Does Mutualism Drive the Invasion of Two Alien Species? The Case of Solenopsis invicta and Phenacoccus solenopsis

Although mutualism between ants and honeydew-producing hemipterans has been extensively recognized in ecosystem biology, however few attempts to test the hypothesis that mutualism between two alien species leads to the facilitation of the invasion process. To address this problem, we focus on the conditional mutualism between S. invicta and P. solenopsis by field investigations and indoor experiments. In the laboratory, ant colony growth increased significantly when ants had access to P. solenopsis and animal-based food. Honeydew produced by P. solenopsis also improved the survival of ant workers. In the field, colony density of P. solenopsis was significantly greater on plots with ants than on plots without ants. The number of mealybug mummies on plants without fire ants was almost three times that of plants with fire ants, indicating a strong effect of fire ants on mealybug survival. In addition, the presence of S. invicta successfully contributed to the spread of P. solenopsis. The quantity of honeydew consumption by S. invicta was significantly greater than that of a presumptive native ant, Tapinoma melanocephalum. When compared with the case without ant tending, mealybugs tended by ants matured earlier and their lifespan and reproduction increased. T. melanocephalum workers arrived at honeydew more quickly than S. invicta workers, while the number of foraging S. invicta workers on plants steadily increased, eventually exceeding that number of T. melanocephalum foragers. Overall, these results suggest that the conditional mutualism between S. invicta and P. solenopsis facilitates population growth and fitness of both species. S. invicta tends to acquire much more honeydew and drive away native ants, promoting their predominance. These results suggest that the higher foraging tempo of S. invicta may provide more effective protection of P. solenopsis than native ants. Thus mutualism between these two alien species may facilitate the invasion success of both species.


Introduction
Mutualistic interactions occur commonly between hemipteran and invasive ants, which are notorious for their aggressiveness and high colony density [1][2][3]. Ecosystem domination by invasive ants is effectively strengthened by the collection and exploitation of honeydew and plant extrafloral nectar [2][3][4][5]. Colony growth of both insects is facilitated by the interactions between ants and honeydew-producing hemipterans [6]. The growth of honeydewproducing hemipterans has been shown to be facilitated by ant tending [2,[7][8][9][10]. Increases in hemipteran density attract more aggressive and dominant ants, however, resulting in the dislodging of native ants in many cases [11,12].
As one of the most important threats to ecosystems, many invasive species are recognized by their extreme aggression and broad omnivory [2,[13][14][15]. Native communities can be significantly damaged by invasive species [2]. Ecosystems are commonly disrupted by invasive species that exploit existing mutualisms [16]. Hemipteran communities and colonies can be supported by ant tending because certain ant species protect honeydew-excreting hemipterans from their natural enemies [17]. For example, the density of the obscure mealybug, Pseudococcus viburni, in California coastal vineyards significantly increased when tended by the Argentine ant, Linepithema humile, and the density of encyrtidae parasitoids and predators decreased in vineyards with the Argentine ant [18]. Besides reducing the mortality risk of fungal infection [9,19,20], ant tending may enhance aphid colonies by the stimulation of feeding and honeydew excretion [21,22]. Mutualism between ants and aphids may have positive effects on the aphid life cycle. For example, individual Metopeurum fuscoviride Stroyan aphids tended by Lasius niger ants lived longer, matured earlier, had a higher rate of reproduction, and had a higher expected number of offspring than aphids that were not tended by ants [23].
In exchange for tending of the hemipterans, the ants receive large amounts of honeydew in such a consumer-resource mutualism [24]. Tended hemipterans usually supply abundant honeydew, which is considered to be an important food resource for ants because it contains sugars mixed with various amino acids and energy-rich materials [25][26][27][28]. Invasive ant species are usually omnivorous and have enormous populations; they not only utilize animal-based food resources in their surroundings but also feed on plant juices and the honeydew excreted by hemipteran insects, which can facilitate colony growth to some extent [2,3,29].
