The present checklist of ants (Hymenoptera: Formicidae) of Ambon is the first comprehensive overview of ant species recorded on the island during the last 150 years. The species list is based on literature and museum collections’ records combined with data from our field survey in 2010. In total, there are 74 ant species and subspecies representing 34 genera and six subfamilies known from Ambon. Five of the species found in undisturbed forest were exotic and indicate the overall habitat degradation on the island. The largest proportion of Ambon ant fauna are species with affinities to the Oriental region and species of Oriental-Austro-Melanesian origin. At least 20% of the species are regional endemics. In comparison to other islands in the region, the Ambon fauna seems more diverse and better sampled; however it is clear that a large part of it still remains to be described.

Moluccas, Indo-Australia, Melanesia, Indonesia, Wallacea, species distributions, biogeography, taxonomy, habitat preferences, invasive species, biodiversity

The Indonesian region of Moluccas (Maluku) is one of the most biogeographically and biologically complex areas on Earth. Many of its local invertebrate groups remain undescribed, including ant diversity, which is mostly unknown from this region. Despite being a frequent subject of ecological and taxonomic studies in surrounding regions, there are a limited number of ant species records from Moluccas (e.g. Chapman and Capco 1951, Trainor and Andersen 2010, Andersen et al. 2013).

Ambon is a smaller (775 km²), densely populated island (estimate for 2014 is 369 000 inhabitants; BPS-Statistics, Indonesia 2014) at the north of Banda islands arc, adjacent to the much larger Seram to the north. Ambon consists of two parts, Latimor and Hitu, which are connected by a short isthmus. Both parts are characterized by cordierite-rich dacites, granites, ultramafic rocks, and uplifted Quaternary coral reefs (Pownall et al. 2013). There are also two inactive volcanoes present on Hitu.

Ambon and Seram are located in the eastern Indonesia region, which is a geologically complex zone, influenced by the activity of three major plate regions: the Eurasian, Indian-Australian and Pacific-Philippine Sea plates. Boundaries among them form deformation zones that contain many small plates, crustal fragments or terranes and are characterized by the presence of subduction zones (Hall and Wilson 2000). The Banda Arc consists of an inner volcanic arc and an outer non-volcanic arc of islands formed by sedimentary, metamorphic and igneous rocks of Permian and Quaternary age (Hall 2000). Both Ambon and Seram lay within a belt of subducted Australian crust that is south of a deep subduction trench encircling the Banda Sea. Similar to most of the Banda Sea islands, Ambon and Seram are true oceanic islands, having never been connected to the mainland, although the distances between them and the Asian and Australian continents decreased significantly during periods of glaciation (Voris 2000). It is estimated that Ambon and Seram emerged above the sea about six to five million years ago (Fortuin and De Smet 1991, de Jong 1998).

Due to its position between the Oriental and Australian regions, the Moluccas host an interesting mixture of fauna and flora from both biogeographic origins, in addition to an important proportion of local endemics. Which faunal elements prevail is often taxon and island specific and has been a subject of multiple debates (Gressit 1958, Keast and Miller 1996). The distance between New Guinea and Ambon (and Seram) is shorter (140 km) then between Sulawesi (550 km). However, there are several islands between Ambon and Sulawesi which could serve as stepping stones. We can also expect that the exchange of biotas was mostly unidirectional as Ambon, Seram and other islands are younger than New Guinea and Sulawesi (de Jong 1998).

Based on the distribution of butterfly species, Vane-Wright and Peggie (1994) suggest that fauna of northern Moluccan islands (Halmahera, Ternate, Morotai and others) is biologically related to New Guinea, while the Buru, Ambon, Seram arc is related to North-West Australia and the Sula islands in west Moluccas are part of the ‘Sulawesi region’. A similar pattern was confirmed for the butterfly genus, Hypolycaena by Cassidy and Rawlins (2011), although the authors also found links among Ambon, Seram and Sulawesi as well as Mindanao in the Oriental zone. Close relationship between Moluccan and Sulawesi species was also suggested by phylogenetic reconstruction of the grasshopper genus Chitaura (Acrididae: Oxyinae) (Butlin et al. 1998).

A distributional analysis of hawk moths (Lepidoptera: Sphingidae) revealed that Ambon species are nested within Seram fauna and both are close to Halmahera and Sulawesi. They all are however less similar to south Philippines fauna than to New Guinean fauna (Beck and Kitching 1998). Finally, a comprehensive comparison based on the distribution of over 200 Lepidopteran species (de Jong 1998) supports greater similarity of Moluccan islands to New Guinea than to Sulawesi, suggesting a much higher diversity of New Guinean fauna and shorter distance as the main contributing factors. This pattern is also repeated at the generic level (de Jong 1998). These studies suggest a complex pattern of relationships among Mollucan Islands and the surrounding regions, with the most prominent links to Melanesia or Oriental regions (often through Sulawesi).

