Impact statements

Tree snails used to be found throughout the islands of the Pacific Ocean but most were driven to extinction by introduced predators in the late 20th century. The Partula tree snails species of French Polynesia were important in the development of the sciences of genetics and of cultural importance, their shells having been used to make necklaces and ‘crowns’ up until their extinction. They were also ecologically significant, feeding on decaying plant matter and on algae, fungi and bacteria growing on leaves. Since their extinction this vital ecological role has not been filled, and there has been a deterioration in forest health in that time. They were driven to extinction by the introduction of specialist snail predators in a failed attempt at biological control of an agricultural pest species. These predators caused exceptionally high mortality, to which Partula were not adapted. The process of extinction shows how quickly species can be lost and how vulnerable island ecosystems are to removal of such a key component. Some relict populations of Partula have been discovered and studying these has shown that some persist even though they remain vulnerable to predators. Since the extinction of most Partula populations the predators have also declined, and there now exists a new dynamic between predators and prey where the scattered distribution of Partula and their movement enables survival. Knowledge of this is being used to protect the existing populations and to re-establish some species that survive only in the conservation breeding programme. Both of these will contribute to restoring the health of the wider environment. The observation of the extinction process led to the Partula conservation programme. After four decades this now paying off, showing that conservation can work even when the causes of extinction cannot be completely eliminated or reversed.

Lessons learned

• • Historical harvesting of Partula snails for necklace-making represented an unnatural mortality factor but its focus on adults was ecologically less damaging than the indiscriminate mortality that followed the deliberate introduction of predators.

• • Some partulid species survived either at geographical margins (high altitude or on headlands) where the slowing of predator range expansion resulted in lower predation pressure, or due to a combination of relatively high breeding rates and preferences for microhabitats (e.g. the tree canopy) that reduce predation pressure.

• • The history of the relict populations shows that although bottlenecks inevitably reduce genetic variation, population recovery may still be possible if threat factors are reduced.

• • Despite predator resurgences, partulids survive in metapopulations that are resilient to loss of subpopulations.

• • Reintroduction programmes may succeed in creating coexistence between predator and their threatened prey by recreating metapopulation structures.

The biology of extinction

The Partulidae are tree snails found on islands throughout the Pacific Ocean: from the Marianas in the north-west to French Polynesia in the south-east. They comprise 114 species in three (Li et al., Reference Li, Haponski, Bick, Lee, Yeung, Hayes, Li and Ó Foighil2026) to five genera (Slapcinsky and Krauss, Reference Slapcinsky and Krauss2016). The most diverse genera are Partula (80 species) and Samoana (at least 21 species), and geographically the greatest diversity is in French Polynesia (Gerlach, Reference Gerlach2016). There the Society Islands have 51 Partula species and four Samoana, the Marquesas Islands have eight Samoana and the low-lying Austral Islands are occupied by only two Partula species (one introduced). Within French Polynesia all but two Partula species are naturally single island endemics, whereas Samoana has seven single island endemics and five species found on between two and five islands (Gerlach, Reference Gerlach2016).

Within French Polynesia, partulids were historically highly abundant, with densities recorded of up to 20 m⁻² (Johnson et al., Reference Johnson, Murray and Clarke1993). The great majority of partulids are arboreal species, with just four known to have been terrestrial, including two of the Society Island Partula. Samoana are all arboreal and tend to be associated with the canopy, whereas Partula are most abundant 1–3 m above ground (Murray et al., Reference Murray, Johnson and Clarke1982).

The dietary ecology of most Society Islands species has been documented (Gerlach, Reference Gerlach2016): they are mostly generalist consumers of plant epidermal tissue and detritus. The two terrestrial species have both been categorised as generalists. Most arboreal species are also generalists, but this group includes species that specialise on detritus caught in plant axils and on biofilm growing on leaves, especially fungi. Those at mid-heights also include some fungi specialists. In consuming biofilm, partulids were significant components in the ecosystem, removing organisms that grow over plant leaves. In addition, three Society Islands species have been documented to consume small snails as part of their diet (Partula clara, P. hyalina and P. incrassa exigua: Gerlach, Reference Gerlach2016). At least in the case of P. hyalina this is deliberate killing and consumption of prey snails (pers. obs.).

With their generalist diets, most arboreal species are not closely associated with any particular host plant species. Exceptions are Partula mirabilis, P. mooreana and P. tohiveana which are associated with the climbing Freycinetia spp., but these are not exclusive associations. The latter two species are also restricted to higher elevation sites, presumably for climatic reasons. Partula taeniata favours shrubs and P. suturalis favours Hibiscus tiliaceus trees, but these are only tendencies (Murray et al., Reference Murray, Johnson and Clarke1982). Because of this lack of strong association with any particular plant host, and their mostly generalist diets, partulids have been relatively tolerant of habitat invasion by introduced plant species.

