Optimising crop production and nitrate leaching in China: Measured and simulated effects of straw incorporation and nitrogen fertilisation

Europ. J. Agronomy 80 (2016) 32–44

Contents lists available at ScienceDirect

European Journal of Agronomy journal homepage: www.elsevier.com/locate/eja

Optimising crop production and nitrate leaching in China: Measured and simulated effects of straw incorporation and nitrogen fertilisation夽 Kiril Manevski a,c,∗ , Christen D. Børgesen a , Xiaoxin Li b , Mathias N. Andersen a,c , Xiying Zhang b , Per Abrahamsen d , Chunsheng Hu b , Søren Hansen b,d a

Department of Agroecology, Aarhus University, Blichers Allé 20, 8830 Tjele, Denmark Institute of Genetics and Developmental Biology, Chinese Academy of Science, Huaizhong Lu 286, 050021 Shijiazhuang, China c Sino-Danish Center for Education and Research, Zhongguancun College, 271 N 4th Ring Road, Haidian, 100080 Beijing, China d Department of Plant and Environmental Sciences, University of Copenhagen, Thorvaldsensvej 40, Frederiksberg, 1871 Copenhagen, Denmark b

a r t i c l e

i n f o

Article history: Received 26 February 2016 Received in revised form 6 June 2016 Accepted 16 June 2016 Keywords: Double crop rotation Dynamic modelling Maize Model evaluation Nitrogen response curves Soil mineral nitrogen Winter wheat

a b s t r a c t The sustainability of growing a maize—winter wheat double crop rotation in the North China Plain (NCP) has been questioned due to its high nitrogen (N) fertiliser use and low N use efficiency. This paper presents field data and evaluation and application of the soil–vegetation–atmosphere transfer model Daisy for estimating crop production and nitrate leaching from silty loam fields in the NCP. The main objectives were to: i) calibrate and validate Daisy for the NCP pedo-climate and field management conditions, and ii) use the calibrated model and the field data in a multi-response analyses to optimise the N fertiliser rate for maize and winter wheat under different field managements including straw incorporation. The model sensitivity analysis indicated that a few measurable crop parameters impact the simulated yield, while most of the studied topsoil parameters affect the simulated nitrate leaching. The model evaluation was overall satisfactory, with root mean squared residuals (RMSR) for simulated aboveground biomass and nitrogen content at harvest, monthly evapotranspiration, annual drainage and nitrate leaching out of the root zone of, respectively, 0.9 Mg ha−1 , 20 kg N ha−1 , 30 mm, 10 mm and 10 kg N ha−1 for the calibration, and 1.2 Mg ha−1 , 26 kg N ha−1 , 38 mm, 14 mm and 17 kg N ha−1 for the validation. The values of mean absolute deviation, model efficiency and determination coefficient were also overall satisfactory, except for soil water dynamics, where the model was often found erratic. Re-validation run showed that the calibrated Daisy model was able to simulate long-term dynamics of crop grain yield and topsoil carbon content in a silty loam field in the NCP well, with respective RMSR of 1.7 and 1.6 Mg ha−1 . The analyses of the model and the field results showed that quadratic, Mitscherlich and linear-plateau statistical models may estimate different economic optimal N rates, underlining the importance of model choice for response analyses to avoid excess use of N fertiliser. The analyses further showed that an annual fertiliser rate of about 300 kg N ha−1 (100 for maize and 200 for wheat) for the double crop rotation with straw incorporation is the most optimal in balancing crop production and nitrate leaching under the studied conditions, given the soil replenishment with N from straw mineralisation, atmospheric deposition and residual fertiliser. This work provides a sound reference for determining N fertiliser rates that are agro-environmentally optimal for similar and other cropping systems and regions in China and extends the application of the Daisy model to the analyses of complex agro-ecosystems and management practices under semi-arid climate. © 2016 Elsevier B.V. All rights reserved.

Abbreviations: NCP, North China Plain; C, Carbon; N, Nitrogen; C/N, Carbon to nitrogen ratio; SOM, Soil organic matter; ET, Evapotranspiration; DM, Dry matter; DS, Development stage of plant phenology; EONR, Economically optimum nitrogen rate; RMSR, Root mean squared residuals; SVAT, soil–vegetation–atmosphere transfer. 夽 Capsule: Agro-environmental analysis of maize—winter wheat double crop rotation in the North China Plain. ∗ Corresponding author at: Department of Agroecology, Aarhus University, Blichers Allé 20, 8830 Tjele, Denmark. E-mail address: [email protected] (K. Manevski). http://dx.doi.org/10.1016/j.eja.2016.06.009 1161-0301/© 2016 Elsevier B.V. All rights reserved.

K. Manevski et al. / Europ. J. Agronomy 80 (2016) 32–44

1. Introduction Food security in China has a high political priority as sufficient food production is needed for a growing population. Since 2004 the Chinese government has been subsidising farmers in order to intensify their crop production, leading to intensive use of fertilisers and pesticides (Sun et al., 2012). The North China Plain (NCP) is the most important region for agricultural production in China, supplying 40% and 60% of the national maize and winter wheat productions, respectively. Both crops are harvested within a year, with maize planted in June and harvested in October, and winter wheat planted few days after and harvested the following June. The continuous overuse of nitrogen (N) fertiliser in the maize—winter wheat double crop rotation, ranging typically between 400 and 600 kg N ha−1 and mostly supplied from urea, has resulted in large soil N surpluses and nitrate leaching out of the root zone (Ju et al., 2009) and has increased nitrate concentrations in the upper groundwater and in river and lake waters (Chen, 2010). This is not sustainable in relation to either current and future drinking water requirements or the ecology of the freshwater systems. As China must address the joint challenges of food production and environmental degradation expeditiously, effective methods are required that can balance the use of N fertiliser against the need for both high crop yield (food security) and low nitrate leaching (protection of freshwaters). Field experiments are essential in quantifying effects of field management such as N fertiliser application or straw incorporation on crop production and nitrate leaching. Crop yield response to fertiliser rates is often described by statistical models in order to determine the fertiliser requirement of a crop in relation to soil type and its nutrient status, quality of the fertiliser and economic considerations (Fageria and Baligar, 2005; Ju and Christie, 2011; Valkama et al., 2013). The nitrate leaching response to fertiliser rates may also be described with statistical models, though this is often not enough to explain the variation in leaching or to link it with field management. Some studies report large increases in leaching already at low fertiliser rates (Simmelsgaard and Djurhuus, 1998), while others report that leaching starts to increase at or slightly below the economically optimal rate (Delin and Stenberg, 2014; Goulding, 2000). Individual and linked processes affect nitrate leaching, such as mineralisation of soil organic matter (SOM) and nitrification during seasons of limited crop growth with water surpluses leading to percolation that increases leaching, or denitrification due to water logging that decreases leaching. Moreover, optimisation of the fertiliser rate is intrinsically linked with the incorporation of crop residues and straw promoted in the last decades in the NCP, which has steadily increased the soil organic matter (SOM) content (Zhang et al., 2013). Increasing the SOM pool and its potential mineralisation affects the crop N supply. Process-based models of agro-ecosystem dynamics are useful to quantitatively assess the N cycle and better understand the links between N sources, losses and crop N demands in space and time. Field-scale, process-based models differ primarily in their objective, i.e., some focus on simulating crop growth and yield with comparatively crude descriptions of soil processes, while others include details of the flows of water, heat, carbon (C) and N between the crop and the soil, as well as their interactions; fewer models integrate the soil–vegetation–atmosphere transfer (SVAT) component in their framework, allowing essential mechanisms of crop growth and responses to weather, soil and management to be captured. Several studies have quantified crop production and environmental impacts of the maize—winter wheat double crop rotation in response to N inputs in the NCP with process-based models (e.g., Cui et al., 2014; Hu et al., 2006; Michalczyk et al., 2014). Regardless of their level of complexity and assumptions, the models in these studies have been calibrated using field data from the NCP in order to decrease parameter uncertainty related to the

33

pedo-climatic conditions of application (Palosuo et al., 2011). Daisy is a process-based SVAT model primarily designed and parameterised using soil measurements and cropping systems under the sub-humid temperate climate of Northern Europe (e.g., Doltra et al., 2011; Manevski et al., 2015; Salazar et al., 2013; Svendsen et al., 1995). While many other models simulate nitrate leaching rates in relation to soil water content above field capacity, Daisy treats leaching as process driven by the integrated effect of the redistribution of soil water and subsequent convection-dispersion of nitrate, which are especially important under semi-arid climates, such as in the NCP, where even small quantities of water below field capacity affect crop growth and soil nutrient status. As such, the model is an attractive tool for studying agro-environmental problems in China, particularly for balancing crop yield with nitrate leaching. However, Daisy has not been set up for other pedoclimatic environments such as for the semi-arid NCP and its agronomic practice of an annual double crop rotation. The objectives of this study were to: i) collect field data of crop production and nitrate leaching from a maize—winter wheat double crop rotation grown on the typical silty loam soils in the NCP, ii) use the field data to set up the Daisy model for the Chinese pedoclimatic environment and cropping system and iii) utilise the field data and the calibrated model for a robust response analyses to determine an optimal field management in relation to straw incorporation and N fertiliser rate for the crops that will return high yields and low nitrate leaching.

