Modeling of Cowpea (Vigna unguiculata) Yield and Control Insecticide Exposure in a Semi-Arid Region

The aim of this study was to evaluate the adaptability of different genotypes of cowpea (Vigna unguiculata L. Walp.) in the edaphoclimatic conditions of a semi-arid region. In the experimental design, a completely randomized split-plot (2 × 8), with 3 repetitions (blocks) was used. The experiment comprised 7 new genotypes and 1 local genotype as the first main factor and application of insecticide as a secondary factor. Two-factor analysis of variance (two-way ANOVA) determined the differences between the treated and untreated plots. The results obtained in the experiment showed that the introduced genotypes V3 (IT07K-181-55), V7 (H4), and V5 (IT97K-556-4M) adapted well to the edaphoclimatic conditions of the study area and their yields were respectively 1019, 1015, and 841 kg/ha of grains in treated plots and 278, 517 and 383 kg/ha in untreated plots. Multivariate analysis revealed that the most important parameter was the germination rate. Finally, the best yield was obtained with the genotype V3 (IT07K-181-55), subjected to the use of insecticide, and with the V7 (H4) genotype in untreated plants. The findings presented in this research should be useful in crop system agricultural programs, particularly in the terms of selection of cultivating systems suitable for high-yielding cowpea.


Introduction
Agricultural plants play an important role in the nutritional balance in semi-arid areas [1][2][3][4][5][6][7][8][9][10]. Currently, many researchers have focused a lot of effort into improving yields and increasing the production of crops [7,[11][12][13]. Cowpea (Vigna unguiculata L. Walp.) is an extensively cultivated plant, widely used as a legume in native tropical and subtropical zones or the semi-arid regions of South Africa [14]. Vigna unguiculata in rural areas (known as "Macunde beans") is a source of green manure, organic matter [15] for hay, forage and silage [16]. The goal of minor farming practices is to improve a cultivation system that is highly adaptable to changes in the soil condition [17,18]. Studies have documented that V. unguiculata can grow in agroforestry and orchards as an important protein plant to the human diet [19]. Other research showed a valuable improvement in yields [20][21][22][23]. In agroecological zone of Cameroonian Guinea, Sudano, and the Sahelian savanna, V. unguiculata is often combined with sorghum [24]. Therefore, it is one of the protein-rich plants able to combat food insecurity, malnutrition and poverty in sub-Saharan African populations [25,26]. Furthermore, many factors limit crop production, including biological ones [27][28][29][30]. Moreover, the low production of local genotypes, lack of improved genotypes and the use of traditional agricultural techniques in arid farming conditions strongly decrease the crop potential to grow [28,[31][32][33][34]. In the western and central regions

Grain Production
Highly significant differences (p < 0.001) were observed both between genotypes and between treatments with insecticide and without insecticide (Table 1). Mean values of grain yield of the insecticide use on grain yield of different genotypes of V. unguiculata are presented in Figure 1. Grain yield of the genotypes ranged from 307 to 1019 kg/ha with the application of insecticide and 113 to 517 kg/ha without treatment with an insecticide. Thus, in relation to grain production, three genotypes were highlighted: IT07K-187-55 (V3), H4 (V7), and IT97K-556-4M (V5). These produced respectively 1019, 1015, and 841 kg/ha of grain with the application of insecticide and 278, 517, and 383 kg/ha of grain without application of insecticide. Canonical Variety Analysis (CVA) plotted to graph detected relation between genotypes according to grain yield. CVA led to enhanced detection of grain patterns of variation across all analyzed groups. 68% of the total variation was accounted for the first two canonical axes ( Figure 2). Table 1. Germination rate for examined genotypes. The average of the genotypes indicated by the letters, are not significantly different at the rate level of 5% and coefficient of variation (CV).

Seed Production
Seed yields ranged widely, from 293 to 992 kg/ha of seeds with insecticide application and 103 to 501 kg/ha of seeds in genotypes untreated with insecticide. In fact, these genotypes were: IT07K-187-55 (V3), H4 (V7), and IT97K-556-4M (V5) which respectively produced 992, 991, and 811 kg/ha of seeds with insecticide application and 251, 501, and 364 kg/ha of seeds untreated with insecticide. The effect of insecticide on seed yield of different genotypes of V. unguiculata is presented in Figure 3. In the case of grain production, highly significant differences were also observed both among genotypes (p = 0.005) as well as among treated/untreated with insecticide (p = 0.0027), with respect to seed yield. Principal component axes PCA 1 (40.81%) and PCA 2 (21.85%) explained the variation ( Figure  4) of crop parameters revealing loading factors (Table 2) and eigenvectors (Table 3). Germination rate was ranked as the most important across all the studied plant traits ( Figure   Figure 2. CVA distinguished the best genotypes according to yield (grain). Cycles represent genotype H4, square determine genotype IT07K-181-55 and diamonds indicate genotype IT97K-556-4M. All data is prepared for treated with insecticide.

