Soil type, crop and irrigation technique affect nitrogen leaching to groundwater
California Agriculture 67(4):231-241. https://doi.org/10.3733/ca.E.v067n04p231
Published online October 01, 2013
Many groundwater resource in California are degraded by high concentrations of nitrate, most of which was transported to the groundwater in water percolating below the root zone of agricultural fields. Factors that affect the rate of water percolation — including soil type, crop and irrigation — along with nitrogen application influence the probability of this type of groundwater degradation. UC scientists have developed several useful tools, including the Nitrogen Hazard Index (NHI) and the ENVIRO-GRO (E-G) model, for use in developing best management practices (BMPs) to achieve high crop yields while minimizing groundwater degradation. We report the results of E-G simulations that quantify the effects of irrigation, soil type and organic and inorganic nitrogen (N) application amounts to corn yield and the amount of leached N. Simulation results indicate that a nitrate management strategy that also includes water management will be more effective in reducing N loading to groundwater. The research findings are discussed in the context of the track and report concept in comparison to the BMP approach.
The downward percolation of nitrate-laden water from agricultural fields is a major contributor to the high levels of the contaminant found in many California groundwater resources (Viers et al. 2012). Many assume that this condition results from the excessive application of nitrogen (N) fertilizer to crops.
The word “excessive” can have any of several connotations, and because the term usually is not clearly defined in this context it can be taken by any number of people to mean any number of things. Excessive application could mean that more N is applied to the soil than can be removed by the crop, and there is no question that most agricultural applications could be included in this definition. Another definition would say that excessive application means that more fertilizer is applied than would be required to achieve high yields and maximum profits.
The ENVIRO-GRO model simulates the consequences of irrigation water salinity and management practices on crop yield and nitrate leaching. Simulations indicate that strategies to minimize groundwater degradation must also include water management practices to be effective.
High yields and maximum profits almost always require the application of more N to the soil than is removed by the crop. Whether growers have historically applied more N than was necessary to obtain maximum profits is not clear and probably cannot be determined.
Other management factors (e.g., irrigation) have a great impact on the relationships between the amount of fertilizer applied, the crop yield, and the deep percolation of nitrate. Strategies that are intended to reduce nitrate degradation of groundwater but that ignore complex dynamic relationships with other management factors are likely to fail.
Nitrate reaches groundwater only by being transported by water that percolates through the soil, a factor often disregarded when assessing the relationship between fertilizer application and nitrate degradation of groundwater. Every crop requires sufficient water to meet its evapotranspiration (ET) needs, and any irrigation or precipitation that exceeds the soil's water-holding capacity in the root zone will cause soluble chemicals, including nitrate, to leach into deeper groundwater. The amount of N that is leached varies with time and with the amount of water flow and the N concentration in the soil water at the time leaching occurs.
The rate of N uptake by a crop varies with its growth stage and, in cases of N deficiency, may also depend on the N concentration in the soil water. Total plant dry matter production usually has a linear relationship to ET. Therefore, if plant growth is reduced because there is too little water, too much salinity, or too little N, the plants will have less dry matter production and less ET, which means that any given irrigation regime will result in more leaching (Pang and Letey 1999).
Both positive and negative feedback loops between plant growth and soil condition can be observed, depending on circumstances. For example, if salinity in the soil reduces plant growth, the reduction in plant growth will reduce ET, resulting in greater leaching of salts, which will to some degree remedy the salinity problem. This is a positive feedback mechanism. However, if both plant growth and ET are reduced by a lack of adequate N or by other factors, leaching of nitrate will increase, further reducing the nitrate content of the root zone, thereby intensifying the problem. One consequence of this negative feedback mechanism is that any attempt to decrease nitrate leaching by reducing N applications may be counterproductive if the reduced N input further reduces plant growth, which would in turn increase N leaching. Groundwater degradation by nitrate is related both to time-dependent fertilizer and to water management.
