AGRONOMIC PERFORMANCE OF COWPEA ELITE LINES IN THE STATES OF MINAS GERAIS AND MATO GROSSO , BRAZIL 1

The objective of this work was to evaluate the agronomic performance of erect and semi-erect cowpea elite lines grown in the States of Minas Gerais and Mato Grosso, Brazil. The experiments were conducted in Jaíba, Janaúba, Sete Lagoas (Minas Gerais), Primavera do Leste and Nova Ubiratã (Mato Grosso), with 20 genotypes of cowpea, arranged in a randomized block design with four replications. The data were subjected to individual and joint analyses of variance and means grouped by the Scott-Knott test (p<0.05). The plant growth habit, lodging, cultivation value, pod length and bean yield were evaluated. The lines MNC04768F-21 and MNC04F-795F-168 had the highest bean yield in the experiments conducted in Mato Grosso, and their growth habit, lodging and cultivation value were similar to those of the cultivar BRS-Guariba. The lines MNC04-769F-30, MNC05-795F-154 and MNC04-769F-49 and the cultivar BRS-Tumucumaque had the highest bean yield in the experiments conducted in Minas Gerais. However, the cultivars BRS-Guariba, BRSTumucumaque, BRS-Novaera, BRS-Itaim and BRS-Cauamé and the lines MNC04-792F-143 and MNC04792F-144 showed satisfactory bean yield and good growth habit, plant lodging and cultivation value in the State of Minas Gerais, especially in the northern region of the state (Jaíba and Janaúba), representing good alternatives for the genotypes recommended for this region.


INTRODUCTION
The world planted area of cowpea is 10.4 million hectares, which is located mainly in tropical and subtropical regions of America, Asia and Africa; the world production of cowpea beans is about 5.5 million Mg, and Nigeria is the world's largest producer (FAO, 2016).
Cowpea is mainly grown in Brazil by small farms using low technological level, thus, the average national crop yield is very low, approximately 401 kg ha -1 (EMBRAPA, 2016).However, the demand for cultivars with appropriate characteristics to planting and harvesting mechanization is increasing due to the interest of more producers in this crop (FREIRE FILHO, 2011).Thus, researches have sought to develop genotypes of more erect growth habit, compact architecture and more adapted to mechanization, that result in shorter maturation cycle, greater plant densities and higher yields.
The Northeast of Brazil is the main producing region of cowpea, however, its production area has expanded to the Brazilian Southeast and Center-West regions, especially to the State of Mato Grosso, where it is produced on approximately 140,000 hectares (CONAB, 2016), focused mainly in international markets.Cowpea crops are conducted with high technology in these new areas, with pest and disease control, balanced soil fertilization and mechanized harvesting.The northern State of Minas Gerais (Brazilian Southeast region) has an expressive consumer market and production of cowpea, however, this production is limited by several factors, especially by the low technology level used and current cultivars recommended for planting in this region, which are based on results of researches carried out in other regions of the country.
In this context, the objective of this work was to evaluate the agronomic performance of erect and semi-erect cowpea elite lines in the States of Minas Gerais and Mato Grosso, Brazil.

