marchi, A., Vera, J.C., Orionte, E., Bernardo´n, A., w 1980. Manejo racional de un campo en la regio´na´rida de Los Llanos de La part of Santiago del Estero, that category ( catchment scale) is Rioja. INTA, Bs. As. not useful for perceiving the spatial variation of the vegetation Angueira, M., 1994. Evaluacio´n de tierras (Esquema FAO) Lavalle, Tapso, from a practical point of view. Frı´as. INTA EEA Santiago del Estero, 43 p. ‘Range site’ (‘hillslope’ scale) significantly affected Angueira, C., Vargas Gil, J., 1993. Suelos de Lavalle, Tapso y Frı´as. INTA botanical composition and forage yield, and it should be used Carta de suelos de la Repu´blica Argentina. EEA Santiago del Estero. for management planning. Rather than using soil differences, Bailey, R., 1996. Ecosystem Geography. Springer, Berlin. Bestelmeyer, B., Brown, J., Hvastad, K., Alexander, R., Chavez, G., Herrick, J., we employed an indirect approach: land physiography. Based 2003. Development and use of state-and-transition models for rangelands. on the resulting runon–runoff phenomenon, we predicted that Journal of Range Management 56, 114–126. different plant communities exist along a topographic gradient, Bianchi, A., Yan˜ez, C., 1992. Las precipitaciones en el noroeste argentino (2da a fact confirmed by the field data and the ANOSIM analysis. edicio´n). INTA EEA Salta. A strata of trees, shrubs and shade adapted grasses Blo¨schl, G., Sivapalan, M., 1995. Scale issues in hydrological modelling: a review. Hydrological Processes 9, 251–290. characterize the upland site, while grasses and shrubs are the Boletta, P.E., 1988. Clima. Cap.1, pag. 7–21. In: Casas, R.R. (Ed.), Desmonte y features of the lowland site, which also receives extra water. habilitacio´n de tierras en la Regio´n Chaquen˜a semia´rida. Red de The latter also possess the most valuable attributes from the Cooperacio´nTe´cnica en el uso de los Recursos Naturales en la Regio´n point of view of cow–calf operations, such as high forage yield. Chaquen˜a Se mmia´rida, FAO, Santiago, Chile. The intermediate site could be considered an ecotone between Bordo´n, A., 1983. Comentarios e ideogramas sobre la vegetacio´n de la Pcia del the two extremes, dominated by woodland. Shrub thickets are Chaco emergentes de una muestra de descripciones de vegetacio´nen relacio´n a series de suelo. Bol. No 86, INTA EERA Sae´nz Pen˜a. also a common plant community in the study area, associated Bork, E., Hudson, J., Bailey, A., 1997. Upland plant community classification with poor range condition and degradation, irrespective of the in Elk Island national park, Alberta, Canada, using disturbance history and site considered. This ’homogeneization’ of the vegetation physical site factors. Plant Ecology 130, 171–190. physiognomy and thence of the landscape may lead to the Bravo, S., Kunst, C., Gime´nez, A., Moglia, G., 2001. Fire regime of a Elionorus confusion of range sites in the field, a fact that supports the use muticus Spreng. savanna, western Chaco region, Argentina. International Journal of Wildland Fire 10, 1–8. of physiographic rather than botanical features for site Bray, J., Curtis, J., 1957. An ordination of the upland forest communities of mapping. southern Wisconsin. Ecological Monographs 27, 325–349. The factors that cause the existence of the plant communities Briske, D., Fuhlendorf, S., Smeins, F., 2003. Vegetation dynamics on in the study area are physiography, fire and past and current use. rangelands: a critique of the current paradigms. Journal of Applied Ecology A state-and-transition model plus thresholds is better suited to 40, 601–614. explain relationships among these communities and supports Briske, D., Fuhlendorf, S., Smeins, F., 2005. State-and-transition models: thresholds, and rangeland health: a synthesis of ecological concepts and future research on ‘active’ tools of rangeland improvement. perspectives. Rangeland Ecology and Management 58, 1–10. Bucher, E., 1982. Chaco and Caatinga. South American arid savannas, Acknowledgements woodlands and thickets. In: Huntley, B., Walker, B. (Eds.), Ecology of Tropical