2. Materials and methods 2.1. Field experiments Fertilisation experiments with maize (Zea mays L.) and winter wheat (Triticum aestivum L.) grown in a double crop rotation were established in 1998 at the Luancheng Agro-Ecological Experimental Station (37◦ 53 15 N, 114◦ 40 47 E, elevation 50 m), Chinese Academy of Sciences, located in the Shijiazhuang Prefecture in the NCP. The climate is continental and semi-arid, with cold and dry winters and hot and rainy summers. The mean temperature ranges from −5 ◦ C in January to 28 ◦ C in July, and precipitation is around 500 mm year−1 , most of which (>70%) occurs in the summer (July to September). The potential evapotranspiration is around 1000 mm year−1 . The data used in the present study were obtained from two fields: field A during 2007–2013 and field B during 2000–2004. Although both fields have a silty loam soil (Haplic Cambisol according to the FAO classification, about 55% silt content across the 2.0 m soil profile), there are different clay contents of 17% and 4.6% in the topsoil, and 30% and 15% in the subsoil, for field A and field B, respectively. Table 1 summarises the field experiments and the data used in the study, and Table 2 describes the soil characteristics in the fields. Field A had straw incorporated N response plots measuring 7 × 10 m in a complete block design with three replications. The plots were top-dressed with annual fertiliser rates of 0, 200, 400 and 600 kg N ha−1 from urea (46-0-0), hereinafter referred to as N0, N200, N400 and N600 treatments, respectively. Half of the fertiliser amount was applied on maize during flowering in August, while the other half was applied in a split dosage to winter wheat at planting in October and at stem elongation in April. Detailed crop measurements were conducted for one double cropping season, i.e., from June 2012 to June 2013. Crops were sampled from small sampling area at the center of each plot in three replications at juvenile, flowering and harvest. Leaf area index (LAI) was measured by LiCor3100 (Li-Cor Environmental, USA). Dry matter (DM) of leaf, stem and grain was determined after oven-drying; their total N content was determined by Kjeldahl System 2300 (Tecator, Sweden). In addi-

34

K. Manevski et al. / Europ. J. Agronomy 80 (2016) 32–44

Table 1 Outline of the field experiments with a maize—winter wheat double crop rotation at Luancheng station, the North China Plain, used in the study. Nitrogen (N) fertiliser is annual amount for the rotation. Year runs from 15 October to 14 October. Field

Year Precipitation (mm) Irrigation (mm) Temperature (◦ C) N fertiliser (kg N ha−1 ) N0 (0) N200 (200) N400 (400) N600 (600) a

A (straw incorporated)

B (straw removed)

Validation

Calibration

Validation

Calibration

2006−07 523 274 13.2

2007−08 596 240 12.6

2008−09 534 240 13.0

2009−10 366 330 11.8

2010−11 338 390 12.4

2011−12 414 360 12.5

2012−13 570 295 12.1

2001−02 327 698 13.2

2002−03 445 420 12.0

2003−04 525 330 12.6

Ca C C C

C C C C

C C C C

C C C

Cn Cn Cn

CnDN CnDN CnDN

CnwDN CnwDN CnwDN CnwDN

CnwDN CnwDN CnwDN

CnwDN CnwDN CnwDN

C C C

C = crop yield; n = crop nitrogen, w = soil water content, D = annual soil water drainage (0–2 m); N = annual nitrate leaching (0–2 m).

Table 2 Soil characteristics for the studied fields at Luancheng station, the North China Plain. Dash denotes no data. Soil depth (cm)

0−20

20–40

Field

A

B

A

B

A

B

A

B

Clay (%) Silt (%) Sand (%) Organic matter (%) Bulk density (g cm−3 ) Available N (mg N kg−1 ) Available P (mg P2 O5 kg−1 ) Available K (mg K2 O kg−1 )

17.0 57.1 24.1 1.8 1.47 148 10 96

4.5 52.8 41.5 1.2 1.22 80–90 10–15 80–100

18.1 59.6 21.2 1.1 1.43 72 5 66

6.5 58.2 35.5 0.8 1.44 – – –

14.7 53.3 31.5 0.5 1.43 60 0.7 39

13.5 57.0 29.4 0.05 1.48 – – –

30.0 55.1 14.5 0.1 1.51 39 0.3 25

14.0 64.0 22.0 0.01 1.56 – – –

tion, by digging two soil pits and exposing the soil profiles from 0 to 200 cm depth, maize roots were sampled in October 2013 according to Ahmadi et al. (2011). Maize root length density (RLD) and maximum rooting depth were determined according to Tennant (1975) and Gerwitz and Page (1974). Winter wheat RLD and maximum rooting depth were adopted from Zhang et al. (2009). Field B had straw removed N response semi-lysimeter plots measuring 2.5 × 2.5 m in a randomised complete block design with three replications. The plots were surrounded by concrete walls to 2.0 m below and 0.2 m above the soil surface to prevent lateral flow of water and nutrients. The aboveground and grain DM and their N contents at harvest were measured in three replications for the period 2000–2004 as described above. For both fields, daily soil water drainage at 2.0 m depth (the bottom of the root zone) was calculated with the water balance equation according to the procedure in Moreno et al. (1996), assuming no surface runoff and negligible upward water movement (for details see Li et al., 2007). Since the ammonium concentrations were negligible, the nitrate leaching was calculated by multiplying the estimated drainage by the nitrate concentrations determined from suction cups installed at 2.0 m depth. Annual (accumulated) soil water drainage and nitrate leaching were calculated from 15 October until 14 October the following year. For both fields, maize was planted in mid-June and harvested around the first week of October; winter wheat was planted the following week after harvesting maize. The straw in field A was returned to the soil after harvest by chopping, even broadcast on the field and incorporation up to 20 cm soil depth by mouldboard ploughing, whereas in field B the straw was completely removed after harvest. The crops in both fields were fully irrigated by flooding with pumped groundwater after N fertilisation, and supplemented throughout the season when needed. Pest and disease management followed the common practice. Details about crop planting, irrigation, fertilisation and harvesting at the two fields are presented in Table S1 of the supplementary material.

40–110

110–210

2.2. Setup of the Daisy model for the North China Plain conditions Daisy (version 5.19 used in this study) simulates water, C and N dynamics in agro-ecosystems. The soil hydrology is simulated by water transport (Richard’s equation), heat fluxes (Fourier’s law), and reference ET (FAO Penman-Monteith equation). Crop phenology is simulated with temperature-dependent development rates for the vegetative and reproductive period, modified by vernalisation and photoperiod for wheat and by photoperiod for maize during vegetative period. Photosynthesis is simulated with calculation of light distribution within the canopy and light response curves, modified by temperature, soil water and N, and senescence functions. Assimilate (a net result of photosynthesis and growth/maintenance respiration) is allocated to root, leaf, stem and storage organ (grain) as a function of phenology and root/shoot ratio. Root distribution depends on root mass and decreases exponentially. The SOM turnover (mineralisation and immobilisation) is represented by three main discrete pools, namely SOM, added organic matter (AOM) from crop residues, rhizodeposition and organic fertilisers, and soil microbial biomass (SMB). These pools are further subdivided according to first order kinetics into slow (indexed 1, e.g. SOM1) and fast (indexed 2, e.g. SOM2), in addition to one inert pool (SOM3) that does not contribute to the turnover. Immobilisation occurs to the pool with a C/N ratio smaller than the C/N ratio of the other pools, otherwise organic N mineralises to ammonium that is nitrified to nitrate and dissolved in the soil water. Nitrification-denitrification is a function of soil temperature, ammonium-, nitrate- and water content. Transport of soil nitrate and ammonium, and hence their leaching, is simulated by the convection-dispersion equation. Ammonia volatilisation is given as a user-defined percentage of ammonium fertiliser at the time of application. Atmospheric N input to the soil is based on ammonium and nitrate concentrations in the rain, which are user-defined in the weather file. Details of the model equations and assumptions are available elsewhere (Hansen, 2002; Hansen et al., 2012).

K. Manevski et al. / Europ. J. Agronomy 80 (2016) 32–44

2.2.1. Model inputs and sensitivity analysis Inputs necessary to run Daisy include daily weather, soil, crop and management information. Daily air temperature, solar radiation, wind speed, relative humidity and precipitation were recorded at the Luancheng station; the respective values for wet and dry deposition of 7.5 ppm and 10 kg N ha−1 for ammonium and 2.5 ppm and 5 kg N ha−1 for nitrate were approximated from Zhang et al. (2011b). Soil description (texture, organic matter, bulk density) was taken from Wang et al. (2013) for field A and Li et al. (2007) for field B; the soil parameters describing water retention and unsaturated hydraulic conductivity at different soil water pressure potentials according to the van Genuchten-Mualem model were optimised against measured water retention in field A in 2007 (UMS GmbH, Germany, unpublished) with the RETC code (van Genuchten et al., 1991) and used as input for the Richards’ equation in Daisy. For the crops, “Pioneer maize” and “Winter wheat” were selected as they initially provided the most reasonable simulation results against the measured DM dynamics compared to the other parameterisations in the Daisy library based on specific experiments in Denmark. Daily management practice included the records on time of crops planting and harvest, time and rate of N fertiliser and irrigation, tilling date and depth, and fraction of crop straw returned to the soil. In order to assess the model behaviour under the “new” local conditions, and to help identify the most sensitive input parameters that should be focused on for calibrating crop production and nitrate leaching, a mono-factor sensitivity analysis was conducted on key parameters directly related to crop development, leaf photosynthesis and net mineralisation (topsoil, 0–20 cm). The model was run with the actual climate for 2000–2013, soil, crop and management for the N400 treatment at the Luancheng station. The sensitivity tests were conducted by systematically increasing and decreasing a single parameter value by 10% while keeping all other input parameters constant in order to diagnose the response of crop yields and nitrate leaching. Parameter interactions were accounted for by testing the sensitivity of crop yields to changes in selected soil parameters, and that of nitrate leaching by changing the crop parameters. 2.2.2. Model calibration and validation The field data were split up according to the “hold out” method (Bennett et al., 2013) and those used for calibration included the highest volume of details (Table 1). The model was calibrated by ‘trial and error’ in an integrated modelling manner, by first fitting the soil water dynamics, then the crop growth and N uptake patterns, and lastly soil mineralisation, with iteration processes in between. The soil water dynamics were calibrated by adjusting saturated hydraulic conductivity (Ksat, matching point in the Mualem equation) and the l-parameter until the simulated soil water contents matched the measured values with the lowest possible error. Simulated evapotranspiration and soil water drainage were also followed as they largely depend on the soil hydraulics. The crops were calibrated following three main steps: (i) First, a set of parameters for phenology, canopy development and partitioning of DM between crop organs was parameterised based on field measurements; (ii) Second, a set of parameters detected by the sensitivity analysis and related to photosynthesis and assimilate production was calibrated by iterative change until the dynamics of the crop DM and N content were simulated with the highest accuracy; (iii) Last, sensitive parameters from steps i and ii were fine-tuned to correctly simulate the crop growth patterns in both fields in all calibration years, aiming at a final set of parameters for maize and for winter wheat. The soil mineralisation processes were calibrated by altering SOM fractions, a parameter describing the distribution of SOM between the slow, fast and inert pools at the start of the simulation. The N0 treatment was used to obtain the background

