# Agronomy Journal - Article

1. OPEN ACCESS

Accepted: May 23, 2017
Published: July 13, 2017

* Corresponding author(s):
View
Permissions
Share

doi:10.2134/agronj2016.12.0739

# Can Organic Amendments Support Sustainable Vegetable Production?

1. Daniele De Rosa *a,
2. Bruno Bassoab,
3. David W. Rowlingsa,
4. Clemens Scheera,
5. Johannes Bialaa and
6. Peter R. Gracea
1. a Institute for Future Environments, Queensland Univ. of Technology, Brisbane, QLD 4000, Australia
b Dep. of Earth and Environmental Sciences, and W.K. Kellogg Biological Station, Michigan State Univ., 288 Farm Lane, East Lansing, MI 48823
Core Ideas:
• Accounting for the N release from organic amendments improves N use efficiency and promotes soil C storage in horticultural soils.
• Regional N fertilizer recommendations are affected by a high degree of uncertainty.
• Crop simulation model can help to develop efficient site-specific N management.

## Abstract

Application rates of synthetic fertilizer to agricultural fields can be reduced through better understanding of N supplied by organic amendments (OA). Field and simulation experiments were performed to quantify the effect of N released from OA application on crop production and selected soil properties in an intensively managed vegetable crop rotation. The SALUS crop model was used to simulate yield, soil N, and soil organic carbon (SOC) dynamics under different combinations of composted or raw OA and synthetic N fertilizer application rates. SALUS accurately simulated aboveground crop biomass production (r2 = 0.91, RMSE = 1.7 t ha–1) and crop N uptake (r2 = 0.96, RMSE = 15 kg N ha–1) under different N management strategies as well as SOC level (r2 = 0.51, RMSE = 1 t C ha–1) and soil mineral N (r2 = 0.58, RMSE = 56 kg N ha–1). No difference in crop biomass production was found with N fertilizer reductions up to 27% of the conventional N fertilizer rate when combined with OA application. A 12-yr scenario analysis using SALUS indicated that conventional N fertilizer can be further reduced by up to 50% while sustaining crop biomass production, thereby potentially reducing N losses to the environment. Data gathered from the field study and simulation scenarios highlighted the positive effect of composted OA to maintain soil C levels. This contrasts with average annual SOC losses of 3.7% observed in long-term simulation scenarios in systems with only N fertilizer or raw OA applications.

### Abbreviations

Co, composted manure; CONV, conventional nitrogen fertilizer rate; Ma, raw manure; OA, organic amendments; PAN, plant available nitrogen; RMSE, root mean squared error; Rd, reduced nitrogen fertilizer rate; SOC, soil organic carbon

Nitrogen is the primary plant macronutrient that commonly limits crop production. Farmers generally supply N to agricultural soils by applying synthetic N fertilizer (N fertilizer) or OA such as raw and composted plant and animal manures. In vegetable farming systems, the application of OA is often associated with supplemental addition of N fertilizer due to uncertainty in accounting for the N that is released from these products over time (De Rosa et al., 2016). Significant N surplus can result from the tendency to underestimate the N contribution of OA to fertilizer rates, especially in vegetable cropping systems where high N inputs are often applied (>220 N kg ha–1 per season (Diao et al., 2013, Rezaei Rashti et al., 2015). The tendency of horticultural producers to overfertilize has been confirmed in a survey conducted by Chan et al. (2007).

Excessive N application in a farmer’s field may increase nitrous oxide emissions to the atmosphere (Smith et al., 2007) and nitrate leaching to the ground water (Giola et al., 2012). In contrast, a balanced application of OA and N fertilizer that precisely meets the N demands of a crop improves N use efficiency and reduces the potential for environmental pollution. This has been demonstrated in the production of rice (Oryza sativa L.) (Liu et al., 2008) and corn (Zea mays L.) (Efthimiadou et al., 2010) and results in more sustained release of N in the soil due to the higher soil organic matter levels after repeated application of OA. Soil organic carbon (SOC) plays a pivotal role in sustaining crop production (Lal et al., 2004) by conserving soil chemical, biological and physical fertility, by increasing soil water retention and nutrient availability, and by reducing the risk of soil erosion and crust formation (Diacono and Montemurro, 2010). The intensive tillage required to cultivate horticultural crops (Scheer et al., 2017) has been addressed as of one of the factors that promotes degradation of SOC (Lal et al., 2004) while the application of OA has been shown to increase SOC levels (Maillard and Angers, 2014).

The strategic application of OA and N fertilizer in horticultural cropping systems, however, is complex because the growth cycle of specific crops within these systems is comparatively short, 2 to 5 mo, and the N demand of these crops is relatively high (Hernández et al., 2014).

For an annual vegetable crop rotation, De Rosa et al. (2016) demonstrated that it is possible to substitute a proportion of the synthetic N fertilizer requirement with OA and at same time maintain the soil plant available nitrogen (PAN) close to the regional N fertilizer recommendation rates. This analysis was completed using N mineralization rates from the literature. However, it has been reported that only about 25% (Pettygrove and Heinrich, 2009) of the organic N added to soil is released in the first year after application (Diacono and Montemurro, 2010). This demonstrates the difficulty in developing long-term strategies for combined OA and N fertilizer application because the cumulative residual effect of OA can last several years. Moreover, N dynamics in the soil after repeated OA application can be influenced by managements practices (e.g., tillage intensity) and climatic conditions that can vary greatly from 1 yr to another (Chalhoub et al., 2013). Crop simulation models can be used to assess the long term residual effects and temporal interactions between the application of OA and climate on crop production, soil PAN, and SOC. Models have been extensively tested in cereal production to develop best N management practice (Basso et al., 2016; Chalhoub et al., 2013; Giola et al., 2012), but there is only limited data on the long- term effects of combined application of OA and N fertilizer on vegetable production.

