Crop Science Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published in Crop Sci 39:1385-1393 (1999)
© 1999 Crop Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (24)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Jeuffroy, M.-H.
Right arrow Articles by Bouchard, C.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Jeuffroy, M.-H.
Right arrow Articles by Bouchard, C.
Agricola
Right arrow Articles by Jeuffroy, M.-H.
Right arrow Articles by Bouchard, C.
Crop Science 39:1385-1393 (1999)
© 1999 Crop Science Society of America

CROP ECOLOGY, PRODUCTION & MANAGEMENT

Intensity and Duration of Nitrogen Deficiency on Wheat Grain Number

Marie-Hélène Jeuffroya and Christine Boucharda

a Unité d'Agronomie INRA - INAPG, BP01, 78 850 Thiverval-Grignon, France. Sté Chimique La Grande Paroisse S.A. and I.N.R.A

jeuffroy{at}bcgn.grignon.inra.fr


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 Materials and methods
 Results
 Discussion
 Conclusion
 REFERENCES
 
In humid temperate climates, N is still a major limiting factor for wheat (Triticum aestivum L.) production. Our objective was to understand and quantify N deficiency effects on the crop grain number to develop optimum N fertilizer management strategies for wheat. In this aim, several experiments were conducted on various soil types and climates with `Soissons' winter wheat. Rates and dates of N fertilizer application were varied in each experiment. This resulted in highly variable dynamics of N accumulation in plants, leading to various N deficiencies throughout the crop cycle. Deficiencies were characterized by a N nutrition index (NNI). Seven criteria describing the deficiency (beginning of deficiency, BD; end of deficiency, ED; duration of deficiency, DD; intensity of deficiency, ID; the product ID x DD = IDD; the lack of nitrogen accumulation at anthesis, LNA; and the NNI at anthesis, NNIa) were estimated for each treatment. Large ranges were obtained for each criterion. Treatments also resulted in highly variable grain numbers. For a N deficient crop, the grain number decrease relative to the control treatment in the same experiment (RGN) was analyzed according to the deficiency criteria. Whatever the grain number component affected (spike number per per square meter or grain number per spike), the RGN appeared to depend on the history of the deficiency, the main explicative variable being IDD, that is, the product of the duration and the intensity of the deficiency. The equation RGN = 1.00355–0.00110 x IDD (R2 = 0.929) allows the prediction of grain number for wheat crops subjected to various N fertilization strategies.

Abbreviations: GN, grain number per square meter • NNI, N nutrition index • BD, beginning of deficiency • ED, end of deficiency • DD, duration of deficiency • ID, intensity of deficiency • NNIa, N nutrition index at anthesis • LNA, lack of N accumulation at anthesis • RGN, relative grain number • IDD = DD x ID, integrated N nutrition index • Nc, critical N concentration • Nm, measured N concentration


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
Materials and methods
 Results
 Discussion
 Conclusion
 REFERENCES
 
WHEN WHEAT IS GROWN in temperate climates, there is a difference between the kinetics of the crop's N requirements and mineralization by microorganisms which release N into the soil (Bergström et al., 1991; Groot and Verberne, 1991; van Keulen and Stol, 1991; Recous et al., 1997). For the last 30 yr, the aim of N fertilization has been thought to be to supply sufficient N to ensure maximal growth, preventing deficiency throughout the crop cycle. However, this aim is unattainable in many fields. Constraints of work organization on commercial farms can prevent N from being applied on optimal dates (Meynard, 1986; Aubry et al., 1998). Environmental constraints may also lead to reduced N applications and prevent insurance fertilization, so that the amount of unused N in the soil which may be potentially leached is reduced (MacDonald et al., 1989). In organic agriculture, the cost of organic N fertilizer is too high to ensure N conditions are not limiting for optimal growth throughout the cycle (David, 1997). Therefore, in humid temperate climates, N is still a major limiting factor for wheat (Boiffin et al., 1981; Frederick and Marshall, 1985; Mascianica and Walden, 1986).

