In arid ecosystems, vegetation controls water and nitrate movement in the soil, reducing solute transport to aquifers. Here we analyzed nitrate distribution and transport throughout the soil profile and to the groundwater under different ecologic (vegetation type) and topographic (upland/lowland) situations across sand dune ecosystems with shallow water tables, subject to domestic grazing in the Monte desert. Based on vertical nitrate distributions in deep soil profiles we found that dune uplands (deep groundwater, low productivity) lost relatively more nitrogen than lowlands (shallow groundwater, high productivity), likely reinforcing productivity contrasts along these topographic positions. The traditional practice of nighttime animal concentration in corrals may affect nitrogen transport, with poorly vegetated interdunes at livestock posts showing higher subsoil nitrate concentrations than a well-vegetated nonsettled interdune. Vegetation left its imprint on the vertical distribution of nitrate, as suggested by the presence of a depletion zone that matched the depth of maximum root densities, followed by an underlying zone of accumulation. To explore how nitrogen exports to groundwater could affect water quality and nutrient supply to phreatophyte plants, we characterized groundwater flow patterns based on a potentiometric map and sediment characteristics, and measured groundwater electric conductivity, nitrate and arsenic concentration, and stable isotopes across 29 wells (5.8–12 m deep). Under the present land use and climate conditions, nitrate leaching does not seem to have an important and widespread effect on water quality. Nitrate concentration exceeded established limits for human consumption (45 mg L−1) in only one well, while arsenic concentration exceeded the established limits (10 μg L−1) in all but one well, reaching extreme values of 629 μg L−1. Yet, our analysis suggests that nitrate exports from corrals can reach the aquifer in localized areas and be transported to the surrounding vegetation in a relatively short time. Vegetation access to groundwater could allow ecosystems to recover part of this nutrient loss, buffering the effects of land use.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Drylands, where water availability is the main constraint of primary productivity and human development, cover about 41% of the Earth land surface, are inhabited by more than 2 billion people (about a third of the world population), support about 78% of grazing worldwide, and are subjected to land use change and increasing land use pressures because of population growth [Asner et al., 2004; Corvalan et al., 2005] (see also www.fao.org). Access to groundwater resources in arid regions by plants and humans allows a higher primary productivity that contributes to sustaining human activities and culture. In the central Monte desert, Argentina, shallow groundwater is exploited by deep rooted trees, increasing primary productivity and coupling the surface with the saturated zone. These groundwater coupled ecosystems found in low topographic areas where the water table is at 6–15 m depth may have 35–99% higher productivities than ecosystems in the same region without access to groundwater resources [Jobbágy et al., 2011]. This high productivity results in the presence of Prosopis flexuosa woodlands that, together with groundwater resources, sustain local pastoralist communities with a rich pre-Hispanic Huarpe identity. Arid regions may have carbon dioxide uptakes comparable to temperate forests and higher carbon storage efficiencies than global averages because dry conditions limit respiration and decomposition [Rotenberg and Yakir, 2010; Schimel, 2010]. Arid groundwater coupled ecosystems may sustain an even higher productivity because groundwater availability increases CO2 uptake rates without having a direct effect on surface soil decomposition and respiration. In these systems, nutrients, regulated mainly by surface water availability [Austin et al., 2004] and land use can become the main constrains of ecosystem productivity.
 Water movement through the soil after precipitation events distributes mobile ions in the horizontal and vertical planes, modifying the spatial nutrient distribution originally caused by litterfall. Uptake of water and nutrients by plant roots “lifts” limiting nutrients to surface soil layers and decreases water and nutrient movement to deep soil horizons, producing shallower distributions for limiting elements [Jobbágy and Jackson, 2001, 2004]. Nitrate ions that “escape” plant absorption in the root zone may accumulate in deep soil layers, forming deep soil nitrate reservoirs in deserts. The scarce organic matter and microbial activity, and the aerobic conditions of these deep sediments prevent further N transformations and gaseous losses, maintaining these nitrate reservoirs over tens of thousands of years [Walvoord et al., 2003], without being available to plants.
 Plants from xeric environments have adaptations to maximize water extraction, reducing the downward movement of water and solutes across the bottom of the root zone [Seyfried et al., 2005]. Reduced deep drainage limits local groundwater recharge by precipitation and transport rates of surface contaminants to the aquifer. Vegetation dynamics has been controlling the water cycle in interdrainage desert areas in the southwestern United States, maintaining an upward water flow since the last glacial period (10,000 to 15,000 years ago) and preventing groundwater recharge under Holocene climate conditions [Scanlon et al., 2005, 2006]. A similarly low or negligible groundwater recharge during Holocene conditions occurred in the Kalahari desert [de Vries et al., 2000]. However, climate variability and land use change associated to cultivation and deforestation may increase groundwater recharge in arid areas up to two orders of magnitude [Scanlon et al., 2006]. Groundwater extraction for different human activities (i.e., irrigation and pastoralism) lifts water and solutes to the surface environment, connecting the aquifer with the soil surface.
 In areas of the Monte desert with shallow groundwater, daily water table fluctuations and stable isotopic composition (δ18O and δD) of groundwater, soil, and sap water suggest that groundwater absorption by the vegetation is higher than local precipitation. This results in a net upward water flux from the aquifer to the atmosphere, with insignificant local recharge rates limited to denuded areas with scarce vegetation. Groundwater is mainly recharged by precipitation in the Andes, more than 100 km away [Jobbágy et al., 2011]. Vegetation absorbs water and nitrate from the soil because both are necessary for plant growth, but some plants have the capacity to exclude chloride ions in the root membrane, decoupling nitrate and chloride vertical distributions [Lambers et al., 1998]. Chloride profiles in the region indicate the occurrence of deep water percolation in denuded areas, mainly in dune crests and slopes [Jobbágy et al., 2011]. Nitrate in arid areas is often produced in pulses associated with peaks of microbial activity during and after rain events [Austin et al., 2004]. Nitrate can be absorbed by the vegetation or lost to deep soil layers or the aquifer, depending on the partition between plant water absorption and drainage.
