Review ARTICLE t Deep vadose zone demonstrates fate of nitrate in eastern San Joaquin Valley

Thomas Harter Yuksel S. Onsoy Katrin Heeren Michelle Denton Gary Weissmann Jan W. Hopmans William R. Horwath t

The sustainability of water resources is key to continued prosperity in the San Joaquin Valley and California. The vadose zone is an often-ignored layer of wet but unsaturated sedi- ments between the land surface and the . It plays an important role in recharge and Nearly 3,000 feet of continuous cores were obtained between July and October 1997 us- in controlling the flux and attenua- ing the Geoprobe direct-push drilling method. This method allowed for complete recovery of tion of nitrate and other potential undisturbed cores throughout the 52-foot deep vadose zone. Above, UC Davis vadose zone hydrology professor Jan Hopmans (left) and Fresno State geology undergraduate Anthony groundwater contaminants. In a Cole operate the coring equipment. former orchard at the UC Kearney Re- search and Extension Center, we in- when the Clean Water Act, the Safe refer to the unsaturated zone above the vestigated the processes that control Drinking Water Act, the Federal Insec- water table as the vadose zone. In gen- the movement of water, nitrate and ticide, Fungicide, and Rodenticide Act eral, only the uppermost 4 to 6 feet (1 other contaminants through the deep and other legislation related to water to 2 meters) is described in soil surveys and investigated in soil studies; the vadose zone. These processes were pollution were enacted. Since then, countless efforts have been mounted deep vadose zone below the root zone found to be controlled by the alluvial by both the scientific-technical commu- remains largely outside the area of re- sedimentary geology of the vadose nity and the agricultural industry to bet- search and regulatory activity. zone, which is highly heterogeneous. ter understand the role of agricultural Yet, the vadose zone below the This heterogeneity should be consid- practices in determining the fate of fer- root zone stores significant moisture. ered when interpreting soil and deep tilizer and pesticides in watersheds (in- All water and contaminant transport vadose zone monitoring data and cluding groundwater) and to improve from the land surface to groundwater assessing of the leaching potential of agricultural management accordingly. passes downward through the vadose Much of the scientific work related to zone. Few studies have investigated agricultural chemicals. The transport subsurface nitrate and pesticide leach- the fate or potential fate of, for ex- of contaminants through the vadose ing has focused on two areas: docu- ample, nitrate and other contaminants zone may be significantly faster than menting the extent of contamination in in such deep vadose zones. Key ques- previously assumed, while denitrifi- groundwater; and investigating the fate tions include, What is the time of cation is likely limited or insignificant of these chemicals in the soil root zone travel through the deep vadose zone? in the oxic, alluvial vadose zone of (including the potential for groundwa- And, is there significant denitrification the eastern San Joaquin Valley. ter leaching) as it relates to particular (natural attenuation) of nitrate in the agricultural crops and management deep vadose zone? practices. Rarely are these two research Pioneering work on nitrate in deep or decades, the leaching of agricul- areas linked within a single study. soil profiles was presented by Pratt et al. tural chemicals (fertilizer, especially In California’s valleys and basins, (1972) who investigated nitrate profiles Fnitrate, and pesticides) has been a particularly in Central and Southern in a Southern California citrus orchard concern of agronomists, soil scientists California, groundwater levels are fre- to depths of 100 feet (30 meters); they and hydrologists. Federal legislation quently much deeper than 10 feet (3 estimated that it would take between 10 first recognized the potential impacts meters) and sometimes as deep as 150 and 50 years for nitrate to leach to that to water resources in the early 1970s, feet (45 meters) or more. Hydrologists depth. Average nitrate-nitrogen levels DRAFT124 CALIFORNIA AGRICULTURE, VOLUME 59, NUMBER 2 Preferential flow paths, which are responsible for most of the water and solute transport from the root zone to the water table, quickly flush nitrogen to deeper portions of the vadose zone and to the water table, allowing for little or no denitrification.

