Rapid range expansion of an invasive flatworm, Kontikia andersoni, on sub-Antarctic Macquarie Island

Spanning the Southern Ocean high latitudes, Sub-Antarctic islands are protected areas with high conservation values. Despite the remoteness of these islands, non-native species threaten native species and ecosystem function. The most ubiquitous and speciose group of non-native species in the region are invertebrates. Due to their cryptic habits and ambiguous establishment history, the impacts of non-native invertebrates on native species and ecosystems in the region remains largely unknown. Understanding how non-native invertebrate species are transported, disperse, establish and colonise new habitats is key to understanding their existing and future impacts. This research is fundamental to improving biosecurity practise and informing future management of Southern Ocean islands. We undertook invertebrate surveys on Macquarie Island to determine the current status of four non-native macro-invertebrates—Kontikia andersoni and Arthurdendyus vegrandis (Platyhelminthes: Geoplanidae), Styloniscus otakensis (Isopoda: Styloniscidae) and Puhuruhuru patersoni (Amphipoda: Talitridae). Arthurdendyus vergrandis was not intercepted in our surveys, while we found S. otakensis and P. patersoni had not expanded their range. In contrast, K. andersoni has more than doubled its previously mapped area and expanded at a rate of ~ 500 m-yr since 2004. We discuss the possible underlying mechanisms for the dramatic range expansion of K. andersoni and consider the implications for the future management of Macquarie Island.


Introduction
Invasive species have devastating impacts on global biodiversity, particularly on islands (Mack et al. 2000;McCreless et al. 2016). These impacts may occur directly through predation (Courchamp et al. 2003;Angel et al. 2009;Wanless et al. 2012;Dilley et al. 2018;Lebouvier et al. 2020), or indirectly, for example via habitat transformation (Croll et al. 2005; Abstract Spanning the Southern Ocean high latitudes, Sub-Antarctic islands are protected areas with high conservation values. Despite the remoteness of these islands, non-native species threaten native species and ecosystem function. The most ubiquitous and speciose group of non-native species in the region are invertebrates. Due to their cryptic habits and ambiguous establishment history, the impacts of non-native invertebrates on native species and ecosystems in the region remains largely unknown. Understanding how non-native invertebrate species are transported, disperse, establish and colonise new habitats is key to understanding their existing and future impacts. This research is fundamental to improving biosecurity Fukami et al. 2006;Mulder et al. 2009), hyper-predation (Courchamp et al. 2000;Travers et al. 2021), changes to soil fertility and decomposition processes (Fukami et al. 2006;Wardle et al. 2009;Towns et al. 2009), mutualisms between non-native species (Convey et al. 2010;Leinass et al. 2015) and in many other complex ways. In any case, the flow-on effects in island ecosystems can be profound. This is especially so for sub-Antarctic ecosystems which are characteristically simple and lack representatives of many functional groups (Vernon et al. 1998;Chown and Convey 2016;. Consequently, in sub-Antarctic ecosystems not only do invasive mammalian predators have particularly devastating consequences (Courchamp et al. 2003;Frenot et al. 2005;Angel et al. 2009), but so can invasive invertebrate predators, pollinators, herbivores, and macro-detritivores (Jones et al. 2003;Smith 2007;Chown et al. 2008;Greenslade et al. 2007Greenslade et al. , 2008Convey et al. 2010;Lebouvier et al. 2020), although more research into the extent of such impacts is required (Houghton et al. 2019a).

