Sustainable Management of Acromyrmex octospinosus (Reich): How Botanical Insecticides and Fungicides Should Promote an Ecofriendly Control Strategy.

The leaf-cutting ant Acromyrmex octospinosus (Reich) causes serious damage to crops and protected areas due to its foraging activity. The main method of control of this species consists of the use of synthetic insecticides that can lead to environmental damage and negative side effects on human health. Consequently, alternative strategies, such as biopesticides, are needed. Insecticide evaluation by ingestion assays was performed using A. octospinosus in vitro bioassay and laboratory nests. Chemical analyses were also performed to know the contents of plant extracts. This study showed that Mammea americana L. is the most promising insecticidal plant extract in the control of A. octospinosus. Indeed, the lethal concentrations (LC50 and LC99) an the lethal dose (LD99) of the M. americana extract (51.31 mg.mL-1, 131.92 mg.mL-1, and 17.36 mg/g of ant respectively) were the closest to those of Fipronil, 0.03 g/kg, the commercial insecticide used as positive control.

The leaf-cutting ants [genera Atta and Acromyrmex (Hymenoptera: Formicidae: Attini)] in general and Acromyrmex octospinosus (Reich) in particular, cause serious damage on fields crops, pastures and plantations due to their foraging activities for its symbiotic fungus cultivation (Pérez et al 2011). Estimated damage was, for example, several million dollars per year in USA and Brazil (Cameron & Riggs 1985). A. octospinosus is native to South and Central America and exotic to Guadeloupe. This species was introduced in Guadeloupe in 1954 and progressively colonized the entire territory (Boulogne et al 2014). Tremendous losses have also been observed in agricultural and protected areas. Indeed, a Regional Federation of Defense against Pests (FREDON) survey carried out in 2008 indicated that vegetable and fruit crops account for 90.9 % of attacks. The 1995 cyclone favored ant invasion in natural areas where some plant species, such as the arborescent ferns of the genus Cyathea, are now threatened and might completely disappear due to their endemism (Boulogne et al 2014). The United States Department of Agriculture (USDA) classifies this ant among the most serious pests of tropical and subtropical America (Pollard 1982). The main leaf-cutting ant control method is the application of granulated toxic baits, which are basically attractive citric matrices that contain a synthetic active ingredient that exerts a delayed action on workers (i.e., dechlorane, fipronil, sulfluramid GX071HB, or sulfluramid GX439) (Nagamoto et al 2007).
Previous studies explored the activities of active ingredients (allelochemicals) from plants. These (Boulogne et al 2012b). An exhaustive literature search was conducted to identify the published papers related to insecticidal and fungicidal chemical compounds that stem from plant species. This meta-analysis revealed that alkaloids, phenolics, and terpenoids are the three main chemical classes that are most often cited for insecticidal and fungicidal activities (Boulogne et al 2012c). To date, very few studies have used artificial nest bioassays (Hebling et al 1996(Hebling et al , 2000 or described the lethal doses, concentrations and times of botanical pesticides that are used for the control of leaf-cutting ants. Therefore, the aims of this study were i) to determine the lethal doses, concentrations and times of four preselected insecticidal plant extracts, ii) to manage a preliminary artificial nest bioassay, and iii) to characterize their biochemical contents in terms of alkaloids, phenolics, and terpenoids, in order to determine the most promising insecticidal and fungicidal plant extracts for use in controlling entire colonies.

Rearing conditions
Two adult A. octospinosus nests (all casts, queen and symbiotic fungi) were collected from the field in different locations in Guadeloupe (French West Indies) and were bred in the laboratory by housing each colony in an artificial nest. The colonies were maintained for one month in the lab after collection before the initiation of the experiments and were supplied with flowers, leaves, sugarcane, corn flakes, and water daily. The workers used in the ingestion assays were collected from these laboratory nests. Some worker vouchers were deposited in the ASTRO Lab at the National Institute of Agronomic Research (INRA) of the French West Indies and Guiana.

