Introduction

During the second half of the 20th century crop yields increased significantly, however, during the last decades it has declined (^{Brisson et al., 2010}; ^{Fischer and Edmeades, 2010}). Coupled with these reductions in yield gains, increases in population mean that per capita food production also declines, which leads to the need for yield increases of around 50% in the near future (^{Reynolds et al., 2009}).

Due to the above, it is necessary to study the variability in yield of the best germplasm, so that the breeders can count on elite progenitors to include in new crosses that allow to increase the potential yields of wheat. In relation to this need, the wheat yield consortium (WYC) is created in 2009, which brings together a group of world experts in wheat (physiologists, geneticists, agronomists) and within its priorities establishes The study of elite wheat germplasm (^{Reynolds et al., 2011}). Before this the is formed CIMCOG (CIMMYT Mexico Core Germoplasm), which integrates a set of 60 wheat genotypes that has excelled due to its good behavior and agronomic adaptation in some regions of the world.

For many years the yield improvement in wheat has been strongly associated with increases in the number of grains per unit area (NG) and harvest index (IC) (^{Shearman et al., 2005}), the latter being considered as the proportion Of the total biomass that represents the harvested grain (^{Zhang et al., 2012}). The increases in the IC were due to reductions in plant height, which modified the biomass produced and consequently its distribution to the reproductive organs, increasing NG (^{Brancourt-Hulmel et al., 2003}; ^{Shearman et al., 2005}; ^{Zhou et al., 2007}). Although there have been substantial increases in NG, many evidences in the literature mention that potential wheat yield is limited by a small NG (demand), mainly in modern varieties (^{Miralles and Slafer, 2007}; ^{Zhang et al., 2010}). The above would indicate that there is still potential to increase wheat yield by increasing demand and thus increase IC without substantial changes in total biomass.

There are currently studies that identify elite wheat lines that could significantly contribute to increased wheat yield, identifying agronomic and physiological characteristics to be included in genetic breeding programs (Reynolds et al., 2015). In this sense, studying the genotypic variability in the physiological and numerical components of the yield of elite cultivars could be very useful in breeding programs. Despite this, crops are vulnerable to environmental variation, which can be considered as an obstacle to improve yield potential in wheat (^{Reynolds et al., 2004}).

One of the numerical components of yield is the grain weight (PG) which is defined during the period between flowering and physiological maturity, and can be affected by both biotic and abiotic stresses (^{Estrada-Campuzano et al., 2008}). For many years it has been mentioned that PG is not associated with changes in grain yield (^{Shearman et al., 2005}; ^{Peltonen-Sainio et al., 2007}), however, it has been recently discovered that some combinations in some lines of CIMMYT germplasm combine high yields with high grain weight (^{Sayre et al., 1997}; ^{Rattey et al., 2009}).

Therefore, evaluating the effect of changes in the environment on NG and PG in elite wheat lines would help to understand the mechanisms behind the distribution of assimilates from the leaves and stems to the spike during the period of grain filling. Based on the above, the objective of the present study was to evaluate the effect of source limitation during the grain filling stage (anthesis to physiological maturity) on the physiological and numerical components of wheat yield through defoliation and water stress.

Materials and methods

Results and discussion

The analysis of variance showed significant effects of treatments on the variables evaluated in both experiments (Table 1). In this sense, the effect of repetitions was only significant in biomass at physiological maturity (BioMF), RG, NG and NE, and only for the case of experiment 1. The genotypic effect was highly significant for all variables studied in both experiments, while the environments evaluated (defoliation and drought) caused significant changes in the characters studied. In the treatment of defoliation the variables BioMF, RG, NGPE and IC were significantly influenced by the genotype*environment interaction, which was manifested in a differential behavior of the genotypes in each of the environments studied.

Table 1 Values of F and their statistical significance for different wheat lines evaluated under contrasting environmental conditions (defoliation and drought) in Toluca, Mexico  

The coefficients of variation fluctuated between 3.1 and 12.9% corresponding to the variables NE and NGPE, respectively. In experiment 2, the genotype*environment interaction was only significantly present in the NGPE. The coefficients of variation were less than 10% in most of the variables shown in the experiment (Table 1). When different genotypes are exposed to contrasting environmental conditions it is common to detect significant effects of cultivars, environments and their various interactions (^{Koutroubas et al., 2012}; ^{González et al., 2014}).

