Forage yield and morphological traits of cactus pear genotypes

. The objective of this study was to evaluate the forage yield and morphological responses of 34 cactus pear ( Nopalea cochenillifera and Opuntia ficus indica ) genotypes. The genotypes were planted in a randomized complete block design, with 34 treatments and three replicates. After 330 days of cultivation, morphological characteristics, plant mortality and pest and disease occurrence on the plants were assessed, and plant cutting was performed. Multivariate analyses were performed to assign the 34 cactus genotypes to homogenous groups. The average Euclidian distance was adopted as a dissimilarity measurement for use with the standardized data Variables with relatively high levels of independence among the agronomic categories and with biological importance to animal nutrition and forage yield were assessed. To identify associative effects, Pearson’s correlation analysis was performed among all the studied variables. The formation of five distinct groups was observed, and some morphological responses correlated with the productivity characteristics. Green mass production was positively correlated with dry mass production, water accumulation, water-use efficiency, water accumulation and cladode number per plant. The following genotypes had remarkably high yields of dry and green mass, a greater ability to accumulate water and relatively high water-use efficiency and support capacity: Negro Michoacan (V07), Tamazunchale (V12), California (V14), Orelha de Elefante Mexicana (V17), and Amarillo 2289 (T32).


Introduction
Cacti have potential for use in animal feeding. In arid and semiarid regions, cacti are preferable over other forage due their high adaptability and capacity to produce herbaceous biomass with little water availability. Due to historical and cultural factors, the main genera of cultivated cactus pear is Opuntia spp.; the two dominant genotypes are Gigante and Redonda, but these genotypes are susceptible to pests and diseases (Dubeux Júnior, Santos, Cavalcante, & Santos, 2013).
Due to the particular characteristics of the semiarid region of northeastern Brazil, one of the main difficulties that producers face is the lack of food for herds in drought periods, given the lack of pasture development. The alternative is to use a food source that is better adapted to water scarcity conditions, such as the cactus pear in its natural form, rather than grass or silage.
To identify promising cactus genotypes that may be used as a basis for genetic improvement, studies that assess adaptive, morphological and structural characteristics, such as the number and dimensions of cladodes, which are positively correlated with forage yield, are necessary (Pinheiro et al., 2014). Barbosa et al. (2018) evaluated the morphological characteristics of three cactus pear genotypes (IPA Sertânia, Miúda, and Orelha de Elefante Mexicana) and found that the growth characteristics (plant height and plant width) were strongly or very strongly positively correlated with the morphological characteristics of the cladodes (basal cladode length, cladode width, cladode thickness, cladode perimeter, and cladode area), contributing to the development of the plant as a whole. Amorim, Martuscello, Araújo Filho, Cunha, and Jank (2015) evaluated the agronomic characteristics of 24 varieties of cactus pear and verified that the agronomic and morphological characteristics (plant height, plant width, cladode area, and total number of cladodes) were positively correlated with cactus pear mass.
We hypothesized that by evaluating the morphologic and productive characteristics of several cactus pear genotypes belonging to two different genera, it would be possible to identify the genotypes that had relatively high levels of productivity.
The agronomic potential and morphological characteristics of 34 genotypes of cactus pear cultivated for forage were evaluated.

Experimental location and period
The experiment was conducted in Experimental Station Benjamim Maranhão, in the city of Tacima, Paraíba State, Brazil. The station is in the mesoregion of Paraíba and the microregion of Curimataú Oriental (coordinates: 6º 29' 16" East and 35º 38' 13" West), has an altitude of 168 m, and belongs to the Empresa Paraibana de Pesquisa, Extensão Rural e Regularização Fundiária (EMPAER). Weather data were collected throughout the experimental period, resulting in a cumulative rainfall of 586 mm and an average temperature of 25°C from January to November 2016 ( Figure 1).

