Top Navigation Bar

Article Request Page ASABE Journal Article

Water and Nitrogen Budget Dynamics for a Maize-Peanut Rotation in Florida

M. I. Zamora Re, S. Rath, M. D. Dukes, W. D. Graham


Published in Transactions of the ASABE 63(6): 2003-2020 (doi: 10.13031/trans.13916). Copyright 2020 American Society of Agricultural and Biological Engineers.


Submitted for review on 18 January 2020 as manuscript number NRES 13916; approved for publication as a Research Article and as part of the National Irrigation Symposium 2020 Collection by the Natural Resources & Environmental Systems Community of ASABE on 19 October 2020.

The authors are Maria Isabel Zamora Re, Post-Doctoral Associate, Sagarika Rath, Graduate Research Assistant, Water Institute, Michael D. Dukes, Professor, Department of Agricultural and Biological Engineering, and Wendy D. Graham, Director, Water Institute, University of Florida, Gainesville, Florida. Corresponding author: Maria Zamora Re, 1741 Museum Road, Gainesville, FL 32611-0570; phone: 352-392-1864, ext. 234; e-mail: mzamora@ufl.edu.

Highlights

Abstract. Nitrogen (N) is an essential element for crop growth and yield; however, excessive N applications not taken up by crops can result in N leaching from the root zone, increasing N loads to waterbodies and leading to a host of environmental problems. The main objective of this study was to simulate water and N balances for a maize-peanut (Zea mays L. and Arachis hypogaea L.) rotational field experiment with three irrigation treatments and three N fertilizer rates. The irrigation treatments consisted of mimicking grower irrigation practices in the region (GROW), using soil moisture sensors to schedule irrigation (SMS), and non-irrigated (NON). The N fertilizer rates were low, medium, and high (157, 247, and 336 kg N ha-1, respectively) for maize with a constant 17 kg ha-1 for all peanut treatments. DSSAT maize genetic coefficients were calibrated using the SMS-high treatment combination under the assumption of no water or N stress. The other eight treatment combinations were used as independent data for model validation of the crop coefficients. All soil hydrologic parameters were specified based on measured values, and default DSSAT peanut genetic coefficients were used with no calibration. For the irrigated treatments, DSSAT models had good performance for N uptake, biomass, and yield (average nRMSE of 8%) and moderate performance for soil water content (average nRMSE of 18%). Soil nitrate RMSE was 21% lower than the standard deviation of the observed data (5.8 vs. 7.2 mg kg-1). For the rainfed treatments, DSSAT had greater error (average nRMSE of 15% for N uptake, biomass, and yield, and average nRMSE of 31% for soil water). Soil nitrate RMSE was 11% greater than the standard deviation of the observed data (8.0 vs. 7.2 mg kg-1), and nRMSE was >30% during the crop rotation. Simulations estimated that N leaching over the crop rotation was reduced by 24% on average when using the 247 kg N ha-1 fertilizer rate compared to 336 kg N ha-1 across the irrigation treatments. Furthermore, N leaching was reduced by 37% when using SMS to schedule irrigation and the 247 kg N ha-1 fertilizer rate for maize and 17 kg N ha-1 for peanut compared to conventional practices (GROW and 336 kg N ha-1 for maize and 17 kg N ha-1 for peanut). Moreover, this management practice reduced N fertilizer use by 26% and irrigation water use by up to 60% without negative impacts on yield. Observed and simulated soil N increased during maize and peanut residue decay, with simulations estimating that this soil N would leach below the root zone during the fallow season. This leaching could potentially be reduced if a cover crop or cash crop were planted between the maize and peanut crops to take up the mineralized N.

Keywords. Agricultural best management practices, Bare fallow, BMPs, Maize-peanut rotation, N balance, N fertilization, N leaching, Sandy soils, Sensor-based irrigation scheduling, Water balance.

Materials and Methods

Experimental Site and Design

A field experiment was conducted at the North Florida Research and Education Center - Suwannee Valley (NFREC-SV), near Live Oak, Florida (30.3058979° N, -82.900987° W) to evaluate the effects of irrigation and N fertilizer rate on crop growth, N uptake, and final yields (Zamora-Re, 2019; Zamora-Re et al., 2020). Experimental data collected from a maize - bare fallow - peanut - bare fallow rotation research study (2015 to 2017) was used to calibrate and validate the CERES-Maize and CROPGRO-Peanut models within DSSAT 4.7 (Hoogenboom et al., 2019).

The field experimental design consisted of a randomized complete block arranged in a split plot with four replicates per treatment. The irrigation treatments were the main plots, and the N fertilizer rates were the sub-plots. The experimental units (plots) were 12.2 m long and 6.1 m wide, with 6.1 m alleys between the plots and 12.2 m alleys among the blocks. A two-span Valley 8000 linear irrigation system with end feed (Valmont, 2015) and a variable-rate irrigation (VRI) package was used to apply the irrigation treatments to both crops.

During the field experiment, the 2015 maize season spanned from 4 April to 18 August (planting to harvest), and the 2016 peanut season spanned from 13 May to 4 October. The field was left as bare-fallow during the intercropping periods (i.e., from 19 August 2015 to 12 May 2016 and from 5 October 2016 to 20 March 2017) following the typical maize-peanut rotation practice in the region. Maize hybrid Pioneer 1498 YHR/Bt was planted at 76.2 cm row spacing and 16.5 cm plant spacing for a total density of approximately 79,500 plants ha-1, and peanut cultivar Georgia 06G was planted at 76.2 cm row spacing and 6 cm seed spacing for a plant density of approximately 219,000 seeds ha-1. Three irrigation treatments (GROW, SMS, and NON), three maize N fertilizer treatments (low, medium, and high = 157, 247, and 336 kg N ha-1), and a constant peanut N fertilizer rate (17 kg N ha-1) were selected from the field experiment to be used in the model simulations in this study. The SMS-high plant growth data were selected to calibrate the maize crop genetic coefficients under the assumption of no water or N stress, while the other eight treatment combinations were used as independent data for crop coefficient validation. A full explanation of the data collection and treatments is provided by Zamora-Re et al. (2019), and a brief summary is provided in the following sections.

Irrigation Treatments

Growers’ Irrigation Practices (GROW)

The GROW treatment mimicked growers’ irrigation practices for the region. Information from local growers was collected from extension agents and the Suwannee River Water Management District to develop this irrigation strategy. The individual irrigation events were 10 mm, and the target irrigation rates varied based on growth stage:

A similar calendar-based irrigation schedule was applied during the 2016 peanut growing season:

Soil Moisture Sensor-Based Irrigation Scheduling (SMS)

Sentek drill and drop probes (Sentek, 2003), consisting of capacitance probes with nine sensors placed every 10 cm from 5 to 85 cm, were used to monitor volumetric water content (VWC). Probes were installed in the row between two plants in three of the four blocks (B2, B3, and B4) of the field experiment. Irrigation was triggered at a maximum allowable depletion (MAD) of 50% of the difference between field capacity (FC) and permanent wilting point (PWP). Irrigation volume was estimated to refill the active root depth with irrigation according to guidelines proposed by Zotarelli et al. (2013).

The active root depth varied as the crop was growing, and the total VWC was adjusted based on root development through the season (i.e., sum of VWC from sensors in the most active root zone). During the maize growing season, three root zones were used: 30 cm (i.e., initial vegetative growth stages ~V3 to V6), 40 cm (i.e., peak of growth in vegetative stages ~V7 to VT), and 60 cm (i.e., tasseling, when the crop is mostly developed and reproductive stages begin). Similarly, the peanut root zones were 30 cm (i.e., vegetative stages), 45 cm (i.e., initial flower/pod set), and 60 cm (i.e., from beginning of pod fill to full seed fill). All irrigation events were 10 mm.

Soil physical properties obtained from the SSURGO database (NRCS, 2016) for the predominant Chipley-Foxworth-Albany soil type in the field (FC = 9.1% by volume, 50% MAD = 6.3%, available water holding capacity (AWHC) = 5%, and PWP = 3.5%) were compared with field-measured values (VWC using the probes in 2015 and following the guidelines of Zotarelli et al., 2013) and were found to have similar values (e.g., average measured FC = 9.3% (±0.2%) in a 30 cm root zone depth versus 9.1% FC from SSURGO). For both crops, irrigation in the SMS treatment was triggered when the VWC for any of the probes was greater than or equal to the 50% MAD threshold.

Non-Irrigated/Rainfed Plots (NON)

In both crops, the non-irrigated/rainfed plots were irrigated only for periods directly following each of the granular fertilization events. All plots (including NON) received on average 7.6 mm of irrigation after granular fertilizer applications to ensure incorporation of the fertilizer into the soil and to provide adequate soil water conditions for nutrient uptake.

Nitrogen Fertilizer Rates

Three N fertilizer rates applied in maize were evaluated:

The low and high N rates deviated ±36% from the medium rate. In peanut, minimal N fertilization (17 kg N ha-1) was applied to all treatments. Generally, fertilization in maize consisted of a starter application (initial 34 kg N ha-1 liquid application of N-P-K 16-16-0) at planting 5 cm below and 5 cm to the side of the seed in the row in all treatments to limit N losses and maximize uptake (Wright et al., 2003). Subsequently, two granular applications (at ~V3 and V6 growth stages) and four liquid sidedress applications (weekly from ~V8 to VT) were performed. This application pattern is representative of a grower practicing precision agriculture. Ideally, all fertilizer was applied before tasseling to ensure adequate nutrition prior to reproduction. All N fertilizer treatments included the starter application, with different treatment rates starting at the first granular application (table 1).

