The terrestrial high-Arctic has, so far, escaped the worst impacts of non-native plant establishment. However, increasing human activity and changing climate raise the risk of introductions and establishment, respectively. The lack of biosecurity in the terrestrial Arctic is thus of concern. To facilitate the development of biosecurity measures on the rapidly warming and highly trafficked archipelago of Svalbard, we generated ecological niche models to map the bioclimatic niche potential of 27 non-native established or door-knocker vascular plant species across Svalbard, identify species with a high risk of widespread occupancy, and locate hotspots of potential current and future invasions. Under the current climate the three species with the highest threat in terms of broad potential area of occupancy and known invasion potential were Deschampsia cespitosa, Ranunculus subborealis subsp. villosus and Saussurea alpina. However, under future climate, most of the considered species have potentially wide distributions across the archipelago. Remote eastern islands were a hotspot region for broader potential establishment of non-native species under the current climate. Our results suggest that many non-native plant species have a broader macroclimatic niche on Svalbard than they currently occupy, and that other factors probably limit both dispersal and establishment outside their current localised distributions. Environmental management on Svalbard has a limited window of opportunity to act early in containing and preventing the spread of non-native plant species beyond the few settlements where they currently exist. Moreover, preventing introductions and establishments on the remote and rarely visited islands of Edgeøya, Barentsøya and Bjørnøya could be also a priority action to safeguard sanctuaries of the archipelago’s natural ecosystems.

Bioclimate, Climate change, Distribution, Neophyte, Svalbard, Tundra

Biodiversity redistribution processes related to the Anthropocene are rearranging regional biotas to higher and cooler elevations and latitudes (Pecl et al. 2017). The spread of non-native species is a major threat to biodiversity and ecosystem functioning (Vilà and Hulme 2017; IPBES 2019). The establishment of multiple non-native species can lead to facilitatory and accumulating changes in habitats known as invasional meltdown (Simberloff and Von Holle 1999). Invasional meltdown processes disproportionally threaten remote, species-poor ecosystems with low biotic resistances to invasions (limited functional redundancies). Meltdown is thus expected to manifest in species-poor systems such as islands (Moser et al. 2018), but remains poorly documented in polar systems (Pertierra et al. 2023). Identifying the most vulnerable areas to the establishment of multiple non-native species and potential impacts or meltdown can be narrowed down using macroecological principles: identifying ‘areas of risk’ with highest aggregated suitable establishment potential by the pool of alien species introduced to a site. In this regard species distribution models (SDMs) are a foundation for identifying regions of concern for potential invasions, facilitating further research and management.

The terrestrial Arctic is characterised by species-poor communities under strong ecological determination by abiotic conditions (Daniëls et al. 2013; Ims et al. 2013). The Arctic is undergoing a higher degree of warming than anywhere else on Earth (Serreze and Barry 2011; Box et al. 2019). At the same time, increasing human activity within the Arctic (Meier et al. 2014) increases both propagule pressure and ecosystem disturbance, respectively facilitating the introduction and establishment of non-native species. Non-native vascular plant species are continually arriving in the Arctic, yet to date, most have not established (i.e. transient or casual species) roughly following the rule of tens (Williamson and Fitter 1996). However, with warming more species can be expected to establish (Alsos et al. 2015). Amelioration of climatic conditions may proceed beyond a point of no return (a tipping point), where the stressors are no longer limiting factors for wider establishment of non-native vascular plants and their associated biota, potentially leading toward invasional meltdown of the Arctic tundra ecosystem. Wasowicz et al. (2020) found that 341 non-native vascular plant species have been recorded in the Arctic, of which 188 are considered naturalised in at least one of the 23 floristic regions of the Arctic. Only 11 taxa are currently considered invasive, and they are only known from Alaska and Iceland.

