Paper No: 200000

Introduction

An automated system for on-the-go mapping of soil pH has been developed and tested under field conditions (Adamchuk et al., 1999). While traveling across the field, a soil sampling mechanism (Figure 1) located in a toolbar-mounted shank scoops 5-10 g of soil from approximately 10 cm depth and brings it into firm contact with the sensitive membranes of two ion-selective electrodes. After the reading stabilizes (typically 5-15 s), a new sample is obtained and the electrode surfaces are rinsed. Every measurement is related to the GPS location of the sample within the field.

Ion-Selective Electrodes Holder

Gauge Wheel

Water Nozzles

Sampling Platform

Figure 1: Soil-sampling mechanism with two ion-selective electrodes

An agro-economic analysis of automated soil pH mapping has shown that higher resolution maps can significantly decrease estimation errors and result in higher potential profitability of variable rate liming. A comparison with commonly used 1 ha (2.5 acre) grid point sampling has shown that automated mapping resulted in $6.13/ha higher net return over the cost of liming during a four-year growing cycle in a corn-soybeans rotation (Adamchuk et al., 2001). The ability to obtain high-resolution maps of soil nitrate and potassium levels at the time of pH mapping could further expand the potential for economic and environmental impacts of site-specific crop management.

Soil Residual Nitrogen

Nitrogen is the major essential soil nutrient for plant growth. Crops take up most nitrogen in mineral form as ammonium (NH₄⁺) or nitrate (NO₃^-) ions. Nitrogen deficiency is the most common soil fertility problem for grain crops. Insufficient N availability results in declining chlorophyll level in leaves and, thus, inadequate absorption of sun energy (Blackmer, 2000). This results in significant yield decrease.

Also, nitrogen losses from soils negatively impact both water and air quality. “However, despite the importance of N in agricultural production and environment quality, a widely accepted method to test soils for plant-available N, particularly in humid regions, has not been developed. The reasons for this center around the complex N cycle transformation” (Sims, 2000).

Pre-plant testing of soil nitrate is used to predict crop available nitrogen in many states of the Great Plains region. In Nebraska, for example, residual nitrate in the soil profile is a key component of the N recommendation algorithm for corn (Shapiro et. al., 2001):

(1)

where Nitrate = recommended nitrogen fertilizer application (lb/acre)

EY = expected yield, determined as 105% of 5-year average (bu/acre)

NO₃^--N = root zone soil residual nitrate-N in 60-120 cm (2-4 ft) depth (ppm)

OM = soil organic matter (%)

NCR = other nitrogen credits

Nitrate carryover can exist due to excess fertilizer use, decreased crop growth because of climatic factors; or, in low-rainfall areas, low potential for nitrate loss through leaching and denitrification. In general, the pre-plant soil nitrate test consists of measuring NO₃^--N in some portion of the crop root zone, and the using obtained values to credit against the nitrogen fertilizer recommendation (Gelderman and Beegle, 1998).

Spatial variation in soil properties has been used in past attempts of variable rate fertilizer-N application. However, implementation of variable rate N management has given mixed results when compared with the conventional uniform treatment. Studies with irrigated maize in Nebraska found no economic or environmental benefits of variable rate N application over a homogeneous N rate (Ferguson et al., 2002). In this approach, variable N rates were prescribed based on spatial variation in soil organic matter and soil nitrate as obtained by destructive soil sampling and interpolation. Better simulation models, in-season application and higher quality thematic soil maps are believed to improve both the economic and environmental benefits of variable rate nitrogen management (Doerge, 2001). In addition, reliable ways must be found to spatially vary the yield goal within a field because most N recommendation algorithms are highly sensitive to this component.

Currently, residual soil nitrate maps are produced through soil sampling and chemical analysis, which consists of extraction and an analytical method. Deionized water, different ionic strength adjustment solutions or 2 M KCl are typically used to extract nitrate ions. Then either an ion-selective electrode or a colorimetric method, such as cadmium reduction, is used to compare the NO₃^- ion concentration in the extract with that in standard solutions. Other methods, such as steam-distillation or micro-diffusion, have also been used in the past (Mulvaney, 1996).