Table 1. Agronomic practices and nitrogen fertilizer rates (high, medium, and low) dates, amounts, and types applied during the 2015 maize and 2016 peanut growing seasons at NFREC-SV.
Crop and YearDateAgronomic PracticeFertilizer
Composition
NutrientsN Fertilizer Rate (kg ha-1)[a]
LowMediumHigh
Pre-Planting 201520 MarchHerbicide (Round-up)
30 MarchHarrow
Maize 20153 AprilPlanting day(16-16-0)N343434
P343434
K000
17 AprilFirst granular(33-0-0)N92534
(0-46-0)P848484
(0-0-60)K110110110
1 MaySecond granular(33-0-0)N112745
(0-46-0)P505050
(0-0-60)K868686
8 MayFirst liquid sidedress(28-0-0)N264056
15 MaySecond liquid sidedress(28-0-0)N264056
22 MayThird liquid sidedress(28-0-0)N264056
29 MayFourth liquid sidedress(28-0-0)N264056
Total N appliedN157247336
18 AugustHarvest day
Fallow 2015-2016[b]11 MayHarrow
Peanut 201613 MayPlanting day
23 JuneGypsum applicationCa493493493
S359359359
24 JuneGranular applicationN171717
P393939
K157157157
Total N appliedN171717
4 OctoberHarvest day
Fallow 2016-2017[b]5 JanuaryHarrow

    [a] N rates in maize: high = 336 kg N ha-1 , medium = 247 kg N ha-1 , and low = 157 kg N ha-1 . In peanut, a constant rate of 17 kg N ha-1 and fertilizations were equally applied to all plots. In discussion, N rates in peanut correspond to the rates applied in the previous (2015) maize growing season.

    [b]Fallow periods were harvested one day before the following crop planting date. The model assumed the day after harvest as the fallow planting date (input not required).

DSSAT Crop Simulation Models

The CERES-Maize and CROPGRO-Peanut simulation models within DSSAT (Jones et al., 1986) were used to evaluate the N and water dynamics in the maize-fallow-peanut-fallow rotation. DSSAT models use the one-dimensional tipping-bucket soil water balance, which predicts soil water flow and root water uptake for each of up to ten soil layers (Ritchie, 1985, 1998) and includes processes such as infiltration of rainfall and irrigation, soil water evaporation, root water uptake, drainage of water through the root zone, and possible maximum soil water evaporation and plant transpiration partitioned from reference ETo (Allen et al., 1998). Actual values of transpiration and soil water evaporation are limited by soil water availability and crop factors such as growth stage. The N balance includes processes such as daily N uptake, N2-fixation (CROPGRO), mobilization from vegetative tissues, rate of N use for new tissue growth, rate of N loss in abscised parts, as well as the turnover or soil organic matter and the decay of crop residues with the associated mineralization and/or immobilization of N, nitrification of ammonium, and N losses associated with denitrification (Boote et al., 2008; Godwin and Singh 1998).

Model Inputs and Methods

Field data collected at the NFREC-SV during a maize-fallow-peanut-fallow rotation were included in the model to create four required input files in DSSAT: soil, weather, cultivar genetic coefficients, and crop management.

Soil

Nine random pre-plant soil samples were collected across the experimental field using the cylindrical core method (Arshad et al., 1997) and analyzed by the UF/IFAS Analytical Services Laboratory (ANSERV, 2011). Soil physical characteristics such as sand (91.6% to 97.6%), silt (1.0% to 5.0%), clay (1.4% to 3.4%), bulk density (1.3 to 1.7 Mg m-3), as well as chemical characteristics such as initial NO3-N (0.3 to 4.0 mg kg-1), initial NH4-N (0.7 to 5.4 mg kg-1), TKN (90.5 to 731.2 mg kg-1), pH (5.2 to 6.6), and organic matter (0.1% to 2.1%) obtained from pre-plant sampling were included as inputs to the model. Details about the pre-plant sampling results are provided by Zamora-Re et al. (2020). To account for the spatial and temporal variability across the field, approximated values for the drained upper limit (DUL), available water content (AWC), and saturation (SAT) (i.e., equivalent to FC, water content between FC and PWP, and saturation points, respectively) were obtained using the Sentek probes averaged across all treatments. Chipley-Foxworth-Albany was the predominant soil type identified in the crop rotation; therefore, other soil characteristics (approximate values for lower limit (LL) equivalent to PWP and saturated hydraulic conductivity) were obtained from the Web Soil Survey (NRCS, 2016).

Weather

Observed (2015 to 2017) daily weather data (i.e., minimum and maximum air temperature, solar radiation, and rainfall) were obtained from the on-site Florida Automated Weather Network (FAWN) weather station located in Live Oak, Florida (30.305° N, -82.8988° W, at an elevation of 50.3 m) (FAWN, 2017).

Cultivar Genetic Coefficients

Maize hybrid Pioneer 1498 YHR/Bt and peanut cultivar Georgia 06G were evaluated in the NFREC-SV field experiment. The cultivars McCurdy 84aa and Georgia Green, which are suitable for Florida conditions, were selected from the DSSAT database for maize and peanut, respectively. The maize genetic coefficients for growth and yield of McCurdy 84aa were calibrated for the SMS-high treatment (table 6 and table A1 in the Appendix) to approximate simulated values that were closer to the measured values following the systematic approach described by Hunt and Boote (1998) (see the Model Calibration section). These calibrated values were then validated using crop growth measurements from the remaining eight treatments. The DSSAT default peanut genetic coefficients (table A2 in the Appendix) were able to simulate peanut aboveground (AG) N uptake, biomass, and yield in good agreement with the measured data (average nRMSE = 7.6%); therefore, no calibration of the peanut parameters was conducted.

Crop Management

The model followed the field experiment planting specifications as well as the fertilization and irrigation practices (see the Experimental Site and Design section). Fertilizer applications are summarized in table 1. The total N applied across seasons was 336, 247, and 157 kg N ha-1 in maize and 17 kg N ha-1 in peanut following maize. Table 2 shows the irrigation applied and cumulative rainfall during those seasons. Field management information was used to supply the required minimum model crop management inputs.

Table 2. Cumulative irrigation applied per treatment and cumulative rainfall during the 2015 maize and 2016 peanut growing seasons.
Crop and YearCumulative Irrigation (mm)Rainfall
(mm)
GROWSMSNON
Maize 201532115115545
Peanut 201654419318659

The CENTURY (Parton) method was used for the soil organic matter module. Using observed organic matter values from the initial sampling, total organic C was estimated for the model initiation (total organic C = organic matter / 1.72). All simulations were performed using the Finesand LOUSUF007 soil, which was developed in DSSAT using observed data. This soil file was constructed using four depths (0 to 15 cm, 15 to 30 cm, 30 to 60 cm, and 60 to 90 cm), the same as the soil N data collected in the experimental field. The simulation start date was 2 April 2015 (i.e., one day prior to the maize planting date in 2015). The crop previous to the rotation was defined as Bahia grass in the model to represent a mix of grasses, predominantly Bahia grass, during the period prior to planting maize in the experimental field. Mechanical operations (e.g., harrowing before planting and after harvest) to prepare the field and/or incorporate crop residue (table 1) were simulated by adding tillage applications. In the model, the fallow periods (2015 to 2016 and 2016 to 2017) extended from immediately after the crop harvest to the day before the following crop planting date. Model outputs were evaluated and compared to the measured field values.

The DSSAT model simulations were performed using the methods described in table 3, the planting and harvest dates described in table 1, and the initial conditions and input values defined in table 4.

Table 3. Methods used in DSSAT-CERES maize model for crop rotation simulations.
ProcessSimulation Method
EvapotranspirationFAO-56
InfiltrationUSDA Soil Conservation Service
PhotosynthesisLeaf photosynthesis response curve
HydrologyRitchie water balance[a]
Soil organic matter (SOM)CENTURY (Parton)[b]
Soil water evaporation methodSuleiman-Ritchie[c]
Soil layer distributionUnmodified soil profile[d]

    [a]Ritchie (1998).

    [b]Parton et al. (1988) and Parton and Rasmussen (1994).

    [c]Suleiman and Ritchie (2003).

    [d] Soil profile created using four depths: 0 to 15 cm, 15 to 30 cm, 30 to 60 cm, and 60 to 90 cm.

Table 4. Initial conditions from field experiment used in DSSAT CERES-Maize model for crop rotation simulations.
Depth to Base
of Layer
(cm)
Bulk
Density
(g cm-3)
Organic
Carbon
(%)
Total
Nitrogen
(%)
pH in
Water
Extractable
Phosphorus
(mg kg-1)
Volumetric
Water Content
(cm3 cm-3)
Ammonium
(NH4)
(g N Mg-1 soil)
Nitrate
(NO3)
(g N Mg-1 soil)
151.50.760.05661.90.11.82.5
301.50.670.04649.80.11.71.7
601.50.660.025.9390.111.6
901.50.340.015.733.50.10.91.4

Model Outputs

To evaluate the DSSAT model performance, the model outputs were compared with the measured field values for independent AG biomass and grain yield data collected in the other eight treatment combinations (excluding SMS-high) and for AG N uptake, soil water, and N content for all nine treatment combinations. The observed field data collected during the crop rotation, measurements, and sampling consisted of the following.

Soil Water

Sentek drill and drop (Sentek, 2003) capacitance probes were used to measure VWC at nine sensors at different depths across the soil profile every 30 min during the 2015 to 2017 crop growing seasons. The VWC data were summed across the soil profile (0 to 90 cm) and compared with DSSAT simulations across irrigation treatments.

Soil N

To determine N levels throughout the soil profile, soil samples were collected using a hand or power auger at depths of 0 to 15 cm, 15 to 30 cm, 30 to 60 cm, and 60 to 90 cm biweekly during the crop growing season and monthly after harvest. Subsamples were air-dried for 48 h, sieved, and analyzed for NO3-N and NH4-N following the EPA 353.2 and EPA 350.1 procedures, respectively. Further sampling details are provided by Zamora-Re (2019). The sum of soil nitrate-N (NO3-N) across the soil profile (0 to 90 cm) was compared with DSSAT simulations. Soil sampling events were scheduled to track the movement of nitrate before and after the fertilizer applications during all crop seasons (i.e., nitrate present in the soil, N uptake, and N leaching), subject to personnel availability and weather conditions.