35

mineralisation using the simulated and measured crop N contents at harvest in the calibration procedure. The obtained SOM fractions parameter was then used to simulate the other treatments as the background mineralisation was assumed to be unaffected by the first-year fertilisation treatments. The C/N ratio of the topsoil ranged from 8 to 12 between N treatments (Hou et al., 2012; Peng, 2011), the C/N ratio of the SMB pools was set to 4.8 according to Dong et al. (2012), and the C/N ratio of the crops straw (leaf+stem) was set to 60 for maize and 100 for winter wheat according to Cui et al. (2014). Ammonia volatilisation was simulated assuming that increased temperature, soil water before application and amount of urea increase soil pH and thus lead to relatively higher fertiliser losses (Holcomb et al., 2011; Ma et al., 2010); thus, respective losses from 50, 100, 150, 200, and 300 kg N ha−1 fertilisation rates were set at 5, 10, 12, 15, and 17%. The model was validated on the rest of the field data (Table 1). Re-validation for the purpose of the study was performed for crop yields and topsoil (0–20 cm) organic C measured from 1979 to 2009 in a field at the Luancheng station with a texture similar to field A (for details see Zhang et al., 2011a, 2013). For this long-term simulation, the model was initialised with 1.1% topsoil organic matter according to Zhang et al. (2013) and N fertiliser amounts for maize and winter wheat were, respectively, 80 and 100, 120 and 150, 150 and 200, and 200 and 210 kg N ha−1 for the periods 1979–1990, 1991–1999. 2000–2003 and 2004–2009. The timing of N fertilisation was set as in the calibration, i.e., for maize before flowering in July and for winter wheat in split-doses at sowing in October and at stem elongation in April. Furthermore, the straw was removed for 1979–1990, incorporated after wheat harvest for 1991–1998, and incorporated after both crops harvest for 2000–2009, and for all years, the soil was ploughed until 18 cm before sowing wheat. Irrigation was set to full water supply as in the calibration. As the cultivars of maize and winter wheat have changed over time and those used in the long-term field experiment were common cultivars widely used in the area (Zhang et al., 2011a), three crop parameters were slightly altered in order to mimicked “decadal” cultivars in the 1980s, 1990s and recent 2000s, while all other parameters were maintained. Weather data for the Shijiazhuang station (about 20 km away) were used since all weather data at the Luancheng station were not recorded before 1991. For all simulations: (i) daily precipitation, irrigation and potential ET were set as upper boundary, and deep groundwater was set as lower boundary, (ii) five years prior the experimental year with known N input from fertiliser and crop residues were included in order to approximate the annual net mineralisation, and (iii) visual performance analysis and four objective measures were used to evaluate model goodness of fit, namely root mean squared residuals (RMSR), mean absolute deviation (Dev), Nash-Sutcliffe model efficiency (ME) and coefficient of determination (R2 ), as described in Bennett et al. (2013). 2.3. Response of crop yield, nitrate leaching and soil mineral nitrogen to straw incorporation and nitrogen fertiliser rate The measured crop yields and N content for the two fields were plotted against N fertiliser rate. To these yield scatter-plots were fitˆ + bx + c), Mitscherlich (y = a(1 − expˆ(−bx + c)) ted quadratic (y = ax2 and linear-plateau (y = ax + b; y = c) models in Matlab R2013a and their standard error of estimates (SEE) and R2 were compared. Except for the linear-plateau model, the economically optimum N rate (EONR) was calculated by equating the first derivatives of the model equations to the ratio between the cost of fertiliser ($0.31 kg−1 urea) and grain price ($0.34 kg−1 for maize and $0.42 kg−1 for wheat) and solving for x; for the linear-plateau model, the EONR was identified by locating the intersection of the two lines. The measured nitrate leaching was also plotted against N

36

K. Manevski et al. / Europ. J. Agronomy 80 (2016) 32–44

Straw incorporated 10 8

200

200

150

150

100

100

Polynomical

2

Mitcherlich Linear-plateau

0 Straw removed

10 8

Nitrate leaching (kg NO3 -N ha -1 )

Grain yield (Mg ha -1 )

Wheat

Maize

4

Grain nitrogen (kg N ha -1 )

6

50 Maize Wheat

0 200 150

6

Nitrate leaching Exponential

50 0 200 150

100

100

50

50

4 2 0

0 0

50

100

150

200

250

300

0 0

Crop nitrogen fertiliser (Kg N ha -1)

100

0

50 100 150 200 250 300 Crop nitrogen fertiliser (Kg N ha -1)

200

300

400

500 600

Annual nitrogen fertiliser (Kg N ha-1 )

Fig. 1. Measured response of field crop yield, nitrogen content and nitrate leaching (2.0 m depth) to nitrogen fertiliser rate in the experiments at Luancheng station, the North China Plain, for straw incorporated and straw removed. The annual nitrogen fertiliser input is sum of crop fertiliser rate (maize + winter wheat). The error bars are standard deviations of mean measured values.

fertiliser rate and to these data an exponential model (y = a expˆ(bx)) was fitted. In order to study the long-term effects of straw incorporation and N fertiliser rate on crop yields and nitrate leaching, the calibrated Daisy model was run for field A, a dominant agricultural soil in the region, for the period 1991–2013 with maize—winter wheat double crop rotation receiving annual fertiliser rates from 0 to 600 kg N ha−1 at 50 kg N ha−1 increments; the model was initialised with 1.3% topsoil organic matter and crop straw was either returned to the soil after each crop harvest (straw incorporated scenario) or was completely removed from the field (straw removed scenario). To these scenario simulation results, the same statistical models as described above were fitted in order to determine average crop yields and nitrate leaching responses, and EONR. The effect of straw incorporation and fertiliser N rate on soil mineral N was investigated by plotting the slope of the linear regression of the post-harvest soil mineral N (6 October for maize, 12 June for wheat) over the simulation period; a negative slope shows annual

depletion of the soil mineral N pool, whereas a positive slope shows annual gain that can be partly lost to the environment or accumulated in the soil. 3. Results 3.1. Analyses of the field data The measured crop yields were overall 1.4 Mg ha−1 higher for field A (straw incorporated) compared with field B (straw removed; Fig. 1), primarily due to the finer texture (higher water-holding capacity) in the former. Yet, crop yields and N contents at N0 were also higher in field A, clearly showing the benefit from straw incorporation in N limited conditions. N fertiliser obviously increased yield and N content of the crops, although N400 (maize and wheat yield of 8.1 and 6.9 Mg ha−1 in field A, 6.4 and 5.0 Mg ha−1 in field B, respectively) did not produce notably higher yields than N200 (8.0 and 6.1 Mg ha−1 in field A, 6.4 and 4.5 Mg ha−1 in field B). The mea-

Table 3 Fitting parameters of the statistical models describing the field response of crop yield and nitrate leaching to nitrogen fertiliser rates in the experiments with straw incorporated and straw removed at Luancheng station, the North China Plain. SEE is the standard error of estimate, R2 is the coefficient of determination, EONR is the economically optimum nitrogen rate. Straw

Incorporated

Modela

Response variable

−1

Maize yield (Mg ha

)

Wheat yield (Mg ha−1 )

Nitrate leaching (kg N ha−1 ) Removed

Maize yield (Mg ha−1 )

Wheat yield (Mg ha−1 )

Nitrate leaching (kg N ha−1 ) a

SEE

R2

Parameter value

EONR

a

b

c

(Kg N ha−1 )

Quadratic Mitscherlich Linear-plateau Quadratic Mitscherlich Linear-plateau Exponential

0.42 0.16 0.12 0.47 0.14 0.09 16.53

0.93 0.97 0.99 0.95 0.96 0.99 0.94

−0.0001 8.2170 0.0438 −0.0001 6.5020 0.0363 1.3453

0.0455 0.0417 3.660 0.0390 0.0286 2.550 0.0074

3.921 0 8.100 2.654 0 6.200 –

223 144 100 213 196 102 –

Quadratic Mitscherlich Linear-plateau Quadratic Mitscherlich Linear-plateau Exponential

0.74 0.30 0.21 0.40 0.09 0.22 22.14

0.97 0.92 0.96 0.96 0.93 0.99 0.95

−0.0001 6.8740 0.0389 −0.00008 5.0190 0.0330 1.2469

0.0433 0.0329 2.510 0.0365 0.0232 1.230 0.0080

2.638 0 6.500 1.325 0.010 4.900 –

212 170 108 224 222 112 –

ˆ + bx + c; Mitscherlich: y = a(1 − expˆ(−bx + c)); Linear-plateau: y = ax + b; y = c; Exponential: y = a expˆ(bx). Quadratic: y = ax2

K. Manevski et al. / Europ. J. Agronomy 80 (2016) 32–44

37

Fig. 2. Relative sensitivity of soil and crop parameters to simulated crop yield (Mg ha−1 ) and nitrate leaching (kg N ha−1 ) with the Daisy model. For convenience, the changes of conversion efficiencies (E Leaf, E Stem and E SOrg) and respiration coefficients (r Leaf, r Stem, r SOrg) are represented by one working parameter (respectively E Crop and r Crop), which is the average change of the three parameters. Description of the parameters is given in the text.

sured nitrate leaching was slightly lower in field A compared with field B but was more variable, reflecting complex N turnover and transport processes in finer textured soil. For both fields, nitrate leaching was low at N200 and increased markedly at N400 and N600 (Fig. 1). The overall conclusion from this was that an annual fertiliser rate of 400 kg N ha−1 was excessive in respect of both crop production and nitrate leaching. Additional analyses by fitting quadratic, Mitscherlich and linearplateau models to the measured yield were conducted and the models parameters appear in Table 3. For both crops, the Mitscherlich and linear-plateau models had smaller SEE and higher R2 values, thus indicating a better fit than the quadratic model. The quadratic model estimated an EONR of around 220 kg N ha−1 per crop for both fields, while with the Mitscherlich model the figure was about 150 kg N ha−1 for maize and 200 kg N ha−1 for winter wheat, with a slight variation between fields, and with the linear-plateau model it was about 105 kg N ha−1 per crop. Yet, optimising the N fertiliser should include numerous rates between 0 and 100 kg N ha−1 for more conclusive yield response, which can be achieved with the use of a robust and a well-calibrated model.