Extracts: source and preparation
The plants used for extract preparation were collected in different locations in  The M. americana seed maceration, the two extracts with leaves of N. oleander and the decoction of dried leaves of N. tabacum were prepared as previously described in Boulogne et al (2012a). All the aqueous plant extracts obtained and fresh leaves of S. alata were freeze dried, ground with a coffee mill, and sieved at 0.5 mm. The residues represented 99, 139, 56, 216 and 295 grams per kilogram of fresh plants of M. americana, N. oleander (crushed), N. oleander (decoction), N. tabacum, and S. alata, respectively.

Statistical analyses
Lethal concentrations (LC 50 and LC 99 ), lethal dose values (LD 50 and LD 99 ) and lethal times (LT 50 and LT 99 ) (concentrations, dose and time that killed 50% and 99% of ants) and their 95% confidence intervals were calculated with logistic regression by Probit analysis. For in vitro insecticidal and preliminary artificial nests bioassays (records three times daily during 28 days for each nest), non-parametric analyses were performed with the Friedman test with multiple comparison method of Nemenyi. For chemical analysis, non-parametric analysis was performed with the Kruskal-Wallis test, multiple comparisons were performed with the Dunn method, and Bonferroni corrections. All these tests were made with XLSTAT ® software.

Ingestion bioassay
Regarding control methods used against leaf-cutting ants, our objective was to find a toxic and attractive extract that exerted a delayed action on the workers. With this objective, ingestion bioassays were performed with groups of ten ant workers, which belong to the same colony (the bioassays were then repeated once again with an other colony). The ants were placed in 30 jam bottles (volume 324 ml, diameter 82 mm). Six bottles were used for each of the concentrations of the lyophilized plant extracts (i.e., 1, 5 and 10 mg.mL -1 ), 6 bottles were negative controls, and 6 bottles containing Blitz® commercial insecticide (Granular bait, Fipronil 0.03 g/ kg, 06/08/2001, 9800377) were used as positive controls. The ants were fed daily over a period of 21 days with an autoclaved artificial diet placed in plastic caps. This artificial diet was composed of glucose (50 mg.mL -1 ), peptone (10 mg.mL -1 ), yeast extract (1 mg.mL -1 ), agar (15 mg.mL -1 ) (Bigi et al 2004), and the freeze-dried plant extracts, distilled water (negative control), or fipronil (0.03 g/kg) q.s. to 1 L. The experiments were performed at 25 °C and 70-80% relative humidity on a 12:12 h light:dark photoperiod. Each day, the number of dead ants in each jam bottle was recorded, the caps were removed and weighed, and new caps with fresh artificial diet were offered.
The amounts of food eaten daily in g were corrected for the numbers of live ants. Mortality was analyzed on the based on the percentages of dead ants corrected by means of Abbott formula. In all insecticidal bioassays, the concentrations used for concentration response estimates were 1, 5 and 10 mg.mL -1 . The delayed action of an extract was defined according to Camargo et al (2006) as follows: a weak action occurred when 50% of the ants died before 48 h, a medium action when 50% of ants died between 48 h and 72 h, and a high action when 50% of ants died after 72 h.
The LD 50 of the N. tabacum extract (0.87 mg/g of ant) was the closest to that of Fipronil, 0.03 g/kg (1.48 mg/g of ant), whereas the LD 99 of the M. americana extract (17.36 mg/g of ant) was the closest to that of Fipronil, 0.03 g/kg (5.05 mg/g of ant, Table 1).
The LT 50 of the N. oleander extracts at 10 mg.mL -1 (47.2 and 69.98 hours) were the closest to those of Fipronil, 0.03 g/ kg at 5 and 10 mg.mL -1 (52.78 and 62.8 hours), whereas the LT 50 of the M. americana extract at 5 and 10 mg.mL -1 (102.21 and 122.95 hours) were the closest to that of Fipronil, 0.03 g/kg at 1 mg.mL -1 (111.94 hours) ( Table 2). The LT 99 of the M. americana extract at 10 mg.mL -1 (311.96 hours) were the closest to that of Fipronil, 0.03 g/kg at 1 mg.mL -1 (220.31 hours) (Table 2). Moreover, linear regressions (with R 2 and equations) between mortalities of the ants and concentrations of the plant extracts and the positive control at 24h, 48h and 72h, ( Figure  1) indicate a high delayed action for the M. americana extracts, a weak delayed action for most of the N. oleander extracts, a weak delayed action for N. tabacum and a "medium to high" delayed action for Fipronil, 0.03 g/kg (positive control).
The quantities of artificial food eaten each day by the ants during the insecticidal bioassays with all of the extracts and the positive control (Fipronil 0.03 g/kg, Blitz©) at 10 mg.mL -1 and most of extracts at 5 mg.mL -1 did not differ from the quantities consumed in the negative control condition (without extracts) with values between 0.2 and 0.3 g. The extracts at 1 mg.mL -1 were more attractive than the control with values between 0.4 and 0.45 g ( Figure 2). Data on Table 3 demontrate that the toxicity/appetency and delayed action characteristics of the M. americana seed extract were the most similar to those of Fipronil, 0.03 g/kg. This extract was thus chosen for the preliminary artificial nest bioassay. Although the extract of N. tabacum seemed to have interesting characteristics, its lack of delayed action and some toxic properties on mammals do not lead us to choose it. Indeed, its well known alkaloid (nicotine) is known for its high level of toxicity in mammals in miming the activity of acetylcholine and binding to post-synaptic receptors to cause stimulation and subsequent depression of the central nervous system, autonomic nervous system, and muscular nerve endings (Philogène et al 2008).   . The lethal concentrations (LC50) are indicated as follows: +, LC50 > 50 mg.mL-1; ++, LC50 between 50 mg.mL-1 and 25 mg.mL-1; and +++, LC50 < 25 mg.mL-1. The lethal concentration (LC99) are indicated as follows: +, LC99 > 100 mg.mL-1; ++, LC99 between 100 mg.mL-1 and 50 mg.mL-1; and +++, CL99 < 50 mg.mL-1. The lethal doses (LD50) are indicated as follows: +, LD50 > 5 mg/g; ++, LD50 between 5 mg/g and 2 mg/g; and +++, DL50 < 2 mg/g. The lethal doses (LD99) are indicated as follows: +, LD99 > 30 mg/g; ++, LD99 between 20 mg/g and 30 mg/g; and +++, LD99 < 20 mg/g. The appetencies are indicated as follows: +, daily consumed quantity < 0.20 g; and ++, daily consumed quantity > 0.20 g. The delayed actions are indicated as follows: +, 50% of the ants died before 48 h; ++, 50% of ants died between 48 h and 72 h; and +++, 50% of ants died after 72 h for the highest concentration (according to LT50 showed in Table 2).

