INTRODUCTION

The cowpea [Vigna unguiculata (L.) Walp], due to its nutritional potential together with its ease of production and accessibility, is the most important food legume, especially for populations in the north and northeast of Brazil, which have been characterised historically by protein-energy and mineral deficiencies (^{BEZERRA et al., 2014}).

Modern cultivars, with a higher potential for grain yield and responsive to cropping technology, have helped expand the cowpea to areas of the Cerrado and the Central-West region, as the main or off-season crop. However, expression of the production potential of the crop depends on the favourable combination of a set of factors, noteworthy among them being population density, which directly influences the use of technological, environmental and management resources.

Leaf-area development is fundamental to the interception of light, and is strongly related to physiological processes, such as the addition, expansion and senescence of the leaves, and consequently the accumulation of biomass (^{RICAURTE et al., 2016}). According to ^{Oroka and Omoregie (2007)}, sufficient increases in the population density of the cowpea can significantly enhance the interception of solar radiation and efficiency in its use. ^{Mendes et al. (2005)}, working with cowpea, found that the percentage of intercepted light and the leaf area index increased by 50% and 206.5% respectively, comparing populations of from 41,666 to 166,666 plants ha^-1.

Determining the ideal plant density for a minimum of intraspecific competition is essential to optimise the use of water and nutrients per unit area, resulting in increased productivity (^{EWANSIHA et al., 2015}; ^{HELMY; HASSAN; IBRAHIMET, 2015}).

General recommendations for spacing, without considering the different characteristics of cultivars and production systems, may contribute to low grain yields in the cowpea (^{KAMARA et al., 2014}). Thus, with the development of modern cultivars for technified and/or traditional systems, there is a need for more precise information about imputed changes in the morphological and productive characteristics of these cultivars when submitted to different planting densities.

The aim of this study was to evaluate the morpho-physiological and productive characteristics of four modern, semi-erect cultivars of the cowpea under five planting densities.

MATERIAL AND METHODS

The experiment was carried out from May to August 2013 in the experimental area of Embrapa Meio-Norte in Teresina, in the State of Piauí, Brazil (05°02'17'' S and 42°47'83'' W, at an altitude of 75 m), using a sprinkler irrigation system with at average application rate of 9 mm and irrigation frequency of two days. The climate of the region, according to ^{Thornthwaite and Mather (1955)}, is C1sA'a', characterised as dry sub-humid, megathermal, with a moderate water surplus in the summer.

The soil of the experimental area is classified as a Dystrophic Red-Yellow Argisol (^{SANTOS et al., 2013}) and presented the following chemical attributes in the 0-20 cm layer: pH (CaCl₂) 5.9; MO, 9.2 g kg^-1; P, 12.6 mg dm^-3; K, 117 mg dm^-3; Ca, 2.62 cmol_c dm^-3; Mg, 0.65 cmol_c dm^-3; Al, 0.02 cmol_c dm^-3; H+Al, 2.06 cmol_c dm^-3; CEC 2,92 cmol_c dm^-3 and base saturation of 2.94 cmol_c dm^-3. The experimental area was prepared by harrowing twice and levelling, with no chemical fertiliser being applied.

Five plant populations (10⁵, 2x10⁵, 3x10⁵, 4x10⁵ and 5x10⁵ plants ha^-1) were evaluated, corresponding respectively to 5, 10, 15, 20 and 25 plants m^-1, and four improved varieties (BRS Guariba, BRS Potengi, BRS Novaera and BRS Tumucumaque), whose main characteristics are shown in Table 1.

A randomised complete block design was used in a 5 x 4 factorial scheme, with four replications. One experimental plot of 12.5 m² consisted of five rows of 5.0 meters in length, spaced 0.5 m apart, for a working area of 6.0 m².

Manual sowing was carried out in holes, using labelled string for the pre-established plant populations. Each hole was 2.0 to 3.0 cm in depth, and contained two seeds to ensure the initial stand. Thinning was performed 15 days after planting, when the surplus plants were cut below the cotyledon node, to avoid sprouting and damage to the root system of the remaining plant.