Svalbard is a high-Arctic archipelago, located between 73 and 81°N. Svalbard is undergoing a high degree of climatic warming, with annual mean temperatures predicted to increase by between 3 °C and 10 °C from 1971–2000 to 2071–2100 and annual precipitation predicted to increase by 40% to 60% over the same time period (Hanssen-Bauer et al. 2019). It is heavily trafficked relative to other high-Arctic regions. The combination of warming and human activity in a remote, species-poor ecosystem (Duffy et al. 2017), places Svalbard at high risk of invasional meltdown. To date, there are more than 100 species of non-native vascular plant species that have been reported from Svalbard (Alsos et al. 2015; Wasowicz et al. 2020) although only 13 have established as of 2023 (Artsdatabanken 2023; Table 1) and these are limited to Svalbard’s few settlements and around old cabins, all on the main island of Spitsbergen. About 75% of the introduced species are considered seed contaminants imported with livestock fodder, animal bedding, shipping of goods (Artsdatabanken 2018) and transported on visitors shoes (Ware et al. 2012). While the intentional introduction of non-native species is forbidden on Svalbard (Svalbardmiljøloven ‘Svalbard environmental protection act’; Klima- og miljødepartementet 2001), there are no biosecurity measures in place to prevent unintentional introductions. This is in stark contrast to other polar (i.e. Antarctica; Hughes and Pertierra 2016) and remote (e.g. Australia; Anderson et al. 2017) regions where stringent measures have been put in place to limit import of non-native species. Importantly, island systems harbour a distinct biota under processes of isolation and speciation with low biotic resistance, making them highly vulnerable to the arrival of non-native species. Indeed, Svalbard has several endemic and near-endemic vascular plant species (Elven et al. 2022).

To facilitate policy development on Svalbard, identification of regions suitable for future establishment of non-native species is required. By encompassing the entire pool of non-native species currently known from the region, broad scale patterns can be identified that can be related to potential ecosystem impacts. The objective of this study is to carry out a risk assessment of macroclimate suitability for non-native vascular plant species on Svalbard by identifying regions within the potential bioclimatic niche under both current and future climates. We aim to: 1. Rank non-native species based on both bioclimatic suitability and current ecological impact assessment category (from the 2023 alien species list; Artsdatabanken 2023). 2. Identify regions with high numbers of potential non-native species under current and future climate scenarios, in order to determine areas where increased vigilance or management is required. 3. Detect low risk zones under both present and future conditions which are less bioclimatically suitable for the focal non-native species and could be considered for reinforced protection. 4. Identify assemblages of non-native species in different regions of the archipelago, in order to understand potential bioregionalization of non-native species and potential impacts.

Climate suitability models were trained for all 27 species based on global occurrences, and these were projected across Svalbard. Model evaluation scores varied between species and modelling methods. Generally, the machine learning methods (RF and GBM) had the highest AUC and TSS values, whilst the simplest method, surface range envelopes, had the lowest evaluation scores (Suppl. material 1: fig. S4). The species split roughly into two groups; one with relatively high evaluation scores (median AUC across models of >0.8 and median TSS of >0.5), and another group with relatively low evaluation scores (Suppl. material 1: fig. S4). The species with the higher evaluation scores were generally those with more restricted geographic ranges while globally widespread species had lower evaluation scores by both AUC and TSS (Suppl. material 1: figs S1, S5).

For some species, one of the bioclimatic variables was by far the most important in determining the distribution. For example, temperature of the warmest quarter for Saussurea alpina, mean diurnal range for Tripleurospermum maritimum and maximum temperature of the warmest month for Veronica longifolia (Suppl. material 1: fig. S6). While for other species, multiple variables are of roughly equal importance (Suppl. material 1: fig. S6). Many species had a unimodal or negative response to growing season temperatures, and a positive response to temperature annual range and a negative response to mean diurnal range (Suppl. material 1: fig. S7).

Under current climate, climate suitability varied across species (Fig. 1). Some species have high potential niche suitability across much of Svalbard such as Deschampsia cespitosa, R. subborealis subsp. villosus and Saussurea alpina. When the projections were thresholded to maximize TSS (Suppl. material 1: fig. S8), these were also the species predicted to be most widespread, with an area of potential occupancy of 100% of the non-glaciated parts of Svalbard for D. cespitosa and R. subborealis subsp. villosus and 75% for S. alpina (Fig. 2, Suppl. material 1: table S2). However, of these three species, only D. cespitosa has an invasion potential (Suppl. material 1: table S1, Artsdatabanken 2023), thus this is ranked as the top risk species, followed by Alchemilla wichurae and Alchemilla subcrenata, also assessed to have invasion potentials (Fig. 2). In contrast, climate suitability for other species is very low under current climate including Poa pratensis, Taraxacum sect. Ruderalia and Achillea millefolium.

