Geographical location of the experiment

The experiment was carried out during the autumn-winter 2016-2017 cycle in a plot of the San Lorenzo Valley in Culiacán, Sinaloa, located at 24° 27’ 27.53’’ north latitude, 107° 16’ 02.32’’ west longitude of the meridian Greenwich and an altitude of 38 m (Google Earth, 2018).

Physical characteristics where the experiment was developed. The physical characteristics of the soil were determined by laboratory analysis, from which the irrigation sheet was estimated, as shown in Table 1.

Determination of the water needs of the onion

To estimate the distribution of the irrigations throughout the crop cycle, the consumptive use of the onion was adapted to the conditions of the planting date and the environmental conditions of the area of the experiment. For this, the closest meteorological station to the study was taken as the base, which in this case was the station located in the Quila Community, Culiacán, Sinaloa (Mexico), as shown in Table 2.

Onion fertilization

For the onion fertilization the drip irrigation system was used, for which appropriate fertilizers were applied for this type of irrigation. The amount of fertilizers was applied as indicated in Table 3.

Experimental design

An experiment was carried out changing in pairs to five recommended herbicides in pre-sowing to control onion weeds. There were 16 treatments, including the control without herbicide application, as shown in Table 4. The experimental design was a randomized complete block experiment, with three replications.

The experimental unit consisted of five grooves 3.5 m long by 4 m wide. Two rows of plants with a separation of 15 cm between them and 15 cm between plants were placed in each bed. Each groove presented a drip irrigation tape, with a 20 cm dropper separation and a water consumption of 0.5 L h^-1 per dropper.

Data on weeds and cultivation were obtained from the three central furrows, leaving the furrows on the banks as a buffer zone (edge effect), as well as one meter from each end of the bed to be evaluated. In the control treatment, no manual or mechanical control of the weeds was performed. The doses applied for each type of herbicide were according to the corresponding technical data sheet. Which were used both when the herbicide was applied alone or in combination with another (Table 5).

In the Table 6 describes the mode of action and chemical family of the herbicides used in this investigation.

The order of establishment of the crop was as follows: 1) application of the herbicide mixture to the soil; 2) application of irrigation. The irrigation tape with emitters arranged every 15 cm, with an expense of 0.5 L h^-1; 3) treatment of the bulb with fungicide (thiabendazole); and 4) transplantation of the bulb.

Weed density

To estimate the density of weeds, the number of weeds of the total species per m² was counted, the weed count was performed every week, for this the central 6 m² of each experimental unit (made up of 14 m²) were considered, for which which, 6 squares of 1 m² each were formed with thread. As an inclusion criterion, the stem of the weed was considered to be within the table formed in the experimental unit (^{Kalhapure et al., 2013}).

Weed biomass

To determine the biomass of weeds, these were dried in a drying oven at 80 ºC until obtaining a constant weight according to the technique suggested by ^{Nielsen (2003}). Weed harvesting was done at the end of the crop cycle. The dry weight of the plant was obtained using an electronic scale (0.1 mg precision).

Weed control efficiency

This variable was calculated according to the procedure proposed by ^{Mani et al. (1973}); (^{Ramalingam, 2013}) from the following formula:

, where: EFM= efficiency of weed control (%); BMc= biomass of weeds (g m^-2) in the control plot. BMt= weed biomass (g m^-2) in the treated plot.

Onion crop yield. The bulbs from each useful plot were harvested and weighed, from which the weed index was calculated, under the procedure proposed by ^{Gill and Kumar (1969}); (^{Ramalingam, 2013}) from the following formula:

, where: X= yield (kg ha^-1) of the plot without weed competition; Y= yield (kg ha^-1) of the experimental unit of the evaluated treatment.

Loss of biological effectiveness of herbicides

This variable was calculated based on the weed density, measured over the course of the experiment. With the obtained data, a linear regression analysis was performed (^{Zar, 2010}), where the independent variable was time (in weeks) and the dependent variable was plant density.

Analysis of data

To analyze the data, a normality and homoscedasticity test was performed, to apply an analysis of variance of the density, biomass and onion yield; however, in the first two cases, the data did not have a normal distribution, although they were transformed. Regarding the crop yield, there was a normal distribution, but the homoscedasticity test showed that the variances were not equal. Given this situation, it was decided to carry out a non-parametric statistical analysis, applying the Kruskal Wallis test (^{Krebs, 1999}).

To analyze the loss of biological effectiveness, a linear regression was performed, where the independent variable was the combination of the different pre-emergent herbicides along a time gradient, while the dependent variable was the plant density. Xlstat statistical software for Excel (^{Addinsoft, 2018}) was used in this analysis.

