How Do Scale Insects Settle into the Nests of Plant-Ants on Macaranga Myrmecophytes ? Dispersal by Wind and Selection by Plant-Ants by

This report elucidates the process of settlement by Coccus scale insects into Crematogaster plant-ant nests formed inside the hollow stems of a myrmecophytic species, Macaranga bancana, in a tropical rain forest. We collected wafting scale insect nymphs from the canopy using sticky traps and characterized the DNA sequence of the trapped nymphs. In addition, we experimentally introduced first-instar nymphs of both symbiotic and nonsymbiotic scale insects to M. bancana seedlings with newly formed plant-ant colonies. Nymphs of symbiotic species were generally carried by ants into their nests within a few minutes of introduction. Most nymphs of nonsymbiotic species were thrown to the ground by ants. Our results suggest that in Crematogaster–Macaranga myrmecophytism, symbiotic coccids disperse by wind onto host plant seedlings at the nymphal stage, and plant-ants actively carry the nymphs landing on seedlings into their nests in discrimination from nonsymbiotic scale insects.


INTRODUCTION
Mutualisms between ants and honeydew-producing hemipterans are sometimes so developed and intensive that both partners have evolved extensive coadaptations (Way 1963;Buckley 1987).Myrmecophilous hemipterans appear to adapt to associations with ants by modifying their behaviors, body structures, and life cycles (Way 1963;Buckley 1987;Gullan 1997;Gullan & Kosztarab 1997).Hemipterans that are frequently associated with ants tend to have poorly developed structures for movement and defense against natural enemies (Way 1963;Buckley 1987).Meanwhile, partner ants sometimes have behaviors adapted to the hemipterans.For example, some ants have been reported to not only tend, but also transport, their hemipteran symbionts (Hölldobler & Wilson 1990).In some species, workers carry symbiotic hemipterans to the appropriate part of the host plants (Way 1963;Maschwitz & Hänel 1985;Hölldobler & Wilson 1990;Gullan 1997).In some other species, foundress queens carry symbiotic hemipterans during the nuptial flight and colony foundation (Hölldobler & Wilson 1990;Klein et al. 1992;Gullan 1997;Gaume et al. 2000;Johnson et al. 2001).These transport behaviors are thought to play an important role in ensuring the foundress queens maintain symbioses with hemipterans that can provide the ants with a nutritious food resource in the form of honeydew.
Ant-hemipteran mutualisms are observed in ant nests in some myrmecophytes, plants that provide their partner ants (plant-ants) with nesting spaces (domatia) (Ward 1991;Gullan 1997).In a few cases of mutualism between hemipterans and "plant-ants, " i.e., in ant species symbiotic with a myrmecophyte, the foundress queens transport the hemipteran symbionts during the nuptial flight or colony foundation (Klein et al. 1992;Gaume et al. 2000).However, how most hemipteran species colonize the nests of their partner plant-ants remains unknown.
In this study, we aimed to elucidate the process of settlement by coccids into plant-ant nests on Macaranga myrmecophytic species.Many species of scale insects in widespread taxa are known to be wind-dispersed as first-instar nymphs (Washburn & Washburn 1984;Gullan & Kosztarab 1997).Therefore, we hypothesized that coccids inhabiting plant-ant nests on Macaranga myrmecophytes might disperse by wind.To test this hypothesis, we used sticky traps in the forest canopy to catch small airborne insects such as dispersing symbiotic coccids of Macaranga myrmecophytes.Next, we experimentally introduced first-instar nymphs of both symbiotic and nonsymbiotic scale insect species onto the surface of Macaranga seedlings and observed the behavior of plant-ant workers in response to each.

Study area and Macaranga myrmecophytes
The study was carried out in Lambir Hills National Park, Sarawak, Malaysia (4°2' N, 113°50' E; 50-200 m above sea level), which is mainly primary lowland mixed dipterocarp forest.The uppermost canopy layer ranges from 30 to 40 m, with emergent trees penetrating the layer to heights of >70 m (Kato et al. 1995).Rainfall and temperature do not follow clear seasonal patterns (Ashton & Hall 1992;Kato et al. 1995;Roubik et al. 2005).
More than 16 species of Macaranga plants, including at least 10 species of myrmecophytes that can reach 20 m in height (Davies et al. 1998;Davies 2001), are found around the study site (Itioka 2005).The stems of most Macaranga myrmecophytes species become hollow when seedlings reach approximately 10-20 cm in height, allowing an alate foundress queen of Crematogaster ant species to settle into the stem and produce workers (Fiala & Maschwitz 1990).Workers then make exit holes on the exterior surface of stems.
Coccids have rarely been observed inside the nests of plant-ant colonies in the early stage of foundation just after the workers begin to emerge on the surface of seedlings (Handa & Itioka, personal observations on >200 seedlings at this stage).First-instar "crawling" nymphs of the superfamily Coccoidea disperse over relatively long distances by wind or walking.Older nymphs and adults, except for alate males, tend to stay at a site or move only short distances (Miller & Kosztarab 1979;Gullan & Kosztarab 1997).