We need to understand the effects of N deficiency on yield according to the crop stage at which it occurs and its duration and intensity, to optimize N fertilization under these constraints. In particular, it is essential to determine whether all N deficiencies are detrimental to yield, or whether some deficiencies can be tolerated because they have only minor effects on yield. It is already known that in various environmental conditions, including those in which N levels vary (Boiffin et al., 1981; Abbate et al., 1995), wheat yield is largely determined by the grain number per square meter (GN) of the crop. However, few studies have quantified differences in GN in response to N nutrition, or in relation to the intensity and duration of the N deficiencies.

Little is known about the effects of N nutrition conditions on wheat GN. Abbate et al. (1995) have shown that if N nutrition is limiting and continuous, i.e., the wheat crop is subjected to N deficiency from early in the crop cycle until after anthesis, GN is strongly and linearly correlated with the amount of N in the spikes at anthesis. However, the effects of fluctuations in N nutrition on GN, such as temporary N deficiency, are unknown. In particular, the effects may vary according to the stage of occurrence of the deficiency, its duration, its intensity, and whether there is a break in the deficiency before anthesis. Several authors have analyzed differences in GN components according to the N nutrition status of the wheat plants. Early N deficiency stops tiller appearance and growth of developing tillers (Masle, 1981a,b), thereby reducing spike number per square meter. Late N deficiency does not affect spike number, but grain number per spike can be decreased, because of flower abortion or reduction in floret primordium initiation or total spikelet number (Langer and Liew, 1973; Whingwiri and Kemp, 1980; McMaster, 1997). Moreover, there may be compensation between the various components of grain number. Thus, the final consequences for GN of changes in N supply at various stages within the period of seed set are unknown. Therefore, most of available crop models for wheat do not simulate GN in sub-optimal N nutrition conditions sufficiently precisely to be used for adjusting fertilization strategies (see for examples O'Leary and Connor, 1996b; Landau et al., 1998).

We characterized N deficiencies throughout the period of grain formation, using a N nutrition index (NNI) proposed by Lemaire and Gastal (1997). This ratio is computed by dividing the observed N concentration of the aerial parts by a critical N concentration (Nc). The latter value Nc corresponds to the minimum N concentration of the aerial organs required to ensure maximal growth. Lemaire and Salette (1984) and Lemaire and Gastal (1997) showed that this critical concentration changes during the growing season and is strongly correlated with the aerial biomass of the crop. The curve proposed by Justes et al. (1994) for wheat holds for a wide variety of soil and climate conditions, crop growth rates, developmental stages until anthesis, and cultivars. If NNI is higher or equal to 1, the N nutrition status of the wheat crop does not limit its growth; if NNI is lower than 1, the crop is N deficient; the lower the NNI, the more intense the deficiency. This indicator has already been successfully used in analysis of the effect of N deficiency on cumulative intercepted photosynthetically active radiation, radiation-use efficiency, and on the growth of a tall fescue (Festuca arundinacea Schreber) sward (Bélanger et al., 1992), and for a posteriori characterization of experimental data, for crop management studies, and analysis of yield variation in farm networks (Lemaire and Meynard, 1997).

In this study, we analyzed wheat grain number per square meter for crops with various N nutrition conditions, using the NNI to characterize crop N status throughout the period of grain set. The GN for a wheat crop was analyzed as a function of the characteristics of the N deficiency affecting this crop.


   Materials and methods
TOP
 ABSTRACT
 INTRODUCTION
 Materials and methods
Results
 Discussion
 Conclusion
 REFERENCES
 
Experiments
Four field experiments were conducted with winter wheat, cultivar Soissons. Experiment 1 was carried out in 1990-1991 near Laon (49.33 °N, 3.37 °E) on a sandy clay loam (Luvisol in Baize and Girard, 1995). In 1991-1992, 1994-1995, and 1995-1996, Exp. 2, 3, and 4, respectively, were carried out at the experimental station of the Institut National de la Recherche Agronomique (INRA) in Grignon (48.9 °N, 1.9 °E) on a clay loam (Luvisol in Baize and Girard, 1995).

We aimed to create a diversity of crop N nutrition conditions, differing in the period of N deficiency. Thus, various N availability conditions were set up by including treatments that were diverse in terms of soil and climatic conditions, preceding crops (Table 1) and dates and rates of N application (Table 2) . The time of important growth stages is given for each year (Table 2). In each trial, a control treatment was set up in which fertilization was managed such that NNI remained not significantly different from 1 throughout the period of seed set. For each control treatment, the total amount of N-fertilizer to be applied was estimated by the balance-sheet method (Rémy and Hébert, 1977), with three or more applications between tillering and anthesis.