Figure 1.

Ensemble projections of the potential macroclimatic niche distribution of all non-native vascular plant species across the archipelago of Svalbard under current climate. Species labels follow Artsnavnebase (Species Nomenclature Database; Artsdatabanken). See Table 1 for scientific names corresponding to the data downloaded from GBIF.

Figure 2.

Area of potential range occupancy of 27 non-native vascular plant species across the Svalbard archipelago under current and future (SSP2-45 and SSP5-85) climate scenarios (see Suppl. material 1: figs S7, S9 for the corresponding maps). These are estimated using 30’ second rasters. The total ice-free area of non-glaciated land available using this dataset is 24 210 km² and is denoted by a dashed horizontal line. Species are ranked first by their ecological impact category (and separated by a vertical line), next by the current area of potential range occupancy and lastly by the area of potential occupancy under the future intermediate climate scenario. All Low impact (LO) species score 1 for ecological effect (range 1-4) except Anthriscus sylvaticus (3, not currently present in Svalbard), Festuca rubra subsp. rubra and Stellaria media (both 2; Suppl. material 1: table S1, Artsdatabanken 2023).The areas and percentages of the non-glaciated parts of Svalbard are included in Suppl. material 1: table S2. Species labels follow the Norwegian Artsnavnebase (Species Nomenclature Database; Artsdatabanken). See Table 1 for scientific names corresponding to the data downloaded from GBIF.

For most species the climate suitability is predicted to be higher under both future climate scenarios than under the current climate (Suppl. material 1: figs S9, S10). However, Poa palustris and Ranunculus acris do not have suitable climate available in Svalbard under neither present nor either of the future scenarios (Fig. 2, Suppl. material 1: figs S11, S12). The area of potential occupancy for several species including Poa annua, Poa pratensis, Ranunculus repens, Rumex longifolius, Trifolium pratense and Trifolium repens increases from around 0% under current climate to almost 100% of the non-glaciated regions of Svalbard under both future scenarios (Fig. 2). In contrast, the area of potential occupancy of Saussurea alpina decreased by around 3000 km² under the severe future climate scenario, but increased under the moderate future climate scenario by around 5000 km² compared to the current. Meanwhile the potential occupancy of Veronica longifolia decreased from a current potential area of around 5000 km² to below 500 km² under both future scenarios (Fig. 2).

Potential richness of non-native vascular plant species is greatest in the remote eastern islands of the Svalbard archipelago such as Edgeøya. The current climate here is suitable for up to 13 non-native vascular plant species (Fig. 3). Current climate is suitable for fewest species in the eastern part of the main island of Spitsbergen and northern Nordaustlandet, with some parts having suitable climate for only two species (Deschampsia cespitosa and R. subborealis subsp. villosus). The southern island of Bjørnøya also has suitable climate for many non-native species (Fig. 3).

Figure 3.

Potential species richness of non-native vascular plant species across Svalbard under current and two future climate scenarios. Potential species richness is estimated as the sum of species with potential occurrence based on thresholding of model predictions maximizing the true skill statistic.

The potential non-native species richness was far higher under both future climate scenarios than under the current climate. Some regions including Edgeøya had suitable climate for 25 non-native vascular plant species, out of a pool of 27 species under both future scenarios (Fig. 3). Potential species richness increased the most between current climate and the future scenarios in the central Spitsbergen region, where human activity is greatest (Fig. 3).

The projected models indicate a clear bioregionalization of potential assemblages of non-native species, with Eastern, Central Fjord and Western clusters being apparent under both current and future climates (Fig. 4). The cluster prevalent in the central region has lowest climate suitability for most species under both current and future climates (Suppl. material 1: fig. S13). The Western cluster had high climate suitability for Poa pratensis, Festuca rubra and Achillea millefolium under all climate scenarios, while the Eastern cluster was widespread with high climate suitability for many species under the current climate. However, under the future climate scenarios, the geographic extent of the Eastern cluster was further to the East (Svenskøya and around), and the climate suitability was low for most species, and the island of Edgeøya shifted from being grouped with the Eastern cluster under the current climate, to the Western cluster under future climate scenarios (Fig. 4).

Figure 4.

