Two Cleptoparasitic Ant Crickets (Orthoptera: Myrmecophilidae) that Share Similar Host Ant Species Differentiate Their Habitat Use in Areas of Sympatry in Japan

Habitat segregation may enable potentially competing organisms to coexist in a given region (Schoener, 1974). Habitat segregation has been well-studied in species pairs of various taxa, including herbivorous mammals (e.g. Main & Coblents, 1990; Conradt et al., 1999), birds (e.g. Lynch et al., 1985), and aquatic organisms (e.g. Hearn, 1987; Leibold, 1991). However, such segregation can be interpreted as a result of either interspecific competition (e.g. Chiba, 1996) or differences in habitat preference (e.g. Steen et al., 2014). There are only a few cricket taxa that are myrmecophilous and Myrmecophilus (Orthoptera: Myrmecophilidae) is one of them (Kistner, 1982; Ingrisch, 1995). These ant crickets live in ant nests and exploit debris, ant eggs and larvae, and other food resources in diverse ways (Wheeler, 1900; Wasmann, 1901; Hölldobler, 1947; Henderson & Akre, 1986; Sakai & Terayama, 1995; Akino et al., 1996; Komatsu et al., 2009). Some Myrmecophilus species mimic the ant colony’s chemistry by acquiring cuticular hydrocarbons from the ants via physical contact, Abstract Myrmecophilus crickets (Myrmecophilidae, Orthoptera) are typical ant guests. In Japan, about 10 species are recognized on the basis of morphological and molecular phylogenetic frameworks. We focused on two of these species, M. kinomurai and M. kubotai, and compared their host and habitat use. Previous work based on a limited sampling effort suggested that these two species share some ant species as hosts, but that their habitat preferences (open versus shaded) differ. Here, on the basis of exhaustive sampling across Japan, we confirmed that M. kinomurai and M. kubotai do not differ in their host ant preferences: both prefer formicine ants as hosts. As for habitat preferences, M. kubotai occurred significantly more often in open habitats than in shaded ones (P < 0.05). In contrast, M. kinomurai showed no habitat preference in areas where M. kubotai did not occur. However, M. kinomurai showed an obvious preference for shaded environments in areas of sympatry with its potential competitor M. kubotai. This pattern suggests that interspecific competition between M. kinomurai and M. Kubotai is a factor causing habitat differentiation in areas of sympatry. Sociobiology An international journal on social insects

In Japan, at least 10 species of Myrmecophilus have been recognized on the basis of body surface structures, and each of these species has been collected from nests of specific ant species (Maruyama, 2004).By using molecular phylogenetic methods, Komatsu et al. (2008) detected seven well-supported mtDNA lineages in Japanese Myrmecophilus; these lineages do not completely agree with the morphological taxonomy.Komatsu et al. (2008) were able to group the lineages into at least two categories on the basis of their host specificity: specialists, which are commensally associated with a few ant species, and generalists, which are commensally associated with many ant species or genera.Additionally, each mtDNA lineage appeared to show a habitat use preference (Komatsu et al., 2008).In this study, we investigated host and habitat use by the two most commonly occurring species, M. kinomurai (mtDNA lineage D+G of Komatsu et al., 2008) and M. kubotai (lineage E+F of Komatsu et al., 2008), on the basis of exhaustive sampling across Japan.
Myrmecophilus kubotai is found only in Honshu and Shikoku, whereas M. kinomurai is distributed widely from Hokkaido to Kyushu.Thus, there is partial overlap in their distribution areas (Maruyama, 2006).Komatsu et al. (2008) investigated the host and habitat preferences of these two cricket species by random sampling in Honshu, where they coexist, and reported that both M. kubotai and M. kinomurai most frequently use formicine ants as host, and one particular formicine ant species, Lasius japonicus, was used by both cricket species (Komatsu et al., 2008).Moreover, Komatsu et al. (2008) reported that the collected M. kubotai samples were from open habitats such as grassland or wasteland whereas the collected M. kinomurai samples were from shaded habitats such as forests.These results may indicate that M. kubotai and M. kinomurai are in a competitive relationship for adequate host ant species and that they differentiate their habitat to avoid encountering each other.However, because Komatsu et al. (2008) sampled these two cricket species mainly in Honshu, where they coexist, they could not determine whether their habitat preferences would differ between the areas of coexistence (Honshu and Shikoku) and areas inhabited by only one (M.kinomurai) of the two species (Hokkaido and Kyushu).
In this study, we conducted exhaustive sampling, mainly in Honshu, where these two cricket species coexist, but also in Hokkaido and Kyushu, where M. kubotai does not occur, and examined their habitat use and how it differed in different areas.We then compared host-ant and habitat use between the two cricket species.

