Anti-predator defense of the biological control agent Oxyops vitiosa is mediated by plant volatiles sequestered from their host plant Melaleuca quinquenervia
G. S. Wheeler
Abstract.— The weevil Oxyops vitiosa is an Australian species imported to Florida, USA for the biological control of the invasive species Melaleuca quinquenervia. The larvae of this species feed on the leaves of their host and produce a shiny orange secretion that covers their integument. When this secretion is applied at physiological concentrations to a dog food bait, fire ant consumption and visitation are significantly reduced. Gas chromatographic analysis indicates that the larval secretion resembles qualitatively and quantitatively the terpenoid composition of their host foliage. When the combination of the ten major terpenoids from the O. vitiosa secretion were similarly applied to dog food bait, fire ant consumption and visitation were reduced. When these ten terpenoids were similarly tested individually, the sesquiterpene viridiflorol was the most active component decreasing fire ant consumption of dog food bait. Fire ant visitation was initially (15 min after initiation of the study) decreased for dog food bait treated with viridiflorol, and the monoterpenes 1,8-cineole, and a-terpineol. Fire ants continued to avoid the bait treated with viridiflorol at 18 µg/mg dog food for up to 6 hrs after the initiation of the experiment. Moreover, ants avoided bait treated with 1.8 µg/mg for up to 3 hrs. The concentrations of viridiflorol, 1,8-cineole, and a-terpineol in larval washes were about twice that of the host foliage suggesting that the larvae sequester these plant-derived compounds for their defense against generalist predators.
The Australian weevil Oxyops vitiosa, Pascoe (Coleoptera: Curculionidae) was introduced in south Florida in 1997 for the biological control of the invasive weed Melaleuca quinquenervia (Cav.) S. T. Blake (Myrtaceae) (Center et al., 2000). This weevil species has since been established throughout the infested area of Florida. The larvae feed on the young leaves and have a characteristic odor that resembles that of the foliage of their host tree. The bright orange larvae are diurnally active and feed exposed during all instars on the foliage then drop to the ground where they form a protective pupal cell in the soil (Purcell and Balciunas, 1994). The cuticle of the larvae is completely covered with a shiny oily substance that may function in defense against generalist predators (Montgomery and Wheeler, 2000). However, the chemical nature of this defense has yet to be determined. Although few weevil species have been reported to produce defensive secretions (Pavis et al., 1992), similar integumental slime found on the slug-like sawfly larvae of Caliroa cerasi reduced the incidence of predation by generalist ants (Eisner, 1994).
The red imported fire ant, Solenopsis invicta Buren (Hymenoptera: Formicidae) constitutes one of the most invasive invertebrate predators in the southeastern U.S. invading forests, pastures, crop lands and natural areas (Lofgren et al., 1975). This ant is among the most common invertebrate predators in the southeastern U.S. (Elvin et al., 1983; Kharboutli and Mack, 1991). Generalist predators, including ants, have prevented the establishment or reduced the effectiveness of many weed biological control agents (Goedon and Louda, 1976) and this ant species in particular has been directly implicated in reducing the efficacy of Tyta luctuosa, a biological control of field bindweed (Ciomperlik et al., 1992).
Biological control efforts may benefit by recruiting agents that sequester plant defensive compounds that impart protection from generalist predators. Sequestration of plant defenses may be more common than appreciated by biological control practitioners, as this phenomenon is common among herbivores with narrow host ranges (Bowers, 1990). Several weed biological control agents are known to sequester plant defensive compounds, namely the pyrrolizidine alkaloids of the Senecio jacobaea/ - Tyria jacobaeae association (Aplin et al., 1968), the quinolizidine alkaloids of the Scotch broom – aphid association (Wink et al., 1982) and Chrysolina beetles derive the bisanthraquinone hypericin from their host plant Hypericum (Rees, 1969). Additionally, these ants are well known for the painful stings in humans (Rhoades et al., 1989) and these ants can cause decreased weight gain in livestock (Wilson and Eads, 1949). They can displace native wildlife species (Drees, 1994; Lockley 1995; Pedersen et al., 1996), thereby decreasing the biodiversity of natural areas. Fire ants can cause damage to electrical circuits as they are attracted to electrical currents (Vinson and MacKay, 1990). The purpose of this study was to determine the nature of the defensive secretions produced by the larvae of O. vitiosa toward the generalist predator S. invicta.