Treatments and Experimental design
The experiment was established as randomized full blocks, including 34 treatments and three replicates, with 20 plants per experimental unit. The treatments consisted of 34 cactus pear (Opuntia spp. and Nopalea spp.) genotypes from the germplasm bank of the Empresa Pernambucana de Pesquisa Agropecuária (IPA), Pernambuco State, Brazil, which were donated for research to EMPAER. The genotypes are described by genera below. The common names are given followed by the identification number of the genotype and the reference paper in parentheses: Opuntia

Soil characteristics and planting of the cactus genotypes
Before planting, random soil samples representing the experimental area were collected at depths of 0-20 cm to characterize the soil fertility (Table 1). Mineral fertilization was performed by applying 50 kg of P 2 O 5 ha -1 in triple superphosphate (5 g of fertilizer per plant), 70 kg of K 2 O ha -1 in potassium chloride to the soil (6 g of fertilizer per plant) and 50 kg of N ha -1 in urea. The addition of urea was divided into two applications: one at 60 days after planting with 40% N and the other at 180 days after planting with 60% N (2 g of fertilizer per plant in the 1 st application and 3 g of fertilizer per plant in the 2 nd application), according to the fertilization recommendations for the soil type at the experimental location (Cavalcanti, 1998). The planting occurred in November 2015, and the experimental period began in January 2016 and ended in November of the same year. The plantings consisted of a cladode per hole in a vertical position, with 1.0 m between rows and 0.4 m between the plants. There was a total of 25,000 plants per hectare. When necessary, periodic weeding was performed to control the occurrence of weeds.

Evaluated variants
The genotypes were evaluated by the number of cladodes per plant (NCP), widths of the cladodes (WCs), lengths of the cladodes (LCs), diameters of the cladodes (DCs), thicknesses of the cladodes (TCs), heights of the plants (HPs), widths of the plants (WPs), dry matter (DM), green mass production (GMP), dry mass production (DMP), water-use efficiency (WUE), water accumulation (WA), and support capacity (SC). Each variable was evaluated for five plants chosen at random by plot. The correlation among the morphological and productive characteristics and the grouping analysis was also evaluated through the dissimilarity of the characteristics among the cactus pear genotypes evaluated (Ward, 1963).
To determine the NCP, a total count of cladodes per plant was performed, and the values were used to represent each cactus genotype. For the quantification of the WCs, LCs, and DCs (in cm), a graduated measuring tape with a length of 150 cm was used. To quantify the TCs, a digital caliper was used.
To measure the HPs, a tape measure was used, and the measurement was taken vertically, from the base of the plant to the highest point, and expressed in cm. A tape measure was used to quantify the WPs, which are expressed in cm, with the measurement being taken horizontally, from one point to another at the widest part of the plant.
A subsample of each treatment was collected for predrying in an oven with forced ventilation at 55ºC until the material reached a constant dry weight to quantify the DM (expressed in g kg -1 of natural matter) according to the methodology recommended by method 934.01 (AOAC, 1990).
To determine the GMP in t ha -1 , cutting was performed 330 days after planting. The GMP (in t ha -1 ) was obtained by multiplying the green mass obtained per harvested plant and the population of plants per hectare and dividing the product by "1,000".
To determine the DMP, the plants were cut 330 days after planting. Then, a subsample of each treatment was taken for predrying at 55ºC in a greenhouse with forced ventilation until the material reached a constant dry weight. The level of DM (expressed in g kg -1 of natural matter) was quantified. The DMP (t ha -1 ) was estimated from the GMP/plant (kg) multiplied by the population of plants/ha, and the product was divided by 1,000 to determine the result in tonnes. The level of DM was calculated from the green mass product (kg) obtained from the harvested plants converted to the total plants ha -1 .
The WUE (kg DM mm -1 ) was estimated by the division of the DMP (t ha -1 ) by the quantity of accumulated rain during the experimental period. The WA (t ha -1 ) was estimated from the GMP (t ha -1 ) multiplied by the water percentage of the plant, which was calculated by subtracting 100 from the DM level; after this, the result was divided by "1,000".
