Biological material and experimental area

The Missouri variety with an erect growth habit was used. The research was carried out on farmers' fields, in the municipality of Cereté-Colombia, during the end of the rainy season of the second half of 2022 (October 27 to December 27, 2022: dry season) and during the beginning of the rainy season of the first half of 2023 (April 24 to June 24, 2023: rainy season). The town is located between the Geographic coordinates (8°57′24.7″ N and 75°45′10.3″ W), altitude of 12 m a.s.l., average annual precipitation of 1,300 mm, average temperature of 28ºC and 6-7 h average sunshine. In 2022B, during the crop cycle, a rainfall of 147 mm was recorded, while in 2023A, it was 327 mm. The soil analysis showed the following: texture: silty-clay; pH=5.6; OM=2.32%; S=1.1 mg kg^-1; P=7.9 mg kg^-1; Ca=14.68 cmol₍₊₎ kg^-1; Mg=7.31 cmol₍₊₎ kg^-1; K=0.61 cmol₍₊₎ kg^-1; CEC=22.7 cmol₍₊₎ kg^-1. The soil was prepared conventionally, with plowing and harrowing.

Experimental design

A randomized complete block design was used, with eight treatments and four replications. The treatments consisted of four-time intervals of weeds control (WC): 0-10, 0-20, 0-30 and 0-50 days after the emergency (DAE) and the same time intervals for coexistence (Co) of the crop with weeds (0-10, 0-20, 0-30 and 0-50 DAE). A total of 32 experimental units, consisted of 4 rows 5 m long, with spacing between rows and between plants of 0.40 and 0.25 cm, respectively, and a planting density of 100.000 sites/ha, 2 seeds were sown per site and the useful plot was made up of the two central rows. The CPC was evaluated with the variable grain yield (YIELD) for all treatments. Dry mass of weeds (DMW) and percentage of cover (COV) were evaluated in three seasons of coexistence of the crop with weeds: 0-10, 0-20 and 0-30 DAE with readings on a grid of 0.25 × 0.25 m, in each experimental unit. The composition of the weed community was recorded, according to the taxonomic classification, and the density of weeds per m² (DW); and the index of occurrence (IO) was calculated as the percentage per species concerning the total found.

Analysis of data

For the yield variable, analysis of variance was carried out by semester and combined, orthogonal contrasts and linear, quadratic, exponential, logarithmic, potential and non-linear (logistic) regression methodologies. The orthogonal contrasts tested and estimated were: C1: weeds control vs. coexistence with weeds; C2: control 10, 20 and 30 DAE vs. control 50 DAE; C3: control 10 and 20 DAE vs. control 30 DAE; C4: control 10 DAE vs. control 20 DAE; C5: coexistence 10, 20, 30 DAE vs. coexistence 50 DAE; C6: coexistence 10 and 20 DAE vs. coexistence 30 DAE; C7: coexistence 10 DAE vs. coexistence 20 DAE. For the DMW and COV variables, analysis of variance and Tukey's multiple comparisons at 5%. Compliance with the assumptions of normality, homogeneity of variances and additivity was verified for the individual and combined analyzes of variances. The regression models were evaluated according to the criteria: Anova, coefficients of determination, residual analysis and significance of the predictors. Simple linear, polynomial, exponential, logarithmic, potential and logistic regression models were adjusted, and those with the best fit were selected for the estimation of the CPC, estimating a 5% loss in relation to the maximum yield estimated with the regression models. For the DMW and COV variables, analysis of variance and Tukey's multiple comparisons at 5%. And SAS software version 9.0 (2002) was used.

Grain yield in the dry season

The contrast of the YIELD between 0-10, 0-20 and 0-30 DAE, as a whole, and 0-50 DAE of weed control, was significant (P=0.0005), in favor of 0-50 DAE, i.e., the maintenance of weed-free cultivation (Tab. 1). This could be costly, for 50 d represent 83.3% of the crop cycle which, for the conditions of the humid Caribbean region of Colombia, is approximately 60 d. However, weeds control in the interval of 0-30 DAE was better compared to the intervals of 0-10 and 0-20 DAE, as a whole. This indicates that, under the conditions of the dry season of the year and 147 mm of precipitation during the cycle, the strategy of keeping the crop free of weeds during the first 30 d is the best. Similar results were reported by ^{Campos et al. (2023)}, but differ from ^{Cerna and León (2015)} and ^{Castro et al. (2019)}, mainly due to the environmental conditions prevailing when the evaluations were carried out, as well as the duration of the cycle of the cultivar used.