Insects. Larvae of O. vitiosa were collected (N = 20) from a field site infested with M. quinquenervia in Ft. Lauderdale, Broward Co. Fla. Solvent washes were conducted on fourth instars by dipping each larva in 1 or 2 ml of chloroform (CHCl3) for 15 seconds. The CHCl3 larval wash was filtered through glass wool and dried over NaSO4 prior to analysis by gas chromatography.
Plants. Leaves were collected from the same source as the larvae described above. Tip leaves were clipped from young trees and brought to the laboratory where they were frozen (-10 °C) for further processing. The leaf components were extracted by a modified microwave technique (Southwell et al., 1995; Degen and Stadler, 1998; Gomez et al., 1999). Fresh leaf samples (75 – 150 mg) were immersed in 1 ml EtOH (95%) and exposed to 60 sec of microwave (750 watts) irradiation. A sample (500 µl) of the EtOH extract was mixed with deionized water (500 µl), followed by an equal volume of CHCl3. The sample was vortexed for 1 min and then centrifuged for 10 min at 10,000 RPM. A sample (200 µl) of the CHCl3 layer was removed and dried over NaSO4. Internal standards were added as 100 ng each of n-tridecane and n-eicosane in 10 µl CHCl3 to the extract prior to analysis by gas chromatography.
Chemicals. Terpenoids were purchased from commercial sources, except where mentioned, and were of the highest purity available (Table 1). The list included the primary compounds reported by Brophy et al., (1989), Ramanoelina et al., (1994). These included (+) a-pinene (98%), (-) b-pinene (99%), a-terpinene (98%), (-) limonene (92%), 1,8-cineole (100%), g-terpinene (98%), terpinene 4-ol (97%), a-terpineol (98%), trans-nerolidol (95%). Additionally, viridiflorol (86%) was generously provided by Dr. I. A. Southwell.
Gas chromatography. Samples were analyzed with either a Hewlett-Packard model 6890 or 5890 gas chromatograph. Data collection, storage, and analysis were conducted with the ChromPerfect data system. Helium at a linear flow rate of 37 cm/sec was used as a carrier gas. All samples were analyzed on three fused silica capillary columns (HP5 Hewlett Packard Company, Wilmington, DE, DB17-MS and a WaxEtr J & W Scientific, Folsom, CA; all 30 m x 0.32 mm i.d., 0.25 micron thick film). Injector temperature was 200 °C and FID temperature was set at 250 °C. The oven temperature was held at 50 °C for 2 min then increased at 8 °C/min to 250 °C where it was held for 5 min.
Compounds identities were confirmed by GC-MS using a Hewlett Packard 6890 instrument fitted with a HP5-MS (30.3 m x 0.25 mm, 0.25 micron film thickness) FSOT colun with helium (36 cm/sec) as a carrier gas, injector port (split 1:50) at 250°C, mass selective detector (HP 5973) at 250°C (source) and 150°C (quad) with transfer line 280°C and ion source filament voltage of 70 eV. Component identification was made on the basis of mass spectral fragmentation, retention index with n-paraffins and comparison with authentic constituents and mass spectral and retention matching with commercial (NIST, Wiley, and Adams) libraries. Compounds were quantified relative to the peak areas of the internal standard and values were adjusted by calculation of their response factors relative to n-tridecane (Debbrecht 1985).