A simulation of the SC of one hectare of the cactus genotypes (CGs) was performed based on the confinement of sheep for a period of 90 days. We know that DMP is expressed in t ha -1 . A sheep with an average live weight (LW) of 25 kg consumes 3% of its LW x 60% CG in a diet with 40% DM. The formula SC = (DMP t ha -1 )/(individual consumption x 90 days of feedlot) was used, in which SC is the support capacity (i.e., the quantity of the animals).

Statistical analyses
Multivariate analyses were performed and Ward's (1963) method was used to create homogeneous groups among the 34 cactus genotypes. The average Euclidian distance was adopted as a dissimilarity measurement to be used with standardized data and variants with relatively high levels of independence among the agronomic categories and with biological importance to animal nutrition and forage production were used. To identify the associative effects, Pearson's correlation analysis was performed among all the studied variants.
After the data were organized, a variance analysis was used to verify the variability among the agronomic characteristics of the 34 cactus genotypes evaluated.
The differences among the cactus genotypes were evaluated by the following model: = µ + + + , in which Yij is the observed value of i-th genotype in the j-th block; µ is the genotype average; αi is the effect of the i-th genotype; i represents 1, 2, 3..., 34; βj is the effect of the j-th block; j represents 1, 2, and 3; and eij is the random error of the j-th observation in the i-th genotype.
For the characteristics for which the analysis of variance was significant, the means were compared by a Scott and Knott test (1974) at a 5% significance level. The statistical analyses were performed using R software (R Core Team, 2018) and the Statistical Analysis System (SAS, 2010).

Results and discussion
There was an effect (p < 0.05) on almost all the morphological variants, except the TCs, which had an average value of 1.8 mm (Table 2). Silva et al. (2010), evaluating correlations between the morphological and productive characteristics of 50 cactus pear clones, did not find an effect for cladode thickness, with an average value of 3.9 cm.
The NCP ranged from 4 to 59. The genotypes that showed the highest numbers of cladodes were California (V14), Nopalea Uruapan (V20), and Tamazunchale (V12), which may be due to these genotypes belonging to the Nopalea spp. The species belonging to this genera are characterized as being more demanding of edaphoclimatic characteristics compared with cactus species from the Opuntia genera and as presenting larger numbers of smaller cladodes.
Evaluating these results practically, it may be inferred that the cultivation of cactus genotypes that produce higher quantities of cladodes may facilitate an increase in cultivation areas, as cacti can be propagated vegetatively. In this system, plantations are filled with clones of the original plants (Amorim et al., 2015;Gebreegziabher & Tsegay, 2015).
The characteristics of the cladode dimensions (width, length and diameter) showed a significant effect (p < 0.05) ( Table 2). For the WCs, variations from 7.6 to 21.5 cm were verified. The genotypes with smaller cladode widths were Nopalea Uruapan (V20) and California (V14), and larger cladode widths were found in the genotypes Orelha de Elefante Mexicana (V17), Rosa S.L.P. (T75), and Oaxaca (F10). The LCs varied from 15.2 to 32.1 cm, and the genotypes with smaller lengths were California (V14) and Vila Nuova (F22); those with larger lengths were Villa Nueva (V22) and Orelha de Elefante Mexicana (V17). The genotype California (V14) had a smaller cladode diameter, with an average value of 37.2 cm, and the genotype Orelha de Elefante Mexicana (V17) had a larger cladode diameter, with an average value of 77.2 cm.
The genotypes with cladodes that had relatively large dimensions were relatively short and narrow, probably as a function of the structure of the plant, because larger cladodes would require plants to invest in nutrient accumulation at the cladode base, increasing thickness to support growth. Plants that have a higher weight per cladode develop horizontally, and the higher fortification of the base of the primary cladodes sustains the weight of the remaining cladodes instead of the tissues being used to support vertical growth; this pattern was observed in the Orelha de Elefante Mexicana (V17) genotype.