The rationale behind this research was that accounting for the release of N from OA over time will improve N use efficiency and promote soil C storage in horticultural soils. The objectives of this study were to: (i) estimate the long-term impact on vegetable production and soil PAN when a proportion of synthetic N fertilizer is replaced with different forms of OA (Co and Ma) when compared to the standard management practice (OA applied in addition to the conventional N fertilizer rate); (ii) identify the best combination of OA and N fertilizer application rates so that N surplus above crop demand is minimized; (iii) evaluate the ability of the SALUS crop model (Basso and Ritchie, 2015) to reproduce experimental results for crop biomass production and soil PAN dynamics under organic and synthetic N fertilization strategies, (iv) assess the long-term effect of repeated applications of different OA forms (Co and Ma) on SOC levels. Calibration and validation of the SALUS crop model were completed with data from an intensively managed vegetable crop rotation where there were repeated OA applications over a period of 2 yr and 6 mo. We also compared long-term scenarios for series of standard OA farm practice and optimized OA application. The field and simulation results identified the potential benefits of using OA as a valid substitute for synthetic N fertilizer in the long term.

## MATERIALS AND METHODS

### Site Description

Data used for the simulations were obtained from a field study that was conducted from September 2013 to February 2016 at Gatton Research Station in the Lockyer Valley, Queensland, Australia (27°32′56″ S, 152°19′39″ E; 100 m above sea level). The Lockyer Valley is a major vegetable producing region in southeastern Queensland, characterized by a humid subtropical climate, with warm to hot summers and mild to cool winters. Annual average temperature is 20°C with an average annual maximum of 27°C and a minimum of 13°C. Summer rainfall accounts for nearly 60% of the total annual rainfall of 787 mm (Bureau of Meteorology, Australia). Chemical and physical characteristics of the soil were determined before the commencement of the experiment from soil samples collected at four random locations across the trial site and analyzed following the procedures described in Rayment and Lyons (2011). The alluvial soil of the study area is classified as a Black Vertisol (FAO, 1998) with 600 g clay kg–1, 220 g silt kg–1, and 180 g sand kg–1. This soil contained no inorganic C (HCl, Chaney et al., 1982) in the uppermost soil layer (0–0.2 m), 0.15 g organic C kg–1, 0.012 g total N kg–1 (analyzed with a Leco Trumac CNS Analyzer, LECO Corporation, St. Joseph, MI), had a pH (H2O) of 7.8, and cation exchange capacity (CEC) of 43.9 cmolc kg–1. The barley (Hordeum vulgare L.) cover crop grown prior to the experiment was cut and removed to minimize any residual N from the cultivation of previous crops.

## Field Management and Experimental Set-up

### Crop Rotation

The crop rotation we investigated was a succession of seven vegetable crops grown on raised beds. The first year’s rotation was green bean (Phaseolus vulgaris L.) followed by sorghum [Sorghum bicolor) (L.) Moench] as a catch crop, broccoli (Brassica oleracea var. Italica) and lettuce (Lactuca sativa L.). The second year’s rotation included sweet corn (Z. mays var. saccarina) followed by broccoli and lettuce and finally sweet corn for the last season (6 mo) of the entire crop rotation. The experimental timeline is shown in Table 1.

View Full Table | Close Full ViewTable 1.

Timeline of crop rotation and details of organic amendments (OA) and N fertilizer management for the field treatments at Gatton Research Facility (2013–2016). CONV = conventional nitrogen fertilizer application rate; Ma = raw manure; Co = composted manure; Rd = reduced nitrogen application rate; 0N = no nitrogen application.

 Management Treatments First year CONV Ma+CONV Co+CONV Ma+Rd Co+Rd 0N 02/09/2013 OA application, t dry wt. ha–1 1.4 12.5 1.4 12.5 Green bean 20/09/2013 Planting+Basal, kg N ha–1 55 55 55 20 20 08/11/2013 Side dressing, kg N ha–1 35 35 35 28 24 2–6/12/2013 Harvest Sorghum 10/12/2013 Planting 17/01/2014 Incorporation Broccoli 11/02/2014 Planting+Basal, kg N ha–1 42 42 42 35 34 14/03/2014 Side dressing, kg N ha–1 39 39 39 36 37 2/04/2014 Side dressing, kg N ha–1 39 39 39 37 37 12–21/04/2014 Harvest Lettuce 10/06/2014 Planting+Basal, kg N ha–1 54 54 54 50 51 09/07/2014 Side dressing, kg N ha–1 23 23 23 23 23 1/08/2014 Side dressing, kg N ha–1 23 23 23 23 23 2/09/2014 Harvest N fertilizer, kg N ha–1 310 310 310 252 249 N fertilizer reduction 57 (–18%) 60 (–20%) Second year CONV Ma Co Ma+Rd Co+Rd 0N 11/09/2014 OA application, t dry wt. ha–1 1.85 9.2 1.85 9.2 Sweet corn 24/09/2014 Planting+Basal, kg N ha–1 72 21 15 22/10/2014 Side dressing, kg N ha–1 48 37 36 10/12/2014 Harvest Broccoli 17/2/2015 Planting+Basal, kg N ha–1 42 28 28 17/03/2015 Side dressing, kg N ha–1 39 33 35 2/04/2015 Side dressing, kg N ha–1 39 33 35 17–29/04/2015 Harvest Lettuce 10/06/2015 Planting+Basal, kg N ha–1 54 47 47 9/07/2015 Side dressing, kg N ha–1 23 23 23 4/08/2015 Side dressing, kg N ha–1 23 23 23 4/09/2015 Harvest N fertilizer kg N ha–1 340 245 242 N fertilizer reduction, kg N ha–1 95 (–27%) 98 (–29%) Third year CONV Ma Co Ma+Rd Co+Rd 0N 16/09/2015 OA application, t dry wt. ha–1 3.05 12.91 3.05 12.91 Sweet corn 28/09/2015 Planting+Basal, kg N ha–1 72 55 38 30/10/2015 Side dressing, kg N ha–1 48 29 31 15/12/2015 Harvest N fertilizer, kg N ha–1 120 84 69 N fertilizer reduction, kg N ha–1 36 (–30%) 51 (–42%) Crop rotation Total N fertilizer, kg N ha–1 770 581 560 Total N fertilizer reduction, kg N ha–1 189 210 Variation, % –24% –27%

### Field Treatments and Agronomic Management

The field treatments were set up in a randomized block design with four replicates (plot size 1.5 by 10 m with 1.5 m buffer) as follow:

• (0N)Zero nitrogen was used to account for background soil mineralization;

• (CONV) Conventional nitrogen fertilizer rate applied as urea (770 kg of N fertilizer ha–1 rotation–1) based on local farm management;

• (Co+Rd) Composted chicken manure plus reduced nitrogen fertilizer rate;

• (Ma+Rd) Raw chicken manure plus reduced nitrogen fertilizer rate;

• (Co+CONV) Composted chicken manure plus conventional nitrogen fertilizer rate only in the first year of study;

• (Ma+CONV) Raw chicken manure plus conventional nitrogen fertilizer rate only in the first year of study;

• (Co)Composted chicken manure only for the second and third year. This treatment replaced Co+CONV;

• (Ma)Raw chicken manure only in the second and third year of study. This treatment replaced Ma+CONV.