Capture of wafting coccids
In January 2008, we selected nine trees of four Macaranga myrmecophytic species (one M. lamellata, one M. trachyphylla, two M. winkleri, and five M. beccariana), which ranged from 3 to 10 m in height, at the Crane site (4 ha, 200 × 200 m) where a crane provided access to the canopy (Ozanne et al. 2003;Roubik et al. 2005).To capture small wafting insects, we strung three 6 × 25 cm 2 sheets coated on one side with adhesive (Shimada Co., Higashiomi, Japan) from three lateral branches of each tree.We collected all 27 adhesive sheets after 7 days.We collected 27 suspected coccid insects from eight sheets that were hung from one M. trachyphylla, one M. winkleri, and three M. beccariana trees.

DNA sequencing and phylogenetic analyses
We extracted DNA from a whole body of each of the 27 wafting insects collected using the DNeasy Blood & Tissue Kit (Qiagen, Hilden, Germany).A partial cytochrome oxidase I (COI) gene (521 bp) of mitochondrial DNA genome was amplified by polymerase chain reaction (PCR) and sequenced.Methods of PCR, DNA purification, and sequencing followed Ueda et al. (2008).We obtained 16 sequences among 27.In the other 11 samples DNA fragments could not be amplified using PCR.We inferred a molecular phylogeny by adding the sequences to the DNA alignment of COI (Dataset 2) in Ueda et al. (2010), which is thought to cover almost all species of symbiotic coccids associated with Macaranga myrmecophytes.

Introduction of symbiotic and nonsymbiotic scale insects to plant-ants
We haphazardly selected one M. bancana tree that was approximately 10 m high and growing in a forest gap.We cut off and dissected a few lateral branches of the tree and collected crawling coccid nymphs from under sessile mature adult coccids in hollow stems.We also collected crawling nymphs of nonsymbiotic scale insect species from five M. bancana trees that had no plant-ant workers on the surface.Plant-ant workers on these saplings were thought to have disappeared or become inactive due to accidental causes such as damage by falling trees, large litterfalls, and floods.We found three species of nonsymbiotic scale insects (Coccus sp.nr.celatus, Pseudococcus sp. 1, and Iceryini sp. 1) on leaves and twigs of M. bancana trees.We collected an additional scale insect species (Iceryini sp. 2) from the leaves of one mature non-myrmecophyte M. gigantea tree.We observed these four species of scale insects being tended by opportunistic (non-plant-ant) ant species, but they were never found in plant-ant nests.
We collected M. bancana seedlings of 10 to 30 cm height from the forest and planted them in pots in a meshed nursery.We selected 45 vigorous seedlings that had active ant colonies.Plant-ants were seen walking on the plant surface and several exit holes occurred near the ant nest.The plant-ant activities and the density of exit holes were clearly higher around the apical part, compared to the other parts.
We introduced one crawling nymph of either a symbiotic or nonsymbiotic scale insect onto one of the 45 seedlings.For each trial, a crawling nymph was placed on the apical part of the stem surface within 3 cm of an exit hole.We observed the behavior of the nymph and the behavioral responses by plantants.Each seedling was used for one to five introduction trials.
Within 3 min after introduction, most nymphs were contacted by ant workers, and the others entered the nests without any contact with ants.The ants' behaviors after contacting introduced nymphs were categorized into two types: "carrying into nests, " in which plant-ants carried nymphs into nests with their mandibles, and "removal, " in which they dropped nymphs off the plant.
We observed all exit holes of a target seedling for 3 min after a nymph entered or was carried into a nest.In addition, for 10 individuals of symbiotic species and two individuals of nonsymbiotic species, we checked the ground for nymphs for 15 min after the nymphs entered a nest.Overall, we obtained observational data for 32 symbiotic coccid nymphs and 73 nonsymbiotic nymphs.

DNA characterization of wafting scale insects
All of the sequences obtained from the 16 individuals constituted one haplotype.The monophyly of the clade including these obtained sequences and those belonging to L6 from Ueda et al. (2010), which consists of Macaranga-inhabiting Coccus scale insects, was strongly supported by all of the three methods: ML bootstrapping (100%), Bayesian posterior probability (100%), and MP bootstrapping (100%).

