Red Imported Fire Ants (Hymenoptera: Formicidae) Cover the Insecticide- Treated Surfaces with Particles to Reduce Contact Toxicity

Chao Wen South China Agricultural University Liming Sheng South China Agricultural University Jian Chen USDA Agricultural Research Service Jianlong Zhang South China Agricultural University Ying Feng Forest Resources Conservation Center of Guangdong Province Zhong Wang Forest Resources Conservation Center of Guangdong Province Xuan Chen Salisbury University Jiacheng Cai Salisbury University Lei Wang South China Agricultural University Yinghao He South China Agricultural University Xiujun Wen South China Agricultural University Tao Ma South China Agricultural University Cai Wang (  wangcai@scau.edu.cn ) South China Agricultural University


Introduction
The red imported re ant, Solenopsis invicta Buren, is a signi cant urban pest that stings people and causes severe symptoms and even death (Vinson 2013; ). This species is also an important agricultural pest because it can damage various crops and livestock (Vinson 2013). S. invicta also competes with native ant species and prey on many invertebrates and vertebrates, posing great threats to the biodiversity of invaded ecosystems (Vinson 2013). Unfortunately, S. invicta has kept spreading (Ascunce et al. 2011). For example, S. invicta colonies have been reported in 390 counties of 13 provinces in China since 2024, just 16 years after it was rst found in Wuchuan, Guangdong, China ).
Two methods have been widely applied to control S. invicta. One is using the oil-based baits containing slow-acting oral toxicants (Oi 2006; ). Another method relies on the contact-based insecticides that can be used in mound and surfaces treatments (Rust and Su 2010). The contact-based insecticides can be either fast-acting that cause the rapid mortality of S. invicta, or slow-acting that provide ants with enough time and chance to horizontally transfer the insecticide among nestmates. Chen and Oi (2020) reviewed the synthetic chemicals or naturally-occurring compounds that can be used as contact-based insecticides against S. invicta. Over 30 products have been registered in US for S. invicta control.
Knowledge in the biology of the targeted insects is always critical to the success of any insect pest management programs. For contact-based re ant control products, tremendous effort has been made in studying the nature of the insecticide (e.g., toxicity and mode of action) and developing new formulations and application methods; however, less attention has been paid to the effect of ant behaviors on the insecticide e cacy. Our previous studies showed that S. invicta workers usually cover the wet, sticky, or repellent surfaces with soil particles, and they can walk on these particles to avoid direct contacting the inaccessible surfaces (Wang and Henderson 2016; Wen et al. 2016Wen et al. , 2020a. These behaviors have provoked our interest in investigating whether S. invicta workers would cover the insecticide treated surfaces with soil particles to avoid the contact with insecticides, and therefore reduce the effectiveness of the insecticides. Herein, we hypothesized that (1) S. invicta workers cover surfaces treated with contact-based insecticides, and (2) such behavior reduces the contact toxicity of insecticides. In the present study, toxicity and behavioral effects of nine insecticides (beta-cypermethrin, thiamethoxam, pronil, indoxacarb, chlorfenapyr, rotenone, spinetoram, avermectin, and chlorantraniliprole) on S. invicta workers were investigated (Table 1). Beta-cypermethrin, pronil, and rotenone have been applied as contact-based insecticides to eliminate S. invicta populations (Chen et al. 2006; Chen and Oi 2020). The potential of thiamethoxam and chlorfenapyr as contact-based insecticides against S. invicta has been con rmed by Wiltz et al. (2010). Although indoxacarb and avermectin have high oral toxicity against S. invicta (Williams 1985; Chen and Oi 2020), their contact toxicity has not yet been reported. We also evaluated two widely-applied insecticides, spinetoram and chlorantraniliprole, for their lethal and behavioral effects on S. invicta. Table 1 Common name, commercial name, percentage of active ingredient (a.i.), manufacturer, and mode of action of insecticides used in the present study. concentrations of