Growth analysis of winter wheat cultivars as affected by nitrogen fertilization / Wachstumsanalyse von Winterweizensorten in Abhängigkeit von Stickstoffdüngung

The effect of nitrogen fertilization on the yield and yield components of various wheat cultivars was studied in a small-plot long-term experiment, with two factors arranged in a split-plot design in four replications. The experiment was carried out in the years 2006/2007, 2007/2008 and 2008/2009 at the Agricultural Institute of the Centre for Agricultural Research in Martonvásár. In the long-term crop rotation experiment, the crop sequence was pea, winter wheat, maize and spring barley. The dose of N fertilizer formed the main plot and wheat cultivar the subplots. The doses of N fertilizer (calcium ammonium nitrate) were 0, 80, 160 and 240 kg ha^-1 (designated as N₀, N₈₀, N₁₆₀ and N₂₄₀, respectively) and were applied in two splits: one-third before sowing and the other two-thirds in early spring at tillering. All the plots were given the same dosage of phosphorus and potassium (120 kg ha^-1 of each). The three Martonvásár wheat genotypes sown in the subplots were Mv Toborzó (extra early), Mv Palotás (early) and Mv Verbunkos (mid-early). The ploughed layer of the chernozem soil, a humus-containing loam, was slightly acidic with moderate supplies of phosphorus and good supplies of potassium.

In the dry year of 2007, the total rainfall during the growing season was only one-third (200 mm) of that in 2008 and 2009 (638 and 617 mm, respectively). The rainfall distribution was also unfavorable in 2007, while in 2008 and 2009 both the quantity and distribution of rainfall were satisfactory (with the exception of lack of rain in April 2009). The mean temperature during the growing season was higher in 2007 (12°C) than in the other two years (10°C), which could be attributed partly to the very mild winter.

The sampling area for each treatment was 13.5 m² (9 m × 1.5 m). At each sampling date, destructive samples consisting of 5 plants were taken randomly from a 0.5 m² area once a week on a total of 25 occasions in 2007, 21 in 2008 and 17 in 2009, covering the whole growing season. Sampling was begun when the wheat reached the two-leaf stage. Leaf area was estimated by measuring the green leaf area of all the leaves with a leaf area meter (Model AM 300, BioScientific Ltd, UK). The dry mass of leaves, stems and spikes was determined after drying in a drying cabinet at 60°C for 48 h. The harvest index was derived from a 0.18 m² subplot. The plants were cut at the soil surface, bundles were weighed and threshed, and grain weights were recorded.

The Hunt-Parsons program (HP curves) (Hunt and Parsons, 1974), which fits first-, second- or third-order polynomial exponential curves to the trends in lnY (dry weight) versus t (time) and lnZ (leaf area) versus t, was used for functional growth analysis. A polynomial exponential function is a polynomial function of the natural logarithm of a growth attribute in relation to time (Causton and Venus, 1981). The output consisted of observed and fitted values of lnY and lnZ and the values of dY/dt, dZ/dt, (1/Y) (dY/dt), (1/Z)(dZ/dt), Z/Y and (1/Z)(dY/dt), together with their standard errors and 95% confidence intervals. The absolute growth rate (AGR), absolute growth rate of leaf area (ALGR), relative growth rate (RGR), net assimilation rate (NAR), leaf area ratio (LAR), crop growth rate (CGR) and leaf area index (LAI) were calculated using the Hunt-Parsons program, while the method of classical growth analysis (Evans, 1972; Hunt, 1982) was used to calculate the harvest index (HI), leaf area duration (LAD), leaf area duration of the flag-leaf (LAD_flag-leaf) and biomass duration (BMD). The growth analysis indices (parameters) were characterized in terms of dynamics over time and average (mean) and maximum (max) values (Causton and Venus, 1981; Hunt, 1982).