 Land use change caused by deforestation and replacement of the native vegetation for forestry use affects the soil vertical distributions of different elements in the soil, such as sodium (Na), potassium (K), manganese (Mn) [Jobbágy and Jackson, 2004], and nitrate. A reduction in vegetation cover and the consequently lower root absorption of surface soil water may facilitate deep drainage, groundwater recharge, and the transport of nitrate to the aquifer [Scanlon et al., 2008]. Excessive nitrate concentrations in drinking water, often found in regions of intensive agricultural and fertilizer use, leads to the sometimes fatal condition in infants known as methemoglobinemia [Almasri and Kaluarachchi, 2004; Knobeloch et al., 2000]. In the Monte desert, the high productivity of groundwater-coupled ecosystems supports local human populations, which exploit vegetation and groundwater for domestic and animal consumption [Roig, 1993; Villagra et al., 2009]. In some areas, the population relies exclusively on groundwater of the phreatic aquifer for all uses. The establishment of pastoral settlements is associated with the removal of the woody vegetation to build houses, corrals and water wells, and the shrub layer to displace insects and poisonous animals. Some trees are kept standing to provide shade, but most of the vegetation is cut, increasing bare soil cover. Animals cause a centripetal transport of nutrients from the woodlands to the corrals, accumulate nutrients as urine and dung in areas with scarce vegetation, which may facilitate nitrate leaching to the groundwater. Water for domestic use is extracted from manually built water wells, often located near corrals or livestock accumulation areas (0–50 m away). The presence of high nitrate concentrations in the aquifer may represent a lower water quality for human consumption associated to health problems, but it may as well represent an additional and more stable nutrient source for phreatophyte vegetation, buffering the potential effects of nutrient losses on productivity caused by land use. In addition to the effects of human activities on water quality, mainly related to nitrate contamination from corrals, the geologic characteristics of the aquifer may cause high arsenic concentrations associated with numerous health problems, although regional studies of groundwater arsenic concentrations are limited to irrigated, more populated areas [Farías et al., 2003]. Because the use of natural resources is changing, often increasing the pressure on arid ecosystems and groundwater resources, understanding the interactions and fluxes among vegetation, surface soil, groundwater, and land use is especially important for conservation, health, and land use management.
 In this study, we analyzed the movement of nitrate in the soil and to the groundwater under different vegetation and geomorphologic conditions in the central Monte desert, and evaluated its effect on groundwater quality. We hypothesized that the absorption of water and nutrients by the surface root system of the vegetation in highly productive interdune valleys minimizes the vertical movement of nitrate in the soil and to the aquifer. In areas with low vegetation cover, nitrate vertical movement would increase, compromising water quality. This may occur in high topographic positions with a deep water table, and in corral and housing areas where the vegetation has been partially removed. We made the following predictions: (1) vertical nitrate profiles along transects crossing different topographic positions (dune crests, midslopes, footslopes, and interdune valleys) will differ, with lower nitrate concentration (nitrate leaching) present in high topographic positions and high nitrate concentration (nitrate retention and accumulation) in low topographic positions; (2) nitrate concentrations in soil profiles will be higher in livestock accumulation areas than in a relatively undisturbed interdune valley; and (3) groundwater nitrate concentrations will be higher than baseline levels in domestic wells, compromising water quality.
 For this purpose, we analyzed nitrate concentrations and water content in vertical soil profiles across different topographic positions (dunes and interdune valleys) and among contrasting land use regimes (settled versus nonsettled interdune valleys). We also analyzed characteristics of the sediments (texture and root distribution) and the aquifer (direction and velocity of the groundwater flow, nitrate and arsenic concentration, electric conductivity, and stable isotope composition) in order to evaluate the effect of nitrate leaching associated to local pastoralism on groundwater quality.
2. Materials and Methods
2.1. Study Site
 The study region is located in the central Monte desert, Mendoza, Argentina. Most of our sampling sites were located in and around the Telteca Natural and Cultural Reserve (32° 20′S; 68° 00′W) (Figure 1), where long-term (1972–2007) mean annual precipitation is 156 mm/yr, mostly concentrated in the austral summer (from October to March), and absolute temperatures vary from −10°C in winter to 48°C in summer, with a mean of 18.5°C [Alvarez, 2008]. Annual precipitation (from July to June) registered in meteorological stations located in sites RB4-La Majada and El Pichón (near P2, Figure 1) during the study period was 203 mm during 2006–2007 in El Pichón, 259 and 179 mm during 2007–2008 in El Pichón and RB4-La Majada, respectively, and 157 mm during 2008–2009 in RB4-La Majada.