Compared with plant-based food resources such as plant juices and honeydew, the effects of animal-based food resources on the population growth of omnivorous ants are much more significant [30]. Animal tissues are rich in protein, which can significantly facilitate the growth and development of ant larvae [31,32]. However, when animal-based food is scarce, honeydew becomes indispensable for sustaining many ants. Other research has indicated that laboratory colonies of S. invicta are substantially enhanced when supplied with honey water [33]. Colonies of S. invicta reared in the absence of other insects ceased brood production entirely, and colony growth was reduced by 60% because of the lack of sugar-water supply [34]. Ant workers cannot digest solid food directly, and honeydew and plant juices are composed mainly of carbohydrates that can supply energy for worker activity [29,35,36].
The red imported fire ant, Solenopsis invicta, is native to South America and is a dangerous invasive species in the United States [37]. Many studies have reported important ecological effects of S. invicta, including decreased local biodiversity and the disruption of native ant communities [37,38]. Similarly, negative effects of S. invicta on agriculture and forestry production, human health and poultry production have been recorded in South China [39]. The mealybug Phenacoccus solenopsis Tinsley (Hemiptera: Pseudococcidae) is native to the US and has spread throughout the world [40]. It has a wide geographic distribution and can be found in Central America, South America and Africa [41,42]. It has been found to cause serious damage to cotton crops in India and Pakistan in 2005 and to Hibiscus rosa-sinensis in Nigeria [43,44]. Recently, P. solenopsis was reported to be an important invasive species in South China [45]. Obviously, it is a common ecological phenomenon for the mealybug, P. solenopsis to form close relationships with ants, similar to that between aphids and ants. Few studies, however, have investigated such a relationship.
The mechanisms promoting invasion is a core issue in invasion biology, and the study of the role of the interaction between species in the invasion process, including competition and facilitation, has become heated in recent years. Liu et al. [46] explored the asymmetric mating mechanism of whitefly invasion from the point of view of species competition. Meanwhile, the effect of mutualism and positive facilitation on successful invasion has been commonly acknowledged [47][48][49]. For example, some studies have found that the invasion of some species may promote other species' successful settlement and invasion by changing the habitat characteristics and species composition of the infested area [47]. The interaction between invasive and native species [50], and the facilitation of the invasion process by the interaction between an alien insect and its symbiotic bacteria [51] have been documented. However, few studies have addressed the mutualism and its contribution to invasion of two alien insect species.
Previous studies have suggested that the invasive ant S. invicta tends the invasive mealybug Antonina graminis (Maskell) extensively and actively constructs shelters around these insects. Additionally, honeydew produced by A. graminis is an important component of the diet of S. invicta [3]. Helms and Vinson [52] indicated that colonies of S. invicta grew substantially larger when supplied with insect prey and honeydew produced by A. graminis than those in other treatments with access to unlimited insect prey. This study also showed that nutritional resources for S. invicta were unlikely to be acquired directly from hemipteran host plants; rather, they were provided indirectly from honeydew. In a previous field investigation, we found that fire ant workers foraged more frequently on plants when mealybugs were present [53], suggesting that S. inivcta in south China may be deriving benefits from another invasive insect, P. solenopsis. Based on these results, we developed the following hypothesis: Compared with native ants, fire ants are able to form closer mutually-beneficial relationships with P. solenopsis, which may enhance colony development and facilitate their invasion success. To test this hypothesis, we conducted a series of experiments to examine the effects of honeydew excreted by P. solenopsis on the growth of S. invicta colonies. We also determined the effects of fire ant tending on the growth of colonies of P. solenopsis in the field. Furthermore, the impacts of ant tending by S. invicta and native ants on single individuals and growth of small colonies of P. solenopsis were examined. The results of these experiments may provide insights into how mutualism between invasive ants and exotic honeydew-producing hemipterans can promote the invasion success of both partners.