Savannas Ecol. Studies, vol. 42. Springer, Berlı´n, pp. 48–79. This research was funded by the Instituto Nacional de Burkart, A., 1969. Flora ilustrada de Entre Rı´os (Argentina): Parte II: Gramı´neas. INTA Coleccio´n Cientı´fica Tomo VI, parte II, Buenos Aires, Tecnologı´a Agropecuaria, Tucuma´n-Santiago del Estero 551 p. Regional Center, Research Plan 420108 ‘Relevamiento y Cabrera, A., Willink, A., 1973. Biogeografı´a de Ame´rica Latina. Serie de evaluacio´n de la receptividad bovina de la vegetacio´n natural Biologı´a, Monografı´a N813, Organizacio´n de los Estados Americanos de Sgo del Estero’, 1993–1997. We thank Ing. Zoot. Juan D. (OEA). Sal and Dr Peter Minchin for their help in data interpretation Chapman, M., Underwood, A., 1999. Ecological patterns in multivariate assemblages: information and interpretation of negative values in ANOSIM and computer packages, respectively. We also express our tests. Marine Ecology Progress Series 180, 257–265. thanks to Jose´ Salvatierra, Gero´nimo Elizondo and Dr Javier Clarke, K., Warwick, R., 1994. Change in marine communities, An Approach Grondona for their help in field and statistical work. We would to Statistical Analysis and Interpretation. Plymouth Marine Laboratory, like also to express our gratitude to all the ranchers that Natural Env. Res. Council UK. 144 p. graciously let us enter in their ranches and farms and sample Cook, C., Stubbendieck, J., 1986. Range Research Methods: Basic Problems their pastures and paddocks; and to the three anonymous And Techniques. Society of Range Management, Denver, CO. Coxon, A., 1982. The User’s Guide to Multidimensional Scaling. Heinemann, reviewers whose suggestions improved the manuscript greatly. London. Lic. Lucio Ahuad and Prof. Carmen Kunst should be thanked Creque, J., Basset, S., West, N., 1999. Viewpoint: delineating ecological sites. for their help in the manuscript layout. Journal of Range Management 52, 546–549. 264 C. Kunst et al. / Journal of Environmental Management 80 (2006) 248–265

Danckwerts, J., Teague, W., 1989. Veld management in the Eastern Cape, McNab, W. Henry, 1993. A topographic index to quantify the effect of Pasture Research Section. Stutterheim, Southern Africa. mesoscale landform on site productivity. Canadian Journal of Forest EA-IGM (Eje´rcito Argentino-Instituto Geogra´fico Militar), 1970–1980. Cartas Research 23, 1100–1107. topo´graficas de Santiago del Estero, Repu´blica Argentina. McNab, W. Henry, Browning, S., Simon, S., Fouts, P., 1999. An Frenguelli, J., 1940. Rasgos principales de fitogeografı´a argentina, Publicacion unconventional approach to ecosystem unit classification in western perio´dica N8 2. Museo de La Plata, La Plata, Argentina. North Carolina, USA. Forest Ecology and Management 114, 405–420. Friedel, M., Bastin, G., Grifin, G., 1988. Range assessment and monitoring in Minchin, P., 1987. An evaluation of the relative robustness of techniques for arid lands: the derivation of functional groups to simplify vegetation data. ecological ordination. Vegetatio 69, 89–107. Journal of Environmental Management 27, 85–97. Minchin, P.R., 1998. DECODA: database for ecological community data. Fumagalli, A., Kunst, C., Pe´rez, H., 1997. Intensificacio´n de la produccio´nde Version 3.00, User Documentation. Anutech Pty. Ltd, Canberra. carne en el NOA. In: Carrozzino, J., Rearte, D. (Eds.), Memoria del 1er. Mitchell, J., 1983. Analysis of forage production for assessments and Congreso Nacional de Produccio´n Intensiva de Carne. INTA-Forum Arg. appraisals. USDA—Forest Service GTR RM-98. de Forrajes, Buenos Aires, p. 261. Morello, J., 1968. Las grandes unidades de vegetacio´n y ambiente del Chaco Gasto´, J., Silva, F., Cosio, F., 1990. Sistemas de clasificacio´n de pastizales de argentino. INTA Serie fitogeogra´fica No 10, Buenos Aires. Sudame´rica. Sistemas en Agricultura 9, 1–92. Morello, J., Saravia Toledo, C., 1959. El bosque chaquen˜o I y II. Review Gasto´, J., Cosio, F., Panario, D., 1993. Clasificacion de ecoregiones y Agrono´mica del Noroeste Argentino 3 (5–81), 209–258. determinacio´n de sitio y condicio´n, Manual de aplicacio´n a municipios y Morello J., Ada´moli, J., 1974. Las grandes unidades de vegetacio´n y ambiente predios rurales. Red de Pastizales Andinos, Santiago, Chile. del Chaco argentino. 