DOI: 10.13102/sociobiology.v61i3.286-292Sociobiology 61(3): 286-292 (September 2014) Open access journal: http://periodicos.uefs.br/ojs/index.php/sociobiology ISSN: 0361-6525 Mutualistic relationships between the shield ant, Meranoplus bicolor (Guérin–Méneville) (Hymenoptera: Formicidae) and honeydew–producing hemipterans in guava plantation I Burikam, D Kantha Introduction The shield ant, Meranoplus bicolor (Guérin–Méneville) (Hymenoptera: Formicidae), is a common ground nesting spe- cies of the subfamily Myrmicinae, and is widely distributed throughout the entire Oriental Region (Schödlh, 1998). The workers not only forage on dead arthropods as scavengers, but also collect honeydew as carbohydrate source from hemipterans, e.g. the cotton aphid, Aphis gossypii Glover (He- miptera: Aphidae) and the striped mealybug, Ferrisia virgata (Cockerell) (Hemiptera: Pseudococcidae), in agricultural eco- system. However, the trophobiotic relationships or mutualism between M. bicolor and honeydew–producing hemipterans are unknown. Mutualism between ants and honeydew–producing hemipterans has been identified as a continuum of relation- ships ranging from mutualistic to antagonistic (Stadler & Dixon, 2005; Billick et al., 2007), and hemipterans tending by Abstract Mutualistic relationships between the shield ant, Meranoplus bicolor (Guérin– Méneville), and two species of hemipteran, Aphis gossypii Glover and Ferrisia virgata (Cockerell), were investigated in an unsprayed guava plot at Kamphaeng Saen, Nakhon Pathom, Thailand. The reciprocal benefits were observed in both field and laboratory studies. M. bicolor activity coincided with peak seasonal activity of both hemipterans during June–August. We indicated two sets of support evidence in M. bicolor honeydew preference: (i) statistically higher value of adjusted honeydew weight collected by ant workers from A. gossypii compared with that from F. virgata (p–value = .005), and (ii) the higher value of the strength of effect (η2 = .62) in the total variance of multi-species association. The physical property on honeydew viscosity was discussed concerning ant preference. We used two–group, ant–tended and ant–excluded, between–subjects multivariate analysis of variance (MANOVA) in order to show hemipteran benefits. Both hemipteran populations increased in the ant–tended treatment, together with lesser amounts of two species of coccinellids, Menochilus sexmaculatus (Fabricius) and Coccinella transversalis Fabricius, and one species of syrphid fly, Pseudodorus clavatus (Fabricius), compared with the ant exclusion treatment (p–value <.001). The facultative mutualistic relationships of M. bicolor and the two hemipteran species were mentioned. Sociobiology An international journal on social insects Kasetsart University, Nakhon Pathom, Thailand Article History Edited by Jacques H. C. Delabie, UESC, Brazil Received 10 Deceber 2013 Initial acceptance 19 May 2014 Final acceptance 20 July 2014 Keywords cotton aphid, striped mealybug, coc- cinellids, syrphid fly, Psidium guajava Corresponding author Intawat Burikam Department of Entomology Faculty of Agriculture at Kamphaeng Saen, Kasetsart University Nakhon Pathom 73140, Thailand E–Mail: intawat.b@ku.ac.th ants are mostly facultative or opportunistic (Delabie, 2001). Generally, ants benefit from associations with hemipterans by obtaining carbohydrate–rich food source in the form of “honeydew” secreted from hemipterans (e.g.: Nixon, 1951; Way, 1963; Hölldobler & Wilson, 1990). Specifically, the benefits to ants have been focused on the foraging behavior of worker ants (Stadler & Dixon, 2005; Grover et al., 2007; Kay et al., 2010). Some have concentrated on fitness benefits in terms of ant colony growth (Grover et al., 2007; Helms & Vinson, 2008; Wilder et al., 2011). In return the benefits, ants may reduce hemipteran contamination of their waste products, removing dead individuals, protecting natural enemies, and transport hemipterans to new feeding sites, resulting in the abundance of hemipteran populations (e.g.: Way, 1963; Nielsen et al., 2010; Stadler & Dixon, 2005). Ants exploit hemipterans not only for their honeydew, but also as a protein source when foraging on them as a common prey (Buckley, 1987; Hölldobler & Wilson, 1990; Delabie, 2001). However, this RESEARCH ARTICLE - ANTS Sociobiology 61(3): 286-292 (September 2014) 287 type of antagonistic relationships will not be treated here; we are looking