The split-plot design from the General Analysis of Variance menu of the GenStat 18 program was applied to analyze the growth parameter data sets, while the relationships between growth parameters were studied by linear regression analysis. Multiple linear regression analysis was used to determine relationships between the yield per plant (g plant^-1) and yield per unit area (t ha^-1) (as dependent variables), and the yield components and growth indices (as independent variables) for all the data (n = 36). The individual and joint effects of independent variables on the yield were determined using the All Subsets Regression menu of multiple regression. Relationships were analyzed between the yield per plant and the following eight independent variables: grain number (GN) per spike, thousand kernel weight (TKW), RGR_mean, AGR_mean, ALGR_mean, NAR_mean, LAR_mean and LAD_{flag_leaf}. The relationships between yield per unit area (t ha^-1) and the following seven variables were analyzed: GN per m², TKW, CGR_mean, LAI_max, LAD_LAI, HI and BMD. The following indices: R² (adjusted multiple correlation coefficient), R² (multiple correlation coefficient), Mallows C_p and AIC (Akaike’s information criterion) were used as criteria in selecting the variables (Afifi et al., 2004).

The dynamics of dry matter accumulation per plant over time was expressed by a third-degree exponential function (Figure 1), the only exceptions being the dry matter accumulation of Mv Toborzó and Mv Verbunkos in the N₀ treatment in 2007, to which a quadratic exponential function was fitted. In all cases, the functions gave a good fit to the measurement data (R² = 94.7–99.3%). The dynamics of dry matter accumulation gave a good reflection of the effect of nitrogen treatments. In response to N fertilizer, the dry matter production increased up to the N₂₄₀ treatment in 2007 and 2008, and up to N₁₆₀ in 2009, the greatest differences generally being observed between the N₀ and N₈₀ treatments. Averaged over the cultivars and years, the maximum values were as follows: N₀: 3.03, N₈₀: 4.01, N₁₆₀: 4.38, N₂₄₀: 4.37 g plant^-1.

In all cases, the dynamics of leaf area growth was depicted with a third-degree exponential function (Figure 2) (R² = 83.6–97.0%). The dynamics in the N₀ and N₈₀ treatments was quite distinct from that in the N₁₆₀ and N₂₄₀ treatments. The maximum value of leaf area per plant was smallest in the N₀ treatment and significantly greater in the N₈₀ treatment, while the highest values were obtained in the N₁₆₀ and N₂₄₀ treatments. Averaged over years and cultivars, the maximum leaf area in the N treatments was as follows (cm² plant^-1): N₀: 84.8, N₈₀: 134.9, N₁₆₀: 160.7, N₂₄₀: 169.5. The maximum leaf area was achieved by the plants immediately before heading. The dynamics of leaf area gave a clear indication of the different maturity dates of the cultivars. As a function of year and N treatment, the maximum leaf area was recorded 173–187 days after sowing (DAS) for Mv Toborzó, 180–194 DAS for Mv Palotás and 187–194 DAS for Mv Verbunkos. Averaged over years and N treatments, the leaf area of Mv Verbunkos was the greatest (144 cm²), followed by Mv Toborzó (135 cm²) and Mv Palotás (134 cm²). The maximum leaf area per plant (averaged over cultivars and N treatments) was considerably lower in 2009 (115 cm²) than in the other two years (2007: 135 cm², 2008: 155 cm²).

Absolute growth rate (AGR), the rate of change in size per unit time, is the simplest index of growth. Figure 3 shows the dynamics of AGR of Mv Palotás in 2008. The dynamics of growth parameters were similar in each year. Tables 1-3 give the detail data of the parameters. The dynamics of AGR could be characterized by a bell-shaped curve. The maximum values were obtained in the period around flowering (ca. 203 days after sowing). In the treatment where no N fertilizer was given, the dynamics of AGR was clearly distinct from that of the fertilized treatments. The value of AGR_mean (g day^-1 10^-2) rose to the N₁₆₀ treatment (Table 1), with the following values in the individual treatments: N₀: 2.26, N₈₀: 3.17, N₁₆₀: 3.54, N₂₄₀: 3.51. There was little difference in the AGR_mean values of the different wheat cultivars: Mv Verbunkos and Mv Palotás: 3.13, Mv Toborzó: 3.09. The mean value of AGR was higher in 2008 and 2009 (3.16 and 3.55, respectively) than in 2007 (2.64).