Invasive isopod and amphipod (crustacea)
The terrestrial Crustacea Puhuruhuru patersoni (amphipod) and Styloniscus otakensis (isopod) (Fig. 2), were discovered on the island around the research station in the 1990s. Prior to its establishment in 1948, the research station was the site of commercial seal harvesting by New Zealand-based operators (Cumpston 1968), thus both terrestrial Crustacea are thought to have been, by association, introduced from their native range in New Zealand between early 1800s and early 1900s (van Klinken and Green 1992;Richardson and Jackson 1995). A comprehensive island-wide survey in 1994 conducted by Davies and Melbourne (1999) across 67 sites did not detect either P. patersoni or S. otakensis, nor did an islandwide survey in 2004 at 693 sites by Greenslade et al. (2007) (see also Greenslade et al. 2008). More finescale surveying in 2004 confirmed that neither the amphipod nor isopod had markedly increased their range, although the isopod had marginally changed its limits within the boundaries of research station (Greenslade et al. 2008). Given that amphipods are highly desirable rodent food (Taylor 1986;Russell et al. 2020), their population is expected to increase following the recent eradication of rodents on Macquarie Island between 2010-2014 (Macquarie Island Pest Eradication Project, 'MIPEP') .
As macro-detritivores, non-native isopods can have considerable impacts on Southern Ocean Island ecosystems (Jones et al. 2003). Neither native nor non-native isopod species are preferred rodent food according to several rodent diet studies on Southern Ocean Islands where they are present (Houghton et al. 2019a). However, they are potentially sensitive to habitat changes, such as vegetation recovery following the cessation of grazing and burrowing since the removal of rabbits during MIPEP (Shaw et al. 2011;Whinam et al. 2014;Fitzgerald 2020). On Macquarie Island, S. otakensis is present in humid habitats with good plant cover, healthy leaf litter and possibly a soil moisture content threshold above 30% (Greenslade et al. 2008). Interestingly, isopod abundance declined at sites monitored for invertebrates between 2015-2018, despite recovering vegetation (post-MIPEP) . This may be due to changing soil properties and vegetation communities following invasive mammal removal .

Invasive flatworms (platyhelminthes)
Flatworms were first discovered at two topographically restricted creek-side sites near Lusitania Bay (for island place names, see Fig. 3) on the southeastern coast of Macquarie Island in 1997 by R. Blakemore whilst conducting earthworm surveys (Greenslade et al. 2007). Commercial sealing and  Greenslade et al. (2007), and b) this work, which included a combination of twenty four island-wide invertebrate moni-toring sites (2015-2018) and a dedicated flatworm survey in 2018. Note that Greenslade et al. (2007) surveyed a further 511 sites island-wide with negative results for K. andersoni but spatially explicit data for these were not recorded penguin harvesting by New Zealand-based operators occurred in the Lusitania Bay area until 1919 (Cumpston 1968), thus flatworms were likely introduced from New Zealand to Macquarie Island more than 100 years ago in association (Greenslade et al. 2007). The flatworms collected in 1997 were later identified as Kontikia andersoni and Arthurdendyus vegrandis (Winsor and Stevens 2005) (Fig. 2). Arthurdendyus vegrandis was described as a new species by Winsor and Stevens (2005), however it is likely to be con-specific with species from the South Island of New Zealand (Greenslade et al. 2007) where many Arthurdendyus species are undescribed (Yeates et al. 1997). Greenslade et al. (2007) conducted a baseline survey for flatworms in 2004, which included 182 sites in the south-east 'target area', and an additional 511 sites across Macquarie Island. No surveys were conducted in restricted access areas, such as the south-eastern coast of the island south of Lusitania Bay. The 2004 survey confirmed that both flatworm species had spread from their likely region of introduction. The location of the 511 additional island-wide sites reporting negative results for both flatworms were not available to us. The island-wide invertebrate survey in 1994 which included timed search counts at 67 sites also did not detect either A. vergrandis or K. andersoni (Davies and Melbourne 1999). Greenslade et al. (2007) reports that these 67 sites were outside the 'target area' for flatworm searches. The precise locations of these sites were also not available to us.