Final Total AG Biomass and N Uptake

In-season maize tissue samples were collected at 12 DAP, at 38 DAP (during vegetative stages), at 68 DAP (~80% tasseling), at 100 DAP (~dough stage), and the final sampling was collected at 135 DAP (mature stage close to harvest). In peanut, in-season tissue samples were collected at 60 DAP (~peak pod formation and beginning maturity), at 100 DAP (~primary seed maturity), and at 147 DAP (mature stage close to harvest). In-season samples were collected only in the SMS treatment across all three N fertilizer rates, while the end-of-season samples were collected across the three irrigation treatments and three N fertilizer rates. The tissue sampling and analysis procedures are described in detail by Zamora-Re et al. (2020). Nitrogen uptake of the different plant tissues was calculated using the percentage N concentrations obtained from total Kjeldahl nitrogen (TKN) laboratory analysis and multiplied by the corresponding biomass dry weight to obtain total AG N uptake (kg ha-1). The final AG biomass dry weight (i.e., sum of leaves, stems, and ears/pods in kg ha-1) and corresponding N uptake were compared with the DSSAT simulations.

Yield

Maize grain yield was measured on 18 August 2015 and adjusted to 15.5% standard moisture content. Peanut pod yield was measured on 4 October 2016 and adjusted to 10.5% standard moisture content. Final yields were compared with model yield outputs.

Model Calibration

Data collected (i.e., phenology, biomass, and yield) in the SMS-high treatment (maize in 2015) was selected to calibrate the maize genetic coefficient parameters, under the assumption of SMS-high being a non-stressed treatment capable of achieving maximum growth and yield potential. Maize cultivar McCurdy 84aa (in the DSSAT database and suitable for Florida conditions) was selected for calibration. Following the systematic approach described by Hunt and Boote (1998), cultivar coefficients were obtained sequentially, starting with the phenological development parameters influencing crop growth cycle and dry matter accumulation, followed by the parameters influencing seed size and seed dry weight (table A1). The DSSAT default peanut genetic coefficients (table A2) were able to simulate peanut AG N uptake, biomass, and yield in good agreement with the measured data (average nRMSE = 7.6%); therefore, no calibration of the peanut parameters was conducted. Similarly, the model simulated soil water content within the range of variability of the observed data using soil hydrologic parameters specified based on measured values with no calibration.

Model Performance Statistics

The DSSAT CERES-Maize and CROPGRO models have been extensively used in crop production to simulate growth, yield, soil water, and nutrient dynamics as well as other interactive processes (Li et al., 2015; Liu et al., 2011; Ma et al., 2006; Naab et al., 2004; O’Neal et al., 2002; Paredes et al., 2014; Paz et al., 1999). Crop model performance was estimated by comparing the simulated (i.e., model outputs) and measured soil water content and soil nitrate and the final AG N uptake, AG biomass, and yield for each crop season across the three irrigation treatments and N fertilizer rates evaluated. Two statistical indices were employed for evaluation.

The root mean square error (RMSE) measures the differences between values predicted by a model and the observed values. The RMSE is calculated as the square root of the variance of the differences between the predicted and observed values. The RMSE is scale-dependent and serves to aggregate the magnitudes of the errors in predictions for various times into a single measure of predictive power for a particular dataset (Hyndman and Koehler, 2006).

To facilitate comparison between different datasets or models with different scales, the normalized root mean square error (nRMSE), a non-dimensional statistic (Liu et al., 2013; Yang et al., 2013), was calculated for final AG N uptake, final AG biomass, yield, and soil water across all treatments; nRMSE = RMSE / mean, with the following ranges: <15% (“good”), 15% > nRMSE < 30% (“moderate”), and >30% (“poor”). After evaluating the DSSAT model performance across the different variables, simulated water and N balances were evaluated, with a particularly focus on drainage and N leaching within the crop rotation.

Results and Discussion

Calibration of Cultivar Genetic Coefficients

For the crop growth cycle, the flowering date (P1) and maturity date (P5) genetic coefficients were calibrated using the SMS-high plant growth data. Based on field observations, flowering occurred at 57 DAP, and physiological maturity occurred at 115 DAP. Thus, P1 and P5 were manually fitted in order to achieve crop phenology equal to the measured data (i.e., P1 decreased to fit the flowering date and P5 increased to achieve a sufficiently long life cycle in comparison to the original cultivar). A comparison of simulated versus observed AG in-season biomass accumulation was performed, resulting in RMSE values within the range of variability of the observed data and good model performance for both the calibrated and validated treatments (i.e., RMSE < STDobs and nRMSE = 4%) (table 5); thus, no further adjustment was needed of the parameters that affect the crop growth cycle and development.

Afterward, the genetic parameters affecting seed size and mass were calibrated, i.e., maximum number of kernels per plant (G2) and kernel filling rate (G3). A range of parameter values were selected, compared, and adjusted to be close to the SMS-high observed final biomass and yield. For G2, values ranging from 800 to 920 in increments of 10 units were evaluated; for G3, values ranging from 8 to 10 in increments of 0.25 were evaluated. The combination of genetic coefficients was used to simulate biomass and yield for a total of 21 iterations. The calculated RMSE values for biomass and yield were 1,744 and 62 kg ha-1, both lower than the standard variation from the observed data. Final calibration results of the maize cultivar genetic coefficients (G2 = 800, G3 = 10) were selected based on the lowest RMSE compared to average observed data for biomass and yield and nRMSE values of 6% and 0.5%, respectively, indicating good model performance (table A1). The maize cultivar calibrated genetic coefficients in the CERES-Maize model are described in table 6.

Table 5. Summary of model performance indices (RMSE and nRMSE) for in-season biomass, total aboveground (AG) N uptake and biomass, and yield. Calibration was performed on SMS-high (maize in 2015), and the other eight treatment combinations were used for validation.
Crop and TreatmentIn-Season BiomassTotal AG N UptakeTotal AG BiomassYield
RMSE
(kg ha-1)
nRMSE
(%)
RMSE
(kg ha-1)
STDobs
(kg ha-1)
nRMSE
(%)
RMSE
(kg ha-1)
STDobs
(kg ha-1)
nRMSE
(%)
RMSE
(kg ha-1)
STDobs
(kg ha-1)
nRMSE
(%)
Maize calibration
SMS-high52044014141,4081,78053993
Maize validation
GROW-low--38418925,52047541,2336
GROW-medium--4649232,9303,175141,2091,43610
GROW-high--4332201,2082,77755181,0394
SMS-low6413560171,5794,87971,9821,71719
SMS-medium1,84413124941,2755,35651,1832,14110
SMS-high52044014141,4081,78053992,5183
Irrigated80963459141,6804,30571,1391,7669
NON-low--529301,7221,072101,4481,14718
NON-medium--6838331,0514,64961,2774,85415
NON-high--8349447154,59647402,1687
Rainfed--6935361,2363,49971,1943,03113
Peanut validation
GROW-low--85321,5181,267102406353
GROW-medium--1413334022,07824905076
GROW-high--4854112,2412,271154955076
SMS-low9481398021,3291,475987675411
SMS-medium231368019541,70764735946
SMS-high52154122116791,0024947431
Irrigated4106308671,3311,7979507752 6
NON-low--7558172,2497351578139616
NON-medium--128031262,17311,0441,14320
NON-high--158348791,090775795715
Rainfed--4571111,3961,5921087081817

Model Outputs and Performance

To evaluate the performance of the model, outputs for final AG biomass, N uptake, yield, soil water content, and soil nitrate in the soil profile (0 to 90 cm) were compared with observed data collected in the GROW, SMS, and NON irrigation treatments across the three N fertilizer rates. After evaluating the model performance results and building confidence in the model, simulated water and N balances were performed.

Biomass

Table 6. Cultivar coefficients calibrated for maize production simulations in CERES-Maize model.[a]
Cultivar CodeCultivar NameEcotype CodeP1P2P5G2G3PHINT
IB0035McCurdy 84aaIB00012600.311008001043

    [a]P1 = Thermal time from seedling emergence to the end of the juvenile stage, expressed in degree days above a base temperature of 8°C, during which the plant is not responsive to changes in photoperiod.

    P2 = Development delayed per hour increase in photoperiod above the longest photoperiod at which development proceeds at a maximum rate, expressed as days. The maximum rate was considered to be 12.5 hours.

    P5 = Thermal time from silking to physiological maturity, expressed in degree days above a base temperature of 8°C.

    G2 = Maximum number of kernels per plant.

    G3 = Kernel filling rate during grain filling stage under optimum conditions (mg d-1).

    PHINT = Phylochron interval, i.e., the interval in thermal time (degree days) between successive leaf tip appearances.

The model had good agreement between simulated and observed AG biomass for maize and peanut (average RMSE 1,167 kg ha-1 < STDobs 2,787 kg ha-1 and average nRMSE = 6%), providing a good representation for biomass accumulation, except in the GROW-high and the NON-low treatments for which the agreement was moderate (nRMSE = 15%) (table 5). Similar performance indices were reported by López-Cedrón et al. (2008) after testing the CERES-Maize model to predict maize biomass under a range of water-limiting conditions (mean RMSE = 2,020 kg ha-1 and nRMSE = 7%). However, Tojo Soler et al. (2007) found biomass nRMSE values ranging from 24% to 33% for irrigated maize and from 10% to 25% for rainfed maize across four maize hybrids grown in Brazil, thus indicating moderate to poor model performance for irrigated maize and good to moderate model performance for rainfed maize biomass. Some disagreements between observed and simulated biomass, particularly in water-limited conditions, were reported by Nouna et al. (2000).

Yield

During the 2015 maize season, yield simulations resulted in good agreement with field data for all irrigation treatments and N rates (average RMSE 869 kg ha-1 < STDobs 2,067 kg ha-1, and average nRMSE = 7%), with the exception of SMS-low, NON-low, and NON-medium, which resulted in a moderate agreement on final yields (average nRMSE = 17%) (table 5). There was good agreement between simulated and observed irrigated peanut pod yields (average RMSE 507 kg ha-1 < STDobs 752 kg ha-1, and average nRMSE = 6%) and moderate performance for rainfed pod yields (average RMSE 870 kg ha-1 < STDobs 818 kg ha-1, and average nRMSE = 17%) (table 5).