Potential non-native vascular plant species assemblage clusters across Svalbard under current and future climate scenarios. For all, the optimum number of clusters was three. These are illustrated using colours. The clustering was undertaken independently for each scenario, and the cluster groups have no direct correspondence between present and future climate scenarios. Species associations with clusters are shown in Suppl. material 1: fig. S11. The southern island of Bjørnøya is shown as an inset and ice-covered regions are excluded.

Biological invasions are a growing threat in polar ecosystems, where milder temperatures and increasing human activity heighten the risk for introduction and establishment of many globally invasive species (Duffy et al. 2017). The relative simplicity of communities in polar ecosystems increases the likely ecological impacts of establishing non-native species. So far Svalbard, in the high-Arctic, has largely escaped impacts of non-native vascular plants despite a relatively high number of species’ introductions. All established or identified door-knocker species are currently confined to disturbed soils in settlements or around cabins, and all are assessed to present low impact or no known impact (Table 1, Suppl. material 1: table S1). There is thus an opportunity for preventive biosecurity management measures to be applied. To facilitate this, we mapped the current and future macroclimatic niche potential of the non-native vascular plants on Svalbard. We found that many species have a wide potential climatic niche on Svalbard today, with a high number of potential niches in some regions of the archipelago beyond the current sites of occurrence. Under future climate change many more non-native vascular plant species have a broad potential climatic niche across Svalbard.

The finding that many of the non-native species considered here are inside their realised climate niche on Svalbard today, shows that factors other than macroclimate likely limits these species occurring more widely across the archipelago than their current distribution. Previous studies in Antarctica indicated that species like Poa annua and Poa pratensis were living at the farthest ends of their niche centroids (Pertierra et al. 2017); our results suggest that many non-native vascular plant species are not at their climate niche limits on Svalbard. In contrast, under current climatic conditions, Deschampsia cespitosa, Ranunculus subborealis subsp. villosus and Saussurea alpina have the widest potential distribution on Svalbard, while Deschampsia cespitosa, Alchemilla wichurae and Alchemilla subcrenata are the top ranked risk species with known invasion potential (Fig. 2, Suppl. material 1: table S1).

All species of non-native vascular plants on Svalbard have highly localised current distributions, on nutrient-enhanced and/or disturbed soils within the settlements and around some remote cabins. The introduced non-native plants on Svalbard may possibly still be under a lag phase of acclimatation-adaptation where their dispersal capabilities are limited. Lag phases are not uncommon in polar ecosystems; for example Pertierra et al. (2013) showed that Poa pratensis struggled to flower in Antarctica, while it flowers so late in the season in Svalbard that seed-production may not always be possible (Elven et al. 2020; Westergaard et al. 2023). In contrast, the discovery of colonies of Poa annua in Antarctica from the 2010s (Molina-Montenegro et al. 2014) suggests that P. annua has overcome a lag phase of 30–40 years, and without the strict biosecurity and eradications in place, it could be more firmly established and spreading in Antarctica (Pertierra et al. 2023). Understanding lag phases in establishment in Svalbard is even more important considering future climate scenarios, where 19 of the 27 considered species have climatic potential to spread to over 75% of the non-glaciated parts of the archipelago.

The potential richness of non-native plants on Svalbard was found to be greatest in Eastern regions such as the large island Edgeøya, as well as the island of Bjørnøya, far south of the main archipelago. These parts of Svalbard are remote and more rarely visited than the central fjord regions. The high number of non-native vascular plant species with potential climatic ranges here suggests that these regions may be at greater risk of establishment by several non-native plant species. Importantly, the potential number of non-natives escalate the risk of causing functional disturbances to native plant communities (Beaury et al. 2023). There is thus a need to increase vigilance measures in these regions. For example, they could potentially be designated as sanctuary regions with limited human visits and increased biosecurity requirements. Future climate change will increase the number of species that can potentially survive in the central regions of Spitsbergen around the largest settlements (which are also the main potential import sites), notably Longyearbyen and Barentsburg. Thus, the key risk areas under future climate overlap with zones under different land-use such as settlements, heavily visited tourism and research sites and protected areas. The difference in potential species richness was greater between current conditions and the moderate future scenario, than between the moderate and severe future scenarios, indicating that non-native vascular plant species are likely to be an increasing issue in the high-Arctic, regardless of the magnitude of climatic change.

