A reappraisal of the critical nitrogen concentration of wheat and its implications on crop modeling

Field Crops Research 164 (2014) 65–73

Contents lists available at ScienceDirect

Field Crops Research journal homepage: www.elsevier.com/locate/fcr

A reappraisal of the critical nitrogen concentration of wheat and its implications on crop modeling Zhigan Zhao a,b , Enli Wang b, *, Zhimin Wang a, *, Hecang Zang a,c , Yunpeng Liu a , John F. Angus d,e a

Department of Agronomy and Biotechnology, China Agricultural University, Yuanmingyuan West Road 2, Beijing 100193, China CSIRO Land and Water and CSIRO Sustainable Agriculture Flagship, GPO Box 1666, Canberra, ACT 2601, Australia c Agricultural Economy & Information Research Center, Henan Academy of Agricultural Science, Zhengzhou 450002, China d CSIRO Plant Industry, CSIRO Sustainable Agriculture Flagship, GPO Box 1600, Canberra, ACT 2601, Australia e EH Graham Centre for Agricultural Innovation, Charles Sturt University, Locked Bag 588, Wagga Wagga, NSW 2678, Australia b

A R T I C L E I N F O

A B S T R A C T

Article history: Received 15 January 2014 Received in revised form 7 May 2014 Accepted 9 May 2014 Available online xxx

The concept of critical nitrogen (N) concentration (Ncc) has been used for both diagnostic purposes and modelling of wheat-N relations. Ncc has been derived with two contrasting approaches: one against above ground biomass (Ncc-biomass), and one against developmental stages (Ncc-stage). While the former has been claimed in diagnostic use, both approaches are adopted in wheat simulation models. This paper provides data from North China Plain (NCP) to re-exam the Ncc-stage relationships used in two widely used wheat models (APSIM and CERES) and to compare the Ncc-biomass vs. Ncc-stage relationships. The results revealed significant higher maximum and critical N concentrations in leaves of wheat in NCP than the values used in the APSIM-wheat model. Recalibration of the APSIM model with the new N concentrations led to improved simulations for wheat biomass and N uptake, particularly under low N input. Our results also show that the Ncc-stage relationship appeared to be more robust than the Ncc-biomass relationship, and it helped explain the variations in wheat Ncc-biomass curves from different regions. This likely reflects the fact that Ncc-stage curve captures the stage-driven formation of structural biomass and carbohydrate reserves of wheat, which is the main cause for N dilution. The implications of the findings on modelling of wheat–nitrogen relationships and on nitrogen management practices are also discussed. ã 2014 Elsevier B.V. All rights reserved.

Keywords: Maximum N concentration Shoot biomass Developmental stage N dilution curve APSIM

1. Introduction Nitrogen (N) concentration in shoots and leaves of field crops has been widely used as an indicator for crop nutritional status to maintain optimal growth (Cui et al., 2009; Greenwood and Barnes, 1978; Justes et al., 1994; Lemaire and Salette, 1984; Sheehy et al., 1998; Yue et al., 2012a). The concept of critical N concentration (Ncc) in aerial biomass (shoots) was defined as the minimum concentration of N necessary to achieve maximum above ground biomass at any moment of vegetative growth (Lemaire and Salette, 1984). This concept has been used in both physiologically-based crop simulation models (Brisson et al., 2003; Porter, 1993; Ritchie et al., 1985; Wang et al., 2002) and in the development of diagnostic tools to assist nitrogen management for crops (Cui et al., 2009; Justes et al., 1994; Ziadi et al., 2010). However, two different

* Corresponding author. Tel.: +61 2 6246 5964; fax: +61 2 6246 5965. E-mail addresses: [email protected] (E. Wang), [email protected] (Z. Wang). http://dx.doi.org/10.1016/j.fcr.2014.05.004 0378-4290/ ã 2014 Elsevier B.V. All rights reserved.

approaches have been used to derive the critical N concentration and its change with crop growth and development. One approach relates the critical N concentration of a crop to its developmental stage, while the other approach derives critical N concentration as a function of above ground biomass (as a biomass dilution curve). Little efforts have been made to reconcile these two approaches so far. Some crop models also define maximum and minimum N concentrations, in addition to the critical N concentration, for the simulation of crop N demand and N stresses for various processes. We refer these N concentrations (maximum, critical and minimum) to as threshold N concentrations. The Agricultural Production Systems model (APSIM) (Keating et al., 2003; Wang et al., 2002) and the CERES model (Jones et al., 2003; Ritchie et al., 1985) are two of the most commonly used models that define threshold N concentrations as functions of developmental stages of crops. For wheat crop, CERES-wheat model (Ritchie and Otter, 1985) uses the critical and minimum N concentrations for shoots derived separately for spring and winter wheat from literature data.