Seed Production
Seed yields ranged widely, from 293 to 992 kg/ha of seeds with insecticide application and 103 to 501 kg/ha of seeds in genotypes untreated with insecticide. In fact, these genotypes were: IT07K-187-55 (V3), H4 (V7), and IT97K-556-4M (V5) which respectively produced 992, 991, and 811 kg/ha of seeds with insecticide application and 251, 501, and 364 kg/ha of seeds untreated with insecticide. The effect of insecticide on seed yield of different genotypes of V. unguiculata is presented in Figure 3. In the case of grain production, highly significant differences were also observed both among genotypes  (Table 2) and eigenvectors (Table 3). Germination rate was ranked as the most important across all the studied plant traits ( Figure 5). In the present study, the correlation analysis showed that the measured parameters affected plant growth (Table 4). Furthermore, a generalized additive model (GAM) described that seed yield may be a qualitative variable used as indicator of Zonocerus variegates occurrence (Table 5, Figure 6a,c). Additionally, grain yield, exposure of disease, the weight of selected seeds, and insect risk have determined the main parameters affecting genotypes (Table 6, Figure 6b,d).

Influence of Pesticides on Germination and Anthesis
Damage caused by pests can reach up to 80-100% in the case of agricultural treatment [61][62][63][64] for effective plant management [65][66][67][68][69][70]. The results of this study showed that we detected a highly significant difference between the genotypes in relation to the percentage of germination (Table 1). In general, the genotypes IT07K-311-1 (V4), IT04K-221-1 (V1), DIAMANTE (V6), IT07K-187-55 (V3), and IT89KD-288 (V2) gave the highest significant rate, while genotypes IT97K-556-4M (V5) and H4 (V7) presented a below-average germination rate. The findings of this study lodged that the application of insecticide did not influence the number of days until flowering, the number of gaining maturity days, and the weight of 100 seeds. However, significant differences were observed between the genotypes in relation to the number of days needed until ripeness. The number of days until flowering (44-53 days, the average of days was 50), and the number of days to maturity (72-82 days, the average of days was 75), as well as the weight of 100 seeds (10-18.1 g), were linked to the genotype.
Data presented in Table 4 revealed that plants untreated with insecticide exhibited a decrease in yield when compared to those treated with insecticide. Nevertheless, results indicated that there were significant differences between the genotypes for the risk of disease (p < 0.05). The results showed that there was no significant difference when comparing the treatment with insecticide and without regard to the number of days until matureness (Table 6). Therefore, the average number of pods per plant was in the range of 8-12 per plant and was entirely different from data (6.0-52.0 pods per plant) published [71]. Thus, among the genotypes, the number of days before maturity varied between 72-82 days after sowing. Meanwhile, the local insecticide-treated genotype was able to grow with Plants 2021, 10, 1074 9 of 16 no signs of pest damage and virus disease. It should be strongly noted that the untreated genotype was susceptible to attacks by Zonocerus variegatus.

Selection of the Best Genotypes for the Field
In general, independently of insecticide application, the highest grain and seed yields of V. unguiculata were obtained with the genotypes IT07K-187-55 (V3), H4 (V7), and IT97K-556-4M (V5). Data recorded from sowing to harvest showed that all studied genotypes have a middle biological cycle (Figure 4). In addition, there were significant differences between treated and untreated genotypes according to yields of seeds and grains. The application of insecticide contributed to an increase of yield of 154.14% of grains (684 kg/ha) compared to the treatment without application of insecticide (261 kg/ha). Moreover, insecticide use also influenced the increase in seed yield of 173.78% (616 kg/ha), followed by those subjected to untreated with insecticide (225 kg/ha). The increase in grain and seed production may be explained by one factor controlling the impact of insects on insecticide application before and after flowering ( Figure 5), and therefore improved the average number of pods per plant. The results revealed that different genotypes of plants treated with insecticide, produced on average 10.7 pods per plant, whereas, untreated with insecticide 8.4 pods per plant. This research is in line with those of [72], who observed that pests are the main constraints to the production of V. unguiculata in many parts of Africa. The effect of plant yield was analyzed using the Generalized additive models (GAM). The following parameters: the threshing yield, the weight of seeds selected, the germination rate, the weight of 100 grain, the yield (grain), the number of harvested plants and the number of pods per plant, proved that seed yield may be considered as a biomass indicator and have an impact on the growth (Table 5). Our results showed the increase of germination rate (Figure 6a), and central lines (Figure 6b). Additionally, we reported number of harvested plants since abundance of Z. variegatus in short fallows and adjacent fields was no observed (Figure 6c), and the decrease for number of live plants in the border (Figure 6d). In addition, the DCA exhibited that qualitative variables were the germination rate, the weight of 100 grain, the threshing yield, and the grain yield. Thus, selection based on these traits could be applied in this semi-arid area.

