| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
a Station de Génétique Végétale, INRA-UPS-INAPG-CNRS, Ferme du Moulon, 91190 Gif/Yvette, France
b Syngenta Seeds, 12 Chemin de L'Hobit, BP 27, 31790 Saint-Sauveur, France
* Corresponding author (gallais{at}moulon.inra.fr).
 |
ABSTRACT |
|---|
 TOPNOTES ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Abbreviations: ASI, anthesis-silking interval • h2, heritability • HI, harvest index • NHI, nitrogen harvest index • NK, kernel nitrogen amount • NNI, nitrogen nutrition index • NUE, nitrogen use efficiency • NUTE, nitrogen utilization efficiency • QTL, quantitative trait locus • r, phenotypic correlation • RIL, recombinant inbred line
a Station de Génétique Végétale, INRA-UPS-INAPG-CNRS, Ferme du Moulon, 91190 Gif/Yvette, France
b Syngenta Seeds, 12 Chemin de L'Hobit, BP 27, 31790 Saint-Sauveur, France
* Corresponding author (gallais{at}moulon.inra.fr).
In maize (Zea mays L.), grain protein yield is the result of two nitrogen fluxes: N remobilization from stover to the kernels and N allocation to kernels from postsilking N uptake. Nitrogen-15 labeling was used to study these two fluxes. Genetic variation for N remobilization and postsilking N uptake was studied in testcrosses derived from a population of recombinant inbred lines. On average, from a 2-yr experiment, 28.3% of whole-plant N was taken up after silking, and 93% of this postsilking N uptake was allocated to kernels. Nitrogen remobilization represented around 61% of total grain N. However, there was greater variation for postsilking N uptake than for N remobilization. Consequently, N grain yield was more highly correlated with the amount of postsilking N uptake than with the amount of N remobilization. The amount of N remobilization was significantly correlated with both the whole-plant N amount present at silking and the proportion of N remobilized, whereas N from N uptake within kernels was only correlated to the postsilking N uptake. Variation for the proportion of postsilking N uptake allocated to kernels was low in comparison to that of postsilking N uptake. There was a negative correlation between the amount of N remobilization and the amount of postsilking N uptake, which appears to have a physiological basis. Finally, use of 15N labeling provided a better description of variation for N accumulation in kernels than the classical balance method.
Abbreviations: ASI, anthesis-silking interval • h2, heritability • HI, harvest index • NHI, nitrogen harvest index • NK, kernel nitrogen amount • NNI, nitrogen nutrition index • NUE, nitrogen use efficiency • NUTE, nitrogen utilization efficiency • QTL, quantitative trait locus • r, phenotypic correlation • RIL, recombinant inbred line
|
INTRODUCTION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
In maize, although varying due to genotype, environmental conditions, and methods of estimation, an average of more than 50% of the grain N originates from the stover (Bertin and Gallais, 2000; Gallais and Coque, 2005). The remaining grain N originates from postsilking N absorption. Remobilization and postsilking N uptake are thus essential components of NUTE. Consequently, to develop varieties with better NUE, it is useful to evaluate the contribution of both N sources (Gallais and Coque, 2005). Unfortunately, the various N fluxes within the plant are difficult to measure. The classic approach for evaluating the contribution of the two sources of grain N is based on the comparison between N quantity in the whole plant at silking and in both grain and stover at maturity (see as examples, Moll et al., 1982; Di Fonzo et al., 1982; Rizzi et al., 1993, 1996; Rajcan and Tollenaar, 1999a,b; Bertin and Gallais, 2000). This approach, called the "balance method," leads to biased results because it neglects the contribution of roots to N remobilization, and it assumes that all postsilking N uptake is allocated to grain. Gallais et al. (2007) have proposed a new method for estimating the proportion of remobilized N by using 15N labeling in the field at the beginning of stem elongation, with a single sampling occurring at maturity. This method is expected to lead to less biased and more accurate estimates than the balance method of the proportion of remobilized N. A second 15N labeling at silking allows estimation of the proportion of postsilking N uptake allocated to the kernels.