 Geomorphologically, the region has been shaped by aggradation processes influenced by the lifting of the Andes. The study area is included in the eastern plains of Mendoza, clearly differentiated from the western Andes mountains. The plain fillings are composed of continental Tertiary and Quaternary deposits of pumice and andesitic constituents [Gonzáles Díaz y Fauqué, 1993]. Heterogeneity in substrate, topography, and vegetation is evident at different scales. At the regional scale, the study area includes two geomorphologic units that differ in topography, sediment sizes, and vegetation, with a probable impact on surface hydrological processes (runoff, infiltration), local groundwater recharge, and groundwater flow patterns. Based on Landsat ETM+ images (bands 1, 2, and 3) (http://glcfapp.glcf.umd.edu) we delimitated the two geomorphologic units of the area: the old river bed of Mendoza river (old river bed unit, RB) and the aeolian plain with sand dunes (sand dunes unit, SD) (Figure 1). Two livestock posts (RB7-El Rosal and RB8-Las Llantas) presented the same geomorphologic characteristics as the river bed unit (i.e., fine sediments, plane relief) evident in the field but not detected in the images. The old river bed unit has a variable width from 30 to 2500 m, including the terrace level of 1–3 m, and old outflow areas. It is formed by alluvial and aeolian sediments reworked by the winds and holds an open vegetation dominated by shrubs, with few large individuals of P. flexuosa. The main access, paved road that connects irrigated areas with the desert was established mainly along the river bed unit, providing an easier access to commercialization centers for their products, drinking water transported from irrigated areas, and materials for different uses. Some of these posts have electricity for electric water pumps and refrigeration. The aeolian plain with sand dunes unit is formed by aeolian sediments that developed dune-interdune systems oriented NNW–SSE, with active sand dunes partially stabilized by the vegetation. It hosts discontinuous valleys with elevation gradients of 10–30 m. This unit holds the highest cover of large P. flexuosa trees (http://www.cricyt.edu.ar/ladyot/lava_carto/index.htm) of the region, localized in interdune valleys with proximity to the water table (6–15 m). At the local scale, the sand dunes unit presents heterogeneity in the topographic positions of dune-interdune systems (dune crests, midslopes, and footslopes), with different vegetation cover and structure in areas with different distances to the water table, making dune crests and slopes disconnected from groundwater resources. Dune crests and midslopes have a lower vegetation cover and productivity than interdune valleys [Contreras et al., 2011], resulting in different conditions for water and solutes vertical movement. Interdune valleys have larger P. flexuosa and Bulnesia retama trees than dune crests and midlslopes (height, crown area, trunk diameter), given by the access to groundwater resources. Roots of P. flexuosa trees are horizontally more extended in interdune valleys than in midslopes [Guevara et al., 2010]. Although P. flexuosa is a legume, and it is associated to higher litter and soil nutrient contents than neighboring plants [Abril et al., 2009; Guevara et al., 2010], the occurrence and importance of N2 fixation by this or other species have not been evaluated in the region. Livestock posts in this unit are more isolated from access roads, commercialization centers, and drinking water resources, and they do not have electricity, relying in most cases on manually extracted groundwater for all uses.
 Local groundwater is mostly recharged by precipitation that fell in the Andes [Jobbágy et al., 2011], and is located at 6–15 m depth in interdune valleys. This shallow groundwater is used by phreatophyte vegetation, dominated by P. flexuosa trees (5–10 m height), coexisting with the small tree species Geoffroea decorticans and Bulnesia retama, and shrub species as Larrea divaricata, Atriplex lampa, and Suaeda divaricata. Dunes are dominated by Larrea divaricata, Tricomaria usillo, small individuals of P. flexuosa (lower than 4 m), and the grass Panicum urvilleanum [Alvarez et al., 2006; Villagra et al., 2005]. These woodlands have been used by human populations during several centuries and are still used today by the local inhabitants. In the past, trees were exploited for railroad and vineyard construction without any planning or management, causing irreversible changes in some areas where trees have been completely eliminated. The difficult access to motorized vehicles in some areas given by the sand dunes prevented deforestation in the study region, where old P. flexuosa individuals still remain. Presently, the woodlands are protected by law and wood can only be extracted by the local population for fuel, houses, wells, and corrals construction. Livestock (mainly goats and a lower number of horses and cattle) feeds on the local vegetation and relies on groundwater extracted from hand-dug wells. Most of these wells are cased with a wooden frame of P. flexuosa trees, and water is taken by hand or by pushing with a horse, mule, or donkey. A few domestic wells have a cemented frame, solar pumps, or windmills to extract the water. The “jagüel,” present in a few settlements, consists of a wooden framed ditch that allows animals to walk to the water table level. Livestock corrals and wells are often close to each other (0–50 m), creating a potential risk of groundwater contamination.
2.2. Experimental Design
 Our study addressed three different aspects of nitrate and water dynamics:
 1. Soil nitrate distribution on different topographic positions with different vegetation cover and structure. We inferred patterns of nitrate vertical movement by analyzing soil nitrate vertical distribution in three transects across contiguous dune crests and their interdune valley [Jobbágy et al., 2011]. The topographic positions selected, dune crests, midslopes, footslopes, and interdune valleys, covered a range of distances to the water table, with consequent changes in vegetation cover and structure. Dunes have a lower above and belowground vegetation cover and trees with lower height and biomass than interdune valleys [Contreras et al., 2011; Guevara et al., 2010]. We sampled one or two soil profiles in each topographic position (with a total of six dune crests, five midslopes, six footslopes, and four interdune valleys) along each transect to obtain full toposequences. This study was done during the winter of 2007. Although the areas covered by the three transects (P1-La Penca, P2-Telteca, and P4-Altos Limpios; Figure 1) are visited by livestock occasionally, they do not host livestock posts.
 2. Site and temporal effect on nitrate and soil moisture. We performed a more intensive sampling from spring of 2008 to winter of 2009, focusing on one dune and three interdune valley sites with different land use histories and vegetation situations. We obtained 14 soil profiles: one profile per site in the dune slope, unsettled, and settled but abandoned interdune valleys in December 2008, March 2009, May 2009, and July 2009, and one profile in an active post in an interdune valley in December 2008 and March 2009. We always sampled in exposed areas between tree canopies to avoid the effect of a particular plant on the soil profiles, attempting to sample in the same location per site during the different sampling times, a few meters away from previous sampling points to exclude the soils disturbed by the previous sampling. One of the interdune valley sites (P1-La Penca, relatively undisturbed) did not have a livestock post, had 35 large trees per hectare, and was located about 0.5 km away from the nearest (abandoned) livestock post. A second interdune valley site (SD13-La Penca, abandoned livestock post with fewer trees than P1-La Penca) was located at an abandoned livestock post that had no domestic animals for at least 6 years and showed intermediate degradation, with 24 large trees per hectare. A third site (SD10-Las Hormigas, active livestock post, substantially fewer trees than P1-La Penca) was located in an active livestock post that hosts three families and has been under continuous use for at least 60 years, resulting in a highly degraded vegetation with only nine large trees per hectare. Although livestock visit P1-La Penca and SD13-La Penca sites occasionally, current grazing pressure is low in these sites. SD10-Las Hormigas is heavily grazed and subject to intense animal concentration in corrals. We also analyzed soil texture in the surface, middle, and deeper samples of the soil profiles. Additionally, we described the root distributions down to a depth of 1 m in the midslope of a dune and an interdune valley in P1-La Penca, in order to infer the effect of root absorption on nitrate concentrations.