While the cadmium reduction procedure is commonly used in soil analysis laboratories because of its sensitivity, robustness, and automatic implementation of in flow injection analyzers (FIA), ion-selective electrode methods can also be used according to the following standard procedure (Gelderman and Beegle, 1998):

1. Measure 20 g of soil into a 100 mL cylindrical container.

2. Add 50 mL of extracting solution.

3. Shake for 5 minutes on a reciprocal shaker.

4. Read the potential while the suspension is being stirred with magnetic stirrer.

5. Record the millivolt reading (if using a calibration curve technique) or read the NO₃^--N concentration directly from pH/ion meter.

Several modified procedures were attempted for measuring residual nitrate on-the-go. Adsett et al. (1999) developed a soil sampler with a NO₃^- monitoring system that utilized a conveying unit to deliver soil to an extraction unit where the measurement took place. They indicated that 95% accuracy could be obtained within 6 s. However, the authors did not report on the correlation between field measurements and the corresponding laboratory analysis.

Artigas et al. (2001) used pH, Ca²⁺, K⁺and NO₃^- ion-sensitive field effect transistor (ISFET) based sensors to measure ion concentration in soil by manually inserting the electrode in pots with loam and peat soil. Even though they concluded that the ISFET sensors were feasible, a relatively poor correlation was obtained in their reported data comparing NO₃^- measurements with those obtained using spectrophotometry.

Hummel and Birrell (1995) indicated that ISFET sensors in combination with flow injection analysis represent a robust alternative for measuring nitrate ion concentrations in manually extracted solutions (r² > 0.9). Price et al. (2000) described their attempt to develop a nitrate extraction system that could use ISFET technology to map soil nitrate on-the-go. No field testing results have been reported yet.

The possibility of rapidly sensing soil mineral-N content using near infrared (NIR) reflectance was also investigated (Ehsani et al., 1999). Simulation studies were conducted to determine the ability of Partial Least Squares (PLS) and Principal Components Regression (PCR) techniques to relate NIR spectral data to soil nitrate content in the presence of interfering effects and experimental noise. The simulation studies revealed that both PLS and PCR techniques were quite robust in predicting soil nitrate content provided the calibration set included the same interfering effects. These techniques failed completely if the prediction set contained interfering effects that were not included in the calibration set. This implies that a site-specific calibration is necessary for this technique to work successfully. Laboratory tests using Yolo loam and Capay clay soil samples as well as verification tests using field soils (Yolo loam and Capay clay) mixed with nitrogen fertilizer indicated that soil mineral-N content can be determined using the NIR technique provided site-specific calibration is used.

Using ion-selective electrodes to measure NO₃^- ion concentration directly on a naturally moist soil is an alternative to the research previously addressed in this paper. However, every on-the-go mapping concept limits the ability to analyze deep soil samples when a minimum of 61 cm (2 ft) long deep sample is required to obtain a representative value for a fertilizer-N recommendation (Shapiro et al., 2001). On the other hand, data presented by Goderya et al. (1996) suggest that the majority of the spatial structure of residual nitrate content exists in the upper layer of soil (<30 cm deep). Moreover, in well-managed systems with high nitrogen use efficiency, levels of residual nitrate in the subsoil tend to be small and vary little from year to year despite much larger fluctuations in the upper 10 to 30 cm of soil that may be caused by seasonal variation in soil N mineralization, crop N removal, and fertilizer use. Therefore, maps of topsoil residual nitrates may be used as an indirect measures of residual nitrate variability and combined with other spatial data such as yield potential, soil organic matter, and soil N mineralization indices to improve variable rate nitrogen management. For example, Nemeth et al. (1993) measured residual soil NO₃ at 0-30 cm depth with electroultrafiltration (EUF) after the summer wheat harvest. They successfully predicted nitrate concentration in the whole 0-90 cm profile in the fall through multiple regression.

Plant Available Soil Potassium

Potassium is absorbed by plant roots from the soil solution as K⁺ ion. Therefore, plant K uptake is closely related to the soil solution concentration and the ability of soil to replenish this concentration as plant roots deplete it. Potassium is associated with many enzymes involved in photosynthesis, organic compound synthesis, and translocation of organic compounds (Foth and Ellis, 1988). When K is limiting, characteristic deficiency symptoms appear in the plants (white spots on the leaf edges, weakening of straw, etc.). As a result, crop performance may suffer (Havlin et al., 1999).