Globally, flatworms have been, and continue to be, inadvertently but readily introduced to new regions alongside human activity (Sluys 2016). Once introduced, they often remain undetected for a time, spread easily, and become invasive (Sluys 2016). Their success is attributed to their cryptic nature, generalist predation, unpalatability to predators, ability to regenerate when cut in two and asexually reproduce via fission (fragmentation) (Winsor et al. 2004;Sluys 2016). Being voracious generalist predators of native soil invertebrates, including earthworms, snails, slugs, insect larvae, isopods, springtails and more, their invasion can have serious consequences for local ecosystems, affecting nutrient cycling regimes, threatening native species, and altering plant communities (Sluys 2016). Arthurdendyus species are all native to New Zealand and prey heavily on earthworms (Yeates et al. 1997). Two species of Arthurdendyus are invasive pests in the United Kingdom (UK)-A. triangulatus in particular is a major pest causing a reduction or temporary elimination of native earthworms with serious implications for soil functioning (Boag et al. 1997;Boag and Yeates 2001;Cannon et al. 1999;Murchie and Gordon 2013), and drastic declines in other animals that rely on earthworms as prey (Alford et al. 1996;Boag 2000). Kontikia andersoni is also a pest in the UK. Its feeding habits are not well known (Winsor et al. 2004), except that it specialises in predation of Collembola (springtails), and consumes slugs and lumbricid worms (Flatworms of Cornwall 2021). The impacts of Kontikia and Arthurdendyus species in the UK are considered so great that it is an offence under the UK's Wildlife and Countryside Act to knowingly distribute them (Eversham 2013; Sluys 2016).
We used historical data on the distribution of the two flatworm species (Greenslade et al. 2007) and the two terrestrial Crustacea (Greenslade et al. 2008) on Macquarie Island together with contemporary surveys of each species to assess their current distribution and status. The data sources were (1) an islandwide invertebrate trapping program (2015-2018), (2) focused searching in areas where these species had been previously detected (i.e. by Greenslade et al. 2007Greenslade et al. , 2008, and (3) a dedicated island-wide survey for flatworms. We discuss mechanisms by which these species may spread, make recommendations for reducing the risk of further intra-island transfer, and highlight the biosecurity implications for Australia.

Island-wide invertebrate monitoring
Comprehensive invertebrate surveys were carried out at 24 sites across Macquarie Island over three consecutive austral summers (2015)(2016)(2017)(2018). At each site, a variety of trapping techniques were used-pitfalls, sweeping, 20-min counts and litter sampling. Vegetation beating was added in the second season (2016-2017), yellow pans in the third season (2018).
Replicates of each trap type were used within sites (five pitfalls, three yellow pans, three beatings, three sweeps, two counts, three litter samples), and trapping events at each site for pitfalls, beatings, sweeps, and counts were repeated multiple times during each season-2015/2016, 2016/2017 three events, 2018 three events, except for pitfalls with two events (see Houghton et al. 2019b and Houghton 2020 for detailed trapping and surveillance methodology). The 24 sites were selected to represent five key vegetation communities on the island ('Herbfield', 'Stilbocarpa', 'Short grassland', 'Tall grassland' and 'Feldmark', as per Selkirk et al. 1990), and were defined by 10 × 10 m vegetation surveys (Houghton 2020) (Survey data available via the Australian Antarctic Division Data Centre, doi:10. 26179/k24z-nx57).
As the distributions of the four target non-native invertebrate species began to emerge from the islandwide monitoring, it became clear that (1) focused searching would be sufficient for the terrestrial crustacea, (2) A. vergrandis was not readily found, and (3) K. andersoni appeared more widespread. These latter revelations prompted a dedicated island-wide survey for flatworms.

Terrestrial crustacea survey
Based on the 2004 baseline distribution of P. patersoni and S. otakensis (Greenslade et al. 2008), focused searching was conducted for P. patersoni around the isthmus south of the research station (the northern tip of the island) and for S. otakensis to the south and west of Halfway Hill (north-east corner of the island) (Greenslade et al. 2008). Areas that were searched by Greenslade et al. (2008) were revisited. Both P. patersoni and S. otakensis are relatively large (up to 5.5 and 11 mm respectively-Greenslade 2006; Richardson and Jackson 1995), thus dedicated searches in suitable habitat were sufficient to indicate presence or absence.
