Weed control in melon with preemergence herbicides

– The objective of this work was to evaluate the effectiveness and selectivity of herbicides applied at preemergence to the melon ( Cucumis melo ) crop. The experiments were carried out from 2017 to 2018, in a randomized complete block design, with four replicates, in the state of Rio Grande do Norte, Brazil. The treatments consisted of the preemergence application of: 35 g ha -1 a.i. flumioxazin, 25 g ha -1 a.i. flumioxazin, 250 g ha -1 a.i. sulfentrazone, 480 g ha -1 a.i. metribuzin, 240 g ha -1 a.i. oxyfluorfen, 240 g ha -1 a.i. oxyfluorfen + 960 g ha -1 a.i. S-metolachlor, 360 g ha -1 a.i. clomazone, and 360 g ha -1 a.i. clomazone + 240 g ha -1 a.i. oxyfluorfen; in 2018, metribuzin did not show selectivity and was substituted by 480 g ha -1 a.i. ametryn and 1,600 g ha -1 a.i. diuron. Fruit yield and quality, as well as weed control and dry mass, were evaluated. The metribuzin, ametryn, and diuron herbicides caused melon plant death. The best efficiency in weed control was obtained with the mixture clomazone + oxyfluorfen, followed by oxyfluorfen, oxyfluorfen + S-metolachlor, and clomazone. The treatments with oxyfluorfen, oxyfluorfen + S-metolachlor, and clomazone + oxyfluorfen were not considered selective. Clomazone was the only selective herbicide in the two years of evaluation and can be an alternative for weed control in melon crops.


Introduction
Brazil is one of the largest producers of melon (Cucumis melo L.) worldwide, with a production of 540,000 tons in 2017 (FAO, 2018). The production system used usually involves raised seedbeds covered with plastic mulch, which has been widely adopted in the country because it improves yield, increases water and fertilizer use efficiency, and reduces weeds (Samtani et al., 2017).
One of the main challenges in melon production is weed control (Bairambekov et al., 2016). In mulch production systems, broadleaf and grass weeds emerge in the planting holes and between beds (Boyd, 2016). Although cucurbits are susceptible to herbicides used in other agronomic crops (Abouziena & Haggag, 2016), currently, only herbicide fenoxaprop-P-ethyl and the combination clomazone + carfentrazone-ethyl are registered for the selective control of weeds in melon (Agrofit…, 2020).
Fenoxaprop-P-ethyl is a herbicide inhibitor of the enzyme acetyl-coenzyme, a carboxylase (ACCase inhibitor) that acts in the postemergence control of grasses (Takano et al., 2020). Clomazone is a synthase inhibitor (DOXP inhibitor), which is a pivotal component for plastid isoprenoid synthesis (Cabral et al., 2017), whereas carfentrazone-ethyl causes the inhibition of the protoporphyrinogen oxidase enzyme (PPO inhibitor) (Brusamarello, 2016). The commercial premix of clomazone + carfentrazone-ethyl is registered for the pre-and postemergence control of grass and broadleaf weeds in several crops. However, this premix is not efficient against weeds considered essential and difficult to control in cropping systems in Brazil, such as Amaranthus spp. and Merremia aegyptia (L.) Urb. (Raimondi et al., 2010).
An alternative for weed control in melon crops is applying residual herbicides during the pretransplantation of melon seedlings. Herbicide application would be carried out after the formation of beds and before the addition of mulch, allowing weed control in the first seven weeks when the crop is more sensitive to the presence of weeds (Monteiro et al., 2021). It is essential to highlight that the efficiency of residual herbicides depends on the physical and chemical properties of the molecule used, the rate applied, the edaphoclimatic conditions of the region, and the weed species present in the cultivation area (Carneiro et al., 2020;Lins et al., 2021). Some herbicides -such as fomesafen and flumioxazin (protoporphyrinogen oxidase inhibitors), clomazone (carotenoid biosynthesis inhibitor), S-metolachor (celldivision inhibitor), and fomesafen + S-metolachlor chlorine or dimethenamid-P -have potential to be applied to cucurbits (Lins et al., 2018). However, it is important to evaluate the efficiency and selectivity of these herbicides in different cropping systems. For a sustainable melon production, it is key to identify selective herbicides with an acceptable weed control effectiveness.