López-Cedrón et al. (2008) reported similar RMSE values for maize yield (1,127 and 2,954 kg ha-1 for irrigated and rainfed, respectively) after evaluation of CERES-Maize across seven irrigated treatments for simulating water deficit. Yang et al. (2013) evaluated the model performance over long-term continuous maize growth for three N rates: no N (N0), 165 kg N ha-1 from synthetic fertilizer (N165), and 50 kg N ha-1 from synthetic fertilizer plus 115 kg N ha-1 from manure (N165M). Simulated yields generally agreed with measured values; however, large disagreements were found in a few years of the evaluation. RMSE values were 1,146, 1,482, and 1,749 kg ha-1, resulting in nRMSE values of 36%, 22%, and 23% for the N0, N165, and N165M treatments, respectively, indicating poor performance for N0 and moderate performance for the N165 and N165M treatments (Yang et al., 2013). The results of those studies support the model performance indices obtained for maize and peanut yields across the irrigation and N fertilizer rates in this study, in which better model agreement with the observed data was achieved for irrigated treatments with high N inputs versus rainfed and low N inputs.

N Uptake

The RMSE for N uptake in the irrigated 2015 maize season (34 kg N ha-1) was less than the variation of the observed data (59 kg N ha-1), and the nRMSE was 14% (table 5). In contrast, the RMSE for rainfed maize (69 kg N ha-1) was greater than the variation of the measured data (35 kg N ha-1) and resulted in a 36% nRMSE (table 5). Overall, RMSE and nRMSE indicated reasonable results to represent the observed N uptake in the irrigated maize; however, N uptake was overpredicted under rainfed conditions. Researchers have found model overpredictions of N uptake due to overestimations of N mineralization (Bowen et al., 1993; Jiang et al., 2019; Liu et al., 2012).

The peanut N uptake (i.e., N uptake from soil + N fixation) RMSE values were 30 and 45 kg N ha-1 for the irrigated and rainfed treatments, respectively, which were lower than the observed variation in the field data (86 and 71 kg N ha-1, respectively). Average nRMSE values were 7% and 11%, indicating good model performance for peanut N uptake under irrigated and rainfed conditions, respectively (table 5).

Soil Water

Simulated daily soil water content in the soil profile (0 to 90 cm) during the 2015 maize and 2016 peanut growing seasons remained within the range of variability of the observed data during most of the crop rotation (fig. 1). However, greater soil water was simulated during the early stages of maize in 2015 (late April through early May) and during the reproductive stages of peanut in 2016 (mid through late June). Over the crop rotation, soil water resulted in RMSE less than the variation of the observed data; however, mostly due to the differences described above, the model had moderate and poor agreements for the irrigated and rainfed treatments, respectively (nRMSE = 18%, 17%, and 31% for GROW-high, SMS-high, and NON-high, respectively) (table 7). Most of the model disagreement in maize occurred

Figure 1. Daily observed (dots) and DSSAT-simulated (lines) total soil water in the profile (0 to 90 cm) in GROW-high, SMS-high, and NON-high treatments. Observed data are means of three Sentek probes in different blocks. Error bars are standard deviations across the three blocks.

due to underestimation of soil water evaporation losses from top soil layers during early stages of the crop. In peanut, the disagreement was due to underestimation of crop peak water uptake occurring during reproductive stages. During the period from 23 April to 3 May 2015 (~V3 to V6 maize vegetative stages), the model predicted on average 90% of the ET losses partitioned to maize transpiration and only 10% to soil water evaporation losses. However, the maize canopy coverage was only about 35% to 40% during this period; therefore, it is likely that actual soil water evaporation losses were greater. Nouna et al. (2000) reported model overestimation in soil water content during early stages of crop growth. In peanut, peak growth rate and water uptake occurred during reproductive stages, which started in mid-June; however, the model simulated the peak water uptake after 7 July 2016, resulting in greater soil water content than observed. In the NON treatment, the model underpredicted soil water from early June to mid-July, and although rainfall occurred, the simulated total soil water was less than the observed values (fig. 1).

Table 7. Summary of model performance indices (RMSE and nRMSE) for soil water and nitrate-N across GROW, SMS, and NON irrigation treatments with low, medium, and high N rates over crop rotation.
TreatmentSoil WaterSoil Nitrate-N
RMSESTDobsnRMSE
(%)
RMSESTDobsnRMSE
(%)
(mm)(mg kg-1)
GROW-low---5.56.565
GROW-med.---4.96.662
GROW-high1518185.38.267
SMS-low---4.24.662
SMS-med.---5.24.971
SMS-high1320179.912.382
NON-low---4.75.860
NON-med.---7.35.688
NON-high1923311210.2126

Similar studies have found good CERES-Maize model agreement with observed data (nRMSE <15%) when simulating soil water at different soil depths (Liu et al., 2013; Meng and Quiring 2008; Tojo Soler et al., 2007). Tojo Soler et al. (2007) studied the impact of planting dates on maize performance under rainfed and irrigated conditions. The model had good agreement with observed soil water content at four depths; however, simulated seasonal soil water presented larger variations in the rainfed experiment compared to the irrigated experiment. Meng and Quiring (2008) found accurate simulations of the annual cycle of soil water and of wetting and drying in response to weather conditions, but inaccurate soil water simulations in the upper soil layers. Liu et al. (2013) evaluated a model to predict the root zone soil water dynamics in a soybean-maize rotation under three tillage practices and found good to moderate agreement with soil water content measurements in the top 20 cm. A recent inter-comparison study among 28 maize growth models indicated that DSSAT models that used the Ritchie method (Ritchie, 1972) to calculate soil water evaporation (Es) often outperformed the newer approach described by Suleiman and Ritchie (2003), particularly during early season ET. However, the opposite occurred for ET data from 41 days after sowing to maturity over eight years of evaluation. The Ritchie (1972) method limits Es to water contents in the top soil layer, while the Suleiman and Ritchie (2003) method includes an upflux from the deeper soil layers (Kimball et al., 2019). These findings support the model performance results obtained in this study for maize in 2015 and suggest the use of the Ritchie (1972) method during early stages of the crop when Es was from the top soil layers is predominant.

Soil Nitrate-N

Soil mineral N is dynamic over time due to the daily changes in weather, crop growth, soil microbial activities, and soil water movement. In general, simulated NO3-N followed the observed trend and was within the range of variability of the measured data during most of the crop rotation; however, some disparities occurred after maize N fertilization and during the intercropping fallow periods (figs. 2 to 4).

Soil N was simulated with high error (nRMSE >30%); however, generally, the predicted values were within the standard deviation of the measured data. High nRMSE values could be a result of the non-normality of the soil N data. The GROW, SMS, and NON high N rate RMSE values were 5.3, 9.9, and 12.0 mg kg-1, respectively, which were lower than the standard deviation of the measured data, except for the NON treatment (STDobs = 8.2, 12.3, and 10.2 mg kg-1, respectively), reflecting good agreement of the simulated and observed data in the irrigated treatments and larger error in the rainfed treatment. Overall, the model tended to overestimate soil NO3-N after maize fertilizer applications and after peanut decay, and to underestimate soil NO3-N after maize residue decay. After N fertilizer applications, simulated soil NO3-N concentrations remained high, showing a delay compared to the observed data, particularly in the NON medium and high treatments, and thus overestimating the crop N demand and uptake (i.e., N uptake nRMSE = 33% and 44%, respectively) (table 5). Similar results were reported by Salmeron et al. (2014) and by Jiang et al. (2019), in which soil N resulted in a poor agreement between simulated and measured values (nRMSE > 30%).

Figure 2. Total observed (circles) and DSSAT-simulated (line) nitrate-N in the soil profile (0 to 90 cm) in the SMS-low, GROW-low, and NON-low treatments during the maize 2015 - fallow - peanut 2016 - fallow rotation. Observed data are means across four blocks. Error bars are standard deviations across replicates.

Greater levels of soil NO3-N due to peanut residue decay were simulated during the 2016 to 2017 fallow period across all treatments. Due to the sampling frequency (i.e., monthly after harvest), it is possible that greater soil NO3-N values were not captured during subsequent sampling in this fallow period when the greatest soil NO3-N values were simulated.

Figure 3. Total observed (circles) and DSSAT-simulated (lines) nitrate-N in the soil profile (0 to 90 cm) in the SMS-medium, GROW-medium, and NON-medium treatments during the maize 2015 - fallow - peanut 2016 - fallow rotation. Observed data are means across four blocks. Error bars are standard deviations across replicates.

Although the observed data had an appreciable increase in soil NO3-N after fertilization events and organic matter decomposition, it had a smaller magnitude compared to the simulations. Conversely, the model tended to underestimate soil NO3-N after maize residue decay during the 2015 to 2016 fallow period, particularly for the irrigated treatments, while the observed data had greater soil NO3-N concentrations due to the incorporation of maize residue (i.e., disk, harrow, and plow). Overall, a similar trend was observed for the low and medium N rates across the irrigation treatments, which had RMSE values close to the standard deviation of the measured data (figs. 2 and 3). Similar model performance indices were obtained by Gijsman et al. (2002), who evaluated the applicability of the DSSAT-CROPGRO model for soil N and soil organic matter decomposition in low-input systems using a Brazilian experiment with seven leguminous residue types. The RMSE values for mineral N concentration ranged from 3.0 to 11.3 mg kg-1 for the 0 to 15 cm top layer and from 1.6 to 9.0 mg kg-1 for the 15 to 30 cm soil layer (Gijsman et al., 2002).

Figure 4. Total observed (circles) and DSSAT-simulated (lines) nitrate-N in the soil profile (0 to 90 cm) in the SMS-high, GROW-high, and NON-high treatments during the maize 2015 - fallow - peanut 2016 - fallow rotation. Observed data are means across four blocks. Error bars are standard deviations across replicates.