Control Management of Insecticide in Semi-Arid Region
The total seed yield with and without insecticide was 6.338 kg/ha and this seed yield was lower than that related by [73], where 6.732 kg/ha was obtained. The average grain yield (475 kg/ha) and seed (452 kg/ha) were higher than the average yield of 300-400 kg/ha as reported [73]. However, these average yields were different in relation to research proposed [74,75] where average yields were of 724-844 kg/ha, and 526.5-7645.1 kg/ha. The average seed yield treated with insecticide (654.3 kg/ha) and untreated with insecticide (266.1 kg/ha) in our study was lower than the value achieved [76], reaching 60.7-1184.2 kg/ha. Conversely, the yields obtained in grains (684 kg/ha) were lower than the potential yield (1500 kg/ha of grains) with insecticides used for genotypes recorded by the National Institute of Research and Agronomic Studies (INERA) [77]. The low yield achieved was possibly influenced by edaphoclimatic factors: an abundance of rainfall during the experimental period, nutrient insufficiency, soil texture (sandy clay), and soil acidity slightly elevated in relation to that required by the V. unguiculata. The best yields of the V. unguiculata are obtained in drained sandy soils, little drained clay soils with a pH ranging between 6 and 7 [78].
Food security in the Republic of Angola is based mainly on agricultural production. In the Dange-Quitexe municipality (Province of Uíge), the effective increase in the production of V. unguiculata should be considered with decisive techniques in the cultivation of this crop in the subsequent experimental agricultural system [79]. When used as a groundcover, V. unguiculata constitutes in semiarid and arid region most widely covered by native legume plant and Africa's basic nutrition. In the semi-arid region, we recommend this plant as a grain legume, vegetable, and fodder crop.
The study area is characterized by sandy loam soil, with pH ranging from 5-6. The average annual temperature in the rainy season is 23 • C and 21 • C in the dry season. In the rainy season air humidity reaches 85-90%, while in the dry season it is 70-75% [79]. The lowest annual average precipitation is 100 mm and the highest is 1750 mm. The experiment was carried out during the rainy season within first agricultural season in Angola from September to February (Table 7).

Sampling Method Experimental Layout and Design
The seven genotypes (DIAMANTE, H4, IT04K-221-1, IT0K-187-55, IT07K-311-1, IT89KD-288 and IT97K-56-4M) were obtained from the Institut National pour l'Etude et la Recherche Agronomiques (INERA) 13, Av de Cliniques c/Gombe KINSHASA Répulique Démocratique du Congo. The crop was established in local agro-pastoralist perspectives for sustainable production proposed by [77]. We used 7-new genotypes + 1 local genotype. Seeds were sown 5 to 7 cm deep with 18 cm row spacing. Each plot comprised rows of 2.25 m long and 3 m wide, and a spacing of 1.25 m width between rows and 28.5 m between blocks (Figure 7). Insecticide were applied against Z. variegatus [80] to each plot at the rate of 20 mL per 16 L (water) in the form of an aerosol, which was pipetted at planting and drifted into plants (two times per week). Selective substance-cypermethrin (25%) was applied [81]. The total area of the field was 826.5 m 2 . The data collection area comprised two central crop-systems, with 0.5 m of each end of the sub-plots and the two lateral lines used as borders. The data collected in the area of each field plot was based on 64 plants randomly extracted from the total population of 128 plants. The experiment was performed according to a randomized complete block design presented as a split-plot (2 × 8), with 3 replicates (blocks). In this cropping system, we incorporated two fundamental factors, namely genotype, as the main factor and insecticide (cypermethrin) use as the second factor. Each block was composed of two plots subdivided into 16 subplots, treated (eight) and untreated (eight) with insecticide that were denominated as T1 (treatment without the use of insecticide), and T2 (treatment with the use of insecticide).
Phenological characteristics related to the plant yield as well as seed production in agriculture system [82] and pesticide interaction were detected [83]. The aphid-mosaic virus in seed was detected by a growing-out test. Risk of disease has been evaluated on a 1-9 scale (>7 severity of insect risk). Phytosanitary control was reported according to the rating scale: 1 = no attack 2-3 = slightly 4-6 = moderate 7-9 = severe Plants were harvested to estimate seed and grain yield per plot, which was subsequently converted to tonne per hectare (t ha −1 ). Meanwhile, more experimental details are showed in Figure 2. Only African variegated grasshopper Zonocerus variegatus (L.) (Orthoptera: Pyrgomorphidae) was simultaneously present in the experimental design. Germination rate was obtained by the number of germinated seeds. Flowering was determined according to the number of days from sowing until flowering. Maturity was determined based on the number of days from sowing until maturation. A number of pods per plant (NPP) were obtained by the average number of pods collected at random from 10 plants. Grain yield (GY) was determined by the amount of grains not selected which means combination of low and high quality. Seed yield (SY) was determined by the number of selected grains, which presented better quality (bigger size, vigorous and higher weight). Weight of 100 grains (WHG) was obtained by the means of the total weight of seeds per treatment. Total yield (TY) was determined as a function of the total productivity of the studied area.  Phenological characteristics related to the plant yield as well as seed production in agriculture system [82] and pesticide interaction were detected [83]. The aphid-mosaic virus in seed was detected by a growing-out test. Risk of disease has been evaluated on a 1-9 scale (>7 severity of insect risk). Phytosanitary control was reported according to the rating scale: 1 = no attack 2-3 = slightly 4-6 = moderate 7-9 = severe Plants were harvested to estimate seed and grain yield per plot, which was subsequently converted to tonne per hectare (t ha −1 ). Meanwhile, more experimental details are showed in Figure 2. Only African variegated grasshopper Zonocerus variegatus (L.) (Orthoptera: Pyrgomorphidae) was simultaneously present in the experimental design. Germination rate was obtained by the number of germinated seeds. Flowering was determined according to the number of days from sowing until flowering. Maturity was determined based on the number of days from sowing until maturation. A number of pods per plant (NPP) were obtained by the average number of pods collected at random from 10 plants. Grain yield (GY) was determined by the amount of grains not selected which means combination of low and high quality. Seed yield (SY) was determined by the number of selected grains, which presented better quality (bigger size, vigorous and higher weight). Weight of 100 grains (WHG) was obtained by the means of the total weight of seeds per treatment. Total yield (TY) was determined as a function of the total productivity of the studied area.