The objective of this work was to evaluate the effectiveness and selectivity of herbicides applied at pre-emergence to the melon crop.

Materials and Methods
The experiments were conducted from October to December, in 2017 and 2018, at the experimental farm of Universidade Federal Rural do Semi-Árido, located in the municipality of Mossoró, in the state of Rio Grande do Norte, Brazil (5º3'39.8"S, 37°23'44.6"W, at 78 m altitude).
The climate of the region, according to Köppen's classification, is of the BSwh type, hot and dry, with an average rainfall of 674 mm (Alvares et al., 2013). The average data on temperature, humidity, and precipitation during the melon growing seasons were obtained at the automatic meteorological station of Instituto Nacional de Meteorologia, located in the state of Rio Grande do Norte ( Figure 1). There was no precipitation during the entire study period in both years.
The soil at the experimental field was classified as a Latossolo Vermelho distrófico (Santos et al., 2018), i.e., an Oxisol (Soil Survey Staff, 2014, with 130 g kg -1 clay, 840 g kg -1 sand, and 30 g kg -1 silt. The soil was prepared with plowing and harrowing, using a rotary hoe to form the seedbeds. According to the crop's needs, soil fertilization was carried out based on the soil analysis (Table 1) (Cavalcanti, 2008).
Each year, two separate experiments were conducted to determine the effectiveness of weed control and the selectivity of the herbicides applied before melon transplantation. The experimental design used was randomized complete blocks with three replicates. In 2017, ten herbicide treatments were selected for application (Table 2). In 2018, the treatment with metribuzin was withdrawn from the trial due to its lack of selectivity for melon plants and was substituted by two other herbicides (ametryn and diuron), totaling 11 treatments in this year. The herbicides were applied one day before the melon seedlings were transplanted. Two controls were included: weedy (no wedding) and weed-free (wedding) plots; the weed-free plots were weeded manually at weekly intervals for the entire growing season.
Pests and diseases were controlled using the pesticides recommended for the crop (Agrofit…, 2020). The drip irrigation system was adopted, with emitters presenting a flow of 1.7 L h -1 and spaced at 0.3 m. Irrigations were performed daily according to the estimated evapotranspiration (ETc) of the crop (ETc = reference evapotranspiration x Kc), where Kc corresponds to the melon development stages; the reference evapotranspiration was obtained by the Penman-Monteith equation (Allen et al., 1998). The total water depth applied during the crop cycle was 552 mm.
The herbicides were applied to the planting lines using the CO 2 pressurized research sprayer Pulverizador Pesquisa (Herbicat, Catanduva, SP, Brazil) equipped with a bar with two XR11002 spray nozzle tips (TeeJet, Aabybro, Denmark), spaced 0.5 m apart, at the height of 0.5 m from the soil. The used sprayer pump pressure of 3.0 MPa and velocity of 3.6 km h -1 allowed an application volume of 200 L ha -1 Figure 1. Average temperature and air relative humidity in the two years of melon (Cucumis melo) cultivation. Data obtained at the automatic meteorological station of Instituto Nacional de Meteorologia and using a rain gauge installed at the experimental farm, in the state of Rio Grande do Norte, Brazil.  To avoid drift, the herbicides were applied during higher relative humidity, lower temperatures, and slower wind speed. Before herbicide application, the planting lines were irrigated until the soil reached field capacity. After the herbicides were applied, black and white mulch film (Negreira, Arujá, SP, Brazil) was placed on the beds, with the white side facing up and black one facing down. The holes for planting seedlings in the mulch had a 5 cm diameter and were spaced at 0.4 m.