Plants
Hibiscus rosa-sinensis, a Chinese native species cultivated worldwide, was purchased from a commercial horticultural farm. Each plant was approximately 25-30 cm in height and had 25-30 true leaves. All plants were cultivated in plastic flowerpots (the diameters of the upper and lower edges were 18 cm and 14 cm, respectively, with a height of 17 cm) in greenhouses.

Insects
Colonies of P. solenopsis were collected from the campus of South China Agricultural University and fed on H. rosa-sinensis. The 1 st  instar mealybug nymphs were inoculated on each plant and raised for several generations. H. rosa-sinensis plants with established mealybug colonies were used for subsequent experiments. All colonies were reared in the laboratory with the temperature maintained at 2762uC and a relative humidity of 60-70%. We collected 16colonies of Tapinoma melanocephalum, a highly competitive and sugar-feeding species [54] from the campus of South China Agricultural University. This species is a worldwide invader whose native range is unknown but is believed to come from Africa or Asia [55]. It has been in southern China for a long time where it is fully established as a resident species. For the purposes of this paper it is considered a presumptive native of southern China.
A total of 51 newly established colonies of S. invicta were collected from a suburb of Guangzhou and reared in plastic boxes (116 L). All colonies were separated from the soil by dripping water into plastic boxes until the colonies floated (Jouvenaz et al. 1977). Each colony was divided into more than 20 sub-colonies (approximately 1.0 g) using a microbalance (Sartorius, BS, 224S). Each sub-colony included one queen and adult workers, pupae, larvae, and eggs. The ants were placed in a 9-cm plastic Petri dish, which served as an artificial nest. T. melanocephalum sub-colonies were maintained with tubes filled with distilled water plus 10% honey solution. Fire ants were given fresh live Tenebrio molitor worms, frozen crickets, and a 10% solution of honey mixed with water (50 ml) weekly.
For each experiment, we used a number of different colonies. Each colony was divided into a number of sub-colonies, with only one sub-colony per colony used for each treatment. Thus, each colony was represented by only one sub-colony per treatment in each of the experiments.

Experimental design
Experiment 1: Effect of food composition on colony growth of S. invicta. We placed a small fluon-coated plastic case (40 cm628 cm622 cm) that was loaded with soil into a fluoncoated plastic box (52 cm639 cm630 cm). Each small plastic case was seeded with a colony of S. invicta. By 24 h, the ants had constructed a new nest in the soil. One hundred 1 st instar mealybugs were transferred to potted plants. A plastic hose was used to build a bridge between the ant nest and the base of the plant stem for worker foraging. Pots were coated with fluon to prevent ant escape. Our experiments included the following treatment groups: (1) water supply only; (2) one potted H. rosasinensis plant placed in the plastic drum; (3) live worms plus frozen cockroaches; (4) one potted H. rosa-sinensis plant plus live worms and cockroaches; (5) one H. rosa-sinensis plant infected with P. solenopsis; and (6) one H. rosa-sinensis plant infected with P. solenopsis plus live worms and cockroaches. One test tube (1561.5 cm) filled with distilled water and sealed with a cotton plug was placed into a small plastic case as a constant source of water. To prevent drying, the soil in each small plastic case received sprayed water every 3 days. Each treatment was replicated 15 times. The experiments lasted for 8 weeks after which time, all surviving ants were extracted, counted, weighed, and measured using a microbalance (Sartorius, BS, 224S).