below the root zone varied from 15 to 35 subplots (0, 100 and 325 pounds per in the three subplots. The analysis milligrams per liter (mg/L) under a acre) were selected in 1997 (2 years showed that the annual nitrogen leach- 50 pounds per acre (lb/ac) treatment after termination of the experiment), ing of excess nitrogen fertilizer from and from 35 to 55 mg/L under an exces- which we refer to as the control, stan- the root zone into the deep vadose sive 350 lb/ac treatment. Based on gross dard and high subplot, respectively. zone was 51 (± 19), 83 (± 30) and 245 mass-balance estimates, denitrification To start, we conducted a conventional (± 42) pounds per acre. Water losses at that site was estimated to account for root-zone nitrogen (N) mass-balance from the root zone to the deep vadose up to 50% of nitrate losses in the thick analysis from application and harvest zone in the flood-irrigated orchard unsaturated zone profile where applica- records for the 12-year experiment amounted to 1,100 (± 180) millimeters tion rates were high. Lund et al. (1974), supported later by ‘Fantasia’ nectarine Gilliam et al. (1978), Klein and Bradford orchard at the UC Kearney Research and (1979) and Rees et al. (1995), argued that Extension Center, the nitrate losses in the deep vadose zone experimental site of (due to denitrification) were strongly this study (the orchard correlated with the textural properties was removed in 1998). of the soil. High losses were found in with pans or textural discontinui- ties, while losses were limited in rela- tively homogeneous, well-draining soils in other areas of Southern California. In contrast, Rolston et al. (1996), using isotope analysis at sites in the south- ern Sacramento Valley and the Salinas Valley, found little evidence of signifi- cant denitrification, even in thick un- saturated zones. To study deep unsaturated zone hy- drology, we established a research site in a former ‘Fantasia’ nectarine orchard at the UC Kearney Research and Extension Center (KREC) in Fresno County. The objective of our work is to provide a comprehensive assessment of the fate of nitrate in a 52-foot (16-meter) deep allu- vial vadose zone that is typical of many agricultural areas in California. The assessment included detailed geologic, hydraulic and geochemical characteriza- tion, using nitrate as an example. Field sampling A 12-year fertilizer management experiment (from 1982 to 1995) was implemented in a ‘Fantasia’ nectarine orchard (Johnson et al. 1995). The fer- tilization experiment consisted of five application treatments in a random block design with triple replicates. Treatments included annual nitrogen application rates ranging from 0 to 325 Fig. 1. In 1997, extensive core drilling was conducted, 2 years after the completion of three pounds per acre (0 to 365 kilograms per orchard nitrate-management trials at KREC with annual fertilizer rates of 0, 100 and 325 hectare [kg/ha])(fig. 1). For the vadose pounds nitrogen per acre. The complete random block design of the management trial and zone characterization, three treatment the three subplots selected for the deep vadose zone drilling are shown. DRAFThttp://CaliforniaAgriculture.ucop.edu • APRIL–JUNE 2005 125 Fig. 3. The 10 major lithofacies identified at the two east-west cross sections, 140 and 144 feet from the southern edge of the orchard. The lithofacies, classified in the field according to color, texture and cementation, exhibit vertically varying thicknesses, yet are laterally Fig. 2. Schematic diagram of the Kings River alluvial fan and continuous over the experimental site. Sandy loam is the most fre- its geologic elements. quent textural unit, while clay is the least.

per year (Onsoy et al. 2005). Assuming kilometers) west of the current river KEY (figs. 3, 4): The major lithofacies are: uniform flow conditions throughout channel. The quaternary alluvial sedi- the deep vadose zone at an average ments (that is, sediments deposited by SL1 — recent Hanford sandy loam content of 25%, the travel a stream) are derived exclusively from C — clay, very thin Var1 — variable sedimentary structures, time through the deep vadose zone the hard, crystalline Sierran bedrock. predominantly sand was projected to be 3.2 (± 0.5) years. Stratigraphically, the quaternary de- HP1 — shallow paleosol (hardpan), red Based on the leaching rate and travel posits in this part of the valley can Var2 — various textures, sandy loam time, and taking into account that the be divided into five units (Marchand to clay loam experiment ended 1 year prior to drill- and Allwardt 1981): the post-Modesto S — medium sand ing, we estimate that the deep vadose (youngest), Modesto, Riverbank, Upper C-Si-L — clay/clayey silt/clay loam, zone nitrogen storage at the time of and Lower Turlock Lake deposits. fine-textured floodplain deposits drilling would be on the order of 195, Except for the post-Modesto, which is SL2 — sandy loam and 233 and 426 pounds nitrogen per acre less than 10,000 years old (Holocene), HP2 — deep clayey paleosol (hardpan), red in the control, standard and high sub- these deposits are of Pleistocene age plot, respectively. (2 million to 10,000 years old). To confirm this estimate and to de- Most of the stratigraphic