View this table:
[in this window]
[in a new window]
  		Table 1 The period of study began at the date of measurement of residual mineral N in the soil at the end of winter and ended at anthesis. The mean temperature (Tmean), the mean global radiation (RG), and the cumulated amount of water from rain and irrigation are calculated for the study period

 

View this table:
[in this window]
[in a new window]
  		Table 2 Treatments: dates (days of year) and amounts (kg N ha-1) of fertilizer N applications. The mean dates of the main visible growth stages (according to Zadok's scale) are given for each experiment

 
Plots were 18.0 by 4.0 m (Exp.1) or 30.0 by 1.75 m (other experiments) in size. They were laid out in a randomized block design with three replicates. In all experiments, soil water content was monitored and maintained by irrigation such that drought stress was prevented. Crops were fully protected against weeds and pests.

Plant measurements began on the day of the first N application at the end of February. From this date until anthesis, aerial biomass and N content of the aerial parts were determined each week by taking four samples (in Exp. 1) and two samples (in Exp. 2, 3, and 4) of 0.175-m2 area within each plot. The samples were dried in an oven (48 h at 80°C), weighed, and ground. Total N concentration in shoots was determined by the Dumas (1831) method. This involves the combustion of dehydrated and ground plant tissue at about 1800°C, reduction of N oxides by reduced Cu at 600°C, and analysis of N2 by catharometry (1500 analyzer; Carlo Erba NA, Les Ulis, France). This method makes it possible to estimate the total N content of the plant, including nitrate. The spike and grain numbers were counted at harvest, on four 0.175-m2 areas per plot. The grain number per spike was estimated per plot by the ratio of grain number to spike number. The grains were dried (48 h at 80°C) and weighed to determine yield.

Characteristics of the Periods of N Deficiency
The periods of N deficiency were characterized in terms of NNI, calculated as the ratio between measured N concentration (Nm in %) of the aerial biomass (BM) and wheat critical N concentration (Nc in %), as determined by Justes et al. (1994):



Analysis of variance was performed with the GLM procedure of SAS (SAS Institute, 1987). Means were compared by the least significant difference (LSD) at the 0.05 probability level.

Seven characteristics of N deficiency were then defined for each crop (Fig. 1) : (i) the beginning of deficiency (BD), (ii) the end of deficiency (ED), (iii) the duration of deficiency (DD = BD - ED), (iv) the intensity of deficiency (ID), corresponding to the minimum NNI observed during the growth cycle (ID = 1 - NNImin), (v) the product ID x DD = IDD, which is close to half the area between the curve of NNI dynamics and the curve of NNI = 1, (vi) the NNI measured at anthesis, considered by Justes et al. (1997) to be an overall criterion for the whole period of deficiency (NNIa), and (vii) the lack of N accumulation (LNA), measured at anthesis. The lack of N accumulation (LNA) was calculated as the difference between the amount of N accumulated in the aerial parts for a treatment in which deficiency occurred, and the amount of N that would have accumulated in a crop with maximal growth and critical N content throughout the crop cycle. The dates of the beginning and end of deficiency were the dates when NNI reached a value of 0.90. Linear interpolation between the two values just on either side of 0.90 was used to determine precisely the moment when NNI reached this threshold value. The dates of the beginning and end of deficiency were then expressed in cumulative degree-days (base 0°C) before anthesis, and the duration of the deficiency in degree-days. The various treatments were carried out to obtain large ranges of values for each of these criteria, without strong correlations between them in all situations.



View larger version (19K):
[in this window]
[in a new window]
  		Fig. 1 Scheme of the various characteristics of N deficiency. BD = beginning of deficiency (when NNI falls under 0.90), ED = end of deficiency (when NNI increases above 0.90), DD = duration of deficiency (=BD-ED), ID = intensity of deficiency (= 1 - NNIminimum observed)

 

   Results
TOP
 ABSTRACT
 INTRODUCTION
 Materials and methods
 Results
Discussion
 Conclusion
 REFERENCES
 
Change in NNI Over Time for the Various Treatments
The treatments resulted in a wide variety of changes in NNI from the end of winter until anthesis (Fig. 2) . In each trial, NNI of the control was kept at about 1 throughout the cycle. These controls received the highest rate of N fertilizer, applied on several dates. Thus, control treatments suffered no period of N deficiency, as expected.