Acta Scientiarum. Agronomy, v. 43, e51214, 2021 Rocha, Voltolini, and Gava (2017) studied the productivity of three types of cactus (Orelha de Elefante Mexicana, Miúda and IPA 20) and verified that at 12 months after cultivation, the Miúda genotype had a greater number of cladodes per plant. Similar results were obtained by Silva et al. (2015), who evaluated the morphological and productivity characteristics of three types of cactus (Orelha de Elefante Mexicana, Miúda, and IPA Sertânia).
The shapes and dimensions of the plants and cladodes of different genotypes belonging to the genera Nopalea and Opuntia differed. Studying how these characteristics are related may facilitate the understanding of how the plants responded under different genetic and environmental conditions. The plants with smaller cladodes may invest in vertical growth and distribute their cladodes accordingly, leading to relatively tall, narrow plants, while plants with larger cladodes may invest in horizontal growth, structuring the arrangement of biomass as needed to support this growth (Neder, Costa, Edvan, & Souto Filho, 2013). Barbosa et al. (2018) evaluated the morphological and productivity characteristics of three cactus genotypes (IPA Sertânia, Miúda, and Orelha de Elefante Mexicana) and verified that the characteristics of the cladode influenced canopy formation. The genotype Orelha de Elefante Mexicana had a relatively large cladode diameter, width and length, but these values spread in relatively tall, narrow plants.
The height and width characteristics of plants directly influence the productivity characteristics, such as GMP and DMP. Tall plants can be grown with more intensive spacing because, as the plants grow, they do not limit the growth of other plants or make it difficult to carry out culturing treatments. The wider cactus genotypes are indicated to be grown with larger spacing to facilitate culturing treatments and harvest, because in these genotypes, the competition between plants for water, light and nutrients is greater than it is in the narrower plants (Arantes, Donato, & Silva, 2010).
The canopy structure may be defined by various characteristics, such as size, shape, number of cladodes, gender, species, and environmental variants, which may define the structure of the plant. Because plants exhibit phenotypic plasticity in response to handling and cultivation conditions, the accurate measurement of these characteristics is difficult, and plants of the same species may be structured in different ways (Lüttge, 2010;Pinheiro et al., 2014).
There was an effect (p < 0.05) on the DM, with the genotypes Moradillha (V03), Copena v1 (V04), and Amarilla Vach (T42) having lower levels of DM, with average values of 72.6, 76.7, and 84.3 g kg -1 in natural matter, respectively. In contrast, the genotypes Polotitlen (V09), Vila Nuova (F22), and Oxaca (V10), had higher levels of DM, with average values of 184.4, 179.70, and 173.1 g kg -1 of natural matter, respectively (Table 3). Table 3. Average values of dry matter (DM) levels, green mass production (GMP), dry mass production (DMP), water use efficiency (WUE), water accumulation (WA) and support capacity (SC) of 34 cactus genotypes. The results obtained in the present study are in accordance with the results obtained by Santos, Lira, Farias, Dias, and Silva (2006), who evaluated genotypes belonging to the genera Opuntia (Gigante, Redonda, Copena v1, Copena f1, and IPA 20) and Nopalea spp. (Miúda) in different locations. The authors verified that the Redonda and Gigante varieties had lower levels of DM (from 8.8 to 9.3%), while the Miúda variety had a markedly higher level of DM (11.7%).
There was an effect on the WUE and WA (p < 0.05), and the Rosa Liso (T63) and Rosa (T64) genotypes had relatively low values of both variables, with WUE of 3 kg DM mm -1 and WA of 13 t ha -1 . The genotypes Rosa Liso (T63) and Rosa (T64) had relatively low values of WUE. The California (V14) and Copena V1 (F02) genotypes presented higher water-use efficiency and conversion to forage mass, with average values of 52.6 and 43.3 kg DM mm -1 , respectively.
Three genotypes presented relatively high GMP and DMP: Negro Michoacan (V07), Tamazunchale (V12), and California (V14). However, these genotypes do not have higher levels of DM than the other genotypes; therefore, the high production relative to the other genotypes was due to the relatively high growth rate of the plants. More cladodes were found in the Tumazunchale (V12) and California (V14) genotypes, while the Negro Michoacan (V07) genotype presented fewer cladodes but larger dimensions (i.e., width and height of the plant), indicating that these three genotypes had high growth levels, which also explains why they had higher WUE, higher WA, and higher SC than the other genotypes.
