Introduction

The mutualistic relationship between ants and some hemipterans, such as aphids and mealybugs, favors the growth of their populations due to the care that ants provide them, since they prevent the action of parasitoids and predators (Calabuig et al., 2015; Sankovitz et al., 2024). This type of association is common in grape cultivation, where an increase in the population of Planococcus ficus (Signoret) (Hemiptera: Pseudococcidae) has been detected (Mgocheki & Addison, 2010; Parrilli et al., 2021; Sime & Daane, 2014). Therefore, ant management is essential to reduce mealybug populations.

Ant control using contact insecticides has limitations due to their persistence on foliage or fruit (Keklik et al., 2024), which represents risks for beneficial fauna (Mansour et al., 2018) and human health (Alokail et al., 2024). Furthermore, these products only affect foraging worker ants at the time of application, and they are quickly replaced by new workers generated by the queen (Daane et al., 2006; Mgocheki & Addison, 2009a).

A promising alternative is the use of toxic baits, as they reduce the impact on other insects (Suiter et al., 2021) and allow workers to introduce the toxic bait to the nest (Hoffmann et al., 2023; Klotz et al., 2003). This method uses low doses of insecticides, but they are lethal to the colony (McCalla et al., 2020; Nelson & Daane, 2007; Tollerup et al., 2004). In this way, trophallaxis behavior is exploited to distribute the toxicant among the nest population, including the queen (Meurville & Leboeuf, 2021). This mechanism allows the control for several generations (Cooper et al., 2008; Le et al., 2024; Nelson & Daane, 2007).

The use of liquid baits has been effective in controlling some ant species associated with Hemiptera in citrus and grapevines, particularly those species that forage for honeydew (Mercer et al., 2025; Parrilli et al., 2021). For example, in California, USA, alginate hydrogels have been evaluated to control the Argentine ant (Linepithema humile [Mayr]), which is mutualistically associated with mealybugs (McCalla et al., 2020; Mercer et al., 2025). The hydrogel (250 g) was soaked in a solution containing sugar water (0.25 %) and thiamethoxam (0.0001 %), applied every three weeks at the base of citrus trees, and reduced the presence and activity of L. humile by 70 % (McCalla et al., 2020). Likewise, the application of 250 g of hydrogel, soaked in sugar water (0.25 %) and spinosad (0.01 %), on three occasions (weeks 1, 4, and 7) decreased the population and activity of L. humile, compared to treatments with fewer applications (Milosavljevic et al., 2024).

On the other hand, solid baits have been more effective in attracting and controlling protein-foraging ant species, such as generalist predators or scavengers of arthropods, that also feed on honeydew or nectar (Tollerup et al., 2004, 2007). For example, a solid bait composed of anchovy powder, corn cob (20 g), and the insecticide imidacloprid (0.005 %) was effective in reducing the presence of the desert ant Formica perpilosa on grapevines in the Coachella Valley, California (Tollerup et al., 2004).

In some vineyards of Valle de Guadalupe in Ensenada, Baja California, Mexico, P. ficus is considered the main pest of the grapevine, with variable levels of infestation and presence in approximately 2 100 ha of that municipality (Baja California State Committee for Plant Health [CESVBC], 2022). Recently, 10 ant species were identified in mutualistic interactions with P. ficus, being Formica perpilosa and F. francoueri the most abundant (Castro-Alvarez et al., 2023). Therefore, developing control strategies targeting these species is a priority. Given this, the objective of this study was to evaluate the effect of toxic baits in bait stations for controlling ants associated with P. ficus in vineyards in Ensenada, Baja California.

Materials and methods

Experimental plots

The experiment was conducted in vineyards in Ensenada, Baja California, Mexico, with the presence of Formica perpilosa and F. francoeuri, according to samples taken in December 2021, and January, June, and August 2022. Likewise, to homogenize the experimental conditions, vineyards with the Cabernet Sauvignon variety were chosen. The experimental plots were distributed in five vineyards in the Francisco Zarco area, two with organic management: Anatolia (32° 07’ 18.9’’ N, 116° 32’ 27.4” W) and Viña Alegre (32° 06’ 49.8” N, 116° 29’ 51.2” W), and three with conventional management: Olé (32° 06’ 53.3” N, 116° 31’ 12.0” W), Paralelo (32° 06’ 42.7” N, 116° 32’ 05.7” W) and Agua Honda (32° 07’ 34.5” N, 116° 31’ 06.7” W). In each vineyard, 3 ha were delimited to evaluate the treatments.