View larger version (33K):
[in this window]
[in a new window]
  		Fig. 2 Changes in N nutrition index (NNI) over time, for the various treatments in the four experiments

 
For the other treatments, there was always some highly variable period of time with a NNI significantly lower than 1 between the end of winter and anthesis. Deficiency began between day 55 (e.g., 3N1) and Day 137 (e.g., 1N21), corresponding to various developmental stages, from the mid-tillering period until the inflorescence emergence, as indicated in Table 2. Some deficient crops received a N application during the period of deficiency. This caused an increase in NNI (e.g., treatments 3N2 and 3N4). For some treatments, NNI was still not significantly different from 1 after this N application (Fig. 2). For the crops that received no N application during the deficiency period, the deficiency continued until anthesis, with NNI decreasing continuously until this stage. Thus, a range of dates for the end of deficiency was obtained, from Day 104 (e.g., 1N2) to Day 149 (e.g., 1N1), apart from those treatments that remained deficient until anthesis. The minimum value of NNI for each treatment was between 0.28 and 0.99.

N Deficiency Characteristics
A large range of values was obtained for each criterion of the periods of N deficiency, for the various treatments (Table 3) : BD was from 0 (treatment without deficiency) to 894 degree-days before anthesis, ED from 0 (treatments with no deficiency or with a deficiency lasting until anthesis) to 670 degree-days before anthesis, DD from 0 (no deficiency) to 894 degree-days, ID from 0.01 (treatment without deficiency) to 0.72, NNIa from 0.31 to 1.10, and LNA from +24 kg N ha-1 (Control3, without deficiency) to -194 kg N ha-1 (Treatment 3N1).


View this table:
[in this window]
[in a new window]
  		Table 3 Characteristics of the periods of nitrogen deficiency for each treatment

 
The correlations between criteria (Table 4) were very small (e.g., r = -0.008 between ED and ID) or high (e.g., r = 0.989 between LNA and NNIa). Of the seven criteria studied, the coefficients of correlation higher than 0.9 were those between LNA and NNIa and LNA and IDD, three integrative criteria measured at anthesis, and between IDD and ID, as expected since IDD = ID x DD. The correlation coefficients between the three basic indicators (BD, ED, and ID) were all below 0.8, indicating that our experimental treatments successfully dissociated deficiency characteristics.


View this table:
[in this window]
[in a new window]
  		Table 4 Correlation coefficient (r) between the characteristics of nitrogen deficiency (presented in Table 3) and relative grain number (RGN)

 
Consequences for Grain Number
Grain number per square meter (GN) differed greatly among treatments (Table 5) , from 6812 (3N1) to 28835 (1N6). Almost 91% of the variation in yield was accounted for by GN variability (Table 5). In each trial, the highest GN was obtained for treatments with no N deficiency throughout the cycle, but the maximums were not always for the control treatments. However, grain number for the controls was not significantly different from that of the treatment with the highest GN. The reduction of GN because of N deficiency resulted from the reduction of either the spike number per square meter (Treatments 1N1, 1N3, 1N7, 1N8, 1N9, 2N1, 3N3, 3N5, and 3N6) or the grain number per spike (Treatment 4N4) or both of them (Treatments 2N2, 2N3, 3N1, 4N1, and 4N2). The absence of significant difference from GN of the control was sometimes linked to a reduction in spike number per square meter and an increase in grain number per spike (e.g., 1N4 and 4N5) or a reduction in spike number per square meter and no significant difference in grain number per spike (Treatments 1N5, 1N10, and 1N12).