The treatments designated as Co and Ma only commenced in the second year to assess the ability of the SALUS model to accurately simulate the residual effect of the OA applied in the first year on soil N mineralization.

The OA were applied in September prior to the planting of green bean in the first year (2013) with the rate calculated to provide 35 kg ha–1 of PAN (NO3 + NH4+) across treatments. The OA application rates in September 2014 and 2015 prior to the planting of sweet corn were calculated to provide approximately 30% of the CONV annual crop rotation N requirements. This calculation was obtained by considering the initial fraction of PAN in the OA and the estimated mineralized N from OA over time (Table 2).

View Full Table | Close Full ViewTable 2.

Composition of chicken manure (raw manure, Ma) and composted chicken manure (composted manure, Co) on dry weight basis. Organic amendments application rates with predicted amount of mineral N released over time.

 Variable First year Second year Third year Ma Co Ma Co Ma Co H2O, % 60 34 69 34 45 30 Organic C, % 24.2 23 24.9 21.5 23.2 31.2 Total N, % 7.2 1.9 8.3 3.12 6.40 2.62 C/N ratio 3.3 12 3 6.9 3.62 11.9 NO3––N, mg kg–1 115 93.9 154 130 122 107 NH4+–N, mg kg–1 24,845 2819.7 27,190 6012 4929 2451 Application organic amendments, t dry wt. ha–1 1.4 12.5 1.85 9.2 3.05 12.91 Organic C, kg N ha–1 340 2884 460 1978 707 4027 Total P, kg P ha–1 28 278 40 218 42 406 Total K, kg K ha–1 91 431 157 312 112 492 Total N, kg N ha–1 102 240 154 287 195 337 Organic N, kg N ha–1 67 205 104 231 180 304 Inorganic N (PAN, NO3−–N + NH4+–N) applied at basal, kg N ha–1 35 35 51 56 15 33 Quartile 3-mo Mr, %† Estimated mineralized PAN, kg N ha–1‡ Ma Co Ma Co Ma Co Ma Co t1 10 5 6.7 10.25 10.4 11.55 18 15.2 t2 10 4 6.7 8.2 10.4 9.24 t3 8 2 5.36 4.1 8.32 4.62 t4 5 1 3.35 2.05 5.2 2.31 t5 1.5 0.8 1.005 1.64 1.56 1.848 t6 5.3 2.1 3.551 4.305 t7 5.3 2.1 3.551 4.305 t8 3 1.2 2.01 2.46 t9 1 0.6 0.67 1.23 t10 1.8 1 1.206 2.05 Estimated mineralized N from OA, kg N ha–1 34.1 40.6 35.9 29.6 18 15.2 Crop rotation Estimated total PAN supplied by OA, kg N ha–1 Ma 189 Co 210
Mr = estimated mineralization rate for the considered quartile (tx) derived from De Rosa et al. (2016); Eghball et al. (2002); Hartz et al. (2000).
PAN = plant available nitrogen.

The PAN delivered by OA during the cropping seasons, was estimated by multiplying the amount of organic N applied at OA incorporation, with mineralization coefficients (Mr) as shown in Table 2 (De Rosa et al., 2016; Eghball et al., 2002; Hartz et al., 2000) divided into four (3-mo) quartiles per year (a total of 10 quartiles for 2.5 yr) using Eq. [1]:where PAN is plant-available nitrogen, t is the annual quartile considered (PAN applied at basal OA application corresponds to t = 0, Table 2) and y is the reference of previous yearly OA application (applicable only for second and third years). The Mr used for the temporal PAN calculation and the estimated cumulative PAN delivered by OA during the entire crop rotation are shown in Table 2. The reduced N fertilizer rate (+Rd) at basal OA application was calculated by subtracting the initial fraction of PAN from the CONV N fertilizer rate while the +Rd rate at each top dressing fertilization event was calculated by subtracting the PAN delivered by each annual OA application from the CONV N fertilizer application rate as shown in Fig. 1 and using Eq. [2].where +Rd is the reduced fertilizer rate, f indicates the N fertilization event, CONV is the N fertilizer rate applied following the conventional farm practice and PAN is the estimated plant available nitrogen delivered by OA.

Fig. 1.

Conceptual model for integrated N supplied from organic amendments and mineral fertilizer in an intensive vegetable crop rotation. The solid black line (CONV) represents conventional N fertilizer application. Solid areas represent the potential N released over the time by OA. Area with forward diagonal lines indicate the supplementary N fertilizer application (+Rd).

The total estimated PAN for the entire crop rotation supplied by OA was 189 and 210 kg ha–1 rotation–1 for Ma and Co, respectively (Table 2). The CONV treatment rate of 770 kg N fertilizer ha–1 rotation–1 was consequently reduced to 581 kg N fertilizer ha–1 rotation–1 (–24%) for the Ma+Rd treatment and 560 kg of N fertilizer ha–1 rotation–1 (–27%) for Co+Rd (Table 1). A basal N fertilizer application was incorporated to 20-cm soil depth at the beginning of each crop while top-dressed N fertilizer was surface applied, and followed immediately by irrigation. The timing and amount of N fertilizer applied to each crop in the rotation for the entire crop rotation are listed in Table 1.

Phosphorus and K were added at basal fertilizer application for each crop in the form of K sulfate and super phosphate at the rates of 171 kg P ha–1 and 460 kg K ha–1 for the rotation, respectively.

Plots were typically irrigated weekly depending on rainfall with a single application of 30 mm from an overhead sprinkler irrigation system according to standard farming practice. Crop residues were incorporated at the end of each crop growing phase with a rotary hoe to 20-cm depth.