Soil potassium exists in four forms: solution, exchangeable, fixed (non-exchangeable), and structural (mineral) K. There are equilibrium and kinetic reactions between these four forms that affect the level of soil solution potassium at any particular time, and thus, the amount of readily available K for plants (Helmke and Sparks, 1996). Various studies have shown that only 7% or less of the total K crop requirement is in the immediate vicinity of the roots (McLean and Watson, 1985). The potassium requirement must be met by some mechanism that moves additional K⁺ from the exchangeable form to the root. The quantities of potassium cations extracted in most soil test procedures are simply referred to as exchangeable K. The potassium level in soil is important for determining the appropriate rates of supplemental K (Warncke and Brown, 1998).

The current standard procedure in the North-Central region (Warncke and Brown, 1998) for measuring exchangeable K involves six steps:

1. Scoop 2 g of soil (dry, ground and sieved through a 2 mm mesh).

2. Add 20 ml (the amount can be different as long as 1:10 ratio is maintained) of extracting solution (1M NH₄OAc at pH 7.0 or Mehlich-3).

3. Shake for 5 minutes on the shaker at 200 excursions/min.

4. Filter the suspension through Whatman No. 2 filter paper.

5. Set up the atomic absorption/emission spectrometer for K by emission. Determine the standard curve using the standards and obtain the concentration of K in soil extracts.

6. To convert K concentration (ppm) in the soil extract solution to ppm in soil, multiply by 10.

In general, extracts of soils can be analyzed for K⁺ using atomic absorption spectroscopy (AAS), inductively coupled plasma-mass spectroscopy (ICP-MS), flame emission spectroscopy, ion chromatography, or ion-selective electrodes. These techniques require liquid samples. Solid soil samples can be analyzed by neutron activation analysis (NAA) or x-ray fluorescence (XRF) (Helmke and Sparks, 1996). In another approach, electroultrafiltration (EUF), soil K release is measured at different temperatures and voltages applied to soil suspensions. This method is widely used in the sugar beet industry in Europe (Nemeth, 1979).

Nair and Talibudeen (1973) used an ion-selective electrode to measure K⁺ activity in the root zone of winter wheat. They used a 1M NaCl salt bridge and KCl solution standards. Also, 2:1 soil/water ratio and 30 sec equilibrium time were found to be satisfactory. The electrode readings were compared to flame photometry measurements used as a reference (from reported data R² = 0.99).

Wang et al. (1988) used ion-selective electrodes (ISE) to study potassium quantity-intensity relationships. They found that ISEs offered a simple and rapid alternative to AAS. They detected compatible (r ³ 0.999) potassium related characteristics measured in three Iowa soils.

Wang and Huang (1990) studied the feasibility of using a potassium ion-selective electrode to monitor changes in K⁺ concentration in soil suspensions over time. They described factors affecting the efficiency of the ISE method including the electrode response time, influence of suspended soil particles, shaking speed, and ionic strength of the system.

Farrell and Scott (1987) compared ion-selective electrodes and atomic absorption spectroscopy measurements (R² = 0.97). The exchangeable K in 30 samples was extracted with neutral solutions of 1M NH₄OAc and 0.5M BaCl₂. They listed the opportunity for automated analysis as one of the advantages of the ISE method over AAS.

The references listed suggest certain benefits from the use of potassium ISEs instead of other, more complicated, analytical procedures. However, automated measurement performed on naturally moist soil samples omits the extraction of exchangeable potassium and results in the ability to detect only soluble K⁺. This measure may not be adequate for reliable recommendation of fertilizer-K application rates. Therefore, additional studies to relate the need for potash fertilizer to the activity of K⁺ in specific soils are necessary. Most likely, an on-the-go sensed map of soil solution K must be integrated with other measurements, collected at a lower spatial density (soil texture, soil exchangeable and non-exchangeable K, CEC), to reliably predict plant K needs.

Objectives

The goal of this work was to evaluate the feasibility of measuring both K⁺ and NO₃^- ion activities on naturally moist soil using ion-selective electrodes. Stability, repeatability and reliability of polymer membrane, combination, flat-surface, ion-selective electrodes are the main focus of the current research.

Materials and Methods

Polymer membrane ion-selective electrodes consist of different ion exchange materials in an inert matrix such as PVC, polyethylene or silicone rubber. The membrane is sealed to the end of a PVC tube. The potential developed at the membrane surface is related to the concentration of the ions of interest. Such electrodes are usually used to measure activities of ions such as K⁺, Ca⁺, Na⁺, NH₄⁺, Cl^-, or NO₃^- in different media.