3.2. Evaluation of the Daisy model The sensitivity analysis showed that a 10% change in organic matter content (humus) and C/N ratio (C per N) affect notably the simulated nitrate leaching, whereas leaf photosynthetic capacity and photosynthetic quantum efficiency (Fm and QEff), as well as vegetative and reproductive development rates (DSRate1 and DSRate2) affected the simulated crop yields (Fig. 2). The parameters interaction showed that the simulation of nitrate leaching is strongly affected by parameters governing crop vegetative development and growth (DSRate1 and QEff), whereas the simulation of crop yields was not affected by the change in any of the studied soil parameters. Based on the sensitivity analysis and the field data, a number of soil and crop parameters required by Daisy were calibrated to conform to the study condition in the NCP (Table 4). In addition, important crop growth processes controlled by piecewise linear functions in relation to crop phenology and temperature were also adjusted (Fig. 3). The partitioning of total assimilate (Partit) was set as 50% to the root before the crops reach early flowering (DS = 1), with the remaining partitioned between leaf, stem and grain; thereafter, assimilate of stem and leaf decreased at the cost of increased assimilate in grain. The potential N dilution curves (CrpN) for the

Fig. 3. Crop parameters used in Daisy to simulate maize and winter wheat fields at Luancheng station, the North China Plain. Development stage (DS) equals 0, 1 and 2 for, respectively, emergence, flowering and physiological maturity.

38

K. Manevski et al. / Europ. J. Agronomy 80 (2016) 32–44

Table 4 Main parameters in Daisy used to simulate maize and winter wheat fields at Luancheng station, the North China Plain. Soil parameter (symbol, unit)

0–20 cm depth A

Hydraulic conductivity (Ksat, cm h−1 ) Saturated soil water (Theta sat, vol%) Residual soil water (Theta res, vol%) van Genuchten ␣ (alpha, cm−1 ) van Genuchten n (n) Tortuosity factor (l) Soil organic matter distribution (SOM fractions) Soil carbon-nitrogen ratio (C per N) Microbial biomass carbon-nitrogen ratio (C per N) Denitrification factor (water factor)b End of root zone (MaxRootingDepth, cm) Total C added (Input, kg C ha−1 y−1 ) Root C added (Root, kg C ha−1 y−1 ) Crop parameter (symbol, unit)

40–110 cm depth

A

1.56 4.65 10.3 0.39 0.45 0.44 0.06 0.01 0.06 0.008 0.004 0.009 1.332 1.252 1.248 1.45 1.75 −1.26 (0.45 0.45 0.10) (0.25 0.25 0.50) 8–12 8–12 12 4.8 (0.89 0.01) (0.98 0.01) (1.00 0.01) 190 1550 (field A), 1000 (field B) 750 (field A), 600 (field B)

Emergence soil temperature sum (EmrTSum, C d) Vegetative development rate (DSRate1, day−1 ) Maximum leaf photosynthesis (Fm, g CO2 m−2 h−1 ) Quantum efficiency (QEff, (g CO2 m−2 h−1 )/(Wm−2 )) Specific leaf weight (SpLAI, (m2 m−2 )/(g DM m−2 )) Conversion efficiencies (E Leaf/E Stem/E SOrg) Respiration coefficients (r Leaf/r Stem/r SOrg) Root maximum penetration (MaxPen, cm) Max NH4 -N uptake (MxNH4Up, g cm−1 h−1 ) Max NO3 -N uptake (MxNO3Up, g cm−1 h−1 ) Specific root length (SpRtLength, m g−1 ) Leaf weight modifier at DS: 0 (emergence) (LeafAIMod) 0.5 (juvenile) 1.0 (flowering) 2.0 (maturity)

100–210 cm depth

B

A

B

A

B

2.5 0.42 0.01 0.004 1.23 1.23

22.3 0.43 0.12 0.005 1.186 0.72 (0.0 0.0 1.0) 12

1.0 0.4 0.05 0.003 1.288 1.05

0.12 0.41 0.12 0.002 1.188 0.88 (0.0 0.0 1.0) 12

0.1 0.41 0.15 0.002 1.25 1.15

“Pioneer maize”

◦

a

20–40 cm depth B

“Winter wheat”

Default

Calibrated

Default

Calibrated

100 0.0245 6.0 0.04 0.03 0.68/0.66/0.68 0.016/0.010/0.010 120 2.5 × 10−8 2.5 × 10−8 100 1 1 1 1

200 0.0265 7.4 0.045 0.02 0.68/0.60/0.75 0.015/0.015/0.005 150 1.5 × 10−7 1.5 × 10−7 130 0.7 0.8 1.2 0.2

100 0.026 5.0 0.05 0.022 0.68/0.66/0.70 0.016/0.010/0.010 120 2.5 × 10−7 2.5 × 10−7 100 1 1 1 1

100 0.026 7.0 0.052 0.0201 0.78/0.78/0.75 0.010/0.010/0.005 210 2.5 × 10−7 2.5 × 10−7 100 0.8 0.9 0.7 0.5

M—measured, C—calibrated, S—sensitivity, L—literature. b Fitting parameter with points (x,y) where first values (x) is fraction of maximal soil water content and second value (y) is denitrification factor.

crop organs were also set to reflect the measurements (g N g DM−1 ), with winter wheat having higher values compared to maize, especially for grain.

Crop nitrogen (kg N ha -1)

Crop dry matter (Mg ha -1) 12

calibration

10

validation

8 6

Evapotranspiration (mm month -1)

200

200

150

150 100

100 Calib. RMSR 0.9 Dev -0.3 0.93 ME 0.84 R2

4 Simulated values

The comparison of the measured and simulated results by pooling the data for the calibration and the validation datasets is shown in Fig. 4. The standard deviation of the crops measurements at harvest across the calibration years was in the range 1.6–3.2 Mg ha−1

2 0 2

0

4

6

8

10

Valid. 1.2 1.1 0.67 0.25

Calib. RMSR 19.8 Dev -5.5 0.83 ME 0.86 R2

50 0

12

0

50

100

150

Valid. 25.9 15.7 -0.71 0.41

RMSR Dev ME R2

50 0

200

0.5

100

150

0.4

80

125

0.3

60

100

150

Valid. 38.2 -10.8 0.67 0.73

200

Nitrate leaching (kg NO 3 -N ha -1 year -1)

Soil water drainage (mm year -1)

Soil water content (vol. %)

50

0

Calib. 30.3 -8.2 0.75 0.82

100 75

0.2

Calib. RMSR 0.19 Dev -0.01 0.65 ME 0.62 R2

0.1 0.0 0.0

0.1

0.2

0.3

0.4

Valid. 0.72 0.01 -0.17 0.51

0.5

40

Calib. RMSR 9.3 Dev -5.5 ME 0.87 R2 0.80

20 0 0

20

40

60

80

Valid. 13.9 -9.1 0.62 0.11

100

50

RMSR Dev ME R2

25 0

0

25

50

75

100

Calib. 10.1 1.9 0.87 0.94

Valid. 16.7 -8.6 0.71 0.81

125 150

Measured values Fig. 4. Fit between simulated and measured crop and soil variables. Values for crops dry matter (DM) and N content are at harvest. Values for soil water content are pooled for 20, 100 and 180 cm soil depths. Values for soil water drainage and nitrate leaching out of 200 cm soil depth are cumulative annual from 15 October to 14 October. Dashed lines indicate 1:1 relation.

K. Manevski et al. / Europ. J. Agronomy 80 (2016) 32–44

39

Table 5 Modification of crop parameters used in Daisy to simulate long-term grain production during 1980s, 1990s and 2000s for maize and winter wheat fields at Luancheng station, the North China Plain. Description of parameters is given in the text. Parameter

Maize cultivars

Fm QEff E SOrg

Maize yield (Mg ha -1)

12 10 8

1990s

2000s

1980s

1990s

2000s

5.0 0.040 0.60

7.0 0.045 0.75

7.4 0.045 0.75

5.0 0.050 0.68

5.5 0.050 0.70

7.0 0.052 0.75

water use efficiency. Therefore, in the re-validation, the values of two photosynthetic and one assimilate conversion parameters were slightly modified to mimic the decadal cultivars, while all other crop parameters were maintained (Table 5). Hence, the model satisfactorily simulated the impacts of climatic variability on grain DM yield trends in the period 1979–2009. Also, the measured topsoil C content corresponded to the overall field management throughout the decades, and it was declining during 1980s under straw removal, but it started to increase significantly from 1990s onwards due to straw incorporation and the model captured these dynamics well (Fig. 5). Considering the inherent complexity of the crop growth, soil N transport and transformation processes in agricultural fields, it is clear that the calibrated Daisy overall satisfactorily simulates crop production, nitrate leaching and long-term dynamics of crop yields and SOM for maize—winter wheat fields on silty loam soil in the NCP. The slight over- or underestimations found for some variables can be solved by further calibration and improved field measurements in the future.

