STRUCTURAL CHARACTERISTICS AND YIELD OF 'GIGANTE' CACTUS PEAR IN AGROECOSYTEMS IN THE SEMI-ARID REGION OF BAHIA, BRAZIL

The adaptation capacity of forage cactus (Opuntia ficus-indica Mill) plants to edaphoclimatic conditions and plant responses to changes in management systems contribute to increase the use of this species in agriculture and the exploration of its productive potential in semi-arid regions. The objective of this work was to evaluate the structural characteristics and cladode yield of forage cactus plants grown under different agroecosystems in the semi-arid region of Bahia, Brazil. Structural characteristics of plants and soils attributes were analyzed. The traditional information on the crop management and its correlations with cladode yield were used to identify the best yield indexes, considering the peculiarities of each agroecosystem. Plant height, cladode thickness, and number of cladodes of the forage cactus plants evaluated were less affected by the agroecosystem than by the production systems. Cladode width, length, and area were more affected by the agroecosystems. The forage cactus crop yields, expressed by the annual cladode fresh matter yield, were positively correlated with the plant structural characteristics: plant height and thickness, and cladode width and length. The cladode weight per plant and fresh matter yield per area were the yield components most affected by the management system adopted by traditional producers.


INTRODUCTION
The semi-arid region of the state of Bahia, Brazil, covers 446,021 km 2 , equivalent to 39.52% of the total area of the Brazilian Semi-arid region (BRASIL, 2017). The resilience of agroecosystems in this region contributed to the thriving of traditional producers, despite the adverse climate conditions and intense pressure on natural resources (RESENDE;LANI, 2002). Although resistant, forage cactus production systems are affected by dry periods (LUCENA et al., 2016). The high risk of occurrence of droughts and some technical criteria mean annual rainfall depths lower than 800 mm and aridity index lower than or equal to 0.50 -currently define the concept of semi-arid regions in Brazil (BRASIL, 2017).
Forage cactus crop systems are affected by water deficit, irregular rainfall distribution, environmental characteristics, producer profile, technological level, and social, economic, and cultural aspects (OLIVEIRA JÚNIOR et al., 2009;DONATO et al., 2014b;BARROS et al., 2016). Forage cactus plants are highly dependent on the crop environment. Their nutrient absorption capacity and vegetative development are dependent on edaphoclimatic factors, crop system, and plant genotype used (BLANCO-MACÍAS et al., 2010;DONATO et al., 2014b).
The challenge of establishing a better plantenvironment-human relationship to increase the crop resilience strengthens the potential of adaptability and longevity of plants (DONATO et al., 2014a;PADILHA JÚNIOR et al., 2016;SILVA et al., 2016;DONATO et al., 2017). Information on local environments, sharing of promising experiments, and recognition of crop responses to different environments and managements contribute to improvements in yield indexes and to the sustainability of forage cactus plants in the Brazilian Semi-arid region (SILVA et al., 2012).
The specificities of agroecosystems denote the need for different management processes and adjusts for their sustainability, mitigation of environmental limitations, and exploration of their potentials, since different environments demand different managements (RESENDE et al., 2007;RESENDE;REZENDE, 2017).
Considering the importance of maintaining herds when water is scarce and the need for ensuring feed availability and improving the use of natural resources in the Brazilian Semi-arid region, the objective with this work was to evaluate the structural characteristics and cladode yield of forage cactus (Opuntia ficus-indica Mill cv. Gigante) plants under different agroecosystems in the semi-arid region of the state of Bahia, Brazil.

Location and general characteristics of the study area
The study was conducted in different agroecosystems in the Guanambi microregion, which is under the Pediplano Sertanejo domain, in a degraded bare plain surface, at downstream of the Rio das Rãs River sub-basin. These landscapes evolved on the geology of the Guanambi Complex, which is found in the east part of the middle São Francisco River Basin, in a large plain region, whose flatness is disturbed by large smooth and sparse inselbergs. Surfaces with detritus from the Tertiary and Quaternary periods were found in small isolated flat areas of interfluves on the Santa Isabel Complex, at upstream of the Rio das Rãs River sub-basin (BRASIL, 1982).