Water Balance

The DSSAT-simulated water balance was calculated to evaluate the possible yield reduction caused by soil and plant water deficits during the crop rotation (table 8) and to evaluate the effects of the irrigation treatments. The model inputs correspond to irrigation, precipitation, and initial soil water, while the model outputs correspond to deep drainage, runoff, soil water evaporation, transpiration, and potential ET. The simulations obtained negligible values for runoff (except during excessive and heavy rainfall events, especially influenced by Hurricane Irma). In addition, during the fallow periods, no transpiration or irrigation occurred.

Simulated deep drainage corresponds to water leaving the root zone as a result of rainfall and/or irrigation. Generally, deep drainage occurring in the irrigated treatments was greater than in the NON treatment because soil water is kept at increased levels (i.e., close to FC), allowing less storage for rainfall events. The efficiency of the irrigated treatments at keeping soil water within the soil profile was evaluated.

Maize 2015 and Fallow 2015 to 2016

The irrigation applied in the GROW, SMS, and NON treatments during the 2015 maize season was 321, 151, and 15 mm, respectively. The SMS and NON treatments resulted in 53% and 95% water savings, respectively, compared to GROW. During the growing season, 54 precipitation events occurred, totaling 545 mm. In general, these events did not exceed 35 mm, were uniformly distributed during the growing season, and most occurred during reproductive stages when maize is highly susceptible to water stress. Thus, rainfall distribution and intensity provided favorable soil water conditions, particularly for the NON treatment. Cumulative deep drainage was 326, 158, and 87 mm for the GROW, SMS, and NON treatments, respectively, averaged across N fertilizer rates. Therefore, the sensor-based irrigation scheduling method reduced cumulative deep drainage by 51% compared to calendar-based irrigation scheduling. Slight variations in transpiration occurred between N rates (i.e., ~3% reduction in transpiration occurred with the low N rate compared to the high and medium rates); however, a 15% reduction in transpiration occurred in the rainfed treatments versus the irrigated treatments.

During the subsequent fallow period (from late August 2015 after maize harvest until early May 2016 before the peanut season), precipitation totaled 617 mm; however, greater losses due to deep drainage and soil water evaporation occurred (average of 370 and 252 mm, respectively, across all treatments).

Peanut 2016 and Fallow 2016 to 2017

During the peanut growing season, total irrigation applied was 544, 193, and 18 mm in the GROW, SMS, and NON treatments, respectively. Thus, 64% and 97% water savings resulted in the SMS and NON treatments compared to the GROW treatment. Precipitation totaled 659 mm, almost equal to the potential ET (673 mm) for this period; however, due to high-intensity and non-uniformly distributed rainfall events, most of the precipitation was not effective, resulting in greater deep drainage lost from the soil profile. Cumulative deep drainage was 673, 366, and 306 mm in the GROW, SMS, and NON treatments, respectively; thus, using SMS to schedule irrigation resulted in a 46% deep drainage reduction compared to calendar-based irrigation scheduling. Simulated runoff was on average 11 mm due to the high-intensity rainfall events occurring during this season. The model simulated 22% less transpiration for rainfed versus well-watered conditions.

During the subsequent fallow period (i.e., late October 2016 after peanut harvest until early March 2017), precipitation totaled 242 mm, which was 60% less than in the 2015 to 2016 fallow period. On average across all treatments, 98 mm was lost as deep drainage, and 94 mm was lost as soil water evaporation.

Table 8. Simulated water balance (mm) for GROW, SMS, and NON treatments during 2015 maize and 2016 peanut growing seasons and fallow periods from 2015 to 2016 and from 2016 to 2017.
Crop and
Year
Water Balance
Component
GROWSMSNON
LowMediumHighLowMediumHighLowMediumHigh
MaizeIrrigation321321321151151151151515
2015Precipitation545545545545545545545545545
Deep drainage326326326159158158878886
Runoff000000000
Soil water evaporation817168776564777071
Transpiration422433435424437437361366366
Reference ETo585585585585585585585585585
FallowPrecipitation617617617617617617617617617
2015-2016Deep drainage371370370371370371370370370
Runoff111111111
Soil water evaporation252252252252252252252252252
Reference ETo837837837837837837837837837
PeanutIrrigation544544544193193193181818
2016Precipitation659659659659659659659659659
Deep drainage673672673366366366306306306
Runoff131313101010999
Soil water evaporation120119120888888808080
Transpiration442442442435435435339339339
Reference ETo673673673673673673673673673
FallowPrecipitation242242242242242242242242242
2016-2017Deep drainage103103103100100100919191
Runoff000000000
Soil water evaporation949494949494939393
Reference ETo427427427427427427426426426

N Balance

As inputs, the simulated N balance considers initial and final soil N (total NH4-N and NO3-N), mineralized N, and N applied through fertilizer. The outputs of the simulated N balance correspond to N uptake from the soil, nitrate leaching, N denitrified, ammonia volatilization, and N immobilized. Only appreciable (>5 kg N ha-1) inputs and outputs were reported (table 9). Time series of the N balance components (i.e., soil N, N uptake, N mineralized, and N leached) are shown in figure 5.

Maize 2015 and Fallow 2015 to 2016

Simulated initial N was 39 N kg ha-1 due to the mineralization and incorporation of previous grasses planted in the experimental field. In general, the model captured the effects of the three N fertilizer rates, in which greater simulated total AG N uptake was found for the largest N rates, and vice versa. As the N rate increased, greater variation in simulated N uptake was found across irrigation treatments, while similar N uptake values were found for the low N rate. The effectiveness of the irrigation treatments and N fertilizer rates on N uptake and N leaching during the crop rotation in sandy soils was evaluated through the developed water and N balances.

Figure 5. Simulated nitrogen dynamics (total soil N (0-90 cm), N uptake, N leached, N applied (fertilizer), and N mineralized) in the SMS, GROW, and NON treatments and high N fertilizer rate during the maize 2015 - fallow - peanut 2016 - fallow rotation.

Similar N uptake values were obtained for the GROW and SMS treatments receiving the low N rate (232 and 238 kg N ha-1, respectively). Increasing N application from the low to the medium N rate resulted in 8% and 22% greater N uptake, and increasing N application from the low to the high N rate resulted in 12% and 38% greater N uptake for the GROW and SMS treatments, respectively. However, increasing N applications from the medium to the high N rate resulted in 4% and 14% greater N uptake, respectively. The N uptake was significantly greater only for the SMS-high N, while the percentage increase for the other treatments was very small or negligible, exceeding crop N demand. Simulations indicated that the SMS low, medium, and high N rates resulted in 3%, 16%, and 27% greater N uptake than GROW at the same rates.

Crop N uptake is simulated by examining the potential supply of N from the soil to the crop and the demand of the crop for N in order to fulfill the critical N concentration (i.e., the lowest concentration at which maximum growth occurs) (Godwin and Singh, 1998). The combination of the initial N, the N applied through fertilization, and the N obtained through mineralization processes, in addition to the magnitude and distribution of rainfall events that allowed soil N availability, provided enough N for crop demand, resulting in a high simulated crop N uptake during the season.

Although rainfall contributions during the 2015 maize growing season allowed adequate N to be in solution for uptake, excessive irrigation resulted in the opposite effect, increasing N leaching from the root zone. During the maize growing season, N leaching was 16, 83, and 161 kg N ha-1 in the GROW low, medium, and high treatments, respectively. In comparison, N leaching in the SMS treatments was 50%, 51%, and 45% less compared to the GROW treatments receiving the same N rates. Thus, high-frequency irrigation resulted in less N uptake and larger N leaching amounts (i.e., GROW-high), while not overirrigating but maintaining adequate soil water for plant growth allowed greater N uptake and reduced N leaching (i.e., SMS high and medium). Furthermore, applying the high N rate compared to the medium N rate did not contribute to significantly greater N uptake during the growing season; however, it increased N leaching nearly two-fold in the irrigated treatments. There was a synchrony between the changes in soil nitrate-N content related to the crop N uptake, which was also affected by the soil N supply and soil water availability during critical stages of the crop. It is evident that 39 kg N ha-1 resulted in a large contribution to the uptake of N during initial stages. Thus, potential N contributions from previous crops should be considered in crop fertilization programs.

During the subsequent fallow period, maize residue (i.e., biomass except ears) was left in the field after harvest, resulting in 198 and 220 kg N ha-1 of cumulative mineralized N on average in the GROW and SMS irrigation treatments across N rates, respectively. However, an average of 115 kg N ha-1 was immobilized, and 66 and 79 kg N ha-1 of leaching occurred in the same treatments on average across N rates. Simulations resulted in large N leaching amounts during this fallow period, particularly for the high N rate (68 and 96 kg N ha-1 in GROW and SMS, respectively).

The cumulative N leaching occurring during the growing season plus the subsequent fallow period was reduced by 36% on average when using the medium N rate compared to the high rate (247 vs. 336 kg N ha-1). In comparison to conventional practices (GROW-high), the SMS low, medium, and high treatments resulted in 67%, 50%, and 19% less N leaching.

The field remained fallow until May 2016; thus, the N that potentially could have been absorbed by the crops was mostly lost to leaching. If a winter crop were planted during this period, the N leaching amounts might be reduced, if the crop can actively take up N as mineralization processes occur.

Peanut 2016 and Fallow 2016 to 2017

After the winter fallow period, the field was prepared for peanut production (May 2016). Initial mineral N was on average 26 kg N ha-1, and minimum N fertilization (17 kg N ha-1) was applied during the peanut growing season. At the end of the season, cumulative mineralized N was 109 and 94 kg N ha-1 for the GROW and SMS treatments, respectively, on average across N fertilizer rates (i.e., rates applied in the 2015 maize season). During the peanut growing season, simulated peanut N uptake from the soil (i.e., excluding N fixation) averaged 61 and 74 N ha-1 for the GROW and SMS treatments, respectively, across N rates, and the corresponding N leaching was 53 and 38 kg N ha-1, respectively. Greater N leaching occurred due to frequent irrigation events; thus, using SMS to schedule irrigation reduced N leaching by 35%, 22%, and 27% compared to GROW low, medium, and high.