A successful attempt at automated mapping of chemical soil properties on-the-go means that measurements performed in the field should be repeatable, correlated with commonly used laboratory methods, and take only a few seconds to perform while moving across the field. According to the automated soil pH mapping (Adamchuk et al., 2002), field measurements were performed directly on naturally moist soils (Direct Soil Measurement – DSM). The measurements obtained were compared to standard laboratory procedures involving preparation of 1:1 soil/water solution (Soil Solution Measurement – SSM).

When adopting the same measurement concept to map soil nitrate and potassium content, five primary issues must be addressed:

· Stability – minimum time required for obtaining a stable sensor output

· Calibration – equation that relates measured quantity and electrical output

· Repeatability – error level due to test replications

· Accuracy – correlation between the tested and referenced measurement method

· Reliability – operational lifetime of a single probe

Four nitrate ISEs (PVC-membrane, combination, flat-surface, epoxy body, jell filled) and one potassium ISE (liquid filled) have been tested in laboratory conditions using two sets of air-dried soils from various locations in Nebraska (nitrate electrodes) and Indiana (potassium electrode). Nitrate electrodes included: NO31 NO31505-003B (pHoenix Electrode Co., Houston, TX), directION 360-75/GEL (Sentek Limited, Essex, UK), directION 3021BN (Nico Scientific, Inc., Huntingdon Valley, PA), and A0NOELECTRODER-CIX (EID Corporation, Bridgeport, CT). The potassium electrode was IS-K001502 (LAZAR Research Laboratories, Inc., Los Angeles, CA).

Stability of nitrate electrodes was evaluated by recording electrode response every second for 60 s starting at the time of contact between soil or calibration standard solution and the surface of electrodes. Every electrode was rinsed with distilled water between measurements.

Calibration was performed using three standard solutions with known concentrations of nitrate or potassium ions. Nitrate standards were prepared at concentration of 3, 30, and 300 ppm of NO₃^- using NO3AS02 1000 ppm standard and NO3IS01 ionic strength adjustment solutions (pHoenix Electrode Co., Houston, TX). Other concentrations (1, 10 and 100 ppm) and sources (1.371 g/L NaNO₃, 1.629 g/L KNO₃, and 1.274 g/L NH₃NO₃) have been tested as well. In every case the difference in electrode output caused by the source of calibration standard was found to be lower then the replication error. The potassium electrode was calibrated using standard KCl solutions (1, 10, and 100 K⁺ ppm). Ionic strength adjuster, 5M NaCl, was added to keep a constant background ionic strength.

The equation to predict nitrate-nitrogen concentrations in soil samples using ion-selective electrodes was:

(2)

Therefore the following calibration equation was used to determine the slope and the intercept:

(3)

where mV = electrode output (mV)

NO₃^- = nitrate ion concentration (ppm)

NO₃^--N = nitrate-nitrogen concentration (ppm)

mV_0pN = electrode output (intercept) at 1 ppm of NO₃^- (pN = log(NO₃^-) = 0)

mV_pN = mV output increase (slope) due to increase of NO₃^- by 10 (pN = 1)

Similarly, the following prediction and calibration equations were used to measure potassium ion concentrations (activity):

(4)

(5)

where K⁺ = concentration of potassium ions (ppm)

mV_0pK = electrode output at 1 ppm (pK = log(K+) = 0)

mV_pK = mV output increase due to increase of K+ by 10 (pK = 1)

Repeatability of electrode response in both calibration standards and soils was calculated as mean squared error of electrode output due to measurement replication during a single experiment. Each calibration and test was reproduced at least three times. Change of electrode’s performance with time was not addressed in this research. However, the results obtained suggested a change in both calibration parameters from test to test.

Accuracy of direct soil measurement was addressed when comparing results against reference measurements. A set of 15 Nebraska soils was analyzed using the cadmium reduction procedure (Maynard and Kalra, 1993) in three commercial soil laboratories. A standard error of approximately 3% was observed when comparing reports from these labs. The measurements obtained by the Soil and Plant Analysis Laboratory (University of Nebraska-Lincoln, Lincoln, NE) were used as the reference.