## Field Data Collection

Soil samples at four depths (0–0.2, 0.2–0.4, 0.4–0.7, and 0.7–1 m) were collected at four random locations across the trial site prior to OA application and analyzed for pH, CEC, texture, total N, and organic C. To monitor mineral N levels in the topsoil (0–0.15 m), four subsamples per plot were collected on a weekly to fortnightly basis both during the crop growing seasons and before and after incorporation of crop residues. All samples collected were analyzed for NO3 and NH4+ by a commercial laboratory using a 1:5 soil/2 M KCl extract. Prior to each OA application and at the end of the field experiment four subsamples per plot were collected and analyzed for total N and soil organic carbon (SOC) using a Leco Trumac CNS Analyzer (LECO Corporation, St. Joseph, MI).

### Crop Yield and Biomass Production

Yield and crop residue biomass at each harvest event were measured by collecting whole plant samples for each treatment. A representative plant sample per plot was oven-dried for 24 to 48 h at 70°C, and subsequently ground and analyzed for total N and C content with a Leco Trumac CNS Analyzer (LECO Corporation).

## Crop Simulation Model

The SALUS model (System Approach to Land Use Sustainability [Basso et al., 2010; Basso and Ritchie, 2015]) was used to assess the long-term effect on soil PAN of the standard and a rationalized combination of chemical N fertilizer and organic amendment. The SALUS model is designed to continuously simulate soil and plant, water and N interactions under different management strategies on a daily time-step for multiple and consecutive crop cycles (Basso et al., 2010). The data required to run the model include: (i) daily weather values for maximum and minimum temperature, rainfall, and solar radiation; (ii) soil data (by layer) including texture, organic matter, organic C, pH, total N, mineral N, and water holding characteristics; and (iii) field/crop management information regarding tillage, planting, harvest, organic residues, N fertilizer application, and irrigation.

The SALUS-Simple crop model (hereafter referred to as SALUS) is based on the ALMANAC simulation model (Kiniry et al., 1992). Crop growth is based on a growing degree days approach, the development of leaf area index (LAI) is S-shaped (based on plant-specific parameters), and biomass production is based on the radiation use efficiency approach (Dzotsi et al., 2013).

The current version of SALUS has only been parameterized for sorghum and sweet corn, so green bean, broccoli, and lettuce were parameterized using field data that included LAI, biomass production, and cumulative thermal time from planting to maturation collected over the entire crop rotation. The parameters for each crop development stage were determined by plotting the normalized thermal time (0 for planting to 1 complete maturation) values against the normalized LAI development curve (S-shaped) using the equation described in Dzotsi et al. (2013).

The SALUS soil organic matter (SOM) and N modules originated from the Century model (Parton et al., 1987) with several modifications to run on daily time-steps. The immobilization/mineralization of SOM and N are simulated from the labile, intermediate and resistant SOM pools, which vary in their turnover and respective C/N ratio. Exogenous fresh organic matter (i.e., crop residues and organic amendments) is partitioned into structural and metabolic pools, to represent recalcitrant and easily decomposable residues based on the residue lignin cellulose/hemicellulose and N fractions. The decomposition rate of this organic matter, and therefore N mineralization, is influenced by soil texture, moisture, and temperature as well as tillage factors. Tillage intensity and the type of implement used for tillage directly influences soil structure and the rate at which fresh organic matter decomposes. For this study the SOC pools were initialized with procedures described in Basso et al. (2011a) so that it was comprised of labile (5%), intermediate (65%), and resistant (35%) pools.

### SALUS Model Simulation Scenarios

Simulations were performed in a continuous sequential mode to account for the carry-over effect of N from 1 yr to the next. The scenarios simulated were:

• Crop model validation- Initial simulations were completed for the period of the experiment field rotation (3-yr crop rotation) to evaluate the model’s ability to simulate measured field data;

• Long-term assessment (12 yr) of the influence of (i) standard OA farm practice (Ma+CONV and Co+CONV), (ii) strategic combination of N fertilizer and OA (Ma+Rd and Co+Rd), and (ii) conventional N fertilizer rate (CONV) on soil PAN, SOC content, and crop biomass response. These simulations evaluated four consecutive 3-yr crop rotations using historical weather data (2004–2016) obtained from a station adjacent the experimental field.

• A third set of long-term rotational simulations (12 yr) were completed to assess the effect of further N fertilizer reductions from OA treatments under +Rd N fertilizer rate (Co+Rd and Ma+Rd). The optimized reduction (Ma+Opt; Co+Opt) from the +Rd treatments (Ma+Rd; Co+Rd) consisted of a nil application of N fertilizer in the first crop following the annual OA application in green bean for the first year and sweet corn for the second and third years of the crop rotation. Consequently, the Ma+Rd N fertilizer rate (581 kg N ha–1 rotation–1) was further reduced in the Ma+Opt treatment by 190 kg N ha–1 rotation–1 and the Co+Rd N fertilizer rate (560 kg N ha–1 rotation–1) was further reduced in the Co+Opt by 164 kg N ha–1. This resulted in N fertilizer reductions when compared to the CONV treatment (770 kg N fertilizer ha–1 rotation–1) of 49% for Ma+Opt (total N fertilizer applied 391 kg N fertilizer ha–1 rotation–1) and 48% for Co+Rd (total N fertilizer applied 396 kg N fertilizer ha–1 rotation–1). The +Opt N fertilizer rate was selected based on (i) the carry-over effect of consecutive OA applications; (ii) the mineralization of high N content of crop residues that were re-incorporated (i.e., broccoli and lettuce); and (iii) the applicability of the N management practice at the farm scale.

### Statistical Analysis

The Tukey post hoc test was used to evaluate field data to determine the influence of treatment on crop biomass accumulation and soil mineral N. Post hoc tests were only performed when the ANOVA yielded P values < 0.05. Differences in temporal mineral N dynamics between OA treatments and the CONV treatment were calculated as differentials (Δs) for each soil sampling event by subtracting the OA soil PAN (NO3–N + NH4+–N) values from the CONV value and by same block position.

Model performance was evaluated using the root mean square error (RMSE) which calculates the total error between observed and simulated values as a percentage. The RSME was calculated with the following equation:where μObs is the average of all measurements, Obs are the observed values in the ith measurement, Sim are the simulated values, and n the number of observations.

The association between measured and simulated values was assessed by calculating the coefficient of determination (r2) and the significance of association value (F) compared to the P values (P = 0.05) given in table of F distribution with the following equation:where n is the number of simulated and measured pairs compared. Statistical analyses and graphical presentations were performed in R environment (R Core Team, 2015).