MATERIAL AND METHODS
The experiments were conducted in Jaíba, Janaúba and Sete Lagoas in the State of Minas Gerais (MG) and in Primavera do Leste and Nova Ubiratã, in the State of Mato Grosso (MT), in 2013 (Jaíba, Primavera do Leste and Nova Ubiratã) and 2014 (Janaúba and Sete Lagoas) during the summer-fall crop season, with planting in February to March.
These locations are considerable different in edaphoclimatic conditions; they include the Cerrado, Cerrado-Amazon and Cerrado-Caatinga biomes.
The experiments were conducted with 20 erect and semi-erect cowpea genotypes, consisted of 15 elite lines and five cultivars (controls).These genotypes were selected by the Cowpea Breeding Program of the Embrapa Meio-Norte and subjected to cultivation and use value experiments in 2013 to 2014 (Table 1).
A randomized block experimental design with four replications was used, with plots consisted of four 5-meter rows, spaced 0.50 m apart, totaling an area of 10 m 2 , and evaluation area of 5 m 2 (two central rows).
Conventional system of soil preparation was used in all locations evaluated, with one plowing and two harrowings at pre-planting.The soils of the areas were then furrowed and fertilized with a fertilizer equipment with spacing and rates adjusted as required.Sowing was carried out with manual seeders.The recommendations of Melo, Cardoso and Salviano (2005) for fertilization of cowpea crops was followed.
Approximately 15 seeds per meter were sown and after their germination a thinning was carried out, leaving approximately ten plants per meter.Manual weeding with hoes was performed from the 20 th to the 30 th day after emergence.
Irrigation was performed by a conventional sprinkler system from planting to the beginning of maturation in Jaíba, Janaúba and Sete Lagoas.The experiment in Nova Ubiratã and Primavera do Leste were conducted in the rainfed system.Harvesting was carried out manually, according to the maturity of each genotype.
Plant growth habit, lodging, and cultivation value, which was visually attributed according to a scale of grades (Table 2), were evaluated at the physiological maturity stage.Pod length and bean yield were evaluated after harvesting.Twenty pods, randomly chosen in each plot, was measured during the harvesting to assess the mean pod length.The beans harvested in the evaluation area of each plot were weighed to estimate the bean yield, correcting their weights for beans with 13% of moisture and transforming their results to kg ha -1 .The data were subjected to individual and joint analyzes of variance.The homogeneity of the residual variances was evaluated for the joint analysis.It was found that the ratio between the highest and The mean square of the residue was greater than Seven, indicating that the residual variances were homogeneous (PIMENTEL-GOMES, 2009).Like this, adjustment of the degrees of freedom of the Mean and GxE interaction, according to the method of Cochran (1954).
The relevance of each source of variation in the total variation of the mean was estimated by the ratio between the sum of squares of the source of variation i (SQ i ) and the total sum of squares (SQ t ), thus defining the coefficient of determination (R 2 ).The means of the genotype were grouped by the Scott-Knott test (p<0.05).All analyses were performed using the program GENES (CRUZ, 2006).

RESULTS AND DISCUSSION
The analysis of variance showed a significant effect of the genotype (G) and environment (E) on all characteristics evaluated.The genotype x environment interaction (GxE) affected most of the characteristics evaluated, except pod length (Table 3).
According to the coefficient of determination (R²) found, the effect of the genotypes was responsible for most of the total variation of the pod length means.The environment as a source of variation was the main responsible for the variation of the growth habit and bean yield means in Janaúba, Jaíba, Sete Lagoas, Nova Ubiratã and Primavera do Leste, and the GxE caused the greatest variation in the means of plant lodging and cultivation value of the genotypes evaluated (Table 3).Table 3. Joint analysis of variance of growth habit, lodging, cultivation value, pod length and bean yield of 20 erect and semi-erect cowpea genotypes, evaluated in five environments of the States of Minas Gerais and Mato Grosso, Brazil.*= significant at 5% and ns = not significant by the F test.R 2 = Coefficient of determination (%); CV = coefficient of variation (%); DF = degrees of freedom; MS = mean square.
According to the results of the coefficient of determination, the pod length was greatly affected by the genotype and little affected by the environment.This result is due to the high heritability of the pod length, which is an important and desirable characteristic for the selection of lines (TEIXEIRA et al., 2007).
Bean yield was significantly affected by G, E and G-E, which were responsible for 38.68% (E), 31.90%(G-E) and 9.22% (E) of its variations (Table 3).These results were probably because the genetic similarity between the genotypes, since most of them have common parents (Table 1).Moreover, bean yield is a polygenic quantitative characteristic, and thus highly affected by the environment (SANTOS et al., 2014).
Growth habit is an important characteristic and its improvement is one of the main objectives of cowpea breeding researches, since it selects more adapted cultivars to mechanization, which has been adopted by producers.According to Araújo and Watt (1988), low temperatures can prolong the vegetative phase of cowpea with emission of new branches, and a greater number of trifoliate leaves, favoring the lodging of cowpea plants.Sete Lagoas had the lowest average and minimum temperatures among the environments evaluated, and the highest grades for plant lodging (Table 5), denoting the lower resistance of the genotypes to lodging in this environment.This result denotes the effect of the environment on the growth habit of the lines and the need for recommendation of cultivars based on tests in different producing regions.
Genotypes with erect growth habit are selected for cultivars of low plant lodging and more compact and erect growth habit, which are suitable to grown in high technological fully-mechanized production systems.Thus, the lines MNC04-792F-143 and MNC04-795F-155, and the cultivars BRS-Novaera and BRS-Itaim stood out with more erect growth habit and lower plant lodging in most of the environments evaluated.
The cultivation value represents the general aspects of the plant, such as architecture, pod loading, pod and bean quality and phytosanitary condition.Cultivation values of 3.0 is the minimum grade for commercial genotypes, indicating that the genotype has most of the suitable characteristics for commercial cultivation (Table 2).The cultivation values of the genotypes had significant differences in all environments evaluated (Table 6).According to the grouping test, the genotypes with the highest grades showed groups of 11 (Jaíba), 8 (Janaúba), 4 (Sete Lagoas), 6 (Nova Ubiratã) and 11 (Primavera do Leste) genotypes.
Thus, genotypes that stood out with good Plant lodging was affected by the genotypes, except in Sete Lagoas.The mean test used grouped the genotypes according to grades attributed to the lodging.Thus, genotypes were grouped into four groups in Jaíba.The group with the lowest grades consisted of nine genotypes, with maximum grade of 1.25.The cultivars BRS-Novaera and BRS-Itaim, and the line MNC04-792F-143 had the lowest grade in Janaúba, with maximum grade of 1.50.Twelve genotypes with the lowest grades were grouped in Nova Ubiratã, with maximum grade of 1.25, and 11 in Primavera do Leste, with maximum grade of 2.25.The cultivars BRS-Novaera and BRS-Itaim were in the group of genotypes that had the lowest grades for plant lodging in all environments evaluated, denoting their greater resistance to lodging compared with the other genotypes evaluated (Table 5).
The ideal pod length for erect cowpea genotypes has not yet been established.However, Freire Filho (2011) reported average pod length of 18 cm for cultivars released between 1991 and 2010.The pod length of the lines MNC04-795F-168, MNC04-769F-30, MNC04-769F-62 and MNC04-795F-159, and the cultivars BRS-Itaim, BRS-Novaera and BRS-Guariba are within this range (Table 7).Smaller pods have been preferred by producers, since they facilitate harvest mechanization, and prevent contact of the pods with the soil and consequently, their rotting.Moreover, smaller pods are generally lighter and reduce bending and breaking of the peduncle at harvesting (SILVA; NEVES, 2011).The bean yield varied in different environments with a same genotype (Table 8).(2011) stated that these differences show the presence of genetic variability among the genotypes and the effect of the environment on them.
The bean yield found in this work can be considered high in most situations, since it was well above the national average of 401 kg ha -1  (EMBRAPA, 2016).However, the national average often denotes the effect of the low technology applied to the crops, which affects bean yield.Moreover, a number of experiments (FREIRE FILHO et al., 2005;ROCHA et al., 2007;TEIXEIRA et al., 2010;VALADARES et al., 2012;BEZERRA et al., 2012;SANTOS et al., 2016) conducted in similar conditions, reported average yields between 1,015 kg ha -1 and 1,390 kg ha -1 , which are close to those found in the present work.BEZERRA, A. A. C. et al.Comportamento morfoagronômico de feijão-caupi, cv.BRS-Guariba, sob diferentes densidades de plantas.Revista de Ciências Agrárias, Recife, v. 55, n. 3, p. 184-189, 2012.COCHRAN, W. G.The combination of estimates from different experiments.Biometrics, Washington, v. 10, n. 1, p. 101-129, 1954 According to Freire Filho et al. (2005), Rocha et al. (2007) and Matos Filho et al. (2009), differences in the same genotype in different environments can also occur as a function of biotic factors and genetic response to the environmental variations.Zilio et al.