66

Z. Zhao et al. / Field Crops Research 164 (2014) 65–73

APSIM-wheat model uses threshold N concentrations derived separately for leaves and stems from literature data which were subsequently modified with experimental data obtained in Australia (Wang et al., 2002). The current version of APSIM-v7.5 does not distinguish spring and winter wheat for threshold N concentrations. Other crop models using a similar approach include Daisy (Hansen et al., 1991), AFRCWHEAT2 (Porter, 1993) and SPASS-wheat (Wang and Engel, 2002). Such an approach normally assumes that the critical N concentration of wheat changes only with developmental stages, and it does not change between different wheat genotypes. While both the APSIM-wheat and CERES-wheat models have been frequently applied to simulate wheat response to N applications, to the best of our knowledge, no studies have further checked the critical N concentrations used in the models with recent experimental data. Due to the decline of N concentration (Nc) with increasing crop biomass, the critical N concentration has also been derived as a negative power function of biomass, called a dilution curve. For wheat, critical N concentration dilution curves were developed for winter wheat in France (Justes et al., 1994), for spring wheat in Canada (Ziadi et al., 2010) and for winter wheat in China (Yue et al., 2012a; Zhao et al., 2012). Based on the biomass-derived critical N concentration, various diagnostic tools were established for the assessment of N status of wheat crop for the purpose of improving N management practices (Cui et al., 2009; Justes et al., 1994; Ziadi et al., 2010). This approach has also been used in crop models like STICS (Brisson et al., 2003) and CropSyst (Stockle et al., 2003). A common feature of these N dilution curves is that they describe the declining critical N concentration in wheat shoots with increasing above-ground biomass from a biomass around 1 Mg/ha up to the flowering stage of wheat. However, the derived critical N dilution curves are not consistent, they differ between geographical regions and even between different wheat cultivars (Angus, 2007; Zhao et al., 2012). These differences have been attributed to the variation in both climatic conditions and wheat genetics. In addition, previous N dilution curves based on dry weight always refer to the shoot N concentration and pay no attention to the differences between wheat organs such as leaves and stems. In this paper, we aim to present data on N concentration of winter wheat collected in the field experiments at Wuqiao, Hebei Province in the North China Plain (NCP), and compare them with the threshold N concentrations used in both APSIM-wheat and CERES-wheat models as well as the N concentration dilution curves in several other studies. The objectives are to: (1) analyse whether significant differences exist between measured threshold N concentrations for a modern winter wheat at NCP and those reported previously as well as those used in the two wheat models, (2) study how possible changes in threshold N concentrations could impact on wheat growth simulations in APSIM, (3) compare the Ncc-biomass dilution curves vs. Ncc-stage relationships where possible, and (4) discuss the need for reconciling these two Ncc approaches and the implications on crop modelling and N management practices.

2. Material and methods 2.1. Study site The study site was Wuqiao (WQ) (37
290 –37
470 N, 116
190 – 116
420 E, altitude 14–23 m above sea level, groundwater table 6– 9 m) in the middle of Heilonggang catchment in Hebei province. The average annual rainfall at the site was 550 mm (1961–2010), most of which falls in the summer months from July to September. The mean annual temperature is 12.9
C. The main cropping system is a winter wheat and summer maize rotation. The growing season for wheat is from mid-October to early June, and for maize from mid-June to early October. Average maximum and minimum temperature during the wheat cropping season from 1981 to 2010 were 13.6 and 1.6
C, respectively. The soil at the site is classified as a Calcaric Fluvisol (FAO, 1990) with a sandy clay loam texture and a potential plant available water holding capacity of 452 mm down to the depth of 2 m. On average, the topsoil (0–20 cm) had a pH of 8.12 and contained about 11.2 g kg1 organic matter, 1.1 g kg1 total N, 49 mg kg1 Olsen-P, and 132 mg kg1 exchangeable K. 2.2. Experimental data collection All the data used in this study are from field experiments (Table 1) conducted at Wuqiao site in 2009–2010 and 2010–2011 wheat seasons aimed to study the response of wheat growth to irrigation water and nitrogen inputs. All the experiments were conducted with randomized complete block design with irrigation water supply ranging from 75 mm to 375 mm per season and fertilizer-N (urea-N) application rates ranging from 0 to 330 kg/ha, each with three or four replicates. The wheat cultivar ‘SJZ15’ was sown early to mid October with plant densities ranging from 350 to 600 plants m2. Weeds, insect pests and diseases were properly controlled and the crops were not limited by other nutrients. Crop samples were collected 5–7 times from 0.2 m2 quadrates at main growth stages of over-wintering (just before frost), turning-green, jointing, booting, flowering, grain filling, and maturity. Measurements included LAI, above-ground biomass, grain yield and yield components. From the jointing stage, the above-ground biomass was measured separately for leaves, stems (including leaf sheath), glumes and grain. All plant samples were oven dried at 70
C to constant weight to measure biomass. The N concentrations of leaves, stems, glumes and grain were determined using the standard Kjeldahl method (Horowitz, 1970). Nitrogen content of leaves, stems, glumes and grain were calculated from the N concentrations and biomass. 2.3. Developmental stages of wheat The observed developmental stages of wheat in the experiments can be expressed as the Feekes stages (Large, 1954) and the decimal code or DC stages (Zadoks et al., 1974). Both the DC stages and Feekes stages are used here to investigate the measured N

Table 1 Water, nitrogen and sowing date treatments in experiments at Wuqiao. Experiment

Year

Irrigationa

N Applicationb (KgN/ha)

Sowing Dates (day/month)

Exp1 Exp2 Exp3 Exp4

2009 2009 2010 2010

W2, W3, W4 W3 W1, W2, W3, W4, W5 W2, W3

192, 270 0, 123, 192, 261, 330 123, 192, 279 0, 60, 123, 157.5, 185, 226.5, 261

10/10, 16/10, 22/10 12/10 10/10, 17/10, 24/10 12/10, 15/10

a W1-W4 represents one to four times of irrigation, each of 75 mm applied at sowing, jointing, flowering and mid grain filling, respectively. W5 – 75 mm applied at sowing, upstanding, jointing, booting and mid grain filling. b All the N was applied at sowing if N rate was less than 160 kgN/ha. Otherwise, 123 kgN/ha was applied at sowing, and the rest at jointing.