at a concrete evidence of mutually benefits among both partners. In this study we verified, in both field and laboratory experiments, the reciprocal benefits of M. bicolor and two species of honeydew–producing hemipterans, A. gossypii and F. virgata. We concentrated for over three–month period in the guava plantation of Horticulture Department, Kasetsart University, Nakhon Pathom, Thailand, observing the mutu- alism of ant–hemipterans including the abundance of natural enemies, mainly predators. We tested three hypotheses: (i) ants receiving benefits in terms of honeydew from mutualistic associations in guava agroecosystem; (ii) ants protecting he- mipterans from natural enemies therefore the densities of natu- ral enemies decrease in the presence of ants; and (iii) in con- sequence of the two hypotheses mentioned earlier, resulting in the increments of hemipteran densities in ant–hemipteran associations compared with the ant–exclusion arrangement. Materials & Methods Study species The study was conducted during April–December 2012 in the unsprayed varietal collection plots (varieties: Phant Si Thong, Kim Ju, and Vhan Pi Roon), consisting of 336 guava trees, Psidium guajava, of Horticulture Depart- ment, Faculty of Agriculture at Kamphaeng Saen, Kasetsart University, Thailand (14.0358 ºN, 99.9826 ºE). The predom- inant ground–nesting ant species in the study area was the na- tive M. bicolor, with only a few colonies of the invasive ant species the tropical fire–ant, Solenopsis geminata (Fabricius) near the perimeter of the plantation. The honeydew–producing hemipterans were A. gossypii and F. virgata. The natural enemies, mainly predators, were two species of coccinellid beetles, Menochilus sexmaculatus (Fabricius) and Coccinella transversalis Fabricius (Coleoptera: Coccinellidae), and one species of syrphid fly, Pseudodorus clavatus (Fabricius) (Diptera: Syrphidae). Ant benefit and honeydew preference The direct benefit of M. bicolor was obtained by weigh- ing a certain number of foraging ants, and then calculating the difference of weight gain between foragers descending and ascending the guava branches. We measured weight gains of M. bicolor after visiting hemipteran colonies as honeydew receiving. We randomly chose foraging ants from the field to weigh for honeydew loading; 50 ant foragers ascending the guava branch before reaching hemipteran colonies, and the other 50 individuals descending the branch with full load of honeydew. Honeydew loads were measured from the weight differences of ants filled with honeydew and ants ascending the guava branch. Individual worker of M. bicolor was cap- tured in an empty hard gelatin capsule (size 0; outer diameter 7.65 mm, height 21.7 mm, and volume 0.68 ml of Torpac Inc., NJ), and shortly after, the capsule containing the arrested ant was weighed on a digital balance. The actual ant weight was obtained from the subtraction of the capsule weight. We weighed, from field collected, two sets (n = 200) foraging ants visiting A. gossypii, and one set (n = 100) of ants visiting F. virgata. The honeydew loads were confirmed with the labo- ratory set up by feeding of M. bicolor workers with honey- dew. A set of field collected workers (n = 100) leaving their nests for foraging were randomly chosen, holding in captivity for 24 hours without food, and subsequently captured inside the gelatin capsule for weighing. After weighing, half of the 24–hour arrested M. bicolor was offered with guava leaves occupied by honeydew exudates of A. gossypii, and the other half of ants with honeydew from F. virgata. The ants were allowed to feed on honeydew until they either refused to feed or left the guava leave. All M. bicolor workers were weighed for the second time in order to obtain honeydew loads before releasing back to their former habitats. Hemipteran benefits We randomly selected 30 guava trees, age 6 years old, approximately 1.65 m in height and 2.5–2.75 m in diameter from the pesticide–free guava plot as our study units. One of two similar branches was randomly chosen from each selected tree to perform ant–exclusion treatment, using sticky barrier around the base of the branch covering 20 cm in length. The target branch was first wrapped around with plastic wrap, and then applied with generic horticultural glue (colorless and odorless). The objective of the gluey barrier is to prevent ants and other crawling insects from reaching hemipteran colonies