This region presents a rainy season from November to April, and the dry season has six months (May to October). The lowest water availability is found in June to August. The mean annual rainfall depth is lower than 800 mm. The predominant climate in the Guanambi microregion, according to the Köppen classification, is BSwh, which corresponds to a hot climate of Caatinga, with rainfall in the summer and a well-defined dry period; a small area in the east presents an Aw climate, tropical rainy of forests, with dry winter and rainy summer (SEI, 2014).
The soil chemical and physical attributes of each agroecosystem were analyzed (Table 1). Three simple soils samples of the 0-0.20 m layer were collected using a hoe in the evaluation area of each plot and replication. P, K, Na, Cu, Mn, Fe, and Zn contents were determined by Mehlich-1; Ca 2+ , Mg 2+ , and Al 3+ contents were determined by the extractor KCl 1 mol L -1 ; H+Al was determined by the extractor calcium acetate 0.5 mol L -1 at pH 7.0; soil organic matter was determined using the Walkley-Black factor (SOM = organic carbon × 1.724); remaining phosphorus was determined by the P concentration in the soil solution after shaking bulk soil for 1 hour, using a CaCl 2 10 mmol L -1 solution containing 60 mg L -1 of P, at the ratio of 1:10; sulfur was extracted by monocalcium phosphate in acetic acid; boron was extracted in hot water.
Four traditional producers in representative areas of the production systems of the region were selected in each of the five agroecosystems, totaling 20 properties (Table 2). Harvesting for sampling in the production systems was carried out in August and September 2017.
Ceraíma 2013 Agreste 2014 Junquinho 2016 ---- Massal 2013 Sacoto 2005   The areas and producers presented some specificities: the Maniaçu region has high incidence of parrots that feed on forage cactus seeds; producers 1, 3, and 4 used insecticides without technical monitoring; producers 5 and 7 used urea and bovine manure for soil fertilization; producer 7 started irrigation in July 2017; producer 12 used ammonium sulfate and bovine manure for soil fertilization; producer 17 used urea for soil fertilization every 2 years; producers 1, 3, 9, and 12 used mineral oil for pest and disease control.

Characterization of forage cactus production systems of each producer
Selected producers were interviewed through semi-structured questionnaires, according to legal conditions (Resolution no. 466 of December 12, 2012 of the Brazilian National Health Council). This survey showed information of the history of the area, production data, and management system used, such as: planting time; soil fertilization; pest, disease, and weed control; and harvest time and method ( Table  2).
The forage cactus produced in the region is used, mainly, as feed for bovine and ovine herds. Traditional crop systems incorporate experiences shared between generations of producers, combined with technical information provided by education, research, and extension institutions, nongovernmental organizations of technical assistance, and class representation entities (unions). These crop systems present low use of external inputs, but demand agroecological and technical practices, which are developed by researches considering specificities of the sites to improve the maintenance, resilience, and sustainability of the activity in the semi-arid region.
The field survey was carried out using simple language to establish a horizontal and constructive dialogue with representants of the traditional communities (MATOS et al., 2014).

Evaluation of plant structural characteristics and yield
The structural characteristics of four randomized plants in each of the three replications (12 plants) of each of the 20 properties were evaluated, totaling 240 plants (GUIMARÃES et al., 2019). These characteristics consisted of: cladode thickness (CT), determined in the middle part of all cladodes; cladode width (CW), measured in the highest widest part of the cladode; cladode length (CL), measured in the longest part of the cladode; total number of cladodes (TNC) per plant; plant height (PH), distance from the top of the highest cladode and the soil; cladode area (CA) [CA (cm 2 ) = CC x LC x 0.693); 0.693 is the correction factor due to the elliptical form of the cladode (PINTO et al., 2002)]; total cladode area (TCA); cladode area index (CAI), which is the TCA of both sides of the cladodes divided by the area used by the plant (AUP) (m 2 of cladode per m 2 of soil).
