The confluence of higher amounts of C and NO3 – moving into a reduced zone could be the reason that the matrix surrounding the preferential flow channel has higher denitrification rates, while the regions further away from the preferential flow channel have lower amounts of microbially available C and NO3 – . In contrast, residence times are too short in the channel to allow for reducing conditions to develop. The ability of the entire vadose zone to denitrify would depend on the overall surface area of preferential flow paths to the rest of the surrounding matrix in the zone of flooding. Overall, we find that low permeability zones alone or embedded within high flow zones demonstrate highest denitrification rates across all soil profiles.Because the ERT column more closely approximates the heterogenity of our agricultural field site, we use this column to demonstrate the impact of hydraulic loading and application frequency on nitrogen fate and dynamics. Simulated profiles of liquid saturation, NO3 – , NO3 – :Cland acetate for the simplified ERT stratigraphy for scenarios S2 and S3 are shown in Figure 9 and A3. It is interesting to note that AgMAR ponding under scenarios S2 and S3 resulted in fully saturated conditions to persist within the root zone only. In comparison, the 68 cm all-at-once application for scenario S1 resulted in fully saturated conditions to occur at even greater depths of 235 cm-bgs . This resulted in the NO3 – front moving deeper into the subsurface to depths of 450 cm-bgs under S1 compared to 150 cm-bgs for scenarios S2 and S3 . Much lower concentrations of NO3 – were found at 450 cm-bgs in scenarios S2 and S3 compared to S1 .
Thus,large pots plastic larger amounts of water applied all-at-once led to NO3 – being transported faster and deeper into the profile. Surprisingly, model results indicate 37% of NO3 – was denitrified with scenario S1, while 34% and 32% of NO3 – was denitrified in scenarios S2 and S3, respectively. For scenarios S2 and S3, denitrification was estimated to occur only within the root zone. This was confirmed by NO3 – :Clratio that did not show any reduction with depth for these scenarios. A reason for this could be that acetate was not estimated to occur below the root zone, preventing electron donors from reaching greater depths for denitrification to occur. In contrast, model results for S1 indicate that acetate was leached down to 235 cm-bgs below the limiting layer. Overall, model results indicate that NO3 – did not move as fast or as deep in scenarios S2 or S3; however, the ability of the vadose zone to denitrify was reduced when the hydraulic loading was decreased. The main reason for this was that breaking the application into smaller hydraulic loadings resulted in O2concentrations to recover to background atmospheric conditions faster than the larger all at-once application in scenario S1. In fact, the O2 concentration differed slightly between S2 and S3. Because O2 inhibits denitrification, we conclude that these conditions resulted in the different denitrification capacity across application frequency and duration. In summary, we find that larger amounts of water applied all-at-once increased the denitrification capacity of the vadose zone while incremental application of water did not. However, NO3 – movement to deeper depths was slower under S2 and S3.Because initial saturation conditions impact nitrogen leaching, we also simulated the impact of wetter antecedent moisture with 15% higher saturation levels than the base case simulation for the ERT profile. Simulated profiles of liquid saturation, NO3 – , NO3 – :Cland acetate for the simplified ERT stratigraphy under wetter conditions are shown in Figure 10. Model results demonstrate that the water front moved faster and deeper into the soil profile under initially wetter conditions for all three scenarios.
Within the shallow vadose zone , across AgMAR scenarios, O2 concentrations were similar initially, but began differing at early simulated times, with lower O2 under wetter antecedent moisture conditions than with the base-case simulation. In addition, both oxygen and nitrate concentrations showed significant spatial variation across the modeled column. Notably, nitrate concentrations were 166% higher in the preferential flow channel compared to the sandy loam matrix under wetter conditions, while only 161% difference was observed under the base case simulation . Nitrate movement followed a pattern similar to water flow, with NO3 – reaching greater depths with the wetter antecedent moisture conditions. Under S1, however, at 150 cm-bgs, NO3 – decreased more quickly under the wetter antecedent moisture conditions due to biochemical reduction of NO3 – , as evidenced by the decrease in NO3 – :Clratio, as well as by dilution of the incoming floodwater. In the wetter antecedent moisture conditions, 39%, 31%, and 30% of NO3 – was denitrified under S1, S2, and S3, respectively. For S1, where water was applied all at once, more denitrification occurred in the wetter antecedent moisture conditions, however, the same was not true of S2 and S3 where water applications were broken up over time. This could be due to the hysteresis effect of subsequent applications of water occurring at higher initial moisture contents, allowing the NO3 – to move faster and deeper into the profile without the longer residence times needed for denitrification to occur. Thus, wetter antecedent moisture conditions prime the system for increased denitrification capacity when water is applied all at once and sufficient reducing conditions are reached, however, this is counteracted by faster movement of NO3 – into the vadose zone. Simluations from our study demonstrate that low-permeability zones such as silt loams allow for reducing conditions to develop, thereby leading to higher denitrification in these sediments as compared to high permeability zones such as sandy loams. In fact, the homogenous silt loam profile reported the maximum amount of denitrification occurring across all five stratigraphic configurations .