View this table:
[in this window]
[in a new window]
  		Table 5 Grain number per square meter, spike number per square meter, grain number per spike, and yield

 
The highest GN differed between years, so relative grain numbers were estimated to assess the effect of the deficiency periods in all situations. Relative grain number (RGN) for a treatment is calculated as the ratio between GN for this treatment and GN for the control treatment of the same trial. For treatments with a period of N deficiency, a diversity of RGN was observed. The treatments with the lowest grain numbers were those in which N deficiency occurred early in the growth cycle and did not cease before anthesis, i.e., those with the longest periods of N deficiency (3N1, 3N3, 4N1). GN in treatments with short deficiency did not significantly differ from the control for the same year (e.g., 3N2, with DD = 224 degree-days and ID = 0.43). In any one year, for the same starting date of deficiency (BD) and similar intensity (ID), GN was more affected when the duration of deficiency was longer (e.g., 3N4 and 3N5). For the same duration and the same BD, GN was lower if the intensity of the deficiency was higher (1N1 and 1N3). The three basic criteria (BD, ED, and ID) seemed to have an effect on GN. However, none can account for all the variability of GN, if considered alone. The correlation coefficients between the deficiency criteria and RGN (Table 4) were -0.607 (BD), 0.278 (ED), -0.808 (DD), and -0.853 (ID), meaning that the duration and intensity of the deficiency had a far higher effect on RGN than the beginning and end of deficiency. With the integrative criteria NNIa, LNA, and IDD, the correlations were higher, 0.920, 0.947, and -0.963, respectively (Table 4).

Relationships between Relative Grain Number and Deficiency Characteristics
We determined the criteria most explicative of the decrease in grain number, and analyzed their relative importance, using a multiple progressive regression with a forward procedure from SAS (SAS Institute, 1987), as it includes only the variables which are significant in the regression. The relative grain number was the variable to be explained. The explicative variables included in the analysis were BD, ED, DD, ID, NNIa, IDD, and LNA.

The variables entered in the regression, by order of importance, were IDD, LNA, ID, and DD, all being significant for RGN. The final regression accounts for 97% of the variability in RGN (Table 6) . The sign of the parameter for the variables IDD (negative parameter) and LNA (positive parameter) are consistent with the effect of the deficiency: the grain number is all the more reduced (RGN lower) when the deficiency is all the longer or all the more intense (IDD higher), and the lack of nitrogen deficiency is more important (LNA lower, as it is in negative value). The effect of ID and DD is low (partial r2 = 0.0131 and 0.0060, respectively). The fact that BD is not a significant variable in the regression does not mean that the period of occurrence of the N deficiency within the cycle has no effect on grain number, because among our treatments, there is a significant correlation between BD and DD (r = 0.800, Table 4), DD being significant in the regression. However, we can say that, for a given duration of deficiency (DD), the period of deficiency (BD) has no significant effect on the reduction of grain number. It is illustrated in the comparison between Treatments 1N11 and 3N4: they have close DD (respectively, 407 and 387 degree-days), different BD (respectively, 407 and 667 degree-days), and yet close RGN (0.95 and 0.94, respectively). The same conclusion can be drawn from the comparison between Treatments 1N9 and 1N11: they have the same DD (407 degree-days) and same BD (407 degree-days) and yet different RGN (respectively, 0.83 and 0.95). It appears that IDD, the interaction between the duration and the intensity of the deficiency, is the most explicative variable, as it accounts for 93% of the variability in RGN (partial r2 = 0.929).


View this table:
[in this window]
[in a new window]
  		Table 6 Estimation and evaluation of two models simulating the relative grain number (RGN) according to the deficiency characteristics (IDD, LNA, ID, DD)

 
The variables LNA, ID, and DD, yet significant in the model, have a low influence on RGN in comparison to IDD, as the model with IDD only and the model with the four explicative variables have close coefficients of correlation (r2 = 0.929 and r2 = 0.968, respectively, Table 6) and close mean errors of simulation (the root square of the mean square error, MSE, is 0.0491 and 0.0353, respectively, Table 6). To test the interest of keeping LNA, ID, and DD in the final model, we compared the performance of prediction of the model with IDD only to that of the model with the four explicative variables. In this aim, we evaluated the two models using the "leave-one-out cross validation" method (Linhart and Zucchini, 1986). We successively realized the four steps: (1) estimating the parameters of each model on 29 out of 30 situations, (2) estimating the predicted RGN with each model on the last situation not included in the first step, (3) performing Steps (1) and (2) for each of the 30 situations of the initial sample, and (4) estimating the mean square error of prediction (MSEP) of each model in comparing the observed RGN and the predicted values from Step (2). It resulted that the root square of the MSEP is only 20% lower with the complex model: 0.05121 for the model including IDD only and 0.04026 for the model including the four variables, indicating close performance of prediction of the two models.