Unfortunately, no comparable data exist for Moluccan ant fauna. Although a number of species were described early on from specimens collected by A. R. Wallace (Smith 1859, 1860, 1871) and the regional list was expanded in the following 150 years, a detailed overview of the regional fauna is missing. In this respect, the most relevant studies are by Karavaiev (1925, 1927, 1929, 1930, 1935) and by Chapman and Capco (1951) which provide summaries of ant species from the Oriental and Austro-Melanesian regions.

In recent years, studies focusing on the ants of Timor Island were conducted, contributing important knowledge of ants of the region (Trainor at al. 2010; Andersen et al. 2013). In Timor, the largest proportion of ant species (76%) had South-East Asian (Oriental) affinities, while Australian taxa accounted for 18% of the fauna and the proportion was even smaller for other neighboring islands. In addition, a distributional analysis of ants of genus Polyrhachis confirms that at least some other Moluccan islands have high levels of endemism, even higher than plants and birds (Andersen et al. 2013). However, it is unclear to what extent these patterns are representative of northern Moluccan islands, as they are positioned closer to New Guinea which contributes Austro-Melanesian taxa, as well as to Sulawesi, which could serve as a source of Oriental fauna. Understanding the distribution patterns and origin of the ant fauna of Moluccas is desirable, since ants can provide another line of evidence for reconstructing the biogeographic history of the region. Due to their diverse biological roles, ants are ideal for assessing the effects of life strategy on dispersal and diversification patterns across oceanic islands.

Here, we assemble all available information on ants occurring on Ambon Island and combine it with new data from a recent field survey. The main objectives of our study are 1) to create a checklist which can be used as a reference for follow up studies; 2) to identify new species records collected during our survey of Sirimau forest in Ambon; and 3) to compare habitat preferences of collected ants, analyze biogeographical affinities, and asses the levels of endemism of ant fauna currently known from Ambon.

Based on our literature survey, 48 species have been reported to occur on Ambon. In our field survey, we collected 31 ant species, 16 of which could be identified, while 15 did not match any described taxa on the basis of morphological or molecular characters. Of the 16 described species, only 5 species were previously reported. Our results suggest there are at least 74 ant species from 34 genera and six subfamilies on Ambon Island. The combined checklist of described ant species currently known from the island is 59.

Habitat types differed in species richness: 30 ant species were found in undisturbed forest, while 19 ant species were found in secondary forest. All species occurring in secondary forest were also recorded in undisturbed forest; the overlap as expressed by Jaccard Index was 0.63. Shannon index of diversity for undisturbed forest was 2.96, and 2.6 for secondary forest. The habitats differed significantly in species occurrences, calculated as the proportion of visited traps (Shannon diversity t-test p < 0.01, t = 10.9, d.f. = 1010.5). The accumulation curves suggest that both habitats were sufficiently sampled, as both curves and the statistical indicators (Mao Tau, ICE) reached an asymptote (Figure 1).

Figure 1. 

The cumulative number of species (Sobs) for 450 samples collected at both habitat types: A undisturbed forest B secondary forest. Each curve represents the average of 1000 replicate randomizations of samples order. Richness estimator ICE (mean) and average number of singletons are shown with dashed lines.

The most frequent species in both undisturbed and disturbed forest was Pheidole sp. AMB01 (cf. P. submonticola (Eguchi, 2001) from Borneo). The other most frequent species in undisturbed forest were Pheidole sp. AMB02, Polyrhachis dives (Smith, F., 1857), Oecophylla smaragdina (Fabricius, 1775), Dolichoderus thoracicus (Smith, F., 1860) and Carebara sp. AMB01 (cf. C. diversa (Jerdon, 1851)). In secondary forest plots, the most frequently encountered were Anoplolepis gracilipes (Smith, F., 1857) Lophomyrmex opaciceps (Viehmeyer, 1922), Monomorium pharaonis (Linnaeus, 1758) and Odontoponera denticulata (Smith, F., 1858). Such species composition is typical for secondary and disturbed forest vegetation in the Indo-Australian region.

The comparisons of biogeographic associations of described species suggest that the ant fauna of Ambon is dominated by species with Oriental affinities (13 spp., 22%) and by species distributed equally across Oriental and Austro-Melanesian regions (13 spp., Figure 2). The proportion of species of solely Melanesian origin is slightly smaller (10 spp.) and there are 12 species (20%) which can be considered as local or regional endemics (restricted to Ambon and/or Moluccas). In addition, we detected five species which are most likely introduced on the island. A smaller proportion of species are either widespread (occurring in five or more biogeographic regions) or have more restricted distribution throughout Moluccas (Figure 2).

Figure 2. 

Biogeographic associations of valid ant species occurring on Ambon Island. The distribution categories reflected main areas of occurrence (see text). The abbreviations refer to following biogeographic regions and their combinations: AU – Australia, MEL – Melanesia, MIC – Micronesia, MOL – Moluccas, ORI – Oriental region, PAL – Palearctic region, POLY – Polynesia.

