Introduction

The bell pepper (Capsicum annuum L.) is one of the most important peppers produced in Mexico. In 2014, the value of pepper exports to the United States reached US $929 million (Secretaría de Agricultura, Ganadería, Desarrollo Rural, Pesca y Alimentación [SAGARPA], 2016) and, according to SAGARPA (2012), the area sown under protected farming conditions for this crop accounts for 16 % of the total, surpassed only by the 70 % devoted to the cultivation of tomato (Solanum lycopersicum L).

Pepper is a plant of indeterminate growth, reaching at least 2 m in height (^{Jovicich, Cantliffe, & Stoffella, 2004a)}. Greenhouse yields with intermediate technology can reach up to 130 t·ha^-1 and with high technology up to 250 t·ha^-1 (^{Fundación Mexicana para la Investigación Agropecuaria y Forestal [FUMIAF], 2005}).

Two main production systems are used. The first, which is the most preferred in southern Europe, is to let the plants grow freely with all their stems, which have sympodial growth and on each branching produce flowers, usually solitary ones. The first six to eight flowers become fruit, but all demand a high amount of assimilates, so in the flowers that form later a high abortion percentage occurs. Once the first fruits finish their growth and are harvested, the availability of sugars increases and allows the vegetative growth, fruit set and growth of four to eight more fruits to continue. Between one harvest flow and another, two months elapse (^{Cruz-Huerta, Sánchez-del Castillo, Ortiz-Cereceres, & Mendoza-Castillo, 2009}; ^{Marcelis, Heuvelink, Hofman-Eijer, Bakker, & Xue, 2004}), which results in a complete crop cycle lasting from eight to ten months after transplant, yielding between 50 and 80 t·ha^-1 (^{Jurado & Nieto, 2003}).

In the second cultivation system, which was developed in countries of Northern Europe and North America, the plant is restricted to two stems. This is the most common system used in Mexico for pepper production. In this system the branches of each bifurcation are pruned leaving only the flower formed in the fork, at a density of 2 to 3 plants·m^-2, so 4 to 6 stems·m^-2 growing up to 3 m in height are managed (^{Jovicich, Cantliffe, & Vansickle, 2004b).} By limiting the number of fruits growing simultaneously, the source-sink relationship is modified, so that production can be continued during almost the whole year. The system generally requires high technology greenhouses to control the indoor environment and, although annual yields may exceed 200 t·ha^-1, the cost of production per kilogram is very high (^{Paschold & Zengerle, 2000}).

Recently, research has been conducted to develop an alternative production system, consisting of late transplanting and early blunting to stop the growth of plants above the fourth branching of their stems. The purpose is to shorten the crop cycle, by reducing the time from transplant to final harvest to less than four months and to achieve three crop cycles per year instead of only one (^{Cruz-Huerta et al., 2009}; ^{Reséndiz-Melgar, Moreno-Pérez, Sánchez-del Castillo, Rodríguez-Pérez, & Peña-Lomelí, 2010}). The lower yield obtained per plant is offset in part by managing a higher population density (6 plants·m^-2 instead of two or three as in the conventional system); therefore, the annual yield is greater since in involves the production of three cycles (^{Cruz-Huerta, Ortiz-Cereceres, Sánchez-del Castillo, & Mendoza-Castillo, 2005}). ^{Reséndiz-Melgar et al. (2010)} evaluated this system by comparing 17 varieties at two population densities and found that the most outstanding variety was Orion with 7.6 kg·m^-2 and 7.75 fruits set per plant, out of a total of 15 possible in the first four bifurcations. On the other hand, ^{Cruz-Huerta et al. (2005}, ²⁰⁰⁹) studied this system with the Ariane variety and reported that three crop cycles per year could obtain similar or even higher yields than those attained by the production system used in European countries, but with lower production costs.

The limitation on producing more in each cycle, with this production system, has been the high fruit abortion percentage, attributed to the competition between plants to intercept the incident photosynthetically active radiation (PAR) more evenly throughout the canopy, and the simultaneous growth of several fruits per plant that cause high demand for photo-assimilates that cannot be fully satisfied (^{Cruz-Huerta et al., 2009}; ^{Reséndiz-Melgar et al., 2010}). Therefore, strategies to make more efficient use of the amount of incident sunlight and the way in which it is intercepted by the leaves are sought (^{Jovicich et al., 2004a)}. One way is to appropriately manage the population density of the crop (^{Papadopoulos & Pararajasingham, 1997}). Another is to adjust the arrangement of the plants in the greenhouse by placing the rows of plants at a different height to form a stair-like canopy arrangement (^{Sánchez-del Castillo, Bastida-Cañada, Moreno-Pérez, Contreras-Magaña, & Sahagún-Castellanos, 2014}).

