The native ant, Tapinoma melanocephalum, improves the survival of an invasive mealybug, Phenacoccus solenopsis, by defending it from parasitoids

Mutualistic ants can protect their partners from natural enemies in nature. Aenasius bambawalei is an important parasitoid of the the invasive mealybug Phenacoccus solenopsis. We hypothesized that mutualism between native ants and mealybugs would favor survival of mealybugs. To test this, we examined effects of tending by the native mutualistic ant Tapinoma melanocephalum on growth of P. solenopsis colonies on Chinese hibiscus, Hibiscus rosa-sinensis, in a field setting. Ant workers with access to honeydew of mealybugs lived much longer than those provisioned only with water in the laboratory, and number of ant workers foraging increased significantly with growth of mealybug colonies in the field. In later observations, there were significant differences in densities of mealybugs between ant-tended and -excluded treatments. Survival rate of mealybugs experiencing parasitoid attack was significantly higher on ant-tended plants than on ant-excluded plants. When the parasitoid was excluded, there was no difference in survival rate of mealybugs between ant-tended and -excluded plants. In most cases, ants directly attacked the parasitoid, causing the parasitoid to take evasive action. We conclude that native ants such as T. melanocephalum have the potential to facilitate invasion and spread of P. solenopsis in China by providing them with protection from parasitoids.

Since P. solenopsis secretes honeydew, it often engages in mutualistic interactions with ants, e.g., the red imported fire ant, Solenopsis invicta Buren 14 . Another study suggested that S. invicta utilized shelters constructed by the cotton leaf roller, Sylepta derogata F. (Lepid.: Pyralidae), to help shield P. solenopsis from its natural enemies 40 . However, these studies did not address whether this invasive ant provided direct protection of P. solenopsis against parasitoids or predators 14,40 .
Native ant species are abundant in the areas of China invaded by P. solenopsis and the ghost ant, Tapinoma melanocephalum (F.), is one of the most dominant. This species can be observed tending P. solenopsis and sometimes transporting individuals from lower to upper leaves on H. rosa-sinensis plants 41 , but we have not observed any predation of the mealybugs. To date, native ants retain a broader distribution in China than the red invasive fire ant, Solenopsis invicta Buren, so P. solenopsis may have many opportunities to establish mutualisms with native ants in the process of range expansion. Here, we address the question of whether mutualisms can evolve between native ants and P. solenopsis, and whether their aggressive behavior towards parasitoids can improve the survival of invasive mealybug colonies. We hypothesized that P. solenopsis colonies able to establish mutualistic relationships with native ants would grow larger and suffer less parasitism than colonies prevented from doing so. To test this hypothesis, a series of experiments were conducted to monitor the growth and survival of P. solenopsis colonies with and without the benefits of tending by T. melanocephalum workers. The results of these experiments may help us anticipate the potential role of native mutualistic ants in facilitating the invasion and range expansion of exotic honeydew-producing hemipterans.

Results
Mutualism in the field. There was a significant positive correlation between the numbers of ant workers foraging and numbers of P. solenopsis over time (F = 147.36; df = 1,8; P < 0.0001) with variation in one variable explaining almost 95% of variation in the other (Fig. 1). Densities of P. solenopsis differed significantly between ant-tended and ant-excluded treatments (F = 126.45; df = 1,4; P = 0.0004) and among sampling dates (F = 161.79; df = 9,72; P < 0.0001) and there was a significant treatment*sampling date interaction (F = 12.62; df = 9,72; P < 0.0001). Differences became significant on the fifth sampling date and remained different thereafter (Fig. 2).

Discussion
There are numerous species of native ants with extensive distribution throughout China, providing P. solenopsis with many opportunities to establish mutualisms that may enhance its invasion and spread. We chose a dominant native ant species, T. melanocephalum, to test whether mutualism with a native ant would improve survival of this invasive mealybug. The greater increase in population density of P. solenopsis in the ant-tended treatment than in the ant-excluded treatment demonstrates the benefit of ant tending for the mealybugs, and the increased longevity of workers in the laboratory feeding trial suggests  a reciprocal benefit for the ants. Furthermore, the positive correlation between numbers of P. solenopsis and numbers of ants foraging indicates that T. melanocephalum workers recruit in direct proportion to the numbers of P. solenopsis present ( Figure S1). Many previous studies have shown that hemipterans provide honeydew for ants that supports the growth of their colony [5][6][7]9,11,14 . For example, the red imported fire ant, S. invicta, can utilize honeydew produced by P. solenopsis to promote colony growth 14 .
In return for carbohydrates, ants often protect their mutualistic partners from natural enemies 2,5,11,17,18 . For example, the role of ants as guardians of aphids against predators and parasitoids is well documen ted 3,10,13,18,19,[42][43][44][45] and similar mutualisms have evolved between ants and other hemipterans 5,6,9,14,17 and butterflies 2,46-48 . In the present study, the parasitoid A. bambawalei parasitized more P. solenopsis on ant-excluded plants than on ant-tended plants, such that the colonies of P. solenopsis grew significantly larger on the latter. When the parasitoid was excluded, there were no differences in numbers of P. solenopsis between ant-excluded and ant-tended plants, clear evidence that T. melanocephalum acts to protect P. solenopsis from A. bambawalei. These results are consistent with the findings of previous work in India 49,50 .
Differences in levels of interspecific aggression between the ant and the parasitoid were evident in the behavioral observations. A high level of interspecific aggression was directed toward A. bambawalei by T. melanocephalum that was not reciprocated; the parasitoid responded to ant encounters mostly  with evasive movements and defensive postures. In summary, the present study provides several lines of evidence to indicate that native ants will likely reduce the effectiveness of the introduced parasitoid A. bambawalei and promote the successful invasion and spread of P. solenopsis in China.