Fire Ants. Red imported fire ants were collected at the University of Florida, Ft. Lauderdale Research and Education Center as described previously (Montgomery and Wheeler, 2000). Colonies, consisting of brood, eggs and workers, were established in plastic boxes (20 x 12 cm) and provided with deionized water in a glass tube (45 x 4 mm) plugged with cotton. Ants were fed kibbled dog food (Pedigree Prime) freely for 2-5 d prior to each test. The food was removed 24-36 hr prior to each test. Newly collected ant colonies were used for each test.
Bioassay of Defensive Secretions. The material washed from the O. vitiosa larvae or select commercial terpenoids was dissolved in 50 µl of CHCl3 and applied to kibbled dog food cut into 50 mg pieces. The solvent was allowed to dry for 0.5 hr in a fume hood under ambient conditions. The treated dog food bait was stored for 12-15 h at 4 °C. Each test consisted of offering the ants a choice between a solvent (CHCl3)-treated control and either a crude larval wash or a select commercial terpenoid formulated individually or in a mixture. To determine the percent loss during the experiment, treated dog food samples were CHCl3-extracted before and after a 6 h test and analyzed by GC. By comparing peak areas the concentration of each terpenoid was found to decrease during the experiment from 53.9 (± 2.5%; viridiflorol) to 93.7 (± 0.9%; a-pinene).
Four choice-tests were conducted, all under ambient conditions between 9 am to 3 pm. For all tests, ant consumption was determined by comparing the dog food weight before and after each 6 hr test. Additionally, the number of ants was counted visiting each dog food bait at regular intervals. In study 1, choice-tests were conducted between a CHCl3-treated control and the crude O. vitiosa larval wash (one-larval-equivalent) applied to a piece of dog food (N = 10). In study 2, choice-tests were conducted between a CHCl3-treated control and a mixture of the ten major terpenoids found from GC analysis of the larval washes (N = 6). These were formulated to include the range of concentrations of each compound found in the crude larval washes (Table 1). Additional treatments consisted of a 10-fold and 100-fold dilution of these mixtures. In study 3, choice-tests were conducted between a CHCl3-treated control and the same ten terpenoids tested individually at the highest concentrations tested in study 2 (N = 4). In study 4, choice tests were conducted between CHCl3-treated controls and three different viridiflorol concentrations using the same concentrations as in study 2 (N = 6).
Statistical Analysis. All data, S. invicta consumption and the number of fire ants visiting each dog food bait were analyzed with SAS (SAS Institute, 1990). The consumption results were analyzed with ANOVA and the fire ant activity data were analyzed with a Kruskal-Wallis.
|
Table 1. Concentration of the 10 terpenoid components applied to kibbled dog food and fed to red imported fire ants workers. Each piece of dog food (50 mg) was treated with one (High), 0.1 (Intermediate) or 0.01 (Low) larval equivalent of these 10 components. N = 6. |
|||||||
|
Concentration (mg/mg dog food) |
|||||||
|
Compound |
High |
Intermediate |
Low |
||||
|
1,8-cineole |
22.00 |
2.20 |
0.22 |
||||
|
viridiflorol |
18.00 |
1.80 |
0.18 |
||||
|
a-terpineol |
5.47 |
0.55 |
0.05 |
||||
|
a-pinene |
5.06 |
0.51 |
0.05 |
||||
|
trans-nerolidol |
5.02 |
0.50 |
0.05 |
||||
|
limonene |
3.09 |
0.31 |
0.03 |
||||
|
b-pinene |
1.70 |
0.17 |
0.02 |
||||
|
terpinene 4-ol |
1.80 |
0.18 |
0.02 |
||||
|
g-terpinene |
1.67 |
0.17 |
0.02 |
||||
|
a-terpinene |
1.43 |
0.14 |
0.01 |
||||
Statistical Analysis. All data, S. invicta consumption and the number of fire ants visiting each dog food bait were analyzed with SAS (SAS Institute, 1990). The consumption results were analyzed with ANOVA and the fire ant activity data were analyzed with a Kruskal-Wallis.