The forage cactus crop yields were based on the cladode harvest of all plants in August and September 2017. Each plot presented a mean evaluation area of 14 m 2 and 16 plants. The cladodes were cut at their insertion point in the plant. The weight of all cladodes harvested were determined in the field for the plots and producers, and used to determine the cladode yield. The variables analyzed to determine the yield of the forage cactus crops were: annual cladode yield of each forage cactus crop in the different properties (ACY); annual cladode yield per plant (AYP); fresh matter yield annual or biannual, depending on the time of the last harvest (FMY); dry matter yield (DMY), calculated by multiplying the dry matter content of the treatment by its FMY.
Considering the dependency of the data of each production system within the agroecosystems, the hierarchical model was used. This design includes the structure of the factors and their levels; it is used when the levels of a factor B only occur in determined levels of a factor A. The factor A in the present work (Figure 2) is the regions or agroecosystems, and the factor B is the production systems represented by the producers and their properties within a same region. The characteristics evaluated (structural and yield) for forage cactus plants (Opuntia ficus-indica Mill cv. Gigante) in the agroecosystems represent a specific production system, thus satisfying the hierarchical condition (RIBEIRO JÚNIOR;MELO, 2008). In addition to the analysis of variance, the hierarchical model also estimates the variance of components and investigates the composition of the total variance, i.e., determines the explanation of the variance by the different factors of hierarchical levels (DIAS; BARROS, 2009).
The statistical analysis of the data was carried out in the SAEG 9.0 (System of Analyses Statistical) program of the Federal University of Viçosa, MG, Brazil, using the nested ANOVA/Hierarchical Model procedure (RIBEIRO JÚNIOR; MELO, 2008). When the variances were significantly different from zero, indicating the existence of at least one difference between agroecosystems and between production systems within each agroecosystem, the Tukey's test (p≤0.05) was applied to compare the means of the variables evaluated.

RESULTS AND DISCUSSION
According to the correlation analysis, the yields of the forage cactus crops, expressed by the annual and fresh matter yields of forage cactus plants, present significant, positive, and highmagnitude correlations with plant height and cladode thickness, width, and length (Table 3). The higher planting density (Table 5) used in the Iuiu and Ceraíma agroecosystems, combined with the higher soil fertility in these environments (Table 1) and use of irrigation (Table 2), resulted in the highest forage cactus yield (Tables 4 and 5).
The plant height (PH) and cladode thickness (CT) of the forage cactus crops were similar between agroecosystems and between production systems within each agroecosystem (p≤0.05) (Tables 4 and 5); 57.61% and 69.89% of the total variance for PH and CT, respectively, were explained by the production systems (Table 4).
The plants presented mean height of 1.03 m with coefficient of variation (CV) of 12.38%, and mean cladode thickness of 1.50 cm with CV of 18.59%. These are low variabilities according to the classification of Pimentel-Gomes and Garcia (2002).   Cladode length (CL) and cladode width (CW) presented significant variations between agroecosystems, probably due to responses to environmental differences (Table 1), as shown by the composition of the total variance (Tables 4 and 5). These results corroborate those of Barros et al. (2016); however, Mondragón-Jacobo and Pérez-Gonzáles (2001) reported that cladode size is determined by genotype and, at a lesser extent, by plant spatial arrangement and soil fertility. The use of a single cultivar and vegetative propagation in the same microregion, did not exclude the occurrence of differences in clones or even somaclonal variation, which may have been summed to environmental differences and increased genetic variability of the Opuntia ficus-indica Mill cv. Gigante used in the production systems.
The harvest cycle interval showed differences, with plants that were not harvested since the planting and others that had completed one or two years after the last harvest, when all plants were harvested to collect the data for the present work ( Table 2). The Morrinhos and Ceraíma agroecosystems had higher cladode yield per plant (24 to 28). Increasing planting density decreases the number of cladodes per plant and contributes to increases in forage cactus yield up to its biological potential.
