Introduction

Tomatoes (Lycopersicon esculentum) are the second most important vegetable crop in the United States with over 8 million tons produced annually in the U.S. (Brunke et al. 2003). The economic viability of tomatoes is threatened by numerous insect pests, including lepidopteran larvae, thrips (Thysanoptera: Thripidae), stinkbugs (Hemiptera: Pentatomidae), whiteflies (Hemiptera: Aleyrodidae) and aphids (Hemiptera: Aphididae) (Nault & Speese 2002; Kemble et al. 2004). Aphids can be particularly serious pests of tomato because aphid feeding reduces the yield and fruit quality of tomatoes and, more importantly, aphids are the primary vectors for plant viruses that can devastate tomato production (Tomlinson 1987). For example, epidemics of Cucumber mosaic virus (CMV) have significantly reduced tomato production in the southeastern U.S. and in China, France, Italy, and Spain (Jorda et al. 1992; Kaper et al. 1990; Sikora et al. 1998). Plant viruses like CMV are often difficult to manage because they may have extremely large host ranges (Edwardson & Christie 1991) and can be transmitted by dozens of species of aphid species in a nonpersistent manner (Palukaitis et al. 1992). Viruses transmitted by aphids in a nonpersistent manner are acquired in seconds by the aphid while probing an infected epidermal cell and are retained on the tip of the aphid's stylet. The virus can then be immediately transmitted in the same general process, i.e., a quick probe of a plant's epidermal cell by the aphid. Because virus transmission of nonpersistent viruses like CMV by aphids is rapid, chemical pesticides are ineffective because aphids transmit the virus before they are intoxicated. Additionally, some pesticides actually enhance the spread of plant viruses by increasing aphid probing and movement among plants (Lowery & Boiteau 1988; Perring et al. 1999). Because aphids are the primary vectors of most plant viruses, any factor that increases aphid abundance is likely to increase the spread of the virus and the potential for a virus epidemic (Madden et al. 2000; Hull 2002; Jeger et al. 2004). For example, the abundance of alate aphids is positively correlated with the incidence of Beet mosaic virus in sugar beet, Beta vulgaris (L.) (Dusi et al. 2000), with the incidence of Barley yellow dwarf virus in cereals (Power et al. 1991; Chapin et al. 2001; Fabre et al. 2003), and with the incidence of Potato virus Y and Potato leafroll virus in potato, Solanum tuberosum (Baskey 2002).

Our work in southeastern agricultural systems has identified an important ecological interaction that may alter the abundance of virus vectors: fire ant--aphid mutualisms. Red imported fire ants, Solenopsis invicta (Buren) are widespread, invasive ants that are notoriously aggressive and voracious (Vinson 1997; Holway et al. 2002). They were originally introduced through the port of Mobile, Alabama, in the early 1900s, spread across the southern United States from North Carolina to Arizona, and have recently invaded California (MacKay & Fagerlund 1997; Vinson 1997). Fire ants, like many ant species, readily form facultative mutualisms with honeydew-producing insects such as aphids, scales, and whiteflies (Reilly & Sterling 1983a, 1983b; Vinson 1997; Helms & Vinson 2002). Fire ants are attracted to plants infested with aphids and are effective at protecting aphids from predators. Fire ants, for example, preferentially forage on aphid-infested cotton plants compared with aphid-free cotton plants, and fire ant and cotton aphid abundances are positively correlated in the field (Kaplan & Eubanks 2002, 2005). Fire ant predation of lady beetle larvae is twice as high on aphid-infested plants as on aphid-free plants (Kaplan & Eubanks 2002) and predator larvae are more abundant in cotton fields with suppressed densities of fire ants than in control fields with large fire ant populations (Kaplan & Eubanks 2002, 2005; Diaz et al. 2004).

The positive effect of ants on aphid population density may affect the movement of aphids. Under low density conditions, developing aphids do not produce wings when they molt and become apterous or wingless adults. When crowded, however, nymphs develop wings as they molt and become alate or winged adults (Dixon 1977; Muller et al. 2001). Dispersing aphids typically search for new host plants with no or few aphids (Dixon 1977; Hodgson 1991). If fire ant protection of aphids results in a 3 to 5-fold increase in aphid population density (i.e., Kaplan & Eubanks 2002, 2005), then it is likely that fire ant protection will result in a 3 to 5-fold increase in the number of alates. This could result in an increase in the movement of aphids among different plant species and ultimately increase virus spread and incidence.