## RESULTS

### Assessment of Nitrogen Fertilization Strategy on Temporal Trend of Soil Plant Available Nitrogen and Soil Organic Carbon

The 95% confidence intervals (CI) of the differentials (Δs) of soil PAN calculated between OA treatments and CONV are presented in Fig. 2. Null Δs PAN values for OA+Rd treatments (Δs PAN of 0 kg N ha–1 within OA+Rd Δs PAN ± CI) were obtained on multiple occasions over the entire crop rotation (Fig. 2). Significantly higher PAN Δs (OA+Rd Δs PAN ± CI > 0) were observed for the +Rd treatments after annual OA applications in September 2013, 2014, and 2015. During these periods, PAN Δs ranged from 163 to 290 kg ha–1 for Ma+Rd, and 62 to 149 kg of PAN ha–1 for Co+Rd (Fig. 2). This resulted in an average PAN Δs for the entire crop rotation of +15.7 and +15.0 kg ha–1 for Ma+Rd and Co+Rd, respectively.

Fig. 2.

The 95% confidence intervals (CI) of the differentials (Δs) of soil plant available nitrogen (PAN) content (kg N ha–1) calculated by difference between treatments that received organic amendments (OA) and CONV at Gatton Research Facility, Queensland (Australia). Arrows indicate annual application of OA. Positive values indicate that soil PAN obtained in OA treatments are superior to CONV treatment. CONV = conventional N fertilizer application rate; Ma = raw manure; Co = composted manure; Rd = reduced nitrogen application rate.

Soil PAN determined in OA treatment plots that received the +CONV amount of urea fertilizer in the first year were, on average, higher than the CONV treatment (OA+CONV Δs PAN ± CI > 0) and showed yearly Δ averages of +24.8 and +27.5 kg PAN ha–1 for Co+CONV and Ma+CONV, respectively. The Ma treatment showed positive PAN Δs of +180 to +151 kg PAN ha–1 in the periods following the second and third OA application, respectively. In the Co treatment, positive PAN Δs were only observed in the second year (+244 kg PAN ha–1) while negative PAN Δs were observed after the third OA application (–268 kg PAN–1). Negative PAN Δs in the second and third cropping year for Ma and Co (OA only) were observed in the periods long after the annual OA application. Negative values for the treatments that received OA only, ranged from –165 to –6 kg PAN Δ ha–1 for Ma and –268 to –5 kg PAN Δ ha–1 for Co. The temporal trend of the actual soil PAN data for five treatments during the experimental period are shown in Fig. 3.

Fig. 3.

Measured (black dots) and simulated (red lines) soil plant available nitrogen (PAN) temporal trend (0–15 cm) for five treatments during the experimental period at Gatton Research Facility, Queensland (Australia). CONV = conventional nitrogen fertilizer application rate; Ma = raw manure; Co = composted manure; Rd = reduced nitrogen application rate.

Soil organic C increased across treatments after 3 yr of OA application from an initial value of 27.5 t C ha–1 in September 2013 to a maximum value of 33.4 t C ha–1 (Co+Rd) in January 2016 (Table 3). However, the SOC increase between September 2013 and January 2016 was only significant in treatments that received composted OA. The SOC content (0–0.15 m) values in January 2016 ranged from a low of 28.4 t C ha–1 in the Ma treatment to 33.4 t C ha–1 in the Co+Rd treatment which was statistically higher than all other treatments (Table 3).

View Full Table | Close Full ViewTable 3.

Observed (mean ± SD) and simulated annual soil organic carbon (SOC) using SALUS model for the Gatton Research Facility experiment (2013–2015). Means denoted by a different letter indicate significant differences between treatments (p < 0.05). Asterisks (*) indicates significant difference at p < 0.05 for SOC values between September 2013 and January 2016. Root mean squared error (RMSE) and coefficient of determination (r2) between observed and simulated values. CONV = conventional nitrogen fertilizer application rate; Ma = raw manure; Co = composted manure; Rd = reduced nitrogen application rate; 0N = no nitrogen application.

 Treatment SOC (0–0.15 m) Measures Simulated t C ha–1 September 2013 27.5 ± 1.3 27.8 September 2014 CONV 25.9 ± 1.12 28 Ma+CONV 27.5 ± 1.30 28.2 Ma+Rd 26.5 ± 1.30 28.2 Co+CONV 28.3 ± 1.90 27.8 Co+Rd 28.7 ± 1.90 28.5 September 2015 CONV 27.8 ± 1.3 28 Ma 28.9 ± 1.7 27.7 Ma+Rd 27.9 ± 1.9 28 Co 30.5 ± 2.9 28.8 Co+Rd 31.0 ± 2.0 28.3 January 2016 CONV 28.6 ± 0.8b 28.5 Ma 29.6 ± 2.0b 30.5 Ma+Rd 28.4 ± 1.9b 27.7 Co 31.2 ± 1.2ab* 29.4 Co+Rd 33.4 ± 1.7a* 31.2 RMSE, % 4.7 F 6.9 Fcritic (0.05) 4.1 r2 0.52

### Assessment of Plant Available Nitrogen on Crop Biomass Production and Crop Nitrogen Uptake

No significant differences were observed in annual aboveground biomass production between all treatments that received N over the 3 yr (Table 4).

View Full Table | Close Full ViewTable 4.

Observed and simulated annual aboveground (AG) crop biomass production and crop N uptake (mean ± SD) at Gatton Research Facility (2013–2016). Means denoted by a different letter indicate significant differences between treatments (p < 0.05). Root mean squared error (RMSE) and coefficient of determination (r2) between observed and simulated values. CONV = conventional nitrogen fertilizer application rate; Ma = raw manure; Co = composted manure; Rd = reduced nitrogen application rate; 0N = no nitrogen application.