Most previous studies that investigated genetic variation of N remobilization and postsilking N uptake on a large number of genotypes (see references in Gallais and Coque, 2005) have used the balance method. This paper presents the results of a study using the 15N method to evaluate the genetic variation for N remobilization and postsilking N uptake in a population of recombinant inbred lines (RILs) evaluated for their testcross performances. Besides the evaluation of genetic variability of the various traits characterizing N remobilization and N uptake, and the validation of our proposed method of 15N labeling, the objectives of the present paper are (i) to study the relationship between N remobilization and postsilking N uptake; (ii) to determine whether N remobilization or postsilking N uptake is more important in improving grain protein yield; and (iii) to identify traits related to N remobilization and postsilking N uptake. A population of RILs was used for this study for the purpose of quantitative trait locus (QTL) detection. Results from line per se evaluation and QTL detection with both per se and testcross populations are not presented here.
|
MATERIALS AND METHODS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Trait Evaluation
Grain Yield and Its Components
At maturity, all plants of each plot were harvested to determine the grain yield. A sample of kernels was taken to determine the thousand kernel weight. Grain yield, thousand kernel weight, and the exact number of plants harvested were used to calculate the average number of kernels per plant.
Traits from the Balance Method
At silking and maturity, six to eight plants were sampled from each plot. For the sampling at silking, the whole plants were chopped and a sample of 700 g of fresh matter was dried at 70°C for three days. At maturity, ears were separated from the stover. Sampling of stover was prepared as for whole-plant sampling at silking. After drying at 70°C, ears were shelled; cobs were not considered. Dried samples were ground, and powders were analyzed for N content. This sampling procedure allowed the determination of (i) at silking, dry-matter yield, whole-plant N content, whole-plant N amount; and (ii) at maturity, harvest index (HI), N content in stover, kernels, and whole plant. Whole-plant dry-matter yield was computed by dividing grain yield by HI. Nitrogen grain yield and whole-plant N yield were then derived. Nitrogen harvest index (NHI) was computed as the ratio of kernel N amount (NK) to total N amount in the aerial part of the plant (Nwhole-plant). At silking, N nutrition index (NNI) was determined according the method of Lemaire and Gastal (1997) as the ratio of the observed N content to a critical N content corresponding to the minimum N content allowing the maximum dry-matter yield. The amount of N remobilization (NKremB), where B = balance method, was derived from the difference between the whole-plant N amount at silking (Nsilk) and the N stover amount at maturity (Nstover). The proportion of N remobilized was then estimated by the ratio tremB:(Nsilk – Nstover)/Nsilk. Postsilking N uptake (Nup) was estimated by the difference Nwhole-plant – Nsilk. Nitrogen utilization efficiency (NUTE) was also computed (i) at the whole-plant level, by the ratio whole-plant yield/whole-plant N yield (which is the reciprocal of whole-plant N content), and (ii) at the grain level, by the ratio grain yield/whole-plant N yield.
Traits from Nitrogen-15 Labeling
Two types of 15N labeling were performed in both years: during vegetative growth and at silking. The methods are described in detail in Gallais et al. (2006, 2007). For labeling during vegetative growth, an average of 1 mg 15N per plant was sprayed around the plant with a solution of KNO3 at 4.91% 15N atom excess. For labeling at silking, 2.5 mg 15N was applied per plant with a solution of KNO3 at 4.91% 15N atom excess. At maturity, six to eight labeled plants were sampled per plot for both types of 15N labeling. For labeling during vegetative growth, labeled plants were also sampled at silking.
After drying, weighing, and grinding samples of each plant part as described above, a subsample of 2.5 mg powder was used to determine total N content and 15N abundance by an elemental analyzer (N analyzer NA1500 Carlo-Erba, Italy) coupled to an isotope ratio mass spectrometer (Optima, Micromass, Manchester) calibrated for measuring 15N natural abundance. The atom percent excess and the total N content per organ (stover and kernels) allows computation of the 15N quantity in each organ. Both the proportion of remobilized N (trem1) from the stover to the kernels (for labeling during vegetative growth) and the proportion of postsilking N uptake (tupK1), which is allocated to the kernels (for labeling at silking), were derived from the ratio of kernel 15N amount to whole-plant 15N amount (Gallais et al., 2007). From these parameters, we derived an estimate of the amount of postsilking N uptake allocated to the kernels NKup1 = Nup tupK1, and an estimate of the N amount translocated from the stover to the kernels NKrem1 = Nsilk trem1.