Island-wide flatworm survey A comprehensive survey was designed to map the distribution of flatworms on the island in more detail. Permits were acquired for all restricted coastal areas except for the sensitive areas at the far southwestern corner of the island. Sites were selected around each 1 km centroid in coastal areas, from which the closest suitable habitat (i.e. vegetated and damp) within 100 m was identified to conduct a timed hand search. Additional to the coastal sites, four crossisland transects following major drainage areas were also surveyed, with sites spaced every 500 m. A GPS coordinate was taken to mark the centre point of each survey site. Wetness of the ground surface was determined by assessment in the field (MH), utilising sight and touch. A five-point scale was used with the ground surface defined as being dry to touch (1), being damp to touch (2), being wet to touch (3), being very wet with water visible (4) and being saturated and severely waterlogged (5). Dominant vegetation at each site was noted, as were other significant features including drainage, aspect, nearby water sources, bare surface, and animal activity. A total of 59 sites were surveyed from the eastern half of the island, and 44 from the western half (includes coastal and inland areas). Surveys (timed searches by 1, 2, or 3 people) were conducted in daylight hours between the 22nd February and 16th March 2018. Due to their relatively large size (up to 2 cm), flatworms are easily detected by the naked eye. Search time was initially 40 min (sites F1 -F36, excepting F6), but it became clear that flatworms, if present, were detectable in under 20 min. Thereafter, depending on the number of people conducting the survey, search time was reduced to 20 or 21 min at each site, allowing more sites to be surveyed. When more than one person attended the survey, each searched separately from the site centre, and the total search time was divided by the number of people (i.e. two people searching 10 min each totalled 20 min; three people searching 7 min each, totalled 21 min). Searching involved overturning stones and rocks, teasing apart wet leaves and litter, looking at the base of plants. The search carried on until a flatworm was detected (upon which the search time was recorded), or until the total search time had elapsed. If no flatworms were found within this time limit, a negative result was recorded. In total, 103 sites were surveyed across Macquarie Island of which 48 were searched by a single person (MH), 17 by a team of two people, and 38 by a team of three people (doi:10. 26179/k24z-nx57).

Mapping
To calculate the change in distribution of K. andersoni over time, we first divided the island into 1 km grid squares using Manifold GIS Systems (Version 8). Both the 2004 and the 2015-2018 survey data were overlaid onto the 1 km grid to quantify occupancy by K. andersoni at these time periods. We then quantified the change in the number of occupied 1 km x 1 km cells and the proportion of the island occupied by K. andersoni over time.

Isopod and amphipod
Focussed low intensity searching in 2018 for S. otakensis and P. patersoni, combined with results from the 24 island-wide invertebrate monitoring sites (2015)(2016)(2017)(2018), revealed that these species had not markedly expanded their range since they were last surveyed in 2004. Neither of these species were incidentally detected during the island-wide flatworm surveys which covered suitable habitat.

Flatworms
A. vergrandis was not detected at any of the 24 island-wide invertebrate monitoring sites, nor during the island-wide flatworm survey, despite additional focussed searches in its last recorded range in 2004 (doi:10.26179/k24z-nx57).
K. andersoni was found at six of the 24 islandwide invertebrate monitoring sites, of which four were outside its previously known distribution (doi:10.26179/k24z-nx57). Two of the new sites were on the west coast, at which only a single K. andersoni was detected during the three years of monitoring. At the other two new sites (one on the east coast, one on the southern coast), K. andersoni was abundant across years.
During the island-wide flatworm survey, K. andersoni was found at 29 of the 103 sites (Fig. 1, doi: 10.26179/k24z-nx57), with 24 outside its previous mapped distribution in 2004. Our survey found K. andersoni occupying most south-eastern coastal areas of Macquarie Island, as well as inland areas in the central region of the island around Green Gorge tarn and the eastern slopes of Mt Law. K. andersoni was detected at 24 eastern sites, the northernmost being ~ 500 m north of Brothers Point in the northeast of the island (Fig. 1, see Fig. 3 for island place names). Eighteen of these were from coastal areas (below the escarpment edge), while six were found at inland sites. K. andersoni was also found at five west coast sites, the northern-most being at Sellick Bay.