Preliminary artificial nest bioassay
To determine whether the effects of our selected extracts could be efficient against entire colonies, preliminary artificial nest bioassays were conducted. For this purpose, ten colonies were observed in field conditions in order to select the 5 most active extracts for use in the preliminary artificial nest bioassays. These colonies were placed in artificial nests and supplied with flowers, leaves, sugarcane, and corn flakes daily prior to the initiation of the bioassays. Each artificial nest was composed of 3 plastic boxes (60 ´ 40 ´ 40 cm) that were linked together by short hoses. The central box was non-transparent and hence suitable for the placement of the colony. The other two boxes on either side were transparent to simulate the outdoor environment, and the daily rations were placed in these boxes.
The nests were exposed to each of the following treatments prepared in artificial diet (one nest by treatment): M. americana extract at 10 mg.mL -1 , S. alata extract at 2 mg.mL -1 , M. americana extract combined with S. alata extract at 10 and 2 mg.mL -1 , respectively, a positive control (Fipronil, 0.03 g/kg), and a negative control (no extracts).
The artificial diet was the same as that used in the laboratory ingestion bioassays with the addition of the lyophilized plant extracts, distilled water (negative control) or Fipronil, 0.03 g/kg. Dried Citrus sinensis (L.) Osbeck (Rutaceae) peels and Dioscorea alata L. (Dioscoreaceae) leaves (ground with a coffee mill and sieved at 0.5 mm) were added to this artificial food at 10 g.L -1 because of the natural appetency and attractiveness of these components to leaf-cutting ants (Verza et al 2006). This attractive mixture has previously been tested, and its lack of insecticidal and fungicidal activities has been verified according to established protocols of ingestion bioassay and antifungal test (Boulogne et al 2012b). The diets were placed in plastic caps and offered to the ants daily. Before and after the treatments, the nests were supplied daily with flowers and leaves for 7 days. During the treatments, the nests were supplied daily for 14 days exclusively with plastic caps containing the artificial diet with or without the active extracts. The foraging activities (number  of ants incoming the nest with plant material or artificial diet during ten minutes) and quantities of food eaten were recorded three times daily at the same time for each nest before, during, and after the treatments (Lopez & Orduz 2003).
Only the nests treated with M. americana extract and Fipronil, 0.03 g/kg exhibited significant reductions in foraging activity (by minute) after the treatments. Similarly, only these nests exhibited significant reductions in the quantities of artificial food eaten daily in grams after the treatments. The quantities of artificial food eaten daily by these ants were significantly lower during the treatments than before or after the treatments with the exception of the nest treated with S. alata (Figures 3-A  and B). Notably, 6 weeks after the treatment, all of the nests died out with the exception of the negative control nest (which lasted more than 2 months after this last observation). The nests treated with Fipronil, 0.03 g/kg and the S. alata extract contained no surviving ants or fungus gardens. The nest treated with the M. americana extract exhibited no surviving ants, and its fungus garden was infested with fungal competitors. The nest treated with the M. americana extract combined with the S. alata extract contained no ants and no fungus garden and was infested with fungal competitors (Figure 3-C).