Plants.
To assess the benefits of honeydew consumption by ants, seeds of Chinese hibiscus, Hibiscus rosa-sinensis L., calabaza squash, Cucurbita moschata Duch. Ex Lam, tomato, Solanum esculentum L. and sunflower, Helianthus annuus L., were sown in plastic flowerpots (18 cm diam) filled with a loamy clay soil. Plants were grown in a greenhouse at 26-30 o C, 65 ± 5% RH, and a 14:10 L:D photoperiod, watered once every four days, and fertilized (N:P:K = 13:7:15) twice a month. These plants were infested with mealybugs when they reached a height of about 20 cm. Plants of H. rosa-sinensis for cage experiments with ants were grown in the same manner, but were used in experiments when they reached a height of 50 cm.
Insects. None of the study species are protected in China, so no specific permits were required for collections or field activities. All source material was collected from plants of H. rosa-sinensis on the campus of South China Agricultural University, Guangzhou, Guangdong Province, China. A colony of P. solenopsis was established on H. rosa-sinensis plants that had been grown in 18 cm diameter plastic flowerpots in a greenhouse. The potted plants were used in experiments when they were ca. 50 cm tall with 50-60 true leaves. Each plant was infested by transferring ca. 60 first instar mealybug nymphs directly to the leaves. Each plant was then isolated in a ventilated aluminum frame cage (60 cm × 60 cm × 60 cm). Four generations were reared to obtain sufficient insects for use in experiments. All mealybug colonies were reared in the laboratory at 27 ± 1 o C and a relative humidity of 60-70%. Third instar nymphs were used in experiments.
Mealybug nymphs parasitized by A. bambawalei were collected from H. rosa-sinensis plants on the campus of South China Agricultural University and reared out in the laboratory. Parasioids emerging from mummified mealybugs were identified and raised for four generations using ca. 100 adult females to start each generation. All wasps used in experiments were newly emerged adults. All parasitoid colonies were reared in the laboratory under the same physical conditions as the mealybugs with a 10% honey solution provided on cotton balls as food for the parasitoid adults.
We found a total of 11 native ant species on the university campus, with T. melanocephalum the most abundant species 41 . Workers of T. melanocephalum were collected from Hibiscus rosa-sinensis plants on the campus and fed on a 10% honey solution. Six newly established colonies of T. melanocephalum were collected from the campus of South China Agricultural University, each including one queen and a collection of adult workers, eggs, larvae and pupae. Each ant colony was housed in a 1.5 L plastic box and provisioned with a 10% honey solution in tubes with cotton wicks. plant to allow foraging workers to access the plant 14 . For treatments including parasitoids, once ants had colonized the plant, eight pairs of A. bambawalei adults were released in the cage and left to forage for 24 h. After two weeks, the number of the mealybugs was counted on all plants in the experiment. In addition, the numbers of parasitized mealybug nymphs were recorded on plants with and without ants.

Ant-parasitoid behavioral interactions. For each experimental replicate, a fresh H. rosa-sinensis
leaf was placed in a glass petri dish (Φ 9.0 cm), and ten third instar mealybug nymphs were placed in the center of the leaf. An ant and a parasitoid were introduced into the dish simultaneously, and the behavior of T. melanocephalum toward A. bambawalei directly observed during a 10 min period of interaction. Interactions were scored for each insect as described in 51 : 0 = ignore/mutual non-interference; 1 = touch, antennation or grooming; 2 = avoidance or evasive behavior following contact; 3 = dorsal flexion or other defensive reaction; 4 = fighting, prolonged aggression, sparring, charging, biting and pushing. The experiment was replicated five times with ten pairs of insects in each replication for a total of 50 trials. The mean of all replications was used to tally a score for each interaction category for both ant and parasitoid.
Statistical Analyses. Mutualism in the field: Logarithmic regression was used to describe the relationship between the numbers of ant workers foraging and numbers of P. solenopsis over time. A 2-way ANOVA for repeated measures was used to test for significant differences between treatments and determine the date on which ant-tended plants became significantly different from ant-excluded plants. Longevity of T. melanocephalum workers fed P. solenopsis honeydew: Data on the response variable 'ant longevity' were found to be normally distributed and were subjected to one-way ANOVA followed by Fisher's LSD (α = 0.05) to separate means among independent variables (plant species). Parasitism by A. bambawalei in the presence of ants: A one-way ANOVA was used to analyze treatment effects (presence of ants and/or parasitoids) on the response variables 'no. mealybugs' and 'percent parasitism' . Ant-parasitoid behavioral interactions: A Chi Square, Goodness of Fit test was used to test for differences between ants and parasitoids in proportion of time spent in different categories of aggressive behaviors. All statistical analyses were conducted using SPSS version 16.0 (SPSS Inc., Chicago, IL, USA).