Z. Zhao et al. / Field Crops Research 164 (2014) 65–73

concentration change with developmental stages of wheat at the study site. The reason for using these two systems is that the DC stages were used in both APSIM and CERES wheat models, while the Feekes stages were recorded in the Ncc-biomass data of Justes et al. (1994) and Yue et al. (2012a). This creates an opportunity to cross compare these N concentrations where possible. 2.4. Maximum, critical, and minimum concentrations at key developmental stages The measured N concentrations (Nc) of shoots, leaves, stems, glumes and grains from all the experiments in Table 1 were plotted against the key developmental stages of wheat at Wuqiao to investigate the change in Nc with stage. For each stage, the 95% percentiles of all the N concentration values were quantified as the representative maximum N concentration to avoid the impact of any outliers. For minimum N concentrations, we looked at both the 5% percentiles and the absolute minimum value of all the measured N concentrations at each stage. In addition, the root N concentrations estimated from Zang (2009) were plotted against the key stages to compare with Nc in other wheat parts and the Nc used in APSIM model. The critical N concentration was derived using data from selected treatments of the experiments (W3N192 in experiment 1, 2, and 3, and treatment W3N185 in experiment 4) where the N rates were considered to be optimal, i.e. below which reduction in either biomass at the time of measurement or final grain yield occurred. Among these selected treatments, some N rates might be higher than optimal due to limited intervals of N rates. Therefore, the critical N concentration for shoots was calculated as the 5% percentile of all the N concentrations measured in these selected treatments at each growth stage (to avoid impact of a few outliers). For each stage, a close linear relationship exists between N concentration in shoots and those in leaves and stems. These relationships, together with the calculated critical N concentration for shoots, were used to derive the critical N concentration for leaves and stems at each developmental stage. The derived maximum, critical and minimum N concentrations were then compared with those used in APSIM-wheat and CERESwheat models. In addition, the derived critical N concentrations were compared with the critical N concentrations of Cui et al. (2009), Justes et al. (1994), Yue et al. (2012a) and Zhao et al. (2012) for the same phenological stages where possible to analyse the difference between them. 2.5. Nitrogen concentration curves with increasing biomass (N dilution curves) With all the measurements at Wuqiao, shoot (leaves plus stems) N concentration of wheat was plotted against above ground biomass (W) to derive the N dilution curve using a negative power function, i.e. Nc = aWb. The range of shoot N concentrations and the derived dilution curve were then compared with those in Cui et al. (2009), Justes et al. (1994), Yue et al. (2012a), Ziadi et al. (2010), and Zhao et al. (2012) to analyse the differences between them. 2.6. APSIM modelling of wheat growth response to N fertilizer inputs The APSIM model (Keating et al., 2003; Wang et al., 2002) version 7.5 was used to simulate the biomass growth, grain yield, nitrogen uptake and partitioning to different organs of wheat against the experimental data collected in experiment 2, covering 5 levels of N inputs ranging from 0 to 330 kg/ha. In APSIM, the crop has a defined minimum, critical and maximum N concentration for each plant part (leaf, stem and root), dependent on phenological

67

stages. Demand for N in each part attempts to maintain N at the critical (non-stressed) level. N demand on any day is the sum of the demand from the pre-existing biomass of each part required to reach critical N content, plus the N required to maintain critical N concentrations in newly produced biomass. Total crop N demand is the sum of the N demand in all vegetative parts. Grain N demand starts at anthesis and is calculated from grain number, thermal time and a potential grain nitrogen filling rate. Grain N demand is met by re-translocating N from other plant parts. N is available for re-translocation from leaves and stems until they reach their defined minimum N concentration. N supply from the soil was simulated as the sum of N supply by mass flow (N with transpiration stream) and diffusion processes. If the plant shoot (leaf and stem) N concentrations fall below the critical N concentrations, different N stress factors based on the actual and critical N concentrations were calculated to reduce leaf expansion, biomass growth, and grain N filling. Model calibration involved adjustments to cultivar parameters for correct simulation of phenology for cultivar SJZ15, with the derived vernalisation sensitivity (vern_sens) of 2.3, photoperiod sensitivity (photop_sens) of 3.5 and thermal time for grain filling of 530
Cd. The model was firstly run with the original threshold N concentrations (maximum, critical and minimum). Then the model was re-run again by replacing the original maximum and critical N concentrations with the ones derived in this study using the Wuqiao data. The impact on simulation results were analysed and discussed. 3. Results 3.1. Nitrogen concentration ranges and changes with developmental stages Fig. 1 shows the range and the change of measured N concentration of different wheat organs plotted against wheat developmental stages expressed as the DC stage (decimal code) of Zadoks et al. (1974). For any given stage, observed N concentrations showed significant variation due to various input levels of irrigation and N fertilizer applications. Generally, the N concentration in shoots and leaves increased from tillering stage (DC13) before winter to jointing stage (DC32), and then decreased towards maturity (DC87) (Fig. 1a and b). The peak N concentration of shoots, leaves and stems occurred at jointing stage (Fig. 1a–c). This trend of change in N concentration is consistent with the data of Zhu et al. (2002) in Jiangsu Province, China. N in glumes (Fig. 1d) decreased rapidly from booting (DC47) to maturity. The initial increase in shoot and leaf N concentration up to jointing stage, likely reflects possible leaf senescence before winter and the rapid exploration of soil by roots and high N uptake rates relative to shoot growth after winter (Sinclair et al., 2000). Maximum N concentrations, defined as the 95 percentile of the measurement data at each stage, were higher in leaves than in stems and in shoots (Figs. 1–3 ). From jointing stage to early grain filling stage (10 days after flowering, DC74), maximum N concentration in leaves declined from 5.37% to 3.43%, and that in stems and shoots decreased from 3.97% to 1.25% and from 4.73% to 1.78%, respectively. At maturity, maximum N concentration in leaves, stems, shoots and grains were 1.99%, 1.09%, 1.21% and 2.53%, respectively. The data from Zang (2009) also showed a wide range of N concentrations in roots of wheat due to different cultivars (Fig. 1e). Interestingly, the maximum root N concentration of winter wheat in NCP also seemed to occur at jointing stage, coincident with the maximum leave/shoot concentration. As compared with leaves and stems, roots had lower N concentrations in most part of the growing season.

68

Z. Zhao et al. / Field Crops Research 164 (2014) 65–73

(a) Leaves

Stem N concentraon (%)

Leaf N concentraon (%)

8 6 4 2 0 23

36

49

62

DC stage

75

(c) Roots

3 2 1 0

Shoot N concentraon (%)

The shoot critical N concentration estimated from the Wuqiao data were 3.53%, 2.34%, 1.86% and 0.67% at jointing, booting, flowering and maturity, respectively (Fig. 2). These values were very similar to those used in the CERES-wheat model, except for a slightly higher value at flowering stage (DC65). However, the derived shoot critical Nc at tillering stage (DC13) was much lower than those in CERES-wheat. The maximum shoot N concentrations derived from the Wuqiao data were on average 36.8% higher than the critical shoot N concentrations (Fig. 2).

8

(a) Shoots CERES_Crit (SW) CERES_Crit (WW) CERES_Min WQ_Max WQ_Crit WQ_Min

6 4 2 0 10

23

36

49

62

75

88

DC stage Fig. 2. Comparison of the derived wheat N concentrations (max, min and critical) at Wuqiao with that used in CERES for shoots: SW for spring wheat, and WW for winter wheat.