Seedlings of the Goldex yellow melon (Topseed, Agristar, São Paulo, SP, Brazil) were planted in 200 cell polyethylene trays, with 20 cm 3 substrate per cell, in a greenhouse, being transplanted 12 days after sowing when the first true leaf appeared and their height was approximately 5 cm. The melon seedlings were transplanted 24 hours after the application of the herbicides, at a 0.4 m spacing between plants and 2.0 m between lines, with one line corresponding to one bed. Each experimental plot had three beds that were 2.8 m long and 0.8 m wide, with 21 plants. The two external beds and the two external plants of the central bed were used as a border, and only the three plants of the effective area of each experimental plot were used for fruit harvesting.
Melon plant injury (phytotoxicity) and weed control (efficiency) were evaluated visually at 7, 14, and 21 days after herbicide application (DAA). Phytotoxicity and herbicide effectiveness were assessed visually for each plot and rated on a scale from 0% (no injury or growth reduction compared with the untreated plot) to 100% (complete plant death). Yellow melon fruits were harvested from the plants in the central bed of each plot (central area) at 65 days after transplanting. During the harvest period, the aboveground biomass of the weeds was collected in the central area of each plot; all weeds that grew during the crop cycle were collected and identified and, then, taken to a forced-air circulation oven, at 65ºC, until reaching a constant weight for later weighing and dry matter determination.
After harvest, the fruits were weighed to determine yield (kg ha -1 ). The following fruit postharvest characteristics were evaluated based on a sampling of two fruits per plot: fruit longitudinal and transverse diameters, pulp thickness and firmness, pH, soluble solids content, and titratable acidity.
The dimensions of the fruits (cm) were obtained by measuring longitudinal and transverse diameters and pulp thickness with the aid of a digital caliper. To determine pulp firmness (Newton), the fruits were divided longitudinally into two parts -median region and basal region opposite of the peduncle -, and two readings were performed in each one of them using the PTR-300 fruit hardness tester (SoilControl Table 2. Herbicide treatments applied before yellow melon (Cucumis melo) transplanting in 2017 and 2018, as well as their mechanisms of action, application rates, and chemical groups. To determine pH, soluble solids, and titratable acidity, the fruits were crushed in a blender and filtered on filter paper to obtain the juice used for analyses performed in triplicate. Soluble solids ( o Brix) were determined using the Pallete-series PR-100 digital refractometer (Atago CO., LTD., Tokyo, Japan), with automatic temperature correction, according to the methodology proposed by Zenebon et al. (2008). Titratable acidity was obtained by titrating a 10 g aliquot of juice, adding 50 mL distilled water, and titrating with a NaOH (0.1 N) solution. The end of titration was determined with the aid of a digital potentiometer (pH = 8.1), and the results were expressed as percentage of citric acid, also following the methodology of Zenebon et al. (2008). pH was measured in the fruit juice using a pH meter (Horwitz, 2005).
The weed dry matter data from the experiments in 2017 and 2018 were evaluated by the descriptive analysis and plotted using the Origin Pro, version 8.0, statistical software (OriginLab Corporation, Northampton, MA, USA). Phytotoxicity and weed control were compared at 5% probability. The crop years were assessed separately, as the ametryn and diuron herbicides were not tested in 2017.
Homogeneity of variances was assessed by Bartlett's test, and normal distribution of residues, by Shapiro-Wilk's test. The analysis of variance was performed for phytotoxicity and postharvest characteristics (fruit longitudinal and transverse diameters, pulp thickness and firmness, pH, soluble solids, and titratable acidity), with means being compared by Scott-Knott' test at 5% probability.

Results and Discussion
The main weeds that occurred in the cultivation area were: Trianthema portulacastrum L. Weed control in the two crop years was dependent on herbicide used (Figure 2). Weed biomass was reduced by 100% in both years when 35 g ha -1 a.i. flumioxazin and the clomazone + oxyfluorfen mixture were applied. In 2017, plots treated with metribuzin and oxyfluorfen In 2018, A. spinosus, W. americana, M. pudica, J. tamnifolia, and M. verticillata were unimportant, presenting little dry matter under all herbicide treatments and the control (no weeds). Clomazone + oxyfluorfen, ametryn, and diuron controlled all weeds in the experimental area. In all plots treated with herbicides, the dry matter of M. aegyptia was reduced in comparison with that of the plots with no weeding; clomazone was the least effective herbicide in controlling this species, reducing only 10% of its dry matter when compared with the control.