Chemical analysis
The extracts were submitted to phytochemical analyses for plant secondary metabolites that are known for their potential insecticidal activities, i.e., alkaloids, phenolic compounds, and terpenoids. The quantity of each compound was determined using a spectrophotometer according to the following quantitative methods: alkaloids were determined using Marquis's reagent with a colorimetric method (Szabo et al 2003), total phenolic compounds were determined using the Folin-Ciocalteau colorimetric method (Heilerová et al 2003), and terpenoids were determined using the iron (III) chloride-o-phosphoric acid-sulfuric acid colorimetric method (Zak et al 1956). All values are expressed in micrograms of standard per g of freeze-dried fresh plant sample and three replications were performed.
The N. oleander extracts contained the greatest quantities of total alkaloids, the N. tabacum and N. oleander decoctions contained the greatest quantities of total phenolic compounds and the N. oleander extracts contained the most total terpenoids. Terpenoids were the most abundant compounds in all of the extracts (between 550 and 1400 mg of standard/g of freeze-dried plant sample) ( Figure 4).
Our preliminary chemical analysis revealed that the M. Americana seed extract contained alkaloids, phenolics and terpenoids. Analysis of the literature revealed that the compounds responsible for the insecticidal activities of M. americana are well-known phenolic compounds such as coumarins and particularly mammein (Duke 1989). Methanol and ethyl acetate extractions and HPLC or LC-MS analyses have revealed 4.8 mg of coumarins per gram of seed (Yang et al 2006).
The M. americana seed extract exhibited an LC 50 of 51.31 mg.mL -1 and an LD 50 of 706.65 mg/g of insect. To compare our results with existing data we found a study of the toxicities of aqueous and ethanolic extracts of M. americana on Artemia salina L. (Anostraca: Artemiidae). This study revealed an LC 50 greater than 10 mg.mL -1 for an aqueous extract and an LC 50 of 197 μg/mL for the ethanolic extract (Bussmann et al 2011). Another study of coumarins in M. americana revealed an LD 50 of 4 μg/insect against Phaedon cochleariae Fab. (Coleoptera; Chrysomelidae) (Perez et al 2010). These results might suggest that ethanolic extracts and bioguided fractionations of M. americana seed extracts might increase the toxicity (coumarins are less miscible in water than in ethanol).