Higher GMP and DMP were found in the California (V14), Negro Michoacan (V07), and Tamazunchale  DMP is a characteristic that is studied and highlighted when it is necessary to identify or select genotypes with high dry mass production. DMP is one of the characteristics that are most sought in the genetic improvement of cacti because it directly influences usage efficiency and leads to the optimization of resources in the cultivation of cactus genotypes with high DMP Sales, Leite, Alves, Ramos, & Nascimento, 2013).
The more productive genotypes were also the ones that had higher WUE and WA values. This was due to the relatively high proportion of tissues in these genotypes, which had higher forage value due to their dry mass volume. This means that the higher the water-use efficiency and the larger the production of green mass are, the larger the capacity of the cladodes to store water. These two characteristics, WUE and WA, are interconnected, and the way they are calculated tends to show their dependence on each other.
Water is an essential resource for animals and vegetables; in the case of vegetables, water is mainly responsible for cell expansion. When a plant undergoes hydric stress, relatively little cell and tissue growth occurs, and there is little increase in plant heights and widths. Under hydric stress conditions, plants use survival strategies and paralysis or drastically decrease their growth. Because cacti are cultivated in arid and semiarid environments, offering cacti in natura as feed may provide some of the water needed by the animals (Costa, Suassuna, Melo, Brito, & Mesquita, 2012).
These results demonstrate that cacti are an essential food to be used in animal feeding in arid and semiarid regions. Some of the genotypes are highlighted as being well-suited for this purpose, while others present high levels of SC due to their relatively large production of forage mass under cultivation conditions. The high forage production suggests an elevated adaptability to the cultivation environment and consequently elevated animal production per hectare (Mokoboki & Sebola, 2017).
The correlations between the morphological and productivity characteristics are shown in Table 4. The correlations among the productivity and morphological characteristics may be genetically or environmentally induced. The characteristics that did not present correlations among each other were independent of the other characteristics, because the increasing or decreasing of the values of a given characteristic did not influence those of the other characteristics . GMP: green mass production (t ha -1 ); DMP: dry-matter production (t ha -1 ); WUE: water use efficiency (kg/ DM/mm -1 ); WA: water accumulation (t ha -1 ); NCP: number of cladode per plant (unity); LC: length of the cladode (cm); WC: width of the cladode (cm); DC: diameter of the cladode (cm); TC: thickness of the cladode (cm); HP: height of the plant (cm); WP: width of the plant (cm); SC: support capacity (quantity of animals); * significance at 5%; ** significance at 1%.
Data source: Authors.
There was a positive and significant correlation between the GMP and DMP (r = 0.925**). The correlation between the GMP and DMP is supported by the fact that they are both dependent variables, and as one increases the other also increases. The difference between the GMP and DMP represents the humidity.
There was a significant positive correlation between the GMP and WUE (r = 0.925**) and the GMP and WA (r = 0.998**) and a strong positive correlation between the GMP and NCP (r = 0.632**). The very strong positive correlation of GMP with WUE and WA and the strong positive correlation of GMP with NCP are related to the growth of the plants. An increase in the ability to convert water to forage mass occurs when plants have more cladodes or larger cladodes than usual. This increases the water storage capacity and nutrient accumulation and consequently increases the photosynthetic efficiency of plants due to the utilization of rain water and nutrient extraction from the soil and the ability to convert these resources into forage mass (Pinheiro et al., 2014).
The widths of the plants were positively correlated with the GMP and DMP. The horizontal growth of the plants provided vegetative expansion and an increase in the GMP, DMP, WUE, WA, and NCP due to the relatively high capacity that each individual cladode has to store water and nutrients. The older cladodes serve as a basis to sustain the plant and the new cladodes lead to the growth of the plant (Silva et al., 2015). Silva et al. (2010) evaluated the relation among morphological and productivity characteristics of 50 cactus pear clones in rainfed conditions and found moderate values (varying from 0.560** to 0.610**) of the positive correlations between the GMP and the HP and WP.