Toxic baits

Based on documentary information on the control of F. perpilosa in California (Tollerup et al., 2004, 2007) and previous trials in Ensenada, a bait was prepared with two attractants: protein (70 %) and carbohydrates (30 %), plus a low concentration of insecticide (organic or conventional). The bait preparation consisted of mixing 35 g of ground shrimp (Barajas® Products) and 15 g of granulated sugar, plus a toxic product: a) 1 % boric acid, b) 0.08 g of natural pyrethrin (PYREMAX® 0.2 % PH) or c) 0.0002 % of thiamethoxam (Actara® 25 WG) (Cooper et al., 2008; Tollerup et al., 2004).

The dry bait components were mixed in a 3 L plastic container, sufficient for 25 stations of each treatment (1 250 g). The mixture was homogenized with constant stirring for 1 min. Subsequently, 50 g portions were packaged in brown paper bags (thickness no. 2), which were placed in 250 mL plastic containers (11 cm diameter × 7 cm height), with their respective lids. The containers with the prepared bait were taken to the field for establishment.

Establishment of bait stations

In the organically managed vineyards (Anatolia and Viña Alegre), pyrethrin, boric acid and a control were evaluated, while in the conventionally managed vineyards (Olé, Paralelo and Agua Honda), thiamethoxam, boric acid and a control treatment were evaluated. In all cases, the control treatment consisted of bait stations with food attractants (protein and sugar) without insecticide. Treatments were distributed in a completely randomized design.

The vineyards had a separation between rows of 3.5 to 4.0 m, and 2.0 m between plants. Three experimental units of 1 ha each were delimited in each vineyard, in which 25 bait stations were placed per treatment. Treatments were defined by the type of insecticide used and included a control without insecticide. Each station consisted of a 250 mL container with 50 g of bait. Each container had four U-shaped incisions (1.5 × 6 × 1.5 cm) to allow ant access.

In each hectare, stations were evenly distributed at the base of the vine trunks. For operational purposes, five stations were placed per row, distributed every five rows (approximately every 20 × 20 m). Within each vineyard, experimental units were separated by at least 60 m to avoid interference between treatments. This is because Formica typically forages 15 to 20 m from the nest, although it can be longer depending on available resources (Tollerup et al., 2007).

Ant abundance assessment

The effect of the treatments was evaluated at 10 randomly selected stations per experimental unit. Two evaluations were conducted: one prior to bait placement and another 30 days later. At each evaluated station, five vine plants were inspected: the central plant (where the bait was placed) and four plants taken from each cardinal point, yielding a total of 50 plants per hectare. In each plant, three strata (root, stem, and branches) were examined, in a section of approximately 20 cm per stratum. The abundance of Formica ants was classified into four levels: 0 (absence), 1 (low, 1-3 ants), 2 (medium, 4-10 ants), and 3 (high, more than 10 ants). This was in accordance with the criteria established by the Baja California State Plant Health Committee.

Identification of ant species

Ant specimens were collected from each experimental plot for identification, and in the laboratory we used specialized taxonomic keys (Hölldobler & Wilson, 1990; Mackay & Mackay, 2002). Reference specimens were deposited in the insect collection of the Colegio de Postgraduados.

Data analysis

Ant abundance was reported in graphs as the weighted percentage of infestation (WPI), which was obtained by multiplying the value of each category (0, 1, 2 and 3) by its respective frequency, and then dividing the sum of these products by the product of the maximum category (3) and the total number of plants evaluated (50). Finally, the result was multiplied by 100 (adapted from Townsend & Heuberger, 1943).

The statistical analysis was performed in two stages. In the first, ordinal regression models were fitted to compare the treatments with the control. This comparison was performed both at the pre-assessment and at the 30-day follow-up. Ordinal regression models were used to estimate coefficients (β), odds ratios (ORs), and p-values, using the control as the reference point. In the second stage, the previous assessment was used as a reference to analyze changes in ant abundance after 30 days. For this stage, ordinal regression models were fitted for each treatment. Only vineyards with a constant presence of ants during the experiment (Agua Honda, Paralelo, and Anatolia) were considered. The Olé and Viña Alegre vineyards were excluded from the analysis, as the irregular presence of ants made objective comparisons between treatments difficult. Model fitting was performed using the polr function of the MASS package (Venables & Ripley, 2002) in the R programming language version 4.2.1 (R Core Team, 2025).

Results and discussion

All ant species collected were F. perpilosa or F. francoeuri, which is consistent with the report by Castro-Alvarez et al. (2023) for the same region. In the conventionally managed vineyards (Agua Honda, Olé, and Paralelo), no significant differences in ant abundance were found among treatments in the previous assessment. In Agua Honda and Paralelo, insecticide treatments showed similar abundance to the absolute control (p > 0.05). In Olé, the boric acid treatment showed slightly higher abundance than the control, although without significant differences (β = 0.178, OR = 1.195, p = 0.5256); similarly, the thiamethoxam treatment did not present a significant difference (β = −0.539, OR = 0.583, p = 0.0908). These results indicate that initial conditions were homogeneous, which allows subsequent changes to be attributed to the effects of toxic baits.