the active ingredient (0.05, 0.5, 5, 50, 500, and 5000 ppm) were prepared by dissolving the required amount of the insecticide in distilled water, and 170 µL of the solution was added and evenly smeared on the bottom of the beaker (to reach the nal amount of 2 µL/cm 2 ). The same amount of distilled water was smeared on the bottom of the beaker as the controls. These beakers were air-dried at room temperature (22 ± 2℃) for 24 h. Right before the experiment, a sterile L-shaped cell spreader (Biglogix 65-1010, BIOLOGIX®, Jinan, China) was inserted into the plastic box containing each S. invicta colony, and workers were allowed to climb up on the surface of the pipette, which was then inserted into the beaker (~ 5 cm above the bottom) and gently shaken until 20 medium workers fall onto the bottom of the beaker. Each concentration of each insecticide was repeated 6 times (one replicate for each colony group). To prevent dehydration of ants, the beakers were placed in plastic boxes with 3 layers of paper towel (moistened using 200 mL distilled water) placed on the bottom, and the plastic boxes were capped to maintain the humidity (Mao et al. 2011 The test for each insecticide was repeated 8 times, and each colony group was tested only once. Before the test, solutions with different concentrations of the active ingredient (0.05, 0.5, 5, 50, 500, and 5000 ppm) were prepared as described above. Plastic squares (50 × 50 mm) were prepared by coating a piece of white cardboard paper with a layer of plastic membrane (Qin et al. 2019), and 50 µL solution was evenly smeared onto the square (the nal amount of the solution on each square was 2 µL/cm 2 ). To prepare the untreated (control) squares, the same amount of distilled water was added onto the square and evenly smeared. These squares were air-dried for ~ 8 h. Right before the test, a piece of sausage (10 × 10 × 1 mm, Guangdong Shuanghui Food Co. Ltd., Qingyuan, China) was xed onto the center of each square using an insect pin (Length = 35 mm, diameter = 0.35 mm). Seven squares either smeared with the active ingredient (0.05, 0.5, 5, 50, 500, or 5000 ppm) or distilled water were placed on the plastic box with randomly assigned orders, and adjacent squares were > 1cm apart from each other ( Fig. 1A).
High-resolution pictures were taken for each square at 15, 30, 45, 60 min, and 3, 6, 12, 24 h into the experiment. The number of ants on each square was counted at 15, 30, 45, and 60 min, or until the food was completely transported away from the square within 1 h. The data obtained from different time points were averaged to determine the foraging activities of ants within 1 h (Wen et al. 2020a). The presence of sausages on each square was recorded at 1, 3, 6, 12, and 24 h. Pictures taken at the end of the experiment (24 h) were analyzed to measure the area of squares covered by particles using the regionprops function of MATLAB (MathWorks Inc., Natick, MA, USA). Also, particles on each square were collected and oven-dried at 50℃ for 5 d. The dry weight of particles was measured using a 0.1-mg electronic balance (Mettler Toledo®, Switzerland).

Behavioral Effects of Insecticide-Treated Surfaces on Solenopsis invicta Under Field Conditions
This study aimed to investigate the effect of insecticide-treated surfaces on the foraging and particle-covering behaviors of S. invicta workers under eld conditions. The experiments were conducted in Zengcheng Teaching and Internship Base of South China Agricultural University from 19-23 April 2021. Twenty S. invicta mounds were randomly selected, and weeds and rocks around mounds were removed 4 d before the experiments. Solution of each insecticide (500 or 5000 ppm) was prepared, and 50 µL solution was evenly smeared onto the square. For each concentration (500 or 5000 ppm), squares either treated with insecticide or distilled water were placed 20-cm apart from the edge of the S. invicta mound (Fig. 1B). The order and direction of squares were randomly assigned, and adjacent squares were equidistant from each other. Each concentration was repeated 10 times (each mound was tested only once). The average number of foraging ants within 1 h, the presence of food at 1, 3, 6, and 24 h, and dry weight and covered areas of relocated particles at the end of the experiment (24 h) were recorded as described above.