N results from grower field studies
Based on extensive field research during the 1970s, it is fair to say that the optimal (i.e., profit-maximizing) amount for N application is dictated by the amount of precipitation and irrigation. That research focused on a total of 55 fields drained by tile systems and 31 naturally drained fields that did not have a shallow water table (Letey et al. 1977, 1979). By looking at the rate at which water discharged into the tile systems and the nitrate concentration of water samples collected in the tile systems, researchers were able to calculate how much nitrate in total was discharged into the tile systems. For the natural drainage studies, researchers drilled into the soil and analyzed samples from various depths, usually reaching to a depth of 50 feet. Procedures were then developed to calculate the rate of water flow through the soil profile. This water flow rate, multiplied by the nitrate concentration, provided an estimate of the nitrate leached below the root zone. Researchers obtained information on fertilizer application from the growers.
Results were similar for both systems: the correlation coefficient between the amount of N leached and the drainage volume was greater than the coefficient for the amount of N applied. This suggests that irrigation management is at least equal in importance to, and possibly of greater importance than, fertilizer application in affecting the leaching of nitrate. As expected, the highest correlation coefficient was between the amount of nitrate leached and a combination of drainage volume and fertilizer application, indicating that both factors are important.
Importantly, there was no significant correlation between the nitrate concentration of the drainage water and either the amount of fertilizer applied or the drainage volume. The linear regression analysis for all the tile systems resulted in the equation
where C is the average nitrate-nitrogen concentration (mg/L) and N is the amount of fertilizer N applied (kg/ha) (Letey et al. 1977). Usually only the concentration is measured and there is no measurement of the water flow, making it impossible to calculate the discharge load. By itself, the numerical value of the concentration is of little value, and it may even lead one to make erroneous conclusions.
Growers are very observant when it comes to crop behavior. They may not know the amount of drainage volume from a given field, but they will know the crop yield. Our researchers hypothesized that growers would be likely to apply more N to fields that were N-deficient because the fields had a high drainage volume. Indeed, the experimental data supported this hypothesis. A linear regression analysis for naturally drained fields resulted in the equation
where N is the fertilizer applied (kg/ha/yr) and W is the amount of drainage water (cm/yr). The tile drain systems yielded
Greater drainage flows, therefore, induced growers to increase their N applications.
From the past to the future
If we simply assume that the large quantities of nitrate that migrated to groundwater decades ago were the exclusive result of excessive N applications, we may not be correct. The cause is just as likely to be related to irrigation management as to fertilizer management. Irrigation at the time in question was almost entirely applied as gravity flow rather than through pressurized irrigation systems. With gravity flow, the irrigator has little control over how much water infiltrates the soil, because the infiltration opportunity time (the amount of time when water is flowing over the soil) within the furrow and the hydraulic properties of the soil can vary so much. Pressurized irrigation systems allow more precise control over the amount and uniformity of water application and partially negate the effects of some soil properties, such as infiltration rate.
Another reason growers might purposely apply excess water is that they might be concerned that they could salinate the soil. Historical accounts of growers salinating soils in irrigated, semi-arid regions of the world are well known. Growers were educated about the need to leach salts from the root zone, and they considered this when setting up their irrigation practices. The leaching of destructive salts, though, also leaches out beneficial nitrate. Less efficient irrigation systems and the perceived need to leach salts contributed to high leaching of N and the resultant requirement for additional N application.
One reason to conclude that growers apply more N than is required for high crop yield is the common belief that growers typically apply more N than is recommended by universities and other research organizations. However, because those recommendations are commonly based on research done on small plots with carefully controlled irrigation, they may not apply so readily to the real-world conditions in many growers' fields. According to the results reported above, growers do tend to apply more N on a field that has a higher drainage volume. This supports the conclusion that growers do base their N applications at least partly on their field observations on yield.
Many growers and researchers may not have considered that converting gravity flow systems to pressurized systems provides an opportunity to reduce deep percolation and even reduce the amount of fertilizer applied without reducing crop yield. Field observation can show where too little N has been applied, but for most crops you cannot visually detect signs of excess application.
The availability of soluble commercial N fertilizer has been cited as a cause for the high levels of nitrate that have historically reached groundwater. Some maintain that organic forms of N have less potential to migrate below the root zone than inorganic forms. As will be demonstrated later, this is not always the case, and if the cause of a problem is mis-diagnosed, the prescribed cure may not be effective.