Other aspects of their biology do make partulids more vulnerable to extinction. Almost since their first description, partulids have been known to give birth to live young (Férussac, Reference Férussac1821). As a result, their fecundity is low, with the number of births per year varying between different species, from 4 to 41 (Gerlach, Reference Gerlach2016). It has been assumed that this slow reproductive strategy has arisen from a lack of predators on the islands (Cowie, Reference Cowie1992): no natural predation on partulids has been documented. Their low reproductive rates mean that they have a limited ability to recover from significant mortality events (Mace and Balmford, Reference Mace, Balmford, Entwhistle and Dunstone2000; Cardillo, Reference Cardillo2003; Hutchings et al., Reference Hutchings, Myers, García, Lucifora and Kuparinen2012).

A low reproductive rate probably contributes to restricted genetic diversity (Larson, Reference Larson and Çalışkan2012). Measures of heterozygosity are notably low: 0.07–0.18 in wild populations (Johnson et al., Reference Johnson, Clarke and Murray1977, Reference Johnson, Murray and Clarke1986). These values are considerably lower than for the Hawaiian tree snail Achatinella lila (0.53: Price and Hadfield, Reference Price and Hadfield2014; Price et al., Reference Price, Sischo, Pascua and Hadfield2015), the continental Otala lactea (0.44: Elliott and Pierce, Reference Elliott and Pierce1992) and the endangered Japanese Mandarina species (0.33–0.88: Nishida et al., Reference Nishida, Mimura, Mori and Chiba2022). This limited genetic diversity is likely to reduce adaptability and there is an association between low heterozygosity and extinction risk (Spielman et al., Reference Spielman, Brook and Frankham2004).

The process of extinction

Exploitation: 19th and 20th centuries

For Partulidae, a generalist ecology and tolerance of their habitats being invaded by introduced plants mean that population decline leading to extinction is most likely to be driven by direct mortality factors. These may be driven by human exploitation (cultural and scientific collecting) and introduced predators.

There is no evidence of partulid cultural exploitation before the mid-19th century: none have been found in archaeological sites in the Society Islands (Kahn Reference Kahn2018; Kahn et al., Reference Kahn, Nickelsen, Stevenson, Porch, Dotte-Sarout, Christensen, May, Athens and Kirch2014; Christensen et al., Reference Christensen, Kahn and Kirch2018), although in 2009 two Samoana (Traversat et al., Reference Traversat, Conte and Molle2024) were found in a 13th century midden in the Marquesas Islands (identified as S. gonochila from photographs provided by Gabrielle Traversat); outside of French Polynesia, two Partula hyalina shells have been found in a rockshelter in the Cook Islands dating from the 15th to18th centuries (Kirch et al., Reference Kirch, Steadman, Butler, Hather and Weisler1995, Reference Kirch, Gonzalez, Plourde and Kirch2017). All of these are isolated, unworked shells, suggesting merely accidental inclusion in human-associated deposits. The 18th- and 19th-century descriptions, illustrations and collections of Society Islands clothing and ornaments document flower decorations, but the only mollusc items are seashell earrings referred to by Cook (Reference Cook1821) and Banks and Dalton Hooker (Reference Banks and Dalton Hooker1896). One such earring is in the collection of the Cambridge University Museum of Archaeology and Anthropology (item D1914.7, described by Cook [Reference Cook1821]: ‘Their personal ornaments, besides flowers, are few; both sexes wear ear-rings, but they are placed only on one side: when we came they consisted of small pieces of shell, stone, berries, red peas, or some small pearls, three in a string’). In the Marquesas, though Cook (Reference Cook1821) referred to ‘necklaces and amulets made of shells, &c.’. In 1819, the laws directed by the London Missionary Society suppressed practices such as dances and associated ornamentation (Tagupa, Reference Tagupa1974), but by the late 19th century hats were being decorated and ‘crowns’ of Partula shells were used as hatbands to secure hats in the wind and also worn as headdresses (Huguenin, Reference Huguenin1902). The oldest known Partula artefact from the Society Islands is a crown of P. hebe shells in the Muséum Le Havre from 1880, although there are two S. inflata necklaces from the Marquesas Islands from 1845 (Gerlach and Mu-Liepmann, Reference Gerlach and Mu-Liepmann2026). In 1892, shell necklaces were part of the costumes for a major dance festival held on Tahiti (Gerlach and Mu-Liepmann, Reference Gerlach and Mu-Liepmann2026). It is notable that Andrew Garrett, who was resident in the Society Islands for many years, gave very detailed accounts of the snails, their habits and geography in 1884, but makes no mention of their use at that time (Garrett, Reference Garrett1884). This may suggest that any cultural use was not widespread. However, by 1901, Partula necklaces and crowns were commonly worn by women on Raiatea (Huguenin, Reference Huguenin1902) and in 1904 the island was described as being famous for ‘shell-wreaths’ and necklaces, with three species recorded as being used: P. dentifera, P. faba and P. hebe (Wragge, Reference Wragge1906).