 3. Impact of nitrate dynamics on groundwater quality. In order to evaluate the effect of traditional land use practices on nitrate contamination of the groundwater, we analyzed hydrogeological characteristics of the aquifer, including nitrate concentrations in sites with different livestock pressures. We estimated groundwater flow direction, velocity, and quality, indicated by the depth of the water table, stable isotope composition, salinity, pH, arsenic, and nitrate concentration in 24 domestic wells and five piezometers (research perforations located in relatively undisturbed sites). The domestic wells were located in both geomorphologic units, while the piezometers were only present in the sand dunes unit. Baseline concentrations of nitrate in groundwater vary in different ecosystems, with values below 2 mg L−1 for natural grasslands in temperate regions [Foster et al., 1982], and values of 0.7–1.04 mg L−1 in several groundwaters from the United States and UK [Zaporozec, 2002; Limbrick, 2003]. We determined groundwater nitrate concentrations in the piezometers in order to establish baseline values for our region. Stable isotope composition allows us to determine the geographic origin of groundwater and estimate the importance of local recharge, because local precipitation has 10–17 ‰ and 60–140 ‰ higher values for δ18O and δD, respectively, than precipitation in the Andes [Jobbágy et al., 2011].
2.3. Soil Sampling
 We obtained the soil samples with a manual soil auger (10 cm diameter), applying a PVC casing if needed to avoid borehole collapse. The samples were taken at intervals of 25 cm for the first 50 cm, and at intervals of 50 cm until the maximum sampling depth (3–5 m). A subsample was separated for gravimetric determination of soil moisture, and another subsample was placed in a cooler or refrigerator, and extracted with a two molar potassium chloride solution (2M KCl) within 3 days after sampling for posterior nitrate analysis. During the first sampling season, September–October 2007, 50 g of soils were extracted with 60 ml of 2M KCl, to obtain nitrate concentrations at detectable levels. During the second sampling period (summer 2008–2009), we extracted only 20 g of soil in 60 ml of KCl, expecting the resulting concentration to be enough based on previous results. Soil extracts were filtered in all cases and frozen until analysis. Nitrate was determined by spectrophotometry with the spongy cadmium method [Jones, 1984]. Soil texture affects soil water vertical movement, so we determined particle size by dry mechanical sieving [American Society for Testing and Materials, 1993] in soil samples obtained in 2008–2009 at the depths of 0–25, 100–150, and 300–350 cm. We determined particle size of the sands, which compose the majority of the sediments [Guevara et al., 2010], quantifying the following sand classes: 425, 250, 125, 63; and the fine fraction corresponding to loam and clay, lower than 63 μm. Coarse plant debris was removed, but we did not chemically remove fine organic matter.
2.4. Root Vertical Distribution
 The vertical distribution of roots was analyzed using the profile count technique [Böhm, 1979]. One cubical soil pit (1 m3) was dug in a high dune, and an interdune valley at P1-La Penca and their four wall flanks (north, south, east, and west) were used to count roots intercepting 10 by 10 cm squares. The results are reported as number of interceptions per dm2 in each of the four walls of the pits.
2.5. Groundwater Characterization
 Water samples were obtained in different periods (2007–2008, 2008–2009, and 2010 summers) from water wells located in different livestock posts of the area (Figure 1). The depth of the water table (w) was recorded at each well in order to obtain the depth of the unsaturated zone and a potentiometric surface of the region to estimate the direction of the groundwater flow. Hydraulic heads (h) were calculated based on measured water table depths (w) at the well sites and elevation data (z), with h = z − w. The elevation data were obtained from the Shuttle Radar Topography Mission–Digital Elevation Model (SRTM-DEM) developed from radar data collected during the 2000 [United States Geological Survey, 2004]. Source for this data was the Global Land Cover Facility (http://www.landcover.org), and Google Earth (http://earth.google.es). Because no other elevation data are available for the study region, we compared the SRTM-DEM data with the closest local geodesic studies at 11 sites, about 50 km west from our sites [Lenzano and Robin, 1995]. Hydraulic gradients (i = dh/dx, where x is horizontal distance) obtained from the potentiometric map indicate the slope of the groundwater table. We used values of hydraulic conductivity (K, 5 m day−1) and effective porosity (Pe, 5%) estimated from the literature, including studies of similar sediments [Custodio and Llamas, 1983; Domenico and Schwartz, 1990; Struckmeier and Margat, 1995] and local studies [Centro Regional de Agua Subterránea, 1979]. With these values, groundwater velocity (Vr) was calculated as follows for the two extreme values of i found in the area:
The error of the estimated velocities (dVr/Vr) associated to the error of z, Pe, and K was calculated by propagation of errors. The error of z was the standard deviation of the difference between the SRTM-DEM and geodesic elevation data at the 11 sites used for the validation [Lenzano and Robin, 1995], with a value of 1.3 m. We assumed an error of 1 m day−1 for K and 1% for Pe, which represent a 20% deviation of the estimated values.
 Groundwater samples were refrigerated in the field and analyzed for nitrate and arsenic concentration, electric conductivity, pH, and stable isotope composition. Subsamples for arsenic determinations were collected in amber glass containers and acidified. Samples were filtered through 0.45 μm pore size membrane filters (Millipore Corporation, Bedford, Massachusetts) immediately after return from the field, and arsenic concentrations were evaluated by electrothermal atomic absorption spectrometry (ETAAS) technique, with a Perkin Elmer (Uberlingen, Germany) Model 5100 ZL atomic absorption spectrometer equipped with a transversely heated graphite atomizer, an As Electrodeless Discharge Lamp (EDL), and a Zeeman-effect background correction system. Samples for nitrate determinations were frozen until analysis with the same method as for the soil extracts. Subsamples were sent to the Duke Environmental Stable Isotope Laboratory for δ18O and δD analysis with a Finnigan MAT 240 Delta Plus XL continuous flow mass spectrometer.