4 2

10

23

36

49

62

DC stage

4

75

88

75

88

(d) Grains

3 2 1 0

10

3.2. Critical shoot N concentrations as compared with those in CERES

APSIM_Max APSIM_Crit APSIM_Min WQ_Max WQ_Crit WQ_Min

6

88

Grain N concentraon (%)

Root N concentraon (%)

4

Grain N concentration increased from early to late grain filling (Fig. 1d). The observed grain N concentration at maturity ranged from 1.8 to 2.6%, corresponding to a range of grain protein from 10 to 15%.

(b) Stems

0 10

Fig. 1. The range of measured N concentration in shoots (a), leaves (b), stems (c), glumes (d), roots (e) and grains (f) of winter wheat at key decimal code stages (Zadoks et al., 1974). All the data except those in (e) were from Wuqiao. Data for (e) was from Zang (2009) in North China Plain. The open circle indicates the mean. The solid line indicates the median. The box boundaries indicate the 75 and 25% quartiles, and the whisker caps indicate the 95th and 5th percentiles. The asterisks are the outliers.

8

23

36

49

62

DC stage

75

88

10

23

36

49

62

DC stage

Fig. 3. Comparison of the derived wheat N concentrations (max, min and critical) at Wuqiao with that used in APSIM for stems (a), leaves (b), roots (c) and grains (d) respectively.

The critical shoot N concentrations in CERES-wheat was derived separately for winter and spring wheat based on 15 data sources in the USA, UK, India and Australia (http://nowlin.css.msu.edu/ wheat_book/CHAPTER4.html). It is noticed that the critical N concentrations of winter and spring wheat in CERES-wheat coincided with each other after jointing stage. Before jointing stage, the critical shoot Nc in CERES-wheat had a maximum value of 6.0% for spring wheat, and 4.2% for winter wheat. The maximum value of shoot critical Nc from previous studies in China were around 4% (Yue et al., 2012; Zhao et al., 2012). From this consistent value of 4%, it seems that the derived value of 2.7% from the Wuqiao data was an underestimation, which may also reflect the likely effects of leaf senescence approaching winter. 3.3. Critical and maximum N concentrations as compared with those in APSIM The critical and maximum N concentrations in leaves estimated based on the Wuqiao data were similar to the values used in APSIM at jointing stage, but much lower before and much higher after the jointing stage as compared to the APSIM values (Fig. 3a). From jointing stage onwards, the biggest difference of leaf N concentration was at flowering stage, where the N concentrations measured at Wuqiao were about twice as compared to those used in APSIM. Such discrepancy was not observed in the N concentrations of stems. Both the maximum and critical N concentration curves of stems derived from the Wuqiao data were close to that used in APSIM, except for stages before jointing and at maturity (Fig. 3b). The maximum N concentrations in roots from the WQ data was similar to the value used in APSIM at booting stage, but lower before and higher after the booting stage as compared to the APSIM values (Fig. 3c). Measured maximum N concentrations in roots decreased from 2.6% at jointing to 1.5% at maturity. In APSIM, the critical and maximum N concentrations in roots are a constant value of 2% and the minimum root N concentration is 1%, independent of stages. The maximum N concentrations in grains from the WQ data were lower than the value used in the APSIM (Fig. 3d). From start grain filling to maturity, the measured maximum N concentration

Z. Zhao et al. / Field Crops Research 164 (2014) 65–73

(a)

Nc Data WQ Nc_WQ (regression) Nc_Max_Justes Nc_Min_Justes Nc_Crit_Justes Nc_Crit_Yue Nc_Yue (regression)

6 5 4 3 2 1

5

Crical N concentraon (%)

Shoot N concentraon (%)

7

0

69

(b) Justes et al. (1994) Yue et al. (2012a) Y16 in Zhao et al. (2012) N13 in Zhao et al. (2012) Ziadi et al. (2010) Lamaire et al. (1990) Greenwood et al. (1990)

4 3 2 1

0

2

4

6

8

10

12

1

Above-ground biomass (Mg ha-1)

3

5

7

9

11

Aboveground biomass (Mg ha-1)

Fig. 4. Changes of shoot N concentration with above-ground biomass of wheat (>1 t ha1 until flowering) (a), and comparison of critical shoot N concentrations dilution curves (changes with biomass) of spring wheat (Ziadi et al., 2010), general C3 species (Lemaire et al., 1990; Greenwood et al., 1990) and winter wheat (the rest) from different studies (b). In (a), the open circles show the data measured at Wuqiao (Nc Data WQ, 368 data points) with a dilution curve Nc_WQ = 4.7DM0.36. The other curves show the shoot N concentration of wheat derived by Justes et al. (1994) and Yue et al. (2012a). In (b), Y16 and N13 represent two cultivars of winter wheat mainly sown in Jiangsu Province, China.

3.4. Comparison of critical shoot Nc derived from biomass dilution curves Fig. 4a shows the measured shoot N concentration of wheat at Wuqiao plotted against the above-ground biomass, as compared with the N dilution curves derived in previous studies in Europe (Justes et al., 1994) and NCP (Yue et al., 2012a). All the N dilution curves were derived for above-ground biomass up to flowering stage, i.e. not including data for the reproductive period. For biomass range of 3–6 t/ha, the range of measured shoot N concentrations at Wuqiao in NCP was similar to the range defined by the maximum and minimum N concentration given by Justes et al. (1994). At higher biomass level the Nc at WQ seems to be lower. In fact, the N dilution curve derived using the Wuqiao data (regression curve) was very similar to the critical N concentration curve of Justes et al. (1994), but significantly higher than the N dilution curve (regression) and the critical shoot N concentration derived by Yue et al. (2012a) in NCP. Both the N dilution curve (regression) and critical shoot N concentration of Yue et al. (2012a) coincided with each other. The lower N concentration values measured at Wuqiao when above ground biomass was below 23 tonne/ha were likely caused by the onset of leaf senescence before winter. When critical N concentration dilution curves from different studies were compared, differences were noticed not only between studies from different geographical regions (e.g. Justes vs. Yue and Zhao), but also between winter wheat cultivars in the same region (e.g. Zhao_Y16 vs. Zhao_N13) (Fig. 4b). For the same amount of above ground biomass, the critical shoot N concentration of spring wheat was much lower than that of winter wheat. For winter wheat, the critical shoot N concentrations derived in

