Amaranths (Amaranthus sp.) include a group of versatile food crops exhibiting high adaptability to new environments, even in the presence of different biotic and abiotic stresses (Rana et al., 2007). Amaranths are dicotyledonous herbaceous plants including approximately 70 species, of which seventeen produce edible leaves and three produce food grains (Jansen, 2004). The edible amaranth is a popular leafy vegetable in the South-East Asia and is becoming increasingly popular in the rest of the continent and elsewhere due to its attractive leaf color, taste and nutritional value. The genus has also been reported to have anticancer properties (Dusgupta & De, 2007). Unlike other leafy vegetables, amaranth does not require a cold climate and can be cultivated during mild summers (Singh & Whitehead, 1996). It can also grow successfully under varied soil and agro-climatic conditions (Katiyar et al., 2000; Shukla & Singh, 2000). Amaranth leaves are a rich and inexpensive source of dietary fibre, proteins, vitamins and a wide range of minerals (Prakash & Pal, 1991; Shukla et al., 2003; Routray et al., 2012; Venskutonis & Kraujalis; 2013). They serve as an alternative source of nutrition for vegetarian people in developing countries where the bulk of the population has little access to protein-rich food.

The species Amaranthus tricolor L. grown as leafy vegetable is loosely termed as vegetable amaranth; it is a self-pollinated C4 crop with wide genetic diversity and phenotypic plasticity (Rajan & Markose, 2007). In Bangladesh, A. tricolor is grown year-round and it is the only crop available in the hot summer months when no other foliage crop grows in the field. The species used as vegetable has short plants with large smooth leaves, small auxiliary inflorescences, and succulent stems. In this country, we have found in this plant a high variability respect to antioxidant, yield and yield-related traits (Sarker et al., 2014). Generally the success of any crop improvement program largely depends on the magnitude of genetic variability, heritability, genetic advance, and character association. Genetic variability is important for selection of parents with transgressive segregation (Patro & Ravisankar, 2004). Heritability estimates provide information on the proportion of phenotypic variance that is due to genetic factors for different traits, but these estimates alone are not a sufficient measure of the level of possible genetic progress. Effective selection can be made when the value of broad sense heritability estimates is considered together with the selection differential or genetic advance (Ibrahim & Hussein, 2006). Information on the amount and direction of association between yield and yield-related characteristics is important for rapid progress in selection and genetic improvement of a crop (Asish et al., 2008). Correlations between two or more plant characters and yield provide suitable means for indirect selection for yield.

Extensive research efforts have been carried out to ascertain the mineral composition of vegetable amaranth. Although some reports on its nutritional aspects are available (Shukla et al., 2003, 2005, 2006), there are few works on mineral composition of leaves along with qualitative improvement of foliage with special reference to leaf attributes (Wu-Leung et al., 1968; Freiberger et al., 1998). So, the present investigation was carried out (i) to estimate quality, biological yield and composition of minerals in 43 different cultivated genotypes of vegetable amaranth available in Bangladesh, and (ii) to find out possible ways for improvement of protein, dietary fibre, K, Ca and Mg compositions without compromising biological yield.

The experiment was conducted at the experimental field of Bangabandhu Sheikh Mujibur Rahman Agricultural University, Bangladesh. The experimental site was located in the centre of the Madhupur Tract (AEZ-28), about 24°23´N 90°08´E, with a mean elevation of 8.4 masl. The experimental field was a high land having silty clay soil. The soil was slightly acidic (pH 6.4) and low in organic matter (0.87%), total N (0.09%) and exchangeable K (0.13 cmol/kg). The site falls under the subtropical zone and has mean temperatures of 29°C (summer) and 18°C (winter).

In a preliminary study, we collected 122 genotypes of vegetable amaranth (Amaranthus tricolor L.) from different eco-geographic regions of Bangladesh (unpublished data). From these, 43 accessions were selected due to their high yield potential as well as quality aspect; they were locally well adapted and cFultivated as varieties by local farmers.

The genotypes were sown in a randomized complete block design (RCBD) with five replications, during three successive years (2012, 2013 and 2014). Each accession was sown in two unit plots, one of 1 m² for the biological yield and other of 0.6 m² for the mineral, quality and agronomic traits study. The spacing was 20 cm from row-to-row and 5 cm from plant-to-plant, respectively. Total compost (10 ton/ha) was applied during land preparation. Urea, triple super phosphate, murate of potash and gypsum were applied at 200, 100, 150 and 30 kg/ha, respectively. Appropriate cultural practices were also maintained. Thinning was done to maintain appropriate plant density within rows. Weeding and hoeing was done at 7 days interval. Day temperature during experimental period ranged from 25 to 38°C. Irrigation (by narrow hose pipe from underground water) was provided at 5-7 days interval. To record the data on biological yield, plants were cut at the base of the stem (base of ground-level).