2.6. Statistical Analyses
 Soil nitrate and moisture contents were analyzed with linear mixed effects models with the lme4 package (http://CRAN.R-project.org/package=lme4) in the R statistical environment [R Development Core Team, 2009]. We calculated p values and upper and lower 95% high probability densities for the mixed model parameters with Markov Chain Monte Carlo (MCMC) models using the languageR package (ftp://18.104.22.168/software/rsoft/languageR/html/languageR-package.html). We made separate analysis for the data obtained from the topographic transects, and the data obtained from the spatial and temporal sampling in 2008–2009. For the topographic transects, we considered soil depth (0–25 cm, 25–50 cm, 50–100 cm, and additional 50 cm intervals), and topographic position (dune crest, midslope, footslope, and interdune valley) as fixed factors, and sampling time and site (P1-La Penca, P2-Telteca, and P4-Altos Limpios), as random effects. The three transects were considered three replicates of the main factor to be evaluated, “topographic position.” For 2008–2009 spatial and temporal sampling in sites with different land use intensities, we considered soil depth and site (dune, nonsettled interdune valley, abandoned livestock post, and active livestock post) as fixed factors, and sampling time as a random effect. In this case, because of the lack of replication of the land use gradient, we could only assess statistical differences among the sites, but could not attribute them statistically to the effect of land use. Groundwater nitrate concentrations were also analyzed with linear mixed effect models and MCMCp values, considering land use (research piezometers and domestic wells) and geomorphologic unit (river bed and sand dunes units) as fixed factors, and sampling time as a random effect. Soil moisture data had a normal distribution, while nitrate data were transformed logarithmically to approximate normality. We used the pvals.fnc function of languageR (ftp://22.214.171.124/software/rsoft/languageR/html/languageR-package.html) with 50,000 iterations to estimate significance for each fixed factor.
3.1. Topographic Transects
 Soil nitrate and water content profiles showed variations among the different sites, topographic positions, and soil depth (Figures 2 and 3). The topographic position had significant effects on nitrate concentration (Table 1). Upland positions (dune crests and midslopes) had lower concentrations than lowland positions (interdune valleys and footslopes) (Figures 2 and 3). Soil depth did not have statistically significant effects on soil nitrate, the random factor “site” explained only 5% of the total variance of nitrate concentration, and sampling time did not contribute to the variance (Table 2). Two of the three topographic sequences analyzed presented the highest nitrate concentrations in the footslopes (from 0 to 2 m depth) of the dunes compared to the other topographic positions. Soil moisture contents of the soil profiles from different topographic positions were neither explained by the different sites considered (1.5% of variability was caused by “site”), soil depth, nor topographic position (Table 1). The total amount of nitrate present in the soil down to 250 cm depth was higher in the interdune valleys than in the dune crests (Figure 2).
Table 1. Effect of Fixed Factors on Soil Nitrate and Moisture for the 2007 Topographic Transects and the 2008–2009 Sites with Different Land Use Pressuresa
The three topographic positions and sites from the table were compared to the relatively undisturbed interdune valleys. Model parameters and their standard errors for each factor level were calculated using the “lmer” function of the lme4 package for R. Markov Chain Monte Carlo mean (MCMCmean), upper and lower 95 % high probability densities (HPDlower and HPDupper) of the parameters, and p values (MCMCp) were computed using “pvals.fnc” function of the languageR package for R, with 50,000 iterations. MCMCp values lower than 0.01, indicating significance for each factor, are highlighted in bold.
Table 2. Contribution of Random Factors to the Total Variance, Calculated With the lmer Function of the lme4 Package
Topographic Transects (%)
Sites With Different Land Use (%)
Topographic Transects (%)
Sites With Different Land Use (%)
3.2. Site and Temporal Effect on Nitrate and Soil Moisture
 The different sites considered (dune, nonsettled interdune valley, abandoned, and active livestock posts) had significant effects on soil nitrate contents (Figures 2 and 4, Table 1). Sampling time explained 52% of the soil nitrate variability, but only 4% of the soil moisture variability (Table 2). Soil moisture was affected by soil depth and site, with the active post having the highest soil water content (Figures 2 and 4). In addition, total nitrate contents from 0 to 250 cm depth were higher in soils from the active and abandoned livestock posts (SD13-La Penca and SD10-Las Hormigas) than soils from the nonsettled interdune valley (P1-La Penca) (Figure 2).
 The linear models used for statistical analyses did not show significant effects of soil depth on soil nitrate concentration. However, there are some clear patterns, although nonlinear, that can be inferred from the data. Soil nitrate profiles obtained at La Penca during 2008 and 2009 showed a marked depletion below the surface (from 25 to 50 cm depth, Figure 4) during the growing season (December and March), especially in interdune valleys. The nitrate profiles analyzed near a corral in Las Hormigas showed overall higher and more variable nitrate concentrations than the other profiles (Figure 4). The rooting density was higher from 10 to 40 cm depth, with a peak from 20 to 40 in the interdune valley soil pit (Figure 5).
 Soil texture was coarser in the surface of the dunes than in interdune valleys, and finer in the abandoned and active livestock posts than in the nonsettled interdune valley (Figure 6). Deeper sediments had a similar texture in all sites. Fine particles were more abundant in the surface of the interdune valleys, with up to 21% of clay and silt (smaller than 63 μm, Figure 6), and a predominance of very fine (63–125 μm) and fine (125–250 μm) sands. Medium sands (250 μm) were not abundant, reaching a maximum of 5% from 0 to 0.25 m depth in the dune. The top of the dunes presented lower clay and silt contents (from 2 to 4%) than interdune valleys. The abandoned and active livestock posts had a finer texture than the other sites, especially in the surface, with up to 47% of very fine sand particles and up to 33% of clay and silt contents (Figure 6).
3.3. Groundwater Characteristics
 The SRTM-DEM elevation data was 4.8 ± 1.33 m (mean and standard deviation) higher than the geodesic elevations, with a coefficient of determination of 0.997 (r2) between the two data sets. The error of the flow velocity estimates given by the error of elevation (1.3 m) resulted in 26%, while adding the assumed error of K and Pe resulted in a total error of 67%. The potentiometric map (Figure 7) shows a West to East groundwater flow direction. The average regional hydraulic gradient (i) was 0.27%, reaching 0.1% locally in the NE of the study region. The groundwater velocities calculated with these values and the estimated K and Pe ranged between 0.1 ± 0.07 to 0.25 ± 0.12 m day−1 (Figure 7). The thickness of the unsaturated zone (groundwater table depths) showed the highest values, 10 m, in old river beds and its surroundings, decreasing in wells located in the interdunes (Table 3).