Ground shrimp powder as a protein source proved to be an attractive bait for Formica under field conditions (unpublished data). This finding supports previous work suggesting that solid baits are more effective against Formica sp. (Tollerup et al., 2004). However, the frequency of bait replacement must be considered due to the humidity in the vineyard region of Ensenada. In this study, the 30-day period was adequate, as the bait retained its characteristics.

The addition of granulated sugar was intended to attract other ant species associated with P. ficus, as has been done for ant control in the United States (Klotz et al., 1998, 2003; Nelson & Daane, 2007). However, no other ant species were recorded.

In conventionally managed vineyards, significant differences (p < 0.05) were observed in some cases 30 days after treatment application. In Agua Honda, boric acid treatment significantly increased ant abundance (β = 0.443, OR = 1.558, p = 0.0428), while thiamethoxam reduced ant abundance (β = −0.761, OR = 0.467, p = 0.0021), compared to the control. At Olé, no treatment showed significant effects: boric acid (β = −0.255, OR = 0.775, p = 0.4756) and thiamethoxam (β = −0.058, OR = 0.944, p = 0.8665), indicating abundance levels similar to those of the control. At Paralelo, both treatments significantly reduced abundance: boric acid (β = −2.071, OR = 0.126, p < 0.0001) and thiamethoxam (β = −1.399, OR = 0.247, p < 0.0001). These results show variable effects depending on the experimental site, highlighting significant reductions in Agua Honda (with thiamethoxam) and Paralelo (with both treatments), while in Olé no significant changes in ant abundance were observed (Figure 1).

Ant abundance in the pre-assessment of organically managed vineyards did not differ between the boric acid and natural pyrethrin treatments compared to the control. In Anatolia, the boric acid treatment showed slightly lower abundance than the control, and natural pyrethrin showed a similar trend, but in neither case was a significant difference recorded (boric acid: β = −0.2459, OR = 0.782, p = 0.2906; pyrethrin: β = −0.1079, OR = 0.898, p = 0.6318). In Viña Alegre, boric acid treatment recorded an abundance similar to the control (β = −0.0323, OR = 0.968, p = 0.8973), while natural pyrethrin recorded a higher frequency in higher abundance categories (β = 0.2773, OR = 1.320), which represented a 32 % increase in the probability of infestation compared to the control, although these differences were not significant either (p = 0.2524). The initial conditions at Viña Alegre did not allow for an objective comparison of treatment effects; therefore, this site was excluded from the plant stratum analysis.

At 30 days after bait exposure, variable effects were observed among treatments depending on the site. In Anatolia, boric acid significantly reduced ant abundance (β = −1.8113, OR = 0.163, p < 0.0001), whereas natural pyrethrin had no significant effect (β = −0.2151, OR = 0.806, p = 0.3106). In Viña Alegre, none of the treatments had significant effects compared to the control: boric acid (β = 0.1692, OR = 1.184, p = 0.5215) and natural pyrethrin (β = 0.1839, OR = 1.202, p = 0.4916). These results highlight a significant decrease in Anatolia with boric acid, while in Viña Alegre no significant changes were observed (Figure 2).

The low concentrations of the insecticides used in the bait were based on research in vineyards with ant problems in California, USA (Tollerup et al., 2004, 2007). These slow acting formulations act through trophallaxis, which allows the active ingredient to reach the interior of the nest.

Given the biology of the ant species involved and the mode of action of these baits, the evaluation period was brief (30 days); however, significant results were observed. It is recommended that evaluations be conducted over a longer period, three to six months, and that further improvements be made to the bait formulation. In previous studies, the effects of boric acid on ant reduction took at least 10 weeks (Klotz et al., 1998).

Thiamethoxam and boric acid have been evaluated and retailed for the control of Linepithema humile and Formica perpilosa in citrus and vineyards (Klotz et al., 2003; McCalla et al., 2020; Tollerup et al., 2004). The concentrations used in this study were higher than those used in liquid baits in California vineyards: 0.5 % boric acid and 0.0001 % thiamethoxam (Cooper et al., 2008; Daane et al., 2006; Greenberg et al., 2006), but similar to those used in hydrogel baits (McCalla et al., 2020). It is important to consider that the concentration of active ingredient may decrease during the trophallaxis process (Rust et al., 2004). In this study, both ingredients demonstrated remarkable effectiveness, representing an alternative for both conventional and organic management, although further evaluation is needed before issuing specific recommendations for Ensenada vineyards.

In Agua Honda and Paralelo vineyards, analysis of ant abundance in the three plant strata (root, stem, and branches) revealed differences between treatments. The control treatment showed an upward trend in ant numbers in all three strata compared to the previous evaluation, although this was not statistically significant (root: β = 0.096, OR = 1.101, p = 0.6608; stem: β = 0.038, OR = 1.038, p = 0.8763; branches: β = 0.430, OR = 1.537, p = 0.1088).