For vegetative phase labeling, the determination of the proportion of remobilized N assumes that there is no postsilking 15N uptake, or at least that it is less than 15% of the total amount of the 15N that is taken up. With a greater relative postsilking 15N uptake, the estimates of the proportion of N remobilized trem will be significantly biased, and thus a correction is required. If the proportion tupK of postsilking N uptake allocated to kernels can be simultaneously estimated with a labeling just after silking, then an unbiased estimation of trem can be derived (Gallais et al., 2007)
 | [1] |
 | [2a] |
 | [2b] |
The solution of this two-equation system gives unbiased estimates of trem and tupK which can be compared to trem1 and tupK1 to test the consistency of these direct estimates by the 15N method. Determination of trem and tupK according to Eq. [2] involves three estimates of amounts of N (NK, Nsilking, Nstover) and three estimates of 15N amounts in kernels, in stover, and at flowering. Consequently, estimates of trem and tupK for a given genotype are expected to have low accuracy. For such estimates, the ANOVA on the pooled data from the two years of the experiment showed no significant effect due to large experimental error. Therefore, to determine an unbiased average for proportions of N remobilization and postsilking N uptake allocated to kernels, we solved the two-equation system (Eq. [2]) using the means of all genotypes, which were known with high accuracy. Thus, in Eq. [2a], NHI and a means were derived from the means of Nsilk, NK, and Nwhole-plant.
Other Traits Observed
Anthesis-silking interval (ASI) was computed as the difference in days between silking date and anthesis date. Leaf senescence was evaluated at maturity by a visual notation, with a scale between 1 and 5 (1 = green, 5 = completely dry).
Statistical Analyses
To simplify the presentation, we use two-year averages except in instances where single year data will be noted to illustrate specific points. The two-year ANOVA model is the following, for a trait Y, a genotype i, a year j, and a replication k(j):
 |
G2, m is the year effect considered as fixed, (Gm) is the genotype x year interaction with variance Gm2, b is the replication effect nested within the year, and e is the residual error with variance e2. For each trait, heritability is estimated from the components of variance. To express all heritabilities on the same basis, we have considered heritabilities of trait means with three replications even when traits were evaluated with two replications. The phenotypic correlations (r) between traits were computed. We will primarily discuss correlations involving proportion or amount of N remobilization and postsilking N uptake. Correlations between the traits studied and date of silking were low or not significant. Thus, we have not given the partial correlations for a given date of silking since the results were approximately the same as simple phenotypic correlations. Significance of the phenotypic correlations given in the text is shown with one, two, or three asterisks according to the significance levels at 0.05, 0.01, and 0.001, respectively. Genetic correlation was only studied for the relationship between N remobilization and postsilking N uptake estimates because of their dependence at the environmental level. When a confidence interval is given for a mean or a genetic correlation, it is given for the 0.05 probability level.
Approximate confidence intervals for the trem and tupK estimates obtained from the two-equation system were derived graphically after computation of the approximate variance of ratios such as 
/
, (1–)/, 
/(1–
), and /(1–). It was also necessary to derive an approximation of the covariance between the two first and the two last ratios. A bar signifies "mean."
|
RESULTS AND DISCUSSION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
, see Table 2
). Thus, to correct the proportion of remobilized N for each genotype, we have used a constant value of tupK = 0.93. This led to an average proportion of N remobilized of 0.60.
|
|
Heritabilities
As previously shown by Gallais et al. (2007), the coefficient of variation for the proportion of remobilized N estimated by the 15N method (trem1) was lower than that estimated by the balance method (Table 1). This indicates lower environmental variability. Consequently, heritability (h2) for the proportion of N remobilized (trem1) was higher for the 15N method estimate (h2 = 0.52) than for the balance method estimate (h2 = 0.27). The estimates of the amount of remobilized N were also measured with better accuracy by the 15N method than by the balance method. However, because there was lower genetic variance, heritability for this trait was not increased by using the 15N method. The percent of postsilking N uptake allocated to the kernels was also estimated with good accuracy (low cv) for a given year. However, due to genotype x year interaction, its heritability was low. The heritability for the amount of postsilking N uptake accumulated within the kernels (NKup1, NKup3) was higher than for the amount of N remobilized, but not by as much as expected on the basis of genetic variances because there was also greater environmental variance for postsilking N uptake.