Collection of samples
Sampling was conducted from Hokkaido to Kyushu, Japan, in 2004-2008 (Appendix 1).Adult or nymph Myrmecophilus crickets were collected from randomlyselected ant nests.We collected as many crickets as possible by excavating the nest if it was subterranean or by spraying insect rejectant (repellent to keep mosquitoes out) into the nest if it was arboreal.The collected cricket samples were immediately preserved in absolute ethanol.When we collected samples, we recorded the habitat type (open versus shaded) where the host ant nests were located.If there were some sort of masking objects (e.g., trees) around the nest entrance (within 10m in radius), we determined the habitat as shaded.If not so, we determined the habitat as open.
The collected cricket samples were identified by using field-emission scanning microscopy (JEOl, JSM-6390) or mtDNA sequencing (see below).The specimens were digitally micrographed without coating.Voucher specimens are deposited in the Institute of Tropical Agriculture, Kyushu University, Fukuoka, Japan.Some of the samples used in this study were also used by Komatsu et al. (2008Komatsu et al. ( , 2010)).

DNA analysis
We used mitochondrial sequences of collected samples to distinguish the target lineages of M. kinomurai and M. kubotai (see below).Komatsu et al. (2008Komatsu et al. ( , 2010) ) used the cytochrome b (cytb) gene for molecular phylogenetic analysis.However, the primers for that gene region did not successfully amplify mtDNA in many of the M. kinomurai samples.Therefore, we used the 16S ribosomal RNA (16S rRNA) gene, which has a lower substitution rate and has been used in studies of several orthopteran insects (e.g.Allegrucci et al., 2005;Lu & Huang, 2006).
DNA was extracted from the hind legs of the crickets by using a DNeasy Blood & Tissue Kit (QUIAGEN); the other body parts were preserved for morphological identification.A 511-bp fragment of the mitochondrial 16S ribosomal RNA (16S rRNA) gene corresponding to positions 12887-13398 in the Drosophila yakuba mtDNA genome was amplified by polymerase chain reaction (PCR) with the primers 16Sbr (5'-CCG GTC TGA ACT CAG ATC ACG T -3') and 16Sar (5'-CGC CTG TTT AAC AAA AAC AT -3') (Simon et al., 1994) and using the following temperature profile: 35 cycles of 95 °C for 30 s, 55 °C for 30 s, and 72 °C for 90 s.After amplification, the PCR products were purified using a QIAquick PCR purification kit (QUIAGEN).Cycle sequencing reactions were performed with BigDye Terminator Ver 1.1 Cycle Sequencing kit on an ABI 3100 automated sequencer.
We used the obtained 16S rRNA dataset and MEGA4 software (Tamura et al., 2007) to reconstruct the phylogeny by the neighbor-joining (NJ) method.Our previous phylogenetic analysis based on the cytb gene revealed two cryptic lineages in both of the morphospecies M. kinomurai and M. kubotai (lineages D and G in M. kinomurai and lineages E and F in M. kubotai;Komatsu et al., 2008Komatsu et al., , 2010)), and we recognized a similar split lineage in our analysis of the 16S rRNA gene.However, one 16S rRNA lineage of M. kinomurai, corresponding to lineage D of Komatsu et al. (2008), was quite rare and was collected from a fairly limited area.In addition, one 16S rRNA lineage of M. kubotai, corresponding to lineage F of Komatsu et al. (2008), almost exclusively used the myrmicine ant, Tetramorium tsushimae, as host.Therefore, we excluded samples of these lineages (corresponding to lineages D and F of Komatsu et al., 2008) from the analysis.

Statistics
Details of the sampled ant crickets (host ant species, collected habitats, etc.) are given in Appendix 1.We used one-way chi-squared test to compare the frequency of occurrence of M. kinomurai and M. kubotai in formicine ant nests relative to other ant subfamilies.We arbitrarily divided Japan into six occurrence areas: Hokkaido, Northern Honshu (Tohoku), Eastern Honshu (Kanto and Chubu), Western Honshu (Kansai and Chugoku), Shikoku, and Kyushu.
We next sorted the areas into two categories: coexistence areas (Eastern Honshu, Western Honshu and Shikoku) where both M. kinomurai and M. kubotai distributed and noncoexistence areas (Hokkaido, Northern Honshu, and Kyushu) where only M. kinomurai distributed.Within each of the two categorized area, the preference of the two cricket species for open or shaded habitats was determined by using a oneway chi-squared test.For example, the proportion of M. kinomurai individuals collected from open habitat among all collected M. kinomurai individuals (observed proportion in open habitat) in a given area was compared to the proportion of investigated ant nests in open habitats among the total investigated ants nests (expected proportion in open habitat).If the crickets inhabited the open habitat significantly more often than expected, we judged that they "preferred" that habitat.The statistical analyses were performed with the R software package (ver.2.3.1;R Development Core Team, 2005).