K. andersoni animals were found both on the ground surface over decaying plant material and within wet leaves and litter up to 1-2 cm deep. They were more often associated with wet, rather than damp or dry sites, meaning the soil at these sites was never dry or pliable, the soil always moist and at times waterlogged (three and four on the wetness scale). At coastal sites, K. andersoni typically occurred in the wet litter of tussock grass (Poa foliosa) and Macquarie Island cabbage (Stilbocarpa polaris). In areas where these species do not occur (e.g. away from the coast, on inland tracks, and in the region of Green Gorge tarn and Mt Law), K. andersoni was typically found in wet moss beds and amongst small Epilobium spp. herbs. All 24 detections of K. andersoni on the east coast were at sites with a wetness index of three or four (i.e. wet, or very wet). On the west coast, the five detections coincided with a wetness index of three.
The detection of K. andersoni at Hurd Point, the southernmost point of the island, indicates that between 2004 and 2018 the flatworm spread south 6.4 km over 14 years -a mean rate of ~ 460 m/ yr. It also spread 7.2 km north from its 2004 limit, representing a mean rate of ~ 515 m/yr. It has also expanded inland to an area of ~ 170 m in altitude around Mt Law. Furthermore, for the first time we found K. andersoni on the west coast of the island, 3.9 km from its previous western extent in 2004. While it was detected over a span of 11.4 km on the west coast, its density was much less than the east coast distribution.

Discussion
Over a 14-year period, the small, flightless and slowmoving invasive flatworm Kontikia andersoni, has greatly expanded its range on Macquarie Island, while the distributions of A. vergrandis, S. otakensis and P. patersoni have not markedly changed. Since Greenslade et al. (2007) calculated invasive flatworms had spread from their presumed introduction site at Lusitania Bay at a rate of ~ 10 m/yr, the invasion of K. andersoni has accelerated south, west and north at around 500 m/yr, and more than doubled its range. It is possible that K. andersoni was present south of Lusitania Bay but was undetected in 2004 as this area was not surveyed due to access restrictions (Greenslade et al. 2007). However, despite this uncertainty, our finding that K. andersoni has rapidly expanded its range since 2004 is supported by flatworm absences previously reported by Davies and Melbourne (1999) and Greenslade et al. (2007) across 578 sites island-wide. Such a lag in the invasion rate for a non-native invertebrate has been detected in the region, notably for the flightless, predatory carabid beetle (Merizodus soledadinus) on sub-Antarctic Kerguelen Island. Introduced more than 100 years ago, in the past few decades the beetle has drastically expanded its range with devastating consequences for endemic invertebrate fauna (Lebouvier et al. 2020). Such lag phases in invasion processes are common, including for invertebrates (see Lebouvier et al. 2020, and references within for discussion), but not regularly documented.
The dispersal mechanism for K. andersoni on Macquarie Island is unknown. Its predominantly coastal distribution is not surprising given the large tracts of unsuitable habitat in-between the east and west coasts. Specifically, the island's interior is colder, drier, higher in altitude, often comprised of feldmark and gravel, and with low vegetation cover. Yet K. andersoni has overcome this barrier, dispersing kilometres from its east coast introduction site to colonise the west coast of the island. Humans are important vectors of non-native plant and invertebrate propagules to Antarctica and sub-Antarctic islands (Frenot et al 2005;Chown et al. 2012;Houghton et al. 2016;Duffy and Lee 2019). Human activities also drive intra-regional transfer of propagules within Antarctic sites and sub-Antarctic islands (Lee and Chown 2011;Hughes et al. 2019;Bartlett et al. 2020;Lebouvier et al. 2020). On Marion Island, cargo movements by helicopters are implicated as likely vectors of introduced slugs (Derocerus panormitanum), which are now widespread and abundant on the island (Smith 1992;Chown et al. 2002). Helicopter transport of people and cargo is frequent during Macquarie Island resupply operations, but these movements rarely occur between field huts, or between the east and west coasts. However, there was a significant increase in helicopter activity between [2011][2012][2013][2014] during the large-scale MIPEP program (Springer 2016). Following the aerial baiting phase of the pest eradication program, a three-year active hunting phase commenced. During this time, high quantities of cargo were moved between field huts, the research station and the ship (Houghton, pers. obs.). Despite this increase in cargo movement, K. andersoni has not been detected at the research station where suitable habitat is present.