RMSR= 1.81 Dev= 0.72 ME= 0.34 R2 = 0.53

6 4 2 0 12

Wheat yield (Mg ha -1)

Winter wheat cultivars

1980s

10 8

RMSR= 1.63 Dev= -0.63 ME= 0.18 R2 = 0.52

6 4 2 0 30 25

RMSR= 1.61 Dev= -1.13 ME= 0.98 R2 = 0.92

3.3. Simulated effects of straw incorporation and nitrogen fertilisation on crop production, soil mineral nitrogen and nitrate leaching

20 15 10 field

5 0

2009

2007

2005

2003

2001

1999

1997

1995

1993

1991

1989

1987

1985

1983

1981

model

1979

Soil organic carbon (Mg C ha-1)

35

Fig. 5. Measured and simulated temporal dynamics of crop yield for maize and winter wheat, as well as soil organic carbon (0–20 cm) at Luancheng station, the North China Plain.

for DM and 27–36 kg N ha−1 for N content for aboveground biomass, and 0.8–2.1 Mg ha−1 for DM and 25–31 kg N ha−1 for N content for grain yield, which is higher than the RMSR values for DM and N content of both crops. The model demonstrated accurate simulations of the crop production, despite the negative ME values for aboveground N that probably occurred because of the small variability in the measured yield at harvest. However, the model was often found erratic (high RMSR, low ME) when predicting the daily dynamics of soil water content and underestimated it by 6.4, 4.0 and 4.7 vol.% for the 20, 100 and 180 cm soil depths, respectively. On a coarser time scale, which was the main intention of the analysis, the simulations were good in both the calibration and validation treatments years for evapotranspiration, soil water drainage and nitrate leaching (Fig. 4). Fig. 5 shows the re-validation of the Daisy model for reproducing the long-term dynamics of crop yields and topsoil (0–20 cm) organic C at Luancheng station. During the 1980s, single-cross local cultivars were grown by the farmers and these were replaced in the 1990s by high-yielding, strong-rooted hybrids released in the NCP to support the farmers’ crop production, and finally by temperature-tolerant hybrid cultivars in the 2000s with higher

The effects on crop production and nitrate leaching of a longterm increase in the application of N fertiliser simulated by the Daisy model are shown in Fig. 6. Overall, crop yields were similar for the two straw management practices (incorporation and removal) as the simulations use the same initialisation, weather and irrigation practices. Yet, simulated N content of crops yields for the N0 treatment were higher with straw incorporated (42 and 18 kg N ha−1 for, respectively, maize and winter wheat) compared to straw removed (34 and 11 kg N ha−1 for, respectively, maize and winter wheat). With rising N fertiliser rates, maximum crop yield was reached at 100 kg N ha−1 for maize and 175 kg N ha−1 for winter wheat. The quadratic model estimated an EONR of 190 kg N ha−1 per crop for both straw treatments, while with the Mitscherlich model the figure was 120 kg N ha−1 for maize and 295 kg N ha−1 for winter wheat, with a slight variation between straw treatments, and with the linear-plateau model it was about 100 kg N ha−1 for maize and 130 kg N ha−1 for winter wheat for both straw treatments (Table 6). The EONR for simulated winter wheat derived by the Mitscherlich model was higher than expected due to the exponential rise to maximum, which again showed the sensitivity of EONR to the method used for its estimation. The simulated nitrate leaching also increased exponentially with N fertiliser rate (Fig. 6). At the EONR, simulated nitrate leaching was 5–12 kg N ha−1 , increasing drastically at higher rates. The slope of the leaching response curve, mimicking the ratio of kg ha−1 nitrate leaching to kg ha−1 crop yield, was 0.08 at the EONR, and reached 2 at N600. At higher N fertiliser rates the simulated leaching showed large variations due to interannual climate variability during the simulation period. Also, the nitrate leaching was often simulated at low level relative to the amount of N fertiliser added, implying soil nitrate accumulation. The simulated post-harvest soil mineral N with straw incorporation decreased for 2–5 kg N ha−1 year−1 throughout the N50

40

K. Manevski et al. / Europ. J. Agronomy 80 (2016) 32–44

Straw incorporated 10 8

200

200

150

150

100

100

Grain yield (Mg ha -1 )

Wheat

Polynomical

2

Mitcherlich Linear-plateau

0 Straw removed

10 8

50 Maize Wheat

0

200 150

6

Nitrate leaching (kg NO3 -N ha -1 )

Maize

4

Grain nitrogen (kg N ha -1 )

6 Nitrate leaching Exponential

50 0

200 150

4

100

100

2

50

50

0 0

50

100

150

200

250

300

00

Crop nitrogen fertiliser (Kg N ha -1)

50 100 150 200 250 300 Crop nitrogen fertiliser (Kg N ha -1)

0 0

100

200

300

400

500 600

Annual nitrogen fertiliser (Kg N ha-1 )

Fig. 6. Simulated response of crop yield, nitrogen content and nitrate leaching (2.0 m depth) to nitrogen fertiliser rate in the experiments at Luancheng station, the North China Plain, for straw incorporated and straw removed. The annual nitrogen fertiliser input is sum of crop fertiliser rate (maize + winter wheat). The error bars are standard deviations of mean simulated values. Table 6 Fitting parameters of the statistical models describing the Daisy response of crop yield and nitrate leaching to nitrogen fertiliser rates in the experiments with straw incorporated and straw removed at Luancheng station, the North China Plain. SEE is the standard error of estimate, R2 is the coefficient of determination, EONR is the economically optimum nitrogen rate. Straw

Incorporated

Response variable

Maize yield

Wheat yield

Nitrate leaching Removed

Maize yield

Wheat yield

Nitrate leaching a

Modela

SEE

R2

Parameter value

EONR

a

b

c

(Kg N ha−1 )

Quadratic Mitscherlich Linear-plateau Quadratic Mitscherlich Linear-plateau Exponential

0.53 0.06 0.35 0.43 0.12 0.36 12.5

0.68 0.97 0.89 0.95 0.94 0.94 0.97

−0.0001 7.9360 0.0445 −0.0001 6.6231 0.0306 0.0526

0.0389 0.1181 4.223 0.0488 0.0182 1.892 0.0133

4.653 1.504 7.700 1.469 0.001 6.210 –

190 118 100 224 290 131 –

Quadratic Mitscherlich Linear-plateau Quadratic Mitscherlich Linear-plateau Exponential

0.56 0.04 0.40 0.25 0.13 0.31 9.40

0.72 0.96 0.86 0.94 0.91 0.95 0.98

−0.0001 7.9383 0.0446 −0.0001 6.6016 0.0291 0.0661

0.0389 0.0878 4.278 0.0455 0.0198 1.932 0.0127

4.710 0.787 7.700 1.557 0.001 6.210 –

190 126 101 238 305 135 –

ˆ + bx + c; Mitscherlich: y = a(1 − expˆ(−bx + c)); Linear-plateau: y = ax + b; y = c; Exponential: y = a expˆ(bx). Quadratic: y = ax2

and N250 treatments, and it started to increase for about 5 kg N ha−1 year−1 at the optimal N300 treatment (Fig. 7). For the straw removal scenario, the simulated post-harvest soil mineral N decreased for 3–12 kg N ha−1 year−1 throughout the N50 and N350 treatments, and a net gain of post-harvest soil mineral N started from N400 treatment. For the fertiliser rates higher than 400 kg N ha−1 year−1 , there was a large increase in the soil mineral N pool, according to the model, which reached more than 100 kg N ha−1 year−1 at N600 regardless of straw management i.e. incorporation or removal. 4. Discussion 4.1. Crop production with straw incorporation and nitrogen fertilisation in the North China Plain The field and model results both agreed that EONR with straw incorporation is around 300 kg N ha−1 year−1 for the dou-

ble crop rotation, i.e., 100 for maize and 200 for winter wheat (Tables 2 and 6). At these rates, acceptable yield of 8 Mg ha−1 for maize and 6.5 Mg ha−1 for winter wheat were achieved. Comparable crop yields were also achieved in the straw removal treatment, with EONR very similar to those for the straw incorporation treatments. The use of straw incorporation to increase crop yields is still under debate since field results across various pedo-climatic environments are inconclusive, partly due to the numerous and complex factors that affect the straw-derived N cycle under field conditions (Mu et al., 2016; Pituello et al., 2015). The marginal effect for the studied soils at Luancheng is probably because the period of mineralisation of the straw after incorporation coincides with slow N uptake by the juvenile crops (Meng et al., 2013; Meng et al., 2016). Accordingly, straw incorporation effect on crop yields is likely to be more evident in coarser-textured soils through a direct nutritional effect and, possibly, an improvement of soil characteristics (Pituello et al., 2015). However, crop yield benefits from straw incorporation are seen for N-restricted conditions such as in unfertilised or

K. Manevski et al. / Europ. J. Agronomy 80 (2016) 32–44

150

Straw incorporated

Change in soil mineral nitrogen (kg N ha-1 year -1)

100

50 After maize harvest After wheat harvest

-10 150

Straw removed

100

50

-10

0

100

200

300

400

500

600

Annual nitrogen fertiliser (Kg N ha -1) Fig. 7. Simulated annual change in residual soil mineral nitrogen (2.0 m depth) under different straw managements and nitrogen fertiliser rates for a silty loam soil in the North China Plain. Each bar represents the slope from linear regression of post-harvest soil mineral nitrogen (6 October for maize, 12 June for winter wheat) over 20-year continuous weather data (1991–2012). The annual nitrogen fertiliser is sum of the crop nitrogen fertiliser (maize + winter wheat).

41

over-fertilisation in the NCP, careful considerations should be made on the choice of model used to estimate crop EONR. It should be mentioned that the different cultivars grown on the two fields i.e. during the two time periods could introduce some uncertainty and should be considered when interpreting of the results. The cultivars of maize and winter wheat in the study were, respectively, LaiYu and GaoYou503 in field B, and XianYu335 and KeNong199 in field A, the former being high-yielding hybrids released in China in the mid-1990s, and the latter being more temperature-tolerant hybrids with large number of grains and higher heat requirement and water use efficiency (Zhang et al., 2013). However, the weather conditions during the time periods for the two fields included similar variability in precipitation and temperature (Table 1). Also, the silty loam fields in Luancheng are known to exhibit spatial variation in clay content within 0–200 cm soil column (Haijing et al., 2016; Zhang et al., 2005), which can also be found across the NCP. While these differences between the two fields cannot be fully eliminated, they were kept to a minimum by analysing average, rather than single-year, treatment responses. The herein developed N response relationships are representative for other field with soil properties and management practices similar to those in this study. They are based on multiple responses of crop yield at multiple rates of fertiliser N that account for the spatiotemporal variability in soil water, nutrient status and yield across fields, thus providing a long-term average important for strategic planning of crop N fertilisation. For operational use on specific field, this variation should be considered either by soil/crop analysis or according to model-based recommendation (e.g. Michalczyk et al., 2014). 4.2. Nitrate leaching and associated soil mineral nitrogen