A survey of partulid artefacts in the world’s museums (Gerlach and Mu-Liepmann, Reference Gerlach and Mu-Liepmann2026) identified 70 Society Island items registered between 1880 and 1988. These comprise items from every island, and include 11 species. Most have shells of only one species, although six contain 2–3 species, including one which mixes two species from Huahine and Raiatea. Necklaces contain around 250 shells each and crowns 500, all live-collected adults (Huguenin, Reference Huguenin1902; Gerlach and Mu-Liepmann, Reference Gerlach and Mu-Liepmann2026). The scale of cultural exploitation cannot be accurately estimated; it appears to have been focussed on certain species: P. hebe of Raiatea makes up 43% of identified artefacts, and P. rosea and P. varia of Huahine make up 21% and 16%. In contrast, there seems to have been very little exploitation from Moorea, with a single item of P. suturalis out of the island’s five species (Figure 2).

Figure 2.

Numbers of artefacts containing French Polynesian partulid shells of different species from the study by Gerlach and Mu-Liepmann (Reference Gerlach and Mu-Liepmann2026). Original data in Supplementary Material.

The species composition also appears to vary over time (Figure 2): P. hebe is the most consistent, with items collected in every decade from 1880 to 1980, with a peak of seven items registered in 1891–1900. The second most abundant species, P. rosea, first appears in 1908 and P. varia in 1920. This might indicate a movement in the focus of cultural exploitation, from Raiatea to Huahine or it may be an accident of museum collection. Specimen notes support the latter; in the Bernice P. Bishop Museum, Hawai’i, a specimen of P. dentifera from Faarepa, Raiatea (BPBM 87282) was annotated by John W. Moore in 1926: ‘much used by native for making hat bands’, corroborating the observation made by Wragge (Reference Wragge1906). However, only two surviving items using this species are known, one from 1897 (Musée de la Nouvelle-Calédonie: 98–4-15) and one from 1915 (Musée d’Aquitaine). This is a white to pale brown species, and is rarely banded, whereas the museum collections are dominated by necklaces of patterned or brightly coloured species (92%). Preservation of patterned items by collectors would account for the three most colourful species making up 69% of collected items.

By the 1930s, trade seems to have been substantial, at least on Huahine. In 1937, C.M. Cooke noted ‘Part. rosea is collected by the tens of thousands in Huahine and shipped to Tahiti for making “leis”’ (note on a necklace in the Academy of Natural Sciences, Philadelphia). If Cooke’s note is accurate, this could account for the observation that in the 1990s Partula were far less abundant on Huahine than on Tahiti or Moorea and had much less genetic diversity (Goodacre, Reference Goodacre2002). The trade continued for a few years after the extinction of wild partulids, with necklaces being commonly sold until 1997. What were purported to be the last available artefacts were purchased in May 2001 (Coote and Loève, Reference Coote and Loève2003), although one further seashell necklace containing some P. rosea was purchased on Huahine in March 2019 (Trevor Coote pers. comm.).

Cultural exploitation appears only to have been significant relatively recently; it may have had some local impact on populations, but based on the available evidence, it is unquantifiable. Scientific exploitation also began in the 18th and 19th centuries, initially on a small scale, but became significant with the arrival of Henry Crampton in 1906. He collected 400,000 partulids across all islands (Crampton, Reference Crampton1916, Reference Crampton1928, Reference Crampton1932; Gerlach, Reference Gerlach2016). The geographic extent of the populations can only be tenuously determined from his field notebooks and, comparing his records with the situation in 1962, Jim Murray and Bryan Clarke (Reference Murray and Clarke1980) were unable to identify any notable changes in distribution or relative abundance that could not be explained by differences in collecting localities. They collected around 100,000 Partula on Moorea in 1962–1968, removing all snails from their plots, with an estimated efficiency of 80% (Murray and Clarke, Reference Murray and Clarke1984). Their samples covered only a small proportion of the terrain and with total population density estimates of 11,200–15,000 per hectare (Murray et al., Reference Murray, Clark and Johnson1993) the collections would account for only a small proportion of the population. Several other collections were made in the 20th century, significant ones being the Mangareva Expedition of 1934 and collections made by Jack Burch and Yoshio Kondo in 1970–1 (Gerlach, Reference Gerlach2016). Although these number into the thousands of specimens, nowhere do they exceed a hundred specimens from any one valley. Both cultural and scientific collecting may have represented significant mortality factors, but only on a very local geographic scale. In contrast, the predatory rosy wolf snail Euglandina and its mortality effects spread across all areas of the islands (Clarke et al., Reference Clarke, Murray and Johnson1984).