Phenotypic Correlations
Correlations between Estimates of Nitrogen Remobilization and Postsilking Nitrogen Uptake by the Balance and Nitrogen-15 Methods
The proportion of N remobilized corrected by the postsilking 15N uptake, and the proportion of postsilking N uptake allocated to kernels (trem2), was strongly correlated with that estimated by a correction, considering variation in postsilking 15N uptake and constancy in the proportion of postsilking N uptake allocated to kernels (r = 0.98). Therefore, for further correlations involving the proportion of N remobilized, we have only considered two parameters, trem1 and trem2, although these two parameters were also highly correlated (r = 0.78). These two parameters were significantly correlated with the proportion of N remobilized estimated by the balance method (tremB), but the phenotypic correlations were only 0.56 and 0.63, respectively (Table 2). As discussed above, this could be due to a greater bias in the balance method than in the 15N method. In contrast, the estimates of the amount of N remobilized from stover to the kernels by the 15N method (NKrem1, NKrem2) were strongly correlated with those estimated by the balance method (NKremB) (r = 0.86 with NKrem1). This is because variation in the amount of N accumulated at silking greater than the variation for the proportion of N remobilized. Indeed, the correlation between the estimates of the amount of N remobilized and the whole-plant N amount at silking was high (r = 0.60 with NKrem2 and 0.76 with NKremB) (Table 3
). However, with the two estimates (NKrem1 and NKrem2) being dependent, we have considered another estimation of the amount of remobilized N derived by subtracting the amount of postsilking N uptake allocated to the kernels from the total N kernel amount. The correlation was still high (r = 0.76).
|
). Such high correlations were due to the low variation in the percent of postsilking N uptake allocated to kernels (tupK) compared with that for the postsilking N uptake. As a consequence, there was no correlation between the proportion of postsilking N uptake allocated to kernels and the amount of postsilking N uptake allocated to kernels. Note that the percent of total N taken up after silking was strongly, positively correlated with the amount of postsilking N uptake (r = 0.92). These two trait estimates are not statistically independent, and the genetic correlation was high (rG = 0.95 ± 0.43). This high correlation means that there was more genetic variation for the amount of postsilking N uptake than for total N uptake. These results for the amount of postsilking N uptake contrast with those for the amount of N remobilized, which was both correlated with the whole-plant N amount at silking and with the proportion of N remobilized (Table 2).
Correlations between Nitrogen Remobilization and Postsilking Nitrogen Uptake
The correlations between the proportion of N remobilized (trem1 or tremB) and the proportion of postsilking 15N uptake allocated to kernels (tupK1) were positive (0.46 and 0.52, respectively) (Table 2). Such correlations between trem and tupK were expected because both of these estimated proportions involve NHI (Gallais et al., 2007). Indeed because the correlations between these two proportions (trem2 and tupK1) and NHI were high (0.69 and 0.67, respectively) (Table 3). However, the correlation between trem and tupK could also mean that the relative sink size measured by the NHI is a limiting factor for both proportions.
By all estimates, the amount of N remobilized and the amount of postsilking N uptake allocated to the kernels (NKup1 and NKup3) were negatively correlated (Table 2). Such a negative correlation could be a statistical artifact or could have a physiological cause. A statistical artifact could be caused by the two estimates not being statistically independent. Indeed, with the balance method, NKup = Nup = Nwhole-plant – Nsilk, and NKrem = Nsilk – Nstover, with also Nup = NK – NKrem; thus, NKup and NKrem estimates are not statistically independent. With the 15N method, NKrem = Nsilktrem and NKup = NuptupK: both estimates involve the estimate of Nsilk. Thus, again, these two estimates are not statistically independent. However, this dependence is lower than with the balance method. Consequently, the correlation between NKrem1 and NKup1 was lower (–0.39) than the correlation between Nup and NKremB (–0.72). A way to suppress the effect of this statistical dependence is to compute the genetic correlation. It was –0.87 ± 0.06 in 2003 and –0.51 ± 0.28 in 2004, whereas the environmental correlations were –0.76 and –0.88, respectively. Therefore, we can conclude that the statistical dependence is not sufficient to explain the observed negative correlation between the estimated amounts of postsilking N uptake and N remobilization.
