Diversity of drosophilid flies on Kume-jima, a subtropical island: comparison with diversity on Iriomote-jima

Study areas

Kume-jima is located in a subtropical region (26.2°N, 126.5°E; Fig. 1), and has an area of 59.1 km², approximately one-fifth that of Iriomote-jima (284.4 km²; 24.2°N, 123.8°E). Relative to Iriomote-jima, the area of forest is much smaller on Kume-jima because of agricultural activities: there is 22.5 km² of forest on Kume-jima and 269.0 km² on Iriomote-jima. Forests on Kume-jima are secondary and are dominated by an evergreen broad-leaved tree, Castanopsis sieboldii Nakai. The winter temperature is approximately 1.5°C colder on Kume-jima than on Iriomote-jima, but the average monthly rainfall does not differ much between these two islands (Fig. 2).

Collections

Insects were collected in four seasons, April, July and October in 2004 and February in 2005 using traps and nets. Trap collections were made using “retainer” (Toda 1977) and “open” type traps. The retainer type traps were deployed in five separate environments: forest stream-side, forest floor, forest canopy, open land, and domestic area. The retainer type traps were baited with banana and dry yeast. Two traps were set in each environment for a week in each season. Traps were set at approximately 0.5–1 m above the ground at all sites except the canopy sites, where traps were set at approximately 10 m above the ground. The open type traps were deployed in four separate environments: forest stream-side, forest floor, open land, and domestic area. The open type traps were baited with banana and dry yeast and/or mushrooms (Flammulina velutipes (Curt. ex Fr.) Sing, Grifola frondosa (Dicks. ex Fr.) S. F. Gray and Pleurotus eryngii (DC. ex Fr.) Quél. Flies that were attracted to the open type traps were collected using nets in the morning and evening for 3 days in each season.

Collections were also made by net sweeping on undergrowth plants, tree trunks, fallen fruits, fallen flowers and mushrooms. Flies found in living flowers were collected using aspirators. Naturally occurring flowers and fruits were collected, placed in vials containing vermiculite, and examined for the emergence of adult flies.

Habitat selection

The habitat selection of each drosophilid species was analyzed based on the results of the collections using the retainer type traps in the five different environments. Unweighted pair-group method using arithmetic average (upgma) analysis (Sneath and Sokal 1973) was performed to group drosophilid species by habitat selection. In this analysis, the similarity of habitat selection between two species was evaluated using the niche overlap index calculated according to the relative circular weighting method of Colwell and Futuyma (1971):

where n_ij (or n_i'j) is the number of individuals of species i (or i′) in sample j, N_i is the total number of individuals of species i, N_j is the total number of individuals in sample j, N is the grand total number of individuals. Species with N < 10 were excluded from this analysis.

Species composition

A total of 39 813 individuals representing 16 genera and 37 species (including a female of an undetermined species of the genus Phortica) were collected from Kume-jima in the present survey. The species diversity was much lower on Kume-jima than on Iriomote-jima, where a total of 95 species were recorded by Okada (1965), Hirai et al. (2000), Chen and Toda (2001), Itoh et al. (2003), and Chen and Aotsuka (2003, 2004) (Appendix I).

To compare the faunal characteristics of the two assemblages, drosophilid species recorded on these two islands were classified into subtropical, temperate and cosmopolitan species. Discrimination between temperate and subtropical species was not easy, since some species are sporadically found in areas where they are not able to maintain their populations year-round. In the present study, species that have not been recorded from the Tohoku district of Japan or still more northerly areas (i.e. at latitudes higher than 37°N) are categorized as subtropical species, those that have been recorded from the Tohoku district or still more northerly areas are categorized as temperate species, and those that are found widely in the world are categorized as cosmopolitan species. The Kanto district is located in a temperate region (i.e. at latitudes between 35 and 37°N), and therefore species occurring there could be categorized as temperate species. However, some species that had been distributed only in subtropical areas in the past have recently been found in urban areas in the Kanto district (particularly in Tokyo), probably due to urbanization or global warming. In the present study, therefore, species that are distributed at least in the Tohoku district are categorized as temperate species. This classification is supported by information on the cold tolerance of these species; that is, among drosophilid species examined for cold tolerance, those categorized as temperate species are always more cold-tolerant than those categorized as subtropical species (Kimura 2004).

The drosophilid fauna of Kume-jima comprised 23 subtropical, ten temperate and three cosmopolitan species (Table 1), whereas the fauna of Iriomote-jima comprises 78 subtropical, 12 temperate and five cosmopolitan species. Thus, the diversity of subtropical species was much lower on Kume-jima. Immigration of subtropical species may be more frequent on Iriomote-jima than on Kume-jima because the former is closer to Taiwan, which has a large diversity of subtropical species. Alternatively, the extinction of subtropical species may be more frequent on Kume-jima, where the winter temperature is a little lower (by about 1.5°C) than on Iriomote-jima. In contrast, the number of temperate species differed little between Kume-jima and Iriomote-jima, but the species composition differed to some extent: four species occurred only on Kume-jima and six species occurred only on Iriomote-jima. We cannot explain this difference.