Extinction: 1982 onwards

At least two species were introduced to French Polynesia under the name Euglandina rosea (now known to be a complex of species: Meyer et al., Reference Meyer, Yeung, Slapcinsky and Hayes2017), first to Tahiti in 1974 in an attempt to control the invasive giant African snail Lissachatina fulica. The history of this failed attempt has been well documented elsewhere (Clarke et al., Reference Clarke, Murray and Johnson1984; Murray et al., Reference Murray, Clark and Johnson1993; Gerlach et al., Reference Gerlach, Barker, Bick, Bouchet, Brodie, Christensen, Collins, Coote, Cowie, Fiedler, Griffiths, Florens, Hayes, Kim, Meyer, Meyer, Richling, Slapcinsky, Winsor and Yeung2021). Key aspects are that Euglandina is a genus of specialised snail predators with a high reproductive rate and short generation time. They are voracious predators of most species of snail, hunting by following prey mucus on the ground and to several metres up trees. It has been calculated that one Euglandina will consume some 45–52 Partula-sized snails per year (Gerlach, Reference Gerlach1994), far exceeding the reproductive potential of most Partula. The first introduction to Tahiti was followed by one to Moorea Island in 1977. Neither introduction was monitored and the first surveys of the predator’s distribution and impacts on partulids did not take place until 1980, on Moorea. At this point Euglandina was detected beyond the reintroduction point and a more extensive survey in 1982 found that it had spread widely. The only Partula found in invaded areas were small numbers (unspecified) of P. exigua (Clarke et al., Reference Clarke, Murray and Johnson1984; Murray et al., Reference Murray, Clark and Johnson1993). In 1985 at least 50 P. mooreana were collected, along with some P. mirabilis and P. suturalis. In the next survey in 1986, the only surviving populations were in still uninvaded Haapiti valley in the extreme south of the island, where only two species were present: P. suturalis and P. taeniata; 6 months later these were extinct too (Murray et al., Reference Murray, Murray, Johnson and Clarke1988). On Tahiti P. diminuta, P. nodosa and P. producta were collected in 1984 but in 1986 survivors of four species (P. affinis, P. clara, P. hyalina and P. otaheitana) were found in only two valleys (Murray et al., Reference Murray, Murray, Johnson and Clarke1988). By 1991, all populations appeared to be extinct on both islands (Pearce-Kelly and Clarke, Reference Pearce-Kelly and Clarke1992).

Euglandina introduction to the other islands occurred later. Surveys in 1991 found partulids to be abundant on Huahine in the absence of predators and 246 were collected from two sites for conservation breeding (17 P. arguta, 8 P. rosea and 221 P. varia). By 1994 Euglandina was present and Partula had been almost completely eliminated; the date of complete extinction has been estimated at 1997 (Coote and Loève, Reference Coote and Loève2003). In 1991 on Bora Bora, Euglandina was well established and partulids were only found as empty shells (Pearce-Kelly and Clarke, Reference Pearce-Kelly and Clarke1992). On Raiatea in 1991 Euglandina had largely eliminated partulids from several valleys but some survived in Hotopuu Valley where invasion was incomplete, and some valleys remained predator-free (Pearce-Kelly and Clarke, Reference Pearce-Kelly and Clarke1992). In Hotopuu, 9 P. faba and 1 P. navigatoria were found in the absence of Euglandina, and 9 live P. hebe remained where Euglandina was present, with abundant dead shells. Fewer predator-free valleys remained in 1992, although single live individuals of P. faba still survived in two invaded valleys (Gerlach, Reference Gerlach1994); no other species were located alive. At this time Euglandina occurred at a density of 3,000 per hectare, and can be calculated to have consumed at least 45,450 Partula per year within that area (Gerlach, Reference Gerlach1994). As the maximum pre-invasion Partula density was 36,667 per hectare, it would be expected that all arboreal Partula would be eliminated in less than a year. Terrestrial foraging by Euglandina is less efficient, giving a consumption of 12,840 Partula per year (Gerlach, Reference Gerlach1994), still sufficient to eliminate the rarer ground-living P. navigatoria (4,000 per hectare). As a result of this high predation pressure, all partulids were concluded to be extinct in 1994 (Coote et al., Reference Coote, Loève, J-Y and Clarke1999).

Records of the numbers of live partulids found on Raiatea in 1991–4 are summarised in Figure 3. The surveys in these years involved manual searching for snails from ground level to 4 m above ground over a one-hour period. Search effort varied due to differing numbers of participants (four in 1991 and 1994, three in 1992) but exactly the same areas were visited on both occasions. The comparisons are therefore indicative but not fully comparable. On each occasion, all live and dead Partula located were identified to species and placed into age categories as juvenile (less than full adult size), subadult (full size but lacking an expanded shell lip at the mouth edge) or adult (full size with an expanded lip). The first sites to be invaded had fewest surviving partulids, and extinction was apparent within 2 years of invasion. This is clearest in the P. faba data whereas P. hebe and P. garrettii were only found alive in 1991 in the absence of Euglandina; in 1992 only abundant fresh shells were present. Partula navigatoria shows the same pattern of decline between 1992 and 1994. For all species, both recently killed and live populations had similar age profiles (Figure 3), indicating that predation was not size selective; all Raiatean partulids, except for the largest P. faba (Gerlach, Reference Gerlach2016), fall within the size category of 4.5–25.0 mm shell length range where Euglandina predation is most efficient (Gerlach, Reference Gerlach1994). The only snails to have been found in the presence of Euglandina were adult P. faba.

Figure 3.

Age profile of Partula species in survey areas. a) the most abundant species on Raiatea in 1991–1994, valleys are in the order of invasion (Avera & Hotopuu in 1991; Tevaitoa in 1992; Hamoa & Temehanirai after 1992); b) monitoring data from Moorea (Opunohu and Maramu-Vaianae) and Tahiti (Faarapa).