Furthermore, the presence of a silt loam channel in a dominant sandy loam column increased the capacity of the column to denitrify by 2%. Conversely, adding a sandy loam channel into a silt loam matrix decreased the capacity of the column to denitrify by 2%. These relatively simple heterogeneities exemplify how hot spots in the vadose zone can have a small but accumulating effect on denitrification capacity . Note that differences in denitrification capacity maybe much greater than reported here because of increased complexity and heterogeneity of actual field sites when compared to our simplified modeling domains. Another observation of interest for silty loams is the prominence of chemolithoautotrophic reactions and Fe cycling observed in these sediments. In comparison, sandy loam sediments showed persistence and transport of NO3 – to greater depths. A reason for this is that oxygen concentration was much more dynamic in sandy loams, rebounding to oxic conditions more readily than in silt loams, even deep into the vadose zone . Dutta et al. found similar re-aeration patterns in a 1 m column experiment in a sand dominated soil,square planter pots with re-aeration occurring quickly once drying commenced. Even with the presence of a limiting layer, defined by lower pore gas velocities and higher carbon concentration, a sandy loam channel acted as a conduit of O2 into the deep vadose zone maintaining a relatively oxic state and thus decreasing the ability of the vadose zone to denitrify. In systems with higher DOC loadings to the subsurface, oxygen consumption may proceed at higher rates creating sub-oxic conditions in the recharge water and more readily create reducing conditions favorable to denitrification in the subsurface . We note here that microbial growth, which was not modeled in this study, could also affect the rates of O2 consumption and re-aeration, which could lead to underestimation of O2 consumption . Overall, denitrification capacity across different lithologies was shown to depend on the tight coupling between transport, biotic reactions as well as the cycling of Fe and S through chemolithoautotrophic pathways. Under large hydraulic loadings , overall denitrification was estimated to be the greatest as compared to the lower hydraulic loading scenarios . The main reason for the higher denitrification capacity was the significant decline in O2 concentration estimated for this scenario, whereas such conditions could not be maintained below one meter with lower hydraulic loadings under scenarios S2 and S3. However, nitrate was also transported deeper into the column under S1 as compared to S2 or S3. Tomasek et al. found the reverse in a floodplain setting, where intermittent indundation with flood water, comparable to our S2 and S3 contexts, resulted in higher rates of denitrification in the zone that was always inundated, due to priming of the microbial community and pulse releases of substrates and electron donors. Future studies examining the impact of AgMAR on denitrification should include processes such as mineralization to see if the same behavior would be observed. It seems that there may exist a threshold hydraulic loading and frequency of application that could result in anoxic conditions and therefore promote denitrification within the vadose zone for different stratigraphic configurations, although this was not further explored in this study.
In another study, Schmidt et al. found a threshold infiltration rate of 0.7 m d-1 for a three hectare recharge pond located in the Pajaro Valley of central coastal California, such that no denitrification occurred when this threshold was reached. For our simulations, we used a fixed, average infiltration rate of 0.17 cm hr-1 for our all-at-once and incremental AgMAR scenarios, however, application rates can be expected to be more varied under natural field settings. Our results further indicate that the all-at-once higher hydraulic loading, in addition to causing increased levels of saturation and decrease in O2, resulted in leaching of DOC to greater depths in comparison to lower, incremental hydraulic loading scenarios . Akhavan et al. 2013 found similar results for an infiltration basin wherein 1.4% higher DOC levels were reported at depths down to 4 m when hydraulic loading was increased. Because organic carbon is typically limited to top 1 m in soils , leached DOC that has not been microbially processed could be an important source of electron donors for denitrification at depth. Systems that are already rich in DOC within the subsurface are likely to be more effective in denitrifying, and thus attenuating, NO3 – , such as floodplains, reactive barriers in MAR settings, or potentially, organically managed agroecosystems .. This finding can also be exploited in agricultural soils by using cover crop and other management practices that increase soluble carbon at depth and therefore remove residual N from the vadose zone . While lower denitrification capacity was estimated for scenarios S2 and S3, an advantage of incremental application was that NO3 – concentration was not transported to greater depths. Thus, higher NO3 – concentration was confined to the root zone. If NO3 – under these scenarios stays closer to the surface, where microbial biomass is higher, and where roots, especially in deep rooted perennial systems such as almonds, can access it, it could ultimately lead to less NO3 – lost to groundwater. While there is potential for redistribution of this NO3 – via wetting and drying cycles, future modeling studies should explore multi-year AgMAR management strategies combined with root dynamics to understand N cycling and loading to groundwater under long-term AgMAR. Simulation results indicate that wetter antecedent moisture conditions promote water and NO3 – to move deeper into the domain compared to the drier base case simulation. This finding has been noted previously in the literature, however, disagreement exists on the magnitude and extent to which antecedent moisture conditions affect water and solute movement and is highly dependent on vadose zone characteristics. For example, in systems dominated by macropore flow, higher antecedent soil moisture increased the depth to which water and solutes were transported . In a soil with textural contrast, where hydraulic conductivity between the topsoil and subsoil decreases sharply, drier antecedent moisture conditions caused water to move faster and deeper into the profile compared to wetter antecedent moisture conditions . In our system, where a low-permeability layer lies above a high permeability layer , the reverse trend was observed. Thus, a tight coupling of stratigraphic heterogeneity and antecedent moisture conditions interact to affect both NO3 – transport and cycling in the vadose zone, which should be considered while designing AgMAR management strategies to reduce NO3 – contamination of groundwater. Furthermore, dry and wet cycles affect other aspects of the N cycle that were not included in this study .