This procedure also made it possible to quantify the decrease in grain number, relative to a crop with non-limiting N nutrition status, in terms of the most explicative criteria. The relationships obtained for each model are given in Table 6. The simple model gives an accurate prediction of RGN for the whole range of observed values and for permanent and temporary deficiencies (Fig. 3) .



View larger version (18K):
[in this window]
[in a new window]
  		Fig. 3 Relationship between observed and predicted values of relative grain number (RGN), for all the crops described in Table 2, classified according to N deficiency (pd = permanent deficiency, td = temporary deficiency, and no d = no deficiency)

 

   Discussion
TOP
 ABSTRACT
 INTRODUCTION
 Materials and methods
 Results
 Discussion
Conclusion
 REFERENCES
 
The treatments used in our experiments resulted in various dynamics of crop N accumulation, and periods of N deficiency. The periods of deficiency differed in date of occurrence, duration, and intensity. Large ranges were obtained for each of the deficiency characteristics. The ranges of these characteristics were similar to those recorded in other studies with wheat or other crops: e.g., the NNI obtained for tall fescue swards was from 0.38 to 1.2 (Lemaire and Gastal, 1997); the minimum value of NNI corresponding to the minimum N content of the aerial parts observed on wheat (Justes et al., 1994) was 0.39; for a network of wheat trials in various locations in France, the values of NNI measured at anthesis were from 0.25 to 1.7, and relative grain number was between 0.37 and 1 (Justes et al., 1997).

Application of N to a deficient crop changed the progression of NNI according to the trial. NNI increased rapidly to values close to 1 (e.g., 3N2 and 3N4), or increased slowly, remaining significantly below 1 until anthesis (e.g., 2N1), or remained constant at the same minimum value over several weeks (e.g., 2N2). These differences in the NNI time course after N application were linked, in a given experiment, to the amount of N applied and the crop phenological stage at which N was applied. The increase was smaller after application of a lower N rate and after applications later in the cycle (e.g., Exp. 3). The closer to anthesis the N application occurred, the shorter the time to allow an increment of NNI.

We compared the effect on grain number of various periods of N deficiency in different years by expressing their beginning and end in degree-days before anthesis. We did this because the effects of the various periods of N deficiency were more likely to be linked to developmental stages than to calendar dates, as observed by McMaster (1997) or for other limiting factors in wheat and other crops. For example, if N becomes limiting for the growth of a wheat plant during the tillering period, the production of new tillers stops and all the tillers which have not yet developed three leaves at this time will stop growing and die without giving a spike (Masle, 1981b). If low levels of radiation occur for a very short period around meiosis, the grain number per spike in wheat is reduced (Demotes-Mainard et al., 1995). In pea (Pisum sativum L.) crops, pods are sensitive to high temperatures during the second part of their period of seed formation (Jeuffroy et al., 1990), which corresponds to a period of around 150 degree-days. In wheat before anthesis, for a given cultivar, the period between two successive developmental stages is constant in cumulative degree-days (Gate, 1995, for the stages 30, 31, 32, 39, 40, and 55 in Zadoks' scale). Thus, a same length of time backwards from anthesis (end of our period of study) corresponds to approximately the same developmental stage for every wheat crop. In contrast, the calendar dates on which the various stages occur are highly variable for different years and locations, because photoperiod and temperature, the two main factors, affect them (Masle et al., 1989).

As we showed, the N stress response of the crop, in terms of grain number, depends on the history of the stress (mainly intensity and duration). In order to quantify this stress history, we proposed to use the evolution of the NNI. In this aim, the use of a single threshold to determine the period of N deficiency is particularly interesting in crop models, because the statistical comparison between the simulated value of the NNI and 1 (method generally used to determine whether the crop is in deficiency) cannot be carried out. It is thus necessary to determine a single value for NNI, below which the crop is considered to be in deficiency. Because of the variability of samples, values of NNI lower than 1, but not significantly different from 1, are often observed. For example, the lowest NNI for crops with no deficiency was 0.79 and the highest NNI for N deficient crops was 0.92. Thus, the single threshold chosen must inevitably be lower than 1. We chose 0.90 as the threshold because this value gave the most consistent results for the comparison of individual means, despite errors of classification due to sample variability. With this threshold, only 5% of the crops in which NNI was significantly lower than 1 were classified as not deficient, and 14% of the crops with NNI not significantly different from 1 were considered to be N deficient. Seven percent of all crops were misclassified, versus 11% with the other two thresholds tested, 0.95 and 0.85. The relationship between relative grain number and the characteristics of the deficiency was most clear taking 0.90 as the threshold. The coefficients of determination of the best multiple regression were 0.950 with 0.95 as the threshold, 0.968 with 0.90 as the threshold and 0.966 with 0.85 as the threshold, for the same explicative variables.