units ranges from clay to small gravel and termine the applicability of the uniform (sediment facies) found at the site are includes a wide spectrum of predomi- flow concept, 60 undisturbed sediment believed to represent separate alluvial nantly silty to sandy sediments. The col- cores were obtained in 1997 by drilling episodes related to several Sierran gla- ors of the sediments range from grayish to the water table at a depth of 52 feet ciations. In cores from the study site, brown to yellowish brown, and more (16 meters) using a Geoprobe direct- these deposits appear as intercalated, randomly to strong brown (no signifi- push drilling technique. After geologic thick and thin lenses of clayey silt, silt, cant reduction zones). The thickness of characterization of the complete core sand and gravel from fluvial deposi- individual beds varies from less than sections, 1,200 samples were collected tion. Channel sediments consist of 0.4 inch (1 centimeter) for some finely (approximately one every 2.5 feet [0.8 moderately to well-sorted, subangular layered clayey floodplain deposits to meters]). Samples were collected for to subrounded sand and gravel. These more than 8 feet (2.5 meters) for sandy each sedimentologic stratum or substra- channel deposits are surrounded by streambed deposits. Sharp as well as tum. The soil samples were preserved muddy sand and silts of floodplain gradual vertical transitions are present and stored for later analysis of their deposits (fig. 2)(Page and LeBlanc 1969; between texturally different units. The texture, hydraulic properties (water Huntington 1980; Weissmann et al. relative proportion of the five major tex- content, unsaturated hydraulic conduc- 2002). Deposits from the various periods tural categories found in the sediment tivity and water retention functions) of Sierran glaciations are vertically sepa- cores was 17.2% sand, 47.8% sandy and biochemical properties (pH, dis- rated by paleosols. Paleosols are buried loam, 13.8% silt loam/loam, 8.3% clay solved organic carbon, nitrate-nitrogen, soil horizons that were formed on stable loam/clay and 12.9% paleosol (see side- 15N isotope analysis). upper-fan or terrace surfaces during bars, pages 128 and 129; fig. 4). interglacial periods, when no sediment Geologic framework Hydraulic properties variability deposition took place (fig. 3). The site is located on the Kings River The vadose zone sediments are most Hydraulic properties of the unsatu- alluvial fan, approximately 2 miles (3.2 easily classified by their texture, which rated zone, such as the hydraulic con- DRAFT126 CALIFORNIA AGRICULTURE, VOLUME 59, NUMBER 2 Fig. 4. Individual core sections were collected in 4-foot (1.2 meter) plastic tubes (see fig. 1). This image shows all 13 sections of a 52-foot (15.6-meter) core, lined up from the ground surface (upper left) to the bottom of the core (lower right). Missing subsections represent soil sampling locations for hydraulic and chemical analysis.

much hydraulic variability within indi- vidual facies as between facies. By far the highest conductivity was observed in sandy facies (“S” and “Var1” in figs. 3 and 4). Heterogeneity of water flow Traditionally, water flow between the root zone (at depths of 6.6 feet ([0 to 2 meters]) and the water table (here a depth of 52.8 feet [16 meters]) has been considered essentially a uniform, vertically downward flow process in a more or less homogeneous vadose zone. Within this conceptual framework, wa- ter from individual rainfall or irrigation events is thought to be initially stored in the root zone. There, it is available for uptake by the roots. Surplus water then gradually drains into the deeper vadose zone. Individual rainfall or irrigation events create pulses of moisture that pen- etrate the root zone profile. Through root water uptake and vertical spreading, the moisture pulse dampens out as it travels downward. In the deeper portions of the ductivity/moisture curve and the water depths at the orchard site. The labora- vadose zone, the downward flow rate retention curve — and the spatial dis- tory tests involved measuring the water has therefore been thought to be equal to tribution of these properties — strongly percolation rates in each core at six to the annual recharge rate. control the flow of water and the trans- 10 different moisture conditions, then However, the highly heterogeneous port of nitrate and other solutes in the determining the hydraulic properties by geology of the alluvial sediments ob- deep vadose zone. (Hydraulic conduc- computer analysis. served at the orchard site, coupled with tivity is a measure of how fast water can Given the large amount of textural the associated heterogeneity of the percolate through the sediments; the variability observed in the cores, it hydraulic properties, suggest that this higher the moisture and the coarser the was not surprising that we found the traditional conceptual framework is sediments, the higher the hydraulic con- hydraulic properties to also vary sig- inadequate to describe how water and ductivity.) Hydraulic properties were nificantly, both with depth and laterally chemicals are transported through the determined in the laboratory on more across the site (Minasny et al. 2003). The vadose zone to the water table. Using than 100 undisturbed sediment core saturated , for our field data and computer simula- samples (3.5 inches [9 centimeters] long example, varied over nearly four orders tion, we reconstructed two-dimensional by 1.5 inches [3.8 centimeters] diam- of magnitude. Within some sedimentary cross-sections of the vadose zone that eter) taken from various locations and layers (facies) we observed nearly as reasonably reflect the spatial variability DRAFThttp://CaliforniaAgriculture.ucop.edu • APRIL–JUNE 2005 127 Major facies in the study-site vadose zone