Host specificity
We collected a total of 880 Myrmecophilus individuals (Appendix 2) of the five species, M. gigas, M. kinomurai, M. kubotai, M. sapporensis, and M. tertamorii.For M. kinomurai and M. kubotai, most individuals were collected from formicine ant nests (M.kinomurai vs. M. kubotai, %, AM: 0 vs. 0, FO: 93 vs. 88, DO: 0 vs. 0, MY: 5 vs. 8, PO: 0 vs. 0, PR: 0 vs. 0).In addition, one M. kubotai cricket individual was collected from a termite nest, and six individuals (including one each of M. kinomurai and M. kubotai) were collected from outside ant nests (e.g. on the ground) (Appendix 2).Thus, we confirmed the previous finding of Komatsu et al. (2008) that M. kinomurai and M. kubotai preferred Formicine ants as hosts.We detected no significant difference in preference toward Formicinae between M. kinomurai and M. kubotai (Chi-squared test for FO and MY, P = 0.54).Further, the most-preferred host ant species was the same (Lasius japonicus) between these two cricket species (M.kinomurai, 25 of 58 hosts; M. kubotai 24 of 52 hosts).Myrmecophilus kinomurai was collected in all six occurrence areas of Japan, but M. kubotai was collected only in Eastern Honshu, Western Honshu, and Shikoku (not in Hokkaido, Northern Honshu, or Kyushu).

Phylogenetic analysis
The DNA lineages of the M. kinomurai and M. kubotai samples were determined by NJ analysis of a 530-bp sequence of the mitochondrial 16S rRNA gene.Of the 110 analyzed samples, 58 belonged to M. kinomurai (corresponding to lineage G of Komatsu et al., 2008) and 52 belonged to M. kubotai (corresponding to lineage E of Komatsu et al., 2008).

Habitat preferences
In areas of sympatry with the closely related M. kinomurai (i.e.coexistence areas), M. kubotai occurred significantly more often in open habitats, whereas M. kinomurai showed an obvious preference for shaded environments (Table 1 did not occur, M. kinomurai did not show an obvious preference for either habitat type (Table 1, Fig 1).These results suggest that interspecific competition between M. kinomurai and M. kubotai is a factor causing M. kinomurai to shift its habitat use in areas of sympatry.This observed habitat shift of M. kinomurai in areas of sympatry with its potential competitor M. kubotai can be interpreted in two ways.One possible interpretation is that the primary host ant species shows a habitat difference between coexistence areas (e.g.western Honshu) and non-coexistence areas (e.g.Hokkaido).In this study, many M. kinomurai individuals were collected from nests of Lasius japonicus (Appendix 2).The studied nests of L. japonicus were found mainly in open habitats both in coexistence areas (number of nests in open vs. shaded habitat, 99 vs. 35) and non-coexistence areas (99 vs. 10) although the habitat preference M. kinomurai was different between the areas (Chi-squared test, P < 0.01).Therefore, it is unlikely that the observed habitat difference of M. kinomurai between the two types of areas reflected a habitat difference of the primary host ant species between them.
Divergence of habitat use resulting from interspecific competition between sympatric species has been reported before (i.e., a form of ecological character displacement; Chiba, 1996).
Because M. kinomurai and M. kubotai are dominant species and share the same host ant taxon (Formicinae), it is plausible that there is competitive interaction between the two species.Because Myrmecophilus crickets use basically the same resource, the co-occurrence of more than one Myrmecophilus species within the same ant nest may threaten the survival of one the cricket species.Thus, M. kinomurai may shift its habitat by shifting its host ant species from an open habitat-dwelling ant species to a shaded habitat-dwelling species in areas in which it coexists with M. kubotai.
Such avoidance may have another advantage.At least in Japanese Myrmecophilus, the morphological differentiation of genital characters among species is too minor to be used for taxonomic differentiation (Maruyama, 2006), which suggests that different species of Japanese Myrmecophilus can potentially mate with each other.Therefore, differentiation of habitat or host use, or both, may function as a premating isolation mechanism in this genus (i.e., a form of reproductive character displacement).The nature of the ecological or reproductive competition between M. kinomurai and M. kubotai remains unknown.The adverse effects of interactions between these species on their fitness and survival need to be investigated by examining interspecific behavioral interactions and their effect on fertility and survival experimentally, for example, by artificially introducing both species into a single host ant colony.

Fig 1 .
Fig 1. Habitat preferences of M. kinomurai and M. kubotai in coexistence and non-coexistence areas of Japan.

Table 1 .
, Fig1).In contrast, where M. kubotai Habitat specificity of the ant crickets determined by chi-squared test for biased habitat preferences of the ant crickets in those areas where only M. kinomurai occurred and where both crickets were distributed.If the crickets inhabited open (or shaded) habitats in a significantly higher proportion than the expected proportion (i.e., the proportion of that habitat among total available ant nests), they were judged to "prefer" that habitat.See text for details.++, P < 0.01; ns, not significant.O, open habitat; S: shaded habitat.