Oceanic islands represent some of the most isolated habitats on earth and their endemic biotas are characterized by small ranges and the absence of highly co-evolved defensive capabilities, such as anti-predator behaviours and morphologies (Paulay, Reference Paulay1994; Vermeij, Reference Vermeij1999; Roff & Roff, Reference Roff and Roff2003; Fullard et al., Reference Fullard, Ratcliffe and ter Hofstede2007). Consequently, island biotas are exceptionally vulnerable to introduced continental predators (Paulay, Reference Paulay1994; D'Antonio & Dudley, Reference D'Antonio, Dudley, Vitousek, Loope and Adsersen1998). This is perhaps best exemplified by Guam's ‘empty forest’ (Redford, Reference Redford1992) phenomenon, where a single introduced predator, the brown tree snake Boiga irregularis, has severely affected the endemic forest fauna (Savidge, Reference Savidge1987; Wiles et al., Reference Wiles, Bart, Beck and Aguon2003; Mortensen et al., Reference Mortensen, Dupont and Olesen2008), prompting extraordinary conservation interventions (Clark & Savarie, Reference Clark and Savarie2012).
Another alien predator, the carnivorous rosy wolf snail Euglandina rosea, has also had an outsized impact on oceanic island endemic faunas, being implicated in the extinction of at least 134 terrestrial snail species (Régnier et al., Reference Régnier, Fontaine and Bouchet2009). A North American native, E. rosea stalks its gastropod prey by following their mucous trails, consuming small prey whole and larger individuals piecemeal (Gerlach, Reference Gerlach2001; Shaheen et al., Reference Shaheen, Patel, Patel, Moore and Harrington2005; Davis-Berg, Reference Davis-Berg2012). It is highly mobile and can climb trees, a detrimental characteristic for many arboreal Pacific island terrestrial snails (Kinzie, Reference Kinzie1992; Meyer & Cowie, Reference Meyer and Cowie2011). Euglandina rosea was introduced to multiple Pacific island archipelagos in a misguided strategy to control outbreaks of the introduced giant African snail Lissachatina fulica, most notably to the Hawaiian Islands in 1955 (Cowie, Reference Cowie1998) and to the Society Islands in 1974 (Coote, Reference Coote2007). It rapidly extirpated large numbers of endemic land snail species on both archipelagos, including members of the Achatinellidae in Hawaii (Hadfield et al., Reference Hadfield, Miller and Carwile1993) and Partulidae in the Society Islands (Clarke et al., Reference Clarke, Murray and Johnson1984).
The loss of the Partulidae of the Society Islands has been of particular concern because this archipelago is home to c. 50% of partulid species diversity (Cowie, Reference Cowie1992). It has also been the main setting for 20th century partulid studies, starting with the classic work of H.E. Crampton (Reference Crampton1916, Reference Crampton1932) and continuing with decades of research by B. Clarke, J. Murray, M. Johnson and associates (Clarke & Murray, Reference Clarke and Murray1969; Murray & Clarke, Reference Murray and Clarke1980; Johnson et al., Reference Johnson, Murray and Clarke1993). The collapse of Society Islands partulid populations following the introduction of E. rosea prompted the emergency establishment of off-archipelago captive populations for 15 Society Island species (Murray et al., Reference Murray, Murray, Johnson and Clarke1988; Tonge & Bloxam, Reference Tonge and Bloxam1991; Pearce-Kelly et al., Reference Pearce-Kelly, Clarke, Walker and Atkin1997). Until recently, only 5 of 61 endemic Society Islands partulid species were thought to persist in the wild (Coote & Loève, Reference Coote and Loève2003) but subsequent field surveys have found scattered extant populations on Raiatea, Moorea and Tahiti representing four additional species (Lee et al., Reference Lee, Meyer, Burch, Pearce-Kelly and Ó Foighil2008, Reference Lee, Burch, Coote, Pearce-Kelly, Hickman, Meyer and Ó Foighil2009). Seven of these surviving taxa (Partula otaheitana, Partula hyalina, Partula clara, Partula affinis, Samoana attenuata, Samoana burchi, Samoana diaphana) occur on Tahiti, the largest and highest island in the archipelago (Coote, Reference Coote2007; Lee et al., Reference Lee, Burch, Jung, Coote, Pearce-Kelly and Ó Foighil2007a, Reference