2da. parte: Vegetacio´n y ambientes de la Pcia. del GSL—Gobierno de San Luis, 1995.Anon., 1995. Plan de desarrollo ganadero Chaco. INTA Serie fitogeogra´fica No 13, Buenos Aires. Provincia de San Luis. San Luis, Argentina. Moussy, M. de., 1873. Description geographique et statistique de la Jax, K., Jones, C., Pickett, S., 1998. The self-identity of ecological units. Oikos Confederation Argentine. 2dxme edition, Atlas. Paris Librairie de Firmin 82, 253–264. Didot Freres. Kashian, D., Barnes, B., Walker, W., 2003. Ecological species groups of O’Brien, R., Johnson, C., Wilson, A., Elsbernd, V., 2003. Indicators of landform-level ecosystems dominated by jack pine in northern Lower rangeland health and functionality in the Intermountain West.US Michigan, USA. Plant Ecology 166, 75–91. Department of Agriculture, Forest Service, Rocky Mountain Research King, R., Nunnery, K., Richardson, C., 2000. Macroinvertebrate assemblage Station, RMRS-GTR-104. response to highway crossings in forested wetlands: implications for Pamo, E., Pieper, R., Beck, R., 1991. Range condition analysis: comparison of biological assessment. Wetlands Ecology and Management 8 (243), 256. 2 methods in southern New Mexico desert grasslands. Journal of Range Klijn, F., Udo de Haes, A., 1994. A hierarchical approach to ecosystems and its Management 44, 374–378. implications for ecological land classification. Landscape Ecology 9, 89–104. Passera, C., Dalmasso, A., Borsetto, O., 1983. Me´todo de “Point Quadrat Koma´rkova, V., 1993. Vegetation type hierarchies and landform disturbance in Modificado”. Memoria del Taller sobre Arbustos Forrajeros en Zonas arctic Alaska and alpine Colorado with emphasis on snowpatches. Aridas y Semia´ridas, Grupo Te´cnico regional de Pasturas, Subgrupo Chaco. Vegetatio 106, 155–181. FAO-IADIZA, Mendoza, Argentina. Kunst, C., Renolfi, R., Pe´rez, H., Dalla Tea, F., 1986. Clasificacio´nde Pen˜a Zubiate, C., Salazar, J., Scoppa, C., Musto, J., 1978. Carta de suelos de los gramı´neas de bosques y arbustales de la regio´n chaquen˜a occidental por sus Departamentos Belgrano y General Taboada, Santiago del Estero. INTA, caracteres ecolo´gicos y productivos. Revista Argentina de Produccio´n Buenos Aires. Animal 6 (Suppl. 1), 6. Pieper, R., Beck, R., 1990. Range condition analysis: modifications to meet Kunst, C., Dalla Tea, F., Pe´rez, H., Renolfi, R., 1987. Vegetacio´n natural y multiple use objectives. Journal of Range Management 43, 550–552. oferta de biomasa herba´cea del sitio ‘Valle fluvial actual del Rı´o Dulce’, Popolizio, E., 1983. Importancia de la cartografı´a geomorfolo´gica en la regio´n chaquen˜a occidental, Sgo del Estero. Revista Argentina de hidrologı´a de llanura. Comite´ Permanente de los Congresos Nacionales del Produccio´n Animal 7, 163–179. Agua. Memorias XI Congreso Nacional del Agua, Tomo 2, Aguas Kunst, C., Bravo, S., Moscovich, F., Herrera, J., Godoy, J., Ve´lez, S., 2003. Superficiales 417-445. Fecha de aplicacio´n de fuego y diversidad de herba´ceas en una sabana de Pyke, D., Herrick, J., Shaver, P., Pellant, M., 2002. Rangeland health atributes Elionorus muticus (Spreng). O. Kuntze. Revista Chilena de Historia Natural and indicators for qualitative assessment. Journal of Range Management 76, 105–115. 55, 584–597. Le Houerou, H., 1984. Rain use efficiency: an unifying concept in arid land Ragonese, A., 1951. La vegetacio´n de las Salinas Grandes. Revista de ecology. Journal of Arid Environments 7, 213–247. Investigaciones Agrı´colas 5, 3–232 (INTA, Buenos Aires). Lopez de Casenave, J., Pelotto, J., Protomastro, J., 1995. Edge-interior Rightmeyr, R., Keys, J., 1991. Utilizing a site suitability guide for site potential differences in vegetation structure and composition in a Chaco semi-arid and species. p. 81-89, In: D. Mengel and D. Thompson Tew: Ecological forest, Argentina. Forest Ecology and Management 72, 61–69. land classification: applications to identify the productive potential of Ludwig, J., Tongway, D., 1995. Spatial organisation of landscapes and its southern forests. Symp. Proc. USDA Forest Service Southeastern Exp. Sta., function in semiarid woodlands, Australia. Landscape EcologyTongway GTR SE-68. 