One possible physiological cause of this negative relationship is a negative relationship between N amount at silking and postsilking N uptake (r = –0.38). The amount of N remobilization was highly correlated with the amount of N accumulated at silking (see Table 3). This could mean that there was a limit in N uptake at the level of the whole plant due to a limit in the soil N availability. If the uptake is low before silking, it could be higher after silking because greater residual N availability. Second, the negative correlation could be due to a mechanical effect of the limit in the sink size. If the total N amount in the kernels is less variable than NKrem or NKup, we expect a negative correlation between NKrem and NKup. The potential of N uptake at the level of the whole plant is highly determined by the sink size (r between protein kernel yield and whole-plant protein yield = 0.80) (Table 3). Indeed, the genetic variance in the kernel protein amount was not significantly different from the genetic variance in the amount of postsilking N uptake (data not shown). Third, the negative correlation between N remobilization and N uptake could result from a relationship between photosynthesis and N uptake. When the growth conditions are favorable for postsilking photosynthesis, they favor N uptake, and thus, N remobilization would be reduced. Reciprocally, when the growth conditions are unfavorable to photosynthesis (like drought stress), they are unfavorable to postsilking N uptake, and thus N remobilization plays a more important role with a degradation of photosynthetic proteins like rubisco, leading to earlier senescence (Triboi and Triboi-Blondel, 2002; Paponov and Engels, 2003). Furthermore, if photosynthesis is insufficient, translocation of carbohydrates toward roots will be insufficient to maintain activity of the roots which contributes to a decrease in N uptake and an increase in N remobilization (Tolley-Henry and Raper, 1991).
Correlations between Nitrogen Grain Yield, Grain Yield, and Their Components
One way to consider N grain yield is as the product of grain yield and grain N content. Nitrogen grain yield was highly correlated with grain yield (r = 0.82), whereas it was nonsignificantly correlated with kernel N content (Table 3). Both yield components, kernel number and kernel weight, were equally related to N grain yield. Second, N grain yield can be considered as the sum of N from N remobilization and postsilking N uptake. Variation in N remobilization appeared to have a low effect on N grain yield variation (r = nonsignificant), whereas variation in the amount of postsilking N uptake appeared to have a high effect (r = 0.64 between N grain yield and N uptake). The negative correlation between N grain yield and senescence at maturity (r = –0.33) can be considered as a negative effect of senescence on N uptake. Stover N yield at maturity was obviously negatively correlated with HI and NHI. It was more negatively correlated with senescence (r = –0.42) than grain yield and also positively correlated with percent of postsilking N uptake (r = 0.44). The whole-plant N yield was about equally determined by N grain yield and N stover yield. It was highly correlated with postsilking N uptake (r = 0.75), and negatively correlated with the proportion of N remobilized (tremB), but not with the amount of N remobilized.