Lee, Burch, Coote, Pearce-Kelly, Hickman, Meyer and Ó Foighil2009). The predominant Tahitian pattern is one of low elevation extirpation and montane persistence: partulid survivors are most common in cloud forest refuges of >1,000 m altitude (Coote, Reference Coote2007; Lee et al., Reference Lee, Burch, Jung, Coote, Pearce-Kelly and Ó Foighil2007a, Reference Lee, Burch, Coote, Pearce-Kelly, Hickman, Meyer and Ó Foighil2009), where predator activity is probably impaired by cooler temperatures (Gerlach, Reference Gerlach1994, Reference Gerlach2001). However, extensive field surveys beginning in 2004 have located small clusters of low elevation survivors in a number of Tahitian valleys (Coote, Reference Coote2007), currently totalling 38 (Fig. 1; Table 1). Of Tahiti's eight species of Partula, two now dominate low elevation extant populations; i.e. 37 of the 38 valleys with known survivors were exclusively populated by P. clara and/or P. hyalina (Fig. 1; Table 1).
1 Several, ⩾2 remnant populations
2 hy, P. hyalina; otah, P. otaheitana; filo, P. filosa; aff, P. affinis; cl, P. clara; prod, P. producta; nod, P. nodosa
Partula hyalina and P. clara are closely-related species, distinguished by shell coloration, which together represent a discrete Tahitian founder lineage of Moorean origin (Lee et al., Reference Lee, Burch, Coote, Pearce-Kelly, Hickman, Meyer and Ó Foighil2009). Their ability to endure almost 40 years of predation by E. rosea in the valleys of Tahiti is surprising because predation models predict partulid extirpation within 3 years of initial predator contact (Gerlach, Reference Gerlach2001). We are interested in understanding what aspect(s) of their biology underlies this survival, not only for their individual conservation but because of the possible implications for the survival of the many endemic land snails across Oceania now threatened by E. rosea (Régnier et al., Reference Régnier, Fontaine and Bouchet2009).
The inverse relationships of population size (Pimm et al., Reference Pimm, Jones and Diamond1988; Schoener & Spiller, Reference Schoener and Spiller1992) and geographical range (Payne & Finnegan, Reference Payne and Finnegan2007; Cardillo et al., Reference Cardillo, Mace, Gittleman, Jones, Bielby and Purvis2008) to extinction risk have been well documented. Our initial hypothesis is therefore that P. hyalina and P. clara have survived because they were the most abundant and/or widespread species in Tahitian valleys and that they will eventually be driven to extinction by the predator. To test this, ideally we need a detailed census of Tahitian partulid populations contemporaneous with the 1974 introduction of E. rosea. Such a resource is not available but we do have access to a century-old dataset of Tahiti's intact partulid populations. During 1906–1909 Crampton (Reference Crampton1916) surveyed and collected Tahitian valley tree snail populations, publishing a detailed account that has been lauded as ‘among the finest work ever done on the evolution of land snails’ (Gould, Reference Gould1994). We therefore have an extensive demographic profile of intact Tahitian partulid populations with individual valley-level resolution. This allows us to calibrate present-day extant populations with their pre E. Rosea introduction equivalents across the island as a whole, as well as for the 23 valleys Crampton surveyed that retain surviving populations (Table 1).
Crampton (Reference Crampton1916) systematically surveyed the partulid populations of Tahiti. He divided the island of Tahiti into five subunits: Tahiti-Nui, comprising four quadrants (north, south, east and west) and the peninsula Tahiti-Iti (Taiarapu). Over four annual surveys during 1906–1909 he surveyed a large fraction of the coastal valleys present in each geographical subunit: 10 western, 20 southern, 10 northern, 10 eastern and 12 in Tahiti-Iti. Crampton (Reference Crampton1916) did not detail his sampling methodology except to state that he walked into each valley along the primary trails during daylight hours, collecting snails from the adjacent trees and vegetation. He was particularly interested in population-level variation and typically obtained large sample sizes (hundreds) from each valley.