organic farming systems where crop N uptake relies solely on N mineralised from crop residues or straw and soil organic matter (Ludwig et al., 2010; Pituello et al., 2015). In this study, the crop yield and N content for the N0 treatment were higher with straw incorporation than with straw removal, though this was more evident for the field data (Fig. 1) than for the model simulations (Fig. 6). For the fertilised treatments, the only marked difference between straw incorporation and straw removal throughout the years of the field experiments and the model simulations was a greater amount of crop N content in the former. It appeared that straw incorporation enabled for the crops to utilise more soil mineral N, and this was more evident for winter wheat compared to maize (Figs. 1 and 6). As crop N content is an important implication for protein and dietary quality, higher amounts with straw incorporation have been reported for cereal crops, compared to straw removal or burning (Dabin et al., 2016; Yang et al., 2015). Under the studied conditions, maize required less N fertiliser to reach maximum yield due to its shorter season and favourable growing conditions (temperature, precipitation, soil N availability), whereas winter wheat season is longer and often accompanied by temperature, water and N stresses (Chen et al., 2016). However, the crop yield response to N fertiliser rates, and thus crop N demand, may be described with more than one statistical model. The EONR estimated with the linear-plateau model was about 100 kg N ha−1 per crop, which is too little for winter wheat in relation to its N content. For comparison, EONR derived from numerous field experiments in the NCP ranges from 70 to 250 kg N ha−1 for maize and 150–350 kg N ha−1 for winter wheat (Cui et al., 2008; Ju and Christie, 2011). Although linear-plateau seems preferable for weak yield responses to N fertiliser rate, this model cannot describe over-fertilisation and associated yield decrease sometimes found at high soil N levels. The quadratic model tends to overestimate the peak in crop yield and hence the EONR, whereas the Mitscherlich model agrees overall better with the data (Belanger et al., 2000; Valkama et al., 2013). Thus, amid the increasing concerns about

There was no significant difference in nitrate leaching between straw incorporation and straw removal treatments. The slightly higher nitrate leaching measured in field B compared with field A (Fig. 1) probably occurred due to higher water input from precipitation and especially irrigation (Table 1). Irrigation can be a particularly important factor for nitrate leaching from silty loam soils in the NCP (Huang et al., 2011; Michalczyk et al., 2014). Generally, the amounts of nitrate leaching as measured and simulated here are similar to those measured in silty loam fields with maize—winter wheat double cropping in the NCP (Fang et al., 2006; Liu et al., 2006). They are also similar to the simulated amounts by Michalczyk et al. (2014) and Cui et al. (2014), who calibrated HERMES and DNDC, respectively, and used these process-based models for maize-wheat double crop rotation on silty loam fields in the NCP to assess effects of different management practices, including fertiliser N rate, on crop yield and nitrate leaching. Both studies found about 310 kg N ha−1 annual fertiliser with straw incorporation and sufficient irrigation to be the optimal input for the system without compromising grain yield but with significant (20–50%) reduction in nitrate leaching. In another simulation exercise, Zhao et al. (2015) calibrated and applied the APSIM model and concluded that 330 kg N ha−1 (180 for maize and 150 for wheat) are required to maintain the high grain yield from the double-crop rotation and to have low nitrate leaching (and N loss due to denitrification), although the straw management is not mentioned in this study. Low leaching at/around the crops EONR is expected since most of the fertiliser N is allocated to crop uptake. Low leaching at higher N fertiliser rates implies accumulation of N in the soil and potential loss by denitrification, volatilisation, and/or leaching. Considerable gaseous N loss by denitrification may be observed throughout the soil column across Luancheng, even until 10 m depth (Haijing et al., 2016), whereas volatilisation may account for 15–25% of N fertiliser rate (Zhang et al., 2004). The simulated annual N losses from volatilisation of 10, 40, 80 and 130 kg N ha−1 for N200, N400, N600

42

K. Manevski et al. / Europ. J. Agronomy 80 (2016) 32–44

and N800, respectively, should be considered as an approximation because this process is represented very simple in Daisy. Yet, losses of a similar magnitude were found in other field studies in the NCP (Ju et al., 2009; Zhang et al., 2010; Zhang et al., 2004). The simulated annual nitrification-born N2 O emissions were 6–11 kg N ha−1 , close to the field estimates of 1–6 kg N ha−1 in the NCP (Cui et al., 2012; Shi et al., 2013). Even under such conditions, the results of the longterm simulations (Fig. 7) clearly implied that the soil N pool during crop growth under N300 treatment is effectively replenished by the N from residual fertiliser N and incorporated straw and the atmosphere, in order to maintain the post-harvest soil N balance. The annual N from atmospheric deposition simulated by Daisy was 62 kg N ha−1 (60–80 kg N ha−1 measured in the NCP; Ju et al., 2009), whereas around 40–50 kg N ha−1 year−1 return was simulated from the incorporated straw (25–80 kg N ha−1 measured in the NCP; Ju and Christie, 2011). High mineralisation of straw, dead roots and abscised leaves is especially present during the maize season when both temperatures and precipitation in NCP are high (Dong et al., 2012). On the other hand, the simulated change in post-harvest soil mineral N for the annual fertiliser rates above N350 increased by 25 to more than 100 kg N ha−1 year−1 , which illustrates that mineral N did indeed accumulate in the soil. Short- and long-term field studies on silty loam soils in the NCP have shown similarly large increases in the mineral N pool below maize and winter wheat, on the background of already high levels of N in the soil due to excessive previous N fertiliser inputs (e.g., Cui et al., 2008; Fang et al., 2006; Ju et al., 2004). From the measured and simulated long-term dynamics (Fig. 5), it is evident that the soil C contents at the top (tillage) layer were increased from 1990 onwards, corresponding to an increase from 1.2% to 1.8% in SOM. The change trends were associated with the management practices of straw incorporation but also of N input that increased by the end of 1990s (Zhang et al., 2011a). Especially during the 1990s, the small tractors and livestock cultivation by the farmers has gradually been replaced by mechanical cultivation, allowing the crops straw to be chopped into small pieces and incorporated into the soil. Hence, contents of SOM were significantly improved. Other studies have shown that straw incorporation has overall positive effects on the soil fertility, especially in 0–20 cm soil layer (e.g. Zhang et al., 2016). In addition, the straw incorporation studied here i.e. incorporation of stubble (and shallow roots) from the topsoil into the subsoil where these can be mineralised and/or stabilised, requires further understanding of the fate of straw N release, including effects of different tillage practices, and it is important for fully managing the double-crop rotation, i.e. maintaining soil fertility and yield and reducing N losses. This study demonstrates how a double crop rotation in the NCP can be optimised for straw management and N fertiliser input, including the dilemmas that arise from a potential depletion of the soil N pool in one season/year, which makes it difficult to achieve high yields the following season, and large amounts of fertiliser N may be required to compensate for such depletion. An annual fertiliser rate of 300 kg N ha−1 with straw incorporation balances the concerns with respect to obtaining a high crop yield and a low environmental impact on the studied silty loam soils in the NCP. This gives an annual saving in N fertiliser use of about 40–50%, compared with the 400–600 kg N ha−1 fertiliser rate applied by the majority of the farmers in the NCP. It is the amount of N fertiliser that primarily needs to be decreased in order to reduce the problem of soil nitrate accumulation and inevitable leaching in the NCP. So far, the artificially low prices of N fertilisers in China for the last two decades have encouraged farmers to apply excessive N rates act as a “cheap form of insurance” against N-limited yield loss, but fertiliser prices ought to be increased. Thus, management and reduction of N in farming requires not only feasibility, as demon-

strated by field and modelling studies, but also adequate political commitment, effective initiatives and implementations. 4.3. Daisy modelling implications A well-quantified response of the agroecosystem to impacts of climate-soil-management should be accompanied by a comprehensive testing in order the model to produce the “right output behaviour for the right reasons” (Kelly et al., 2013) and thus to be included in any decision-making. The good practice of integrated environmental modelling was applied to Daisy for the pedo-climatic and agronomic environment in the NCP (for details see Manevski et al., 2016), and this study presents the evaluation and the application of the model. It should be emphasised that the calibrated crop parameters are not cultivar specific, i.e., they contain general characteristic for each of the two crops. Thus, the newly parameterised maize and winter wheat are a significant outcome of the study and a valuable platform for further adaptation of Daisy to new cultivars, or for setting up a new simulation study on silty loam soils across China. In a similar study in Europe, Heidmann et al. (2008) used similar approach for common parameterisation of potato in the Daisy model using field data from different sites/years. The values of all tested parameters in the sensitivity analysis varied by ±10%, but surely the uncertainty of some parameters may be higher than this range. Soil properties such as organic matter content and C/N ratio are easy to measure but spatially variable within a few meters in a field, whereas the size of the SOM pools and their C/N ratios appear to be largely related to the soil N status, but are difficult to measure (Springob and Kirchmann, 2003). In relation to parameters interactions, the results showed high sensitivity of nitrate leaching to changes in some crop parameters because the model finely couples crop growth and associated N uptake with soil N dynamics. The lower sensitivity of crop yield to the tested soil parameters was most likely due to the sensitivity analysis simulating an excess of soil N for the N400 treatment. However, the importance of the soil parameters should be considered particularly when simulating crop yields in N-restricted systems. The simulation of crop yield was notably affected by photosynthetic parameters (Fm and QEff), which clearly reflects enhanced photosynthesis and crop productivity under warm climate in the NCP compared to the cool climate in Denmark used for the original crop parameterisation. The default photosynthesis module in Daisy (Goudriaan and van Laar, 1978) does not distinguish between photosynthetic pathways, but it is affected by temperature (Fig. 3), with that of maize (C4 photosynthesis) being more responsive compared to winter wheat (C3 photosynthesis; Yamori et al., 2014). Also, the maximum leaf photosynthesis (Fm) had similar values for maize and winter wheat (Table 4), but this type of plant photosynthetic parameters is subject to considerable variability. The calibrated values were within the ranges of 3.2-7.5 g CO2 m−2 h−1 for maize and 4.7–12 g CO2 m−2 h−1 for winter wheat reported for the NCP (Guo et al., 2010). Amid the photosynthesis in Daisy, it should also be mentioned that the default module neither accounts for air CO2 concentration, so the implicit CO2 concentration equals the one in the 1980s in Denmark when the field experiments for the crops parameterisations were conducted (around 370 ppm). Modellers may include the effect of increased CO2 concentrations on crop growth according to Borgesen and Olesen (2011) for climate change simulation studies. The somewhat unsatisfactory validation of the soil water content was partly because of the problems associated with getting true values for soil water retention and hydraulics for both fields. The years involved in the calibration were wet while some validation years were dry (Table 1), so the soil hydraulics were probably calibrated for higher retention of soil water than observed (Manevski et al., 2016). Difficulties in accurate simulation of soil