 Treatments AG plant biomass AG plant N uptake Measured Simulated Measured Simulated t dry wt. ha–1 kg N ha–1 First year CONV 16.7 ± 0.88a 17 606 ± 65a 591 Ma+CONV 18 ± 1a 17.1 631 ± 46a 621.4 Co+CONV 18 ± 1.6a 16.9 654 ± 51a 629.2 Ma+Rd 17 ± 1.8a 17.1 599 ± 71a 560.3 Co+Rd 17.5 ± 1.3a 17.4 655 ± 7a 566.3 0N 9.3 ± 3.4b 10.7 271 ± 52b 271 Second year CONV 28 ± 8a 25.9 505 ± 67a 523 Ma 25 ± 2ab 24.3 412 ± 89ab 460 Co 26 ± 4ab 23.6 439 ± 155ab 446 Ma+Rd 23 ± 3ab 25.8 487 ± 67a 579 Co+Rd 27 ± 4a 25.9 560 ± 102a 525 0N 15 ± 2b 16.9 188 ± 62b 256 Third year CONV 16 ± 2a 16.5 202 ± 21abc 220 Ma 12 ± 1.2a 16 136 ± 33bcd 190 Co 16 ± 2a 16.8 122 ± 50cd 190 Ma+Rd 15 ± 3a 16.5 250 ± 45a 233 Co+Rd 18 ± 5a 16.9 220 ± 49ab 238 0N 11 ± 6a 9.5 70 ± 39d 90 RMSE, % 9.1 11.8 F 1.3.6 14.4 Fcritic (0.05) 4.1 4.1 r2 0.91 0.96

Except for the 0N treatment, no significant differences were observed in annual cumulative biomass production between all treatments in the first year’s crop rotation. First year biomass production ranged between 9.3 t ha–1 for 0N to 18 t ha–1 for Co+CONV. In the second year, 0N (15 t ha–1) plant biomass production was statistically lower than CONV (28 t ha–1) and Co+Rd (27 t ha–1). No significant differences between treatments were observed in the third year where biomass production ranged from 11 t ha–1 (0N) to 16 t ha–1 (CONV).

Aboveground crop N uptake in the first year ranged from 271 kg N ha–1 for 0N to 655 kg N ha–1 for Co+Rd. In the second and third years, Ma and Co were not statistically different from 0N. Crop N uptake ranged from 188 to 560 kg N ha–1 for the 0N and Co+Rd in the second year and from 70 kg N ha–1 for 0N to 250 kg N ha–1 for Ma+Rd in the third year. Detailed crop biomass production and N uptake for each crop in the rotation are presented in Table 5.

View Full Table | Close Full ViewTable 5.

Observed and simulated aboveground (AG) crop biomass, and AG pant N uptake (mean ± SD). Means denoted by a different letter indicate significant differences between treatments (p < 0.05). CONV = conventional nitrogen fertilizer application rate; Ma = raw manure; Co = composted manure; Rd = reduced nitrogen fertilizer application rate; 0N = no nitrogen application.

 Variable Treatments First year CONV Ma+CONV Co+CONV Ma+Rd Co+Rd 0N Green bean AG plant biomass, t dry wt. ha–1 5 ± 0.35 (5.2)† 5.4 ± 1 (5.1) 5.3 ± 1.7 (5.1) 5 ± 1.8 (5.1) 4.7 ± 0.7 (5.1) 5.6 ± 2 (5.1) AG plant N uptake, kg N ha–1 155 ± 17 (144) 135 ± 26 (143.7) 130 ± 40 (143.2) 165 ± 25 (143.3) 139 ± 7 (143.3) 136 ± 42 (142) Sorghum AG plant biomass, t dry wt. ha–1 3.7 ± 1.5 (3.5) 4.5 ± 0.4 (3.6) 4.5 ± 0.3 (3.5) 3.9 ± 1.6 (3.6) 4.4 ± 1.1 (4) – AG plant N uptake, kg N ha–1 125 ± 70 (150) 162 ± 11 (170) 154 ± 30 (170) 116 ± 54 (114) 154 ± 21 (127) – Broccoli AG plant biomass, t dry wt. ha–1 4.8 ± 0.54a (5.1) 5 ± 0.6a (5.2) 5.3 ± 0.8a (5) 4.6 ± 0.8a (5) 5.2 ± 0.7a (5.1) 1.83 ± 0.6b (2.8) AG plant N uptake, kg N ha–1 204 ± 35a (176) 201 ± 31a (185.7) 244 ± 39a (190) 189 ± 14a (175) 212 ± 29a (173) 49 ± 15b (64) Lettuce AG plant biomass, t dry wt. ha–1 3.2 ± 0.13 (3.2) 3 ± 0.26 (3.2) 3.2 ± 0.23 (3.3) 3.3 ± 0.1 (3.4) 3 ± 0.42 (3.2) 2.7 ± 0.2 (2.8) AG plant N uptake, kg N ha–1 134 ± 11a (121) 126 ± 10a (122) 132 ± 10a (126) 137 ± 2a (128) 129 ± 15a (123) 70 ± 11b (65) Second year CONV Ma Co Ma+Rd Co+Rd 0N Sweet corn AG plant biomass, t dry wt. ha–1 20 ± 8.2 (18.4) 20.1 ± 2 (18.4) 19.1 ± 2 (17.8) 15.4 ± 3 (18.5) 19.8 ± 3.5 (18.5) 11.5 ± 2 (11.9) AG plant N uptake, kg N ha–1 271 ± 93 (248) 319 ± 87 (261) 282 ± 80 (250) 253 ± 47 (308) 311 ± 60 (260) 122 ± 60 (111) Broccoli AG plant biomass, t dry wt. ha–1 5.2 ± 0.6a (4.5) 3.6 ± 1ab (2.9) 3.9 ± 1.23ab (2.8) 4.6 ± 0.3a (4.1) 5.2 ± 0.4a (4.2) 2.5 ± 0.2b (2) AG plant N uptake, kg N ha–1 150 ± 28a (152) 95 ± 56ab (109) 106 ± 47ab (76) 158 ± 15a (139) 178 ± 22a (134) 41 ± 2b (55) Lettuce AG plant biomass, t dry wt. ha–1 3 ± 0.4a (3) 1.7 ± 0.4c (3) 2 ± 0.4bc (3) 2.7 ± 0.4ab (3.2) 3 ± 0.25a (3.2) 1.7 ± 0.2c (3) AG plant N uptake, kg N ha–1 84 ± 18a (123) 32 ± 16b (90) 38 ± 12b (120) 76 ± 9a (132) 77 ± 17a (131) 25 ± 3b (90) Third year CONV Ma Co Ma+Rd Co+Rd 0N Sweet corn AG plant biomass, t dry wt. ha–1 16 ± 2 (16.5) 12 ± 1.2 (16) 16 ± 2 (16.8) 15 ± 3 (16.5) 18 ± 5 (16.9) 11 ± 6 (9.5) AG plant N uptake, kg N ha–1 202 ± 21abc (220) 136 ± 33bcd (190) 122 ± 50cd (190) 250 ± 45a (233) 220 ± 49ab (238) 70 ± 39d (90)
Numbers in brackets () are simulated values.