Correlations between Nitrogen Remobilization or Postsilking Nitrogen Uptake and Agronomic Traits, Including Nitrogen Use Efficiency Traits
N remobilization traits (NKrem or trem) were not related to grain yield, whereas amount of postsilking N uptake allocated to kernels (tupK1) was highly correlated with grain yield (r = 0.52)(Table 3). Nitrogen grain yield was significantly correlated with N remobilization amount NKrem2 (r = 0.26), but it was more significantly correlated with the amount of postsilking N uptake allocated to kernels NKup1 (r = 0.61) and NKup3 (r = 0.64). This again means a greater contribution to the variation of total N amount in the kernels of N amount from postsilking N uptake than N remobilization amount. Furthermore, estimates of the amount of postsilking N uptake allocated to kernels NKup1 and NKup3 were positively correlated with kernel number (r 
0.40) but less correlated with thousand kernel weight (r 0.22). Both proportions (trem and tupK) were correlated with grain NUTE (more significantly for the proportion of N remobilized, r = 0.47 between NUTE and trem balance), and highly correlated with NHI (r 0.70). This is an illustration of the relationship between N uptake and sink size. Amounts of N remobilized and N absorbed after silking were also correlated with NHI. Increasing NHI could thus be a way to increase simultaneously N remobilization and postsilking N uptake, which are antagonistic. Kernel N content was not related to N remobilization and postsilking N uptake amounts, but was positively correlated with proportion of postsilking N uptake allocated to kernels tupK1 (r = 0.26). As expected, stover N content was negatively correlated with N remobilization (amount and proportion, r –0.65) and with the proportion of postsilking N uptake allocated to kernels tupK1 (r = –0.56).
Even though postsilking N uptake allocated to kernels represented less than 40% of total kernel N, its genetic variation (i.e., variation in postsilking N uptake) explained most of the genetic variation in N grain yield. Considering that there is generally a high correlation between N grain yield and grain yield, the results from Rizzi et al. (1993), which show a high correlation between grain yield and postsilking N uptake, support our conclusion. The major role of postsilking N uptake also appears in the studies of Tollenaar (1991), Rajcan and Tollenaar (1999a,b), and Coque and Gallais (unpublished results), where genetic advance in grain yield was shown to be highly related to genetic advance in postsilking N uptake. Greater N uptake during the phase just after ovule fertilization in which embryos can abort could explain the relationship between postsilking N uptake and kernel number. Indeed, just after ovule fertilization, embryos need carbon and N assimilates (Below et al., 2000). The kernel number could be determined by the plant growth rate, and then N uptake, during this period (Andrade et al., 2002).
Correlations with Nitrogen Traits at Silking, Anthesis-Silking Interval, and Senescence
The NNI was highly correlated with N amount and whole-plant N content at silking (r = 0.88 and 0.82, respectively) (data not shown). Consequently, both whole-plant N yield at silking and NNI were similarly, positively, correlated with the amount of N remobilization (r = 0.76 and 0.60, respectively, with NKremB) but negatively correlated with the amount of postsilking N uptake allocated to kernels NKup1, as a result of the antagonism between N remobilization and N uptake (r = –0.38 and –0.35, respectively) (Table 3). Nitrogen content at silking also has a favorable effect on the amount of N remobilization and an unfavorable effect on the amount of postsilking N uptake.
Anthesis-silking interval was negatively correlated with N remobilization (proportion and amount), in particular with NKrem2 (r = –0.38), a short interval corresponding to a high N remobilization. The correlation between grain yield and ASI has already been shown, mainly at low N input (Bänziger and Lafitte, 1997; Bertin and Gallais, 2000; Zaidi et al., 2003). Anthesis-silking interval appears to be an indicator of tolerance to stress as shown by the results of Tollenaar (1991) and Bänziger et al. (2002). Genotypes for which ASI does not increase under N or drought stress could have better C and N metabolism, leading to a greater grain yield or grain N yield, mainly at low N input, as a result of better C and N partitioning, just after fertilization, avoiding embryo abortion and ear abortion (Gallais and Coque, 2005). The observed negative correlation between ASI and NNI (r = –0.25), already observed by Bertin and Gallais (2000), supports the physiological basis of ASI.
Senescence was expected to affect both N remobilization and N uptake. However, the correlations were low (Table 3). Senescence was not correlated with the amount of N remobilized (NKrem) but was correlated with the proportion of N remobilized (trem) and with the postsilking N uptake allocated to kernels NKup3, and with the postsilking N uptake, Nup. As a consequence, senescence negatively affected N grain yield (r = –0.33). In Canadian studies (Tollenaar, 1991; Ma and Dwyer, 1998; Rajcan and Tollenaar, 1999a,b) the effect of leaf area duration on postsilking N uptake was clearer. It could also be more important in stress conditions, particularly at low N input (Bänziger and Lafitte, 1997; Bänziger et al., 1997; Presterl et al., 1996, 2002; Bertin and Gallais, 2000; Ding et al., 2005; Coque and Gallais, unpublished results). In our study, the low correlation between senescence and N remobilization and postsilking N uptake could be due to the difficulty in evaluating senescence by a visual notation. The use of a chlorophyll meter at several stages after silking could increase the efficiency of the evaluation.