In boric acid and thiamethoxam treatments, significant decreases in ant abundance were detected in most strata (Figure 3). Boric acid treatment caused a significant reduction in all three plant strata (root: β = −0.5003, OR = 0.606, p = 0.0245; stem: β = −0.8181, OR = 0.441, p = 0.0051; branches: β = −0.536, OR = 0.585, p = 0.0011), with decreases of 29.4, 55.9, and 41.5 %, respectively, compared to the previous assessment. Likewise, treatment with thiamethoxam recorded a decrease in ants in the root, although without statistical significance (β = −0.281, OR = 0.755, p = 0.3264). However, significant reductions were recorded in stem and branches (stem: β = −1.033, OR = 0.356, p = 0.0066; branches: β = −1.310, OR = 0.270, p = 0.0082), with decreases of 64.4 and 73.0 %, respectively (Figure 3).

These results suggest that the treatments had variable effects depending on the stratum, likely due to variations in the structure or dynamics of ant colonies, which were not detected in the previous sampling. However, the significant and consistent reductions in ant abundance in two commercial vineyards support the relevance of continuing this type of trial in the region.

In relation to the organically managed vineyards, only Anatolia had adequate numerical consistency to analyze ant abundance by strata. Overall, the control vineyard showed a significant increase in ant abundance in all three strata compared to the previous assessment (root: β = 0.6525, OR = 1.920, p = 0.0080; stem β = 1.403, OR = 4.067, p = 0.0006; branches: β = 1.446, OR = 4.246, p = 0.0061). In contrast, boric acid treatment significantly decreased ant presence at the root (β = −1.1300, OR = 0.3230, p = 0.0035), representing a 67.7 % reduction compared to the previous assessment. However, no significant effects were observed in stems (β = −0.5480, OR = 0.5781, p = 0.2318) or branches (β = −0.1981, OR = 0.8203, p = 0.7398). Natural pyrethrin did not reduce ant abundance. In fact, an upward trend was observed in all three strata, although it was only statistically significant in the root stratum (β = 2.0760, OR = 7.9725, p < 0.0001). In the stem (β = 0.4837, OR = 1.6221, p = 0.1980) and branch (β = 0.2287, OR = 1.2570, p = 0.5967) strata, the differences were not significant compared to the previous assessment (Figure 4).

These results show that ant abundance varied between conventionally and organically managed vineyards. In the pre-assessment, abundance levels were higher in organically managed vineyards, which could explain the increase observed in the control between the pre-assessment and the 30-day assessment. The increase in ant numbers could be due to the release of pheromones during the establishment of foraging trails (Greenberg & Klotz, 2000). Furthermore, ant activity levels tend to increase when initial populations are high (Nelson & Daane, 2007).

Natural pyrethrin did not significantly reduce ant populations; this could be attributed to at least three factors: 1) higher ant abundance at the start of the evaluation, 2) natural pyrethrin has no effect on ants in toxic baits, and 3) it takes longer to perceive its effect on the ant population. Klotz et al. (1998), evaluating the effect of boric acid, imidacloprid, and thiamethoxam on the control of L. humile, observed that boric acid reduced the ant population by 80 % after 10 weeks.

It is important to note that this study only tested one product formulated as natural pyrethrin in organic vineyards. If this group of insecticides is to be continued, it is necessary to find one compatible with solid matrices for use in toxic baits. On the other hand, boric acid had an effect on the roots in organically managed vineyards, and the results were consistent with those in conventional systems. This compound represents a promising alternative for organic farmers, justifying its evaluation in future research in the winegrowing region.

The use of 25 bait stations per hectare exceeded the density recommended by Daane et al. (2006), who suggested using between five and twenty dispensers per hectare. However, the optimal density may vary depending on the size of the ant population (Cooper et al., 2008), since high ant densities could reduce the effectiveness of a limited number of stations (Daane et al., 2006). Therefore, more bait stations were used in the present study than recommended in the literature.

In vineyards with P. ficus problems, it is essential to consider various factors in bait development, as well as varying their density, to reduce the incidence of ants (Milosavljevic et al., 2024; Rust et al., 2004). These actions would facilitate the implementation of more effective biological control programs against this insect, since it has been proven that natural control of P. ficus can be more effective in the absence of ants, which contributes to more sustainable crop management (Cocco et al., 2021; Mgocheki & Addison, 2009b).

Conclusions

Ground shrimp as a protein source in baits proved to be attractive to ants of the genus Formica. Baits formulated with thiamethoxam and boric acid reduced the incidence of ants in vineyards, although their performance varied depending on the plant stratum. It is necessary to evaluate other attractants and active ingredients to improve the control of ants associated with P. ficus in vineyards in Ensenada, Baja California.