Persistence and recovery: 1994 to the present

Since 1995, surviving Partula have been located at high altitude on Tahiti (P. otaheitana on seven peaks and P. laevigata on one), with isolated individuals seen on Moorea (P. taeniata on Mont Tohiea) and on Raiatea (P. meyeri on Mont Tefatua in 2006, with a fresh shell in 2024: Ravahere Tapuatarai pers. comm.). Samoana populations are also known in the upper parts of Tahiti and Huahine. These persist despite the occasional presence of small numbers of Euglandina. The south-east coastal area of Tahiti was not invaded by Euglandina until the 2000s and still supports three species at Te Pari.

Since 2004, scattered valley populations have been found: eight species in 35 valleys on Tahiti, and three in 11 valleys on Moorea (Table 1). In 2004–2005 monitoring plots were established in nine of the Tahiti valleys (Faarapa, Haapoponi, Maruapo, Matatia, Tereia, Tiapa, Vaioo, Vaipoe, Vaitoare) and four on Moorea (Haumi, Maatea, Morioahu, Moruu). These 5 × 5 m plots were located within identified partulid populations and varied in number according to the number of subpopulations located (Coote Reference Coote2005–2020). Each plot was searched for partulids and predators for 30 min, monthly or annually depending on accessibility. There are also four sites on Tahiti where periodic surveys provide population trend data (Fautaua, Papehue, Te Pari and Mont Marau), and two on Moorea (Opunohu and Maramu-Vaianae). These sites have not been monitored systematically; surveys comprised teams of people (between two and six) recording all partulids found during a 30-min search.

Table 1.

Wild partulid populations recorded since 2004 showing the most recent record of live animals

Table 1. Long description

On Tahiti 40 valley and 7 mountains have supported partulid populations since 2004 and 10 valleys and 1 mountain on Moorea. 8 species of Partula and 3 species of Samoana have been found on those islands.

Note: Dates in parentheses are for recent shells.

The monitoring data were analysed with a Generalised Linear Mixed Model using Euglandina numbers, Platydemus presence, mean mid-day temperature (°C) and monthly rainfall (mm) as predictors of partulid numbers. These were run separately for Tahiti and Moorea, for each partulid species and in combination. No significant associations were found between partulids and any combination of variables (Table 2). This is to be expected given that the data contain a high proportion of zero values, as many monitoring visits failed to locate any partulids, and predator abundance was not fully quantified. Despite these limitations, examination of the population trends and predator presence shows that there are no consistent patterns in monitored valleys (Figures 4 and 5), either in snails in monitoring plots or elsewhere in the same valleys (‘not monitored’ in Figures 4 and 5). Declines in the monitored populations corresponded to Euglandina resurgence in three sites (Maruapo, Matatia and Vaipoe), although additional Partula populations were found elsewhere in all of these valleys. Two declined with flatworm invasion (Tereia, and temporarily in Faarapa), but in two valleys partulids coexisted with abundant flatworms (Papehue and Vaipoe). In four valleys there were declines in the absence of changes in predator populations (Haumi, Maramu-Vaianae, Morioahu and Vaioo). One site has always been free of predators (Opunohu).

Table 2.

Generalised Linear Mixed Model of partulid numbers

Table 2. Long description

The Generalised Linear Mixed Model incorporating Euglandina and Platydemus presence and rainfall and temperature failed to show any significant association between these variables and numbers of live partulids encountered.

Data from Coote (Reference Coote2005–2020) and Taputuarai et al. (Reference Taputuarai, Lenoble and Laroche2024), available at https://doi.org/10.17863/CAM.133520. Note: Weather data from Meteo France (portail- api.meteofrance.fr ).

Figure 4.

Monitoring data from Tahiti. The bar at the top of each graph shows predator presence: light shading – single live Euglandina found, dark shading – multiple live Euglandina present, hatching – P. manokwari presence. Data from (Coote, Reference Coote2005–2020) and Taputuarai et al. (Reference Taputuarai, Lenoble and Laroche2024), available at https://doi.org/10.17863/CAM.133520.

Figure 5.

Monitoring data from Moorea. The bar at the top of each graph shows predator presence: light shading – single live Euglandina found, dark shading – multiple live Euglandina present, hatching – P. manokwari presence. Data from Coote (Reference Coote2005–2020) and Taputuarai et al. (Reference Taputuarai, Lenoble and Laroche2024), available at https://doi.org/10.17863/CAM.133520.