The sand (“S” and “Var1”) is quartz- ish gray in color. The bed thickness rich, and contains feldspar, muscovite, is within a range of a few inches to of the hydraulic properties observed in biotite, hornblende and lithic fragments a foot (centimeters to decimeters). the field, then we simulated water flow consistent with the granitic Sierran Fine-grained sediments often show through this reconstructed vadose zone. source (see figs. 3 and 4). Cross-bedding sharp contacts between the units. Figure 5 illustrates the spatial distribu- at the scale of few inches (centimeters) Changes from one unit to the next ex- tion of the water flux through a hypo- could be observed in some fine-grained ist on small distances. Lamination can thetical cross-section at 40 feet, similar sand samples. The dominant color of more frequently be observed within to that observed at the orchard site. It the sand is a light gray to light brown, silty sediments than in fine sands. captures important features that are char- the brown hue increasing with increas- Root traces and rusty brown–colored acteristic of water flow in heterogeneous ing loam content. The thickness of the mottles are quite common. The depo- vadose zones (Russo et al. 1998; Harter sand beds is as much as 8 feet (2.4 me- sitional environment was presumably and Yeh 1996). Understanding that flow ters), though thickness varies across the the proximal to distal floodplain of occurs in this highly irregular pattern is study site. Very coarse sand and par- the fluvial fan, an area dissected by important for interpreting the field data. ticles up to pebble grain-size (up to 0.4 distributary streams. In particular, figure 5 shows that inch or 1 centimeter) could occasionally The finest sediments are grouped even though the simulated water ap- be observed at the bottom of the sand in the fourth category: silt, clay and plication at the surface was uniform, the units, but were not present in all the clay loam (portions of “C,” “C-Si- heterogeneity of the vadose zone forces cores. The sand units typically show a L” and “Var2”). These are believed water into distinct preferential flow subtle fining-upward succession. The to have been deposited in the distal paths (warm colors), separated by large basal contact is typically sharp. The floodplain and in ponds that devel- areas of relatively stagnant flow (cooler texture and distribution of these sandy oped in abandoned channels. The colors). The preferential flow paths have deposits are consistent with deposition main color is brownish gray to olive highly irregular shapes, but are continu- in a fluvial distributary channel on the brown. Fine, less than 1-millimeter- ous and extend to the water table. The Kings River fluvial fan. One ancient diameter root traces and rusty brown preferential flow paths occupy only a river channel was observed in cores mottles are common in the clay sedi- small portion of the vadose zone. In collected from the orchard site, and it ments. Statistics for the thickness of contrast, areas with relatively stagnant appears to have had a northeast-south- clay layers in the unit between 27 water flow occupy most of the vadose west orientation. The mean thickness and 43 feet (8 and 13 meters) depth zone. Soil moisture differences between of this channel deposit is nearly 6 feet show a mean thickness of 5 inches these two zones are small. Soil moisture (1.7 meters). The basal coarse sand and (12.8 centimeters), but the mode is is therefore not a particularly sensitive pebbles probably represent channel lag about 2.2 inches (3 centimeters). A 20- indicator and may not be useful for deposits that were laid down in deeper inch (50-centimeters) thick clay bed identifying the presence and location of parts of the channels. was observed at approximately 8 feet these preferential flow paths. Sandy loam (“SL1,” “SL2”) is the (2.5 meters) depth in most of the cores. In our study, further field work and most frequent lithofacies within the Paleosols (“HP1,” “HP2”) were computer simulations indicate that the profile. The color is usually light olive recognized in different stages of ma- location of these preferential flow paths to yellowish brown. Some of the sandy turity. They show a brown to strong does not change over time, even though loam sediments are considered to be brown, slightly reddish color, exhibit they may partially or completely dry weakly developed paleosols because aggregates, ferric nodules and con- out between events. A new of their stronger brownish color, root cretions, few calcareous nodules and event will recreate the same set of pref- traces and presence of