Experiment 2:
The effect of food variety on the survival of worker ants. One gram (approximately 950-1000 individuals) of workers was placed in a plastic box (26 cm618 cm68 cm) with distilled water supplied every day, and a petri dish (7 cm diameter) with moist plaster was used as an artificial nest. One hundred 1 st instar mealybugs were transferred to the potted plants. Paraffin wax was used to coat the base of the plants to prevent escape by the mealybugs. We placed the potted plant and artificial ant nest in a large plastic box (50 cm640 cm616 cm) after the mealybugs had colonized the plants (reared previously). At the beginning of the bioassay, a plastic hose was used to build a bridge between the ants' nest and plant seedlings to allow worker foraging. The ants were raised with access to different food supplies: (1) distilled water only; (2) live worms; (3) a potted plant inoculated with P. solenopsis; (4) live worms and a potted plant inoculated with P. solenopsis. The experiments were conducted for 5 weeks. We counted the number and measured the weights of the surviving ants using a microbalance (Sartorius, BS, 224S). Each food supply treatment was replicated 10 times. We calculated the survival rate of the ants as 100% 6 the number of surviving ants after tests/the number of surviving ants before tests. We determined the effects of mealybug density on the survival of the ant workers. The experimental protocols were performed as described above, with 10 replicates per condition. The tests were finished after 5 weeks. The number of surviving ants was counted. Worker survival rate was calculated as 100% 6 the number of surviving ants after the tests/the number of surviving ants before the tests.  (2) plots without fire ant infestation were chosen to exclude colonies of S. invicta, and the base of the main stem of the plants was covered with paraffin, (3) plots without fire ant infestation were chosen to exclude colonies of S. invicta, but the base of the plant stem was not covered with paraffin. Sizable colonies of S. invicta were inoculated on the appointed plots. We then checked whether the transferred colonies had survived, and a second colony of S. invicta was provided if the first colony had not successfully established. In each treatment, a circle with its center point at the ant colony and a radius of approximately 1 m was drawn, and, four H. rosa-sinensis plants were placed within the circle at 90-degree angles. Two neighboring plants had been infected with mealybugs, whereas the other two (also neighboring) plants were uninfected. Individual 1 st instar mealybug larvae were transferred onto the plant via small plastic tubes with cotton plugs prior to their introduction to the study site. For this transfer, we placed four tubes, each containing 100 individuals, on the top branches of each H. rosa-sinensis plant. When the plug was removed, the nymphs crawled out from the tubes and began sucking the tender plant leaves. We investigated the hypothesis that the presence of S. invicta colonies facilitates the growth and spread of P. solenopsis colonies. One mealybug-infected plant was selected randomly for our investigation (a different plant was observed if the colony of P. solenopsis had disappeared). The number of foraging workers moving up or down the plant trunk during a 3-min period was counted, and the density of live P. solenopsis was determined every 7 days. The quantity of mealybug mummies and the probability of P. solenopsis spread were also measured at the end of the investigation. We assumed that the probability of spreading was 100% if both mealybug-absent plants were colonized by P. solenopsis, 50% if one was infected and 0 if neither plant was infected. Each treatment was replicated 10 times.

Experiment 5: Consumption of honeydew by S. invicta
and T. melanocephalum. H. rosa-sinensis seedling leaves were inoculated with 60 3 rd instar mealybugs, which were subsequently reared on the plants. Artificial nests (one queen and 1 g workers) of S. invicta and T. melanocephalum were transferred to individual plastic cases (40 cm628 cm622 cm). The workers began to build the nests immediately. After 24 h, mealybug-infected plants were placed into each plastic case. A plastic hose was used to build a bridge between the ants' nest and the stem of the plant to allow worker foraging. We collected 30 random workers from the bottom stalk at the beginning of the experiment, as they were moving toward the mealybugs, and 30 more were collected after 24 h, as they were returning from the mealybugs. The weight of the ants collected before and after foraging was measured by a microbalance (Sartorius, BS, 224S). Every treatment was replicated 10 times.   invicta and T. melanocephalum. Each colony (one gram workers and one queen) of S. invicta and T. melanocephalum was transferred to a plastic box (26 cm618 cm68 cm). One hundred 1 st instar mealybugs were transferred to H. rosa-sinensis plants. After abundant honeydew has present on plants, each nest of S. invicta and T. melanocephalum received a plastic pipe (with the length and the diameter of 35cm and 0.8cm, respectively) which was used to build a bridge between the nest and plant stem for worker foraging. In order to contain the ants and mealybugs, the base of the plant stems were coated with ceresin wax. A colony of S. invicta and T. melanocephalum was connected with the plants. We recorded the search time (time for the first ant to arrive at the honeydew) and recruitment time (time for 10workers to be present on the plant) of S. invicta and T. melanocephalum. In addition, to determine the level of competition between S. invicta and T. melanocephalum for honeydew, we connected two ant colonies to one plant. We counted the number of foraging workers of the two ant species on the plant every 10minutes. The experiment lasted for 80 minutes. All treatments were replicated 10 times.