Our study provides the first formal checklist for ant species occurring at Ambon Island. In addition to nearly 50 species reported previously from the literature and databases, our field survey contributed new distributional records for 11 species (marked in Table 1). The additional 15 species which did not match with any currently described taxa are likely to expand the checklist once incorporated into taxonomic revisions of corresponding groups. Our survey also confirms the presence of several species recorded more than 100 years ago and provides evidence of several introduced species in the area.

Despite the addition of 26 species to the checklist, the total number of ant species recorded from Ambon (74) is most likely underestimated. Our survey focused on forested area of only one of the two Ambon islands and in the current checklist some of the genera (e.g. Pheidole, Tetramorium, Camponotus, Nylanderia) which are generally diverse in this region are represented by only few species or completely missing. Our study also did not include sampling of more cryptic leaf litter fauna which would probably contribute substantially to the local species list. For example, we anticipate that species diversity of the genera Pheidole, Nylanderia, Hypoponera and Strumigenys will be higher than reported here. On the other hand, we report significantly higher species richness than the number of species recorded from the southern Moluccan islands of Wetar (3600 km²), Pantar (728 km²) and Lembata (Trainor and Andersen 2010) or the Solomon Islands (e. g. Lembata – 70 ant spp., Sarnat et al. 2013) which are larger than or of similar size to Ambon. We anticipate that the ant species richness of Ambon might be even higher than that of other comparably sized islands, as it is adjacent to the much larger Seram, which could potentially serve as a source area for ant species that would otherwise go extinct on small and disturbed islands.

The accumulation curves indicate that both our sites were sampled exhaustively, within the limits of the methods used. This suggests that our survey provides a representative view of the above ground foraging ant fauna of the area. General collecting in the area outside the transects yielded only one additional species (Anochetus graeffei Mayr, 1870). The fact that all the species from secondary forest were also found in undisturbed forest is somewhat surprising and probably indicates a limited regional species pool and relative similarity of both types of habitats. A thorough sampling of other microhabitats (leaf litter, canopy) could contribute additional species.

Both forests were also inhabited by several invasive species not native to the area, including: Anoplolepis gracilipes, Paratrechina longicornis (Latreille, 1802), Solenopsis geminata (Fabricius, 1804) and Monomorium pharaonis. Although, these species mostly did not reach high frequencies in either type of forest (only A. gracilipes reached a high frequency, being the fourth most frequent species in disturbed forest), their presence is indicative of medium to strong habitat disturbance (Invasive Species Specialist Group, www.issg.org, New Zealand biosecurity system, http://www.biosecurity.govt.nz). The most frequently collected taxa in both habitats were several species of Pheidole and Polyrhachis, as well as Oecophylla smaragdina and Dolichoderus thoracicus.

Although the undisturbed forest site was inhabited by 11 more species than the disturbed site, the high abundances of O. smaragdina, D. thoracicus and moderate abundances of M. pharaonis and Tapinoma melanocephalum (Fabricius, 1793) indicate that this area has likely been disturbed in the past, as these species are often associated with secondary or disturbed forests (Invasive Species Specialist Group database, http://www.issg.org, Way and Khoo 1991, Klimes et al. 2012).

Our field survey and literature search revealed that there are over 70 ant species and subspecies found on Ambon Island, of which at least 59 have been previously described. The total number is likely to increase once data from other parts of the island and from understudied microhabitats such as tree canopy or forest leaf litter become available. The occurrence of several non-native ant species recorded in old preserved forest is of conservational concern and suggests that the introduced species are well established on the island. The overall structure of the ant community recorded in the forested area is indicative of a strong effect of habitat fragmentation and large habitat conversion on the island, where urban habitat is prominent. From a biogeographic perspective, the largest portions of Ambon ant fauna are species with Oriental affinities and species native to the Oriental-Austro-Melanesian region. Interestingly, the portion of Austro-Melanesian species is smaller which contrasts to patterns observed in Moluccan Lepidoptera. Our study provides the first basis for future research on ants of Ambon Island and surrounding areas. As habitat conversion in this region continues, the unique and endemic fauna of Moluccas is particularly endangered and in urgent need of proper conservational efforts. Ants, along with moths and butterflies can serve as useful indicators of habitat quality and uniqueness and for identification of conservational priorities.

We thank the following funders for supporting this work: Marie Currie Fellowship (PIOFGA2009-25448), Czech Science Foundation (P505/12/2467), US National Science Foundation (DEB 0515678). We thank Eric Youngerman for his assistance with specimen sorting and corrections of the text. We are grateful to Campbell O. Webb for facilitating our collaboration and to Yuliza Tafoya Alvarado for assistance with specimen sequencing. We thank an anonymous reviewer for the helpful comments on the earlier version of the manuscript.