aggregates. hard, cemented layers. They also erential flow paths. Mean bed thickness is 20 inches (50 display a sharp upper and a gradual Preferential flow paths are not only centimeters), though individual beds lower boundary as is typical for pa- created by the sediment heterogene- can be as much as 7 feet (2 meters) leosols (Retallack 1990). Clay content ity. Other conditions may also trigger thick. The sorting is moderate to good. decreases downward in the paleosols. and support preferential flow paths: Clay flasers and thin (fractions of an Another feature is fine root traces, macropores in the root zone; flow in- inch, 0.5 to 1 centimeter) clay laminae though these are typically obliter- stability during infiltration into sand or occur in some sandy loam units. Sandy ated in the more mature paleosols. loamy sand soils (Wang et al. 2003); flow loam sediments are assumed to have Paleosols formed in periods of stasis through or above embedded clay or sand developed at the edge of channels, as marked by nonerosion and nonde- lenses (“funneling”)(Kung 1990); and levee or as proximal floodplain depos- position, during the interglacials “fingering” as a result of a sharp textural its near the channels. (Weissmann et al. 2002). The thickness boundary within the vadose zone where Silt loam, loam and silty clay loam of the paleosol horizons ranges from the finer-textured (silt/clay) layer is (portions of “C-Si-L” and “Var2”) are 20 inches (50 centimeters) to about located above a coarse-textured (sand) usually slight olive brown to brown- 7 feet (2 meters). layer (Glass and Yarrington 2003). DRAFT128128 CALIFORNIA CALIFORNIA AGRICULTURE, AGRICULTURE, VOLUME VOLUME 59 ,59 NUMBER, NUMBER 2 2 Study-site geologic profile

A number of distinct sedimentologic units are recognized in the vadose zone profile throughout the orchard and are used to construct a field-scale geologic framework for the research site (see figs. 3 and 4). The deepest parts of the cores (be- tween 50 and 52 feet [15 and 15.8 meters]) display a strong brownish-colored, clay- rich paleosol. This paleosol marks the top of the upper Turlock Lake deposits (Weissmann 2002). Directly above this pa- leosol, from depths between 40 and 50 feet (12 to 15 meters) below the surface, the Fig. 5. Hypothetical, computer-simulated, unsaturated water flow rates through a hetero- geneous vadose zone, illustrating how heterogeneity in the unsaturated zone generates main textural units are sandy loam to fine preferential flow paths: Warmer colors (red, orange, yellow) represent high flow rates, and sandy loam, with some coarse sand and cooler colors (blue, violet, black) represent lower flow rates. The unsaturated hydraulic gravel or fine-grained sediments. In the properties used for the simulation are similar to those found at the KREC research site. cores with fine sediment at the bottom of this unit, a coarsening-upward succession Since much of the annual recharge Agency under the Safe Drinking Water was observed in this zone; in the other occurs through these preferential flow Act. More than half of those occurred cores a fining-upward cycle was observed. paths, the actual downward flow rate in the subplot with the highest nitro- Between 27 and 40 feet (8 and 12 meters) is locally much higher than that esti- gen application. Mean nitrate-nitrogen depth, the sediments are vertically and lat- mated when assuming that the entire concentrations (not including nonde- erally quite heterogeneous with relatively vadose zone uniformly participated in tects) in the control, standard and high thin bedding (thickness of a few inches the downward water flow. Hence, sol- subplot vadose zone were estimated to [centimeters] to a couple of feet [few deci- ute travel times through deep vadose be 5.2, 3.3 and 7.4 mg/L, respectively meters]), consisting mainly of clayey, silty zones are likely shorter than if flow (Onsoy et al. 2005). The nitrate-nitrogen and loamy material. Another strong brown- were indeed uniform. Because of the coefficient of variation (CV) ranged ish paleosol occurs at a depth of 30 to 33 feet shorter travel time, nitrogen storage in from 1.6 to 2.4 within each subplot. The (9 to 10 meters). Between 20 and 30 feet the deep vadose zone should be signifi- difference between the control and stan- (6 and 9 meters) below the surface, a distinct cantly lower than estimated based on dard subplots was statistically not sig- sand layer, representing a former stream the uniform flow concept. On the other nificant due to the large variability. But channel bed, is found. This unit has laterally hand, under heterogeneous conditions, the high subplot yielded significantly varying thickness averaging nearly 6 feet relatively old water may be trapped in larger mean nitrate-nitrogen concentra- (1.7 meters). A weak, mostly eroded paleo- the more stagnant portions of the va- tions throughout the profile, consistent sol was developed on top of the sand unit. dose zone for extended periods. Do the with the overapplication of fertilizer. From about 10 to 13 feet (3 to 4 meters) measured nitrate distribution and water Within all three subplots, slightly to 20 feet (6 meters) below the surface, chemistry in the deep vadose zone at higher nitrate-nitrogen