The Maniaçu agroecosystem showed the greatest cladode widths and longest cladode lengths, with similar cladode lengths to those found in Ceraíma (Table 5). These results were found in Maniaçu, despite the soil low natural fertility when compared to the other regions (Table 1). Maniaçu has a mean altitude of 936 m, which is the highest altitude among the studied regions (Ceraíma = 542 m, Iuiu = 507 m, Riacho de Santana = 482 m, and Morrinhos = 843 m); this condition results in nights with milder temperatures, which is proper to the crop physiological demands, favoring the capture of CO 2 (SANTOS et al, 2013). However, based on the percentage of composition of the total variance, TNC was the variable most affected by the management systems used by the producers (Table 4).
Padilha Júnior et al. (2016) evaluated forage cactus in the Ceraíma region (Guanambi, BA, Brazil) and found that cladode width is not affected by soil fertilization; however, Barros et al. (2016) found this effect. In the present work, the differences in cladode width were more significant between agroecosystems (49.90%) than between production systems, which include the effect of soil fertilization (Tables 4 and 5).
The first cladode harvest in most traditional forage cactus crops is done at two years after planting (SILVA; SAMPAIO, 2015). However, harvest intervals in soils with high natural fertility (Tables 1 and 2) can be shorter than a year, depending on the forage demand (DONATO et al., 2014b;DONATO et al., 2017). The cladodes were harvested at one year after the last harvest, except for the producer eight in Iuiu, producer 10 in Maniaçu, and producers 13, 14, and 15 in Riacho de Santana, who harvested the cladodes after two years. Farias, Santos and Dubeux Junior (2005) reported that the biannual harvests provide greater longevity to forage cactus crops; however, is the demand and need to compensate for local edaphoclimatic limitations that define this period.
The number of residual cladodes, left in the plant by producers at harvest, varies. The producers in Iuiu harvested all cladodes, leaving only the main cladode, as the producers nine and 12 of Maniaçu and the producer 14 of Riacho de Santana. In the other agroecosystems, two to three cladodes were preserved to promote a more vigorous regrowth and provide greater longevity to forage cactus crops (Table 2). This practice assists in the maintenance of larger photosynthetic area and reserves in the plant (DONATO et al., 2014a).
The difference in number of cladodes between producers in Morrinhos was not significant (Table 5). This indicates a high similarity of the techniques used in these production systems. Dubeux Junior et al. (2006) evaluated four locations in the semi-arid region of the state of Pernambuco, Brazil, with the forage cactus clone IPA-20 and found higher number of cladodes in the lowest planting densities, constituting an inverse relation due to the larger soil surface explored.
The Ceraíma and Iuiu agroecosystems showed homogeneity in cladode width and length (Table 5). Maniaçu, Riacho de Santana, and Morrinhos presented only two levels of AUP, thus showing less variability in the spacings used by producers in each region, except Riacho de Santana, which presented three spacings adopted by producers (Table 5).
AUP presented significant positive correlation with TNC, but with low magnitude; and TNC presented negative significant correlation with plant density (Table 3). This denotes that the availability of a larger area per plant favors the emission of cladodes.
Among the production systems evaluated, a producer in Ceraíma and other three in Iuiu used irrigation for their forage cactus crops (Table 2), but without technical monitoring of the water depths applied and watering shift used. Despite the use of irrigation, the cladode thickness did not differ (Table  4). However, application of water, even at low quantities, is a viable option to ensure more satisfactory productions under the adverse conditions of the Brazilian Semi-arid region (LIMA et al., 2015). Table 5. Structural characteristics of forage cactus (Opuntia ficus-indica Mill cv. Gigante) crops grown in 20 traditional production systems in five agroecosystems in the Bahia semi-arid region -Guanambi microregion, state of Bahia, Brazil. Agroecosystem = environment of the forage cactus crop; Producer / Agroecosystem = production systems within the crop environment; TNC = total number of cladodes per plant; CW = cladode width; CL = cladode length; AUP = area used by the plant; CA = cladode area; TCA = total cladode area; CAI = cladode area index. Cladodes were harvested at one year after the last harvest, except for producers 8, 10, 13, 14, and 15, who harvested cladodes after two years. Means followed by the same letter in the columns for each agroecosystem are not different by the Tukey's test (p≤0.05). Absence of letters in the columns of a variable indicates not significant differences.