In tomato, several experiments have been conducted in this regard and the results show that with stair-shaped canopy systems a higher yield is obtained than when the plants are established in a uniform arrangement (^{Sánchez-del Castillo, Moreno-Pérez, & Contreras-Magaña, 2012}), and higher than the yield achieved with conventional greenhouse systems (^{Resh, 2001}; ^{Sánchez-del Castillo et al., 2012}). However, this system has not been evaluated for the cultivation of bell pepper.

Based on the above, the aim of this research was to compare four bell pepper production systems generated by two canopy arrangements (stair-shaped and uniform) and two planting densities (6 and 8 plants·m^-2). It was hypothesized that it is possible to increase the number of fruits set per plant and per unit area, and therefore the pepper crop yield, with plants blunted just above the fourth bifurcation and by the formation of stair-like canopies in order to intercept the incident PAR more homogeneously throughout the canopy.

Materials and methods

This research was carried out from March to September 2015 in a metal-frame, chapel-type greenhouse with a polyethylene cover allowing 80 % light transmission. The greenhouse is located in the Experimental Field of the Universidad Autónoma Chapingo, Texcoco, State of Mexico (19° 29’ North latitude, 98° 53’ West longitude and 2,251 masl). The greenhouse had a cooling pad, extractors and a heating system that allowed for adequate temperature control, as well as windows protected with anti-aphid mesh, roll-up curtains and white polypropylene fabric used as ground cover over the floor of the aisles.

Cannon variety was used; it is characterized by being of indeterminate growth and producing blocky-type fruits of 150 to 200 g per fruit. This variety is appreciated in the market for its thick wall, large caliber and good taste.

Sowing was done in 60-cavity polystyrene trays, depositing one seed in each cavity. The substrate for sowing was peat moss mixed with perlite at a ratio of 1:1 (v:v). In the first eight days, the irrigations applied were with water alone. Once emerged, and up to 10 days later, the plants were watered with a nutrient solution diluted to 50 % of its normal concentration. Subsequently, and until the end of the cycle, they were watered with the 100 % solution. The nutrient solution used contained the following elements and concentrations in ppm (mg·L^-1): N = 200, P = 50, K = 200, Ca = 235, Mg = 40, S = 160, Fe = 3, B = 0.5, Mn = 0.5, Cu = 0.1 and Zn = 0.1, with electrical conductivity of 2.5 dS∙m^-1 and pH between 6 and 6.5.

To supply the nutrient solution, a drip irrigation tape with emitters spaced 20 cm apart was used, placing a tape in each row. The average solution flow ranged from 3 to 6 L·m^-2·day^-1.

The transplant was carried out 50 days after sowing (das). The duration of the crop cycle, from transplant to final harvest, was 120 days. A hydroponic system consisting of cultivation beds 1 m wide and 30 cm high were used at floor level to form a uniform canopy, and a set of two lower beds and an upper middle one (40 cm high with respect to the lower ones) 30 cm wide was used to form a stair-like canopy (Figure 1). The cultivation beds were filled with red tezontle sand with particles from 1 to 4 mm in diameter. The middle bed was 40 cm above the others, for which a structure with plastic boxes supported by blocks was assembled (Figure 1).

Figure 1 Treatments with different plant arrangements: A) stair-like canopy with three rows of plants, B) stair-like canopy with four rows of plants, C) uniform canopy with three rows of plants and D) uniform canopy with four rows of plants. 

Two arrangements (plant layouts) and two population densities were evaluated. A split-plot randomized complete block experimental design was used with four replicates. The canopy arrangement, stepped and uniform, was in the large plots, while in the subplots the population densities were tested: 6 and 8 plants·m^-2 for each type of arrangement. The small-plot experimental unit was nine plants. The arrangement of plants in combination with the densities is described below:

The plants developed without pruning until reaching the fourth branching, at which time the growth apex was removed (blunting). This practice was performed between 50 and 55 days after transplant (dat). The flowering of the first branching was removed from all plants. The training consisted of supporting the plant with polypropylene (raffia) threads tied to wires that ran along the greenhouse structure at 2 m high.

The characters evaluated were:

Analyses of variance and Tukey's range test (P ( 0.05) were performed using the ^{Statistical Analysis System (SAS, 2002)} package.

Results and discussion

The analyses of variance of the yield variables and components (Table 1) indicated significance (P ( 0.01) between the canopy arrangements, except for mean fruit weight. By contrast, densities did not affect the variables evaluated. The interaction between canopy arrangements and population densities was significant only for mean fruit weight.