A set of 24 soil samples obtained from various locations in Indiana was previously analyzed using five different procedures with atomic absorption spectroscopy (AAS) (Soil Analysis Laboratory, Purdue University). The concentration of potassium ions was measured using 1M NH₄OAc extraction, standard laboratory practice described by Warncke and Brown (1998), or 0.01 N CaCl₂ (Cassman et al., 1990) extractions. The measurements were also performed in 1:1 weight-to-weight soil/water solution, saturated soil paste (aqueous methods modified from Rhoades (1982)), and in liquid obtained using the column displacement method (modified from Adams (1974)). Even though an R² of 0.54 was found between ion concentrations in NH₄OAc extracts (standard method) and in 1:1 soil/water solutions measured with AAS, it appears that the correlation between these two methods may be affected by differences in texture and organic matter content. Therefore, the AAS measurement performed on 1:1 soil/water solution were used as the reference for ISE tests.

All four nitrate ion-selective electrodes were simultaneously connected to a PMU-8PH data logger (LAZAR Research Laboratories, Inc., Los Angeles, CA). The potassium electrode was used with an Accumet^® model 25 pH/Ion meter (Denver Instrument Company, Arvada, CO). Both units were used in mV output mode.

Direct soil measurements (ISE-DSM) were performed on samples prepared by adding 4 ml of distilled water to 20 g of air-dried and sieved (2 mm mesh) soil (during nitrate ISEs test), and 5 g of soil and 2 ml of deionized water (during potassium ISE test). Therefore, target gravimetric moisture content was 0.2 g/g. The actual gravimetric soil moisture measured after the test ranged from 25 to 33 g/g. Soil solution measurements (ISE-SSM) were also conducted in 1:1 soil/water solutions the same way as ISE-DSM.

Reliability of an ion-selective electrode can be defined as the number of samples or the storage time prior to a significant decrease in the electrode’s slope, which makes the sensor incapable of distinguishing between calibration standards. However, some manufacturers suggest that only a 2 mV potential deviation from theoretical slope (approximately 57 mV) over time should be expected (at constant temperature); a much greater reduction was observed upon contact between soil and the sensor. No formal tests of reliability were conducted during this study.

Results and Discussion

Nitrate ISEs Testing

In order to monitor the electrode response and to define proper stability criteria, mV output from four selected electrodes was recorded at the rate of one reading per second. Figure 2 shows two common responses obtained from a calibration standard (left) and a soil sample (right).

Figure 2: Nitrate ISE step response in a calibration standard (left) and in a soil sample (right)

It appears that rather than a classic first order response, the biggest change occurred approximately 5 s after electrode-soil/solution contact. A relatively small drift while approaching a steady value usually followed this sudden “jump”. The final reading differed for each electrode as the result of the differences in the electric potential.

Figure 3 shows the reduction in measurement error (MSE between current value and the average of five last measurements) and instability (standard deviation of five consecutive measurements). These data were pooled from 15 measurements of calibration standards (top) and 45 measurements of soil samples (0.2 g/g target gravimetric moisture content) (bottom).

Figure 3: Measurement error (left) and instability (right) as a function of response time while placing electrode in contact with standard solutions (top) and soil samples (bottom)

A response time of 15 s was used throughout this study, as results showed that the majority of the response occurred before 10 s (sooner for standard solutions). It appears that approximately 10 mV error should be expected due to the short measurement time. This results in over- or underestimating the steady-state value by approximately 0.2 log(NO₃^-), a prediction range of approximately –37% to +58% of the [B1] measured concentration. This error is one of the major factors that affect both repeatability and accuracy of individual measurements.

The repeatability of electrode responses at 15 s was compared in Table 1. The mean squared error was calculated as the square root of the average variance calculated for three replications. It appears that Probe 4 could reproduce its previous measurement best with a mean squared error of 0.21 log(NO₃^-) for ISE-DSM and 0.07 log(NO₃^-) for ISE-SSM method.

Table 1. Parameters characterizing the repeatability of nitrate ISEs 15 s response

Parameter	Probe 1	Probe 2	Probe 3	Probe 4	Probe 1	Probe 2	Probe 3	Probe 4
Parameter	ISE-DSM Test				ISE-SSM Test
	Calibration Standards
MSE, mV	7.04	6.73	6.09	7.90	9.05	9.16	6.05	4.35
MSE, log(NO₃^-)	0.19	0.19	0.13	0.16	0.32	0.27	0.13	0.10
	Soil Samples
MSE, mV	11.39	10.67	13.47	10.27	3.96	6.90	3.81	3.11
MSE, log(NO₃^-)	0.31	0.31	0.28	0.21	0.14	0.20	0.08