The control of invasive plants was by the pre-emergent application of the herbicide S-Metolachlor (960 g L^-1), at a dose of 1.25 L of commercial product per hectare. Insect pests were controlled with Thiamethoxam (250 g kg^-1), at a dose of 150 g of commercial product per hectare, for whitefly (Bemisia tabaci race B), thrips (Frankliniella schultzei), and the green leafhopper (Empoasca kraemeri); for control of the cotton aphid (Aphis gossypii), dimethoate (400 g L^-1), at a dose of 600 mL of commercial product per hectare, was used.

The following characteristics were evaluated: primary branch length (PBL, in cm), distance from the root collar to the apex of the main stem; number of lateral branches (NLB), count of the primary lateral branches; leaf area index (LAI), ratio of plant leaf area to the area of ground occupied by the plant; light interception (LI, as a %), reading carried out between 1100 and 1300, using an LDR-225 digital lux meter; crop growth rate (CGR, in g m^-2 day^-1 - Equation 1); net assimilation Rate (NAR, in g m^-2 day^-1 - Equation 2 ); 100-grain weight (HGW, in g), grain randomly chosen per plot, with 13% moisture; and grain yield (GY), total grain yield in the working area of the plot, with 13% moisture, converted to kg ha^-1.

where: MS₁ and MS₂ - the mean value for shoot dry matter of four plants per plot at 25 and 35 DAS respectively; S - area of ground occupied by the plant; ln AF₁ and ln AF₂ - Neperian logarithm of the leaf area (LA) at 25 and 35 DAS respectively; and t₁ and t₂ - time in days corresponding to 25 and 35 DAS respectively.

The data were submitted to analysis of variance (p<0.05) with polynomial regression for plant population, and Tukey's test (p<0.05) to compare the mean values for the cultivars. For the statistical analysis, the ASSISTAT 7.6 Beta software (^{SILVA; AZEVEDO, 2009}) was used.

RESULTS AND DISCUSSION

In relation to the cultivars, significance was found by F-test for PBL, NLB, HGW and GY only, showing variability between the cultivars relative to these variables. There was no interaction between factors for the variables under analysis, demonstrating that the four cultivars respond similarly to the effects of the different populations (Table 2).

The cultivar BRS Guariba had the highest value for PBL (70.9 cm) and the lowest for HGW (19.0 g) and GY (981.0 kg ha^-1). The highest value for HGW (23.0 g) and GY (1,240.0 kg ha^-1) were obtained with the cultivar BRS Novaera, which did not differ statistically (p>0.05) from BRS Potengi, where GY was 1,230 kg ha^-1 (Table 3). High values for GY, and large grains (HGW ≥ 20 g), meet the requirements of producers and the preferences of the domestic and especially the international market, thus increasing the export potential of the product.

The regression analysis showed a quadratic fit for PBL and NLB in response to the increases in plant population (Figure 1). PBL had a maximum of 58.5 cm for 300 thousand plants ha^-1, corresponding to an increase of 11.5% in PBL when compared to 100 thousand plants ha^-1. However, there was a 91.2% reduction in NLB when compared to the populations of 100 and 500 thousand plants ha^-1, with 82.4% of this reduction occurring in the range of 100 to 300 thousand plants ha^-1. ^{Kumar et al. (2012)} and ^{Helmy, Hassan and Ibrahimet (2015)} obtained greater lengths for the primary branches and a smaller number of lateral branches in cowpea cultivars when subjected to higher plant densities. ^{Nwofia, Nwanebu and Agbo (2012)}, studying populations in 12 varieties of cowpea, concluded that PBL shows a significant and negative correlation with grain yield.