The high correlation between yield and grain number in our experiments is consistent with many previous results (Boiffin et al., 1981; Echeverria et al., 1992; Fischer, 1993; Abbate et al., 1995). Yet in our experiments grain number was linked to spike number in some cases and to grain number per spike in other cases. Maximum grain number differed among years. It is known that potential GN is strongly correlated with radiation intercepted by the crop during grain set (Masle, 1985), and with the ratio between intercepted radiation and temperature during this period (Fischer, 1985). Therefore, the observed differences in maximum grain number were probably due to climate variability, as indicated by the high coefficient of correlation between GN and the mean ratio of global radiation to temperature between the end of winter and anthesis (r = 0.728).

Analysis of the N nutrition status of a wheat crop throughout the growth cycle makes it possible to determine various characteristics of the N nutrition of the crop. For the whole range, the decrease in grain number can then be deduced from these criteria. Thus, NNI is a simple criterion which is very useful for the prediction of grain number in various N nutrition conditions, as it was already shown to be a good indicator for the prediction of radiation interception, radiation-use efficiency, and growth in such conditions (Bélanger et al., 1992). The relationship proposed for quantifying grain loss in terms of ID, DD, and LNA holds for both long, permanent deficiencies and temporary ones, and for both deficiencies reducing spike number and grain number per spike. According to this relationship, whatever the crop stage at which the deficiency occurs, similar values of IDD result in similar values of RGN. However, in our situations, similar values of IDD were obtained at various periods with different reductions of the amount of fertilizer applied relative to the optimal rate applied to the controls (Fig. 4) .



View larger version (18K):
[in this window]
[in a new window]
  		Fig. 4 Relationship between IDD, the product of the intensity with the duration of the intensity, and the reduction in amount of N applied from that applied to control

 
Among the few crop models simulating grain number, most of them simulate grain number as a function of the crop or spike biomass at anthesis (Weir et al., 1984; O'Leary and Connor, 1996a,b). This gives sometimes a poor estimate of GN in conditions of sub-optimal N nutrition, as shown by O'Leary and Connor (1996b) and Abbate et al. (1995) for monotonous N nutrition conditions. This may be because the effect of N deficiency on grain number is not purely dependent on its effect on crop growth (Abbate et al., 1995). Dry matter partitioning to growing spikes may also be affected by N stress as observed by Fischer (1993). In the relationship proposed herein to account for grain loss in terms of the characteristics of the N deficiency, grain loss does not depend only on crop growth, and this seems to be more effective and rather simple to account for grain number in wheat crops in which N nutrition is limiting. Before including this relationship in such models, it should be tested if this relationship gives also a good account for grain number in situations including other limiting factors, such as water stress, a frequent limiting factor in the main areas of growing wheat, and for other cultivars.

The relationship predicting RGN from four characteristics of N deficiency indicates that, whatever the stage at which N deficiency appears, it will decrease GN and yield, particularly if it lasts a long time (in degree-days) or is intense (minimum NNI is low). The influence of the date of beginning of the deficiency depends on the end of the period: for deficiencies which are not stopped by a N application before anthesis, the duration is all the longer and the reduction of grain number is all the higher that the deficiency began earlier. For such deficiencies, the relative grain number is highly linked to the beginning of the deficiency. But, when N is applied during the period of deficiency, the influence of the beginning of deficiency is reduced, as RGN is generally not reduced in these treatments, although in those crops a period of deficiency occurred but RGN was not significantly different from 1. It thus seems that some deficiencies can be tolerated without reducing grain number and yield. These deficiencies all have a low value for IDD because of the duration of the period of deficiency being short or the intensity of deficiencies too low. All these cases correspond to treatments for which there was a maximum of 40 kg ha-1 less than for treatments with maximal growth and critical N content, whatever the crop stage was when deficiency occurred.