Of the subtropical species, there are seven species that have been collected from the main islands of Japan and Iriomote-jima but were not found on Kume-jima. Among them, six species are only rarely found on Iriomote-jima and/or the main islands (Toda 1984; Hirai et al. 2000; Beppu 2006), suggesting that their niches are small. Thus, they may be unable to maintain their populations on a small island like Kume-jima where their niches could be lost from time to time. The temperate species Liodrosophila aerea is rather common on both Iriomote-jima and the main islands (Toda 1984; Hirai et al. 2000; Beppu 2006) but, interestingly, is not distributed on Kume-jima.

An undescribed species, Drosophila sp. aff. asahinai, has been found on this island and surrounding islands including Okinawa-jima (M. Watada, unpublished data). This species and its sibling species are allopatrically distributed in this archipelago and neighboring regions: D. lacteicornis occurs on Iriomote-jima and neighboring islands; D. sp. aff. asahinai on Kume-jima, Okinawa-jima and neighboring islands; D. asahinai on Amami-oshima and surrounding islands; D. rufa on the main islands of Japan, Korea and neighboring islands; and D. tani on the Chinese continent (Okada 1964, 1965; Cheng & Okada 1985; Hirai et al. 2000; Beppu 2006; M. Watada, unpublished data). In this archipelago, allopatric patterns of species distributions have been identified for terrestrial vertebrates such as amphibians, reptiles and mammals, which have a low capacity to travel across the sea (Ota 1998, 2000; Otsuka & Takahashi 2000). However, drosophilids have a rather high capacity to disperse (Kimura 1992), so it is interesting that this group (i.e. the rufa species complex) has undergone speciation in this archipelago, whereas the other groups of drosophilids have not.

Only one individual of the cosmopolitan species D. simulans has been collected so far on Iriomote-jima (Itoh et al. 2003); however, this same species was rather abundant on Kume-jima, and also on the Bonin Islands, another archipelago located in a subtropical region (Toda 1976; Toda et al. 1987). It is interesting that D. simulans is almost absent from Iriomote-jima.

Feeding habits

Larval feeding habits are presented in Table 1 on the basis of the results from previous studies (Kimura et al. 1977; Nishiharu 1980; Hirai et al. 2000). Fermenting fruit and fungi would be major breeding sites on both Kume-jima and Iriomote-jima. In addition, some species breed on living flowers, fresh figs and herbage (decayed plant parts). The numbers of fruit-feeding, fungus-feeding and herbage-feeding species were compared between Kume-jima and Iriomote-jima. In this analysis, cosmopolitan species and species for which the breeding site was unknown or ambiguous were excluded. Although the breeding sites of some Hirtodrosophila and Mycodrosophila species have not been confirmed, species of these two genera are categorized as fungus-feeders since no exceptions have been observed. In addition, Dichaetophora sp. aff. emeishanensis, cited as an undescribed species of Lordiphosa in Hirai et al. (2000), has unknown feeding habits, but is assumed to be a herbage-feeder on the basis of the feeding habits of its relatives (Kimura et al. 1977; Nishiharu 1980). Figure 3 shows the results of the feeding habit analysis. There were slightly fewer fruit-feeders on Kume-jima than on Iriomote-jima, but much fewer fungus-feeders on Kume-jima. This difference may be attributable to differences in the abundance of resources. Fungi seem to be less abundant on Kume-jima, where forests are smaller and more arid. In fact, fungi were seldom observed during our collections on Kume-jima. On the other hand, fruiting trees and bushes were rather abundant on Kume-jima, even in lightly wooded areas and human-associated environments, which would explain the similar numbers of fruit-feeders. Herbage-feeders were rare on these two subtropical islands, but they are abundantly observed in temperate areas (Kimura et al. 1977; Nishiharu 1980; Toda 1984; Beppu 2006). This is because herbaceous plants are scarce on the forest floor in subtropical areas, where evergreen trees dominate, but abundant in temperate regions, where deciduous trees dominate.

Habitat selection

Species that were abundantly collected in the retainer type traps were analyzed for habitat selection (Table 2). They are classified into several groups by upgma analysis as shown in Figure 4. These species had basically similar habitat selection on Iriomote-jima if they occurred there (Hirai et al. 2000). However, Scaptodrosophila bryani, D. takahashii, D. bipectinata and D. annulipes, which were rare in the forests of Iriomote-jima, were rather abundantly collected in the forests of Kume-jima. This may be due to the forests on Kume-jima occupying only small areas.

Seasonality

Species that were abundantly collected by the retainer type traps were also analyzed for seasonality (Fig. 5). Species were divided into three categories: (i) species that were abundant only in April (D. takahashii); (ii) species that were abundant only in October (Sd. dorsocentralis, D. suzukii, D. bipectinata and D. bizonata), and (iii) species that were abundant for longer periods (the remaining species). The species of the last group were further subdivided into five groups: species that were more abundant in April (Sd. coracina and D. sp. aff. asahinai), species that were more abundant in October (Sd. bryani), species that were more abundant in July (D. albomicans), species that were rare in July (D. melanogaster and D. simulans), and species that were abundant in April and October (D. ficusphila). Scaptodrosophila bryani, D. bizonata and D. albomicans showed similar seasonality on Iriomote-jima (Hirai et al. 2000). Scaptodrosophila dorsocentralis and D. bipectinata showed abrupt expansion of populations on both islands, but their expansion was delayed 2 or 3 months on Iriomote-jima. On the other hand, D. takahashii had very different seasonality on Iriomote-jima. These seasonal patterns should be analyzed with respect to the seasonality of the species' resources and enemies.