S. alata
The nests treated with the S. alata extract contained no fungus garden after 6 weeks of treatment. This effect was consistent with our previous study, which showed that the S. alata foliar extract was fungicidal against L. gongylophorus (Boulogne et al 2012b). We also reported a preliminary chemical analysis, which revealed that S. alata foliage contains alkaloids, phenolic compounds and terpenoids. Previous studies exhibited fungicidal activity against several other fungi and showed that the fungicidal activity of S. alata foliar extracts can be attributed to phenolic compounds such as anthraquinones (Somchit et al 2003) like chrysophanol (Palanichamy & Nagarajan 1990).

Artificial nests bioassay
We were aware that our artificial nest bioassay was really preliminary and all results of differences among extracts should be treated as preliminary and with caution. Indeed, the experimental design of the test had no replicates to the treatments and it could bias the results (differences might be attributed to differences in nests themselves and not to treatments). However, this preliminary assay is crucial and important to this study line since they show that these extracts are attractive to ants and consumed by them.
Concerning the length of treatment, similar bioassays revealed that Atta sexdens rubropilosa nests that are supplied daily with a diet containing Ricinus communis leaves exhibit gradual decreases in fungus gardens and substantial worker mortality after 6 weeks of treatment (Hebling et al 1996). Another study of Atta sexdens nests that were fed daily with a diet containing Canavalia ensiformis L. (Leguminosae) leaves reported complete nest extinction after 11 weeks of treatments (Hebling et al 2000).

Ant baits in field conditions
Regarding delayed action, the seed of M. americana seemed to naturally possess this property. It would be interesting to increase this delayed action of the substance with a digestible polymer using microencapsulation techniques (Benita 2005). Microcapsules could be made to contain a freeze-dried ethanolic or aqueous extracts of M. americana seeds and placed in a mix of Citrus sp. pulp and dried Dioscorea alata leaves to increase the appetency based on our laboratory results and existing data (Verza et al 2006). These granular baits (microcapsules and attractant mixtures compressed in granular form easy to apply) might be protected in biodegradable and compostable plastics to preserve them against sunlight and adverse weather conditions. Thus, like others well known commercial formulation containing botanical extracts (e.g. pyrethrins, rotenone, sabadilla, ryania, nicotine, azadirachtins or limonene) (Weinzierl 2000), a field study should be conducted with these kinds of baits to improve our work in field conditions. However compared to these previous studies, the force of our argument is to use mixtures of active phytochemicals to reduce conventional resistance compared with the selection pressure exerted by single pure molecule (Arnason et al 1993) and make the choice of extraction type and solvent with the greatest sustainability (water or eventually ethanol extractions) (Boulogne et al 2012c). As the example of M. americana extract, toxicity may be increased and transformed in granular baits (microcapsules and attractant mixture) protected in a plastic. As the same way, to improve our fungicidal results (in vitro and with artificial nest) in field conditions, S. alata foliar extracts may also be transformed in granular baits.
Our results might be useful in the control of A. octospinosus and might be applicable to the control of other leaf-cutting ant species. However, we need to keep in mind that, in Guadeloupe (FWI), this species is exotic. This ant might behave different in its exotic range than in its native range (e.g. the case of Linepithema humile (Human & Gordon 1996), and the results found might only apply for non-native populations. Thus, these results should be treated with caution before their generalization.
Our study is the first report of the toxicities of M. Americana seed, N. oleander leaf, and N. tabacum leaf freezedried extracts due to ingestion by Attini. It allowed us to determine the lethal doses, concentrations and times of four insecticidal plant extracts that were selected for examination based on previous studies. This study revealed that the M. americana seed extract was the most similar in terms of toxicity/appetency and delayed action to the commercial bait Fipronil, 0.03 g/kg. Our analyses also revealed that these extracts contained alkaloids, phenolics, and terpenoids. The preliminary artificial nest bioassays showed that the most promising insecticidal (M. americana) and fungicidal (S. alata) plant extracts might be useful in the control of A. octospinosus in Guadeloupe and where the species is exotic.
Further studies should be conducted to optimize the toxicities of these extracts against A. octospinosus, to verify that they lack toxicity against non-targeted organisms and to confirm their activities in entire nests in natural conditions.