The results revealed significant effects of the different environments (defoliation and drought) evaluated on the characteristics that determine grain yield, as well as the genotypic effect was responsible for substantial changes in the observed variability along with genotype*environment interaction.

The depletion significantly affected grain yield through reductions in biomass production, as the crop index remained constant in both environments (0.26 to 0.56); in this sense, reductions in yield were better explained by changes in biomass to physiological maturity (r²= 0.44 p< 0.01, Figure 1a) than for changes in the IC (r²= 0.3 p< 0.01, Figure 1b). The defoliation reduced grain yield by 15% compared to treatment without defoliation and the effect of this type of stress depended on the cultivar, with the yield range for the defoliated plants being 260.8 to 568.8 g m^-2, whereas for the control (without defoliation) were maintained between 338.1 and 608.4 g m^-2.

Figure 1 Relationship between grain yield and biomass at physiological maturity (a), and crop index (b). In 60 wheat lines grown in two treatments of manipulation of the source-demand (defoliado and without defoliar).  

Although IC had values higher than 0.5 in both treatments, it was not reflected in a higher grain yield. Similar results were reported by ^{Bijanzadeh and Emam (2010)} who, when evaluating 5 wheat cultivars: Shiraz, Bahar, Pishtaz and Sistan (wheat bread) and Yavaros (durum wheat), found that defoliation of all leaves decreased yield grain and its components more drastically, compared to just leaving the flag sheet. They also observed that with the exception of Pishtaz, all cultivars significantly decreased grain yield of the main stem by defoliation treatments; in the Shiraz cultivar the defoliation of all the leaves decreased the grain yield of the main stem in 40.75% compared to the control which could demonstrate that Shiraz was sensitive to the restriction of the source. On the other hand, ^{Castañeda-Saucedo et al. (2009)} when evaluating the effect of environmental conditions on grain yield and seed quality on barley and wheat, observed that the IC in wheat was superior, however it was not related to a higher yield.

The numerical components of grain yield showed broad genotypic variability (Figure 2), which were differentially affected by source-demand manipulation treatments. The results suggest that the observed changes in grain yield by effect of genotypes and environments were due to changes in NG rather than to PIG. Under normal growth conditions (without defoliation) the number of grains varied from 6 937 to 13 577 grains m^-2, while the individual grain weight ranged from 33.5 to 57.5 mg grain^-1.

Figure 2 Relationship between grain yield and number of grains per m² (a) and individual grain weight (b).In 60 wheat lines subjected to defoliation and nondefoliation conditions.  

The wide genotype variability in the number of grains per unit area was in line with the observed changes in the number of grains per spike, which turned out to be the component that best explained the observed changes in the number of grains (r²= 0.55 p< 0.001) (Figure 3). The alteration in the source (defoliation) relation affected marginally the components of the number of grains, so that the grain per spike was only affected by 5%, compared to the control environment. The restrictions to which the crops were submitted reduced the NGPE, which is in agreement with ^{Bijanzadeh and Emam (2010)}, who evaluated the cultivars Shiraz, Bahar, Pishtaz, Sistan and Yavaros, indicated that the number of grains by ear was significantly affected by the defoliation treatments of all the leaves of the plant and even leaving only the flag leaf; the cultivars Sharaz and Bahar had a reduction of 18.97 and 11.07%, sistan was the least affected (0.43 to 0.91%). Meanwhile, ^{Chowdhary et al. (1999)}, reported that removing all leaves in spring wheat caused a reduction of 17.17 and 13.27% for the number of grains per spike and weight of 100 grains respectively.

Figure 3 Relationship between number of grains and number of grains (a), number of grains and number of grains per spike (b). In 60 wheat lines grown under conditions of defoliation and without defoliation.  