at the guava shoots, allowing only the entering of air–borne insects, including winged aphids, mealybug crawlers, lady- bugs, and syrphid flies. The barriers were examined periodi- cally, and reapplied the glue as required, in order to maintain the effectiveness as ant barriers throughout the experimental period. The other branch was left unmanipulated as the ant– presence treatment. There was the total of 60 experimental units. This ant–exclusion/presence experiment was started in May, beginning with equal numbers of both A. gossypii and F. virgata between the two treatments on the same guava tree. Insect observations were made during peak seasonal activi- ties of both hemipterans and their natural enemies in June– August 2012. On each chosen guava tree, we randomly selected one terminal shoot from the total of 3–5 shoots of each experi- mental unit, in order to make observations. All terminal shoot belonging to each experimental unit had an equal chance to be picked on each data collection day. The number of he- mipterans: A. gossypii, F. virgata; and larvae of predators: M. sexmaculatus, C. transversalis, and P. clavatus, occupying the branch terminal side of 30 cm in length of both presence I Burikam, D Kantha - Meranoplus bicolor–hemipteran mutualism in guava288 and absence of M. bicolor were counted at various intervals throughout the duration of the experiment from April–De- cember 2012. We counted the insects at interval of 3–5 days, with the total of 7 times per month during June–August, co- incided with the peak activities of both hemipterans and their predators, and every 15 days in other months. However, the observation data or multivariate responses of the five depen- dent variables were derived from the average of 7 times x 3 months = 21 field observations during peak activities of the insects in June–August 2012. We recorded the number of M. bicolor moving up or down (bidirectional) passed a fixed point on the treatment branch with no gluey barrier for 3–min period, to ascertain ant activity throughout the overall exper- imental period from April–December 2012. All observations in the field were done during 08:30–11:30 hours. Statistical analyses To answer the question on the difference of honeydew weights or ant’s honeydew preference between M. bicolor collecting A. gossypii honeydew compared with those of F. virgata, we used analysis of covariance (ANCOVA) of IBM SPSS Statistics (Verma, 2013; Meyer et al., 2013). Body weight of M. bicolor workers with empty stomach (24–hour unfed workers) or weight before receiving honeydew was treated as covariate, and the criterion variable or dependent variable was ant weight after eating honeydew from each he- mipteran species. The analysis of covariance approach was used in order to adjust the initial variations of M. bicolor worker size. The honeydew–producing hemipteran benefits were demonstrated by interference of ants, predominantly M. bicolor, with sticky barrier applying around the base of the main branch in order to exclude the ant. The abundance of hemipterans and natural enemies were compared between presence and absence of M. bicolor. We anticipated more hemipterans and less natural enemies in the ant–attended guava branches. Most studies of ant–hemipteran interactions included either ant or hemipteran removals from the study plants, and made comparisons with the unmanipulated partners. The con- clusions, in general, relied on statistical analysis by the uses of univariate analysis of variance (ANOVA), which concen- trated on one dependent variable, with attempts to make find- ings from multiple analyses of ANOVA (e.g.: Flatt & Weiss- er, 2000; Billick et al., 2007; Daane et al., 2007; Mgocheki & Addison, 2009; Styrsky & Eubanks, 2010). Herein we used multivariate analysis of variance (MANOVA) of IBM SPSS Statistics (Meyer et al., 2013; Rencher & Christensen, 2012), in order to draw one solid conclusion of ant–hemipteran mu- tualism based on the comparison of five dependent variables from two groups, presence and absence of M. bicolor on gua- va branches. These five dependent variables or multivariate responses were number of insects: i.e. nymphs and adults of A. gossypii; nymphs and adults of F. virgata; larvae of M. sexmaculatus; larvae of C. transversalis; and larvae of P. clavatus. All insect counts were transformed into log (y + 1) format; where y = number of insect, in order to agree with statistical