China (Yue et al., 2012a; Zhao et al., 2012) was constantly lower than that derived in Europe (Justes et al., 1994) in the whole biomass range (Fig. 4b). It is obvious that the shoot Ncc-biomass curves derived in Europe will not be applicable to China and vice versa. However, when the shoot critical N concentrations were plotted against development stages, the curves from different studies tended to converge, particularly after stem elongation stage (Feekes stage 6–7 or DC stage 30–32) (Fig. 5). Thereafter, the critical shoot N concentrations derived from four sources of data (Justes et al., 1994; Yue et al., 2012a; Zhao et al., 2012, and data from this study) nearly coincided. At biomass levels of 2, 6 and 9 tonne/ ha, the differences in the derived shoot Ncc between those of Justes et al. (1994) and Zhao et al. (2012) were 0.77%, 0.49% and 0.41%, respectively (Fig. 4b). However, at DC stage 31, 39 and 65, the corresponding differences in shoot Ncc became 0.4%, 0.2% and 0.2%, and there was no systematic departure between the values from the four studies (Fig. 5b). This clearly indicates that the shoot Ncc-stage relationship has the potential to unify the shoot Ncc for different wheat cultivars and from different climatic regions. The bigger difference between shoot critical N concentrations of Justes et al. (1994) in Europe and Yue et al. (2012a) in NCP at early stages (Feekes stage 4–5) might be caused by inconsistencies and difficulties in measuring biomass and recording growth stage when the plant size was very small. Comparing Fig. 5a and b it is noticed that Feekes stage 4 and 5 are very close to each other. From definition, they are also very difficult to distinguish. The derived critical shoot Nc values from this study are consistent with the critical shoot N concentration derived by Justes

Crical N concentraon (%)

in grains increased from 2.2% to 2.6%. In APSIM, the critical and maximum N concentration in grains is given as constant value of 3%, and the minimum grain N concentration is 1.4%, representing the threshold N concentrations at maturity. As expected, the minimum N concentrations derived from the Wuqiao data were much higher than those used in APSIM-wheat. They cannot represent the true minimum threshold of N concentrations in wheat, which was defined as a lower limit at which metabolism of tissues ceases to function (Penning de Vries, 1982). The values used in APSIM are likely to better reflect the true minimum values than the derived values because of the relatively high N input levels and N content in soil in the experiments at Wuqiao. In subsequent simulations with APSIM, the minimum concentrations in APSIM were used, instead of those derived from the Wuqiao data.

6

(a)

5

6 WQ Justes et al. (1994 ) Yue et al. (2012 a)

4

Zhao et al. (2012 )

(b)

5 4

3

3

2

2 1

1 3

5

7

9

Feekes stage

11

13

25 31 37 43 49 55 61 67

DC stage

Fig. 5. Comparison of critical shoot N concentrations of winter wheat from studies of Justes et al. (1994), Yue et al. (2012a) and Zhao et al. (2012) against Feekes stage (Large, 1954) of wheat (a) and against decimal code stages (Zadoks et al., 1974) of wheat (b). , together with the data from Wuqiao in NCP.

70

Z. Zhao et al. / Field Crops Research 164 (2014) 65–73

16

20 16 12

13 10 7 4

8 12

16

20

15

Simulated (g/m2 )

(d) Biomass N

30 20 10 10

20

30

9 6

10

13

3

16

Measured (t/ha)

Measured (t/ha) 40

(c) Grain DM

3 7

4

24

40

30

(e) Straw N

10 5 0 0

Measured (g/m2 )

5

10

6

9

12

Measured (t/ha)

Simulated (g/m2 )

8

Simulated (g/m2)

12

(b) Straw DM

Simulated (t/ha)

(a) Biomass

Simulated (t/ha)

Simulated (t/ha)

24

15

Measured (g/m2)

(f) Grain N

24 18 12 6 6

12

18

24

30

Measured (g/m2)

Fig. 6. Comparison of simulated and observed dry matter and N content in above ground parts (a, d), straw (b, e), and grain (c, f) of wheat at Wuqiao in NCP, China: blue solid circles with original version of APSIM 7.5 and red solid circles with APSIM v7.5 adjusted for critical and maximum N concentrations in leaves and stems based on the Wuqiao data. Blue dashed lines were the 1:1 lines.

et al. (1994), Yue et al. (2012a) and Zhao et al. (2012) (Fig. 5). In addition, at maturity (DC87), the estimated critical N concentration in shoots was 0.67% (Fig. 2), and was also very close to the result of Cui et al. (2009) who showed the critical straw N concentration of wheat in NCP was 0.68%. 3.5. Impact of critical and maximum N concentrations on APSIM simulation results Based on our results in Fig. 3, we tried to modify the threshold N concentrations in APSIM and test the impact of modification on simulation results. For leaves and stems, we adopted the maximum and critical N concentrations derived from the WQ data, but the minimum concentrations from APSIM (Fig. 3a and b) for reasons described in previous sections. For roots and grains, we kept the APSIM threshold N concentrations unchanged, because the root data did not allow us to derive critical N concentrations, and the maximum grain N concentration is only used to estimate grain N demand. Modifications to the critical and maximum concentrations of leaves and stems in APSIM based on the data at Wuqiao led to several changes in the simulation results (Figs. 6–8 ), although both versions could well capture the crop response to N input levels. Compared to the results from the original APSIM v7.5 setup, the modified model simulated less straw and grain biomass (Fig. 6a–c), less grain N (Fig. 6f), but slightly more N in straw (Fig. 6e). There was a noticeable overestimation of the biomass with APSIM 7.5 and underestimation of the grain yield with the modified model under low N input (Fig. 6a and c), though no significant changes in the N content in biomass and grain (Fig. 6d and f). In general, the modifications led to improved simulation results (Table 2, Fig. 7), particularly for the low N (N0) treatment. They resulted in a better match of simulated dynamics of LAI (Fig. 7a) and biomass (Fig. 7e) under the N0 treatment and of leaf N content in both the N0 and N330 treatments (Fig. 7c and d). The impact on total N uptake in the above ground biomass was insignificant (Fig. 7g and h). These improvements of simulation results for LAI, N in leaf and biomass were caused by the increased N stress for leaf area and biomass growth (Fig. 8) as a result of increased critical N concentrations in leaves since jointing stage (Fig. 3a). The modified