The subsequent fallow period extended from mid-October after peanut harvest until late March of the following year before maize planting. Simulated mineralization processes occurred at a rapid rate after peanut biomass was left on the ground, resulting in a cumulative mineral N average of 192 kg N ha-1, and immobilized N averaged 64 kg N ha-1 across all treatments (less than in the 2015 to 2016 fallow period after maize residue had decayed). Therefore, greater organic matter from peanut residue started mineralization processes providing large N available for uptake and less for N immobilization. The field remained fallow for approximately five months; as a consequence, most N was lost as N leaching by the end of this period. On average, simulated N leaching was 69 kg N ha-1 across irrigated treatments and N rates during the 2016 to 2017 fallow period. Based on these results, it is important to recognize that N potentially available for plants can be leached when not taken up or retained in the soil for the following crop season. Although the field data and model simulations presented greater discrepancies for soil nitrate during this fallow period, field soil samplings indicated an increase in soil N after the decay of plant residue, which could be used for crop N uptake or lost as N leaching, regardless of the model overpredictions. Furthermore, rainfall amounts and distribution play an important role in N leaching. Most of the simulated nitrate leaching events occurred after the fertilizer applications as well as after heavy rains.

Conclusions

A comprehensive evaluation of the DSSAT CERES-Maize and CROPGRO- Peanut models with field measurements on a maize-fallow-peanut-fallow rotation provided good model agreement for N uptake, AG biomass, and yield for the irrigated treatments during both maize and peanut growing seasons. Simulations followed the trends and remained mostly within the variation of the measured data, resulting in moderate model performance for peanut rainfed yields, soil water content, and soil NO3-N for the irrigated treatments during the crop rotation. Poor performance occurred for maize N uptake, soil water content, and soil NO3-N in the rainfed treatments.

The use of these crop simulation models along with field observations allowed better estimation of N processes and N fate and evaluation of the effectiveness of irrigation and N fertilizer rate BMPs within a maize-fallow-peanut-fallow rotation. The GROW and SMS treatments resulted in similar maize AG N uptake, biomass, and yields for both the medium and high N rates (247 and 336 kg N ha-1). However, during the maize growing season and subsequent fallow period, simulated N leaching was reduced 50% with the SMS-medium treatment compared to the GROW-high treatment while applying 26% less N fertilizer and reducing irrigation water by up to 53%. In comparison, for the peanut growing season and subsequent fallow season, SMS irrigation treatments resulted in an average of 65% irrigation reduction and 17% N leaching reduction, without negative impacts on yield.

Overall, during the maize-fallow-peanut-fallow rotation, simulations suggested that N leaching was reduced by 24% on average when using the medium N rate (247 kg N ha-1) compared to the high rate (336 kg N ha-1). Furthermore, fertilizer, irrigation, and N leaching reductions of 26%, 60%, and 37%, respectively resulted when using SMS to schedule irrigation, the medium N rate in maize production, and the minimum N fertilization in peanut production compared to conventional management practices (GROW-high + min N peanut), without negative impacts on yields.

Observed and simulated soil N increased during maize and peanut residue decay. The simulation results indicated that this mineralized N then leached during the fallow season. If a cover or cash crop were planted immediately following the maize and peanut harvest, this fallow period leaching could potentially be reduced. In addition, N contributions from both cash and cover crop residue mineralization processes could benefit the following crop’s growth and development, reducing N fertilizer requirements at planting and improving profits. However, the residue N supply and the subsequent crop N demand must be synchronized for successful N uptake and thus reduce the risk of environmental losses.

Acknowledgements

This work was supported by Florida Department of Agriculture and Customer Services (FDACS) (Contract No. 21894, 2015-2018) and partially based upon work that is supported by the USDA National Institute of Food and Agriculture under Award No. 2017-68007-26319. The authors wish to express their gratitude to Ben Broughton, Mike Boyette, Michael Gutierrez, Marc Thomas, as well as all staff from the North Florida Research and Education Center - Suwannee Valley (NFREC-SV) and the University of Florida Department of Agricultural and Biological Engineering who helped make this project possible. Thanks to Dr. Ken Boote for his guidance and advice in using the DSSAT models.

Appendix

Table A1. Cultivar genetic coefficients (G2 and G3) calibration for McCurdy 84aa used in CERES-Maize model.
Cultivar Genetic
Coefficients
BiomassBiomass
Observed
(kg ha-1)
Simulated
(kg ha-1)
RMSE
(kg ha-1)
nRMSE
(%)
Observed
(kg ha-1)
Simulated
(kg ha-1)
RMSE
(kg ha-1)
nRMSE
(%)
G2G3
920827,52626,0901436513,06613,4653993
910827,52624,9452581913,06612,2148527
900827,52624,83426921013,06612,0949727
890827,52624,72428021013,06611,97410928
880827,52624,61429121113,06611,85412129
870827,52624,50330231113,06611,734133210
860827,52624,39331331113,06611,613145311
850827,52624,28232441213,06611,493157312
840827,52624,17233541213,06611,373169313
830827,52624,06234641313,06611,253181314
820827,52623,95135751313,06611,133193315
810827,52623,84036861313,06611,012205416
800827,52623,72438021413,06610,892217417
8008.2527,52623,60939171413,06610,772229418
8008.527,52623,92935971313,06611,109195715
8008.7527,52624,23832881213,06611,445162112
800927,52624,54829781113,06611,782128410
8009.2527,52624,85726691013,06612,1189487
8009.527,52625,1652361913,06612,4556115
8009.7527,52625,4742052713,06612,7922742
800[a]10[a]27,52625,7821744613,06613,128620

    [a] Final combination of genetic coefficients (G2 = 800, G3 = 10) selected based on lowest RMSE compared to biomass and yield average observed data across four blocks.

Table A2. Genetic coefficients for peanut cultivar Georgia Green (cultivar code = A0001, ecotype code = Runner).
CoefficientValue[a]Definition
CSDL11.84Critical short day length below which reproductive development progresses with no daylength effect (for short-day plants) (h).
PPSEN0Slope of relative response of development to photoperiod with time (positive for short-day plants) (h-1).
EM-FL21.2Time between plant emergence and flower appearance (R1) (photothermal days).
FL-SH9.2Time between first flower and first pod (R3) (photothermal days).
FL-SD18.8Time between first flower and first seed (R5) (photothermal days).
SD-PM77.3Time between first seed (R5) and physiological maturity (R7) (photothermal days).
FL-LF85Time between first flower (R1) and end of leaf expansion (photothermal days).
LFMAX1.45Maximum leaf photosynthesis rate at 30°C, 350 vpm CO2, and high light (mg CO2 m-2 s-1).
SLAVR270Specific leaf area of cultivar under standard growth conditions (cm2 g-1).
SIZLF18Maximum size of full leaf (three leaflets) (cm2).
XFRT0.95Maximum fraction of daily growth that is partitioned to seed + shell.
WTPSD0.69Maximum weight per seed (g).
SFDUR42Seed filling duration for pod cohort at standard growth conditions (photothermal days).
SDPDV1.65Average seeds per pod under standard growing conditions (seeds/pod).
PODUR28Time required for cultivar to reach final pod load under optimal conditions (photothermal days).
THRSH80

    Maximum ratio of (seed / (seed + shell)) at maturity. Causes seed to stop growing as their dry weights increase until shells are filled in a cohort (threshing percentage).

SDPRO0.27Fraction protein in seeds (g protein g-1 seed).
SDLIP0.51Fraction oil in seeds (g oil g-1 seed).

    [a]Default coefficient values in DSSAT database, not calibrated.

References

Allen, R. G., Pereira, L. S., Raes, D., & Smith, M. (1998). Crop evapotranspiration: Guidelines for computing crop water requirements. FAO Irrigation and Drainage Paper No. 56. Rome, Italy: United Nations FAO.

Andraski, T. W., Bundy, L. G., & Brye, K. R. (2000). Crop management and corn nitrogen rate effects on nitrate leaching. J. Environ. Qual., 29(4), 1095-1103. https://doi.org/10.2134/jeq2000.00472425002900040009x

ANSERV. (2011). UF/IFAS Analytical Services Laboratory (ANSERV Labs). Gainesville, FL: University of Florida IFAS. Retrieved from https://arl.ifas.ufl.edu/ARL%20Analysis.asp

Arshad, M. A., Lowery, B., & Grossman, B. (1997). Physical tests for monitoring soil quality. In J. W. Doran & A. J. Jones (Eds.), Methods for assessing soil quality (pp. 123-141). Madison, WI: SSSA. https://doi.org/10.2136/sssaspecpub49.c7

Arthur, J. D., Wood, H. A., Baker, A. E., Cichon, J. R., & Raines, G. L. (2007). Development and implementation of a Bayesian-based aquifer vulnerability assessment in Florida. Nat. Resour. Res., 16(2), 93-107. https://doi.org/10.1007/s11053-007-9038-5

Asadi, M., & Clemente, R. (2003). Evaluation of CERES-Maize of DSSAT model to simulate nitrate leaching, yield, and soil moisture content under tropical conditions. Food Agric. Environ., 1(3-4), 270-276. Retrieved from https://www.wflpublisher.com/Abstract/446

Basso, B., Dumont, B., Maestrini, B., Shcherbak, I., Robertson, G. P., Porter, J. R., ... Rosenzweig, C. (2018). Soil organic carbon and nitrogen feedbacks on crop yields under climate change. Agric. Environ. Lett., 3(1), 1-5. https://doi.org/10.2134/ael2018.05.0026

Boote, K. J., Sau, F., Hoogenboom, G., & Jones, J. W. (2008). Experience with water balance, evapotranspiration, and predictions of water stress effects in the CROPGRO model. In L. A. Ahuja (Ed.), Response of crops to limited water: Understanding and modeling water stress effects on plant growth processes (pp. 59-103). Madison, WI: ASA-CSSA-SSSA. https://doi.org/10.2134/advagricsystmodel1.c3

Bowen, W. T., Jones, J. W., Carsky, R. J., & Quintana, J. O. (1993). Evaluation of the nitrogen submodel of CERES-Maize following legume green manure incorporation. Agron. J., 85(1), 153-159. https://doi.org/10.2134/agronj1993.00021962008500010028x