We conducted the field studies in areas where fire ants and mealybugs occur in the field and no specific permits were required for the described field studies. The land used in the study area is not privately-owned or protected in any way and the field studies did not involve endangered or protected species.

Statistical analysis
To compare the differences in colony weight and survival rate between treatments, survival rate between densities, quantity of foraging ant workers per plant and probability of P. solenopsis spread between ant-including and ant-excluding plots, and the developmental duration, lifespan, searching time and recruitment time of workers, all data were tested for normal distribution by Shapiro-Wilk test and for homogeneity of variances by Levene's test. If the data were normally distributed and had similar variances, then one-way analysis of variance (ANOVA) using Type III sum of squares was performed to compare means among all measured variables. When ANOVA results were significant, multiple comparisons of means were performed with Tukey HSD post-hoc analysis. If the data did not have similar variances, the non-parametric Kruskal-Wallis test for comparing the median was applied and also the Mann-Whitney test (or the two-sample Kolmogorov-Smirnov test) for multiple comparisons among the different groups if the results of the Kruskal-Wallis test showed significant differences at the 0.05 significance level.
The density of P. solenopsis per plant, quantity of mealybug mummies on plants in ant-including and ant-excluding plots, changes in ant weight and honeydew consumption between the two ant species, and differences in reproductive quantity among  single mealybugs and small colonies of P. solenopsis tended by S. invicta and T. melanocephalum were analyzed with independentsamples t-tests. We used paired tests to compare worker weights before and after foraging.
In addition, all proportion data, such as proportion spread or survival, were binomially distributed and analyzed after arcsin square root transformation. All statistical analyses were conducted using SPSS version 14.0 (SPSS Inc., Chicago, IL, USA).

The effect of food composition on colony growth in S. invicta
The variable food composition resulted in markedly different colony masses under different treatments (x2 = 65.187, p,0.001). Colony mass was not obviously different between the animal food only treatment and the animal food with plants treatment (Mann-Whitney test, U = 97.00, P = 0.520). However, colony mass in the treatment provided with animal food with plants and mealybugs was significantly greater than that receiving only animal food (Mann-Whitney test, U = 64.00, P = 0.044, Table 1). In addition, colony growth of mealybugs was greater when fire ants were provided with animal source of foods than without (t = 25.629, P,0.001, Fig. 1).

The effect of food variety on survival of worker ants in S. invicta
The survival of S. invicta workers differed in different food varieties (F = 17.841, P,0.001, Fig. 2). Worker ants provided with distilled water only had the lowest survival. There was a significant difference between the water-only and animal-based food conditions (P = 0.01). The survival of workers was significantly increased when worker ants had access to mealybugs or mealybugs plus animal-based food compared with the other treatments (P,0.01, Fig. 2); there was no difference, however, in the survival between these two treatments (P = 0.638).
3 The effect of P. solenopsis density on ant worker survival in S. invicta The survival of workers increased significantly with greater mealybug density (F = 21.645, df = 3, P,0.001, Fig. 3).