Despite having tested four replicates and four treatments, it was possible to identify statistical differences in fruit yield for the arrangement factor (housed in large plot). However, in planting densities (small plot), there were no significant variations in this character. These results can be considered reliable against the coefficients of variation, less than 16 %, despite the reduced number of degrees of freedom of error b, a situation that can be attributed to the large experimental plot used, being a factor which favors the precision of the effect estimates (^{Steel, Torrie, & Dykey, 1997}).

Table 2 shows that the yield obtained per plant and per unit area were lower with the uniform canopy arrangement, while with stair-like canopies the yield per unit area increased by an average of 32 %. Similar results were reported in tomato (^{Sánchez-del Castillo, Moreno-Pérez, Coatzín-Ramírez, Colinas-León, & Peña-Lomelí, 2010}; ^{Sánchez-del Castillo, Moreno-Pérez, & Cruz-Arellanes 2009}; ^{Vázquez-Rodríguez, Sánchez-del Castillo, & Moreno-Pérez, 2007}), by observing that in stair-like canopies an increase in yield per unit area was achieved with respect to the uniform canopy.

Table 3 shows that the arrangement with lower density had a significantly higher yield per plant, as a consequence of greater fruit set per plant and higher mean fruit weight. However, in number of fruits per m² and yield per m², there were no significant statistical differences, indicating that the density of 6 plants·m^-2 can be used commercially for a greater yield, as already indicated by ^{Reséndiz-Melgar et al. (2010)}. The final plant dry weight was also higher with the lowest population density.

Table 4 shows that in both densities the yield and number of fruits per unit area and per plant were higher in the stair-like canopy, compared to the uniform canopy. These differences were 3.5 kg·m^-2 in yield and 22 fruits∙m^-2. It can also be seen that the significant interaction between arrangements and densities indicated in the mean fruit weight was due to the fact that in the uniform arrangement this character decreased significantly when increasing from 6 to 8 plants·m^-2. By contrast, in the stair-like arrangement the fruit weight was statistically similar at both densities.

The results of yield and number of fruits obtained with the uniform canopy arrangement are consistent with those reported by ^{Reséndiz-Melgar et al. (2010)}, with the same bell pepper production system managed at high density and blunting above the fourth bifurcation. ^{Reséndiz-Melgar et al. (2010)} obtained yields of 7 kg·m^-2 in a four-month growing cycle. They point out that the limitation on yield per unit area was the high fruit abortion percentage per plant (approximately 50 %). On the other hand, ^{Cruz-Huerta et al. (2009)} reported similar fruit abortion percentages with high population density management, by performing the blunting after either the third or fourth bifurcation. In both papers, they attribute this to a strong demand for assimilates by the first fruits in growth, coupled with limitations in the interception and homogeneous distribution of PAR in the canopy caused by competition among plants.

Solar radiation is the most important factor by which plants compete in an environment without restrictions, as is the case in greenhouse cultivation and hydroponics; for this reason, several cultural practices carried out under these conditions seek to increase PAR interception and improve its distribution among the canopy leaves (^{Jovicich et al., 2004a}; ^{Papadopoulos & Pararajasingham, 1997}). The results obtained in the present research are similar to those reported by ^{Jolliffe and Gaye (1995)}, who mention that, up to a certain limit, increasing population density in a favorable environment, such as that of a greenhouse, increases yield per unit area. The lack of increase in yield with increasing population density is explained by the fact that the mutual shading between plants negatively affects the production of photo-assimilates, to the degree that the number of fruits per plant or the mean weight of each fruit decreases drastically (^{Heuvelink, 1995}; ^{Villegas-Cota et al., 2004}).

With the arrangement of plants in the stair-like system managed at 6 plants·m^-2, it was possible to significantly increase the number of fruits and the yield per plant compared to the uniform canopy treatments; this indicates that with the stair-like arrangement, it was possible to increase the amount of PAR intercepted per plant and, above all, improve its distribution among the canopy leaves, thereby resulting in more efficient photosynthesis.

The density increase from 6 to 8 plants∙m^-2 in the stair-like arrangement did not increase yield per unit area, mainly because the number of fruit set per plant decreased significantly, indicating that the low density is the most suitable for commercial management.

Numerous studies carried out in tomato with plants blunted at three clusters in stair-like canopy arrangements (^{Méndez-Galicia, Sánchez-del Castillo, Sahagún-Castellanos, & Contreras-Magaña, 2005}; ^{Sánchez-del Castillo et al., 2014}, ²⁰¹² and ²⁰¹⁰; ^{Vázquez-Rodríguez et al., 2007}) also show significant increases in yield per unit area, thanks to the fact that this arrangement has allowed increasing the population density with respect to uniform canopies without a significant decrease in mean fruit size and weight.