Bush, P. W., & Johnston, R. H. (1988). Ground-water hydraulics, regional flow, and groundwater development of the Floridan aquifer system in Florida and in parts of Georgia, South Carolina, and Alabama. USGS Paper No. 1403-C. Reston, VA: U.S. Geological Survey. Retrieved from https://doi.org/10.3133/pp1403C

Donner, S. D., Kucharik, C. J., & Foley, J. A. (2004). Impact of changing land use practices on nitrate export by the Mississippi River. Global Biogeochem. Cycles, 18(1), article GB1028. https://doi.org/10.1029/2003gb002093

EPA. (2016). Clean Water Act Section 303(d): Impaired waters and total maximum daily loads (TMDLs). Washington, DC: U.S. Environmental Protection Agency. Retrieved from https://www.epa.gov/tmdl

FAWN. (2017). Florida Automated Weather Network: Data access. Gainesville, FL: University of Florida IFAS. Retrieved from https://fawn.ifas.ufl.edu/data/reports/

FDACS. (2015). Water quality/quantity best management practices for Florida vegetable and agronomic crops. Tallahassee, FL: Florida Department of Agriculture and Consumer Services. Retrieved from https://www.fdacs.gov/content/download/77230/file/vegAgCropBMP-loRes.pdf

FDEP. (2013). Rule 62-302.530: Surface water quality standards. Tallahassee, FL: Florida Department of Environmental Protection. Retrieved from https://www.flrules.org/gateway/RuleNo.asp?ID=62-302.530

Ferguson, R. B., Shapiro, C. A., Hergert, G. W., Kranz, W. L., Klocke, N. L., & Krull, D. H. (1991). Nitrogen and irrigation management practices to minimize nitrate leaching from irrigated corn. J. Prod. Agric., 4(2), 186-192. https://doi.org/10.2134/jpa1991.0186

Gehl, R. J., Schmidt, J. P., Maddux, L. D., & Gordon, W. B. (2005). Corn yield response to nitrogen rate and timing in sandy irrigated soils. Agron. J., 97(4), 1230-1238. https://doi.org/10.2134/agronj2004.0303

Gijsman, A. J., Hoogenboom, G., Parton, W. J., & Kerridge, P. C. (2002). Modifying DSSAT crop models for low-input agricultural systems using a soil organic matter-residue module from CENTURY. Agron. J., 94(3), 462-474. https://doi.org/10.2134/agronj2002.4620

Godwin, D. C., & Singh, U. (1998). Nitrogen balance and crop response to nitrogen in upland and lowland cropping systems. In Understanding options for agricultural production (pp. 55-77). Dordrecht, Netherlands: Springer. https://doi.org/10.1007/978-94-017-3624-4_4

He, J. (2008). Best management practice development with the CERES-Maize model for sweet corn production in north Florida. PhD diss. Gainesville, FL: University of Florida, Department of Agricultural and Biological Engineering.

Hoogenboom, G., Jones, J. W., Wilkens, P. W., Porter, C. H., Boote, K. J., Hunt, L. A., & Tsuji, G. Y. (2015). Decision Support System for Agrotechnology Transfer (DSSAT) ver. 4.6. Gainesville, FL: DSSAT Foundation. Retrieved from https://dssat.net

Hoogenboom, G., Porter, C. H., Shelia, V., Boote, K. J., Singh, U., White, J. W., & Jones, J. W. (2019). Decision Support System for Agrotechnology Transfer (DSSAT). Gainesville, FL: DSSAT Foundation.

Hornsby, D., & Mattson, R. (1997). Surface water quality and biological monitoring network annual report. Live Oak, FL: Suwannee River Water Management District.

Howarth, R., Chan, F., Conley, D. J., Garnier, J., Doney, S. C., Marino, R., & Billen, G. (2011). Coupled biogeochemical cycles: Eutrophication and hypoxia in temperate estuaries and coastal marine ecosystems. Front. Ecol. Environ., 9(1), 18-26. https://doi.org/10.1890/100008

Hunt, L. A., & Boote, K. J. (1998). Data for model operation, calibration, and evaluation. In Understanding options for agricultural production (pp. 9-39). Dordrecht, Netherlands: Kluwer Academic.

Hyndman, R. J., & Koehler, A. B. (2006). Another look at measures of forecast accuracy. Intl. J. Forecasting, 22(4), 679-688. https://doi.org/10.1016/j.ijforecast.2006.03.001

Jiang, R., He, W., Zhou, W., Hou, Y., Yang, J. Y., & He, P. (2019). Exploring management strategies to improve maize yield and nitrogen use efficiency in northeast China using the DNDC and DSSAT models. Comput. Electron. Agric., 166, 104988. https://doi.org/10.1016/j.compag.2019.104988

Jones, C. A., Kiniry, J. R., & Dyke, P. T. (1986). CERES-Maize: A simulation model of maize growth and developmen. College Station, TX: Texas A&M University Press. Retrieved from https://books.google.com/books?id=SCkzAAAAMAAJ

Katz, B. G. (2004). Sources of nitrate contamination and age of water in large karstic springs of Florida. Environ. Geol., 46(6), 689-706. https://doi.org/10.1007/s00254-004-1061-9

Katz, B. G., DeHan, R. S., Hirten, J. J., & Catches, J. S. (1997). Interactions between groundwater and surface water in the Suwannee River basin, Florida. JAWRA, 33(6), 1237-1254. https://doi.org/10.1111/j.1752-1688.1997.tb03549.x

Katz, B. G., Hornsby, H. D., Bohlke, J. F., & Mokray, M. F. (1999). Sources and chronology of nitrate contamination in spring waters, Suwannee River basin, Florida. USGS Report No. 99-4252. Tallahassee, FL: U.S. Geological Survey.

Katz, B. G., Sepulveda, A. A., & Verdi, R. J. (2009). Estimating nitrogen loading to groundwater and assessing vulnerability to nitrate contamination in a large karstic springs basin, Florida. JAWRA, 45(3), 607-627. https://doi.org/10.1111/j.1752-1688.2009.00309.x

Kimball, B. A., Boote, K. J., Hatfield, J. L., Ahuja, L. R., Stockle, C., Archontoulis, S., ... Williams, K. (2019). Simulation of maize evapotranspiration: An inter-comparison among 29 maize models. Agric. Forest Meteorol., 271, 264-284. https://doi.org/10.1016/j.agrformet.2019.02.037

Li, Z. T., Yang, J. Y., Drury, C. F., & Hoogenboom, G. (2015). Evaluation of the DSSAT-CSM for simulating yield and soil organic C and N of a long-term maize and wheat rotation experiment in the loess plateau of northwestern China. Agric. Syst., 135, 90-104. https://doi.org/10.1016/j.agsy.2014.12.006

Liu, H. L., Yang, J. Y., Drury, C. F., Reynolds, W. D., Tan, C. S., Bai, Y. L., ... Hoogenboom, G. (2011). Using the DSSAT-CERES-Maize model to simulate crop yield and nitrogen cycling in fields under long-term continuous maize production. Nutr. Cycling Agroecosyst., 89(3), 313-328. https://doi.org/10.1007/s10705-010-9396-y

Liu, H.-L., Yang, J.-Y., He, P., Bai, Y.-L., Jin, J.-Y., Drury, C. F., ... Hoogenboom, G. (2012). Optimizing parameters of CSM-CERES-Maize model to improve simulation performance of maize growth and nitrogen uptake in northeast China. J. Integ. Agric., 11(11), 1898-1913. https://doi.org/10.1016/S2095-3119(12)60196-8

Liu, S., Yang, J. Y., Zhang, X. Y., Drury, C. F., Reynolds, W. D., & Hoogenboom, G. (2013). Modeling crop yield, soil water content, and soil temperature for a soybean-maize rotation under conventional and conservation tillage systems in northeast China. Agric. Water Mgmt., 123, 32-44. https://doi.org/10.1016/j.agwat.2013.03.001

López-Cedrón, F. X., Boote, K. J., Piñeiro, J., & Sau, F. (2008). Improving the CERES-Maize model ability to simulate water deficit impact on maize production and yield components. Agron. J., 100(2), 296-307. https://doi.org/10.2134/agronj2007.0088

Ma, L., Hoogenboom, G., Ahuja, L. R., Ascough, J. C., & Saseendran, S. A. (2006). Evaluation of the RZWQM-CERES-Maize hybrid model for maize production. Agric. Syst., 87(3), 274-295. https://doi.org/10.1016/j.agsy.2005.02.001

Marek, G. W., Marek, T. H., Xue, Q., Gowda, P. H., Evett, S. R., & Brauer, D. K. (2017). Simulating evapotranspiration and yield response of selected corn varieties under full and limited irrigation in the Texas High Plains using DSSAT-CERES-Maize. Trans. ASABE, 60(3), 837-846. https://doi.org/10.13031/trans.12048

Marella, R. L. (2015). Water withdrawals in Florida, 2012. USGS Open-File Report 2015-1156. Reston, VA: U.S. Geological Survey. https://doi.org/10.3133/ofr20151156

Meng, L., & Quiring, S. M. (2008). A comparison of soil moisture models using soil climate analysis network observations. J. Hydrometeorol., 9(4), 641-659. https://doi.org/10.1175/2008jhm916.1

Mylavarapu, R. S., Wright, D., & Kidder, G. (2015). UF/IFAS standardized fertilization recommendations for agronomic crops. SL129. Gainesville, FL: University of Florida IFAS. Retrieved from https://edis.ifas.ufl.edu/ss163

Naab, J. B., Singh, P., Boote, K. J., Jones, J. W., & Marfo, K. O. (2004). Using the CROPGRO-Peanut model to quantify yield gaps of peanut in the Guinean Savanna zone of Ghana. Agron. J., 96(5), 1231-1242. https://doi.org/10.2134/agronj2004.1231

NASS. (2012). 2012 Census of agriculture - State data. Washington, DC: USDA National Agricultural Statistics Service. Retrieved from https://www.nass.usda.gov/Publications/AgCensus/2012/Online_Resources/County_Profiles/Florida/cp12121.pdf