version simulated much more N stress for leaf growth under low N input levels. 4. Discussion 4.1. Confidence in the derived threshold N concentrations using the WQ data By definition, the critical N concentration was the minimum N concentration below which growth is reduced at any given time during the growing period. While the critical N concentrations derived in this study were from the N treatments with the lowest N input levels below which wheat growth at the time of measurement and/or yield were reduced. From the aspect of biomass reduction at measurement time, the method used in this study (for derivation of Ncc) is consistent with the method used in previous studies (Justes et al., 1994; Lemaire and Salette, 1984). In addition, we also tried to refer to final yield changes as an additional measure to ensure that the derived critical N concentration would not result in yield reduction. However, due to the limited number of N levels in the treatments, it is possible that the values derived from this study (Fig. 3a) may overestimate the critical N concentrations because it was not possible to determine which N treatments were the ‘real’ optimal ones. Nevertheless, the close match of the derived shoot critical N concentration to that in previous studies (Justes et al.,1994; Yue et al., 2012a; Zhao et al., 2012) (Fig. 5) and in CERES-wheat (Fig. 2), together with the close agreement of the derived maximum and critical stem N concentrations to those in APSIM-wheat clearly indicate that the derived values are plausible. We have much more confidence in the maximum N concentration of leaves and stems derived from the WQ data, because they are derived from all the experimental data collected from a region with generally high N application rates. The much higher maximum and critical N concentrations in leaves of wheat cultivar SJZ15 may also reflect the changed N status of the modern new wheat varieties. 4.2. Reconciliation of the Ncc-biomass and Ncc-stage curves The data on N concentration of wheat in NCP presented in this paper, together with the comparisons to previously published data,

Z. Zhao et al. / Field Crops Research 164 (2014) 65–73

APSIM 7.5 APSIM (Modified) Measured

8

LAI (m2/m2)

10

(a) N0

6

8 6

4

4

2

2

0

0

10/10 09/12 07/02 08/04 07/06 06/08

N in Leaf (g/m2)

10 8 6 4 2 0

3 0

Biomass (Mg/ha)

10/10 09/12 07/02 08/04 07/06 06/08

20

(e) N0

15

10

10

5

5

0

0

18 15 12 9 6 3 0

(d) N330

12 9 6

10/10 09/12 07/02 08/04 07/06 06/08

N in Biomass (g/m2)

10/10 09/12 07/02 08/04 07/06 06/08

15

(f) N330

10/10 09/12 07/02 08/04 07/06 06/08

40

(g) N0

(h) N330

32 24 16 8 0

10/10 09/12 07/02 08/04 07/06 06/08

Date (DD/MM)

R2

Treatment

15

(c) N0

10/10 09/12 07/02 08/04 07/06 06/08

20

Table 2 Comparison of R2 and RMSE between APSIM and modified version with measured data.

(b) N330

10/10 09/12 07/02 08/04 07/06 06/08

Date (DD/MM)

Fig. 7. Comparison of APSIM (APSIM 7.5, and APSIM modified with the critical and maximum N concentrations (in Fig. 3)) simulated dynamics of LAI (a, b), nitrogen in leaf (c, d), aboveground biomass (e, f) and nitrogen in aboveground biomass (g, h) with measured data under treatments of no N application (N0) and 330 kg N/ha (N330) in Wuqiao, NCP.

did not indicate that the winter wheat in NCP had in general a lower plant N concentration as compared to wheat in Europe and some other countries (Cui et al., 2009; Yue et al., 2012b), particularly from jointing stage onwards. The differences found in the critical N concentration (Ncc) from N-biomass dilution curves between regions and wheat cultivars are likely to reflect potential issues in the method used. The significant convergence of critical shoot N concentrations derived from different sources when they were redrawn against developmental stages (Fig. 5) clearly indicates some advantages of relating critical N concentration to developmental stages instead of biomass. Reconciling the

Biomass (t/ha) Straw DM (t/ha) Grain DM (t/ha) Biomass N (g/m2) Straw N (g/m2) Grain N (g/m2)

1

1

0.8

0.8

0.6

0.6

0.4 0.2 0 02/01

(b) Leaf expansion

0.4

APSIM 7.5

0.2

APSIM (Modified)

03/03

02/05

Date (DD/MM)

01/07

0 02/01

Shoot biomass (Mg/ha)

N stress factor

1.2

(a) RUE

APSIM (Modified)

APSIM

APSIM (Modified)