The Iuiu and Ceraíma agroecosystems had higher cladode area index (CAI): 3.92 and 3.11 m 2 cladode m -2 soil, respectively (Table 5). According to Nobel (2001), CAI of 4 to 5 indicates that the area of both cladode faces is four-to five-fold higher than the area used by the plant, and that they reached a morphology favorable to a high solar radiation capture and maximum yield. Contrastingly, when the CAI surpasses these indexes, forage cactus crop yield decreases (NOBEL, 2001). The CAI found in the present work were between 0.50 to 3.92, with coefficient of variation of 40.63% (Table 5), denoting very high variability (PIMENTEL-GOMES;GARCIA, 2002). The spatial arrangement of plants affects the CAI, even when maintaining the plant density (DONATO et al., 2014a).
The soils of the agroecosystems had low water and nutrient retention capacity (Table 1), and the producers did not use irrigation (Table 2); thus, the mean number of cladodes needed to reach 1 Mg in forage cactus crops in Riacho de Santana was 4,714 (Table 6). The other agroecosystems presented similarity (p≤0.05) in cladode weight, between 466 and 642 g, higher than the 245 g of the cladodes of Riacho de Santana (Table 6).   TNC  CW  CL  AUP  CA  TCA  CAI  (un) ------(cm) ------(m 2 ) -------(cm 2 ) -------(m 2 m -2 ) 1 Agroecosystem = environment of the forage cactus crop; Producer / Agroecosystem = production systems within the crop environment; ACY = annual cladode yield in each production system; AYP = annual cladode yield per plant; PD = plant density (plant ha -1 ); CMW = cladode mean weight; C/Mg = number of cladodes per Mg; FMY = fresh matter yield; DMY = dry matter yield. Cladodes were harvested at one year after the last harvest, except for producers 8, 10, 13, 14, and 15, who harvested cladodes after two years. Means followed by the same letter in the columns for each agroecosystem are not different by the Tukey's test (p≤0.05). Absence of letters in the columns of a variable indicates not significant differences.
The forage cactus crops within the Ceraíma, Iuiu, and Maniaçu agroecosystem presented no significant difference for number of cladodes per Mg. This denotes a low structural variability of cladode produced between forage cactus crops within these agroecosystems ( Table 6).
The productive potential of forage cactus crops of the Ceraíma and Iuiu agroecosystems showed high annual yield: 131.73 and 101.74 Mg ha -1 year -1 , respectively (Table 6). The overall coefficient of variation of annual yield for the agroecosystems studied was 35.36%, denoting very high variability (PIMENTEL-GOMES; GARCIA, 2002).
These yields denote the productive potential of forage cactus crops based on the cultural practices adopted in the agroecosystems and local edaphoclimatic conditions (Table 4), despite the higher soil fertility in the Ceraíma and Iuiu agroecosystems ( Table 1). The producer is the main factor in this process, because of its interest, capacity to thrive, traditional knowledge, and technical information, which affect the adoption of agricultural practices (RESENDE; CURI; LANI, 2002).
Irrigation for forage cactus crops in the irrigated perimeter of Ceraíma and in the Valley of Iuiu (Table 2) has contributed to increase their yield (FONSECA et al., 2019) to values higher than those found for agroecosystems with low fertility soils and water deficit, such as Riacho de Santana (Table 1). Forage cactus crops in Riacho de Santana had the lowest yield, with a mean of 15 Mg ha -1 year -1 ( Table  6). The low yield per plant (5.54 kg) of forage cactus crops in Iuiu was similar to that found in Riacho de Santana (5.03 kg), despite the higher soil natural fertility in Iuiu (Table 1).