Location of the experiment and crop management

The experiment was established in Chapingo Autonomous University’s Experimental Agricultural Field (19° 29’ 20.4” NL and 98° 52’ 26.7” WL, at 2,250 masl.). Planting was carried out on March 12, 2016 in 200-cavity polystyrene trays with Cosmo Peat^® as substrate. After emergence, seedlings were watered every other day, for three weeks, with 50 % Steiner nutrient solution (^{Steiner, 1984}), then the waterings were daily with 100 % Steiner nutrient solution Steiner.

The crop was established in open field conditions by transplanting on April 16, 2016 under a fertigation system in 1.2 m wide furrows. A 16 mm diameter hose with 1.5 L∙h^-1 self-compensating drippers and 33 cm dripper spacing was placed in each furrow. Bottom fertilization was applied with the commercial products urea, diammonium phosphate and potassium chloride (100-100-50, respectively). During crop development, 50 kg of urea per hectare were applied each week in irrigation. Nutrition was supplemented with applications of Bayfolan Forte^® liquid foliar fertilizer, and pests were controlled with Methomyl.

Evaluated treatments

Manual weeding and the herbicides Bensulide (PREFAR 480-E^®, 5.76 kg_i.a.·ha^-1), Halosulfuron-methyl (SEMPRA 75 GD^®, 75 g_i.a.·ha^-1) and Isoxaflutole (PROVENCE 75 WG^®, 112.5 g_i.a.·ha^-1) were evaluated. Bensulide and Isoxaflutole were applied 10 dat, Halosulfuron-methyl was applied in post-emergence of the weed (21 dat) and manual weeding (always clean control) was done at 21 and 44 dat. The herbicides were applied dissolved in water at a dose of 400 L·ha^-1 using a manual sprayer (model 425, Swissmex^®) with a hollow cone nozzle. In order to identify the weed species present at the site, an additional never-weeded control was also left in place.

Experimental design and unit

The experimental design was randomized complete blocks with 10 replicates per evaluated treatment. The experimental unit consisted of a 1.2 m wide furrow with 22 plants spaced 33 cm apart.

Evaluated variables

Yield was quantified from two fruit cuts, the first at 70 dat and the second two weeks after the first. In both cuts, the fruit yield per experimental unit and the weight of a 10-fruit sample were recorded. At the end, the values obtained in both cuts were added to determine the total yield and the average weight of 10 fruits was calculated with the corresponding data obtained in each cut.

The phytotoxicity of each herbicide on the crop was observed, along with the type of weeds that each controlled or not. For this purpose, the species present in the never-weeded control were identified and the number of individuals of each species was counted in the different experimental units where the treatments were applied. With the data obtained, the weed density was calculated as the number of individuals per square meter.

Statistical analysis

An analysis of variance of the harvest variables was carried out and, subsequently, Tukey’s multiple comparison test (P ≤ 0.05) was performed on the variables that showed a significant effect of the treatments.

The number of individuals of each species, as well as the weed density, was analyzed by the Friedman test (^{Conover, 1999}), with which the three herbicides and the never-weeded control were compared. In each case, multiple comparisons using rank sums were made to identify the best treatment.

Analysis of variance

Table 1 shows that the treatments had a significant effect (P ≤ 0.05) on yield per plant in cut one (YC1), yield per plant in cut two (YC2) and total yield per plant (TYP). For the weight of 10 fruits in cut one (W10FC1), weight of 10 fruits in cut two (W10FC2) and average weight of 10 fruits in both cuts combined (AW10FBC) there was no significant effect of the treatments (P > 0.05). The coefficients of variation presented values comparable with those obtained in other husk tomato studies, with the exception of YC2 (^{Peña-Lomelí et al., 2008}). This could be due to the fact that in the first cut only completely filled tomatoes were collected (when the fruit fills the calyx or husk), while in the second one the rest of the fruits were cut, which increased the internal variability.

Comparison of means

Table 2 shows the comparison of means test of the yield variables. For YC1, the clean control (manual weed control) and the herbicide Isoxaflutole, statistically equal to each other, were superior to the herbicides Bensulide and Halosulfuron-methyl (P ≤ 0.05), with no difference between the latter two. In YC2, the herbicide Isoxaflutole was the best, although it only significantly surpassed Halosulfuron-methyl. This suggests that Isoxaflutole remained effective in controlling weeds longer. The treatment with this same herbicide had the best total yield, which was statistically equal to the control, although of these two treatments only Isoxaflutole significantly outperformed (P ≤ 0.05) the other herbicides. As can be seen in the same table, no treatment had a significant effect on fruit size (evaluated as W10FC1, W10FC2 and AW10FBC), so the differences in yield can be explained by the fruit set, which should be higher in plants with less competition from weeds.