The effect of fire ants in agroecosystems is not always negative. For example, fire ants have been reported as predators of a wide variety of insect pests including the velvetbean caterpillar, Anticarsia gemmatalis Hübner (Lee et al. 1990), rednecked peanutworm, Stegasta bosqueella (Chambers) (Vogt et al. 2001), horn fly, Haematobia irritans (L.), boll weevil, Anthonomus grandis (Sterling 1978; Jones & Sterling 1979), cotton bollworm, Helocoverpa zea (Diaz et al. 2004); beet armyworm, Spodoptera exigua (Hubner) (Diaz et al. 2004), and the sugarcane borer, Diatraea saccharalis (Zehntner) (Adams et al. 1981; Fuller & Reagan 1988; Bessin & Reagan 1993). Thus, the suppression of fire ants in tomato may result in reduced aphid densities, but could increase the abundance of other insect pests of tomato.

Our primary objective was to determine if fire ants increase the abundance of aphids in tomato. If fire ants do increase the abundance of aphids in tomato fields, then the suppression of fire ants in and around tomato fields may reduce the abundance of aphids and potentially alter the spread of aphid-vectored viruses. A secondary objective of this study was to document the effect of fire ant suppression on the abundance of other insect pests in tomato so that any changes in pest abundance could be included in future decisions regarding the suppression of fire ants in tomato.

Materials and Methods

Field experiments were conducted at the E.V. Smith Research Center in Macon County, Alabama, during 2003 and 2004. For both field seasons, tomatoes (Lycopersicon esculentum variety: Floralina) were grown from seeds in the greenhouse for 1 month or until they reached 20 cm in height. They were then placed outdoors for one week of acclimatization, and then transplanted into field plots. In 2003, twelve plots of tomato plants, separated by at least 10 m, were established on bare ground. Each plot consisted of 30 plants, spaced 45 cm apart and organized into three rows separated by 75 cm each. In 2004, tomatoes were transplanted into raised (15 cm) beds of white plastic mulch, following fresh market tomato production standards (Kemble et al. 2004). Twelve plots of tomato plants were established during 2004 with the same plant and row spacing as in 2003.

Plots were randomly divided into 2 treatments: natural or high fire ant density (control) or suppressed fire ant density. Suppression treatments were established with Amdro®, a commercially available, ant-specific bait. Foraging fire ant workers pick up the bait and return it to the colony. The active ingredient of Amdro® is hydramethylnon which blocks the production of ATP and kills ants by inhibiting energy production (Valles & Koehler 1997). Amdro® has very minimal effects on non-target arthropods (Hu & Frank 1996; Eubanks et al. 2002), and has been successfully used to suppress fire ants in similar sized and spaced plots in other studies (e.g., Harvey & Eubanks 2004; 2005). Amdro® was applied weekly to plots assigned to the fire ant suppression treatment, beginning 1 week prior to sampling, and fire ants were allowed to naturally colonize control (high fire ant) plots. Fire ant abundance was quantified weekly within each plot with traps consisted of a 2.5-cm piece of “hotdog” placed in a 50-ml plastic tube. One trap was placed in the center of each plot for ≈ 45 minutes, after which they were collected, sealed, returned to the laboratory, and stored in a freezer until ants were counted. All plants within each plot were visually searched each week for aphids, fire ants, and other arthropods. All abundance data were log (n+1) transformed prior to analysis. Differences in the abundance of fire ants, aphids, non-aphid herbivores, and natural enemies were compared between treatments with repeated measures ANOVA (SAS Proc Mixed with repeated statement; Khattee & Naik 1999). To avoid pseudo-replication (i.e., counting each plant within each plot as a replicate), data for each plot were averaged to obtain mean plot values for analysis.