levels were ob- sandy loam with intercalated sand, clayey the Kearney site support this alternative served in the root zone than in the deep and silty material is found. Different trends conceptual framework of water and sol- vadose zone below the root zone, pos- of upward-fining and upward-coarsening ute flow through the vadose zone? sibly due to the last fertilizer application are found on top of each other and laterally in fall 1996, prior to our drilling. Other next to each other within this unit. Nitrogen distribution than that, we observed no significant Immediately above the unit, at a depth To address this question, nitrate- vertical nitrate-nitrogen trend in the of about 10 to 13 feet (3 to 4 meters), a nitrogen (NO3-N) concentrations were deep vadose zone. nearly foot thick (0.2 meter) to more than measured in 809 subsamples of our The highest number of nondetects 10 feet (1 meter) thick paleosol hardpan cores. We found that the data were in- occurred in the coarse-textured, sandy occurs. This paleosol marks the top of the deed highly variable and log-normally lithofacies and in the sand lithofacies Riverbank formation. Recent ground- distributed (the logarithms of the data (above historic water level). There, ap- penetrating radar surveys indicate that were normally distributed). Nitrate- proximately half of the samples had non- this paleosol is laterally extensive across nitrogen concentrations ranged from detectable nitrate-nitrogen levels. This is the orchard site. less than the detection limit of 0.05 mil- consistent with findings that fingering Sandy loam and subordinated loamy ligram per liter (mg/L)(224 samples or preferential flow is particularly domi- sand and loam are present from the top of or 28% of the sample population) to nant in sandy unsaturated sediments. the paleosol to the surface, and represent more than 100 mg/L (two samples). In other words, much of the sand facies the Modesto deposits at the site. About 8 Approximately 10% of the samples does not participate in the flow (stagnant feet (2.5 meters) below the surface, a later- exceeded the maximum contamination moisture) and would see little or none ally continuous clay horizon with a thick- level for drinking water (10 mg/L), set of the nitrate-nitrogen that is passing ness of few inches (centimeters) is found by the U.S. Environmental Protection through the preferential flow paths. in most of the cores. DRAFThttp://CaliforniaAgriculture.ucop.edu • APRIL-JUNEAPRIL–JUNE 20052005 129129 Most importantly, the total nitrogen mass estimated directly from the mea- sured nitrate-nitrogen distribution in the deep vadose zone was only 46 (± 9), 37 (± 6) and 83 (± 12) pounds per acre annually for the control, standard and high subplots, respectively. This is less than one-quarter of the total nitrogen mass in the deep vadose zone that was indirectly estimated from the mass bal- ance described above, which is based on the conventional uniform flow concept. Role of denitrification Traditionally, such low nitrate-nitro- Fig. 6. Composite nitrate-nitrogen, dissolved organic carbon (DOC) and δ15N profiles from gen mass in the deep vadose zone has borings at the orchard study site. Higher DOC indicates a higher potential for denitrification. been attributed to denitrification (the Higher δ15N levels (above 10) indicate the occurrence of partial denitrification. microbial breakdown process of nitrate by soil microbes). However, the lack of a significant vertical trend in the aver- of -5‰ to 5‰ (Rolston et al. 1996). transport in heterogeneous soils as well age nitrate-nitrogen concentration does Denitrification decreases nitrate- as that predicted by numerical models not support that hypothesis (denitrifi- nitrogen concentrations, but increases of flow and transport in highly hetero- cation in the deep vadose zone would the relative amount of isotopically geneous vadose zones (Harter and Yeh 15 create a nitrate-nitrogen profile that heavy NO3- δ N. Indeed, there was a 1996). Hence, the proposed conceptual shows decreasing concentration with very weak trend (R2 < 0.1) to support framework of preferential flow in the depth). To further evaluate whether the hypothesis that a limited amount of deep vadose zone may provide the denitrification played a significant role denitrification may occur in the vadose answer: preferential flow paths, which in the deep vadose zone, we measured zone: samples with 2 mg/L to 10 mg/L are responsible for most of the water the amount of soluble organic carbon nitrate-nitrogen had relative δ15N lev- and solute transport from the root zone (a microbial food source) and the els of 5‰ to 12‰, while samples with to the water table, quickly flushed amount of δ15N (a rare nitrogen isotope higher nitrate-nitrogen contained from nitrate-nitrogen to deeper portions of that increases in relative abundance 0‰ to 6‰ of δ15N. (No δ15N measure- the vadose zone and to the water table, when denitrification occurs) in the ments could be made on samples with allowing for little or no denitrification. nitrate-nitrogen of samples from four less than 2 mg/L nitrate-nitrogen.) This would explain the occurrence of cores (three from the high subplot and There was significant scatter in high