Table 3. Characteristics of the Wells, Sampling Sites, and Groundwatera
Site Code and Name
Thickness of Unsaturated Zone or Water Table Depth (m)
Type of Perforation
Conductivity (mS cm−2)
Arsenic (μg L−1)
Nitrate (mg L−1)
The pH, conductivity, nitrate and arsenic concentration, and stable isotope composition are groundwater characteristics. Asterisks indicate reference samples, extracted from research perforations (piezometers) with minimum grazing and land use intensity. All domestic wells except “SD13-La Penca,” which is currently unoccupied, are located in active livestock posts, with varying numbers and types of domestic animals. The two geomorphologic units distinguished are indicated by SD (aeolian plain with sand dunes) and RB (old river bed of Mendoza River). Different nitrate concentrations in each well represent samples obtained at different times, from 2007 to 2010.
RB1-Bajo del Vino
Domestic well w/electric pump
RB3-Balde de la Vaca
Domestic well w/electric pump
Domestic well w/wind mill
P1-La Penca *
P3-Altos Limpios (bare lowland)*
P4-Altos Limpios (vegetated lowland)*
 Groundwater was mostly alkaline, with pH values from 7.8 to 8.6. Electric conductivity ranged from 1.42 to 10.29 mS cm−2 and it did not increase in the groundwater flow direction. It was generally lower in the old river bed geomorphologic unit, with most values below 2.5 mS cm−2, than in the sand dunes unit, which reached 10 mS cm−2. Arsenic concentrations ranged between 5 and 629 μg L−1, exceeding the maximum contaminant limit (MCL) established by the Código Alimentario Argentino (CAA)  for human consumption (10 μg L−1) in all, except one water well, RB4-La Majada (Table 3). Neither land use (domestic wells and research piezometers) nor geomorphologic unit had statistically significant effects on groundwater nitrate concentrations (MCMCp values = 0.3286 and 0.9640, respectively). Nitrate concentrations in piezometers (P1, P2, P4, and P5), located in relatively undisturbed sites, ranged from 0.7 to 4.4 mg L−1 (Table 3). Background concentration of nitrate may differ according to the vegetation, climate and depth of the unsaturated zone, so we considered the maximum value found at our research perforations, excluding P3, as the baseline concentrations for our region of study: 4.4 mg L−1. The value from piezometer P3 was excluded from the baseline calculation because the isotope composition of groundwater reflects values of local precipitation, indicating the presence of perched groundwater, disconnected from the regional aquifer. Values above 4.4 mg L−1 were assumed to indicate the occurrence of nitrate leaching from surface soils, associated to human activity. Groundwater of 10 domestic wells contained nitrate concentrations above the baseline level at least during one sampling period, but only in one case did nitrate concentration exceed the limit for human consumption (45 mg L−1 at SD10-Las Hormigas, Table 3). Six wells exceeded nitrate concentration of 10 mg L−1. The research perforation in the nonvegetated interdune valley at Altos Limpios, P3, had the second highest nitrate concentration found in this study, reaching 17 mg L−1. Groundwater stable isotope values were generally lower (more negative) in the sand dunes geomorphologic unit than in the old river bed, reaching values of −139 ‰ for δD and −18 ‰ for δ18O. The highest values of both isotopes were found in the unvegetated interdune valley, P3-Altos Limpios, clearly different from the others (−41.67 ‰ for δD and −4.76 ‰ for δ18O), being followed by the “jagüel” (SD6-Las Cuentas), a large open ditch that receives direct inputs of precipitation, and waters from the old river bed geomorphologic unit (i.e., RB6-El Chañar).
4.1. Nitrate in Different Topographic Positions
 The higher nitrate concentrations in lowland (interdune valleys and footslopes) than in upland (dune crests and midslopes) topographic positions agree with our first prediction, proposing higher nitrate leaching and lower nitrate input from litterfall in uplands than in lowlands. This suggests different nitrogen and water dynamics in these positions, controlled by the strong interactions among climate, vegetation, and hydrology. Vegetation in uplands relies on local precipitations [Jobbágy et al., 2011], resulting in lower vegetation cover, lower growing rates of the dominant tree, P. flexuosa, with a root system distributed locally under plant canopies, and lower stocks of nutrients being produced than in the groundwater subsidized lowlands (interdune valleys) [Giordano et al., 2011; Guevara et al., 2010]. The localized root distribution of dune P. flexuosa under tree canopies [Guevara et al., 2010] results in a lower exploration of the soil environment, with a fraction of water and nutrients unutilized, and available for transport to the subsoil, or to lower topographic positions during large precipitation events. Soil water contents in upland positions (dune crests and midslopes) reached and exceeded those of lowland positions (interdune valleys and footslopes) in 2007 (Figures 2 and 3), indicating the permanence of unused resources in high topographic positions.
 The low nitrate concentrations in the dune crest profiles may therefore indicate either a low nitrate production in the surface (owing to the lower biomass and litterfall present in these areas) or a high percolation of rainfall water. In La Penca, dune midslope soils had a coarser soil texture in the surface (Figure 6), slightly lower bulk density, and a higher wetting velocity after water additions [Guevara et al., 2010] than interdune valley soils. Chloride mass balances [Jobbágy et al., 2011], a higher soil water content, and coarser sediments point to the possibility of deep drainage and nitrate leaching from upland positions. Successive precipitation and deep drainage events could result in local groundwater recharge with precipitations, but deeper soil profiles in uplands or chemical markers would be necessary to detect whether deep drainage of rainfall in these high positions reaches the aquifer. Isotopic composition of groundwater indicates that the rates of local recharge are negligible, because it reflects precipitation values of the Andes, clearly different from local rains (Table 3) [Jobbágy et al., 2011].