If we were to assume, for instance, that the huge, long-term buildup of nitrate in groundwater is a result of a history of excessive N applications rather than a history of excessive water applications, we would be inclined to take poor, and possibly counterproductive, actions in an attempt to improve the situation. Regulations that attempt to reduce groundwater degradation by focusing strictly on the amount of N applied, without consideration for the interactions between the amounts and timing of both fertilizer and water applications, most likely will not achieve their desired goal. Furthermore, each individual crop, soil and irrigation technology comes with its own challenges and opportunities that must be assessed.
The Nitrogen Hazard Index (NHI) was developed by UC scientists and is available online at http://ciwr.ucanr.edu/Tools/Nitrogen_Hazard_Index/ . A farm manager who uses this online tool to input his or her crop, soil and irrigation technology will receive a report that estimates the probability that nitrate will degrade groundwater in the field. The report also ranks the relative significance of effects from the crop, the soil and the irrigation system in terms of their contribution to the overall hazard, so the grower can focus management efforts toward those factors that are doing the most harm. The website also presents guidelines for management practices that minimize degradation according to the specific crop, soil and irrigation technology.
The California State Water Resources Control Board (SWRCB) recently submitted a report to the California Legislature with recommendations that address nitrate problems in groundwater (SWRCB 2013). The report emphasized the quantity of nitrogen applied but gave little recognition to the influence of irrigation management. The report specified high-risk areas for nitrate problems, but identified those areas only on the basis of hydrogeological conditions.
Regulations that focus only on the amount of N applied without considering the interactions between the amounts and timing of fertilizer and water applications may not be successful.
The report's authors cite a map that identifies areas at high risk for groundwater contamination with MTBE (methyl tertiary-butyl ether, a now-banned gasoline additive), which reached groundwater through leakage from underground storage tanks, and go on to assume that areas vulnerable to MTBE are also vulnerable to nitrate. This assumption, however, ignores all of the dynamic interactions that occur in the root zone and control the movement of nitrate below the root zone. Only after the nitrate has migrated below the root zone can its movement be affected by the hydrogeological features that affect the movement of MTBE. If only a small amount of nitrate migrates below the root zone, the risk that significant quantities of nitrate will move through the groundwater is small. The real probability of risk is related to the crop, soil and irrigation system as assessed using the NHI, and that is the proper means for determining likely problem areas.
Farm-level management is the most effective mechanism for reducing the continued degradation of groundwater from nitrate. A more useful report to the Legislature would have focused on best management practices (BMPs) and would have provided a plan by which they would be implemented on the farm. Management factors that influence both the yield of a specific crop and N leaching include irrigation events and the amount and timing of organic or inorganic N applications. There are other significant factors, such as soil hydraulic properties and rainfall, but those cannot be specifically managed.
A major objective of this paper is to present scientific factors concerning the dynamic interactions between soil, crop and irrigation on crop yield and the leaching of nitrate.
ENVIRO-GRO (E-G), a model developed by UC scientists, simulates the consequences of various management factors on crop yield and nitrate movement below the root zone. In this paper, we use E-G to illustrate the effects of organic and inorganic N application amounts, rainfall amounts and irrigation amounts on crop yield and nitrate leaching on two soil types. The effects of soil temperature on the dynamic rate of organic matter mineralization and the implications of this on potential N leaching represent new findings. We discuss these findings as they relate to the NHI and BMP concepts as well as to the proposed track and report system.
The E-G model (Pang and Letey 1998) was developed to simulate (1) water, salt and nitrate movement through soil with a growing plant; (2) plant response to stresses associated with matric water potential, salinity and N deficiency; (3) water, salt and nitrate leaching below the root zone; (4) cumulative relative transpiration and N uptake and (5) consequent crop yields as compared with those of an unstressed crop. The E-G model does not account for denitrification or N immobilization. The model allows us to simulate the consequences of irrigation water salinity and management practices on crop yield and nitrate leaching.
