In an area of elevated nitrate (NO3) groundwater concentrations in the northern Chihuahuan Desert in central New Mexico (United States), a large reservoir of nitrate was found in the subsoil of an arroyo floodplain. Nitrate inventories in the floodplain subsoils ranged from 10,000 to 38,000 kg NO3-N/ha—over twice as high as any previously measured arid region. The floodplain subsoil NO3 reservoir was over 100 times higher than the adjacent desert (59–95 kg NO3-N/ha). Chloride mass balance calculations of subsoils indicate arroyo floodplain subsoils have undergone negative recharge since 2600–8600 yr ago, while the surrounding desert has had negative recharge since 13,000–17,000 yr ago. Compared to the adjacent desert, plant communities are larger and more abundant in the floodplain, though subsoil NO3 is apparently not utilized. We demonstrate that NO3 accumulates in the subsoil of the floodplain through evaporation of monsoon season precipitation funneled into the arroyo. Through a one-dimensional vadose zone model, we show that the NO3 inventories in the arroyo floodplain could be acquired 8 to 75 times faster than through atmospheric deposition through the lateral movement of water from the arroyo channel to the adjacent unsaturated zone. As aquifer recharge occurs through the arroyo, channel migration across the floodplain likely flushes subsoil NO3 to the aquifer. High NO3 concentrations and molar ratios of NO3 to Cl in monitoring wells beneath the arroyo floodplain indicate a subsoil NO3 source. These results have major implications for land-use planning in arroyo and ephemeral stream floodplains as well as arid region soil biogeochemistry.


Increased nitrogen loading on the land surface has led to groundwater contamination and degradation of surface-water ecosystems worldwide (Smith et al., 1999; Schlesinger, 2009). As an essential nutrient for life as well as a potential groundwater and surface-water contaminant, there is considerable interest in understanding nitrogen cycles, reservoirs, and sources (Gruber and Galloway, 2008; Canfield et al., 2010; Houlton et al., 2018). This is especially true in desert environments, where limited water resources may be vulnerable to contamination (Stadler et al., 2008).

In arid regions, natural sources of nitrogen include ammonium (NH4) and nitrate (NO3) in wet and dry precipitation, eolian deposition of NO3 salts, and the uptake of atmospheric N2 gas by nitrogen-fixing organisms (Skujiņš, 1981). Arid soils are generally nitrogen-poor soils (Wang et al., 2010) because nitrogen cycling in the first 1 m of soil is rapid due to plant nitrogen uptake, volatilization to ammonium, erosion, and denitrification (Skujiņš, 1981). Consequently, primary productivity in deserts is generally nitrogen limited (Wang et al., 2010). Despite this, several studies have found that, given thousands of years of atmospheric nitrogen deposition, NO3 leaching from soils to the subsoil zone can result in substantial NO3 reservoirs (up to ∼14,000 kg NO3-N/ha) in deserts (Walvoord et al., 2003; Graham et al., 2008; Scanlon et al., 2008). These NO3 reservoirs can be mobilized by land-use change and can lead to groundwater contamination until NO3 sources are spent (Walvoord et al., 2003; Scanlon et al., 2008). These findings have been limited to deserts and high plains in the southwestern United States, with especially high concentrations found beneath desert pavement (Graham et al., 2008) and subsoils in the Mojave Desert (Walvoord et al., 2003). However, much heterogeneity may exist. For example, in the northern Chihuahuan Desert in New Mexico (USA), the subsoil is generally nitrogen-poor, with plant uptake removing the majority of nitrogen (Walvoord et al., 2003; Jackson et al., 2004).

Arroyos (deeply entrenched ephemeral stream channels) have persisted in the American Southwest since at least 8000 14C yr B.P. (Waters and Haynes, 2001). While aquifer recharge beneath desert scrubland from precipitation is generally minimal to nonexistent due to high evapotranspiration rates and thick unsaturated zones (Walvoord and Phillips, 2004), aquifer recharge can occur through arroyos and is generally the only source of valley floor recharge (Sanford et al., 2000; Moore, 2007; Plummer et al., 2012).