Chemical analysis. Considerable quantitative and qualitative similarity was found between the chemical constituents from leaf extracts and larval washes (Table 2; Fig. 1). However, some of the terpenoids analyzed (e.g., 1,8-cineole, α-terpineol, viridiflorol) were in greater concentrations in the larval washes compared with the leaf extractions. This suggests that as the O. vitiosa larvae consume M. quinquenervia foliage they absorb and deposit the foliar terpenoids on their cuticle. The larval weight averaged (± SE) 101.9 (± 3.0) mg and the amount of material removed by washing each larva averaged 7.9 (± 0.5) mg.
|
Compound |
Leaf extract |
Larval wash |
||||
|
(mg/mg) |
se |
(mg/mg) |
se |
|||
|
1,8-cineole |
3.10 |
0.40 |
6.73 |
0.76 |
||
|
viridiflorol |
6.10 |
0.90 |
11.85 |
1.25 |
||
|
a-terpineol |
0.90 |
0.10 |
2.23 |
0.22 |
||
|
a-pinene |
1.50 |
0.40 |
1.44 |
0.18 |
||
|
trans-nerolidol |
0.10 |
0.10 |
0.06 |
0.01 |
||
|
limonene |
0.40 |
0.10 |
0.66 |
0.08 |
||
|
b-pinene |
1.10 |
0.20 |
0.45 |
0.04 |
||
|
terpinen 4-ol |
0.10 |
0.00 |
0.15 |
0.01 |
||
|
g-terpinene |
0.10 |
0.00 |
0.09 |
0.01 |
||
|
a-terpinene |
0.01 |
0.00 |
0.03 |
0.01 |
||
Bioassay of Larval Secretions.
Study 1. Fire ant consumption and visitation was significantly greater on the control dog food bait (treated with CHCL3) compared with that treated with the O. vitiosa larval wash (Table 3). During the 6 hr test the fire ants consumed an average of 8.7 (± 3.6) mg of dog food in the controls compared with 0.8 (± 0.3) mg in the larval wash-treated dog food. Additionally, fire ants frequently visited the control dog food at both 3 and 6 hrs after initiation of the experiment, whereas they were never observed visiting the larval wash-treated dog food (Table 3).
Study 2. Fire ant consumption and visitation of dog food treated with the mixture of the ten terpenoids (about one-larval-equivalents) were significantly decreased compared with the solvent-treated controls. In both the high (F 1,10 = 8.16; P = 0.0171) and low (F 1,10 = 5.15; P = 0.0466) treatments of the terpenoid mixture, consumption was significantly reduced compared with that of the control-treated dog food (Fig. 2). Furthermore, fire ant visitation was reduced (Fig. 3) during the 15 min observation period in the high (X2 1 = 9.4661; P = 0.0021), intermediate (X2 1 = 8.4255; P = 0.0037) and the low (X2 1 = 8.3662; P = 0.0038) treatments of the terpenoid mixture compared with the solvent-treated controls. Significant differences were also found 3 hrs after the study began in the high (X2 1 = 9.5422; P = 0.0020) and the low (X2 1 = 5.4482; P = 0.0196) concentrations. Finally, significantly fewer fire ants were observed visiting the terpenoid mixture-treated dog food after 6 hrs on the high (X2 1 = 9.4661; P = 0.0021) and the low (X2 1 = 6.2256; P = 0.0126) concentrations.
Study 3. When each terpenoid was applied individually to the dog food the only treatment that reduced fire ant consumption was viridiflorol (Fig. 4; 18.0 µg/mg; F 1,6 = 41.55; P = 0.0007). Fire ant visitation was reduced after 15 min (Fig. 5) in those treated with 1,8-cineole (22.0 µg/mg; X2 1 = 3.6067; P = 0.0575), α-terpineol (5.47 µg/mg; X2 1 = 4.0833; P = 0.0433) and viridiflorol (18.0 µg/mg; X2 1 = 5.4634; P = 0.0194). After 3 hrs (X2 1 = 5.3976; P = 0.0202) and 6 hrs (X2 1 = 5.6000; P = 0.0180) fire ant visitations were reduced only on the viridiflorol-treated dog food bait.