Advantage of Nitrogen-15 Measurements
A low but significant correlation was observed between the N remobilization amount estimated by the 15N method and N grain yield, whereas when the N remobilization amount was estimated by the balance method, the correlation was not significant. The highly negative correlation between the amount of postsilking N uptake and N remobilization estimated by the balance method (r = –0.78 with tremB, r = –0.72 with NKremB) became less significant with the 15N method (not significant for the proportion and r = –0.46 for the amount) (Table 2). Consequently, for the amount and proportion of N remobilized, the 15N method decreased the correlation with traits involving N uptake: grain NUTE, whole-plant N yield, and stover N yield. Similarly, correlation with percent postsilking N uptake was strongly reduced, mainly for the percent of N remobilized, because the 15N method gives an estimate that is more statistically independent of postsilking N uptake than the balance method. It is also probably closer to the true value. Some specific correlations or higher correlations appeared with the 15N method. In particular, thousand kernel weight and grain N yield were correlated with the N remobilization amount when estimated by the 15N method (NKrem2) and not when estimated by the balance method.
Because there was a strong correlation between postsilking N uptake and the 15N estimate of amount of postsilking N uptake allocated to kernels (r = 0.94), there were few changes in correlations for N uptake. This is a result of low variation in the proportion of postsilking N uptake allocated to kernels that appeared to be effected by genotype x year interaction. However, this method must underestimate the proportion of postsilking N uptake allocated to kernels because the entire phase of grain filling was not labeled. This favors genotype x year interaction for this trait because 15N fertilizer was probably taken up in a period after silking that differed in duration in the two years. As there was very little genetic variation for the estimated proportion of postsilking N uptake allocated to kernels, but there was variation in postsilking N uptake, this could indicate genetic variation for N uptake in the last part of the grain-filling period. Indeed, results from the study of genetic variation at the end of the grain filling period support this assumption (Coque and Gallais, unpublished results, 2006).
On the whole, studies of correlations with different estimates of N remobilization and postsilking N uptake allocated to kernels showed the relevance of the 15N estimates despite assumptions that a priori could appear restrictive (Gallais et al., 2006, 2007). The main advantage of measurements using the 15N method alone (proportions of N remobilized and of postsilking N allocated to kernels) is that they can be determined by sampling only at maturity, whereas, the balance method also requires sampling at silking. Furthermore, they are measured with high accuracy. The main challenge is to develop a 15N-labeling protocol that restricts bias in the estimates. The protocol used for labeling just after silking did not allow consistent estimates of the proportion of postsilking N uptake allocated to the kernels. More intense labeling, with irrigation, would be necessary to label during the entire phase of grain filling. In contrast, 15N labeling during the vegetative phase provided estimates of the proportion of remobilized N with a clearer physiological basis than did the estimates from the balance method. As previously noted by Gallais et al. (2007), the main problem with this type of labeling is in limiting the amount of postsilking 15N uptake. It appears that this can be accomplished by an early spray of 15N fertilizer. However, our study shows that even with 18% of postsilking 15N uptake, the biased estimates of the proportion of N remobilized are highly correlated with corrected estimates. Thus, to study a large number of genotypes in the field, 15N labeling during the vegetative phase appears to be an easier tool to use than the balance method for the estimate of the proportion of N remobilized. Furthermore, with an accurate estimate of the proportion of N remobilized by vegetative 15N labeling (with low postsilking 15N uptake) and sampling at silking, it is possible to derive an estimate of the amount of N remobilized from stover to kernels that is more pertinent (less biased) and more accurate than by the balance method. Thus, from this estimated amount and from the total kernel N amount, it will be possible to derive the amount and proportion of postsilking N uptake allocated to kernels with good accuracy.
|
CONCLUSIONS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
NOTES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Received for publication February 21, 2007.
|
REFERENCES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