Table 1

Effect of N fertilization on the mean values of the growth parameters and the maximum leaf area index (LAI) of wheat cultivars, using the functional method of growth analysis (2007-2009)

Tabelle 1. Einfluss der N-Düngung auf die Mittelwerte der Wachstumsparameter und den maximalen Blattflächenindex (LAI) der Weizensorten nach der funktionellen Methode der Wachstumsanalyse (2007-2009)

Relative growth rate (RGR) expresses growth in terms of the rate of increase in size per unit of size. After the plants reached the 4-leaf stage (ca. 110 days after sowing), the dynamics of RGR exhibited an initial, relatively rapid increase. The maximum value was reached between mid-March and mid-April (around 154 days), at the beginning of shooting, after which it gradually declined, dropping to 0 when the foliage withered completely in mid-June (236 days) (Figure 3). The value of RGR_mean (g g^-1 day^-1 10^-2) was the greatest in the N₁₆₀ treatment (Table 1), with the following values per N treatment: N₀: 2.78, N₈₀: 3.07, N₁₆₀: 3.25, N₂₄₀: 3.12. There was no significant difference between the RGR values of the cultivars: Mv Toborzó: 2.97, Mv Palotás: 3.08, Mv Verbunkos: 3.01. In the wetter years of 2008 and 2009, the RGR_mean value (g g^-1 day^-1 10^-2) was considerably higher (3.31 and 3.18, respectively) than in 2007 (2.67), when the rainfall supplies were less favorable.

The absolute growth rate of the leaf area (ALGR) was characterized by two successive bell-shaped curves (Figure 3), the first describing the growth of leaf area and the second describing the dynamics of leaf withering. The maximum value of ALGR was achieved at the end of tillering (174 days after sowing), immediately prior to shooting, with a difference of around a week between the cultivars due to differences in their maturity dates. The mean values of ALGR (cm² day^-1 10^-2) in each N treatment were as follows: N₀: 0.95, N₈₀: 1.67, N₁₆₀: 2.01, N₂₄₀: 2.15 (Table 1). Among the cultivars Mv Verbunkos exhibited a higher value of ALGR_mean (1.79) than Mv Toborzó (1.60) or Mv Palotás (1.69). The mean value of ALGR (cm² day^-1 10^-2) was lowest in 2007 (1.34) and substantially higher in 2008 (1.94) and 2009 (1.81).

Net assimilation rate (NAR) is an index of the productive efficiency of plants, calculated in relation to total leaf area. Starting from the early growth stage (111 days after sowing), NAR increased rapidly for a few weeks, until the side-tillers had developed (Figure 3), after which the rate slowed and remained more or less constant until the foliage was fully developed. Then, as the leaf area decreased (about 174 days), NAR accelerated up to the end of the vegetation period. The mean value of NAR (Table 1) was smallest in the N₀ treatment (2.79), rising with the application of N fertilizer and reaching the highest value in the N₈₀ (2.82) or N₁₆₀ treatment (2.84), depending on the cultivar and year. The NAR values of Mv Palotás (3.05) and Mv Toborzó (2.71) exceeded that of Mv Verbunkos (2.58). The mean value of NAR was highest in 2009 (3.63), while there was little difference between the values recorded in 2007 (2.27) and 2008 (2.45).