The spread of flatworms between and within regions around the world is closely linked to the movement of horticultural material such as soil, pots, plants and garden supplies (Boag et al. 2010;Sluys 2016). While soil and plants are not actively transported around Macquarie Island, human foot traffic around the island is common, by which wet plant and soil material is transported on boots and equipment. Flatworms have been found moving along Macquarie Island walking tracks (Greenslade et al. 2007), and footpads have been identified as dispersal corridors for invasive invertebrate movement elsewhere in the region . Similar observations have been made for invasive plants in the sub-Antarctic (Scott and Kirkpatrick 1994; Le Roux et al 2013;Sindel et al. 2017). Given the flatworm's ability to fragment and regenerate (Winsor et al. 2004;Sluys 2016), human-mediated transport of K. andersoni via foot is highly probable. There was intensive foot traffic on the island during MIPEP (Springer 2016). First mitigation teams were deployed on foot to bury carcasses, followed by 2.5 years of hunting teams industriously traversing the island by foot in search of remaining rodents and rabbits. Five temporary huts were placed in locations distant from the established network of walking tracks to enable field workers working access to more of the island. Annual hunting field teams comprised 12-14 people and up to eleven detection dogs, each walking over all accessible ground on the island (including between coasts) on a daily basis. The flatworm's presence at Mt Law and the southern flank of Green Gorge tarn is close to routes regularly used by hunters and park rangers to travel between the east and west coasts.
Birds may also contribute to the dispersal of K. andersoni on Macquarie Island. Seabirds inadvertently transport non-native biological material between and within landmasses in the lower latitudes, as well as the Antarctic and sub-Antarctic region, including soil microarthropods up to 1.5 mm in size (Krivolutsky et al. 2004), plant seeds (Vidal et al. 2003;Turner et al. 2006;Sindel et al. 2017), bacterial pathogens (Cerdà-Cuéllar et al. 2019) and viruses (Springer and Carmichael 2012). Their activity is a possible dispersal mechanism for non-native invertebrates on Signy Island (Hughes and Worland 2010) and the Kerguelen archipelago (Lebouvier et al. 2020). Birds could transport flatworms or their fragments on their feathers or feet, in mud or guano, as they move between flatworm-suitable coastal habitats around penguin colonies and waterways. Furthermore, frequent seabird travellers along and between the west and east coasts include kelp gulls (Larus dominicanus Lichtenstein, 1823), skuas (Stercorarius antarcticus Lesson, 1831) and giant petrels (Macronectes giganteus Gmelin, 1789, Macronectes halli Mathews, 1912 (Travers 2021). However, bird-mediated dispersal does not explain the northern limits of the flatworm's range, nor its rapid and recent range expansion. Likewise, its distribution cannot be explained by considering ocean currents as a potential dispersal pathway via marine debris or marine animals. Salinity stress experiments on the larvae of the Eretmoptera murphyi, a fly introduced to Signy Island, found they could survive up to two weeks in sea water (e.g. caught in terrestrial vegetation washed into the sea, or trapped in seal fur), and suggested that this might permit local movement and colonisation on a scale of tens of kilometres (Bartlett et al. 2021). Similar studies into the salinity tolerance of K. andersoni would help to clarify its potential to disperse in this way.
Kontikia andersoni has an intermittent occurrence on the west coast of Macquarie Island over a range of ~ 11 0.4 km and was only detected as singletons at our western fixed invertebrate monitoring sites. These observations point to less abundant populations on the west coast of the island, where we generally observed sites with a lower wetness index than eastern sites (doi: 10. 26179/k24z-nx57). Future research is needed to clarify the role of aspect, climate and microclimate on habitat availability and suitability for K. andersoni.