Effect of Particle-Covering Behavior on Toxicity of Contact-Based Insecticides against Solenopsis invicta
Our laboratory and eld study showed that S. invicta workers transported signi cantly more particles onto the squares smeared with pronil or rotenone compared with the control ones (see results). Herein, we conducted a laboratory study to investigate whether such particle-covering behaviors would reduce the toxicity of pronil or rotenone against S. invicta.
Six colony groups of S. invicta were collected from Tianlu Lake Park on 24 February 2021, using the same method as described above. Topsoil was collected from the Zengcheng Teaching and Internship Base of South China Agricultural University where S. invicta activities were detected. The soil was oven-dried at 50℃ for > 5 d. The dried soil was crushed using a wooden hammer and sequentially sifted through 0.5-mm and 1-mm sieves (our preliminary study showed that the diameters of most particles relocated by S. invicta workers were ranging from 0.5-1 mm). Before experiments, the required amount of distilled water was added into the soil and thoroughly mixed to prepare the 10%-moisture (w/w) particles.
The experiment for each insecticide contained 7 treatments (Table 2), and each treatment was repeated 6 times (one replicates for each colony group). One hundred and seventy microlitres of insecticide solution (500 or 5000 ppm) or distilled water was evenly smeared on the bottom of beakers with the inner wall coated with Te on, and air-dried for > 12 h. Certain amounts of soil particles (Table 2) were evenly added onto the bottom of beakers (the density of particles was equal to the mean density of particles relocated onto the squares treated with pronil or rotenone at the concentration of 500 or 5000 ppm in the laboratory study). Twenty medium S. invicta workers were released into each beaker and maintained in the plastic boxes with moistened paper towels as described earlier. The mortality of ants was recorded at 1, 2, 3, 4, 5, 6, 7, 8, 12, 24, 36, 48, 60, and 72 h for rotenone. Since all ants exposed to pronil-treated surface (whether covered by particles or not) died within 12 h, the mortality of ants was no longer recorded after that time point.
At the end of the experiment (24 h), few food items that placed on the squares smeared with beta-cypermethrin (500 or 5000 ppm), thiamethoxam (5000 ppm), or pronil (5000 ppm) were transported by ants (Fig. 4). Signi cantly more particles (measured in both weight and covered area) were found on squares smeared with rotenone (5000 ppm) or avermectin (5000 ppm) compared with untreated squares (Figs. 5 and 6; statistical results are shown in Supp. Materials: Table S3 and S4). In addition, signi cantly heavier particles were found on the squares smeared with pronil (50, 500, or 5000 ppm) compared with the untreated squares (Fig. 5), but the particle-covered areas on pronil-treated and untreated squares were similar (Fig. 6). The weight and covered-area of particles on the squares smeared with 5000-ppm indoxacarb were signi cantly greater than squares smeared with 50-ppm indoxacarb, but they were not signi cantly different from the untreated squares (Fig. 5, 6). Ants also relocated signi cantly heavier particles onto the squares smeared with 5000-ppm spinetoram than that smeared with 0.05-ppm spinetoram, but they were not different from the untreated squares (Fig. 5). No difference in particle weight and covered areas were detected among untreated squares and squares treated with different concentrations of beta-cypermethrin, thiamethoxam, chlorfenapyr, or chlorantraniliprole (Figs. 5, 6).

Behavioral Effects of Insecticides on Solenopsis invicta Under Field Conditions
At the concentration of 500 ppm, beta-cypermethrin and pronil signi cantly reduced the number of foraging ants within 1 h (Fig. 7). No food placed on the square treated with beta-cypermethrin was transported by ants at the end of the experiment, whereas food on other squares was actively foraged by ants (Table 4). In addition, ants relocated signi cantly heavier particles onto the squares smeared with pronil, rotenone, or avermectin than the untreated squares (controls), but only pronil caused signi cantly more particle-covered area compared with the controls (Fig. 7).  At the concentration of 5000 ppm, beta-cypermethrin, pronil, chlorfenapyr, and rotenone signi cantly reduced the number of foraging ants within 1 h (Fig. 7). At 24 h, few food items on the square treated with beta-cypermethrin or pronil were transported by ants (Table 4). Ants relocated more particles (measured in both weight and covered area) onto the squares smeared with rotenone compared with the controls, whereas beta-cypermethrin signi cantly decreased the particle weight and covered area (Fig. 7).