Leaf area ratio (LAR) is a morphological index expressing the leafiness of the plant as the ratio between total leaf area per plant and total dry weight per plant. The LAR values reached a maximum at the end of tillering (160-168 days after sowing), after which they declined steeply until flowering, followed by a further, slower decrease (Figure 3). The mean values of LAR provided a good illustration of the effects of N treatments (Table 1), and exhibited the following values in the individual N treatments: N₀: 82.8, N₈₀: 91.4, N₁₆₀: 95.8, N₂₄₀: 94.3 cm² g^-1. Among the wheat cultivars, the LAR_mean values of Mv Toborzó (93.5) and Mv Verbunkos (94.3) were higher than that of Mv Palotás (85.4) . The highest value of LAR_mean was recorded in 2008 (101.4), while the values in 2007 and 2009 were similar (89.2 and 82.6, respectively).

Crop growth rate (CGR) is an index of agricultural productivity of land in terms of the plant biomass produced per unit area. The dynamics of CGR was similar to that of AGR, being characterized by a bell-shaped curve with maximum values during the flowering period. The mean value of CGR (CGR_mean: g m^-2 day^-1) was lowest in the N₀ treatment, increasing significantly in the N₈₀ treatment (15.4) and achieving the highest value in the N₁₆₀ treatment (17.2), after which it dropped slightly (N₂₄₀: 16.3) (Table 1). Among the wheat cultivars, Mv Toborzó had the highest CGR_mean value (15.3), followed by Mv Verbunkos (14.9) and Mv Palotás (14.4). CGR_mean exhibited the lowest value in 2007 (13.9), with higher values in 2008 (15.6) and 2009 (15.1).

Leaf area index (LAI) is the ratio between the total leaf area of the crop and the total ground area on which it stands. The dynamics of LAI in response to N fertilization was similar to that of the leaf area (Figure 3). The maximum LAI values (Table 1) clearly reflected the effect of N fertilization (m² m^-2): N₀: 4.05, N₈₀: 6.73, N₁₆₀: 8.13, N₂₄₀: 8.06. Averaged over N treatments and years, the LAI_max values were lower for Mv Palotás (6.24) than for Mv Toborzó (6.96) and Mv Verbunkos (7.02). LAI_max was highest in 2008 (7.83), somewhat lower in 2007 (7.14) and much lower in 2009 (5.25).

Leaf area duration (LAD) is a quantitative expression of the length of time over which the plant stand maintains an active photosynthesizing leaf area. Both N fertilization and cultivar had a highly significant effect on the value of LAD_LAI in all three years, and there was also a significant N fertilizer × cultivar interaction (Table 2). The value of LAD_LAI (days) was lowest in the N₀ treatment (248), significantly higher in N₈₀ (388) and N₁₆₀ (448), and the highest in N₂₄₀ (455). Averaged over the years and N treatments, the greatest value of LAD_LAI was obtained for Mv Toborzó (402), followed by Mv Verbunkos (296) and Mv Palotás (356); though in the favorable years of 2008 and 2009, Mv Verbunkos had the highest values. In the individual years, the LAD_LAI value was significantly higher in 2007 (442) than in 2008 (396) or 2009 (317).

The leaf area duration of the flag-leaf (LAD_{flag_leaf}) differed in terms of both N treatments and cultivars (Table 2). The lowest values were recorded in the N₀ treatment, rising with increases in N rate. The highest LAD_{flag_leaf} values were found in the N₁₆₀ treatment in 2007 and 2008 (423 and 864 cm² day, respectively) and in the N₂₄₀ treatment in 2009 (664 cm² day). In 2008 and 2009, the LAD_{flag_leaf} values of Mv Verbunkos exceeded those of the other two cultivars. In terms of the years, the highest LAD_{flag_leaf} value (cm² day) was recorded in 2008 (719), with a significantly lower value in 2007 (581) and the lowest in 2009 (544).

The biomass duration (BMD) takes into account not only how much dry weight develops, but also how long it lasts. The effects of both N fertilizer and cultivar on BMD were significant in all the years, and there was also a significant N fertilizer × cultivar interaction (Table 2). The value of BMD (g day) was lowest in the N₀ treatment (156), rising significantly with increasing N rates to 201.9 in N₈₀, 220 in N₁₆₀ and 227 in N₂₄₀ (Table 2). Averaged over the N treatments, Mv Toborzó had the highest BMD in 2007 and 2009 (252 and 185 g day, respectively), while in the favorable year of 2008, the highest value was recorded for Mv Verbunkos (196 g day). The value of BMD (g day) was the highest in the dry year of 2007 (248), with significantly lower values in 2008 (191) and 2009 (177).