The heights of the plants were not correlated with the productivity characteristics, indicating that this characteristic was independent of the others. The exception was the lengths and diameters of the cladodes, because these measurements directly influence the sustentation of the plant.
The results found by Barbosa et al. (2018) were similar to those obtained by this study, indicating that there is a strong correlation among the dimensions of the cladode. The authors evaluated the morphological characteristics of three cactus pear clones and verified that the lengths, widths and diameters of the cladodes were highly correlated; the thicknesses of the cladodes, however, showed a slightly negative and nonsignificant correlation.
Consistent with the results of the present research, the study by Amorim et al. (2015), which evaluated the agronomic characteristics of 24 cactus varieties of the genera Nopalea, verified that there was a significant positive correlation between the agronomic (GMP, DMP, and NCP) and morphological (HP, WP, DC, TC, LC, and WC) characteristics and the production of forage mass among the cactus varieties.
To evaluate the discriminatory agronomic characteristics, the GMP, DMP, WUE, WA, SC, NCP, HPs, and WPs were used. The results were expressed in a dissimilarity dendrogram (Figure 2). Acta Scientiarum. Agronomy, v. 43, e51214, 2021 According to the dissimilarity dendrogram, there was variability in each group, indicating that some genotypes may stand out due to their agronomic characteristics.
In the descriptive analysis of the data, a hierarchical formation of five groups was observed, with 7% total dissimilarity. The formation of the groups took into account a Euclidian distance of 2% in relation to the genotypes (Table 5).
Data source: Authors.
The grouping analysis identifies the variations that occur in each genotype. These variations are also expressed within each group, with higher homogeneity among the genotypes of a given group and higher heterogeneity among the genotypes of the other groups. From the point of view of the selection and identification of genotypes, this variation is favorable, as it indicates that the genotypes are different from each other, and provides the possibility of the comparison and selection of the genotypes that presented the desired characteristics, such as productivity and/or morphological characteristics (Ferreira et al., 2003).
As the grouping analysis shows, group I included genotypes with higher dry matter and green mass production, which was due to the genotypes in this group presenting relatively high WUE, WA, and heights.
These variables are related to the productivity characteristics; therefore, as one characteristic increases or decreases, the other dependent characteristics respond positively or negatively, depending on the type of correlation among the variables. The genotypes belonging to groups II, III, and IV had intermediate values relative to those of the genotypes of group I and the group V, which had relatively low productivity and morphological characteristics.
According to the classification, groups I and V had relatively low and high average values for the following variables, respectively: GMP, 32 and 185.1 t ha -1 ; DMP, 4.7 and 23.2 t ha -1 ; WUE, 8 and 39.5 kg DM mm -1 ; WA, 27.2 and 161.9 t ha -1 ; SC, 116.5 and 572.8 animals ha -1 ; and NCP, 7 and 58.7 per plant.
The morphological characteristics, such as the heights and widths of the plants, were highly correlated with the productivity characteristics; as a plant increases its yield, it needs to expand its tissues by increasing the cladode dimensions and number, which also influences the height or width of the plant. Therefore, these morphological and productivity characteristics are interconnected, although cactus cultivation produces forage mass. Beyond the factors already described, the forage mass also depends on the edaphoclimatic characteristics and the density of the cultivation. All of these characteristics acting together (i.e., the environmental effect) with the genetic characteristics (i.e., the genetic effect) of the species will determine the response of the cactus to cultivation in distinct situations (Oelofse, Labuschagne, & Potgieter, 2006;Pinheiro et al., 2014).
Relatively short and tall plants were found in the genotypes belonging to groups II and III, with values from 64.7 and 85.2 cm, respectively. For the WPs, group V presented a relatively low value (54.9 cm), and group I presented a relatively high value (104.1 cm). The NCP presented relatively low and high variation for groups I and II, respectively, with a CV of 1.5% for group I and a CV of 69% for group II.