## SALUS Simulation

### Model Testing

The comparison between simulated and measured crop biomass production resulted in a model total error of 1.6 t ha–1 (RSME 9.1%) and 46.1 kg N ha–1 (RSME 11.8%) for aboveground plant biomass and crop N uptake, respectively (Table 4). The F values for crop biomass production (F = 13.6, r2 = 0.91) and N uptake (F = 14.4, r2 = 0.96) were greater than the critical F value at P = 0.05 (4.1), indicating significant association between the measured and predicted values. Simulated values for each crop in the rotation were within the measured average ± SD values except for lettuce biomass production in the second year for the Co, Ma, and 0N treatments (Table 5). The SALUS model reproduced both the magnitude and temporal pattern of soil PAN over the entire crop rotation (Fig. 3 and 4). The calculated r2 for the entire crop rotation was 0.58 with an F value of 115 vs. Fcritic of 3.9 (Fig. 4). Discrepancies between measured and simulated values were associated with those dates where wide confidence intervals were associated with the measured values (e.g., after the corn residues incorporation in the second year crop rotation) as shown in Fig. 3. Simulated SOC values were in accordance with observed values, the average RMSE calculated was 4.7% (1.4 t C ha–1) and an F value of 6.9 vs. Fcritic of 4.1 (r2 = 0.51) as shown in Table 3.

Fig. 4.

SALUS validation of soil plant available nitrogen (PAN) for the 2 yr and 6 mo crop rotation at Gatton Research Facility, Queensland (Australia). Red line indicates y = x.

### Simulation Results of Rotational Scenarios

Long-term simulation scenarios produced consistently higher soil median PAN in those treatments that received OA over the four consecutive 3-yr crop rotations (Table 6). The median soil PAN for each 3-yr crop rotation was used to better represent typical soil PAN value and to negate the effect of the highly skewed frequency distribution of data points. The average of these median values was then taken to determine differences in soil PAN between treatments. The Co+CONV treatment was characterized by the highest average median soil PAN (73.2 kg N ha–1) and the +Opt N fertilizer treatments had the lowest (57.8 kg N ha–1 for Co+Opt and 49.2 kg N ha–1 for Ma+Opt). Treatments under +Rd N fertilizer rate showed average median PAN values of 70.2 and 62.3 kg N ha–1 for Co+Rd and Ma+Rd, respectively.

View Full Table | Close Full ViewTable 6.

Predicted median soil plant available nitrogen (PAN) content (0–15 cm) for the four consecutive crop rotations (3 yr) using SALUS model for long-term simulation scenarios. CONV = conventional nitrogen fertilizer application rate; Ma = raw manure; Co = composted manure; Rd = reduced nitrogen application rate; Opt = optimized nitrogen application rate.

 Rotation Long-term scenarios CONV Ma+Rd Co+Rd Ma+CONV Co+CONV Ma+O Co+Opt Median PAN, kg N ha–1 1 44 66.5 83.9 73.2 83.0 50.6 67.2 2 57.4 76.6 78.5 77.0 81.4 57.3 64.9 3 44.6 57.9 66.0 62.7 72.6 46.3 50.1 4 38.3 48.0 52.2 49.6 55.8 42.5 48.8 Average 46.1 62.3 70.2 65.6 73.2 49.2 57.8 SD 7.0 10.5 12.2 10.6 10.8 5.5 8.3

The trade-off between median soil PAN and crop biomass production in the long-term simulated scenarios was calculated as the difference between the six OA treatments and the CONV treatment (Fig. 5). The highest differential in average crop biomass production was observed in the Co+CONV treatment (+1.42 t ha–1 rotation–1) which showed the highest average median soil PAN Δ (+27.1 kg PAN ha–1 rotation–1). The Ma+CONV treatment resulted in a lower Δ biomass production and soil PAN Δ than Co+Rd (+0.9 vs.+1.2 t ha–1 rotation–1 and +19.5 vs. +24.0 kg PAN ha–1 rotation–1). The Ma+Opt showed marginal negative crop biomass production of only –0.2 t ha–1 rotation–1 and a soil PAN Δ of +3 kg PAN ha–1 rotation–1. On the other hand, the Co+Opt treatment showed positive Δ biomass production of +1.1 t ha–1 rotation–1 with a positive soil PAN Δ of +11.6 kg PAN ha–1 rotation–1 (Fig. 5). The highest Δ N crop uptake calculated as differences between the six OA treatments and CONV, was observed in the Co+CONV treatment with +180 kg N ha–1 rotation–1, followed by Co+Rd (+153 kg N ha–1), Ma+CONV (+146 kg N ha–1), and Ma+Rd (+114 kg N ha–1). Co+Opt and Ma+Opt showed the smallest differences in crop N uptake (+100 and –11 kg N ha–1 rotation–1 respectively).

Fig. 5.

Differentials of model simulation between organic amendments (OA) treatments and conventional nitrogen fertilizer application rate (CONV) of average median plant available nitrogen (PAN) vs. average crop biomass production with associated crop N uptake for the long-term rotation scenarios at Gatton Research Facility, Queensland (Australia) using the SALUS model. Positive values indicate that simulation values obtained for the OA treatments are superior to CONV treatment. Ma = raw manure; Co = composted manure; Rd = reduced nitrogen application rate; Opt = optimized nitrogen application rate.

Changes in SOC varied widely across the 12 annual OA treatments (Fig. 6). Soil organic carbon decreased (on average) by 2.2% per year in the CONV treatment. Treatments that received Ma (+Rd, +CONV, +Opt) also lost an average of 4.3% per year SOC. The average annual decrease in SOC in the Co treatments (0.1%) showed less variation than the Ma treatments over the same time period. Peaks of SOC were observed after incorporation of crop residues and OA.

Fig. 6.

Simulated evolution of soil organic carbon (SOC) (t ha–1) in different treatments for the long-term simulation scenarios at Gatton Research Facility, Queensland (Australia) using the SALUS model. CONV = conventional nitrogen fertilizer application rate; Ma = raw manure; Co = composted manure; Rd = reduced nitrogen application rate; Opt = optimized nitrogen application rate.