Significant linear regression was found between leaf area duration (LAD) and biomass duration (BMD) based on the data of three years (Y = 75.9 + 0.324BMD). The R² value showed that LAD explained 75.9% of the variance in BMD. On the basis of the 3-year data, linear regression was significant at P < 0.1% level between the absolute leaf area growth rate (ALGR_max) and the maximum value of the leaf area index (LAI_max) (Y = 1.97 + 1.56ALGR_max, R² = 79.6%). In each year, significant linear regression was detected between the mean absolute growth rate of dry matter (AGR_mean) and the biomass duration (BMD). Based on R² AGR_mean accounted for 75% of the variance in BMD in 2007, for 95.7% in 2008 and for 95.3% in 2009. Based on the three-year data (n = 36), there was a significant relationship between RGR_mean and its two components, NAR_mean and LAR_mean. The two components explained 62.7% of the variance in RGR_mean at P < 0.1% level. The two parameters had similar effects on the RGR_mean. In all three years and averaged over three years, significant linear regression (P < 0.1%) was found between CGR_max and its components, NAR_mean and LAI_max, which together determined 71.2% of the variance in CGR_max. In all three years, the effect of LAI_max was decisive, being more than three times as great as that of NAR_mean.

Relationships were investigated between the yield per plant (g plant^-1), as a dependent variable, and eight independent variables (Table 3). In decreasing order of R², the individual variables having the greatest significant effect on the yield per plant were GN per spike, RGR_mean, AGR_mean, ALGR_mean and LAD_{flag_leaf}. The two independent variables that had the greatest joint influence on the yield were GN spike^-1 and TKW, with the regression equation: Y = −1.133 + 0.04386GN + 0.02486TKW. This was followed (in decreasing order of R²) by the GN spike^-1 and RGR_mean and the GN per spike and AGR_mean. The three independent variables with the greatest joint effect on the yield per plant were the GN spike^-1, the TKW and RGR_mean, with the regression equation: Y = −1.494 + 0.03971GN + 0.02381TKW + 0.1709RGR_mean. The four independent variables with the greatest combined effect on yield per plant were GN spike^-1, TKW, RGR_mean and LAD_{flag_leaf}. In this case, the regression equation was: Y = −1.029 + 0.04299GN + 0.02438TKW + 0.0896ALGR_mean + 0.000341LAD_{flag_leaf}.

The influence of seven independent variables was examined on the crop yield (t ha^-1) (Table 4). The independent variables that individually had a separate significant influence on the crop yield (in decreasing order of R²) were GN m^-2, CGR_mean, LAI_max, HI, TKW and LAD_LAI. The two independent variables having the greatest effect on the yield in combination were GN m^-2 and LAD_LAI. The regression equation was: Y = 4.072 + 0.00009843 GN m^-2 + 0.002947 LAD_LAI. This was followed (in decreasing order of R²) by GN m^-2 and LAI_max, GN m^-2 and CGR, GN m^-2 and BMD and GN m^-2 and HI. The three independent variables having the greatest combined influence on the yield (t ha^-1) were the GN m^-2, LAI_max and HI, with the regression equation: Y = 1.08 + 0.0000802 GN m^-2 + 0.1398LAI_max + 0.0825HI.

Table 4

Variables significantly influencing crop yield (t ha⁻¹) alone or in combination, based on the stepwise method of multiple regression analysis (n = 36)

Tabelle 4. Variablen, die den Ertrag des Pflanzenstandes (t ha⁻¹) signifikant beeinflussen, allein oder in Kombinationen, nach der schrittweisen Methode der Mehrfach-Regressionsanalyse (n = 36)