assumptions. Several outputs were requested from the MANOVA analysis of IBM SPSS. Box’s Test of Equality of Covariance Matrices expected to see if the dependent variable covariance matrices are equal across the levels of the presence–absence of M. bicolor. Bartlett’s Test of Sphericity was demanded to ascertain sufficient correlation between dependent measures in order to proceed with the analysis. The core MANOVA output was inquired for the multivariate null hypothesis evaluation of no differences between presence and absence of M. bicolor on the composite dependent (number of insects) variate. When the multivariate test is statistically significant, we can pro- ceed with some assessments of each dependent variable. We performed the Tests of Between–Subjects Effects to evaluate the statistical significance of each dependent variable sepa- rately. Bonferroni–corrected alpha level was applied to avoid alpha inflation in order to evaluate these presence and absence of M. bicolor effects. We divided .05 by the number of ANOVAs and obtained .05/5 or a Bonferroni-corrected alpha level of .01. Results and Discussion M. bicolor generally foraged on honeydew of hemipterans as carbohydrate source throughout the year in guava plan- tation at Kamphaeng Saen. Monthly averages (± SE) of M. bicolor activity from April–December 2012 are presented in Fig 1. M. bicolor activity coincided with population fluctua- tions of both hemipterans (A. gossypii, and F. virgata), with peaks seasonal activities in June–August (Fig 1). There were very high correlation coefficients (r’s) between ant activity and either A. gossypii or F. virgata density at r =.97 (p–value < .001; n = 9) and r = .93 (p–value < .001; n = 9), respectively. M. bicolor dominated the other ground–nesting ant species, Solenopsis geminata, in the studied guava plot, al- though S. geminata has been considered as one of the most invasive ant species worldwide (Wetterer, 2011), but not in this guava ecosystem with history of pesticide applications. There were no S. geminata workers observed on the experi- mental guava trees. The tolerance to pesticides of M. bicolor was probably due to the protection of long fine hair covering the entire body (Schödlh, 1998), together with the defensive behavior of Meranoplus by curling up the body and feigned dead when disturbed (Hölldobler, 1988). Ant benefit and honeydew preference In the studied guava plantation, foragers of M. bicolor leaving their nests weighed approximately 2.48 mg (SE = .08; n = 150). After visiting hemipteran colonies, M. bicolor with honeydew loaded, descending the branch back to their nests weighed on average 8.69 mg (SE = 0.1; n = 150). The Sociobiology 61(3): 286-292 (September 2014) 289 honeydew loading is about 6.21 mg (8.69 − 2.48) or roughly estimate around 2.5–fold (6.21 ÷ 2.48) of the mean forager weight departure from their nests. The weighting capacity of M. bicolor workers was reconfirmed in a confined study of laboratory feeding of ant workers to different kinds of honeydew from both hemipteran species. After 24 hours in captivity, M. bicolor workers weighed 3.61 mg (SE = 0.12; n = 100) on average. We selected larger workers with more tolerance and easier for seizing in order to withstand the 24– hour starvation before obtaining honeydew. These workers were fully fed with honeydew from different hemipteran spe- cies, and later weighed approximately 10.70 mg (SE = 0.15; n = 100). The overall expected value of honeydew loading is 7.09 mg (10.70 – 3.61), with an estimate of 2.96–fold (7.09 ÷ 3.61) of the average worker weight after 24 hour in caging. The former 2.5–fold honeydew loading from field foragers was slightly lesser; this was probably due to the offering of honeydew by trophallaxis among workers before returning to their nests (Pfeiffer & Linsenmair, 2007). One–way between–subjects ANCOVA assessing the difference of honeydew loadings from two hemipteran spe- cies of M. bicolor workers showed that the covariate effect or weight of 24–hour captured M. bicolor before honeydew feeding was statistically significant, F (1, 97) = 786.297, p–val- ue < .001. Moreover, a statistically significant effect of honeydew source, from either A. gossypii or F. virgata honeydew, was obtained, F (1, 97) = 8.387, p–value = .005. Mean weight of ants before eating honeydew of F. vir- gata group was higher than that of A. gossypii group, leading to higher full up honeydew loading