nitrate-nitrogen levels and low one from the standard subplot). (The these data, corroborating the concept δ15N levels throughout the depth of the concentration of isotopes is not report- of highly heterogeneous transport. vadose zone profile. ed in absolute values, but rather as a Furthermore, just as there was no sig- Lower nitrate-nitrogen would occur relative concentration, hence the nota- nificant decrease in nitrate-nitrogen in stagnant water zones outside pref- tion “delta” or “δ” 15N. A value of δ15N with depth, there was no significant erential flow paths. Due to their longer = 5‰ indicates that the 15N concentra- increase in δ15N with depth, similar to residence time, nitrate-nitrogen in these tion is 5 permil above normal [1 permil isotopic results at geologically similar zones was apparently subject to a small = 0.1 percent].) Yolo County and Salinas Valley sites amount of denitrification, which would Soluble carbon was found to be very (Rolston et al. 1996). Then why was the explain the higher levels of δ15N that low, not favoring high rates of microbial total nitrogen storage in the deep va- were also found, scattered throughout degradation anywhere in the vadose dose zone so low? most of the profile. zone profile. The δ15N values varied The nitrate distribution pattern The importance of this finding is from 0‰ to 12‰ and averaged 6‰ that we found (fig. 7) was similar to that while limited denitrification took (fig. 6). Without denitrification, δ15N that postulated in other experimental place in the stagnant water areas of levels are expected to be in the range studies specifically designed to assess the vadose zone, the majority of the DRAFT130 CALIFORNIA AGRICULTURE, VOLUME 59, NUMBER 2 Fig. 7. Measured core nitrate-nitrogen data (squares) and three-dimensional, kriged (yel- low) contours of nitrate-nitrogen data. The

kriged isosurfaces (only for NO3-N > 1 mg/L) are obtained from geostatistical analyses of the and nitrate measurements at the sampling locations. The standard sub- plot yielded the largest areas of negligible

nitrate concentrations (NO3-N < 1 mg/L) among the three subplots. High concentra- tions are seen in the upper profile and near the water table.

nitrate-nitrogen transport occurred in preferential flow paths, where no significant denitrification appears to have taken place. Hence, the low aver- age nitrate-nitrogen concentration in the vadose zone pore water should not be interpreted as an indicator for high denitrification and low nitrate impact on groundwater. Rather, it may be the result of swift, unattenuated nitrate- nitrogen transport to the water table deep vadose zone below the root zone water and is not effectively represented (the same may apply to the root zone). may be subject to significant preferential by composite soil samples. It appears that flow patterns with significantly faster the average or composite nitrate-nitrogen Analysis shows variability travel times than would be estimated concentrations are also not appropriate Our detailed geologic, hydraulic and under uniform flow assumptions. for estimating the amount of denitrifica- geochemical analysis of a typical deep Faster travel times not only decrease tion as a closure term to the nitrogen mass vadose zone in the eastern San Joaquin the potential for denitrification, but also balance. We are currently implementing Valley demonstrated that alluvial va- decrease the potential for natural at- detailed heterogeneous flow and trans- dose zones are subject to significant geo- tenuation of pesticides. port simulations to further support these logic variability, which in turn causes Our work suggests that the traditional findings and to develop guidelines for the hydraulic properties and water flow interpretation of deep vadose zone mea- sampling the deep vadose zone. in the vadose zone to exhibit strong surements should be reconsidered. The spatial variability. While such variability assumption of uniform flow is not appli- is expressed to only a limited degree in cable to many alluvial vadose zone sites. the variability of the observed moisture The common practice of compositing soil T. Harter is Associate Groundwater Quality content, it leads to highly variable con- samples taken from immediately below Hydrologist (UC Cooperative Extension), centrations of chemicals, such as nitrate. the root zone provides an estimate of the Department of Land, Air and Water Re- Our research presents new evidence average nitrate-nitrogen concentration sources, UC Davis; Y.S. Onsoy is consul- indicating that unsaturated water flow at that depth. However, our work indi- tant and Doctoral Candidate, UC Davis; and transport of nitrate and other agro- cates that recharge water may constitute K. Heeren is freelance geologist in environ- chemicals (such as pesticides) in the only a minor portion of that vadose zone mental education, Germany; M. Denton is DRAFThttp://CaliforniaAgriculture.ucop.edu • APRIL–JUNE 2005 131 Before it is pumped to the surface for ag- ricultural or other uses, groundwater per- colates through the geologically variable vadose zone. Based on this study, current assumptions about transport of nitrate from fertilizers and other agricultural chemicals need to be reexamined.