Plant-ants' responses to experimentally introduced scale insect nymphs
Of the 32 introduced symbiotic coccid nymphs, 30 were directly contacted by plant-ants.Within 30 s after the first contact of nymphs, plant-ants always grasped nymphs in their mandibles and carried them into hollow stems, except for one case in which a plant-ant dropped the nymph down (Table 1).The other two nymphs walked into nests were not recognized by the plant-ants and entered nests on their own after searching around exit holes.One of these was observed being grasped by an ant just after it entered an exit hole.The proportion of symbiotic nymphs taken into nests was significantly greater than nymphs removed from plants (Table 1; binominal test, p < 0.001).Regardless of the mode of entry, all symbiotic nymphs stayed inside for at least 3 min and none of the 10 target nymphs were found on the ground 15 min later.
Of the 73 nonsymbiotic scale insects introduced, 71 were removed and the other two were carried into nests.For each of the four nonsymbiotic species, the proportion of removed nymphs was significantly higher than that of nymphs carried into nests (Table 1; binominal test, Coccus sp.nr.celatus: p < 0.001; Pseudococcus sp.1: p < 0.001; Iceryini sp.1: p = 0.003; Iceryini sp.2: p < 0.001).Seventy percent of the removed nymphs were dropped off the plant within 10 s after their first contact with plant-ants.The others were carried by plant-ants from 10 s to 7 min before eventually being dropped from the plant.One nymph of Pseudococcus sp. 1 and one nymph of Iceryini sp. 1 were carried into nests and stayed inside for more than 3 min, but were then found on the ground within 15 min.

DISCUSSION
Because no case of foundress queens of plant-ants on Macaranga myrmecophytes carrying their symbiont coccids to new nests had been reported, the symbiont coccids were assumed to disperse independently of their partner plant-ants (Fiala & Maschwitz 1990;Maschwitz et al. 1996).However, the dispersal of the symbiont coccids has not been empirically studied.This study presents some evidence on details of the dispersal of the symbiont coccids.We collected symbiotic coccid nymphs in the canopy dispersing by wind to host plant seedlings, where plant-ants develop new colonies.First-instar Table 1.Behavioral responses of the plant-ants to experimentally introduced first-instar nymphs of symbiotic and nonsymbiotic scale insects on seedlings of Macaranga bancana and the significance probability (p) of binomial tests on the null hypothesis that the two types of behaviors might be randomly adopted.coccid nymphs are difficult to identify to species morphologically.DNA sequencing compared with a molecular phylogeny (Ueda et al. 2010) suggested that many of the trapped coccids were in one of the clades known to be symbionts of plant-ants on Macaranga myrmecophytes.Crawlers of these symbiotic coccids have no physical adaptations for flight, so they are believed to disperse passively by wind.
Our experimental introductions of crawling nymphs suggest that in Crematogaster-Macaranga myrmecophytism, plant-ants actively carry winddispersed nymphs of symbiotic coccids into their nests and clearly discriminate against morphologically similar nonsymbiotic nymphs.Crematogaster plant-ants are known to exclude almost all extraneous materials, including those that would be food for the majority of other ants, by carrying them in their mandibles and throwing them off their host plants (Fiala & Maschwitz 1990;Hashimoto et al. 1997).This habit of plant-ants was confirmed in our experiments; most of the introduced nymphs of non-symbiont scale insects were removed by plant-ants soon after first contact.This indicates that symbiont coccids are among the few organisms not excluded by plant-ants.Symbiont coccid nymphs also seem to be able to enter plant-ant nests by walking into them after alighting on a seedling.In our observations, none of the symbiont coccid nymphs that entered ant nests by either method were excluded by plant-ants, whereas all of the non-symbiont scale insect nymphs were excluded even if they were carried into nests.Thus, plant-ant workers in the Macaranga myrmecophytism appear to actively accept nymphs of only symbiotic coccid species.Some symbiotic scale insects have been suggested to enter the domatia of myrmecophytes without any help from plant-ants (Wheeler 1921;Bequaert 1922;Moog et al. 2005).To the best of our knowledge, no previous report has described plant-ants actively carrying symbiotic hemipterans into their nests.The active transportation of symbiotic coccids by the plant-ants inhabiting Macaranga myrmecophytes suggests they may have a stronger dependence on their symbiotic coccids than in many other ant-hemipteran interactions, except for ones in which foundress queens transport symbiotic scale insects (Klein et al. 1992;Gullan 1997;Gaume et al. 2000).Positive effects of symbiotic coccids on the growth of plant-ant colonies in Macaranga myrmecophytism (Handa & Itioka 2011) may be associated with the plantants' heavy dependence on coccids.
The limited range of symbiotic hemipteran species of Macaranga plant-ants (Heckroth et al. 1998;Ueda et al. 2008Ueda et al. , 2010) ) can no doubt be maintained by such rejecting behaviors as observed in our system.Selection of scale insects by plant-ants may be adjusted in the evolutionary process on the basis not only of food intake from the honeydew or meat of scale insects but also of the total effects, including herbivory damage caused by the scale insects on the host plants.