In the three yield variables (YC1, YC2 and TYP), the results obtained with Isoxaflutole are markedly superior to those obtained with the other two herbicides, and in no case different from those of the clean control. Therefore, Isoxaflutole can be considered as the best active ingredient for weed control in husk tomatoes, as it was just as effective as manual weeding.

Weed identification

Table 3 presents a list of the weed species found in the never-weeded control and the number of individuals of each species. As can be seen, the most abundant species were buffalobur nightshade, common purslane and red-root amaranth, followed by cheeseweed and coco-grass, while oat and chayotillo were scarce. However, the last one is relevant due to its climbing growth habit, as a single plant can cover a large part of the cultivated area.

Weed control and selectivity

In the results obtained with the Friedman test (Table 4), it can be seen that in the seven weeds found there was an effect of the treatments on the number of individuals per experimental unit (P < 0.01). When this is found, it is expected that the greatest number of individuals will be in the weed control and that it will decrease with the application of herbicides. Weed density was also affected by the treatments evaluated. Since a significant effect on this variable was found, the desirable treatment will be the one with the lowest density, as this decreases competition for space, water and nutrients.

In general, the population of different weed species was reduced with the application of herbicides. For cheeseweed, red-root amaranth, buffalobur nightshade and common purslane, which were the four weeds with the highest number of individuals in the never-weeded control, the best control was obtained with Isoxaflutole; consequently, this efficiency resulted in a lower weed density (Table 4). For its part, the coco-grass population decreased almost by half with the application of both Bensulide and Halosufuron-methyl. In the case of oat, the treatment with Isoxaflutole yielded the same number of individuals as the never-weeded control, while with Bensulide and Halosulfuron-methyl it doubled, which is contrary to expectations. A similar phenomenon occurred with chayotillo, as the number of individuals of this species was not statistically different between the never-weeded control and the Bensulide treatment, but increased slightly with Isoxaflutole and grew drastically with Halosulfuron-methyl.

In the last two species described, the increase in the number of individuals with some treatments is explained by the fact that these herbicides are not effective in controlling those particular species, but they are for others. Therefore, when the competition of the species in question is eliminated, they can be developed more fully. The most drastic case occurred with Halosulfuron-methyl on chayotillo, as it is highly invasive due to its climbing growth habit, and in this case the number of individuals grew to such an extent that the plots were practically covered.

Description of the behavior of each herbicide

Provence 75 GD^® (Isoxaflutole). This herbicide was applied in post-transplant in a band and directed to the base of the plant. When applied in pre-emergence of the weeds, it showed good control of broadleaf weeds, but did not control oat or coco-grass, and had partial control of chayotillo, which could be due to the escape caused by band application. The use of this herbicide in husk tomato is safe, since it was applied on the row of plants and did not show any damage, which makes it selective for this crop in post-transplant, which coincides with what was reported by ^{Pérez-Moreno et al. (2014)}. An additional observation is that in the area where the herbicide was prepared and the spray backpack was calibrated, there were grasses, which died after the application, so its effectiveness in the control of grasses is evident.

Prefar 480 E^® (Bensulide). Its application was made in a band directed to the base of the plant and in the borders of the row of plants. This herbicide efficiently controlled chayotillo, but did not control oat, and had partial control of broadleaf weeds (cheeseweed, red-root amaranth, buffalobur nightshade and common purslane) and coco-grass. There was no damage to the crop, as also reported by ^{Pérez-Moreno et al. (2014)} and ^{Roque et al. (1995)}.

Sempra 75 GD^® (Halosulfuron-methyl). In general, it did not show good weed control and was toxic to the husk tomato when it was applied on the plant. In particular, it had an effect on coco-grass, so it is a valuable product for the control of this weed in the husk tomato, as long as it is applied in a band and directed to the base of the plant, as also suggested by ^{Urzúa et al. (2009)}.

Strategy for weed control in husk tomato

Husk tomato is a crop sensitive to excess moisture in the soil, a condition in which it is attacked by fungi such as Fusarium oxisporum; therefore, it is necessary to make two crops and a hilling in order to promote aeration of the roots, in addition to controlling weeds between plant rows, but not within them. In this context, a strategy for effective weed control in husk tomato established by transplant could consist of the following: make a total application of Isoxaflutole in pre- or post-transplant, carry out three cultivation tasks and after the third one (approximately 40 dat) apply the herbicide in a band to "seal" the soil. It is essential that the soil has sufficient moisture or is irrigated after application to ensure that the herbicide acts efficiently. For rainfed conditions it is recommended to apply the herbicide after a rain. In soils where coco-grass is a major weed, it is recommended to apply Isoxaflutole plus Bensulide or Halosulfuron-methyl in pre-transplant.

It is important to note that Isoxaflutole is not sold in Mexico, but can be imported as Provence 75 GD® from Brazil or as Merlin 75 GD® from Central America, where it is used in sugarcane.