 Interdune valley trees have access to groundwater, resulting in high vegetation cover, large P. flexuosa trees with higher growing rates, and a root system extended well beyond tree canopies [Giordano et al., 2011; Guevara et al., 2010; Jobbágy et al., 2011]. The structure and high productivity of the vegetation determines higher soil nutrient contents with a more continuous distribution [Guevara et al., 2010]. However, plant demand and nutrient uptake should also be high, resulting in variable quantities of nitrate accumulated in the subsoil. Two topographic sequences (P4-Altos Limpios and P1-La Penca) showed the maximum nitrate concentrations in footslopes, while the other sequence (P2-Telteca) showed the maximum nitrate accumulation in the interdune valleys (data not shown). The high nitrate concentrations in dune footslopes could indicate the occurrence of downward nitrate transport from high to low topographic positions, driven by water flow in the unsaturated soil.
 Nutrient availability may limit plant growth in young phreatophyte individuals growing in sandy soils with artificial irrigation [Villagra and Cavagnaro, 2000]. Our results indicate that the scarce vegetation in high, nutrient poor topographic positions may allow the export of nitrate to highly productive, nutrient rich interdune valleys, reinforcing the nutrient limitation of plant growth in dune crests and slopes and the patterns of productivity found in the different topographic positions.
4.2. Site and Temporal Effects on Soil Nitrate
 Land use change may increase soil nitrate inventories and drainage in arid and semiarid regions, and consequently increase nitrate levels in underlying aquifers [Scanlon et al., 2008]. Grazing by domestic animals may affect soil nitrogen concentration directly by depositing dung and urine in livestock concentration areas, and indirectly, modifying vegetation structure and productivity. The soil profiles obtained during 2008 and 2009 in interdune valleys with different land use intensity showed higher nitrate concentrations in the valleys associated with livestock activity (abandoned and active livestock posts) than in the relatively undisturbed valley (Table 1, Figures 2 and 4). The soil profiles obtained in the active post presented overall higher nitrate concentrations (Figures 2 and 4) and soil moistures than the other soil profiles, probably because urine and dung deposition increases nitrate inputs, and the removal of the vegetation decreases plant absorption, facilitating water and nitrate movement to the subsoil. Soil moisture contents in the active livestock post soil profiles reached the field capacity values reported for the region (7–11%, [Villagra, 1998]) probably facilitating deep drainage and local recharge during precipitation events.
 The depletion of nitrate concentration found below the surface (from 25 to 50 cm depth, Figure 4) is coincident with the depth of maximum rooting density (Figure 5), indicating the imprint of plants on soil nitrate distribution. However, more than 90% of the nitrate found in soil profiles in 2008 was located in the subsoil, from 50 cm to the maximum sampling depth (calculated from Figure 4), indicating that a considerable amount of nitrate escapes root absorption and is transported to deep sediments, not accessed by surface roots. This unused deep soil nitrate would be lost from the ecosystem and accumulated in deep subsoil reservoirs, as happens in other arid regions [Walvoord et al., 2003], representing a net loss of nutrients from the ecosystem. The lack of deep drainage in arid areas caused by the high consumption of rainfall by evapotranspiration prevents this nitrate to reach groundwater reservoirs [Noy-Meir, 1973; Scanlon, 2005], with the probable exception of denuded areas. Subsoil nitrate may be transported to groundwater if precipitations, vegetation removal, and hydraulic conductivity of the sediments in the unsaturated zone allow water percolation to the aquifer. The unsaturated zone in the study region is formed by fine sands, with an estimated field capacity of 7–11%, and a high hydraulic conductivity, which may be increased by root channels [Archer et al., 2002], facilitating nitrate leaching during precipitation events.
 Although more replicates in different land use gradients would be necessary to attribute the differences of nitrate concentration found in the three sites to the effects of land use, our study suggests that land use (urine and dung accumulation and vegetation removal) from small and traditional livestock posts may allow the movement of water and nitrate to the aquifer. The water survey in 29 wells discussed in the next section provides another strategy to test whether nitrate leaching is facilitated in livestock posts.
4.3. Nitrate and Groundwater Quality
 The chemical characterization of groundwater indicated that traditional pastoralism does not have an important impact on water quality, contrary to our third prediction. Baseline nitrate levels found in the region (0.5–4.4 mg L−1) are higher than reported in studies in relatively undisturbed areas [Foster et al., 1982; Limbrick, 2003; Zaporozec, 2002], and may represent nitrate leaching from surface soils outside animal concentration areas due to nitrification, urea deposition by grazing animals, and small events of local recharge during wet periods. More than half of the domestic wells analyzed had equal or lower nitrate concentrations than baseline levels, indicating either the absence of human-induced nitrate leaching, or that contamination with nitrate occurs at a small spatial scale, not reaching sampling wells. The age (SD15-San Miguel was recently settled), and the low number of animals present in some subsistence livestock posts (i.e., SD4-San Expedito, SD11-El Arbolito) may also explain low nitrate values in groundwater. In two wells contaminated with organic matter, nitrate may have been denitrified (SD13-La Penca well was abandoned, and SD6-Las Cuentas jagüel was contaminated with organic matter from the animals that walked inside to drink). The location of the wells upstream from the corral and a large distance between wells and corrals may cause nitrate leaching to be undetected (i.e., RB1-Bajo del Vino).
 However, 10 of 24 domestic wells had higher nitrate contents than the established background levels (4.4 mg L−1 found in piezometers), and six wells had more than 10 mg L−1 of nitrate, indicating the occurrence, in some cases, of nitrate leaching from the surface in animal accumulation areas. Four of these wells, RB3-Balde de la Vaca, RB4-La Majada, SD14-Las Cañas, and SD16-Primavera, were located very close to the corrals (0–30 m) in large and old livestock posts, with a large number of animals. The only well that exceeded nitrate limits for human consumption (45 mg L−1) [CAA, 2007], SD10-Las Hormigas, was located about 20 m downstream from the corral, in an old livestock post. Nitrate concentrations did not change the suitability of groundwater for human consumption even in this well, because the geologic characteristics of the aquifer make it unsuitable for human consumption in the absence of human activity. Arsenic concentration, which is presumed to have a geologic origin in the area [Smedley and Kinniburgh, 2002], exceeded in all but one case the limits established for drinking water (10 μg L−1), reaching extreme values of 629 μg L−1 (Table 3). In this area, livestock activity, even if it increases nitrate leaching rates, is not likely to substantially modify the suitability of groundwater for human consumption. Management efforts to improve water quality should focus on groundwater treatment to remove arsenic or freshwater transport to the area.