K. Manevski et al. / Europ. J. Agronomy 80 (2016) 32–44

water transport in fields of the same soil type have been reported for Daisy (Kröbel et al., 2010) and other similar models (e.g. Nylinder et al., 2011). Also, validation of a model depends upon its successful calibration based on field data, and the accurate estimation of the specific parameters in a given environment. For instance, the model excludes the effects of ploughing on the mineralisation rates, thus the C/N ratio of 4.8 for the SMB pools (Table 4) aims to represent bacteria-dominated soils (C/N = 3–8) such as ploughed fields, compared to the original value of 7 that represents fungi (C/N = 5–16) more abundant in undisturbed grasslands (Frey et al., 1999). The model was able to well represent the long-term dynamics in the topsoil C as well as crop yield (Fig. 5). Overall, the results confirm the recommendations of Bellocchi et al. (2010) and Kersebaum et al. (2015) that calibration of a process-based model should include multiple variables in order to improve the simulation of soil N processes. 5. Conclusion This study utilised field measurements and Daisy model simulations in order to balance crop production and nitrate leaching using optimal straw and N fertiliser management for maize—winter wheat double crop rotation on silty loam soils in the NCP. The outcomes can be summarised as follows: i) Maize—winter wheat double crop rotation in the NCP is a highly efficient user of N resources (straw incorporation, residual fertiliser, atmospheric deposition). Based on the multi-response analyses, the most suitable N fertiliser rate for maize and winter wheat was, respectively, 100 and 200 kg N ha−1 , with acceptable yield of 8 and 6.5 Mg ha−1 , while low nitrate leaching and soil N accumulation were maintained. ii) Overall, the Mitscherlich and linear-plateau models gave more realistic estimates of the EONR on measured and simulated yield data than the quadratic response, but the choice of yield response model requires careful consideration in order to avoid overestimation of the EONR. iii) The results increase the ability to understand how to simultaneously achieve both high yield and high N use efficiency and adopt research-proven practices that maximise crop yield minimise N losses to the environment. The generated Daisy model setup serves as a powerful tool for assessment of the effects of straw and N fertiliser managements on crop production, nitrate leaching and soil N accumulation in the NCP. Acknowledgements The authors would like to thank the anonymous reviewers for their valuable comments that improved the manuscript. Authors also thank Yuan Haijing and Junqi Yang at Luancheng AgroEcological Experimental Station in Shijiazhuang, China, for help with the field data collection and translations. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.eja.2016.06.009. References Ahmadi, S.H., Plauborg, F., Andersen, M.N., Sepaskhah, A.R., Jensen, C.R., Hansen, S., 2011. Effects of irrigation strategies and soils on field grown potatoes: root distribution. Agric. Water Manage. 98 (8), 1280–1290. Belanger, G., Walsh, J.R., Richards, J.E., Milburn, P.H., Ziadi, N., 2000. Comparison of three statistical models describing potato yield response to nitrogen fertilizer. Agron. J. 92 (5), 902–908.

43

Bellocchi, G., Rivington, M., Donatelli, M., Matthews, K., 2010. Validation of biophysical models: issues and methodologies. A review. Agron. Sustainable Dev. 30 (1), 109–130. Bennett, N.D., Croke, B.F.W., Guariso, G., Guillaume, J.H.A., Hamilton, S.H., Jakeman, A.J., Marsili-Libelli, S., Newham, L.T.H., et al., 2013. Characterising performance of environmental models. Environ. Model. Softw. 40, 1–20. Borgesen, C.D., Olesen, J.E., 2011. A probabilistic assessment of climate change impacts on yield and nitrogen leaching from winter wheat in Denmark. Nat. Hazards Earth Syst. Sci. 11 (9), 2541–2553. Chen, Y., Zhang, Z., Wang, P., Song, X., Wei, X., Tao, F., 2016. Identifying the impact of multi-hazards on crop yield—a case for heat stress and dry stress on winter wheat yield in northern China. Eur. J. Agron. 73, 55–63. Chen, J.Y., 2010. Holistic assessment of groundwater resources and regional environmental problems in the North China Plain. Environ. Earth Sci. 61 (5), 1037–1047. Cui, Z., Zhang, F., Miao, Y., Sun, Q., Li, F., Chen, X., Li, J., Ye, Y., et al., 2008. Soil nitrate-N levels required for high yield maize production in the North China Plain. Nutr. Cycl. Agroecosyst. 82 (2), 187–196. Cui, F., Yan, G., Zhou, Z., Zheng, X., Deng, J., 2012. Annual emissions of nitrous oxide and nitric oxide from a wheat–maize cropping system on a silt loam calcareous soil in the North China Plain. Soil Biol. Biochem. 48 (0), 10–19. Cui, F., Zheng, X., Liu, C., Wang, K., Zhou, Z., Deng, J., 2014. Assessing biogeochemical effects and best management practice for a wheat–maize cropping system using the DNDC model. Biogeosciences 11 (1), 91–107. Dabin, Z., Pengwei, Y., Na, Z., Changwei, Y., Weidong, C., Yajun, G., 2016. Contribution of green manure legumes to nitrogen dynamics in traditional winter wheat cropping system in the Loess Plateau of China. Eur. J. Agron. 72, 47–55. Delin, S., Stenberg, M., 2014. Effect of nitrogen fertilization on nitrate leaching in relation to grain yield response on loamy sand in Sweden. Eur. J. Agron. 52 (Part B(0)), 291–296. Doltra, J., Laegdsmand, M., Olesen, J.E., 2011. Cereal yield and quality as affected by nitrogen availability in organic and conventional arable crop rotations: a combined modeling and experimental approach. Eur. J. Agron. 34 (2), 83–95. Dong, W., Hu, C., Zhang, Y., Wu, D., 2012. Gross mineralization, nitrification and N2O emission under different tillage in the North China Plain. Nutr. Cycl. Agroecosyst. 94 (2–3), 237–247. Fageria, N.K., Baligar, V.C., 2005. Enhancing nitrogen use efficiency in crop plants. Adv. Agron. 88, 97–185. Fang, Q.X., Yu, Q., Wang, E.L., Chen, Y.H., Zhang, G.L., Wang, J., Li, L.H., 2006. Soil nitrate accumulation, leaching and crop nitrogen use as influenced by fertilization and irrigation in an intensive wheat-maize double cropping system in the North China Plain. Plant Soil 284 (1-2), 335–350. Frey, S.D., Elliott, E.T., Paustian, K., 1999. Bacterial and fungal abundance and biomass in conventional and no-tillage agroecosystems along two climatic gradients. Soil Biol. Biochem. 31 (4), 573–585. Gerwitz, A., Page, E.R., 1974. An empirical mathematical model to describe plant root systems. J. Appl. Ecol. 11 (2), 773–781. Goudriaan, J., van Laar, H.H., 1978. Calculation of daily totals of the gross C02 assimilation of leaf canopies. Neth. J. Agric. Sci. 26, 10. Goulding, K., 2000. Nitrate leaching from arable and horticultural land. Soil Use Manag. 16, 145–151. Guo, R., Lin, Z., Mo, X., Yang, C., 2010. Responses of crop yield and water use efficiency to climate change in the North China Plain. Agric. Water Manage. 97 (8), 1185–1194. Haijing, Y., Qin, S., Dong, W., Hu, C., Manevski, K., Li, X., 2016. Denitrification Rates and Controlling Factors for Accumulated Nitrate in the 0-12 m Intensive Farmlands: a Case Study in the North China Plain. Pedosphere (accepted/in press). Hansen, S., Abrahamsen, P., Petersen, C.T., Styczen, M., 2012. Daisy: model use, calibration and validation. Trans. ASABE Am. Soc. Agric. Biol. Eng. 55 (4), 1315–1333. Hansen, S., 2002. Daisy Description−Equation Section One, Copenhagen University (https://daisy-model.googlecode.com/files/DaisyDescription.pdf), Copenhagen, Denmark. Heidmann, T., Tofteng, C., Abrahamsen, P., Plauborg, F., Hansen, S., Battilani, A., Coutinho, J., Doleˇzal, F., et al., 2008. Calibration procedure for a potato crop growth model using information from across Europe. Ecol. Modell. 211 (1–2), 209–223. Holcomb, J.C., Sullivan, D.M., Horneck, D.A., Clough, G.H., 2011. Effect of irrigation rate on ammonia volatilization. Soil Sci. Soc. Am. J. 75 (6), 2341–2347. Hou, R.X., Ouyang, Z., Li, Y.S., Tyler, D.D., Li, F.D., Wilson, G.V., 2012. Effects of tillage and residue management on soil organic carbon and total nitrogen in the North China Plain. Soil Sci. Soc. Am. J. 76 (1), 230–240. Hu, C., Saseendran, S.A., Green, T.R., Ma, L., Li, X., Ahuja, L.R., 2006. Evaluating nitrogen and water management in a double-cropping system using RZWQM. Vadose Zone J. 5 (1), 493–505. Huang, M., Liang, T., Ou-Yang, Z., Wang, L., Zhang, C., Zhou, C., 2011. Leaching losses of nitrate nitrogen and dissolved organic nitrogen from a yearly two crops system, wheat-maize, under monsoon situations. Nutr. Cycl. Agroecosyst. 91 (1), 77–89. Ju, X., Christie, P., 2011. Calculation of theoretical nitrogen rate for simple nitrogen recommendations in intensive cropping systems: a case study on the North China Plain. Field Crops Res. 124 (3), 450–458.