At the northern margin of the Chihuahuan Desert in New Mexico, on Kirtland Air Force Base (KAFB; Fig. 1), we investigated the sources of persistently elevated groundwater NO3 concentrations (>10 mg/L NO3-N; KAFB, 2014). After examining spatial and historical geochemical trends in groundwater, we hypothesized that some NO3 in groundwater could be sourced from naturally occurring NO3 deposits in the subsoils of arroyo floodplains. To test this, we sampled soil and subsoil pore water and groundwater for NO3 and chloride (Cl). Because Cl behaves conservatively in the environment, Cl provides a valuable comparison with NO3, which is subject to redox reactions (Scanlon et al., 2008). The region's climate is semiarid, receiving on average 22 cm of precipitation per year (National Oceanic and Atmospheric Administration, 2020). Past studies of groundwater chemistry in the field area have considered potential anthropogenic sources of NO3 to ground-water, including human wastewater and landfills; however, determination of NO3 sources to the aquifer has been largely inconclusive (KAFB, 2014). The field area is bisected by Tijeras Arroyo, a major arroyo draining 332 km2 (U.S. Geological Survey, 2020a). The arroyo channel has carved a floodplain 30–50 m lower than the surrounding desert (Fig. 1); the floodplain varies in width from 500 to 1000 m through the study site, while the channel is ∼10 m wide. A thick vadose zone (>150 m) and underlying aquifer are composed of Oligocene- to Pleistocene-age Santa Fe Group sediments (Kernodle and Scott, 1986; Plummer et al., 2012). Flow through Tijeras Arroyo (typically 1–3 m3/s) is largely confined to the summer monsoon season during storms (see the Supplemental Material1).


We selected eight sites for sediment core sampling (Fig. 1) in 2018. One in the arroyo channel (AC), five in the arroyo floodplain (AF1–AF5), and two on the adjacent mesa (MT1 and MT2). Sediment samples were collected using a track-mounted Geoprobe® dual-tube hollow-stem auger system, without the use of drilling fluids. Extracted cores were analyzed for water potential, gravitational water content, bulk density, and porosity (see the Supplemental Material). Pore-water concentrations of Cl, NO3, nitrite (NO2), bromide (Br), and sulfate (SO4) were determined from the anion mass mobilized during 18 MΩ deionized water extraction and the gravimetric water content of each sample. Data are available in Table S3 in the Supplemental Material, and in the U.S. Geological Survey National Water Information System database (NWIS; U.S. Geological Survey, 2020b) by using the site identifiers presented in Table S3.

We used chloride mass balance (CMB) to calculate the apparent Cl residence time in the subsoil profiles (Phillips, 1994; Rivett et al., 2008). Age was calculated by dividing the Cl inventory of the soil profile to a specific depth by the annual Cl depositional flux (see the Supplemental Material). This method assumes one-dimensional piston flow and constant Cl deposition. The estimated age is not the true soil or subsoil age but the time over which the system has been characterized by negative recharge (defined as when water potential gradients have been consistently upward).

In 2017, a groundwater sampling campaign was completed by the U.S. Geological Survey through the field area (see the Supplemental Material). Results from this work were compared with historical sampling from 2002 to 2017 completed by the U.S. Air Force Civil Engineer Center (Lackland, Texas) using similar methods (Tables S1 and S2).


Sediment pore-water NO3 concentrations were exceptionally high in the arroyo floodplain, reaching concentrations >4000 mg/L NO3-N (Fig. 2). In contrast, sites MT1 and MT2, located above the floodplain, had mean NO3 concentrations of 37 mg/L NO3-N (n = 36) and greater Cl concentrations (mean = 1947 mg/L, n = 38). Pore water in the arroyo channel (site AC) had low NO3 (mean = 6.6 mg/L NO3-N, n = 34) and Cl (mean = 27 mg/L, n = 22) concentrations. Nitrate inventories were calculated over the length of soil and subsoil profile and ranged from 10,000 to 38,000 kg/ha NO3-N in the arroyo floodplain, i.e., over twice as high as any previously measured arid region soil or subsoil NO3 reservoirs (Walvoord et al., 2003; Graham et al., 2008). This was in stark contrast to NO3 inventories at sites MT1 and MT2 in the adjacent desert (59–95 kg/ha NO3-N; Fig. 2). The arroyo channel site (AC), which is subject to repeated infiltration during rainfall, had relatively low NO3 inventories (200 kg/ha NO3-N). No measurable NO2 was detected at any site. Sediment properties are described in the Supplemental Material (Tables S3 and S4).