Table 1 .
Erect and semi-erect cowpea genotypes subjected to cultivation and use value experiments in 2013 to 2014 and their parental and commercial subclass.

Table 2 .
Scales of grades used to classify the growth habit, lodging and cultivation value of cowpea plants.

Table 4 .
Average grades attributed to growth habit of 20 genotypes of cowpea, evaluated in five environments of the States of Minas Gerais and Mato Grosso, Brazil.
1Means followed by the same uppercase letter in the columns (comparing genotypes) and lowercase in the lines (comparing environments) do not differ by Scott-Knott test (p<0.05).

Table 5 .
Average grades attributed to plant lodging of 20 genotypes of cowpea, evaluated in five environments of the States of Minas Gerais and Mato Grosso, Brazil, from 2013 to 2014.
1Means followed by the same uppercase letter in the columns (comparing genotypes) and lowercase in the lines (comparing environments) do not differ by Scott-Knott test (p<0.05).

Table 6 .
Average grades attributed to the cultivation value of 20 cowpea genotypes evaluated in five environments of the States of Minas Gerais and Mato Grosso, Brazil.Means followed by the same uppercase letter in the columns (comparing genotypes) and lowercase in the lines (comparing environments) do not differ by Scott-Knott test (p<0.05).

Table 7 .
Average pod length (cm) of 20 genotypes of cowpea, evaluated in five environments of the States of Minas Gerais and Mato Grosso from 2013 to 2014.
1Means followed by the same uppercase letter in the columns (comparing genotypes) and lowercase in the lines (comparing environments) do not differ by Scott-Knott test (p<0.05).