Modern-day surveys of Tahitian valleys for surviving partulids largely follow Crampton's (Reference Crampton1916) methodology, except that the snails are much rarer and that more valleys have been surveyed (Coote, Reference Coote2007). Each survey of a valley is restricted to a single day and involves walking along existing forest trails as deep as possible into the valley, stopping at regular intervals for intense searches of adjacent 5-m2 patches of habitat. Where snails are encountered, all individuals within the immediate patch are enumerated within a 20-minute search period (Coote, Reference Coote2007).
At the end of each day's sampling Crampton (Reference Crampton1916) preserved the snails for later analyses (identification, measurement, dissection) in his Columbia University laboratory. In total, Crampton (Reference Crampton1916) collected 24,085 individuals of seven Tahitian species: P. affinis [as P. otaheitana affinis; reclassified by Kondo (Reference Kondo1980)], P. clara, Partula filosa, P. hyalina, Partula nodosa, P. otaheitana and Partula producta. We extracted his frequency data for each valley surveyed, combining data for subspecies categories into totals for each species. Although the scale of Crampton's (Reference Crampton1916) collecting was extraordinary by today's standards, it is unlikely to have been the main driver in their subsequent extirpation. His sampling was restricted to snails adjacent to the main valley paths and, as late as 1970, the valleys of Tahiti continued to support significant populations of partulids (John B. Burch, pers. comm.).
Partulids are ovoviviparous hermaphrodites and adults typically contain a small number of progeny at different stages of development, giving birth to single young at multi-week intervals (Murray & Clarke, Reference Murray and Clarke1966). Crampton (Reference Crampton1916) dissected the adults he collected, recording the number of eggs, embryos and shelled young present in the reproductive tracts of individual gravid females when collected (i.e. instantaneous clutch size). He presented these data as means per valley population for five species: P. affinis, P. filosa, P. nodosa, P. otaheitana and P. producta. As a result of the relatively low abundance of P. clara and P. hyalina, he calculated their mean instantaneous clutch sizes over multiple valleys, grouped into his five geographical subunits.
We compiled Crampton's (Reference Crampton1916) mean instantaneous clutch size data, combining his subspecies data into single species values for each valley/quadrant. We then calculated estimates of mean clutch sizes across the entire island for each species.
Crampton (Reference Crampton1916) calculated the relative frequency of each species he collected across Tahiti from 1906 to 1909 (Fig. 2a). Of the seven species of Partula he collected on Tahiti, P. otaheitana was the most abundant (a total of 18,955 individuals were collected) and the most widespread, being found in 51 of the 62 valleys surveyed. It was the most numerous partulid species in 48 of the 51 valleys in which it was recorded, usually comprising >90%, of all individuals collected in each valley (Supplementary Fig. S1).
Partula nodosa was the second most abundant Tahitian partulid species collected by Crampton (Reference Crampton1916). The 1,922 specimens he collected (Fig. 2a) had a regional distribution within the island, being restricted to seven western valleys and predominating in three of them (Supplementary Fig. S1). Partula affinis was almost as numerous: 1,560 individuals (Fig. 2a) were collected from 10 valleys distributed in the northern, eastern and southern quadrants, as well as on Tahiti-Iti (Crampton, Reference Crampton1916). In eight of these valleys P. affinis predominated, comprising >80% of all tree snails collected (Supplementary Fig. S1). The two rarest species collected, P. filosa and P. producta (Fig. 2a), were both single-valley endemics. They formed minor components of their respective valley partulid totals (P. filosa 17% and P. producta 6%; Supplementary Fig. S1) and both species are now extinct.