Age profiles of monitored populations fluctuated markedly during monitoring periods (Figure 3). Except for Maramu-Vaianae, they had lower proportions of adults than was the case historically: 20.0%–68.8% adults compared to 70.7%–98.4% (Crampton, Reference Crampton1932; Murray et al., Reference Murray, Johnson and Clarke1982). This difference may suggest that current populations are in an expansion phase, whereas the historic data come from stable, largely saturated populations. Most extant populations are very small, but three populations were large enough for comparison. On Tahiti, Faarapa monitoring plots were reinvaded by Euglandina in 2009, accompanied by many dead Partula shells and a drop in live adult numbers. This predation was insufficient to eliminate the population, which was ultimately lost to habitat clearance (Coote, Reference Coote2005–2020). On Moorea, only two Euglandina have been seen near the Maramu-Vaianae population, with no evidence of any predation. This Partula population expanded, by 2012 reaching pre-invasion densities and a similar age profile (76.7%–79.1% adult). It then declined, and the adult proportion dropped to 32.0%. In the absence of any dead shells, this change can probably be attributed to dispersal, especially of adults; a population was found some 300 m away in 2026, which may be the result of this dispersal. Opunohu mangroves on Moorea have not been invaded by predators, but the Partula population has fluctuated in association with occurrences of habitat disruption. There, the adult proportion has also fluctuated, but remains relatively low: 20.0%–66.7%.

Synthesis

Partulid populations have been assumed to be exceptionally vulnerable to predators having evolved in a predator-free environment (Groombridge, Reference Groombridge1992; Gerlach, Reference Gerlach1994; Bick et al., Reference Bick, Pearce-Kelly, Coote and Foighil2018), although the presence of extinct predatory birds has been hypothesised to explain the origin of shell polymorphisms (Johnson et al., Reference Johnson, Murray and Clarke1993). Their vulnerability arises from having low reproductive rates and lacking anti-predator defences. Partulids faced other significant mortality factors in the past: the pollen and sedimentation record of Lake Temae, Moorea suggests that slash-and-burn agriculture was extensive in valleys and had destabilised hillsides around 950 BP (Stevenson et al., Reference Stevenson, Benson and Kirch2017). Partula-specific mortality factors were introduced with the advent of cultural exploitation in the 19th century. This may have reduced local populations; it has been speculated that similar exploitation of Hawaiian tree snails had population-level impacts (Hadfield, Reference Hadfield1986; Cowie, Reference Cowie, DellaSala and Goldstein2022; Cowie et al., Reference Cowie, Bouchet and Fontaine2026), but it is impossible to assess the scale or impact of this exploitation in French Polynesia. However, the evidence suggests that Partula were able to withstand the removal of large numbers of adults from the population.

The introduction of Euglandina in 1974 represented a qualitatively different mortality factor, acting on both adults and juveniles. Even if Partula had faced bird predation in the past, Euglandina represented a different scale of pressure, occurring at extremely high densities and consuming all age categories. During the Euglandina invasion, all Partula species were vulnerable, but the speed of extinction varied. Small numbers of P. exigua and P. faba persisted for a few months after the extinction of all other species. These two differed markedly, the former being a small, partially carnivorous species high on vegetation (Clarke and Murray, Reference Clarke and Murray1969) whereas the latter was the largest Society Island partulid. Partula exigua may have survived briefly due to reduced predation on higher vegetation whereas isolated individuals of P. faba may have survived due to their relatively large size or by chance; this having been the most abundant Raiatean species (Gerlach, Reference Gerlach2016).

Euglandina expansion was not uniform and it was predicted that it would fail to colonise high-elevation sites (Gerlach, Reference Gerlach1994). The putative altitude ceiling at 1,000 m has proven to be incorrect, but partulid populations do persist above this altitude despite occasional Euglandina incursion. Monitoring data from Mont Marau (Coote, Reference Coote2005–2020) demonstrate that Partula otaheitana survives these periodic small-scale invasions. Data from Mont Aorai are more limited but partulids continue to persist there too. On Raiatea, P. meyeri’s refuge on Mont Tefatua was assumed to be predator-free (Coote et al., Reference Coote, Loève, J-Y and Clarke1999), however the collections made by J.W. Moore on the summit in 1927 included species which are now absent and it is probable that Euglandina invaded the summit temporarily, leaving a relict population of P. meyeri (Gerlach, Reference Gerlach2016). A similar situation is found in coastal Te Pari where Euglandina did not arrive until 2001 (Coote, Reference Coote2007) but it remains extremely scarce and Partula survive. At high elevations and on Tahiti’s south-eastern peninsula it is likely that the slow invasion by a diminishing wave of predators never caused the mass mortality seen in other areas. A similar pattern of slowing invasion has been reported on Chichijima Island in the Ogasawara group (Chiba and Cowie, Reference Chiba and Cowie2016).

Samoana species are more resistant to Euglandina presence than Partula and survivors were found as early as 1995 (Coote et al., Reference Coote, Loève, J-Y and Clarke1999). These have always been relatively scarce: Crampton collected only eight individuals from high elevations (Crampton and Cooke, Reference Crampton and Cooke1953). Clarke and Murray (Reference Clarke and Murray1969) noted that Samoana and P. exigua ‘share a tendency to rest higher than taeniata on the branches and leaves of shrubs and trees’, echoing Garrett (Reference Garrett1884) on S. attenuata: ‘Owing to its peculiar habit of living on the foliage near the top of trees, it easily escapes observation’. This habitat preference may have reduced predation pressure on P. exigua, allowing it to persist for a further year, whilst predation on the Samoana species was further reduced by their more sparse distribution. Today the Samoana species are increasing in abundance on Tahiti and Moorea, even in valleys and low on vegetation (Gerlach, Reference Gerlach, Garcia, Chmara and Close2023a). This change in habits could be a result of competitive release following the decline of Partula species, or of some undetected environmental change.