The association of the oxyfluorfen + S-metholachlor herbicides reduced the dry matter of M. aegyptia by 70% in comparison with the plots with no weeding, showing a control efficiency similar to that of oxyfluorfen applied alone. However, the clomazone + oxyfluorfen mixture reduced the dry matter of M. aegyptia by 100%. This highlights the existence of a combined effect of clomazone + oxyfluorfen that broadened the spectrum of weed control, effectively limiting the broadleaf and grass weeds present in the study area. This observation is supported by another research in which there was an increased control of weeds with the application of tank-mixed herbicides, some of which with several modes of action and residual activity in the soil (Boyd, 2016).
Flumioxazin, at both applied rates, and sulfentrazone reduced the dry matter of M. aegyptia by 88, 58, and 82%, respectively, compared with the plots under no weeding. However, these herbicides were ineffective in controlling I. triloba, whose dry matter increased in the treatments with 35 and 25 g ha -1 a.i. flumioxazin, in comparison with the plots in the untreated area. Therefore, the control of annual grasses was not satisfactory in the treatments with both rates of flumioxazin and with sulfentrazone, herbicides which were not effective in controlling U. plantaginea.
The weed species that predominated in the cultivation area were T. portulacastrum, M. lathyroides, M. aegyptia, and J. tamnifolia, which are herbaceous with indeterminate growth and prostate size (Fahad et al., 2014). In addition, M. lathyroides, M. aegyptia, and J. tamnifolia are climbing plants that rapidly intertwine with other ones for their growth (Moreira et al., 2018). In crops with prostrate growth habits, such as melon, the interference caused by these weeds can be even more serious (Chaney & Baucom, 2012); however, the cultivation system used in the present study, with beds covered with polyethylene (mulch), provides a mechanical barrier to weed infestation (Johnson & Mullinix, 2002). Therefore, the integration of this practice with the application of preemergent herbicides at pre-transplantation offers melon producers more effective options for weed management, as observed here.
The greater effectiveness in weed control by ametryn, oxyfluorfen, diuron, clomazone + oxyfluorfen, and oxyfluorfen + S-metolachlor over the days after their application is probably related to their residual control effect. These herbicides have moderate to high persistence in soils (University of Hertfordshire, 2019), allowing weed control for a long time after being applied (Cahoon et al., 2015). The half-life of oxyfluorfen in the soil, for example, has been reported to be between 10 and 31 days (Alister et al., 2009), and it has been observed that the mixture of oxyfluorfen with herbicides of other modes of action and residual activity in the soil improves the effectiveness of weed control (Boyd, 2016). Sulfentrazone shows a very high persistence in soils, with a half-life of 541 days (University of Hertfordshire, 2019); however, this herbicide was not effective in controlling weeds -46% control at 21 DAA -, which explains the decrease found in the percentage of control.
In the two years of cultivation, there was a reduction in the injuries caused by the herbicides in the third phytotoxicity assessment at 21 DAA, with melon plants showing a recovery behavior. This result suggests that melons can metabolize herbicides to fewer toxic metabolites as they grow (El-Nahhal & Hamdona, 2017). The metabolization of herbicides by plants reduces the amount of these xenobiotics that reaches their sites of action (Bakkali et al., 2007), decreasing the damage caused. According to Qi et al. (2015), oxidative damage caused to plants by herbicides can inhibit the activity of antioxidant enzymes that play an essential role in the elimination of reactive oxygen species. Therefore, the lower the inhibition of these enzymes, the greater the metabolism and recovery of plants from the oxidative stress caused by herbicides.