A century ago P. clara and P. hyalina were both widespread in Tahiti, recorded from 43 and 51 of the 62 valleys surveyed, respectively (Crampton, Reference Crampton1916). Although they approached P. otaheitana's extensive range across the island of Tahiti, P. clara and P. hyalina were much rarer; the surveys yielded totals of 819 and 589 individuals, respectively (Fig. 2a). Each of these two species typically composed <5% of the tree snails collected in individual valleys, with their highest incidence being 28% for P. clara and 21% for P. hyalina (Supplementary Fig. S1).
Ongoing field surveys of Tahitian valleys since 2004 have encountered remnant populations of P. clara and/or P. hyalina in 38 valleys (Coote, Reference Coote2007; T. Coote, unpubl. data; Fig. 1; Table 1). We cross-referenced these with the 62 valleys that Crampton surveyed in 1906–1909 and identified 23 valleys containing present-day species that were also collected by Crampton (Table 1). Fig. 2b shows that, a century ago, the relative frequencies of P. clara and P. hyalina among the 23 Tahitian valleys where they survive today were not exceptional, but closely matched their relative frequencies across the island as a whole (Fig. 2a). Partula otaheitana was the most common species and P. clara and/or P. hyalina were minor constituents. This general pattern was maintained at the level of individual valleys, with the exception that P. otaheitana was replaced as the locally dominant species by either P. affinis or P. nodosa in a few valleys (Supplementary Fig. S1). Partula clara survives today in three valleys (Fautaua, Ahonu and Faarapa; Table 1) where it was sufficiently rare a century ago to go undetected by Crampton, despite his intensive collecting (e.g. his Fautaua Valley sample size was 1,084 snails). Given the very low migration rates of partulid tree snails (e.g. 1–10 m per year; Murray & Clarke, Reference Murray and Clarke1984), we consider it likely that these three valley populations of P. clara represent local survivors rather than de novo colonists from other valleys. The formerly locally dominant species in these three valleys, P. otaheitana, has been extirpated, despite having been at least 2–3 orders of magnitude more common than the surviving P. clara.
Fig. 3 is a summary of Crampton's (Reference Crampton1916) mean instantaneous clutch sizes for Tahitian valley partulids. The surviving taxa, P. clara and P. hyalina, had markedly higher clutch sizes than their now-extirpated congeners.
The introduction of E. rosea to Tahiti in 1974 exposed naïve endemic tree snails to an uncontrolled predator–prey experiment in which each valley population represented a discrete iteration. Outcomes in 37 of 38 valleys with known survivors have been strikingly uniform: persistence of two of seven endemic Tahitian species of Partula: P. clara and/or P. hyalina (Fig. 1; Table 1). Our initial hypothesis, that the surviving taxa endured because they were the most abundant and/or widespread species, is clearly refuted by Crampton's (Reference Crampton1916) data. A century ago these two species were relatively rare, typically representing <5% of the original species diversity in most valleys, including those valleys where they still persist (Fig. 2b; Table 1). Although P. hyalina and P. clara were widely distributed, this cannot explain their survival relative to the widespread, co-occurring and much more abundant P. otaheitana, now completely extirpated from the valleys of Tahiti (Coote, Reference Coote2007; Fig. 1).
Molecular phylogenies have shown that P. clara and P. hyalina are two colour morphs of a founding lineage that is distinct from other Tahitian congeners (Lee et al., Reference Lee, Burch, Coote, Fontaine, Gargominy, Pearce-Kelly and Ó Foighil2007b, Reference Lee, Burch, Coote, Pearce-Kelly, Hickman, Meyer and Ó Foighil2009). It is therefore plausible that some shared phylogenetic trait has contributed to their differential survival. One such potential trait is evident in Crampton's (Reference Crampton1916) dataset: P. clara and P. hyalina exhibited similar instantaneous clutch sizes that were markedly higher than those of their extinct congeners (Fig. 3). These data raise an obvious paradox regarding the population structure of tree snails in Tahitian valleys in 1906–1909. If mean clutch sizes in P. clara and P. hyalina were so much higher, why were they so rare relative to three of their congeners, especially P. otaheitana? This discrepancy implies that, a century ago, P. clara and P. hyalina were inferior competitors to their now-extirpated Tahitian valley congeners.