from F. virgata compared with that from A. gossypii (Fig 2). However, the use of AN- COVA approach removed the covariate effect and unveiled the reversal outcome. Mean weight of M. bicolor plus honey- dew from A. gossypii was significantly higher when corrected for weight prior to receiving honeydew (adjusted mean = 10.85; SE = .072; 95% CI = 10.707–10.992) than mean weight of worker ant with honeydew loaded from F. virgata (adjusted mean = 10.55; SE = .072; 95% CI = 10.408–10.693) (Fig 2). This could indicate that M. bicolor workers prefer honeydew from A. gossypii to that from F. virgata. Ants are expected to concentrate their honeydew col- lection activities on hemipteran species offering higher re- ward in terms of both quantitative and qualitative effects. Hemipteran species that produce larger amount of honeydew, or having honeydew with the presence of preferred sugars or amino acids should be more attractive to certain ant species (Cushman, 1991; Völkl et al., 1999; Yao, 2014). Ant preference for particular sugars in hemipteran honeydew can be species spe- cific (Blüthgen & Fiedler, 2004). Several ant species react strongly to honeydew that holds large amounts of melezitose (Völkl et al., 1999), while others prefer sucrose to melezitose (Blüthgen & Fiedler, 2004). On the other hand, A. gossypii honeydew consisted of mainly sucrose, fructose, and erlose (Lawo et al., 2009), with no appearance of melezitose. Honeydew composition of F. virgata is unknown; however, some studies of mealybugs’ honeydew show composition of fructose, glucose, sucrose, and small amounts of melezitose and raffinose, together with a variety of amino acids (Gray, 1952; Salama & Rizk, 1969). Another difference in honeydew quality beside the com- position of sugars and amino acids is a physical property specific� cally honeydew viscosity. In our study, honeydew excreted by F. virgata was more viscous than that by A. gossypii, which their honeydew seemed to be watery liquid. A study in Ar- gentine ant showed that workers fed eightfold longer on gel sucrose composition, and removed fivefold less sucrose than workers feeding on liquid sucrose (Silverman & Roulston, 2001). The later would agree with lesser amounts of honey- dew loading of M. bicolor from F. virgata than that from A. gossypii in this study. Fig 1. Monthly average of Meranoplus bicolor activity (± SE, vertical line), and hemipteran densities (Ferrisia virgata and Aphis gossypii) from 30 ant–tended guava branches at Kamphaeng Saen, Nakhon Pathom, Thailand in year 2012. Fig 2. Mean weight of Meranoplus bicolor workers (mg) with empty stomach (24–hour without food), mean weight with honeydew loading, and adjusted mean weight from different hemipteran species, Aphis gossypii and Ferrisia virgata. Value on top of column chart indicates data label; different letters followed mean values represent statistically significant differences (p–values ≤ .005). I Burikam, D Kantha - Meranoplus bicolor–hemipteran mutualism in guava290 Hemipteran benefits A two–group between–subjects MANOVA was done on logarithmic transformed data [log (y + 1); y = observation data] of five dependent variables: no. of A. gossypii; no. of F. virgata; no. of M. sexmaculatus larvae; no. of C. transversalis larvae; and no. of P. clavatus larvae. The independent variable or treatment was the presence-absence of ants, particularly M. bicolor, in guava plantation. There were two treatments, i.e. ant–tended and ant–excluded. In general, the ant–excluded treat- ment with sticky barrier was quite effective against M. bicolor, the slow–moving ant species. Even though some ants could accidentally reach the colonies of hemipterans on the exclu- sion treatment from adjacent branches due to the contact with nearby branches via wind blowing, however, these ants could not return back to their nests or could not be able to recruit additional ant foragers. The sample consisted of 60 guava branches divided into equal amounts of presence and absence of M. bicolor. The output of Box’s Test of Equality of Covariance Matri- ces was statistically significant (Box’s M = 69.998; p-value < .001), showing that the dependent variable covariance matri- ces were not equal across the levels of the presence–absence of M. bicolor. Therefore, Pillai’s trace was used to evaluate all multivariate effects (Meyer et al., 2013). Bartlett’s Test of Sphericity was statistically significant (approximate chicsquare = 