Partula clara, P. hyalina, P. incrassa and P. taeniata have also been found to persist in some valleys (Coote et al., Reference Coote, Loève, J-Y and Clarke1999; Coote, Reference Coote2007; Coote and Loève, Reference Coote and Loève2003). Bick et al. (Reference Bick, Lee, Coote, Haponski, Blaauw and Foighil2021) proposed that P. hyalina survives due to its association with ‘solar refuges’: edge habitat with high solar exposure, avoided by Euglandina. Whilst this may be a contributing factor, it does not apply to the other species and not all P. hyalina are found in such sites. It has also been hypothesised that the surviving species tolerate predation due to having relatively high reproductive rates (Bick et al., Reference Bick, Ó Foighil and Coote2016, Reference Bick, Pearce-Kelly, Coote and Foighil2018). There is some support for this from captive populations where the species that coexist with Euglandina have annual birth rates of 22.6 (P. taeniata), 40.1 (P. clara) and 45.5 (P. hyalina), compared to 10.0–24.9 for the extirpated species (Gerlach, Reference Gerlach2016). Nevertheless, annual birth rates for species with populations surviving occasional invasion are relatively low: 16.4 for P. affinis and 9.0 for P. otaheitana, supporting the idea that these populations survive due to low predation pressure, rather than their reproductive rate. High fecundity has not been sufficient for survival in all cases; even though P. suturalis and P. garrettii have high values (24.8–24.9) they did not persist in the wild following predation by Euglandina. The high birth rate of P. suturalis may be offset by a longer age to maturity (545 days to first birth, compared to 298 for P. taeniata: Clarke and Murray, Reference Clarke and Murray1969). Data for P. garrettii are limited but suggest maturity at about 298 days (Dave Clarke pers. comm.). The failure of this species to survive Euglandina invasion may be due to increased vulnerability, having occurred near the base of trees (Garrett, Reference Garrett1884).

Surviving Partula have fluctuated or declined within monitoring plots; however, additional populations have appeared elsewhere in the same valleys. This is therefore not a simple pattern of decline. In Fautaua Valley, the few partulids that survived the arrival of Euglandina disappeared following a substantial predator resurgence in 2008 through to 2021. However, several partulid populations have reappeared in various parts of the valley. A similar pattern of local populations appearing and disappearing is seen in Te Pari, despite the absence of significant numbers of Euglandina. There P. hyalina has a clumped distribution: several individuals can be found within areas of approximately 20 m², interspersed with apparently suitable habitat lacking any snails (Gerlach, Reference Gerlach, Pearce-Kelly, Clarke, Buckner, Sincage, Brocherieux, Depierre, Amo and Lenoble2025b). Very few of these isolated populations can be found on sequential visits. This suggests a metapopulation structure which would make them resistant to local increases in mortality, and a high degree of mobility. A clumped distribution of P. hyalina was noted by Crampton (Reference Crampton1916) who described them as living in ‘secluded colonies’, but he did not record whether the same colonies could be located over multiple years. Although P. taeniata was distributed throughout Moorea it also exhibited an uneven distribution; a 200 m transect showed two areas of high density separated by 40 m of very low density (Clarke, Reference Clarke and Drake1968), with a neighbourhood size of 200 m² (Murray and Clarke, Reference Murray and Clarke1984). A mark-recapture study did record P. taeniata as being extremely sedentary, on average moving just 2.80 m in 5 years, with a maximum record of 7.37 m (Murray and Clarke, Reference Murray and Clarke1968). However, only 5% of snails were located after 5 years, attributed to a mortality rate of 50% per year (Johnson et al., Reference Johnson, Murray and Clarke1993). As any snails dispersing beyond the 10 × 10 m sample plot would not have been detected, this 50% loss represents a combination of mortality and dispersal. Thus, Partula may be considerably more mobile than has been assumed.

A further contributor to population decline and extinction is often genetic erosion (van Oosterhout et al., Reference van Oosterhout, Speak, Birley, Hitchings, Bortoluzzi, Percival-Alwyn, Urban, Groombridge, Segelbacher and Morales2026). As numbers of breeding age adults decline, inbreeding will increase and reduce genetic diversity, with the resultant loss in evolutionary adaptability increasing extinction risk (O’Grady et al., Reference O’Grady, Reed, Brook and Frankham2004). Partulids appear to have inherently low heterozygosity (Johnson et al., Reference Johnson, Clarke and Murray1977, Reference Johnson, Murray and Clarke1986), with measured wild values from original populations around a quarter of those of other species, and less than half those of threatened snail species (e.g. Achatinella lilae and Mandarina mandarina: Price and Hadfield, Reference Price and Hadfield2014; Nishida et al., Reference Nishida, Mimura, Mori and Chiba2022). Whilst this should reduce adaptability and increase extinction risk (Spielman et al., Reference Spielman, Brook and Frankham2004), it may also mean that they are adapted to this situation. Within the conservation breeding programme genetic diversity has eroded further and some species declines are suspected to be due to inbreeding (Gerlach, Reference Gerlach2016) but this is speculative. Other species have robust populations despite having declined to crucially low levels, as low as one individual in the case of Partula taeniata simulans. One of the most successful species in the conservation programme is P. tohiveana, now successfully re-established in the wild, despite having passed through a population bottleneck of just four individuals (Gerlach et al., Reference Gerlach and Mu-Liepmann2026). This shows that, in at least some species, adaptability has not been lost even though genetic erosion must have occurred. With an evolutionary history of chance colonisations of islands, probably by single individuals, and hence strong founder effects, it may be possible that partulids are relatively resilient to the predicted impacts of inbreeding.