Effect of Particle-Covering Behavior on Toxicity of Contact-Based Insecticides against Solenopsis invicta
All ants died within 5 h when they were exposed to the surfaces smeared with rotenone (500 or 5000 ppm) and when no particle was added (Fig. 8A). However, when rotenone-treated surfaces were arti cially covered by particles, < 50% ants died at the end of the experiment (72 h). In addition, ants had signi cantly higher mortality when exposed to uncovered surfaces smeared with 5000-ppm pronil than those arti cially covered by particles within 5 h ( Fig. 8B; statistical results are shown in Supp. Materials: Table S5). Covering the surfaces treated with 500-ppm pronil also signi cantly decreased mortality within 8 h (Fig. 8B). For both insecticides and both concentrations, ants exposed to the treated surfaces arti cially covered with particles had signi cantly longer medium lethal time (LT 50 ) compared with uncovered ones (Table 5).

Discussion
Our study showed that (i) surface treatment of nine insecticides had concentration-dependent toxicity against S. invicta workers; (ii) beta-cypermethrin (0.5, 5, 50, 500, or 5000 ppm) and rotenone (0.05, 5, 50, or 5000 ppm) signi cantly decreased the number of foraging ants in the laboratory, and beta-cypermethrin (500 or 5000 ppm), pronil (500 or 5000 ppm), chlorfenapyr (5000 ppm), and rotenone (5000 ppm) signi cantly decreased the number of foraging ants in the eld; (iii) pronil, rotenone, and avermectin signi cantly induced particle-covering behaviors of S. invicta workers under both laboratory and eld conditions; and (iv) using particles to arti cially cover the surfaces previously treated with pronil or rotenone (500 or 5000 ppm) signi cantly decreased the toxicity of these insecticides against S. invicta workers (with signi cantly higher LT 50 values).
Jr. Stringer et al. (1964) de ned the slow acting insecticide as killing < 15% S. invicta workers at one day and > 89% ants at the end of the test period (14 day). On the contrary, a fast acting insecticide caused rapid mortality of ants in a short period of time. The fast-or slow-action of insecticides is concentrationdependent. For example, beta-cypermethrin (≥ 0.5 ppm), thiamethoxam (≥ 50 ppm), pronil (≥ 50 ppm), indoxacarb (500 and 5000 ppm), chlorfenapyr (500 and 5000 ppm), and rotenone (500 and 5000 ppm) caused > 80% mortality of ants within 24 h, but the remained concentrations showed slower speeds of kill.
For the two insecticides that can be used as bait toxicants, indoxacarb caused high mortality of S. invicta workers through contacting, whereas avermectin may be not quite effective as a contact-based insecticide for S. invicta control.
Few studies evaluated the repellency of insecticides against S. invicta. Wiltz et al. (2010) reported that signi cantly lower proportions of S. invicta workers passed the bridges treated with bifenthrin to reach a food source than that passed the bridges treated with water. This result was more likely due to the rapid lethal effect of bifenthrin, which reduced the level of recruitment, rather than the repellent effect of the insecticide (Wiltz et al. 2010). Likewise, we observed some dead ants on or around the squares treated with beta-cypermethrin, which may disturb the recruitment of ants. Our study also showed that rotenone decreased the number of foraging ants under laboratory and eld conditions. In addition, the reduction in number of foraging ants caused by the high concentrations of pronil and chlorfenapyr was evident in the eld, though neither reduced the number of foraging ants in the laboratory. It worth noting that rotenone (except at the contraindications of 500 and 5000 ppm) and pronil had slower action than beta-cypermethrin (Fig. 2), and few dead ants were observed on or around the squares treated with these insecticides. Therefore, the decrease of foraging ants may be due to the "true" repellent effect of rotenone and pronil, instead of the reducing recruitment caused by rapid lethal effects of these insecticides.