99.838; pcvalue < .001), indicating sufficient correlation between the dependent variables to proceed with the MANO- VA. Using Pillai’s trace as the criteria, the combined dependent variable was significantly affected by the presencecabsence of M. bicolor, Pillai’s trace = .807, F (5, 54) = 45.244, p-value < .001. There were reliable multivariate differences between ant–tended and ant–excluded treatments on the combined dependent variate. The partial eta squared = .807 (partial η2), equivalent to the full eta squared (η2) in this two–group de- sign (Levine & Hullett, 2002), indicating that we had a very high proportion of the total variance (.807, or about 81%) explained by the activity of M. bicolor. Each dependent measure or each observed insect density was assessed individually in order to determine the strength of the statistically significant multivariate effect. The result of the tests of the univariate effects is shown in Table 1. We had statistically significance univariate effects on all dependent variables (Table 1; p-values < .001). Of all insect species under investigation, A. gossypii provided the highest effect size (η2 = .62), while M. sexmaculatus had highest ef- fect size in terms of natural enemies (η2 = .55) (Table 1). The descriptive information for the univariate analysis is presented in Fig 3; providing each dependent measure’s ob- served means, and total averages obtaining from 30 guava trees in the study. The presence of M. bicolor–tended hemipterans had a considerable impact on insect populations not only hemipteran themselves, but also their natural enemies. On ant–tended treatment, we detected higher densities of both hemipteran species, together with lesser amounts of all natu- ral enemies compared with the ant–excluded treatment (Fig 3). There was more abundant in density of roughly 7.6–fold [{antilog (1.824) – 1} ÷ {antilog (1.002) – 1} or 68.50 ÷ 9.05 = 7.57] of A. gosypii than F. virgata from 30 guava trees in the study (Fig 3). In general, we would say that M. bicolor preferred to associated with A. gossypii more than F. virgata, in this meaning the preference of honeydew collecting, as indicated by higher value of the strength of effect or effect size (Levine & Hullett, 2002; Meyers et al., 2013), i.e. η2 = .62 and η2 = .52, respectively (Table 1). This could be the second evidence in supporting the previous study of honeydew preference of 24– hour captured M. bicolor. Among the three natural enemies or predators, M. sexmaculatus had more strength of effect (η2 = .55), i.e. would be more effective predator, than the other two competitors (η2s = .39, and .32) in this M. bicolor–hemipteran association (Table 1). Even though the surphid fly, P. clavatus, was more abundant than the other two coccinellid predators, but its appearance in the guava plot was restrict to June till August. In addition, there were no P. clavatus larvae found preying on the striped mealybug, F. virgata, in our study. The mutualistic relationships or trophobiotic interac- tions between either A. gossypii or F. virgata with ants have been classified as facultative and very common phenomenon by Delabie (2001). There are two main reasons from this study in supporting the above mentioned: firstly, both he- mipteran species are polyphagous and cosmopolitan species (Blackman & Eastop, 2000; da Silva-Torres et al., 2013), any mutualistic relationship with ants should be opportunistic or facultative rather than obligatory; and secondly, M. bicolor is the most common species of the genus Meranoplus in the Oriental Region (Schödlh, 1998), and is widely distributed as ground nesting species in disturbed habitats of agricultural Fig 3. Effect of Meranoplus bicolor-exclusion on hemipterans (Aphis gossypii and Ferrisia virgata) and their natural ene- mies (Menochilus sexmaculatus, Coccinella transversalis and Pseudodorus clavatus). Different letters on top of dark columns represent statistically significant differences (p–values < .001); numbers on top of clear columns indicate data labels of total average or grand mean from 30 guava trees. Sociobiology 61(3): 286-292 (September 2014) 291 ecosystem, therefore the acquiring for food in the vicinity should be by selection of most abundant resources. This study showed that ant attending had a considerable effect on hemipterous pest densities in guava plantation. There were more individuals of hemipterans in the ant tending gua- va branches, together