Study 4. Viridiflorol tested alone was most active at decreasing fire ant consumption and repelling ants. Compared with the solvent-treated controls, consumption was reduced only by viridiflorol at the highest concentration (18 µg/mg; F 1,6 = 4.39; P = 0.0360) tested (Fig. 6). Fire ant visitation to viridiflorol-treated bait was reduced at the highest (X2 1 = 9.4661; P = 0.0021) and intermediate (X2 1 = 9.1034; P = 0.0026) concentrations after 15 min (Fig. 7). This activity continued for 3 hrs after the beginning of the study at the highest (X2 1 = 9.4661; P = 0.0021) and intermediate (X2 1 = 3.4797; P = 0.0621) viridiflorol concentrations. After 6 hrs a difference in activity was only found at the highest viridiflorol concentration (X2 1 = 7.1739; P = 0.0074).
These results demonstrate the repellent nature of the larval secretions of the M. quinquenervia biological control agent O. vitiosa toward the red imported fire ant S. invicta. Our results indicated repellent activity from the crude larval washes, mixtures of the ten major terpenoids contained in the larval washes, and individual compounds from this mixture. In all experiments, fire ant consumption and visitation to the treated baits decreased significantly compared with the solvent-treated controls. The repellency of the most active component, viridiflorol, was demonstrated throughout the duration of the experimental observations (6 hr).
In the larval secretions, the three most active compounds, viridiflorol, 1,8-cineole, and a-terpineol were about twice that extracted from the host foliage suggesting that the larvae sequester these plant-derived compounds. In fact, nearly all of the 10 compounds investigated were in greater concentration (on a fresh mass basis) washed from the larvae compared with those measured in the leaf extracts. The most active ant repellents increased the most from the plant to the larvae, namely viridiflorol (1.9-fold), 1,8-cineole (2.2-fold), and a-terpineol (2.5-fold). The proportion of the ingested terpenoids retained by the larvae on the cuticle can be calculated and expressed as the conversion efficiency for each terpenoid (ECDt), analogous to calculation of the food conversion efficiency (ECDf, Slansky and Scriber, 1985). The ECIt values were estimated by: larval mass (mean = 101.9 mg) x concentration of the terpenoid (mg/mg) washed from the larvae (from Table 2) / food consumed (mg) x concentration of the terpenoid (mg/mg) in the leaves (from Table 2). We used consumption (400 mg fresh mass) and food conversion efficiency (ECIf; 7.3% fresh mass) results from a similar study of O. vitiosa larvae fed leaves of M. quinquenervia (Wheeler unpublished data). To determine if select terpenoids were actively sequestered and deposited on the larvae the conversion efficiency of each active terpenoid washed from the cuticle (ECIt) can be compared with the conversion efficiency of the food (ECIf). The results of these calculations suggest that generally more than half the quantity consumed of the most repellent terpenoids was retained and deposited on the larval cuticle. For example, the ECIt values all exceeded 46% for the terpenoids viridiflorol (49.5%), 1,8-cineole (55.3%), and a-terpineol (63.2%). Moreover, with the exception of a-terpinene, which was present in barely detectible levels, the conversion efficiencies of the less repellent terpenoids were considerably lower (< 45%). These results are consistent with the interpretation that these terpenoids were actively sequestered, as these values were considerably greater than the conversion efficiency value for food (ECIf = 7.3%). A similar analysis was conducted, using the approximate digestibility (AD) of the ingested terpenoid 1,8-cineole by a chrysomelid eucalypt-feeder Paropsis atomaria (Olivier) (Ohmart and Larsson, 1989). Our calculations also suggest that sequestration was selective as the ECIt values were greatest for the most repellent compounds. Similar selectivity was found with the larvae of other species where the most emetic iridoid glycoside catalpol was selectively sequestered compared with a less-emetic aucubin (Bowers and Collinge, 1992; Bowers et al., 1993). However, our calculations need to be confirmed with a more direct determination of the terpenoid budgets.