Climate is changing across the region. There have been recent decadal changes in summer water availability, increased wind speed and sunshine hours on Macquarie Island (Bergstrom et al. 2015). It is not yet known how climatic change will influence K. andersoni. Observations and predictive models from other sub-Antarctic islands suggest there may be changes to available habitat and changes in species-species interactions (e.g. Davies et al. 2011). A recent study on Iles Kerguelen found climate change was a key factor in the recent rapid range expansion of an invasive beetle due to increases in suitable habitat (Lebouvier et al. 2020). Multiple studies for the region predict warming climate will influence biodiversity of soil invertebrates (e.g. Bokhorst et al. 2008;Janion et al. 2010;Nielsen and Wall 2013;Andriuzzi et al. 2018). Furthermore, warming climate is predicted to drive expansion of existing non-native species and see the arrival and expansion of new non-native species; plants (Pertierra et al. 2017;Molina-Montenegro et al. 2019;March-Salas and Pertierra 2020) and invertebrates (Lebouvier et al. 2011;Bartlett et al. 2020;Pertierra et al. 2020), or both (Frenot et al. 2005;Duffy et al. 2017;Lee et al. 2017;Duffy and Lee 2019).
Species distribution models and ecological niche models of both native and non-native species under different climate change scenarios point to both range contractions and expansions. For example, the native Antarctic chironomid midge Parochlus steinenii Gerke, 1889 is predicted to expand its distribution on the Antarctic Peninsula as the climate warms (Contador et al. 2020), while ensemble forecasting of four species of non-native Collembola in Antarctica identifies several areas in which environmental conditions will be suitable for their invasion under future warming scenarios (Vega et al. 2021). Such research informs pre-emptive biosecurity, targeted surveillance at high-risk sites, the identification of pathways of introduction, and development of rapid responses in the early stages of invasion, all of which are increasingly required (Wilson et al. 2009;Foxcroft et al. 2011;Duffy et al. 2017). However, to inform the management of established invasive species, information on native species and community functioning is critical (Molina-Montenegro et al. 2019). Monitoring native invertebrate populations on Macquarie Island is essential to determine how K. andersoni is impacting the island ecosystem. Invertebrate monitoring will also broaden our understanding of the role of climate change is impacting the island ecosystem.
Arthurdendyus vergrandis was not detected during the island-wide invertebrate monitoring program between 2015-2018, despite one of these sites overlapping an area where A. vergrandis was found in 2004 (Greenslade et al. 2007). Our monitoring program involved multiple trap methods in replicates each season, including timed searches. A. vergrandis was also not found during the dedicated island-wide flatworm survey, despite searches in both areas of its previous recorded distribution (in 2004). Rocks, stones and debris were regularly upturned and examined during our survey, a technique which had previously detected A. vergrandis (Greenslade et al. 2007). It seems unlikely that having established, A. vergrandis is now absent from Macquarie Island. We assume that it remains at least within the limited areas described by Greenslade et al (2007), but at a much lower abundance than K. andersoni. This finding is consistent with Greenslade et al. (2007), that K. andersoni are more widespread and at least three times more abundant than A. vergrandis.