Weed density

The control without application was the treatment with the highest density of weeds, with 213 plants being statistically different from the rest of the treatments (Table 7). This means that applied herbicides have some degree of control in weed emergence. Furthermore, it is shown that the most effective combination to inhibit weed emergence was the combination of Acetochlor with Flumioxazin. In addition, it is striking that Flumioxazin alone or combined is among the herbicides with the lowest density of weeds.

This suggests that the most effective herbicide within the combinations is Flumioxazin combined with Acetochlor, without statistically differing from all chemical treatments. Flumioxazin, combined with Acetochlor, improves their action because they belong to groups with different modes of action, acting better on certain weed species.

In this regard, ^{Labrada and Parker (1996}), point out that different types of herbicides are used in onion weed control, depending on their type of control. However, the other variables evaluated in this investigation must be considered to conclude which is the best herbicide combination.

Weed biomass and herbicide efficiency

The analysis showed that all the treatments were better than the control. Furthermore, it is shown that the Pendimethalin plus Acetochlor treatment had better performance with respect to the variable biomass of weeds with 0.253 kg m^-2 (Table 8).

One of the reasons that the combination of Pendimethalin with Acetochlor had less weed weight may be that they present different mechanisms of action, since the former inhibits the assembly of the micro tubules which are part of the skeleton of the cells, where they have multiple functions.

While Acetochlor belongs to the group that inhibits cell division (^{Megías et al., 2019}). Furthermore, there is a combination between the two types of herbicides, since Pendimetalin mainly inhibits root development, while Acetochlor is absorbed mainly by coleoptyl and hypocotyl.

It was also found that some herbicides were more efficient acting alone than in combination with another, such is the case where Oxifluorfen was only better than combined with Acetochlor. In this regard, in a study by ^{Villalba et al. (2010}) in the control of weeds in eucalyptus, found that the combination of Oxifluorfen with Acetochlor was less efficient than Oxifluorfen alone, in the control of Digitaria sanguinalis grass.

Onion crop yield and weed index

Table 9 shows that the best combinations in onion yield were Pendimethalin plus Acetochlor, as well as Oxifluorfen only with 2.97 and 2.92 kg m^-2 respectively. In contrast, treatment with Flumioxazin was only the herbicide with the lowest yield.

It is striking that Flumioxazin was only the treatment with the lowest weight of the bulb. In this aspect, it must be considered what dose and what formulation of the herbicide can cause toxicity to the crop depending on its phenological stage (^{Akey and Sousa, 1985}; ^{Karnataka, 2007}; ^{Ramalingam et al., 2013}). In the case of this herbicide, the technical data sheet indicates that the product should be applied about three days after transplantation.

This means that the herbicide was active at the time the bulb was placed on the soil treated with this herbicide, and that when the root arose it was damaged by the herbicide. Additionally, the University of California Integrated Management Program (^{Smith et al., 2019}), notes that the herbicide Flumioxazin is recommended only for garlic, not for onion.

On the other hand, the results also show that not using herbicides, or any other control method, reduced yields 83%. This reinforces the approach of ^{Prakash et al. (2000}), regarding the loss of yield in the onion crop, caused by competition with weeds.

Loss of biological effectiveness of herbicides

When performing the analysis of variance of the regression, statistical significance was observed, with a value of p= 0.004839. Likewise, the R² value was 0.888 and a regression slope value of 0.6477 (Figure 1), which is equivalent to 674 new daily plants per hectare.

Figure 1 Relationship between the time of application of herbicides and the emergence of new weeds. 

Regarding the density of weeds over time, it was observed that in all the treatments it decreases and then there is an increase in the density (Figure 2). This may be due to the fact that, those that emerge during the first days after the application of herbicides are born weak due to the action of the latter. Another reason may be competition for space between and among individuals of the same species (^{Begon et al., 2006}).

Figure 2 Efficiency of herbicides days after herbicide application. 

The emergence of new weed individuals, apart from being related to loss of residuality, is also associated with their characteristics, since they have a discontinuous germination process (^{Baker, 1974}); ^{Altieri (1999}), since the duration of the latency of the weed seed varies according to the biotype and the storage conditions of the seeds in the soil (^{Ferrero, 2004}).

This loss of biological effectiveness reinforces the approach that after one herbicide applied in pre-transplantation another one is applied after transplantation. Likewise, the convenience of performing manual weed control is shown (^{Kalhapure et al., 2013}). In addition to the above, the different combinations with Flumioxazin were the treatments that presented the lowest density of weeds, so it would be expected that this would be reflected in a higher weight of onions as part of the competition was eliminated; however, it was not.

This could be because this herbicide may have caused phytotoxicity in onion plants. In this regard, the technical data sheet of this product (^{Valent, 2017}) indicates that Flumioxazin should be applied without combining it with other herbicides. It should also be taken into account that the dosage and formulation of the herbicide can cause toxicity to the crop depending on its phenological stage (^{Akey and Sousa, 1985}; ^{Karnataka, 2007}: ^{Ramalingam, et al., 2013}).