Data were collected at 30 days after seed sowing, on 10 randomly selected plants in each replication for four agronomic traits such as leaf area (cm²), shoot weight (g), shoot/root weight and stem base diameter (cm). Biological yield was recorded on whole plot basis. Beside this, content percentages of three minerals, K, Ca and Mg and of protein and dietary fibre, were estimated.

Mean performance, coefficient of variation (CV, %) and critical difference (CD) for mineral content, quality and agronomic traits and biological yield for 43 vegetable amaranth genotypes are presented in Table 1. The analysis of variance revealed significant differences among the genotypes for the ten traits studied, indicating the validity of further statistical analysis.

Variability plays a vital role for the selection of superior genotypes in crop improvement programs. Agronomic traits are quantitative in nature, and interact with environment under study, so partitioning the traits into genotypic, phenotypic, and environmental effects is essential to find out the additive or heritable portion of variability. In the present investigation, biological yield, leaf area, shoot/root weight, shoot weight and dietary fibre content had high to moderate genotypic and phenotypic variances along with GCV and PCV values, which indicate scope for improvement in these traits through selection due to predominance of additive gene action for these traits. Variability alone is not of much help in determining the heritable portion of variation. The amount of gain expected from a selection depends on heritability and genetic advance in a trait. Heritability has been widely used to assess the degree to which a character may be transmitted from parent to offspring. Knowledge of heritability of a character is important as it indicates the possibility and extent to which improvement is possible through selection (Robinson et al., 1949). However, high heritability alone is not enough to make sufficient improvement through selection generally in advance generations unless accompanied by substantial amount of genetic advance (Johnson et al., 1955b). The expected genetic advance is a function of selection intensity, phenotypic variance and heritability and measures the differences between the mean genotypic values of the original population from which the progeny is selected. It has been emphasized that genetic gain should be considered along with heritability in coherent selection breeding program (Shukla et al., 2006). It is considered that if a trait is governed by non-additive gene action it may give high heritability but low genetic advance, which limits the scope for improvement through selection, whereas if it is governed by additive gene action, heritability and genetic advance would be high, consequently substantial gain can be achieved through selection. In the present study the heritability and genetic advance values were high for all the traits, indicating preponderance of additive gene effects. The genotypic correlation coefficients were very much close to the corresponding phenotypic values for all the traits indicating additive type of gene action for the expression of these traits. Insignificant genotypic correlation was observed among mineral, quality and agronomic traits and biological yield, except K vs. Mg (0.753), protein vs. dietary fibre (–0.295), and stem base diameter vs. Ca (–0.491). This indicates that selection for high mineral, protein and dietary fibre content might be possible without compromising yield loss. On the other hand, most of the interrelationships among different agronomic traits were significant. Similar trend was observed by earlier works in A. tricolor (Shukla et al., 2006; Sarker et al., 2014). Biological yield had significant positive correlation with leaf area (0.326), shoot weight (0.999), shoot/root weight (0.454) and stem base diameter (0.368), indicating that biological yield of vegetable amaranth could be increased with the increase of leaf area, shoot weight, shoot/root weight and stem base diameter. Sarker et al. (2014) observed that foliage yield was highly associated with plant height, leaf area, leaves/plant stem base diameter and dietary fibre content. Similarly, Shukla et al. (2010) observed a positive association of foliage yield with beta carotene and ascorbic acid. Stem base diameter had a significant positive association with leaf area (0.597), and shoot weight (0.365), whereas these traits showed significant negative association with Ca (-0.491). Shoot/root weight exhibited significant positive interrelationship with shoot weight (0.454) indicating that plant with thick stem contained less Ca, more leaves and shoot weight. Significant positive association was observed between shoot weight and leaf area (0.326).

Considering high genotypic and phenotypic variances along with GCV and PCV values, high heritability coupled with GAMP, five traits (leaf area, shoot/root weight, shoot weight, dietary fibre content and biological yield) could be selected for the improvement of 43 vegetable amaranth genotypes under study. However, the correlation study revealed strong positive association of leaf area, shoot weight, shoot/root weight and stem base diameter with biological yield. Selection based on leaf area, shoot weight, shoot/root weight and stem base diameter could lead to increase the biological yield of vegetable amaranth genotypes. Insignificant genotypic correlation was observed among mineral, quality and agronomic traits except K vs. Mg (0.753), protein vs. dietary fibre (–0.295) and stem base diameter vs. Ca (–0.491) indicated that selection for high mineral, protein and dietary fibre content might be possible without compromising yield loss. Based on mean performance of the genotypes, six vegetable amaranth genotypes VA16, VA27, VA26, VA14, VA18 and, VA13 were identified as high yielding having substantial mineral, protein and dietary fibre content.