Similar amounts of fertilizer applied to a wheat crop do not necessarily result in similar grain numbers and is dependant on the date of application. This is due to the fact that they do not lead to the same dynamics for NNI, and to the same value for IDD, because of differences in apparent or real utilization coefficients (Machet et al., 1987; Recous et al., 1997; Recous and Machet, 1999) for the fertilizer applied, and in growth dynamics and N absorption throughout the cycle. This result is important for fertilization management because it is then possible to estimate the best dates for N application to optimize grain number. Most crop models simulate growth and N accumulation in the aerial organs, so it would then be easy to simulate NNI throughout the growth cycle. The relationship proposed to quantify grain loss is easy to incorporate into such models to simulate efficiently relative grain number in crops with periods of N deficiency, as suggested by Jeuffroy and Recous (1999).


   Conclusion
TOP
 ABSTRACT
 INTRODUCTION
 Materials and methods
 Results
 Discussion
 Conclusion
REFERENCES
 
The time-course change of the NNI on a wheat crop allows one to determine precisely the period of occurrence of a N deficiency, and its intensity during the vegetative phase of the crop cycle. According to the period of deficiency, the grain number of the wheat crop was reduced, either because of a reduction in spike number, a reduction of grain number per spike, or both. In the whole range of deficiencies tested, the reduction in grain number, relative to the control of the same experiment, was strongly correlated to the product of the duration with the intensity of the deficiency. The relationship proposed allows one to predict the grain number of a wheat crop subjected to various fertilization strategies.Baize Girard 1995


   ACKNOWLEDGMENTS
 
We thank Myriam Dauzat, Béatrice Le Fouillen, and Florence Lafouge for excellent technical assistance. We thank Sylvie Recous for management of the experiment in 1991 and the Unité Expérimentale at Grignon for the experiments in 1992, 1995, and 1996. We thank Dr. Jean-Marc Meynard for valuable discussions and helpful comments in the preparation of this paper, and Hervé Monod for his help in the statistical analysis. This work was partly supported by grants from Société Chimique La Grande Paroisse S.A (Paris), and from INRA through its AIP Ecofon.

Received for publication September 18, 1998.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 Materials and methods
 Results
 Discussion
 Conclusion
 REFERENCES
 




This article has been cited by other articles:


Home page
 Agron. J.Home page
W. W. Wilhelm, G. S. McMaster, and D. M. Harrell
Nitrogen and Dry Matter Distribution by Culm and Leaf Position at Two Stages of Vegetative Growth in Winter Wheat
Agron. J., September 1, 2002; 94(5): 1078 - 1086.
[Abstract] [Full Text] [PDF]

Home page
 J Exp BotHome page
M.H. Jeuffroy, B. Ney, and A. Ourry
Integrated physiological and agronomic modelling of N capture and use within the plant
J. Exp. Bot., April 15, 2002; 53(370): 809 - 823.
[Abstract] [Full Text] [PDF]

Home page
 Agron. J.Home page
J. L. Hatfield, T. J. Sauer, and J. H. Prueger
Managing Soils to Achieve Greater Water Use Efficiency: A Review
Agron. J., March 1, 2001; 93(2): 271 - 280.
[Abstract] [Full Text] [PDF]

Home page
 Crop Sci.Home page
S. Demotes-Mainard and M.-H. Jeuffroy
Incorporating Radiation and Nitrogen Nutrition into a Model of Kernel Number in Wheat
Crop Sci., March 1, 2001; 41(2): 415 - 423.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (24)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Jeuffroy, M.-H.
Right arrow Articles by Bouchard, C.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Jeuffroy, M.-H.
Right arrow Articles by Bouchard, C.
Agricola
Right arrow Articles by Jeuffroy, M.-H.
Right arrow Articles by Bouchard, C.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Vadose Zone Journal
Journal of Natural Resources
and Life Sciences Education		 Soil Science Society of America Journal
Journal of Plant Registrations Journal of
Environmental Quality		 The Plant Genome