The highest YIELD, 2,547.5 kg ha^-1, was obtained by keeping the crop weed-free from emergence to 50 DAE, while the lowest YIELD, 887.5 kg ha^-1, was achieved by controlling weeds during the first 10 DAE. Therefore, the decrease in YIELD due to weed competition, with the control strategy for 50 DAE, was 65.2%, lower those that shown by ^{Campos et al. (2023)} of 90% and higher than the results obtained by ^{Castro et al. (2019)} of 39.8% in the semi-erect genotype and 37.27% in the semi-prostrate genotype of cowpea beans, differences were supported by the edaphoclimatic conditions, weed community and cultivar genetics.

According to the coexistence of crop-weeds, when contrasting the intervals 0-10, 0-20 and 0-30 DAE, as a whole, and the interval of 0-50 DAE there was significant difference (P≤0.0001) in favor of the first three intervals (0-10, 0-20 and 0-30 DAE), doubling the yield with respect to the presence of weeds for 50 d (Tab. 1). Likewise, no significant differences were found (P=0.8496 and P=0.6513, respectively) between the interval of 0-30 DAE and those of 0-10 and 0-20 DAE as a whole, nor when comparing the intervals of 0-10 and 0-20. This suggests that, under the conditions of the dry season, the strategy of sowing and maintaining the crop in coexistence with weeds could be viable up to 30 DAE, which would represent a decrease in production costs, flowering without problems and a greater number of pods per plant (^{Castro et al., 2019}).

With the YIELD for each time interval and the two weed management strategies, the best-fit equations were estimated and selected, which were linear for coexistence and potential for control (Fig. 1). These models show the variation of grain yield as a function of the DAE. By estimating 5% losses in relation to the maximum grain yield estimated with the models, the CPC was 14 to 33 DAE. For this species, ^{Lacerda et al. (2020)} and ^{Campos et al. (2023)} in Brazil, have reported CPC of 21 to 32 DAE, while ^{Osipitan (2017)}, places it within the first 40 DAE. Despite being the same species, these differences in terms of the location of the CPC in the cowpea life cycle can be attributed to the climatic conditions of each place, the genetics of the cultivar and the fact that the competition of weeds are closely related to the seed bank of each soil.

Figure 1. Functional relationship between cowpea yield and weed control intervals and coexistence with the crop, end of the dry season of semester B of 2022. 

Grain yield in the rainy season

The contrast of the YIELD between 0-10 and 0-20 DAE, as a whole, and 0-30 DAE, was significant (P=0.0154), in favor of 0-30 DAE, i.e., keeping the crop free of weeds for 30 DAE (Tab. 2). This difference was further evidenced by the non-significance (P=0.5348) of the contrast between the weed control in 0-10 and 0-20 DAE. This indicates that, under the conditions of the rainy season of the year, the strategy of keeping the crop free of weeds during the first 30 DAE, leads to a higher YIELD compared to the first 10 or 20 DAE.

The highest YIELD, 553.5 kg ha^-1, was obtained by keeping the crop weed-free, from emergence to 50 DAE, while the lowest YIELD, 316.9 kg ha^-1, was achieved by only controlling weeds during the first 10 DAE. Consequently, the decrease in the YIELD due to the interference of weeds with the control strategy for 50 DAE was 42.75%, possibly due to competition for light and nutrients, increased presence of pests and diseases, and allelopathic effects caused by fungi and weeds that affect grain production and quality, since the exudates affect the growth of the radicle, size and vigor of the cowpea seedling (^{Lacerda et al., 2020}; ^{Al-Deliamy and Abdul-Ameer, 2023}).

According to the coexistence strategy, the contrast between the treatments 0-10, 0-20 and 0-30 DAE, as a whole, and 0-50 DAE was significant (P≤0.0004), in favor of the first three treatments (Tab. 2). This result is complemented by the significance (P=0.0095) of the contrast between the treatment 0-30 DAE and the treatments 0-10 and 0-20 DAE, taken together, in favor of coexistence during the first 10 or 20 DAE. This suggests that, under the conditions of the rainy season of the year, the strategy of maintaining the crop in coexistence with the weeds was effective until 20 DAE.