Response of yield to water stress

The stress conditions generated by water deficit varied the yield of grain for the different cultivars (Figure 4). The yield of grain in plants under irrigation conditions (control) ranged from 348.5 to 542.9 g m^-2; while under drought conditions they fluctuated between 239.4 and 457.4 g m^-2, representing a 13% decrease due to stress. On the other hand, water stress during grain filling affected biomass production in 10% with respect to control and IC was only affected marginally (4%) (Figure 4). It should be noted that the observed changes in grain yield due to the evaluated treatments were explained by the biomass production at maturity (r²= 0.52 p< 0.001) and by the harvest index (r²= 0.51 p< 0.001).

Figure 4 Relationship between grain yield and biomass yield to physiological maturity (a), and crop index (b).In 20 wheat lines grown under irrigation and drought conditions.  

This coincides with ^{Ortiz et al. (2003)}, who, when studying leaf water parameters in wheat and their use in selection of drought-resistant genotypes, observed that irrigation conditions and genotype influenced significantly the variation in yield; the treatment of drought caused a 50% decrease in grain yield with respect to the control (irrigation) (3 609 and 6 917 kg ha^-1, respectively). The reduction of IC was attributed to the reduction of the size of the “demand”in the treatment of drought resulting in a lower proportion of assimilates, and ultimately as a resource to be retained in the vegetative organs (^{Nicolas et al., 1985}).

The individual grain weight (PIG) was significantly affected by post-anthesis water stress, the response observed to water stress was similar (9% to control) than that observed in defoliated plants (Figure 5). In this sense, the PIG under irrigation conditions was 39.7 mg grain^-1, while under water stress conditions it was 36 mg grain^-1. It was observed that there was no relationship between the grain yield and the PIG, contrary to the variations in yields were more positively associated with changes in NG (Figure 5).

Figure 5 Relationship between grain yield and number of grains per m² (a) and individual grain weight (b).In 20 wheat lines grown under irrigation and drought conditions.  

In spite of the low affection to the NG, these explained greatly the changes observed in the yield, which coincides with ^{Cossani et al. (2009)} who mention that grain yield for all treatments and species was mainly determined by the number of grains per unit area, even under the most stressful conditions of water; Likewise, they observed that there was no relation between weight and grain yield in bread wheat. On the other hand, ^{Rajala et al. (2009)} when studying different treatments of water deficit in spring wheat detected that the number of grains per spike and the individual weight of the grain was clearly affected.

As expected, the number of grains per spike was only marginally affected by the effect of post-parasitic water stress, so that this decrease represented only 4% of the control. Also, changes in the number of grains per unit area per effect, both genotype and water stress were mainly associated with the change in the number of grains per spike (Figure 6). Our results showed no relationship between NG and NE, on the contrary there was a strong positive relationship between NG and NGPE. Similar results report that the number of grains per m² under drought conditions depended on 43% of spikes per m² and 56% of grains per spike (^{Ortíz et al., 2003}). However, ^{Giunta et al. (1993)}, studying the effects of drought on yield and its durum wheat components in Mediterranean-type environments, reported that the main components affected were the number of fertile spikes per unit area and the number of grains per spike.

Figure 6 Relationships between number of grains and number of spikes per m² (a) and number of grains per spike(b). In 20 wheat lines grown under irrigation and drought conditions.  

Thus, the results of the present study indicate that the effect of the interaction of the genotype, the soil and the type of environment determine the yield of the wheat; and is mainly characterized by the limitations of the source to which the crop was subjected. Meanwhile, the component that best explained the variations in yield was the number of grains per m².

Conclusions

There was variability among the genotypes for all variables evaluated, grain yield and its components were affected by the stress conditions to which the crops were subjected. The decreasing imposed defoliated source declined proportionally performance, single grain weight, number of grains per m² and number of grains per spike (15, 9, 5 and 5%). Moreover, under conditions of water deficit it was significantly reduced biomass yield, number of grains per m², single grain weight and harvest index (9, 13, 5, 4 and 4%). The number of grains per m² was the component that was positively related to the number of grains per spike; in this way, variations in yield tended to be associated more by changes in the number of grains than by changes in grain weight.