To understand these aspects in greater detail, an analysis was made of the behavior of plant rows within each treatment (Tables 5, 6, 7 and 8). The comparison of means for the stair-like treatment with 6 plants·m^-2 (Table 5) shows that the three rows of plants behaved similarly in the variables yield, fruits per plant, mean fruit weight and dry weight per plant at the end of the cycle, bolstering the argument that there is similar PAR interception and distribution in each row.

The behavior of the rows within the stair-like treatment with a density of 8 plants·m^-2, by placing four rows of plants instead of three (Figure 1B and Table 6), was different. It was observed that the yield per plant was 33 % lower in the two rows that occupied the lower level with respect to the higher level. This was caused by a lower number of fruits per plant, since the mean fruit weight was similar in all rows. The dry weight per plant also decreased in the lower two rows.

By increasing the population density in the same area, the intercepted PAR per unit area is distributed among more plants, with a corresponding decrease for each one, which affects its dry matter production and final yield; In addition, in the particular case of pepper, it affects the number of fruits that each plant can keep growing (^{Heuvelink, 1995}; ^{Wien, 1999}). These results agree with those of ^{McAvoy et al. (1989)}, in the sense that there is a high relationship between fruit yield, total PAR intercepted by the canopy and its distribution in each plant, particularly during anthesis prior to harvest, which is when the population density pressure is higher because of the greater leaf area that is formed.

The plants that were in the lower level had a greater shade effect; therefore, they were the most affected in their interception of PAR per day, since the lower row oriented towards the West received less PAR in the mornings and the East row in the afternoons. On the other hand, the comparison of rows with a uniform arrangement and 6 plants·m^-2 (Table 7) shows that the three rows of plants had a similar behavior for all variables, which is attributed to fact that the spacing between rows (33 cm) is sufficient to achieve a similar interception and distribution of PAR, including the middle row.

^{Gardner, Pearce, and Mitchell (1990)}, ^{Heuvelink (1995)}, and ^{Sánchez-del Castillo et al. (2014)} indicate that the more homogeneous distribution of PAR among the leaves of the plants that make up the canopy results in an increase in the rate of photosynthesis. The decrease in yield per plant and number of fruits per unit area with respect to the rows of plants arranged in a stair shape at the same density (Table 1) can be explained because, although in both canopies the interception of PAR per plant might be similar, the arrangement of rows allowed a better distribution of PAR among the canopy leaves compared to the uniform arrangement. This is due to the fact that PAR had more impact on the upper leaves than on the lower ones, making the photosynthetic rate less efficient.

The test of comparison of means of the treatment with a uniform canopy arrangement and density of 8 plants·m^-2 (Table 8) also showed no significant differences between rows for any of the variables studied. Therefore, the lowest interception of PAR per plant due to the increase in density is what prevents increasing the yield per unit area in relation to the uniform arrangement with lower density.

If the rows of the high-density, uniform canopy treatment (Table 8) are compared with those of the same density in the stair-like arrangement (Table 6), it can be seen that the two middle rows of the latter treatment produce a higher yield and number of fruits per plant (approximately 70 % more yield and 85 % more fruits per plant) than the same rows of the uniform arrangement. This result supports the argument that this is due to the greater interception and homogeneous distribution of PAR in the plants located in the upper rows, in relation to the location of the four rows of plants in the same level (uniform canopy).

The results obtained in the present research allow us to deduce that with a stair-like arrangement of plants it is possible for pepper producers, in greenhouses with good climate control, to increase their annual yield per unit area. This yield level is even higher than that of producers in North Europe and North America who use the conventional high-tech system and obtain from 200 to 250 t·ha^-1, but with very high production costs (^{Heuvelink, 1995}; ^{Wien, 1999}). Since 12 kg·m^-2 (equivalent to 120 t·h^-1) were obtained in a four-month production period, from transplant to the end of the harvest, it is feasible to obtain three crop cycles per year with annual yields exceeding 300 t·ha^-1.

Conclusions

The stair-like system generated a higher fruit yield than the plants in the uniform canopy, due to the increase in the number of fruits per plant and its ability to maintain fruit weight without significant decreases.

By increasing the density from 6 to 8 plants·m^-2 in the stair-like canopy treatments, the yield per unit area was not increased, so for commercial management it is considered more appropriate to establish three rows of plants blunted at the fourth bifurcation in a stair-like arrangement with 6 plants·m^-2.