Foraging dynamics of S. invicta workers and mealybug density on plants
The average fire ant forging activity was markedly higher in the ant-tended plants than in the ant-excluded plants (t = 24.756, df = 5, P = 0.005, Fig. 4a). Few workers were present on antexcluded plants (Fig. 4a). There was no obvious difference in mealybug density between the ant-tended and ant-excluded plants in the first observation

Effect of ant tending by S. invicta on the quantity of mealybug mummies
Our results indicate that the number of mealybug mummies on ant-excluded plants was significantly greater than that on anttended plants (Mann-Whitney test, U = 13.5, P = 0.006, Fig. 5), which indicated that fire ants drive parasitic wasps away from the mealybugs. The probability of P. solenopsis spread was conspicuously different among the three treatments (F = 9.277, df = 2, P = 0.001, Fig. 6). There was no significant difference in the probability of P. solenopsis spread between the treatments of plants with and without paraffin in ant-excluded plots (Mann-Whitney test, U = 40.0, P = 0.146).

melanocephalum
We recorded the difference in ant worker weights between ants moving in two different directions (i.e., prior to honeydew consumption and after honeydew consumption) in the two species. Paired t-tests showed that returning workers were significantly heavier than outgoing workers on plants with P. solenopsis for both ant species (S. invicta: t = 23.713, df = 9, P = 0.005; T. melanocephalum: t = 25.546, df = 9, P,0.001; Fig. 7). Therefore, both ant species were able to collect abundant honeydew from P. solenopsisinfested plants. However, we detected a significantly greater amount of honeydew consumed by workers S. invicta than by workers of T. melanocephalum (Mann-Whitney test, U = 9.0, P = 0.002; Fig. 8).
7 Effect of S. invicta and T. melanocephalum tending on a single P. solenopsis Single mealybugs tended by S. invicta and T. melanocephalum matured an average of 1.2 and 1.7 days earlier than mealybugs tended by no ants, respectively, but there were no significant differences among the three tending treatments (F = 0.736, df = 2, P = 0.489; Fig. 9). The lifespan and reproduction of single mealybugs were significantly increased by tending by both ant species. Tending by S. invicta and T. melanocephalum extended the lifespan of single mealybugs by 5.6 d and 4.7 d, respectively (F = 3.487, df = 2, P = 0.046; Fig. 9), and the number of offspring increased by 114.2 and 94.7 individuals, respectively (F = 5.190, df = 2, P = 0.013; Fig. 9).
8 The effect of S. invicta and T. melanocephalum tending on small colonies of P. solenopsis There was no obvious difference in the reproduction of colonies of P. solenopsis in the presence and absence of ant tending (F = 0.880, df = 2 P = 0.426, Fig. 10). The number of offspring and oocysts produced by P. solenopsis colonies tended by S. invicta and T. melanocephalum were also similar (F = 0.347, df = 2 P = 0.710, Fig. 10). 9 Offspring production in single and small colony P. solenopsis under tending by S. invicta and T. melanocephalum An independent-samples t-test showed that reproduction of P. solenopsis colonies decreased significantly when tended by T. melanocephalum (t = 2.127, df = 19, P = 0.047, Fig. 11). In contrast, there was no significant difference in the reproduction of colonies on single P. solenopsis when tended by S. invicta (t = 0.958, df = 16, P = 0.353, Fig. 11). Without ant tending, there was no distinct difference in reproduction between colonies and single P. olenopsis (t = -1.298, df = 18, P = 0.211, Fig. 11).   10 Comparison of foraging behavior between S. invicta and T. melanocephalum to honeydew Workers of T. melanocephalum arrived at honeydew more quickly than workers of S. invicta. But there was no significantly difference in searching time between the two ant species (t = 1.516, df = 17, p = 0.148, Fig. 12). Recruiting time of T. melanocephalum was significantly shorter than that of S. invicta (t = 4.520, df = 13.53, p = 0.001, Fig. 12).