with lesser amounts of natural enemies mainly predators because of ant guarding activities. In return the benefit, ants received carbohydrate sources in terms of honeydew from both hemipterans. In general, the results of this ant exclusion experiment using gluey barrier are agreed with previous studies done in fruit orchards; e.g. cherry (Stutz & Schmidt-Entling, 2011), apple (Stewart-Jones et al., 2008; Miñarro et al., 2010; Nagy et al., 2013); and in vine- yards (Mgocheki & Addison, 2009). In conclusion, mutualistic relationships between M. bicolor and honeydew–producing hemipterans were revealed. M. bicolor preferably collected honeydew of A. gossypii more than that of F. virgata, because it was not only easier to find, i.e. more abundant, but also more ingestible, due to the physical property of watery liquid. Other alluring properties of honey- dew to ants could be honeydew composition in terms of sugars and amino acids, which needed further investigations. The ant–exclusion promoted an increment in predator densities, and thus leading to a tentatively conservation biological con- trol of hemipterous pests in guava agroecosystem. Acknowledgment We thank Horticulture Department of the Faculty of Agriculture at Kampkaeng Saen, Kasetsart University, par- ticularly Unaroj Boonprakob and Kriengsak Thaipong, for their support in field experiment. Financial support was par- tially offered to junior author by the Graduate School, Kasetsart University. References Billick, I., Hammer, S., Reithel, J.S. & Abbot, P. (2007). Ant– aphid interactions: Are ants friends, enemies, or both? Annals of the Entomological Society of America, 100: 887–892. Blackman, R.L. & Eastop, V.F. (2000). Aphids on the World’s Crops: An identification and formation guide (2nd edition). Chichester: John Wiley and Sons Ltd. Blüthgen, N. & Fiedler, K. (2004). Preferences for sugars and amino acids and their conditionality in a diverse nectar-feeding ant community. Journal of Animal Ecology, 73: 155-166. Buckley, R.C. (1987). Interactions involving plants, Homoptera, and ants. Annual Review of Ecology and Systematics, 18: 111-135. Cushman, J. (1991). Host-plant mediation of insect mutualisms: variable outcomes in herbivore-ant interactions. Oikos 61:138- 144. da Silva-Torres, C.S.A., de Oliveira, M.D. & Torres, J.B. (2013). Host selection and establishment of striped mealybug, Ferrisia virgata, on cotton cultivars. Phytoparasitica 41: 31–40. Daane, K.M., Sime, K.R., Fallon, J. & Cooper, M.L. (2007). Impacts of Argentine ants on mealybugs and their natural ene- mies in California’s coastal vineyards. Ecological Entomology, 32: 583-596. Delabie, J.H.C. (2001). Trophobiosis between Formicidae and Hemiptera (Sternorrhyncha and Auchenorrhyncha): a review. Neotropical Entomology, 30: 501-516. Flatt, T. & Weisser, W.W. (2000). The effects of mutualistic ants on aphid life history traits. Ecology, 81: 3522-3529. Gray, R.A. (1952). Composition of honeydew excreted by pineapple mealybugs. Science, 115: 129-133. Table 1. Tests of univariate effects of Meranoplus bicolor activity on five dependent variablesa: Aphis gossypii; Ferrisia virgata; Menochilus sexmaculatus; Coccinella transversalis; and Pseudodorus clavatus. Data are visual counts of insects occupying 30–cm length of branch terminal (n = 60) Source Dependent var. Type III SS df MS F p–value Eta squared (η2) Ant A. gossypii F. virgata M. sexmaculatus C. transversalis P. clavatus .466 1.681 1.744 .977 .925 1 1 1 1 1 .466 1.681 1.744 .977 .925 96.188 61.859 72.151 27.876 37.525 .000b .000 .000 .000 .000 .624 .516 .554 .325 .393 Error A. gossypii F. virgata M. sexmaculatus C. transversalis P. clavatus .281 1.576 1.402 2.034 1.430 58 58 58 58 58 .005 .027 .024 .035 .025 Corrected total A. gossypii F. virgata M. sexmaculatus C. transversalis P. clavatus .746 3.257 3.146 3.011 2.355 59 59 59 59 59 a The multivariate test of combined dependent measure was statistically significant; Pillai’s trace = .807, F(5, 54) = 45.244, p–value < .001. b .000 indicates < .001. I Burikam, D Kantha - Meranoplus bicolor–hemipteran mutualism in guava292 Grover, C.D., Kay, A.D., Monson, J.A., Marsh, T.C. & Holway, D.A. (2007). Linking nutrition and behavioural dominance: carbohydrate scarcity limits aggression and activity in Argen- tine ants. Proceedings of Biological Sciences, 274: 2951-2957. Helms, K.R. & Vinson, S.B. (2008). Plant resources and colony growth in an invasive ant: the importance of honey- dew-producing hemiptera in carbohydrate transfer across tro- phic levels. Environmental Entomology, 37: 487–493. Hölldobler, B. (1988). Chemical Communication in Meranoplus (Hymenoptera: Formicidae). Psyche, 95: 139-151. Hölldobler, B. & Wilson, E.O. (1990). The Ants. Cambridge, MA: Harvard Univ. Press. Kay, A.D., Zumbusch, T.B., Heinen, J.L., Marsh, T.C. & Holway, D.A. (2010). Nutrition and interference competition have interactive effects on the behavior and performance of Argentine ants. Ecology, 91: 57-64. Lawo, N.C., Wäckers, F.L. & Romeis, J. (2009). Indian Bt cot- ton varieties do not affect the performance of cotton aphids. PLoS ONE 4(3): e4804. DOI:10.1371/journal.pone.0004804 Levine, T.R. & Hullett, C.R. (2002). Eta squared, partial eta squared, and misreporting of effect size in communication research. Human Communication Research, 28: 612-625. Meyers, L.S., Gamst, G. & Guarino, A.J. (2013). Applied Multivariate Research (2nd edition). Los Angeles, CA: SAGE Publications, Inc. Mgocheki, N. & Addison, P. (2009). Interference of ants (Hy- menoptera: Formicidae) with biological control of the vine mealybug Planococcus ficus (Signoret) (Hemiptera: Pseudo- coccidae). Biological Control, 49: 180-185. DOI:10.1016/j. biocontrol.2009.02.001 Miñarro, M., Fernández-Mata, G., Medina, P., 2010. Role of ants in structuring the aphid community on apple. Ecological Entomology, 35, 206-215. Nagy, C., Cross, J.V. & Markó, V. (2013). Sugar feeding of the common black ant, Lasius niger (L.), as a possible indi- rect method for reducing aphid populations on apple by dis- turbing ant-aphid mutualism. Biological Control, 65: 24–36. DOI:10.1016/j.biocontrol.2013.01.005 Nielsen, C., Agrawal, A.A. & Hajek, A.E. (2010). Ants defend aphids against lethal disease. Biological Letters, 6: 205-208. Nixon, G.E.J. (1951). The Association of Ants with Aphids and Coccids. London: Commonwealth Inst. Entomol. Pfeiffer, M. & Linsenmair, K.E. (2007). Trophobiosis in a tropi- cal rainforest on Borneo: Giant ants Camponotus gigas (Hyme- noptera: Formicidae) herd wax cicadas Bythopsyrna circulata (Auchenorrhyncha: Flatidae). Asian Myrmecology, 1: 105–119. Rencher, A.C. & Christensen, W.F. (2012). Methods of Multivari- ate Analysis (3rd edition). Hoboken, NJ: John Wiley & Sons, Inc. Salama, H.S. & Rizk, A.M. (1969). Composition of honey dew in the mealy bug, Saccharicoccus sacchari. Journal of Insect Physiology, 15: 1873-1875. Schödlh, S. (1998). Taxonomic revision of Oriental Meranoplus F. Smith, 1853 (Insecta: Hymenoptera: Formicidae: Myrmicinae). Annalen des Naturhistorischen Museums in Wien 100B: 361-394. Silverman, J. & Roulston, T.H. (2001). Acceptance and in- take of gel and liquid sucrose compositions by the Argentine ant (Hymenoptera: Formicidae). Journal of Economic Ento- mology, 94: 511–515. Stadler, B. & Dixon, A.F.G. (2005). Ecology and evolution of aphid–ant interactions. Annual Review of Ecology and Sys- tematics, 36: 345-372. Stewart-Jones, A., Pope, T.W., Fitzgerald, J.D. & Poppy, G.M. (2008). The effect of ant attendance on the success of rosy apple aphid populations, natural enemy abundance and apple damage in orchards. Agricultural and Forestry Ento- mology, 10: 37–43. DOI:10.1111/j.1461-9563.2007.00353.x Styrsky, J.D. & Eubanks, M.D. (2010). A facultative mutual- ism between aphids and an invasive ant increases plant repro- duction. Ecological Entomology, 35: 1-10. Stutz, S. & Schmidt-Entling, M.H. (2011). Effects of the landscape context on aphid–ant–predator interactions on cherry trees. Biological Control, 57: 37-43. Verma, J.P. (2013). Data Analysis in Management with SPSS Software. New Delhi: Springer India. Völkl, W., Woodring, J., Fischer, M., Lorenz, M.W. & Hoffman, K.H. (1999). Ant-aphid mutualisms: the impact of honeydew production and honeydew sugar composition on ant preferences. Oecologia, 118: 483-491. Way, M.J. (1963). Mutualism between ants and honeydew– producing homoptera. Annual Review of Entomology, 8: 307-344. Wetterer, J.K. (2011). Worldwide spread of the tropical fire ant, Solenopsis geminata (Hymenoptera: Formicidae). Myr- mecological News, 14: 21-35. Wilder, S.M., Holway, D.A., Suarez, A.V. & Eubanks, M.D. (2011). Macronutrient content of plant-based food affects growth of a carnivorous arthropod. Ecology 92: 325–332. Yao, I. (2014). Costs and constraints in aphid–ant mutualism. Ecological Research, 29:383–391. DOI:10.1007/s11284- 014-1151-4