## DISCUSSION

The wide-ranging benefits to agricultural soils that are reported worldwide from OA application studies are often associated with application rates that are many times larger than what is necessary to supplement N chemical fertilizer (Quilty and Cattle, 2011). The +CONV treatments included in this study that represent the OA standard farm practice, showed no improvement in crop biomass production when compared to the +Rd N fertilizer treatments. Soil PAN values observed in +CONV treatments in the field study were consistently higher than the CONV treatment, thereby creating a high risk of N losses to the environment (De Rosa et al., 2016). The OA application without N fertilizer showed a reduction in crop N uptake compared to OA that received +Rd N fertilizer. Indeed, positive Δs PAN for the Co treatment were only observed in the second year crop rotation. This was most likely caused by the residual soil N N-fertilizer application in the Co+CONV plots from the previous year. The reduction in crop N uptake from the treatments that received OA only could be attributed to the combination between the lower amount of N applied and the slow mineralization of organic N form the OA that can cause an asynchrony between crop N needs and soil PAN supply.

Similar results were observed by Hernández et al. (2014), who observed reduced tomato (Lycopersicon esculentum Mill.) yield when compost was applied as the sole N source, while the combination of OA and N fertilizer improved crop yield and N use efficiency. This N limitation, resulting from the application of OA as the only source of N, can lead farmers to either completely neglect or severely underestimate OA as an N supply in their budget calculations and highlights the need to supplement application of OA with N fertilizer to maximize crop yield and fruit quality.

The N reduction (+Rd) strategy adopted in the field study that lead to N fertilizer reductions up to 210 kg (–27%) of N fertilizer ha–1 rotation–1, demonstrated that it is possible to strategically replace a proportion of the synthetic N fertilizer with OA without compromising yield, and still maintain soil PAN close to regional N fertilizer recommendations. However, regional N fertilizer recommendations are usually developed from empirical data collected over decades without taking into account the spatial and temporal variability that affects interactions among soil, plant, and atmosphere. This high degree of uncertainty, especially in the long term, could lead to an overestimation or underestimation of crop N needs that will reduce a farmer’s net income and increase the risk of N pollution to the environment. The optimal additional N reduction (+Opt, up to 49% N fertilizer reduction) identified in the SALUS long-term simulation exercise showed no differences in crop biomass production and confirms the potential PAN surplus associated with conventional N fertilizer rate in the horticultural farming systems.

The SALUS model was able to reproduce the crop biomass production and crop N uptake with less than 12% error, a value that is in agreement with the results of other long-term simulation studies of cereal crop growth after OA application (Basso et al., 2011b, Giola et al., 2012). These results are consistent despite the complexity of simulated horticultural crop rotations, and the uncertainty linked to the consecutive return of considerable amounts of organic matter to the soil (as both OA and crop residue).

Our simulations identified the economic benefit of N fertilizer reductions up to 49% with +Opt the N rate. Given the average price of N fertilizer as US$0.43 kg–1 N (IndexMundi, 2016), annual savings of up to$54 ha–1 are possible. In addition, the Co+Opt treatment had the greater additional benefits of lower risk of N losses and a higher yield than CONV. The lower median PAN observed in the Co+Opt treatment can be interpreted as a more stationary temporal soil N trend that would reduce the risk of soil PAN surplus or deficit to crop needs. The slower but more constant release of N that is characteristic of composted OA would enhance the synchronicity between crop growth needs and soil PAN that is often a limiting factor to increasing N use efficiency.

These simulations also highlighted the positive effect of applying Co in terms of increased soil C, a strong indicator of soil health (D’Hose et al., 2016). The positive effect of Co on the maintenance of soil C levels over the long-term crop rotation contrasts the constant reductions in soil C obtained for the CONV and Ma treatments (annual average of –3.7%). This trend was already visible in the field study after 3 yr of the OA treatment. A common research finding is an increase of soil C following repeated application of OA, however, there is also evidence that reductions in soil C can occur due to increased C mineralization (Quilty and Cattle, 2011). For our study, the relatively small amount of Ma applied annually (average of 2.08 t ha–1), combined with an average C/N ratio of 3.3 could have promoted a priming effect on the native soil C. Differences between Co and Ma could, however, be attributed to the net amount of C applied and its stability (resistance against degradation), which the amount of C applied was on average six times higher in the Co treatment than Ma.

Composted manure, when compared to Ma, generally contains lower organic C and a higher percentage of high stability, recalcitrant material that is not easily degraded by soil microorganisms (Bernai et al., 1998). The lower level of C applied in the Ma treatments, combined with the opposing effect of increased tillage that promotes C mineralization (Ding et al., 2012) can offset the effect of annual Ma applications and could explain the decline of soil C. Further research is needed to fully account for any possible adverse effects associated with repeated OA application such as heavy metal or P accumulation.

The methodology adopted in this research, which includes a combination of detailed field and simulation scenarios, demonstrates the added value that long-term simulations can bring to shorter term field trials when they are completed with a well calibrated and validated model. The methodology also provides strategies to enhance N use efficiency and reduce uncertainty in defining site-specific best management combinations of OA and N fertilizer rates that will support long-term soil health for an intensely managed vegetable crop rotation. This research also demonstrates the value of decision support tools as an aid in the development of environmentally and economically viable OA systems at the farm scale.

## CONCLUSIONS

This research demonstrated the value of using OA as a partial substitute of N fertilizer in an intensively managed vegetable cropping system because it resulted in both agronomic and environmental advantages. Integration of composted material with reduced levels of N fertilizer (up to 49% per crop rotation) rather than application of raw OA proved to be an effective pathway to improve plant nutrient management and promote C storage. The N strategy adopted in the field study (+Rd) was designed to match the CONV soil mineral N level. The possibility of further reductions in N fertilizer rates (+Opt) lowering yield penalty, however, highlighted the potential overestimation of crop N needs under conventional N management strategy. The capability of a simulation model such as SALUS to project alternative N management over longer periods of time offers the opportunity to develop site-specific management as a means to reduce the negative effects of OA application in vegetable production.

## Acknowledgments

This project was funded by the Australian Department of Agriculture and Water Resources. Some of the data reported in this manuscript were obtained at the Central Analytical Research Facility operated by the Institute for Future Environments (QUT).

## Footnotes

Be the first to comment.