Our results indicated that number of foraging workers of T. However, by the seventh and eighth observation period, the number of foraging workers of S. invicta was significantly greater than that of T. melanocephalum (t = 2.76, df = 14.16, p = 0.015; t = 3.901, df = 11.38, p = 0.002; Fig. 13).

Discussion
The fire ant S. invicta and the mealybug P. solenopsis are two invasive species that have vast populations in South China. H. rosasinensis is popularly cultivated in the parks and greenbelts where S. invicta frequently occurs, therefore it is likely that S. invicta encounters P. solenopsis in the field and that the two species establish a mutualistic relationship. Our study characterized the conditional mutualism between S. invicta and P. Solenopsis to determine whether these interactions may have facilitated the invasion of these two alien species.
Our results demonstrate that S. invicta benefits from the conditional mutualism in our experimental ecosystem. We found that an animal-based food supply could facilitate the growth of S. invicta. Previous studies also demonstrated that animal tissues were essential for ant colony growth [29,30,35]. Ant colonies supplied with live worms and cockroaches as prey and honeydew produced by P. solenopsis exhibited an obviously greater live mass than ant colonies supplied with animal-based food only, which suggests that the honeydew produced by P. solenopsis enhances the growth of S. invicta when insect prey is sufficient. However, the ant colonies lost most of their live mass when they were fed on honeydew only. We also found that access to honeydew strongly promotes worker ant survival. In contrast, the survival of the worker ant barely increased when supplied with animal-based food (Fig. 2). This may be explained by the fact that workers rarely ingest solid food resources [27,36,56]. Consistent with these observations, we also found that a high mealybug density significantly facilitated worker survival (Fig. 3). This result provides further evidence that carbohydrate-rich honeydew plays an important role in worker ant activity [28,29,35].
Such a consumer-resource mutualism not only influences the growth of fire ant colonies but also alters the density of P. solenopsis on H. rosa-sinensis plants. Mealybug survival was greater in treatments where ants were provided with animal-based food than in those without animal-based food (Fig. 1). This result suggests that P. solenopsis colony growth resulting from tending by S. invicta depends on the food supply of S. invicta. One reasonable conclusion based on this finding is that the absence of animalbased food may compel fire ants to exploit honeydew excessively, which may lead to a significant decrease in the fitness of P. solenopsis colonies on host plants. Our hypothesis is supported by a previous report indicating that ant tending has negative effects on developmental rate, growth rate and offspring production of aphid [57]. It seems that the mutualism between fire ants and mealybugs is dependent on the food supply and ant tending level. The variables underlying the interactions between these two invasive species require further study.
Our results demonstrate that tending by S. invicta ants could increase the density of P. solenopsis on H. rosa-sinensis plants by interfering with predation and parasitism by natural enemies (Fig. 5). We found that mealybug nymphs on ant-excluded plants were frequently preyed upon by the lady beetle Menochilus sexmaculata (Coleoptera: Coccinellidae) and infected by two parasitic species, A. bambawalei and Acerophagus coccois Smith. In contrast, the density of mealybug mummies was significantly lower and lady beetles were present less frequently on ant-tended plants, probably because fire ants attack visiting enemies on host plants. Similarly, cotton aphid populations and the predation of sentinel bollworm eggs were greater in the presence of S. invicta than in its absence [58][59][60]. Mealybug population growth was apparently facilitated by Argentine ants in California vineyards [61][62][63]. There were few mealybugs present on H. rosa-sinensis without ant tending and paraffin treatment, whereas more mealybugs were obviously present on ant-tended plants (Fig. 6). The elimination of the paraffins on the plants did not promote the appearance of P. solenopsis, which suggests that ant-tending by S. invicta may have positive effects on the short-range spread of mealybugs. We found that the ants could transport the unserviceable mealybugs (i.e., those that produced little honeydew) to their nests. However, few young mealybug nymphs, such as 1 st and 2 nd instar nymphs, were transferred directly by S. invicta. In fact, young mealybugs crawled more actively on plants than the adults. We observed that 3 rd instar nymphs and adults rarely moved to colonize a favorable location, which indicates that 1 st instar mealybugs have a higher rate of dispersal than other instar nymphs and adults. We inferred that the adults were removed by fire ants indiscriminately, which could increase the risk of spreading P. solenopsis because many unserviceable adults were reproductively active (i.e., contained offspring in their oocysts), even though they did not produce much honeydew. In the course of removal, 1 st instar nymphs were highly capable of invading other plants. It has been shown that foraging ants of Lasius niger commonly removed dead aphids from the colony under laboratory conditions [23]. These results support the conclusion that ants can recognize low-value mealybugs and remove them.
