The butterflies of the Torres Strait Islands: A fauna in review

Abstract

Detailed collation of butterfly collecting and observational data from the Torres Strait islands has allowed comprehensive checklists and distribution maps to be constructed, and analyses of these data for the first time. The results obtained have provided a better understanding of biogeographical relationships between the butterfly faunas of these islands with Papua New Guinea (PNG) to the north, and Australia (Cape York, Queensland) to the south. In addition, this information and these analyses can assist whenever subsequent conservation measures may be required. Recorded here, from 38 islands, islets and cays of Torres strait are 227 butterfly taxa. This tropical monsoonal climatic region comprises a shallow, 150km wide seawater gap between mainland Queensland and PNG, with approximately 280 islands, islets and cays in total. Varying in their geology, topography and floral composition, the islands can broadly be grouped into four physiographic types: granitic and basaltic soils, mud and sand. Butterflies were first collected in Torres Strait on Erub Island during 1843-1845, but since then have been collected on all inhabited islands and several uninhabited islands mainly from the 1980s onwards. This study relies, therefore, on the collation of these faunal inventories. Surveys of the poorly explored islands will likely reveal additional species records to further improve our understanding of the biogeography and conservation status of Torres Strait butterflies. Using the butterfly records for the 17 inhabited and 21 uninhabited islands, generalized linear models were used to test whether the number of butterfly species (richness) from each island correlated with the classic island biogeographical predictor variables of surface area of the island (km2 ) and the shortest ocean distances (km) of each island from the two putative butterfly sources, namely Queensland and PNG. Additionally, the effect of human population size (perhaps a measure of disturbance) was tested for each island. Correlation of islands’ butterfly compositions with the above four predictors, including their four physiographic types, were fitted as the response variable in a canonical correspondence analysis. Dissimilarities between the islands and the two mainlands were assessed using Jaccard dissimilarity index. These analyses were used to determine whether a single natural latitudinal zoogeographic line could be designated within Torres Strait that might delineate the butterfly faunas of Queensland and PNG. It was not possible to designate such a line. However, the butterfly faunas of Dauan, Saibai, Erub and Mer Islands suggest a distinctly PNG provenance, whereas the islands in the southwest, Moa and around Waiben, are more Australian. The greatest influence on an island’s butterfly species composition and richness was found to be their physiographic types, especially the islands with granitic and basaltic soils, which are the oldest in origin. Thus, there was a positive linear relationship between species richness and larger granitic, or basaltic islands. Moreover, the predictor variables of island size and distance from the two mainlands showed less obvious correlation. The relationship between human populations on islands and butterfly richness were disparate. Thus, for almost all Torres Strait butterfly species, island 3 physiography provides the best species richness predictor; secondarily an island’s size. The granitic and oldest islands richest in butterfly species are Dauan, with a low dissimilarity to the PNG fauna, and Moa Island and the island group around Waiben Island, which have low dissimilarity to the butterfly fauna of Queensland. The butterfly fauna of the two basaltic islands to the east of the strait, Erub and Mer have a low dissimilarity to the butterfly fauna of PNG. The other islands, mostly spread across the central Torres Strait zone, are typical oceanic coral cays with depauperate butterfly and floral faunas, quite dissimilar to the richer PNG and Queensland faunas. The mainly mud-built islands of Boigu and Saibai are richer in butterfly species than the coral cay islands, Saibai having a low dissimilarity to the PNG fauna. The 91 butterfly taxa recorded from Saibai suggest that the island may be an extension of the adjacent swampy PNG lowlands. The five islands with the highest richness in butterfly taxa were Dauan (144 recorded taxa), Waiben (126), Moa (116), Mer (103) and Erub (95). Islands with the highest numbers of butterfly taxa do not necessarily have the highest floral diversity and it appears that specific plant species’ communities or compositions, rather than floral richness per se, may determine butterfly richness. Percent butterfly vagrancy was estimated for four islands, Dauan (19%), Erub (25%), Mer (22%) and Waiben (22%); where vagrants were considered to be ephemeral species that either did not breed on the island, or perhaps did so for only a limited period prior to their extinction. Vagrant species’ compositions suggested that, within Torres Strait, these arrive from the south and north. Percent vagrancy estimations were unrelated to the island’s distance (i.e. their remoteness) from the nearest land mass. Based on these estimations, it appears that significant butterfly numbers regularly fly out over Torres Strait to arrive on islands, even those considered remote such as Mer Island. Thus, it seems likely that other vagile insects (e.g. pests), may frequently cross Torres Strait, especially from a northern direction during the monsoonal season. The appreciable numbers of vagrant arrivals estimated for Erub and Mer Islands and their surprisingly rich faunas suggest that these are PNG in origin, despite their relatively small size and perceived isolation. The relationship between butterfly species richness and larger granitic, or basaltic islands was generally consistent. However, the relatively small granite islands, Dauan and Waiben, had the two highest species numbers, thus contradicting larger island area as a key predictor of greater species richness. Waiben lies within a group of predominantly larger islands, (9 in total, including Waiben, comprising a total of 292km2 ), with each island separated by between 0.9 and 7.5km of water. Thus, Waiben may be a part of a collectively larger ‘land mass’. Considering butterfly vagility, their estimated vagrancy, and the relative ease of which butterflies can cover distances between these islands, the combined land area of the group does, therefore, fit better with classic island biogeography theory. In contrast, Dauan Island is surrounded by a large expanse of water and is adjacent to PNG’s floristically more depauperate southern coastline. Thus, Dauan Island is essentially isolated from current day humid tropical forests, and it may be one of several high peak 4 relics (including Moa Island) of a much larger area of humid rainforest that likely covered the region prior to the formation of Torres Strait. Future threats likely to impact Torres Strait butterflies include climate change induced sea-level rises and human land use effects, especially burning regimes and weed invasion. The inhabited Torres Strait islands most likely to be impacted by sea-level rises are Masig, Poruma and Warraber (coral cays), Iama (low granite), and Boigu and Saibai (sedimentary mud), where the most climate change susceptible habitats are ocean strands and mangroves. Butterfly taxa that are not recorded from the Australian mainland but are known from Torres Strait number 69. Of these, 17, including an endemic Torres Strait species, Hypolycaena litoralis Lambkin et al., 2005, are limited to island fringing mangrove, lowland areas and swampy habitats. Therefore, the northwestern end of Dauan, and all of Saibai and Boigu Islands, are at the highest risk of losing several of these habitats (and the 17 taxa associated with them) by sea-level rises and anthropogenic influences, especially weed invasion and fire threat. In summary, the results of this study indicated that even islands in Torres Strait that may be considered quite ‘remote’ were likely not far enough from the two mainlands to inhibit immigration of continental butterflies. Moreover, the Torres Strait islands, which seemed to contradict classic island biogeography theory, are likely not a suitable model system for quantitatively predicting island biogeography, due to their relative closeness to the mainlands of Queensland and PNG; rather, geological composition and size of islands could be used in most cases as a crude measure to predict butterfly species composition and species richness.