44

K. Manevski et al. / Europ. J. Agronomy 80 (2016) 32–44

Ju, X.T., Liu, X.J., Zhang, F.S., Roelcke, M., 2004. Nitrogen fertilization, soil nitrate accumulation, and policy recommendations in several agricultural regions of China. Ambio 33 (6), 300–305. Ju, X.T., Xing, G.X., Chen, X.P., Zhang, S.L., Zhang, L.J., Liu, X.J., Cui, Z.L., Yin, B., et al., 2009. Reducing environmental risk by improving N management in intensive Chinese agricultural systems. Proc. Natl. Acad. Sci. U. S. A. 106 (9), 3041–3046. Kelly, R.A., Jakeman, A.J., Barreteau, O., Borsuk, M.E., ElSawah, S., et al., 2013. Selecting among five common modelling approaches for integrated environmental assessment and management. Environ. Modell. Softw. 47 (0), 159–181. Kersebaum, K.C., Boote, K.J., Jorgenson, J.S., Nendel, C., Bindi, M., Fruhauf, C., Gaiser, T., Hoogenboom, G., et al., 2015. Analysis and classification of data sets for calibration and validation of agro-ecosystem models. Environ. Model. Softw. 72, 402–417. Kröbel, R., Sun, Q., Ingwersen, J., Chen, X., Zhang, F., Müller, T., Römheld, V., 2010. Modelling water dynamics with DNDC and DAISY in a soil of the North China Plain: a comparative study. Environ. Model. Softw. 25 (4), 583–601. Li, X., Hu, C., Delgado, J.A., Zhang, Y., Ouyang, Z., 2007. Increased nitrogen use efficiencies as a key mitigation alternative to reduce nitrate leaching in north china plain. Agric. Water Manag. 89 (1–2), 137–147. Liu, M., Yu, Z., Liu, Y., Konijn, N.T., 2006. Fertilizer requirements for wheat and maize in China: the QUEFTS approach. Nutr. Cycl. Agroecosyst. 74 (3), 245–258. Ludwig, B., Hu, K., Niu, L., Liu, X., 2010. Modelling the dynamics of organic carbon in fertilization and tillage experiments in the North China Plain using the Rothamsted Carbon Model—initialization and calculation of C inputs. Plant Soil 332 (1), 193–206. Ma, B.L., Wu, T.Y., Tremblay, N., Deen, W., McLaughlin, N.B., Morrison, M.J., Stewart, G., 2010. On-farm assessment of the amount and timing of nitrogen fertilizer on ammonia volatilization. Agron. J. 102 (1), 134–144. Manevski, K., Børgesen, C.D., Andersen, M.N., Kristensen, I.S., 2015. Reduced nitrogen leaching by intercropping maize with red fescue on sandy soils in North Europe: a combined field and modeling study. Plant Soil 388 (1–2), 67–85. Manevski, K., Børgesen, C.D., Li, X., Andersen, M.N., Abrahamsen, P., Hu, C., Hansen, S., 2016. Integrated modelling of crop production and nitrate leaching with the Daisy model. MethodsX 3, 350–363. Meng, Q., Yue, S., Chen, X., Cui, Z., Ye, Y., Ma, W., Tong, Y., Zhang, F., 2013. Understanding dry matter and nitrogen accumulation with time-course for high-yielding wheat production in China. PLoS One 8 (7), e68783. Meng, Q., Yue, S., Hou, P., Cui, Z., Chen, X., 2016. Improving yield and nitrogen use efficiency simultaneously for maize and wheat in China: a review. Pedosphere 26 (2), 137–147. Michalczyk, A., Kersebaum, K.C., Roelcke, M., Hartmann, T., Yue, S.-C., Chen, X.-P., Zhang, F.-S., 2014. Model-based optimisation of nitrogen and water management for wheat–maize systems in the North China Plain. Nutr. Cycl. Agroecosyst. 98 (2), 203–222. Moreno, F., Cayuela, J.A., Fernandez, J.E., FernandezBoy, E., Murillo, J.M., Cabrera, F., 1996. Water balance and nitrate leaching in an irrigated maize crop in SW Spain. Agric. Water Manag. 32 (1), 71–83. Mu, X., Zhao, Y., Liu, K., Ji, B., Guo, H., Xue, Z., Li, C., 2016. Responses of soil properties, root growth and crop yield to tillage and crop residue management in a wheat-maize cropping system on the North China Plain. Eur. J. Agron. 78, 32–43. Nylinder, J., Stenberg, M., Jansson, P.-E., Klemedtsson, Å.K., Weslien, P., Klemedtsson, L., 2011. Modelling uncertainty for nitrate leaching and nitrous oxide emissions based on a Swedish field experiment with organic crop rotation. Agric. Ecosyst. Environ. 141 (1–2), 167–183. Palosuo, T., Kersebaum, K.C., Angulo, C., Hlavinka, P., Moriondo, M., Olesen, J.E., Patil, R.H., Ruget, F., et al., 2011. Simulation of winter wheat yield and its variability in different climates of Europe: a comparison of eight crop growth models. Eur. J. Agron. 35 (3), 103–114. Peng, K., 2011. Changes of soil carbon and nitrogen storages under modern industrialization in a rural landscape of North China Plain. Procedia Environ. Sci. 8 (0), 81–89. Pituello, C., Polese, R., Morari, F., Berti, A., 2015. Outcomes from a long-term study on crop residue effects on plant yield and nitrogen use efficiency in contrasting soils. Eur. J. Agron.

Salazar, O., Hansen, S., Abrahamsen, P., Hansen, K., Gundersen, P., 2013. Changes in soil water balance following afforestation of former arable soils in Denmark as evaluated using the DAISY model. J. Hydrol. 484 (0), 128–139. Shi, Y., Wu, W., Meng, F., Zhang, Z., Zheng, L., 2013. et al. Integrated management practices significantly affect N2O emissions and wheat–maize production at field scale in the North China Plain. Nutr. Cycl. Agroecosys. 95 (2), 203–218. Simmelsgaard, S.E., Djurhuus, J., 1998. An empirical model for estimating nitrate leaching as affected by crop type and the long-term N fertilizer rate. Soil Use Manag. 14 (1), 37–43. Springob, G., Kirchmann, H., 2003. Bulk soil C to N ratio as a simple measure of net N mineralization from stabilized soil organic matter in sandy arable soils. Soil Biol. Biochem. 35 (4), 629–632. Sun, B., Zhang, L.X., Yang, L.Z., Zhang, F.S., Norse, D., Zhu, Z.L., 2012. Agricultural non-Point source pollution in China: causes and mitigation measures. Ambio 41 (4), 370–379. Svendsen, H., Hansen, S., Jensen, H.E., 1995. Simulation of crop production, water and nitrogen balances in two German agro-ecosystems using the DAISY model. Ecol. Model. 81 (1–3), 197–212. Tennant, D., 1975. A test of a modified line intersect method of estimating root length. J. Ecol. 63 (3), 995–1001. Valkama, E., Salo, T., Esala, M., Turtola, E., 2013. Nitrogen balances and yields of spring cereals as affected by nitrogen fertilization in northern conditions: a meta-analysis. Agric. Ecosyst. Environ. 164, 1–13. van Genuchten, M.T., Leij, F.J., Yates, S.R., 1991. The RETC Code for Quantifying the Hydraulic Functions of Unsaturated Soils, Version 1.0., U.S. Salinity Laboratory, USDA, ARS, Riverside, California. Wang, Y.Y., Hu, C.S., Ming, H., Zhang, Y.M., Li, X.X., Dong, W.X., Oenema, O., 2013. Concentration profiles of CH4, CO2 and N2O in soils of a wheat–maize rotation ecosystem in North China Plain, measured weekly over a whole year. Agric. Ecosyst. Environ. 164 (0), 260–272. Yamori, W., Hikosaka, K., Way, D., 2014. Temperature response of photosynthesis in C3, C4, and CAM plants: temperature acclimation and temperature adaptation. Photosynth. Res. 119 (1-2), 101–117. Yang, H., Yang, B., Dai, Y., Xu, M., Koide, R.T., Wang, X., Liu, J., Bian, X., 2015. Soil nitrogen retention is increased by ditch-buried straw return in a rice-wheat rotation system. Eur. J. Agron. 69, 52–58. Zhang, Y.M., Chen, D., Zhang, J.B., Edis, R., Hu, C., Zhu, A., 2004. Ammonia volatilization and denitrification loss from irrigated maize-wheat rotation field in the North China Plain. Pedosphere 14 (4), 8. Zhang, Y.M., Hu, C.S., Zhang, J.B., Chen, D.L., Li, X.X., 2005. Nitrate leaching in an irrigated wheat-maize rotation field in the North China Plain. Pedosphere 15 (2), 196–203. Zhang, X., Chen, S., Sun, H., Wang, Y., Shao, L., 2009. Root size, distribution and soil water depletion as affected by cultivars and environmental factors. Field Crops Res. 114 (1), 75–83. Zhang, Y., Dore, A.J., Ma, L., Liu, X.J., Ma, W.Q., Cape, J.N., Zhang, F.S., 2010. Agricultural ammonia emissions inventory and spatial distribution in the North China Plain. Environ. Pollut. 158 (2), 490–501. Zhang, X.Y., Chen, S.Y., Sun, H.Y., Shao, L.W., Wang, Y.Z., 2011a. Changes in evapotranspiration over irrigated winter wheat and maize in North China Plain over three decades. Agric. Water Manag. 98 (6), 1097–1104. Zhang, Y., Dore, A.J., Liu, X., Zhang, F., 2011b. Simulation of nitrogen deposition in the North China Plain by the FRAME model. Biogeosciences 8 (11), 3319–3329. Zhang, X., Wang, S., Sun, H., Chen, S., Shao, L., Liu, X., 2013. Contribution of cultivar, fertilizer and weather to yield variation of winter wheat over three decades: a case study in the North China Plain. Eur. J. Agron. 50 (0), 52–59. Zhang, P., Chen, X., Wei, T., Yang, Z., Jia, Z., Yang, B., Han, Q., Ren, X., 2016. Effects of straw incorporation on the soil nutrient contents enzyme activities, and crop yield in a semiarid region of China. Soil Tillage Res. 160, 65–72. Zhao, Z.G., Qin, X., Wang, E.L., Carberry, P., Zhang, Y.H., Zhou, S.L., Zhang, X.Y., Hu, C.S., et al., 2015. Modelling to increase the eco-efficiency of a wheat-maize double cropping system. Agric. Ecosyst. Environ. 210, 36–46.