In 2004, tomato plants in 4 fresh market tomato fields were sampled in Blount Co., Alabama for the presence of fire ants and other arthropods. Fields ranged from 5 to 7 acres and approximately 20 tomato plants per acre were visually searched for aphids, fire ants, and other arthropods weekly for 7 weeks. Additionally, “hotdog” traps were used to quantify the relative abundance of fire ants (1 trap per acre of tomato). In 2005 the same sampling procedures were repeated on 5 fresh market tomato fields in the same area of Blount County. We found very few non-aphid herbivores and natural enemies on tomato plants on the farms and, consequently, limited our analysis to a correlation of aphid abundance and fire ant abundance in commercial tomato fields with seasonal averages.

Results

In both 2003 and 2004 the abundance of fire ants foraging on the ground in tomato plots was significantly reduced by the application of Amdro®, although fire ant suppression was better in 2003 (2003: 290.67 ± 48.13 fire ants per trap versus 39.1 ± 27.88 fire ants per trap, control versus Amdro® treated plots, respectively, F_1,20 = 27.51; P = <0.0001; 2004: 355.83 ± 61.28 fire ants per trap versus 200.33 ± 31.41 fire ants per trap, control versus Amdro® treated plots, respectively, F_1,29 = 4.63; P = 0.04).

We identified at least 6 different aphid species on tomato plants in our plots, including Aphis gossypii, A. fabae, Aulacorthum solani, Macrosiphum euphorbae, Myzus persicae, and Uroleucon spp. In 2003, aphid abundance was significantly different between treatments for apterous aphids (F_1,30 = 9.82; P = 0.004) and for total aphids (apterous + alate) (F_1,30 = 8.57; P = 0.01) (Fig. 1A). The abundance of alate aphids, however, was not statistically different between treatments (F_1,30 = 1.76; P = 0.19). In 2004, however, the abundance of alate aphids (F_1,29 = 26.4; P < 0.0001) and total aphids (F_1,29 = 27.2; P < 0.0001) were significantly different between the two treatments (Fig. 1B). In contrast to 2003, alate aphids made up over 95% of total aphids in both control and suppression plots in 2004. Likewise, during our survey of fresh market tomato fields, the abundance of fire ants and aphids was positively correlated in 2004 (one-tailed test; r = 0.09; P = 0.04; n = 4) and during 2005 (one-tailed test; r = 0.91; P = 0.02; n = 5) and greater than 95% of these aphids were alates.

In 2003 and 2004, several non-aphid herbivores were found feeding on tomato plants (Table 1), although their overall densities were quite low, especially in 2004. Non-aphid herbivore abundance, however, was not significantly different between treatments in both 2003 (2.25 ± 0.2 per plant versus 2.26 ± 0.2 per plant, control versus Amdro® treated plots, respectively) (F_1,24 = 3.89; P = 0.06) and 2004 (0.32 ± 0.06 per plant versus 0.4 ± 0.05 per plant, control versus Amdro® treated plots, respectively) (F_1,23 = 0.00; P = 0.96).

Several natural enemies also were present on tomato plants in our experimental tomato plots in 2003 and 2004 (Table 2). In 2003, the most abundant natural enemies were spiders and coccinellid larvae (Coccinellidae). Natural enemy abundance, however, was not significantly different between the 2 treatments (0.10 ± 0.02 per plant versus 0.09 ± 0.03 per plant, control versus Amdro® treated plots, respectively) (F_1,24 = 0.01; P = 0.93). In 2004, the most abundant natural enemies were damsel bugs (Nabidae) followed by spiders and minute pirate bugs (Anthocoridae). There was a significant difference (F_1,29 = 4.31; P = 0.04) in natural enemy abundance between suppressed fire ant plots and control plots with significantly more natural enemies on tomato plants in fire ant suppressed plots (0.13 ± 0.03 per plant versus 0.20 ± 0.03 per plant, control versus Amdro® treated plots, respectively).