The E-G model has recently been re-programmed to make it more efficient. Modifications include the addition of compensation for N uptake, a two-pool model for organic matter decay, mass balance calculations, comprehensive output routines and improvements to the transport calculations for salt and nitrate. The E-G program and user manual are available online for free at http://ciwr.ucanr.edu/Tools/ENVIRO-GRO . Running the model does require an understanding of using such models and is not useful for the general practitioner.
When you use the tool, you first input certain information: the potential ET as a function of time, the amount and timing of water addition (irrigation or precipitation), the potential N uptake of the crop as a function of time, the amount and timing of N applications, and soil and plant characteristics. The time and amount of application is sufficient for soluble inorganic N, but not for organic forms of N, since they are not immediately available for plant uptake. For organic N, the model also requires its rate of mineralization into inorganic N. One purpose of this paper is to evaluate factors, including soil temperature, that affect the dynamic rate of organic N mineralization.
Organic material mineralization
Pratt et al. (1973) proposed that one could characterize the mineralization of organic materials applied to soil in terms of a decay series, a sequence of numbers representing the fraction of the current organic N amount that can be expected to mineralize in successive years. For example, the decay series [0.40, 0.20, 0.10, 0.05] would indicate that 40% of the organic N would mineralize the first year, 20% of the remaining organic N would mineralize the second year, and so forth. The decay series is an important practical tool for estimating multiyear N mineralization for manure, compost or other organic N materials (Cusick et al. 2006).
Applications of organic N material should be timed to provide mineralized N when it will be needed by the crop, a condition that is hard to evaluate using decay series. A better choice in this case is a continuous decay function that predicts the production of plant-available nitrogen (PAN). It is this function that is required for models such as E-G that have variable time-stepping with intervals that are usually shorter than one day. The upgraded E-G model includes a two-pool decay model that is represented as
The initial organic N applied is N0 (kg/ha), which is divided into a fraction ?N0 that is assigned to a slow-decay pool and a remaining fraction (1 − ?)N0 that is assigned to a fast pool (P. Vaughan, unpublished manuscript). The decay coefficients are ?1 and ?s for the fast and slow pools, respectively. Numerical values for these coefficients and fraction can be obtained using the decay series.
The relationship between the decay series and equation 1 can be viewed as data points of the decay series and a continuous function that can be fitted to these points. Yearly remaining organic N (Nr) can be calculated from the decay series if one assumes an initial applied amount. The resulting sequence of Nr values can be extended to 10 years under the assumption that decay rates after the final year of the explicit decay series are determined exclusively by the slow pool. The presumed decay coefficient of the slow pool is 0.0101, representing the decay rate of 1% per year that is commonly accepted for soil organic matter (Meisinger et al. 2008). By taking the curve that passes through the Nr values for exclusively slow-pool decay and extrapolating it backward to the application time, one can obtain the value of ?. The remaining unknown, ?1, can be determined by curve-fitting equation 1 to all Nr values using a nonlinear least-squares algorithm.
Although mineralization is known to be a temperature-dependent reaction, the effects of temperature variations have not generally been considered in the estimation of mineralization rates. For our work, we averaged the California Irrigation Management Information System (CIMIS) soil temperature data for 2000 through 2011 at Madera, California (site #145), to obtain daily values and then fitted these data to a sine function (fig. 1). Note that there is a great difference in soil temperature between winter and summer. One would expect this temperature difference to impact the temporal rate of mineralization. Vigil and Kissel (1995) proposed an exponential function to describe mineralization rate in the temperature range of 5°C to 30°C:
where TF is the temperature factor and Ts (°C) is soil temperature. These factors were input data for calculating temperature-dependent decay rates in E-G.
Fig. 1. Daily average CIMIS soil temperature at the 15 cm depth from 2000 through 2011 at Madera, site #145.
Crop and organic material demonstration
Corn (Zea mays) was selected as the crop for demonstration because a comparison had already been made between simulated (E-G) results and actual, observed experimental cornfield results. Pang and Letey (1998) compared the simulated results from E-G with field data reported by Broadbent and Carlton (1979) that included three water application treatments and four nitrogen application amounts. The mean relative yield for all observed treatments was 0.69, and 0.64 for simulated treatments. The mean N uptake was 158 kg/ha (observed) and 159 kg/ha (simulated). The poorest agreement between observed and simulated results involved extreme irrigation treatments that would not ordinarily be applied on a working farm. The E-G simulations were also compared to a cornfield experiment in Israel that included four irrigation water salinities and four irrigation intervals, though no N data were available (Feng et al. 2003). The mean relative yields were 0.68 (observed) and 0.70 (simulated). Overall, the model has been shown to produce values that are comparable to real-world values for corn crops.