Outside of the arroyo floodplain (at sites MT1 and MT2), pore-water NO3 concentrations followed a typical nutrient-type profile (Jobbágy and Jackson, 2001), with concentrations that decreased rapidly with depth due to plant uptake of NO3 (Fig. 2). This contrasts with the arroyo floodplain, where NO3 concentrations were much greater than Cl and followed a solute-type profile, forming high concentration bulges concurrent with Cl (Fig. 2). While the nutrient-type NO3 soil profiles at sites MT1 and MT2 match previous results from the Chihuahuan Desert collected in both desert scrub and semiarid grasslands (Walvoord et al., 2003; Jackson et al., 2004), the large NO3 reservoir in the arroyo floodplain sites is unexpected. The molar NO3/Cl ratios in pore waters in the arroyo floodplain profiles were generally between 10 and 30 (Fig. 3), i.e., higher than the local atmospheric deposition total-N/Cl molar ratio of ∼11 (National Atmospheric Deposition Program, http://nadp.slh.wisc.edu/data/sites/siteDetails.aspx?net=NTN&id=NM07).

Despite the high NO3 inventories in the floodplain relative to the surrounding desert, plants in the arroyo floodplain were larger, dominated by 1–2-m-tall scrub vegetation, whereas the adjacent desert was covered in shorter (∼0.2 m) tussock grasses (see the Supplemental Material). Water potential measurements in all sediment cores except the arroyo channel (AC) were negative and showed upward potential head gradients (see the Supplemental Material; Fig. S2), implying discharge through evapotranspiration and no recharge (Walvoord and Phillips, 2004). Water potential in the arroyo channel was slightly positive, indicating saturated sediment.

Groundwater NO3 concentrations in the floodplain ranged between 0.78 and 71.6 mg/L NO3-N; concentrations outside of the floodplain were between 0.87 and 27.7 mg/L NO3-N (Tables S1 and S2). Molar NO3/Cl ratios were generally much higher in the floodplain, ranging from <1 to 14, whereas outside of the flood-plain, molar NO3/Cl ratios were more tightly constrained and lower (<1–2.6; Fig. 3).


The exceptionally high NO3 inventories in the arroyo floodplain relative to the adjacent desert are likely due to variance in soil age, plant communities, or hydrologic processes unique to the arroyo. Based on CMB calculations, sub-soils in the adjacent desert have been undergoing negative recharge for much longer (since 17,000–13,000 yr B.P.) than those in the arroyo floodplain (since 8600–2600 yr B.P.; Table S4). The older age of subsoils in the adjacent desert likely corresponds to the change in climate at the end of the last glacial period (Connell et al., 2007); the CMB ages in the floodplain may correspond to a depositional event, or the CMB clock may have been reset when the arroyo incised the floodplain, and there were no longer frequent overbank flows to leach through the sediments.

Water potential through the arroyo channel (site AC) was ∼0 m through the entire sampled column, while water potentials were highly negative (−500 m to −1000 m) in the surrounding floodplain (Fig. S2). Hence, following flow in the arroyo, a strong horizontal gradient exists beneath the arroyo channel to the surrounding dry vadose zone that will drive lateral unsatu-rated flow (Fig. 4). We modeled the lateral water and solute flux from the arroyo channel to the unsaturated floodplain sediments using a one-dimensional model constructed in Hydrus 1-D (Fig. 4; see the Supplemental Material; Šimůnek et al., 2012). Given initial Cl concentrations similar to site AC (10–20 mg/L) and modeled unsaturated flow, Cl inventories in the floodplain could be reached within 200–800 yr, i.e., 8 to 75 times faster than atmospheric deposition CMB calculations (Table S4).