 Nitrate leaching and recovery may have an ecological importance if groundwater flow transports nitrate from source areas (denuded or high animal accumulation areas) to the phreatophyte vegetation, and if nitrate is absorbed and returned to the surface environment. The estimated regional groundwater velocities (0.1 ± 0.07 to 0.25 ± 0.12 m/day) indicate that nitrate leached in animal accumulation areas may reach domestic wells and phreatophyte vegetation in less than a year (for a distance of 30 m downstream from the corral). These estimates have to be validated with tracer studies and dynamic hydrogeological models, due to the heterogeneity of the sediments and associated hydraulic conductivities and porosities.
 The two geomorphologic units studied, sand dunes and old river bed, seem to have different local recharge rates, which can affect nitrate leaching to the aquifer. Electric conductivity, indicative of saline content resulting from the contact between groundwater and the sediments, was generally lower in the old river bed unit, indicating a higher recharge with local precipitation. Stable isotope composition, indicative of the geographic origin of the groundwater [Jobbágy et al., 2011], was generally lower in the sand dunes unit (with the exception of P3-Altos Limpios discussed below), while the higher values were found in the old river bed unit, indicating a higher contribution of local precipitations to the old river bed groundwater (Table 3). A higher recharge with local precipitation in the old river bed unit may occur because of the low vegetation coverage, including a lower abundance of large P. flexuosa trees (http://www.cricyt.edu.ar/ladyot/lava_carto/index.htm), and localized water accumulation, infiltration, and percolation to the aquifer given by runoff processes associated to the fine surface sediments and low topographic positions of the old river beds. However, groundwater nitrate concentrations found in this study indicate that the characteristics of each well are more important to determine nitrate leaching than the possible differences in recharge rates in the two geomorphologic units.
 A notable exception of groundwater chemical characteristics is the research perforation P3-Altos Limpios, located in a nonsettled, relatively undisturbed, bare interdune valley, which showed the highest isotope values (−41.67 ‰ for δD and −4.76 ‰ for δ18O), and the fourth highest nitrate concentration of this study (17 mg L−1) (Table 3). This site is located between bare dunes, with almost no vegetation cover during the entire year, conditions that may allow water and solute transport to the aquifer. The nitrate present in this site may have been transported to the area by winds or precipitations over long time periods, because the scarce vegetation and soil organic matter of the site are not likely to produce significant amounts of nitrate. The isotopic composition, reflecting those of local precipitation, soil chloride [Jobbágy et al., 2011], and groundwater nitrate concentrations indicate the occurrence of local recharge by precipitations and downward movement of solutes, allowed by the absence of vegetation and the presumed presence of a confining sediment layer. Consequently, any land use practice that reduces vegetation cover and increases nitrate accumulation, such as settlements with intensive grazing pressures, and increasing precipitations may cause higher levels of groundwater contamination with nitrate. Groundwater nitrate, either originated in animal concentration points or in vegetated areas, might be absorbed by the phreatophyte vegetation and transported back to the surface soil during litter fall. This recovery mechanism may buffer the effects of nutrient losses, maintaining ecosystem productivity even in heavily used interdune valleys. Future studies should focus on quantifying groundwater nitrate transport from livestock concentration areas to vegetated woodlands, detecting groundwater nitrate absorption by deep rooted trees and its effect on productivity and carbon uptake, and deducing the scale at which groundwater access by the vegetation may buffer the effects of land use change.
 The interactions between hydrology and vegetation are evident in the differences between areas with shallow and deep water tables. High productivity, extended root systems, and low water and chloride percolation are found in areas with shallow water tables, while low productivity, reduced root systems, and high water and chloride percolation are found in areas where the water table is deep [Contreras et al., 2011; Guevara et al., 2010; Jobbágy et al., 2011]. These interactions seem to control nitrate dynamics and movement to the groundwater in the central Monte desert. Vegetation affects the spatial distribution of soil nitrate, creating nitrate rich areas in low topographic, highly vegetated positions (footslopes and interdune valleys), and nitrate poor areas in high topographic, poorly vegetated positions (dune crests and midslopes). A nitrate depletion zone observed at soil depths corresponding to the maximum rooting densities, and subsoil nitrate accumulation found below the root zone, reflect the imprint of plants on soil nitrate. Part of the soil nitrate not absorbed by the root system accumulates in the subsoil of vegetated areas. In a few livestock posts, with reduced vegetation cover and high livestock accumulation, nitrate was leached to the groundwater. Under the current climate, hydrogeologic, and land use conditions, water suitability for human consumption was not affected by nitrate leaching, because it was not a widespread process and because high arsenic concentrations determine a naturally poor water quality. Groundwater flow may transport nitrate from leaching areas to the surrounding vegetation in a relatively short time, providing an additional nutrient source to phreatophyte plants and the ecosystem. Land use and climate changes may increase soil nitrate concentration and leaching to the aquifer, and the phreatophyte vegetation may play an important role in recovering and recycling nutrients lost by land use.
 We are thankful to our colleagues who helped us with fieldwork: Alejandra Giantomassi, Carla Giordano, Marcelo Nosseto, Fernando Ríos, and Silvina Ballesteros. We also thank Mariano Gonzalez, Telteca park rangers, Chicho, and his family for their kind hospitality and logistic assistance while working in the area, and Silvana Goirán, Ricardo Mauricio, and Silvina Lassa for laboratory assistance. We also thank the reviewers, who helped us improve our work with their valuable suggestions. This work was financed by the Agencia Nacional de Promoción Científica y Tecnológica (PICT 2007–1222, PBID 1728), a National Geographic Foundation grant given to Esteban Jobbágy, and grants from SECTyP, Universidad Nacional de Cuyo (PID 2009–2011, and PID 2010–2014).