Invasion across the sub-Antarctic region is relatively recent with most of the islands discovered only 200 or so years ago (see Frenot et al. 2005). Invasion studies from the sub-Antarctic and Antarctica (Hughes and Worland 2010;Frenot et al. 2005;McGeoch et al. 2015), and more broadly (Williamson and Fitter 1996;Kolar and Lodge 2001;Sol et al. 2012;Walther et al. 2009), show only a minority of non-native species introduced become invasive, while many other non-native species establish and are 'persistent' neither expanding or contracting their range over decades. Our survey found Arthurdendyus vergrandis had very low or localised populations, and we found no evidence that Styloniscus otakensis and Puhuruhuru patersoni have expanded their ranges, remaining restricted to the north of the island. Despite all three of these species residing on the island for more than 100 years, to date they have persisted in a narrow range. Suitable habitat for both S. otakensis and P. patersoni is available outside and adjacent to their existing range (van Klinken and Greenslade 1992;Richardson and Jackson 1995) and there are no identifiable barriers to their dispersal. Factors limiting their spread therefore remain unclear. Furthermore, the removal of invasive rodents from the island, which are known on sub-Antarctic islands to preferentially prey on amphipods (Houghton et al. 2019a), has not influenced amphipod numbers. Although in 2018 we detected singletons of S. otakensis at sites outside their 2004 range, given the animal's tendency to mass in high numbers where present (Houghton, pers. obs.), up to 4000-6000 animals per square metre (Greenslade et al. 2008), and as multiple years of monitoring did not reveal further individuals at these sites, we suspect these detections were bird-assisted or wind-swept vagrants. The high winds experienced on Macquarie Island are renowned for dispersing invertebrates (Hawes et al. 2013). Similarly, a single S. otakensis detected in 2004, 2 km south of its mapped distribution (Greenslade et al. 2008), may also have been a vagrant, as we found no individuals in this region during our survey. It is possible that some populations of S. otakensis remain undetected by past surveys and our work, as despite high densities where it is found, its distribution is patchy (Greenslade et al. 2008).

Management implications
The impact of a rapidly expanding K. andersoni population on the Macquarie Island ecosystem is unknown. Flatworms are generalist predators and on Macquarie Island they have no known predators and little competition from other invertebrates. K. andersoni could reduce the abundance and diversity of invertebrates on the island, as invasive flatworms do elsewhere (Boag and Yeates 2001;Cannon et al. 1999;Murchie and Gordon 2013;Sluys 2016), altering soil nutrient cycling and impacting vegetation communities. This study highlights the need to establish an appropriate monitoring program of native invertebrates on Macquarie Island in order to identify and quantify impacts of invasive species.
Given there is similar habitat outside of K. andersoni's current range on the island, further expansion is feasible. Proactive development of hygiene measures would help to prevent the inadvertent transport of K. andersoni. Rigorous washing of boots, shoes, walking poles and other equipment that have been in contact with the soil when leaving invaded areas is recommended. This could be done in many of the numerous fresh running streams or accessible water on the coast. The use of suitable biocides should be investigated.
Kontikia andersoni and flatworms of the genus Arthurdendyus are not known to be present in other parts of Australia (Greenslade et al. 2007). Both flatworms present potential environmental and agricultural risks to mainland Australia based on their invasive impacts elsewhere. Of real concern, is that K. andersoni will continue to advance north to the island's research station, where during resupply operations propagules attached to outdoor stored cargo could be transferred via the resupply ship back to Tasmania. In this context there is pressing need to identify appropriate biosecurity measures. Currently, no washing or biocides are used on cargo and equipment returning to Australia.
All four species discussed here are registered as introduced species to Australia (ABRS 2021) as they occur on Macquarie Island, which falls under Tasmanian jurisdiction. However, their presence 1500 km away from mainland Tasmania may have influenced the amount of attention they have received. Biosecurity protocols for transport from Macquarie Island to Tasmania should be reviewed to minimise the risk of inadvertently moving invasive flatworms to mainland Tasmania, and Australia. Our work demonstrates how quickly these invaders can spread, even on a remote world heritage island.
Author contributions All authors contributed to the study conception and survey design. Field data collection and species identifications were undertaken by Melissa Houghton. Analysis was performed by Aleks Terauds, Melissa Houghton and Justine Shaw. All figures are by Aleks Terauds. All authors contributed to drafting of the manuscript. All authors read and approved the final manuscript.
Funding This work was funded by the Australian Government's National Environmental Science Programme (NESP) through the Threatened Species Recovery Hub, and the Australian Antarctic Science program (AAS Project 4035). Additional support was received by Melissa Houghton from the Australian Academy of Science through the Max Day Environmental Science Fellowship.
Data availability All data generated or analysed during this study are included in this published article and its supplementary information files.

Conflict of interests
The authors have no relevant financial or non-financial interests to disclose.