Statistical Data Handling
One-way analysis of variance (ANOVA) has proven as an excellent technique to summarize and determine the differences between the genotypes treated (eight) and untreated (eight) with insecticide, using Statistical Software 8.0. In this study, a total of 7 genotypes were characterized. PCA (Principal component analysis) was a useful technique to emphasize the most important plant parameters. Group I consisted of insecticide-treated plants and group II of untreated samples to allow better visualization of the variation and to present combinations of out/input variables in a specific way. This scrutiny was calculated with PQStat software (ver. 1.6.4). An exploratory method Canonical Variety Analysis (CVA) was applied to cope with collinear high-dimensional data and accordingly grouped variables for yield grain were interpreted. Individual or group mean scores and multiply regression were computed by PaST 3.20 [84]. Multivariate analysis DCA (Detrended correspondence analysis) based on the data matrix revealed the variation patterns in genotypes treated and untreated with insecticide. Variables were computed with Canoco for Windows 4.5. Current work provides data related to multiply regression represented by 8 parameters. GAM (generalized additive model) is defined as the range of models examined step-wise and can be used by adding other covariates. The model was performed with Statistica 13.3. Formally, the GAM framework imposing a particular functional form that may be adapted to the data as a sum of independent variables. V. unguiculata has strong potential as a component of a sustainable crop farming system as well as animal feed and fodder. Consequently, our model was determined by a response variable to get an accurate point of forecasting, and was formally written as: g(E(y i )) = β 0 + f 1 (x i1 ) + . . . + f p x ip + ε i (1) where: g-link function (the identical function), y-response variable (depends linearly on unknown smooth functions), x 1 . . . x n -independent variables (predictor variables), f 1 , . . . , f p -smooth functions (splines), i = 1, . . . , N, β-an intercept, -random error (a constant error variance is assumed).
Interaction between smooth function resulted in the following formula: where: f -smooth functions, q-basis dimension, b-the sum of basis functions, β-corresponding regression coefficients.

Conclusions
The evaluation of different genotypes of V. unguiculata highlighted the adaptability to edaphoclimatic conditions of incorporated genotypes and the interaction between insecticides use and their productivity. Certainly, the use of insecticide contributed to the increase of the yield of V. unguiculata seeds. Our findings showed that genotypes IT07K-181-55 (V3), H4 (V7), and IT97K-556-4M (V5) can be recommended to farmers to optimize the yield and integrated management strategy for phytosanitary control of Zonocerus variegates. Yield seed was positively correlated with pod number, while seed number per pod and mean seed weight remained negatively related to field potential. Yield may be used specifically as plant identification of adaptability to drought-prone conditions. In future research, V. unguiculata should be successfully used as a cover to the subsequent crop and plant protection. Small-scale farmers may provide quantitative traits of field seed for improved cultivar development, large seed size, and resistance to major diseases. Our results showed that seed yield can severely limit the crop system. In the prospect study, we suggest focusing on the selection of suitable genotypes of V. unguiculata that dominate in the cultivated areas, including risk assessment of the Zonocerus variegates.