Discussion

Total aphid abundance in both 2003 and 2004 was significantly greater in tomato plots with high densities of fire ants (control plots) than in plots with suppressed fire ant densities. In 2004 alate aphids were more than twice as abundant in high fire ant plots as in low fire ant plots and there was a positive correlation between the abundance of fire ants and alate aphids in commercial tomato fields in 2004 and 2005. Taken together, these results suggest that fire ants increase the abundance of important aphid pests of tomato. These results are consistent with studies conducted in other agricultural crops indicating that fire ants often increase aphid abundance. For example, fire ant ‘tending’ of aphids increases the abundance of cotton aphids (Aphis gossypii) in cotton (Kaplan & Eubanks 2002, 2005; Diaz et al. 2004) and brown citrus aphids (Toxoptera citricida) in citrus (Michaud & Browning 1999). By essentially doubling aphid abundance in tomato, fire ants may significantly reduce tomato yield and fruit quality, especially if these aphids are carrying plant viruses that infect tomato. All of the aphids that we identified in our field experiment (A. gossypii, A. fabae, A. solani, M. euphorbae, M. persicae, and Uroleucon spp.) have been identified as vectors of CMV and other plant viruses that can devastate tomato production (Sikora et al. 1998; Palukaitis et al. 1992). Our results suggest that suppression of fire ants in and around agricultural fields will significantly reduce the abundance of aphids and may reduce the spread of aphid-vectored plant viruses. We plan to test this hypothesis in the near future.

We found limited evidence that fire ants affected the abundance of non-aphid herbivores in tomato. This is surprising because many previous studies conducted in other crops have found that fire ants suppress herbivores (Eubanks, 2001; Vogt et al. 2001; Diaz et al. 2004; Harvey & Eubanks 2004). For example, Eubanks (2001) found that densities of S. invicta workers were negatively associated with densities of several herbivores in both cotton and soybean. Likewise, Diaz et al. (2004) found that fire ants destroyed significant numbers of bollworm (Helicoverpa zea) and beat armyworm eggs (Spodoptera exigua (Hubner)) in cotton. Additionally, Vogt et al. (2001) found that fire ants forage on many herbivores in peanut fields, suggesting that fire ants may have negative effects on many herbivorous taxa. One reason for the difference in our findings from other studies could be the composition of herbivores present or their relative abundance. The non-aphid herbivores in tomatoes were dominated by whiteflies and lepidopteran larvae (Table 1). Like aphids, whiteflies produce honeydew and are sometimes tended by fire ants (Queiroz & Oliveria 2001), so it is unlikely that fire ants are significant predators of whiteflies in tomato. In 2004 the most abundant herbivores were first and second instar lepidopteran larvae, but they were present at relatively low densities. Predation of caterpillars by fire ants has been shown to be density-dependent and to some extent size-dependent in other crops (e.g., collards (Harvey & Eubanks 2004) and soybean (Styrsky et al. 2006)) and their low densities in tomato may have resulted in low fire ant predation during our plot experiments.

The number of natural enemies was not significantly different between treatments in 2003. These results are inconsistent with results published by Bugg & Dutcher (1989), Eubanks et al. (2002), Kaplan & Eubanks (2002), and Vogt et al. (2001). The conflicting results may be attributed to the low density of natural enemies present and the behavior of the natural enemies. In 2003, only 2 kinds of natural enemies, spiders and lady beetle larvae (Coccinellidae), were found during visual searches and spiders were much more abundant than lady beetle larvae. Some spiders can avoid predation by foraging fire ants by rapidly moving to other parts of the plant, dropping off of the plant on silk lines, or stretching to reduce contact with the plant surface (Eubanks et al. 2002). In 2004, however, fire ants did negatively affect the abundance of natural enemies. A greater diversity of natural enemies was found during visual searches in 2004 than in 2003. Other studies have shown that damsel bugs and minute pirate bugs, the first and third most abundant natural enemies in our tomato plots in 2004, were much more susceptible to fire ants than spiders (Eubanks 2001; Eubanks et al. 2002). Thus, the difference in natural enemies among the 2 years of the study may explain the greater impact of fire ants on beneficial insects in 2004.

In conclusion, fire ants had strong, positive effects on aphid abundance and insignificant effects on non-aphid herbivores. Taken together, the results of this study suggest that fire ants may disrupt biological control of aphids in this crop and the suppression of fire ants will decrease aphid populations in tomato.