The repellency of insecticides can be two-fold for re ant control. On one hand, ants may avoid walking on the surfaces treated with a repelling insecticide, and therefore they would not acquire the lethal dose of insecticide. On the other hand, insecticide with strong repellency may prevent S. invicta from entering the treated area (Rust and Su 2010). Under both laboratory and eld conditions, few food items placed on the squares treated with beta-cypermethrin (500 and 5000 ppm) and pronil (5000 ppm) were transported by S. invicta at 24 h. These results showed that the repellency of beta-cypermethrin and pronil may last longer than other insecticides. In a recent study aiming to screen effective re ant repellents, only N, N-diethyl-m-toluamide (DEET) and eugenol can completely repel S. invicta foragers for 24 h in the eld (Wen et al. 2020a). The present study showed that the repellent duration of beta-cypermethrin and pronil may be comparable to DEET and eugenol. Therefore, area or periphery treatment of high concentrations of beta-cypermethrin and pronil may effectively stop ants from entering or passing the treated area, and therefore protect people from re ant stings.
In our study, signi cantly higher levels of particle-covering behavior were triggered by the high concentrations of pronil (50, 500, and 5000 ppm), rotenone (5000 ppm), and avermectin (5000 ppm) in the laboratory. These results indicate that the particle-covering behavior is concentration-dependent. In accordance with laboratory results, S. invicta workers also covered the squares treated with high concentrations of pronil, rotenone, or avermectin in the eld. Similar behavior was observed when S. invicta workers use particles to cover the surfaces treated with re ant repellents (Wen et al. 2016(Wen et al. , 2020a. Wen et al. (2016) reported that S. invicta workers can access the squares smeared with essential balm by covering the treated zones with particles, but they failed to do so when no particles were available, indicating that particle-covering is essential for S. invicta workers to walk and search for food on the repellent surfaces. In the present study, high concentrations of rotenone (in the laboratory and eld) and pronil (in the eld) repelled foraging ants and induced the relocation of particles. However, the particle-covering behavior of S. invicta workers may not be tightly associated with the repellency of pesticides because no signi cant amount of particles were found on surfaces treated with beta-cypermethrin, which showed strong repellency against S. invicta workers. On the contrary, S. invicta workers relocated signi cantly more particles on the squares treated with the high concentration of avermectin than the control squares, but this insecticide did not signi cantly reduce the number of foraging ants. (2019) reported that cytochrome P450 of S. invicta played an important role in detoxifying pronil by transforming it into pronil-sulfone. Compared to physiological resistance with complex biochemical and genetic mechanisms, behavioral resistance usually has been considered as "simply aversion behaviors either learned or based on simple repellency or avoidance" (Zalucki and Furlong 2017). Our study showed that S. invicta workers not only avoid accessing surfaces treated with some pesticides, but also modify the treated surfaces with particles to reduce the toxicity of certain insecticides (e.g., rotenone and pronil). Some eusocial insects such as ants and termites usually bury or cover the corpses with particles (e.g., Hölldobler and Wilson 1990; Sun and Zhou 2013; Wang et al. 2013). This strategy can effectively isolate microbial pathogens within corpses, and therefore prevent disease transmission (Wang et al. 2013). As far as we know, our study provides the rst evidence that S. invicta workers also used particles to keep themselves off the insecticides, which can be considered as a unique behavioral resistance strategy in ants. We suggest to include this behavior in evaluating the effectiveness of insecticides for re ant control in future studies.    conditions. Note that few food items on the squares smeared with beta-cypermethrin (500 or 5000 ppm), thiamethoxam (5000 ppm), or pronil (5000 ppm)

Declarations
were transported at the end of the experiment (24 h).   Particle weight and covered areas on the square treated with 500-or 5000-ppm solutions of beta-cypermethrin, thiamethoxam, pronil, indoxacarb, chlorfenapyr, rotenone, spinetoram, avermectin, chlorantraniliprole, or distilled water (controls) under eld conditions. Boxes show the 25th percentile, 50th percentile (median), 75th percentile, and whiskers show the maximum and minimum value of the data. Different letters indicate signi cant differences (P < 0.05).