10, 51–63. Rodriguez Iglesias, R., Kothmann, M., 1997. Structure and causes of vegetation Ludwig, J., Tongway, D., Eager, R., Williams, R., Cook, G., 1999. Fine-scale change in state and transition models. Journal of Range Management 50, vegetation patches decline in size and cover with increasing rainfall in 399–408. Australian savannas. Landscape Ecology 14, 557–566. Rowe, J.S., 1991. Forests and ecosystems: implications for their regionaliza- Mauchamp, A., Rambal, S., Lepart, J., 1994. Simulating the dynamics of a tion and classification. In: Mengel, D., Thompson Tew, D.,: ‘Ecological vegetation mosaic. Ecological Modelling 71, 107–130. land classification: applications to identify the productive potential of Maxwell, J., Edwards, C., Jensen, M., Paustian, S., Parrott, H., Hill, D., 1995. A southern forests’. Symp. Proc. USDA Forest Service Southeastern Exp. hierarchical framework of aquatic ecological units in North America Sta., GTR SE-68. (Neartic zone). US Departement of Agriculture, Forest Service, North Sarmiento, G., 1963. Las comunidades vegetales del Chaco semia´rido Central Forest Experiment Station, GTR-NC-176. santiaguen˜o. Tesis. Facultad de Ciencias Exactas y Naturales, Universidad McIver, J., Starr, L., 2001. Restoration of degraded lands in the interior Nacional de Buenos Aires. Columbia River basin: passive vs. active approaches. Forest Ecology and SAS, 1998. Release 6.12 Edition. SAS Institute Inc, Cary, NC. Management 153, 15–28. Scarnecchia, D., 1995. Viewpoint: the rangeland condition concept and range McNab, W., 1989. Terrain Shape Index: quantifyng effects of minor landforms science search for identity: A systems viewpoint. Journal of Range on tree height. Forest Science 35, 91–104. Management 48, 181–186. C. Kunst et al. / Journal of Environmental Management 80 (2006) 248–265 265

Schwartz, H., Walsh, M., 1991. Range unit inventory. In: Schwartz H.J., S. composition. I. Field sampling. CSIRO, Australian Div. Trop. Crops and Shaabani, D. Walther. (Eds.): Range Management Handbook of Kenya, Pastures. Brisbane, Qeeensland, Trop. Agron. Tech Mem. 8. Marsabit District., Vol II (1). Ministry of Livestock Development, Nairobi, Walker, B., 1993. Rangeland ecology: understanding and managing change. Republic of Kenya. Ambio 22, 80–87. SCS, Soil Conservation Service, 1976.Anon., 1976. National Range Handbook. Walter, H., 1977. Zonas de Vegetacio´n y Clima. Ediciones Omega, Barcelona. US Department of Agriculture, Washington, D.C. Wondzell, S., Cunningham, G., Bachelet, D., 1996. Relationships between Stringham, T., Krueger, W., Shaver, P., 2001. States, transitions, and landforms, geomorphic processes, and plant communities on a watershed in thresholds: further refinement for rangelands applications. Agricultural the northern Chihuahuan Desert. Landscape Ecology 11, 351–362. Experiment Station, Oregon State University, Special Report 1024. Wright, R., Murray, M., Merrill, T., 1998. Ecoregions as a level of ecological Stringham, T., Krueger, W., Shaver, P., 2003. State and transition modeling: an analysis. Biological Conservation 86, 207–213. ecological process approach. Journal of Range Management 56, 106–113. Zogg, G., Barnes, B., 1995. Ecological classification and analysis of etland Teague, W., Smit, G., 1992. Relations between woody and herbaceous ecosystem, northern Lower Michigan, USA. Canadian Journal of Forest components and the effects of bush clearing in southern African savannas. Research 25, 1865–1875. Journal of the Grassland Society of Southern Africa 9, 60–71. Zuccardi, R., Fadda, G., 1971. Los suelos del a´rea de Guasaya´n, Santiago del Tothill, J., Hargreaves, J., Jones, R., 1978. BOTANAL. A comprehensive Estero. Revista Agrono´mica del Noroeste Argentino 9, 273–305. sampling and computing procedure for estimating pasture yield and