The decline of partulids in French Polynesia was solely due to the introduction of Euglandina. The patterns of population decline suggest that in the initial Euglandina population explosion in 1980–1994 the predators fed non-selectively and eliminated almost all populations within 2 years of contact. The few partulids that survived due to a relatively protected ecology, higher fecundity or fortuitous geography have started to expand and, in some cases, form viable populations. Subsequent predator resurgences are accompanied by a high mortality of adult Partula, but these resurgences differ from the initial invasion in that they are on a smaller scale, both in abundance and extent, and in affecting adult Partula rather than juveniles. Euglandina hunt by following their prey’s mucus trails (Clifford et al., Reference Clifford, Gross, Johnson, Martin, Shaheen and Harrington2003; Davis-Berg, Reference Davis-Berg2012; Holland et al., Reference Holland, Chock, Lee and Sugiura2012) and by chance encounter. At low prey densities trail-following is likely to be more important than chance encounters, leading to higher mortality of adults than of juveniles due to their larger trails. Partulid populations may persist through the maturation of the surviving juveniles offsetting adult losses.

The recovery of four partulid species in Fautaua Valley indicates that their metapopulations are likely to survive the predation pressures of Euglandina, even with local predator resurgences. The appearance of Platydemus manokwari flatworms has been an additional concern (Gerlach et al., Reference Gerlach, Barker, Bick, Bouchet, Brodie, Christensen, Collins, Coote, Cowie, Fiedler, Griffiths, Florens, Hayes, Kim, Meyer, Meyer, Richling, Slapcinsky, Winsor and Yeung2021) but has not restricted partulid population growth, despite being a significant pressure on reintroduced Partula (Gerlach et al., Reference Gerlach, Garcia, Chmara and Close2023a). During invasion it reaches 30,000 individuals per hectare and, even when stabilised, fluctuates seasonally between 595 and 2,000 per hectare (Gerlach et al., Reference Gerlach, Barker, Bick, Bouchet, Brodie, Christensen, Collins, Coote, Cowie, Fiedler, Griffiths, Florens, Hayes, Kim, Meyer, Meyer, Richling, Slapcinsky, Winsor and Yeung2021, Reference Gerlach, Garcia, Chmara and Close2023a, Reference Gerlach, Capel, Balogh and Maille2023b). Their daily consumption rate of 0.056 g (Gerlach, Reference Gerlach2019) equates to 21.3 Partula-sized snails per year in ideal conditions, but this predator is vulnerable to desiccation. As a result its activity is constrained to wet days and predation is reduced by 63% above ground; it may be insignificant above 2.5 m (Gerlach, Reference Gerlach2019; Anderson et al., Reference Anderson, McIlwaine, Talbot-Ponsonby, Zhao and Gerlach2026). These figures would indicate that the peak predation pressure would be substantial, but so reduced above ground as to make elimination of arboreal Partula populations unlikely. However, terrestrial species, such as P. navigatoria, might only be able to persist at the very lowest flatworm population densities.

Of the 51 French Polynesian Partula species at least 39 did not survive the initial introduction of Euglandina (a further eight species are probably extinct but may survive in poorly surveyed areas). Of these, ten are in the conservation breeding programme and these Extinct in the Wild species are the focus of the reintroduction programme with one, P. tohiveana, being successfully re-established in 2025 (Gerlach et al., Reference Gerlach and Mu-Liepmann2026). These extirpated species were probably more vulnerable to extinction than the surviving ones due to lower fecundity, a more uniform distribution or an association with a more vulnerable microhabitat. The analysis of population decline and recovery offers hope that more species can be successfully re-established in the wild by creating a series of metapopulations that could enable survival and recolonisation even in the continued presence of predators.

The extinction of most French Polynesian Partula is a classic example of the present extinction crisis. It reinforces several well-known lessons: the vulnerability of island populations; the problems of inappropriate biological control measures; the importance of biosecurity to prevent introductions; and the value of conservation breeding. Research since the extinction event has shown the importance of thorough surveys as relict populations may persist undetected long after apparent extinction. These relict populations and the conservation breeding programme demonstrate that even though critically low populations would be expected to suffer inbreeding depression and loss of viability, some species recovery is still possible.