The required model input information for a cornfield is also available from a study in the San Joaquin Valley. The total N uptake was measured as a function of time for 3 years (Feng et al. 2005). Based on these data, the potential N uptake rate as a function of time was computed as 300 kg/ha total.
Ninety percent of the organic material selected for illustration mineralized in 1 year and the other 10%, in the slow pool, mineralized at a rate of 1% per year. This approximates the results that Pratt et al. (1973) reported for chicken manure, with a decay series of 0.90, 0.10, 0.05. An organic N fertilizer that is known to mineralize almost entirely in 1 year was chosen in order to avoid large carryovers of unmineralized N in successive years that would continue to accumulate and require complex multiyear simulations.
The cumulative N uptake by corn and the cumulative amount of mineralized N from an application of manure that contained 370 kg/ha of N were computed as a function of time for manure applications on Jan. 1, April 1, May 15 or Oct. 1. Only the October and April applications are represented in figure 2. The mineralization amounts illustrated are adjusted for temperature-dependent effects (TD) or presented with the assumption of constant temperature (CT). Note that an Oct. 1 application allows enough N to be mineralized before the crop period to satisfy its N requirement. However, whatever mineralized N exceeds the crop uptake is subject to leaching during that time period. Application on April 1 does not allow time for mineralization of enough N to meet crop requirements during the first year, but it may do so in following years if the N is not leached. Note that the temperature adjustment alters the time sequence for mineralization.
Fig. 2. Cumulative crop N uptake and the cumulative amount of plant-available nitrogen (PAN) production for organic material applied on April 1 or Oct. 1. The temperature is assumed constant (CT) for one set of data, and for the second set mineralization is adjusted for temperature dependence (TD) at different times of the year.
Variables for simulations
The organic material data had two application dates and variables for adjustment for temperature (TD) or no such adjustment (CT). Inorganic N was applied one time, between the preplant irrigation and planting. A clay loam soil and a sandy loam soil that differ in hydraulic properties and water-holding capacity were selected. Two ratios of uniform irrigation amount (AW) to potential ET (PET) equal to 1.1 and 1.42 were applied. These would cause expected leaching fractions for a nonstressed crop of 9% and 30%, respectively.
The first annual results are highly dependent on the initial soil conditions at the beginning of the simulation and may not accurately reflect the long-term effects of the treatment. For example, Broadbent and Carlton (1979) found that for the first year, crop N uptake on the plot that received no N application was approximately 75% of what was taken up from the plot that received the highest N application. This ratio dropped to about 25% after about 3 years of treatment. These results emphasize the importance of multiyear field experiments in terms of getting an accurate picture of treatment effects. We ran simulations for 10 consecutive years. The effects of the initial soil conditions were dissipated after the first 2 years, but only the 10-year results are reported. However, one asset of the model is that it allows the effects of changing management to be determined on an annual basis.
The crop was seeded on May 15 and harvested on Sept. 28. Irrigation was applied biweekly on the clay loam and weekly on the sandy loam because of its lower water-holding capacity. The soil profile was not recharged with water at the end of the growing season, but a sufficient amount of water to recharge the profile was applied as a preplant irrigation the next season. The time and amount of rainfall during the fallow season were those recorded at CIMIS station #145, Madera, California, during the calendar year 2006, a relatively wet year that recorded 29 cm (11.4 in) total precipitation; the 10-year average for station #145 was 22 cm (8.7 in). The individual rain event numbers are reported below, in the Results section.
We chose a range of N input amounts for each combination of variables in order to determine how much N would be required to achieve maximum yield and what the relationship was between yield and application amount. The annual amount of N leached was computed for each case. The direction (upward or downward) and rate of water flow and N concentration i