Nitrate Cycling in the Arroyo Streambed

While the vadose zone model shows that rapid solute acquisition in unsaturated sediments in the arroyo floodplain is plausible, NO3/Cl ratios measured in floodplain pore waters are higher than those beneath the arroyo (site AC). In ephemeral streams, repeated wetting and drying cycles cause increased NO3 concentration in surface waters and associated hyporheic zones. This is because drying leads to oxygenated conditions, which favor aerobic microbial processes that stimulate nitrification and depress denitrification (Gómez et al., 2012). Subsequent rewetting releases NO3 pulses to surface waters (Arce et al., 2014) that infiltrate back into the ground. Beneath the arroyo channel, pore-water NO3 concentrations increased with depth from 0.1 mg/L NO3-N at the surface to ∼1.7 mg/L NO3-N below 10 m depth (Supplemental Material). Net nitrification rates in parafluvial sediments near ephemeral streams can be upwards of 8 ng N g–1 h–1 during the summer (Holmes et al., 1994). Assuming a nitrification rate of 8 ng N g –1 h–1, 50–120 d of saturation during a typical summer (U.S. Geological Survey, 2020a), a bulk density of 0.5–1.5 g cm–3, and sediment water content of 0.05–0.15 mL cm–3 (typical parameters in the floodplain; Tables S3 and S4), pore-water NO3 concentrations adjacent to the arroyo streambed could increase by 42–922 mg/L NO3-N, which is more than enough to explain the high NO3/Cl ratios measured in the floodplain pore waters.

As NO3 moves into the unsaturated zone, aeration and drying likely prevent the reduction of NO3. Because Cl is conservative, the high NO3/Cl ratios in the arroyo floodplain may be the result of enhanced nitrogen fixing and nitrification during wetting and drying cycles in the arroyo channel. These cycles could lead to preservation of NO3 as it migrates into the unsaturated zone. Many xeric plant species do not access nutrients below 1 m depth (Evans and Ehleringer, 1994). Consequently, lateral transport of NO3 several meters beneath the arroyo channel surface to the surrounding dry sediments may bypass most plant nutrient uptake and allow high NO3/Cl ratios to persist.

Groundwater Contamination from Natural NO3 Sources

As the arroyo channel migrates over the floodplain (see the Supplemental Material; Fig. S4; Friedman et al., 2014), it may flush solutes in the floodplain to the aquifer. Molar ratios of NO3/Cl in groundwater show that there is a low-Cl, high-NO3 source of NO3 to groundwater beneath the arroyo floodplain (Fig. 3). Critically, similar groundwater chemistry was observed in groundwater samples collected in the Hell Canyon Arroyo flood-plain (Fig. 3), an arroyo located 15 km south of the field area that also lacks surrounding agriculture (Fig. 3). While agricultural runoff generally has high NO3/Cl ratios (Abualnaeem et al., 2018), potential anthropogenic sources of NO3 to the aquifer are largely from human wastewater (KAFB, 2014), which typically has NO3/Cl ratios of ∼0.07 (Fig. 3; Abualnaeem et al., 2018).

Assuming an aquifer saturated thickness of 30 m and a porosity of 0.2, and complete mixing over this depth, flushing all observed NO3 in the vadose zone (10,000–38,000 kg/ha NO3-N) into the aquifer would result in NO3 groundwater concentrations between 167 and 633 mg/L NO3-N. Observed NO3 concentrations in the arroyo floodplain ranged from 3.85 to 71.7 mg/L NO3-N (Fig. 3).


Arid land covers ∼35% of land surface worldwide and ∼25% of the conterminous United States (North et al., 2014); arroyos and ephemeral streams drain much of this land surface. If many arroyo floodplains hold large subsoil NO3 reservoirs, this discovery has major implications for water quality following land use and climate change. Farming, housing development, and dam building within arroyo floodplains could mobilize NO3 and lead to aquifer contamination. While desert plants are generally nitrogen-limited, the deep-rooted vegetation in the arroyo floodplain is not utilizing the available large store of nitrogen. This has biogeochemical and ecological implications and suggests that these plants are phosphorous limited. Further research is merited to determine the biogeochemical causes and extensiveness of arroyo floodplain NO3 reservoirs.


This work was funded by the Air Force Civil Engineer Center (Lackland, Texas, USA). We would like to thank USGS and AECOM technicians for help with sample collection and analyses, as well as Isleta Pueblo, the City of Albuquerque, the New Mexico Environment Department, and Bernalillo County for their help and expertise. We sincerely thank the reviewers of this work. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. government.

1Supplemental Material. Additional information on site background, sampling methods, chloride mass balance calculations, and vadose zone modeling. Please visit https://doi.org/10.1130/GEOL.S.13584950 to access the supplemental material, and contact editing@geosociety.org with any questions.
Gold Open Access: This paper is published under the terms of the CC-BY license.