Figure 1 Primary branch length (PBL, in cm) and number of lateral branches (NLB) in four semi-erect cultivars of cowpea [Vigna unguiculata (L) Walp] submitted to five plant populations 

A drastic reduction in NLB due to the increase in plant population has a negative impact on leaf area, impairing photosynthetic capacity and grain production per plant, considering that the greatest number of pods per plant in these cultivars are found on the lateral branches. According to ^{Choudhary and Bordovsky (2001)}, the number of pods per plant is an important regulator of grain production. ^{Mendes et al. (2005)}, ^{Kumar et al. (2012)}, ^{Bezerra et al. (2009}, ²⁰¹⁴⁾ and ^{Rezaeei et al. (2013)} saw reductions in NLB in cowpea cultivars submitted to higher plant densities.

PBL and NLB are important in determining plant architecture and have an influence on the lodging resistance and yield of cultivars. Thus, the response of the cultivars when at higher densities in relation to these two variables can directly influence their potential adoption in technified systems, which use higher densities than those recommended.

When the populations of 100 and 500 thousand plants ha^-1 were compared, light interception (LI) by the canopy increased from 66.4% to 82.9%, and LAI from 0.75 to 2.16 (Figure 2). It can be seen that for this population range, the increases in LAI (188%) were much more expressive than the increase in LI (24.9%), showing that in the largest populations, only 13.2% of the increase in LAI was converted into increments in light interception by the canopy. This result can be explained by the higher level of self-shading that occurred in the larger populations, which can reduce net photosynthesis and consequently dry-matter accumulation.

Figure 2 Light interception (LI) and leaf area index (LAI) in four semi-erect cultivars of cowpea [Vigna unguiculata (L) Walp] submitted to five plant populations 

Greater light interception at the larger populations may favour greater control of invasive plants and less evaporation of available water, especially in the most superficial layer of the soil, and this joint effect may help to boost productivity in the cultivars.

The capacity to produce dry matter from a crop under satisfactory conditions will ultimately depend on the degree of solar radiation utilised (^{BEZERRA et al., 2014}); according to ^{Taiz and Zeiger (2009)}, only 5% of the incident solar radiation is converted into carbohydrates by the leaf. ^{Mendes et al. (2005)} and ^{Kamara et al. (2016)}, working with different plant densities in the cowpea, found significant increases in LI and LAI with increased plant populations per hectare. ^{Kuapata, Hall and Madore (1990)}, found differences in light interception between two cowpea cultivars, however, higher percentage light interception was seen at the higher plant densities. ^{Ricaurte et al. (2016)}, working with three cultivars of the common bean, obtained values for light interception close to 100% and less than 70% for each cultivar at 50 DAS, for densities of 35 and 5 plants m^-2 respectively.

For CGR, there was adjustment of the linear model, with an increase of from 7.6 to 12.2 g cm² day^-1, and of the quadratic model for NAR, which was reduced from 6.7 to 4.7 g cm² day^-1, considering the populations of 100 and 500 thousand plants ha^-1, representing a 60.5% increase in CGR and a 29.9% reduction in NAR (Figure 3).

Figure 3 Crop growth rate (CGR) and net assimilation rate (NAR) in four semi-erect cultivars of cowpea [Vigna unguiculata (L) Walp] submitted to five plant populations 

In the range of 100 to 300 thousand plants ha^-1, NAR was reduced by 29.9%, but remained stable in the range of 300 to 500 thousand plants ha^-1. The increases of 188% obtained in the LAI in the range of 100 to 500 thousand plants ha^-1 explain the linear growth of the CGR as being due to the greater production of leaf area; these increases also explain the decreases seen in NAR, especially in the range of 100 to 300 thousand plants ha^-1, due to the consequent self-shading, which was intensified with the increase of LAI in the largest plant populations.

There was no adjustment (p<0.05) of the tested models for HGW or GY. ^{Bezerra et al. (2009}, ²⁰¹⁴⁾ and ^{Bisikwa et al. (2014)}, in studies of cowpea populations, found similar results for HGW.

The independence of GY in relation to the increasing plant densities shows that the positive effects of the rises in plant population on LI, LAI and CGR may have offset the negative effects on NLB and NAR, thereby making it possible to maintain the productivity of the cultivars.

CONCLUSIONS