The decrease in YIELD due to the competition of weeds with the coexistence strategy was 80.46%, higher than the achieved with the control strategy. Similar results have been shown by ^{Lacerda et al. (2020)}, with a decrease in cowpea yield by 73.5% when grown with weeds throughout its cycle, under semi-arid conditions, in Brazil. Likewise, ^{Osipitan (2017)} reported a 76% decrease due to weeds in Africa.

With the YIELD for each time interval in both weed management strategies, the best-fit equations were estimated and selected (Fig. 2). These models show the variation of grain yield as a function of the days after emergence. By estimating 5% losses in relation to the maximum grain yield estimated with the models, the CPC was 14 to 29 DAE.

The combined analysis of variance of the YIELD obtained in the two seasons, dry and rainy, shows significant differences for season of the year (P≤0.0001), between treatments (P=0.0235) and in the season-treatments interaction (P=0.0386). This interaction shows that the yield varies according to the season of the year, with a better performance in the dry season, since the aggressiveness of the weeds is reduced by less water availability in the soil. In both seasons, the weed control strategy resulted in higher yields at 30 and 50 DAE, while with the coexistence strategy, it was at 10 and 20 DAE (Fig. 3).

FIGURE 2. Functional relationship between cowpea yield and weed control intervals and coexistence with cultivation, end of the rainy season, 2023A. 

FIGURE 3. Season-treatment interaction for cowpea bean yield and weed control intervals (A) and coexistence (B), dry season 2022B and rainy season 2023A. 

Composition of the weed community

During the two experimental seasons, and cumulatively, 18 species of weeds belonging to 11 families were recorded. Table 3 shows the taxonomic classification, the number of individuals, the IO and the DW (species with at least 10 individuals/m²). Weeds of the Liliopsidae class predominated with 73%, over the Magnoliopsida class with 27%. In this sense, ^{Lacerda et al. (2020)}, obtained similar results when evaluating the CPC of weeds with cowpea beans, where weeds of the botanical class Liliopsida and family Poaceae predominated, this behavior is attributable to environmental conditions with high temperatures and intense solar radiation, conditions that are also present around this research.

The specie Rottboellia cochinchinensis, the one with the highest IO and DW, has a recognized competitive capacity in various crops (^{Castro et al., 2019}), including cowpea beans. Likewise, ^{Peerzada et al. (2016)} and ^{Shabbir et al. (2019)} agreed in stating that Echinochloa colona is a weed responsible for yield losses of up to 50% in crops, its success is attributed to its prolific seed production, rapid growth, allelopathic power and adaptability to a wide range of environments, which leads to decreased number of pods, grain mass and yield in cowpea (^{Maia et al., 2021}).

Additionally, ^{de la Cruz-Zapata et al. (2016)} reported Urochloa fusca as a host of Aeneolamia contigua and Prosapia simulans, considered insect pests in several crops the weed species is very efficient in the accumulation of nutrients and their level of interference depends on the weed community present (^{Marques et al., 2019}). Furthermore, Euphorbia heterophylla has been reported to be more efficient than soybean in the use of nitrogen absorbed from the soil (^{Silva et al., 2015}).

The above explains why inadequate management of weeds negatively influences the chlorophyll content, pod production and protein content in the grain (^{Coelho et al., 2019}).

Dry mass of weeds and cover (%)

The analysis of variance showed significant differences between the two study seasons (P=0.0024), as well as between treatments (P=0.0066) for DMW (Tab. 4). The difference between seasons can be attributed to the frequency and intensity of rainfall, given that the first experiment was set in the dry season of 2022, and the 147 mm of rainfall during the crop cycle, whereas in the rainy season of 2023, rainfall of 327 mm during the cycle was frequent and intense. Similar results were reported by ^{Yadav et al. (2017)} in cowpeas, in India, and highlighted that the weed problem is more serious during the rainy season, since weeds grow very quickly, compete for light, nutrients and space, causing a reduction in yield.

The treatment with the highest DMW, in both seasons, was coexistence during the first 30 DAE (Tab. 4), due to the longer time for the growth and development of the weeds due to the possible allelopathic effect they exert on the bean, which associated with their slow growth affects the final yield and this shows the need for efficient weed control in the first days. Similar results were found by ^{Grazziero et al. (2019)}, who reported that as the dry mass of weeds increases, it shows a linear reduction in grain yield, thus 200 g m^-2 of DMW caused a reduction between 223 and 722 kg ha^-1 in soybeans.