The highest yield (kg ha -1 ) was registered in the weeded plots (free of weeds) in 2017 and 2018, with values of 29,333 and 25,383 kg ha -1 , respectively ( Figure 5). In this year, the plots treated with clomazone did not differ statistically from the weeded plots and the plots treated with oxyfluorfen, oxyfluorfen + S-metolachlor, and clomazone + oxyfluorfen. The treatments with 35 g ha -1 a.i. flumioxazin, 25 g ha -1 a.i. flumioxazin, and sulfentrazone, as well as the untreated plots (no weeding, i.e., with natural weed infestation), showed a significantly reduced yield when compared with the weedy plots. The lightest fruits (12,200 kg ha -1 ) were harvested in the plots treated with 35 g ha -1 a.i. flumioxazin. Flumioxazin, at both rates, and sulfentrazone caused a decrease in yield above 45%. Oxyfluorfen, oxyfluorfen + S-metolachlor, clomazone, and clomazone + oxyfluorfen reduced the yield of yellow melon by up to 24% when compared with the weeded plots. In the case of metribuzin, which caused plant death, there was no fruit production.
In 2018, the weight of the fruits in the plots without weed control was approximately 30% lower than that of the weeded plots. The oxyfluorfen, oxyfluorfen + S-metolachlor, clomazone, and clomazone + oxyfluorfen treatments did not reduce the yield of yellow melon, when compared with the weeded control plots. The weight of the fruits harvested in the plots treated with 35 g ha -1 a.i. flumioxazin, 25 g ha -1 a.i. flumioxazin, and sulfentrazone was of 18,800, 16,050, and 15,475 kg ha -1 , respectively, similar to or lower than that of 17,783 kg ha -1 of the fruits grown in competition with weeds. It should be noted that ametryn and diuron caused the death of melon plants, which led to a 100% reduction in yield.
In 2018, the application of oxyfluorfen, oxyfluorfen + S-metolachlor, clomazone, and clomazone + oxyfluorfen caused phytotoxicity in the melon plants, which, however, recovered, without having their yield affected. The loss of yield caused by 25  g ha -1 a.i. flumioxazin, 35 g ha -1 a.i. flumioxazin, and sulfentrazone was more significant than that in untreated plots due to phytotoxicity and mainly the ineffective control of weeds. In 2017, herbicide treatments influenced the quality of soluble solids in melon fruits (Table 3). However, pH, titratable acidity, pulp thickness, transverse and longitudinal diameters, and pulp firmness were not significantly different between treatments and controls (weeded and not weeded). The soluble solids content of the fruits harvested in the plots treated with 25 g ha -1 a.i. flumioxazin and with sulfentrazone was lower than that of the fruits grown in the weeded plots. Furthermore, the plots treated with oxyfluorfen were statistically equal to those in competition with weeds.
In 2018, herbicide treatments significantly influenced the content of soluble solids and the firmness of the fruits (Table 3). However, the other postharvest characteristics were not affected. Since the treatments with the application of ametryn and diuron caused plant death, there were no fruits for evaluation in this year. The firmness of the fruits was significantly lower in the plots treated with 35 g ha -1 a.i. flumioxazin, sulfentrazone, oxyfluorfen, and clomazone + oxyfluorfen in comparison with the weeded plots. The soluble solids content was of 8.03 and 8.64 °Brix, respectively, for the fruits harvested in the plots treated with clomazone and oxyfluorfen + S-metolachlor, statistically lower than that of 11.12 °Brix found for the fruits grown in the weeded plots, but close to that of 7.73 °Brix for the fruits of the plots in competition with weeds. Despite this reduction in the soluble solids content of the fruits, the values obtained in the plots treated with herbicides and in the controls (weeding and no weeding) are within the standard of 8 °Brix set for yellow melon by the United Nations Economic Commission for Europe (Unece, 2017). The high content of soluble solids in melon is considered one of the most important factors for consumer acceptance. (1) Means followed by equal letters, in the columns, do not differ by Scott-Knott's test, at 5% probability.