2003. Neural networks prediction of soil consultant and obtained a master’s degree References hydraulic functions from multi-step outflow in hydrologic studies, Department of Gilliam JW, Dasberg S, Lund LJ, Focht DD. data. Soil Sci Soc Amer 68:417–29. Land, Air and Water Resources, UC Da- 1978. Denitrification in four California soils: Onsoy YS, Harter T, Ginn T, Horwath WR. vis; G. Weissmann is Assistant Professor Effect of soil profile characteristics. Soil Sci 2005. Spatial variability and transport of Soc Amer 42:61–6. nitrate in a deep alluvial unsaturated zone. of , Department of Geologi- Glass RJ, Yarrington L. 2003. Mechanistic Vadose Zone J 4(1):41–54. cal Sciences, Michigan State University; modeling of fingering, nonmonotonicity, Page RW, LeBlanc RA. 1969. Geology, hy- and J.W. Hopmans is Professor of Vadose fragmentation, and pulsation within grav- drology and water quality in the Fresno area, ity/buoyant destabilized two-phase/unsatu- California. USGS Open-File Rep. 70 p. Zone Hydrology, and W.R. Horwath is rated flow. Water Resour Res 39(3):1058, Pratt PF, Jones WW, Hunsaker VE. 1972. Associate Professor of Soil Biogeochemis- DOI:10.1029/2002WR001542. Nitrate in deep soil profiles in relation to try, Department of Land, Air and Water Harter T, Yeh T-CJ. 1996. Stochastic analy- fertilizer rates and leaching volume. Environ Resources, UC Davis. sis of solute transport in heterogeneous, Qual 1:97–102. variably saturated soils. Water Resour Res Rees TF, Bright DJ, Fay RG, et al. 1995. The California Fertilizer Research and 32(6):1585–95. Geohydrology, water quality, and nitrogen Education Program, California Tree Fruit Huntington GL. 1980. Soil-land form rela- geochemistry in the saturated and unsatu- Agreement, UC Water Resources Center, tionships of portions of the San Joaquin River rated zones beneath various land uses, River- Geoprobe Systems and Studienstiftung and Kings River alluvial depositional systems side and San Bernardino Counties, California, in the Great Valley of California. Ph.D. disser- 1991–93. USGS Water Resour Invest Rep des Deutschen Volkes provided support for tation, UC Davis. 147 p. 94–4127. 267 p. this project. We would like to thank Kevin Johnson RS, Mitchell FG, Crisosto CH. Retallack GJ. 1990. Soils of the Past — An Pope, formerly with Geoprobe Systems, Jean 1995. Nitrogen fertilization of Fantasia nec- Introduction to Paleopedology. Hyman U tarine: A 12-year study. UC Kearney Tree Fruit (ed.). Winchester, MA. 520 p. Chevalier with his crew at KREC, and Dick Rev 1:14–9. Rolston D, Fogg GE, Decker DL, et al. Rice and R. Scott Johnson at the UC Kear- Klein JM, Bradford WL. 1979. Distribution 1996. Nitrogen isotope ratios identify nitrate ney Agricultural Center for their support of nitrate and related nitrogen species in the contamination sources. Cal Ag 50(2):32–6. during the field work; Anthony Cole, Chad unsaturated zone, Redlands and vicinity, San Russo D, Zaidel J, Laufer A. 1998. Nu- Bernardino County, California. USGS Water merical analysis of flow and transport in a Pyatt and Rigo Rios for drilling the 3,000 Resourc Invest Rep 79–60. 81 p. three-dimensional partially saturated hetero- feet of vadose zone cores at the site; Andrea Kung K-JS. 1990. Preferential flow in a geneous soil. Water Resour Res 34(6):1451–6. DeLisle and Timothy Doane for their ana- sandy vadose zone: 1. Field observations. Wang Z, Wu L, Harter T, et al. 2003. A Geoderma 46:51–8. field study of unstable preferential flow dur- lytical chemistry work; Jim MacIntyre for Lund LJ, Adriano DC, Pratt PF. 1974. Ni- ing soil water redistribution. Water Resour implementation of the multistep outflow trate concentrations in deep soil cores as Res 39(4), 10.1029/2001WR000903, 01 April experiments needed to determine the hy- related to soil profile characteristics. Environ 2003. draulic properties at the site; and Atac Tuli Qual 3:78–82. Weissmann GS, Mount JF, Fogg GE. 2002. Marchand DE, Allwardt A. 1981. Late Ce- Glacially-driven cycles in accumulation space for his support in developing the necessary nozoic stratigraphic units, northeastern San and sequence stratigraphy of a stream- software to use in conjunction with the mul- Joaquin Valley, California. USGS Bull 1470. dominated alluvial fan, San Joaquin Valley, tistep outflow experiments. Minasny B, Hopmans JW, Harter T, et al. California. J Sediment Res 72:270–81. DRAFT132 CALIFORNIA AGRICULTURE, VOLUME 59, NUMBER 2