Species with a low intrinsic rate of increase, as a result of factors such as low fecundity, are at increased risk of extinction from stochastic events (Beissinger, Reference Beissinger2000). Island endemics with greater reproductive effort are therefore predicted to have a higher likelihood of surviving the introduction of non-native predators. In the Guam avifauna, for example, species with larger clutch sizes have exhibited better survival (Wiles et al., Reference Wiles, Bart, Beck and Aguon2003). Introduction of the alien predator E. rosea to the Society Islands affected the population dynamics of local partulids by increasing their mortality rates (Clarke et al., Reference Clarke, Murray and Johnson1984). The significantly higher instantaneous clutch sizes of P. clara and P. hyalina relative to their extirpated congeners (Fig. 3) may be a major factor contributing to their continued survival in Tahitian valleys.
However, there are inherent shortcomings in Crampton's (Reference Crampton1916) data that complicate the comparison of reproductive rate among Tahitian partulids. Instantaneous clutch size estimates were compiled from diverse valleys and individual valley-level clutch size estimates are not available for the two surviving taxa, making within-valley comparisons of survivors and non-survivors impossible. In addition, the exact gestation period is unknown for individual Tahitian species. We cannot at present rule out the possibility of longer gestation periods in P. clara and P. hyalina than in the extirpated species, a developmental pattern that could yield higher instantaneous clutch sizes (Fig. 3) but not necessarily higher birth rates.
The impact of introduced predators may vary across island microhabitats; e.g. in addition to larger clutch sizes, survival in the Guam avifauna is associated with the ability to nest in locations inaccessible to the brown tree snake (Wiles et al., Reference Wiles, Bart, Beck and Aguon2003). On Tahiti more low elevation sites with known survivors (Fig. 1; Table 1) contain P. hyalina (31 of 38) than P. clara (20 of 38) and this outcome may reflect microhabitat differences among the two taxa. Crampton (Reference Crampton1916) noted a discrete distributional trait of P. hyalina that distinguished it from its Tahitian congeners. In addition to occurring in dense forest (the typical partulid habitat) he regularly observed P. hyalina at forest edges and in clearings where it was exposed to prolonged direct sunlight. P. hyalina has a distinctively white shell, and a correlation between light shell coloration and an enhanced ability to withstand exposure to direct sunlight is well known among land snail species (Jones, Reference Jones1973, Reference Jones1982; Hazel & Johnson, Reference Hazel and Johnson1990; Ozgo, Reference Ozgo2011). If E. rosea has a lower tolerance of direct sunlight, it is possible that this microhabitat distinction plays an additive role in the survival of P. hyalina.
The persistence of two of seven species of endemic Tahitian Partula under selection pressure from an introduced continental predator has some parallels with the fate of Guam's avifauna (Wiles et al., Reference Wiles, Bart, Beck and Aguon2003). In both cases endemic species with larger clutch sizes exhibited better survival, being able to persist in the presence of the predator for multiple decades (40 years on Tahiti, 60–70 years on Guam). We suspect that this general pattern may also apply to diverse clades of endemic taxa across Oceania. If so, this could guide the prioritization of limited conservation resources for the preservation of Pacific island species that are threatened with extinction. For example, when a novel introduced predator appears on an island, it may be appropriate to give the highest conservation priority to endemic prey species that have lower reproductive potential.
The constructive comments of two anonymous reviewers significantly improved this article. This work was funded by grants from the University of Michigan International Institute Individual Fellowship and the Department of Ecology and Evolutionary Biology to CSB and by National Science Foundation award OCE 0850625 and National Geographic award 9180-12 to DÓF.
C.S. Bick is interested in the ecology and evolution of endemic species and the use of evidence-based conservation to manage the impacts of invasive species on South Pacific oceanic islands effectively. Diarmaid Ó Foighil focuses on molluscan evolution and systematics and has collaborated on a variety of research projects involving terrestrial molluscs. Trevor Coote has been the principal field biologist involved in the conservation of endemic snails in French Polynesia for more than 15 years, working in collaboration with local Polynesians and government agencies as well as biologists interested in conservation on South Pacific islands.