Resource competition has been considered one of the key factors shaping the reduction of native ant diversity by S. invicta [37]. Our result confirms that fire ants exploit resources by disrupting the mutualism between native ants and hemipteran insects. Under laboratory conditions, we have demonstrated that tending by both S. invicta and T. melanocephalum has positive impacts on the fitness of individuals but not small colonies of mealybugs. We did not find a difference between S. invicta and T. melanocephalum in terms of the benefit to mealybugs in the laboratory experiments. Compared with the absence of ant tending, lifespan was significantly extended and the number of offspring was increased when single mealybugs were tended by S. invicta or T. melanocephalum (Fig. 9). Honeydew removal and the stimulation of feeding by ants may be the primary reason for the observed mealybug colony growth. These results are consistent with previous studies indicating that ant tending was beneficial to the fitness of hemipteran colonies [6,9,20,21,64]. However, the benefit to mealybugs in the fire ant inclusion plot was greater than in the exclusion plot (where mealybugs were tended by native ants) indicated that fire ants facilitate mealybugs more than native ants. This may result from better protection provided by S. invicta than by native ants (Fig. 5, 6). Most importantly, our results suggested that foraging activity of workers of T. melanocephalum was more intensive than the activity of S. invicta workers when the two ant species were separated. Workers of T. melanocephalum discovered honeydew and recruited nestmates more quickly than workers of S. invicta. However, foraging intensity of T. melanocephalum was restrained by interference from S. invicta when the two ant species were foraging on one plant. We assigned equal biomass to the two ant species in our experiments. For a given biomass, there were more workers of T. melanocephalum than of S. invicta, which may contribute to the more intense foraging activity of T. melanocephalum in the first two observations. Food preference and olfaction difference are likely other reasons for this result. Workers of T. melanocephalum seemed to be more interested in honeydew than workers of S. invicta. However, with an increasing number of foraging workers of S. invicta, the advantage of greater foraging intensity of T. melanocephalum gradually disappeared (Fig. 13). Workers of T. melanocephalum were not completely driven off the plants by S. invicta in our experiment. Head-on confrontation happened occasionally between the two ant species. Workers of T. melanocephalum initially kept away from S. invicta most of the time. The results indicated that domination of food resources by S. invicta depended on their forceful aggressiveness. Colonies of T. melanocephalum were unlikely to take possession of honeydew when the numbers of foraging workers was nearly equal to that of S. invicta.
In conclusion, the mutualism between S. invicta and P. solenopsis facilitates population increase and fitness of each other. Although native ants may also establish mutualistic relationships with P. solenopsis, S. invicta tends to acquire more honeydew and therefore play a predominant role. Compared with native ants, S. invicta acquires most of the honeydew and protects P. solenopsis more effectively, which may facilitate the invasion of these two alien species in South China. While such mutual invasion success is mainly due to the aggressive behavior of fire ants which have further inhibition effect on native ants. Our results support a facilitative relationship of invasion between two exotic and mutualistic species. In addition, we should pay more attention to the invasion success facilitated by native species on S. invicta and P. solenopsis because their interactions with native species occur extensively when these alien species are introduced independently to a new area. Exploring these interactions with native species will help to further explain the invasion success of S. invicta and P. solenopsis.