0.99 0.95 0.98 0.95 0.66 0.96

0.99 0.95 0.98 0.96 0.72 0.96

1.6 1.5 0.3 37.5 18.4 36.2

1.2 1.4 0.4 34.4 17.5 31.1

Ncc-biomass and Ncc-stage approaches may lead to improved understanding of wheat–N relationships and benefit to modelling of wheat N dynamics. Even under optimal N supply, the N concentration of wheat shoot declines with time from the start of the growing season, i.e. it declines with both increasing biomass and the progressing developmental stages. The declining N concentration reflects an increasing proportion of structural tissue (i.e. stems) with a low N concentration, the dilution of N by carbohydrate reserves, and increasing mutual shading of leaves in the later part of the growing season (Angus and Moncur, 1985). Partitioning of biomass to leaves and stems and the formation of structural tissues (stems, glumes) and carbohydrate reserve (stems, grains) are more stage-dependent than biomass driven, particularly when wheat is grown in different environments under different stress (water or N) conditions. In addition, mutual shading of leaves mostly occurs at later stages when structural tissues (stems) are fast growing. For example, at the same developmental stage, a winter wheat grown in Europe and NCP would have a similar fraction of structural biomass (because it is stage-dependent). However, at the same stage, the wheat crop normally has higher shoot biomass in Europe than in NCP (Fig. 9a), due to relatively longer growing period and more favourable weather conditions (lower spring temperature and more rainfall) in Europe and possible cultivar difference (Wang, 1997; Wang and Engel, 1998). If the critical shoot N concentration is stage-dependent rather than biomass-driven, combining the stage-dependent shoot Ncc derived from the WQ data in Fig. 5b and the shoot biomass data in Fig. 9a should allow us to re-produce the patterns in the Ncc-biomass curves in Fig. 4b. Fig. 9b shows the results of so-calculated dependence of critical shoot N concentration in relation to shoot biomass. It clearly indicates that at a given biomass level the winter wheat in NCP has a lower shoot Ncc than the winter wheat in Germany, mimicking the patterns in the N dilution curves for shoot Ncc in Fig. 4b (Justes et al., 1994; Yue et al., 2012a; Zhao et al., 2012). This explains why the critical shoot N concentrations derived for winter wheat in China (Yue et al., 2012a; Zhao et al., 2012) was constantly lower 25

1.2

RMSE

APSIM

(a)

Crical N concentraon (%)

10

71

20 15 10 5 0

03/03

02/05

01/07

25

38

51

64

DC stage

77

90

4

(b) SJZ15 (China)

3

Kanzler (Germany) Oress (Germany)

2

1

0 0

10

20

Shoot biomass (Mg/ha)

30

Date (DD/MM)

Fig. 8. Comparison of simulated N stress factors for radiation use efficiency (RUE) (a) and Leaf expansion (b) under the N0 treatment (no N application) using the original APSIM v7.5 and APSIM v7.5 modified with the critical and maximum N concentrations (in Fig. 3).

Fig. 9. Relationship between shoot biomass and DC stage (a), shoot critical N concentration and shoot biomass (b) for three winter wheat cultivars grown in different regions under high nitrogen input conditions: SJZ15 in the N330 treatment at Wuqiao, NCP; Kanzler and Orestis in north Germany, data from McVoy et al. (1995).

72

Z. Zhao et al. / Field Crops Research 164 (2014) 65–73

than that derived in Europe (Justes et al., 1994) at the same amount of shoot biomass (Fig. 4b). Our results clearly demonstrate the needs to reconcile the stage-driven and biomass-driven approaches for defining the critical N concentrations of wheat crops. The more stable Ncc-stage relationship found in this study likely reflects the stage-dependent formation of structural tissues and carbohydrate reserves. The ability to explain regional variations in the Ncc-biomass curves by using a constant Ncc-stage relationship also implies a wider applicability of the Ncc-stage relationship. 4.3. Implication on modelling of wheat-nitrogen relationship The differences in N concentration limits (maximum and critical) for leaves and stems of wheat between the values derived from the Wuqiao data and the ones used in APSIM highlight the need for a re-appraisal of the N relationships used in the model. Similar to other most commonly used crop models, some algorithms and parameterisation in APSIM was developed using specific datasets collected more than 10–20 years ago. Rigorous updates with new data and improved understanding are needed in order to meet the increasing demand for the model to be confidently applied in research areas and geographical regions other than those where the model was developed. Use of the newly derived threshold N concentrations in APSIM led to improvement in simulations of both crop biomass growth and N uptake, particularly under low N treatments at the study site. Further validation of the derived critical N concentration of wheat (leaves, stems and roots) is needed with data for modern wheat cultivars from different experiments. The findings of this study implies that a re-analysis of the Ncc-biomass data for Ncc dependency on developmental stages and new studies relating organ N concentration to the developmental stages of wheat, rather than above-ground biomass, would offer greater benefit for improved understanding of the concept of critical N concentration of crops. In addition, separately defining the Ncc for leaves and stems is important because plant growth has to be considered as the sum of metabolic (e.g. leaves) and structural (e.g. stems) compartments, each with its own demand for metabolic and structural N, in order for crop simulation models to capture the genotypic and environmental control of crop N dynamics in a physiologically functional manner (Lemaire et al., 2007). Our results also indicate that with the higher leaf maximum and critical N concentrations, the N stress index on radiation use efficiency (RUE, Fig. 8a) simulated in the APSIM model may need to be lessened and experimentally verified in order to better simulate crop growth and yield formation. This implies that N stress would have less impact on RUE, but more impact on leaf expansion growth as compared to what is currently simulated in the APSIMwheat model. This may be true for water stress as well, and is consistent with the experimental findings of Meinke et al. (1997) who showed that RUE of wheat did not differ between different water and N treatment, while the leaf area index was severely reduced by N stresses. Acknowledgements This study was supported by the National Basic Research Program of China (973 Program, 2012CB955904), by the Earmarked Fund for Modern Agro-Industry Technology Research System (CARS-3), by the Crop High Yield Technology Engineering Program (2011BAD16B14), China, by CSIRO and Chinese Academy of Sciences (CAS) through the research project 'Advancing crop yield while reducing the use of water and nitrogen', and by CSIRO and Chinese Ministry of Education (MoE) through the CSIRO-MoE

PhD Research Program. Financial supports from the above sources are gratefully acknowledged.

