Analysis of Variance and Influence of Number of Grains Per spike on Protein Percentage and Yield in Wheat Under Different Environmental Conditions

Introduction

An understanding of association of various characters under a particular set of environment is necessary of synthesis of well-adapted genotypes. This becomes very important for plant characters directly related to quality and production, such as number of grains per spike, grain protein percentage of grain yield in cereals.

The semi dwarf varieties have significantly increased the yield potential of wheat (Triticum aestivum L., em Thell.) mainly through their better lodging resistance and greater sink size. The yield increase, however, has been accompanied by lower total protein percentage of the grain, which in turn, lowers water absorption capacity of the flour and reduces the loaf of the dough (Busch et al., 1969). This tendency among the wheat varieties is of critical concern both nutritionally and economically.

Previous research has indicated that the greater sink size of semi dwarf varieties is due to their producing more spikes ha^–1, more spikelets/spike and more grains/spike than normal tall cultivars (McNeal et al., 1972; Johnson et al., 1966). McNeal and Davis (1966) observed that grains from the top third of the spike were significantly lower in protein percentage than those of the bottom and middle parts. This suggests that the supply of nitrogenous products needed for protein production may become limited before the additional grains produced by the high yielding varieties mature.

The tendency for lower grain protein percentage in high yielding varieties does not result from undesirable linkages involving the dwarfing gene, but from inverse yield-protein relationship (Pepe and Heiner, 1975). Such yield-protein relationship has been reported by many investigators (Campana and Semp, 1984; Levy and Feldman, 1989; Millet et al., 1988). As yield increase is accompanied by more grains per spike (Camargo, 1987; Li, 1989; Mikheev, 1992), the inverse relationship among number of grains per spike and grain protein percentage has been reported by Mahmood and Shahid (1993) and Wells (1984).

These relationships may change because of the changes in gene expressions that may occur due to the variations in environments. These changes known as genotype-environment interactions, have long been recognized as an important source of variation in the expression of the genotypic characters (Yates and Cochran, 1938; Mather, 1949). It has been found empirically that a linear relation frequently exists between character and environment (Finlay and Wilkinson, 1963) which accounts for most of the variation in genotypic character over environments (Breese, 1969; Jinks and Perkins, 1970). Though several scientists have studied this phenomenon, the information on the subject under local environments is still inadequate. In the present study it is tried to determine the magnitude of relationship of number of grains per spike with grain protein percentage and yield under different environments.

Materials and Methods

Ten commercial bread wheat genotypes developed at three ecologically different wheat breeding centers were evaluated for two years at nine widely distributed climatically distinct locations in Pakistan and Azad Jammu and Kashmir during 1995-96 and 1996-97. Experiments were conducted in randomized complete block design with three replications and 85 g seed of each genotype was used for a plot size of 1.5x5.0 m² with 25 cm row to row distance. There was no supplemental irrigation and normal tillage and cultural practices were adopted.

Grains per spike were recorded at maturity from the spikes of primary tillers of 20 consecutive plants in the middle four rows of each plot. For this purpose plants were tagged early in the season. The tallest and vigorous tiller of each plant was considered the primary tiller for taking observations. The centre 1 meter square area of each plot was harvested to determine plot grain yield (converted to t ha^–1 at 15% moisture level). Grain protein percentage was examined with Kjeltec Auto 1030 analyzer. Amount of nitrogen was calculated by the formula, Nitrogen%= 1.401 x Normality of HCL x (ml titrant-Blank)/Sample weight. Protein contents were calculated by multiplying the percentage of nitrogen contents with protein factor, that is 5.7 for wheat. Three samples from each plot were analyzed with this procedure and average value was worked out.

The genotype x year (σ²_gy), genotype x location (σ²_gl) and genotype x year x location (σ²_gyl) variance components were determined by factorial analysis of variance as mentioned by Johnson et al. (1995). Regression coefficient was determined as proposed by Eberhart and Russell (1966) and association among the characters was calculated with the help of satistical method, as stated by Gomez and Gomez (1987).

Results and Discussion

Table 1 indicated that relative performance of genotypes for all characters is quite inconsistent across the locations as well as years. It suggests that genotypes need to be tested at more locations and years for evaluation of these characters. Large σ²_gl as compared to σ²_gy for grain protein percentage and grain yield reflects that performance of these characters was more inconsistent over the locations than years.

Table 1:	Analysis of variance for number of grains per spike, grain protein percentage and grain yield

Table 2:	Regression coefficeint of number of grains per spike, grain protein percentage and grain yield

Table 3:	Correlation of number of grains per spike, grain protein percentage and grain yield under diffrent agro ecological conditions

, * Significant at 0.05, 0.001

Testing at more locations thus seems more effective than testing over more years for evaluation of relative performance of these characters. Large σ²_gy as relative performance of these characters. Large σ²_gy as compared to σ²_gl in case of grains per spike indicates that performance of this character is more inconsistent across the years than locations. However, as suggested by Saeed and Francis (1984), it would be possible to replace higher number of environments through different agronomic practices such as early and late planting at various locations for an additional year of testing without decreasing the precision of estimating a genotype (Table 2).

Positive association of regression coefficient (b_i) with number of grains per spike (r = 0.474), grain protein percentage (r = 0.358) and grain yield (r = 0.712) revealed that genotypes having more number of grains, high grain protein percentage and high grain yield were more sensitive to changing environments (Fig. 1, 2, 3) as suggested by Faris et al. (1981). Hence, consistency in the performance of these characters may by in those genotypes possessing relatively less number of grains per spike, less protein percentage and low grain yield.


Fig. 1:	Relationship of genotype adaptation (regression coefficient) and number of grains/spike


Fig. 2:	Relationship of genotype adaptation (regression coefficient) and grain protein percentage


Fig. 3:	Relationship of genotype adaptation (regression coefficient) and grain yield

The variation in the relationship of number of grains per spike, grain protein percentage and grain yield at different location (Table 3) reflected significant environmental effect on the performance of the genotypes. Significantly positive association of grains per spike with grain yield at eight out of nine locations indicates that grain yield can be increased by increasing number of grains per spike under most of the environments. Camargo (1987), Li (1989) and Mikheev (1992) also observed similar results.

Significantly negative (P<0.05) association between grains per spike and grain protein percentage at one location and non-significantly negative at six out of nine locations (Table 3) shows that increase in number of grains per spike may decrease grain protein percentage under certain environments. Mahmood and Shahid (1993) and Wells (1984) have reported similar results. As number of grains per spike contribute positively to grain yield therefore, efforts to increase to yield on the basis of number of grains per spike, might result in reduction in grain protein percentage under certain environments which is clear from the significantly negative relationship between grain yield and grain protein percentage at two locations and non-significantly negative at six locations. In previous research, Campana and Semp (1984), Levy and Feldman (1989) and Millet et al. (1988) also recorded inverse relationship between yield and grain protein percentage. The results suggest that due to the inconsistent relationship of these characters, selection for the improvement in grain yield on the basis of number of grains per spike should not be exercised as a routine procedure as it may reduce grain protein percentage. However, non-significant relationship of grain protein percentage with number of grains per spike and grain yield at certain locations is an encouraging fact. This reality and the selection of plants with increase source potential give hope that the plant breeder will be able to minimize the inverse yield-protein percentage relationship in the development of future high yielding wheat cultivars. In this context, It would be desirable to explore the effects of agro-ecological factors such as temperature, precipitation, soil fertility and their interactions on the development of these characters. Such information would help build understanding regarding the nature of their associations and its exploitation in breeding programmes.