Terpenoids are well known ant repellents (Shorey et al., 1992) that when sequestered (e.g., cardenolides, cucurbitacins) may be repellent against many predatory types (Bowers, 1990). Several terpenoids have shown activity against both herbivorous and predacious ant species. For example, plant foliar terpenoids may either repel (Hubbell, et al., 1984; Howard et al, 1989) or be toxic to leaf-cutting ants (Howard et al., 1988). Additionally, insect-produced terpenoids from chrysomelid larvae also repel the imported fire ants (Blum et al., 1978). The plant sesquiterpene farnesol may be applied as a band around citrus tree trunks to exclude Argentine ants Linepithema humile (Mayer) for several months (Shorey et al., 1996). Numerous compounds with a range of volatilities have been reported to have ant repellent activity (Shorey et al., 1992), however to our knowledge, this is the first report of repellent activity at physiological concentrations of the terpenoids viridiflorol, 1,8-cineole, and a-terpineol.
The ecological host range of this biological control agent may be influenced by the availability of foliar compounds that serve as effective repellents against generalist predators. Larvae may escape from predators on plant species that have the compounds that can be sequestered for anti-predator defense (Jeffries and Lawton, 1984). Even though oviposition and larval development may occur on test plants in the laboratory, in field conditions, where natural enemies like fire ants are abundant, host use may be restricted to plant species which provide compounds that can be sequestered and impart defense against generalist predators. Although field observations indicated that this species was only collected on M. quinquenervia (Balciunas et al., 1994), quarantine host testing indicated that neonates complete development on the Australian myrtaceous species Callistemon rigidus and C. viminalis. Additionally, third instars previous fed M. quinquenervia will feed on the North American Myrica cerifera (Balciunas and Buckingham, 1996; Wheeler unpublished data). Additional chemical variation in the O. vitiosa host range exists in the different M. quinquenervia chemotypes, both of which occur in Australia (Ireland, 1999) and Florida (Dray and Wheeler unpublished data). The two M. quinquenervia chemotypes have distinct terpenoid profiles; one has primarily trans-nerolidol and only small concentrations of the primary components found in the present study. Different larval diets influenced the cuticular waxes of Manduca sexta when fed different plant species or an artificial diet (Espelie and Bernays, 1989) and these differences influenced the susceptibility of the larvae to ant predation (Cornelius and Bernays, 1995). The protection O. vitiosa larvae derive from these plant species and chemotypes, and its ecological significance is a subject of an additional series of studies (Wheeler et al., unpublished data).
Predators are well known antagonists of arthropods used for biological control of weeds (Crawley, 1989; Julien and Griffiths, 1998) and predation is so common that it may be assumed that it likely influences the establishment of all agents (Goeden & Louda, 1976). In general, natural enemies, including predators, parasitoids, and diseases interfere with approximately half the biological control projects (Goeden & Louda, 1976). This M. quinquenervia biological control agent O. vitiosa is apparently well protected from one of the most common invertebrate generalist predators in the southeastern U.S. (Montgomery and Wheeler, 2000). However, recent observations indicate that another predator, the pentatomid Podisus mucronatus Uhler, with relatively long mouthparts may penetrate the integumental defenses of O. vitiosa larvae (Pratt, P. USDA/ARS, Ft Lauderdale, Fla. personal communication). Different mouthpart sizes have allowed heteropteran predators with long mandibles (e.g., Nabis americoferus or Podisus maculiventris) to utilize otherwise well-defended prey compared with predators with shorter mandibles (e.g., coccinelids; Olmstead and Denno 1993). However, this biological control agent has become well established in Fla (Center et al., 2000). Several factors may contribute to its success; among them is their integumental terpenoid defense.
References