References Angus, J., 2007. Should nitrogen dilution curves be expressed in relation to biomass or development? In: Bosch, A., Teira, M.R., Vilar, J.M. (Eds.), Towards a Better Efficiency in N Use. Editorial Milenio, Leida: Spain, pp. 305–308. Angus, J., Moncur, M., 1985. Models of growth and development of wheat in relation to plant nitrogen. Crop Pasture Sci. 36, 537–544. Cui, Z., Zhang, F., Dou, Z., Yuxin, M., Sun, Q., Chen, X., Li, J., Ye, Y., Yang, Z., Zhang, Q., 2009. Regional evaluation of critical nitrogen concentrations in winter wheat production of the North China Plain. Agron. J. 101, 159–166. FAO, 1990. FAO-UNESCO soil map of the world. vol. VIII. UNESCO, Paris. Greenwood, D.J., Barnes, A., 1978. A theoretical model for the decline in the protein content in plants during growth. J. Agric. Sci. 91, 461–466. Greenwood, D., Lemaire, G., Gosse, G., Cruz, P., Draycott, A., Neeteson, J., 1990. Decline in percentage N of C3 and C4 crops with increasing plant mass. Ann. Bot. 66, 425–436. Hansen, S., Jensen, H., Nielsen, N., Svendsen, H., 1991. Simulation of nitrogen dynamics and biomass production in winter wheat using the Danish simulation model DAISY. Fertilizer Research 27, 245–259. Horowitz, W., 1970. Official methods of analysis. Association of Official Analytical Chemists, Washington, DC. Justes, E., Mary, B., Meynard, J.-M., Machet, J.-M., Thelier-Huche, L., 1994. Determination of a critical nitrogen dilution curve for winter wheat crops. Ann. Bot. 74, 397–407. Keating, B.A., Carberry, P., Hammer, G., Probert, M.E., Robertson, M., Holzworth, D., Huth, N., Hargreaves, J., Meinke, H., Hochman, Z., 2003. An overview of APSIM, a model designed for farming systems simulation. European Journal of Agronomy 18, 267–288. Large, E.C., 1954. Growth stages in cereals illustration of the Feekes scale. Plant Pathology 3, 128–129. Lemaire, G., Salette, J., 1984. Relationship between growth and nitrogen uptake in a pure grass stand. 1. Environmental effects [nitrogen nutrition, nitrogen content, allometric relations, variability between years, tall fescue]. Agronomie 4, 423– 430. Lemaire, G., Oosterom, E.V., Sheehy, J., Jeuffroy, M.H., Massignam, A., Rossato, L., 2007. Is crop N demand more closely related to dry matter accumulation or leaf area expansion during vegetative growth? Field Crops Res. 100 (1), 91–106. McVoy, C.W., Kersebaum, K.C., Arning, M., Kleeberg, P., Othmer, H., Schröder, U., 1995. A data set from north Germany for the validation of agroecosystem models: documentation and evaluation. Ecol. Model. 81 (1), 265–300. Meinke, H., Hammer, G.L., van Keulen, H., Rabbinge, R., Keating, B.A., 1997. Improving wheat simulation capabilities in Australia from a cropping systems perspective: water and nitrogen effects on spring wheat in a semi-arid environment. Developments in Crop Science 25, 99–112. Penning de Vries, F., 1982. Crop production in relation to availability of nitrogen. In. In: Penning de Vries, F.W.T., van Laar, H.H. (Eds.), Simulation of Plant Growth and Crop Production’. Wageningen, PUDOC (:). Porter, J., 1993. AFRCWHEAT2: a model of the growth and development of wheat incorporating responses to water and nitrogen. European Journal of Agronomy 2, 69–82. Ritchie, J.T., Godwin, D.C., Otter-Nacke, S., 1985. CERES- Wheat. A Simulation Model of Wheat Growth and Development. Texas A and M University Press, College Station, TX, USA. Sheehy, J., Dionora, M., Mitchell, P., Peng, S., Cassman, K., Lemaire, G., Williams, R., 1998. Critical nitrogen concentrations: implications for high-yielding rice (Oryza sativa L.) cultivars in the tropics. Field Crops Res. 59, 31–41. Sinclair, T., Pinter Jr, P., Kimball, B., Adamsen, F., LaMorte, R., Wall, G., Hunsaker, D., Adam, N., Brooks, T., Garcia, R., 2000. Leaf nitrogen concentration of wheat subjected to elevated [CO2] and either water or N deficits. Agric. Ecosyst. Environ. 79, 53–60. Stockle, C., Donatelli, M., Nelson, R., 2003. CropSyst, a cropping systems simulation model. European Journal of Agronomy 18, 289–307. Wang, E., Engel, T., 1998. Simulation of phenological development of wheat crops. Agricultural Systems 58, 1–24. Wang, E., Engel, T., 2002. Simulation of growth, water and nitrogen uptake of a wheat crop using the SPASS model. Environmental Modelling & Software 17, 387–402. Wang, E., Robertson, M., Hammer, G., Carberry, P., Holzworth, D., Meinke, H., Chapman, S., Hargreaves, J., Huth, N., Lean, Mc, G, 2002. Development of a generic crop model template in the cropping system model APSIM. European Journal of Agronomy 18, 121–140. Yue, S., Meng, Q., Zhao, R., Li, F., Chen, X., Zhang, F., Cui, Z., 2012a. Critical Nitrogen Dilution Curve for Optimizing Nitrogen Management of Winter Wheat Production in the North China Plain. Agron. J. 104, 523–529. Yue, S., Meng, Q., Zhao, R., Ye, Y., Zhang, F., Cui, Z., Chen, X., 2012b. Change in Nitrogen Requirement with Increasing Grain Yield for Winter Wheat. Agron. J. 104, 1687– 1693. Zadoks, J.C., Chang, T.T., Konzak, C.F., 1974. A decimal code for the growth stages of cereals. Weed Research 14, 415–421.

Z. Zhao et al. / Field Crops Research 164 (2014) 65–73 Zang, H., 2009. Clustering analysis on root traits and developmental difference between seminal and nodal roots of leading wheat (Triticum aestivum L.) Cultivars in Henan Province. M.Sc. Thesis. Henan Agricultural University (in Chinese with English abstract). Zhao, B., Yao, X., Y, -c., X, -j., W, -x., Zhu, Y., 2012. Accumulative nitrogen deficit models of wheat aboveground part based on critical nitrogen concentration. Chin. J. Appl. Ecol. 23, 3141–3148 (in Chinese with English abstract).

73

Zhu, X., Yan, L., Guo, W., Feng, C., Peng, Y., 2002. Carbon-nitrogen Characteristics of super-high Yielding Wheat after Rice in Huaibei Region. Journal of Triticeae Crops 22, 51–55 (in Chinese with English abstract). Ziadi, N., Bélanger, G., Claessens, A., Lefebvre, L., Cambouris, A.N., Tremblay, N., Nolin, M.C., Parent -É, L., 2010. Determination of a critical nitrogen dilution curve for spring wheat. Agron. J. 102, 241–250.