Nouna, B. B., Katerji, N., & Mastrorilli, M. (2000). Using the CERES-Maize model in a semi-arid Mediterranean environment. Evaluation of model performance. European J. Agron., 13(4), 309-322. https://doi.org/10.1016/S1161-0301(00)00063-0

NRCS. (2016). Web soil service. Washington, DC: USDA Natural Resources Conservation Service. Retrieved from https://websoilsurvey.nrcs.usda.gov/app/WebSoilSurvey.aspx

O’Neal, M. R., Frankenberger, J. R., & Ess, D. R. (2002). Use of CERES-Maize to study effect of spatial precipitation variability on yield. Agric. Syst., 73(2), 205-225. https://doi.org/10.1016/S0308-521X(01)00095-6

Paredes, P., Rodrigues, G. C., Alves, I., & Pereira, L. S. (2014). Partitioning evapotranspiration, yield prediction, and economic returns of maize under various irrigation management strategies. Agric. Water Mgmt., 135, 27-39. https://doi.org/10.1016/j.agwat.2013.12.010

Parton, W. J., & Rasmussen, P. E. (1994). Long-term effects of crop management in wheat-fallow: II. CENTURY model simulations. SSSA J., 58(2), 530-536. https://doi.org/10.2136/sssaj1994.03615995005800020040x

Parton, W. J., Stewart, J. W., & Cole, C. V. (1988). Dynamics of C, N, P, and S in grassland soils: A model. Biogeochem., 5(1), 109-131. https://doi.org/10.1007/BF02180320

Paz, J. O., Batchelor, W. D., Babcock, B. A., Colvin, T. S., Logsdon, S. D., Kaspar, T. C., & Karlen, D. L. (1999). Model-based technique to determine variable-rate nitrogen for corn. Agric. Syst., 61(1), 69-75. https://doi.org/10.1016/S0308-521X(99)00035-9

Prasad, R., Hochmuth, G. J., & Boote, K. J. (2015). Estimation of nitrogen pools in irrigated potato production on sandy soil using the model SUBSTOR. PLoS One, 10(1), e0117891. https://doi.org/10.1371/journal.pone.0117891

Rabalais, N. N., Turner, R. E., Justic, D., Dortch, Q., Wiseman, W. J., & Sen Gupta, B. K. (1996). Nutrient changes in the Mississippi River and system responses on the adjacent continental shelf. Estuaries, 19(2), 386-407. https://doi.org/10.2307/1352458

Ritchie, J. T. (1972). Model for predicting evaporation from a row crop with incomplete cover. Water Resour. Res., 8(5), 1204-1213. https://doi.org/10.1029/WR008i005p01204

Ritchie, J. T. (1985). A user-orientated model of the soil water balance in wheat. In Wheat growth and modeling (pp. 293-305). Boston, MA: Springer. https://doi.org/10.1007/978-1-4899-3665-3_27

Ritchie, J. T. (1998). Soil water balance and plant water stress. In Understanding options for agricultural production (pp. 41-54). Dordrecht, Netherlands: Springer. https://doi.org/10.1007/978-94-017-3624-4_3

Salmerón, M., Cavero, J., Isla, R., Porter, C. H., Jones, J. W., & Boote, K. J. (2014). DSSAT nitrogen cycle simulation of cover crop-maize rotations under irrigated Mediterranean conditions. Agron. J., 106(4), 1283-1296. https://doi.org/10.2134/agronj13.0560

Sentek. (2003). TriSCAN manual ver. 1.2a. Stepney, South Australia: Sentek Pty Ltd.

Strebel, O., Duynisveld, W. H., & Bottcher, J. (1989). Nitrate pollution of groundwater in western Europe. Agric. Ecosyst. Environ., 26(3), 189-214. https://doi.org/10.1016/0167-8809(89)90013-3

Suleiman, A. A., & Ritchie, J. T. (2003). Modeling soil water redistribution during second-stage evaporation. SSSA J., 67(2), 377-386. https://doi.org/10.2136/sssaj2003.3770

Tojo Soler, C. M., Sentelhas, P. C., & Hoogenboom, G. (2007). Application of the CSM-CERES-Maize model for planting date evaluation and yield forecasting for maize grown off-season in a subtropical environment. European J. Agron., 27(2), 165-177. https://doi.org/10.1016/j.eja.2007.03.002

Turner, R. E., & Rabalais, N. N. (1991). Changes in Mississippi River water quality this century: Implications for coastal food webs. Bioscience, 41(3), 140-147. https://doi.org/10.2307/1311453

Upchurch, S. B., Chen, J., & Cain, C. R. (2007). Trends of nitrate concentrations in waters of the Suwannee River Water Management District, 2007. Live Oak, FL: Suwannee River Water Management District. Retrieved from https://www.mysuwanneeriver.com/DocumentCenter/View/130/2007-Nitrate-Trend-Report?bidId=

Valmont. (2015). Valley variable-rate irrigation (VRI). Omaha, NE: Valmont Industries. Retrieved from http://www.valleyirrigation.com/vri

Wright, D., Marois, J., Rich, J., Rowland, D., & Mulvaney, M. (2003). Field corn production guide. SS AGR 85. Gainesville, FL: University of Florida IFAS. Retrieved from https://edis.ifas.ufl.edu/ag202

Yang, J. M., Yang, J. Y., Dou, S., Yang, X. M., & Hoogenboom, G. (2013). Simulating the effect of long-term fertilization on maize yield and soil C/N dynamics in northeastern China using DSSAT and CENTURY-based soil model. Nutr. Cycling Agroecosyst., 95(3), 287-303. https://doi.org/10.1007/s10705-013-9563-z

Zamora-Re, M. I. (2019). Irrigation and nitrogen best management practices in corn production. PhD diss. Gainesville, FL: University of Florida, Department of Agricultural and Biological Engineering.

Zamora-Re, M. I., Dukes, M. D., Hensley, D., Rowland, D., & Graham, W. (2020). The effect of irrigation strategies and nitrogen fertilizer rates on maize growth and grain yield. Irrig. Sci., 38(4), 461-478. https://doi.org/10.1007/s00271-020-00687-y

Zhang, W. L., Tian, Z. X., Zhang, N., & Li, X. Q. (1996). Nitrate pollution of groundwater in northern China. Agric. Ecosyst. Environ., 59(3), 223-231. https://doi.org/10.1016/0167-8809(96)01052-3

Zotarelli, L., Dukes, M. D., & Morgan, K. T. (2013). Interpretation of soil moisture content to determine soil field capacity and avoid over-irrigating sandy soils using soil moisture sensors. AE460. Gainesville, FL: University of Florida IFAS. Retrieved from https://edis.ifas.ufl.edu/pdffiles/AE/AE46000.pdf

DSSAT simulations of final N uptake, biomass, and yield for a maize-peanut rotational field experiment with three irrigation treatments and three N fertilizer rates had good performance for the irrigated treatments (average nRMSE of 9%) but greater error for the rainfed treatments (average nRMSE of 15%).

Experiments and DSSAT simulations demonstrated that N fertilizer and irrigation applications were reduced by 26% and 60%, respectively, when using a 247 kg N ha-1 fertilizer rate and a sensor-based irrigation schedule rather than conventional practices of 336 kg N ha-1 and a calendar-based irrigation method, with no impact on yield.

Simulations demonstrated that N leaching during the crop rotation was reduced by 37% when an N fertilizer rate of 247 kg N ha-1 and sensor-based irrigation scheduling were used versus conventional practices.

Soil N increased (=15 mg kg-1) when maize and peanut residues decayed and then leached during the fallow season. Cover or cash crops planted immediately after the maize and peanut harvests have potential to take up this N and reduce leaching.

Abstract. Nitrogen (N) is an essential element for crop growth and yield; however, excessive N applications not taken up by crops can result in N leaching from the root zone, increasing N loads to waterbodies and leading to a host of environmental problems. The main objective of this study was to simulate water and N balances for a maize-peanut (Zea mays L. and Arachis hypogaea L.) rotational field experiment with three irrigation treatments and three N fertilizer rates. The irrigation treatments consisted of mimicking grower irrigation practices in the region (GROW), using soil moisture sensors to schedule irrigation (SMS), and non-irrigated (NON). The N fertilizer rates were low, medium, and high (157, 247, and 336 kg N ha-1, respectively) for maize with a constant 17 kg ha-1 for all peanut treatments. DSSAT maize genetic coefficients were calibrated using the SMS-high treatment combination under the assumption of no water or N stress. The other eight treatment combinations were used as independent data for model validation of the crop coefficients. All soil hydrologic parameters were specified based on measured values, and default DSSAT peanut genetic coefficients were used with no calibration. For the irrigated treatments, DSSAT models had good performance for N uptake, biomass, and yield (average nRMSE of 8%) and moderate performance for soil water content (average nRMSE of 18%). Soil nitrate RMSE was 21% lower than the standard deviation of the observed data (5.8 vs. 7.2 mg kg-1). For the rainfed treatments, DSSAT had greater error (average nRMSE of 15% for N uptake, biomass, and yield, and average nRMSE of 31% for soil water). Soil nitrate RMSE was 11% greater than the standard deviation of the observed data (8.0 vs. 7.2 mg kg-1), and nRMSE was >30% during the crop rotation. Simulations estimated that N leaching over the crop rotation was reduced by 24% on average when using the 247 kg N ha-1 fertilizer rate compared to 336 kg N ha-1 across the irrigation treatments. Furthermore, N leaching was reduced by 37% when using SMS to schedule irrigation and the 247 kg N ha-1 fertilizer rate for maize and 17 kg N ha-1 for peanut compared to conventional practices (GROW and 336 kg N ha-1 for maize and 17 kg N ha-1 for peanut). Moreover, this management practice reduced N fertilizer use by 26% and irrigation water use by up to 60% without negative impacts on yield. Observed and simulated soil N increased during maize and peanut residue decay, with simulations estimating that this soil N would leach below the root zone during the fallow season. This leaching could potentially be reduced if a cover crop or cash crop were planted between the maize and peanut crops to take up the mineralized N.

"/>