Clustering of the non-native species potential distributions across Svalbard revealed assemblages separated on the longitudinal axis. This clustering reflects the Svalbard’s bioclimate sections (i.e. continentality-oceanity gradient representing oscillating temperature and precipitation) more closely than the bioclimatic zonation (temperature gradient representing thermal limits) (Elvebakk 1997). For some species (for example Trifolium repens and Ranunculus acris), precipitation variables were as, or more, important than temperature variables in the distribution models (Suppl. material 1: fig. S6). These results suggest that future precipitation patterns are also important in determining the distribution of non-native vascular plant species, and composition of non-native assemblages in the Arctic in the future. There are parallels here with importance of soil moisture, in addition to temperature, in driving shrub establishment in the tundra (Myers-Smith et al. 2015). It remains to be studied whether the clusters would also implicate different functional pressures (Beaury et al. 2023).

In this study we only included non-native species assessed for their ecological impact by the vascular plant committee for the Norwegian Alien Species List 2023 (Artsdatabanken 2023). We have not considered species which have never been observed or identified as potential door-knockers by the committee, which was expert-led and based on monitoring data and a literature-based horizon scan (Suppl. material 1: table S1). However, many other species could find suitable climatic niches on Svalbard, and a full, data-driven horizon scan is required. Such exercises have been carried out for lower latitude regions including both climate suitability and import potential (Bayón and Vilà 2019).

Our study focussed on macroclimatic niche potential. In harsh environments, the microclimatic conditions that plants experience can substantially differ from macroclimatic conditions. In many cases this can offer refugia for species originating from less harsh environments (Lembrechts et al. 2018), thus our results may actually underestimate potential occupancy. Furthermore, species may undergo climatic niche shifts in their non-native ranges (Bates and Bertelsmeier 2021). Climate is unlikely to be the only factor limiting species potential establishment. Other limiting factors like soil development and nutrients (Doetterl et al. 2022), soil disturbance and biotic interactions with native species will play a major role in determining establishment potential for many non-native species. In addition, wind and other dispersal vectors will shape the speed of niche filling within the archipelago from the established non-natives. While our study presents the potential climate niche of the species (based on the global realised niche), the realised niche on Svalbard is likely to be far smaller. Conversely, some species such as Achillea millefolium, Barbarea vulgaris, Poa palustris and Ranunculus acris are identified as having no suitable current climate on Svalbard, yet all exist and disperse locally within settlements, so here the Svalbard realised niche is broader than the potential climatic niche. Microclimate as well as other factors, such as soil nutrients, disturbance or enemy release may lie behind this. It should also be noted that Svalbard’s current climate differs from the current ranges of the species considered here, particularly in having a cooler summer temperature. Therefore, the models represent some degree of extrapolation in climatic space.

There are also taxonomic challenges in using global occurrence data to model regional distributions. Three of the most taxonomically complex groups in the northern hemisphere are included in this study. Firstly, Norwegian material of Poa pratensis s.lat. is considered to include five closely related species, all present in Svalbard: the native P. alpigena and P. colpodea, and the introduced P. angustifolia, P. humilis, and P. pratensis s.str. (Elven et al. 2022). Our data comprises Poa pratensis s.str. and our results show P. pratensis climate suitability is very low under current climate.

Secondly, Festuca rubra is not only one of the most widely distributed grasses of the northern hemisphere, but also among the taxonomically and morphologically most complicated of all northern grasses. Its main Arctic subspecies richardsonii is consistently accepted and is considered widely distributed in a wide range of habitats in Svalbard. Morphological characterization of the Svalbard material into two subspecies (the alien F. rubra subsp. rubra and the native F. rubra subsp. richardsonii) is, however, ambiguous, with potential transitional forms between the subspecies (Elven et al. 2020). The introduced F. rubra subsp. rubra is known from areas within the settlements and around old cabins, but has in recent years only been found in Longyearbyen, Pyramiden and Barentsburg. In addition, there are some morphologically less studied native populations in bird-cliffs along the west coast of Spitsbergen that may belong to F. rubra subsp. rubra. Our results suggest that both current and future Svalbard climates are suitable for F. rubra, and the species shows an affinity for the western clustered assemblage.

Thirdly, Taraxacum sect. Ruderalia (or T. officinale aggregate; Elven et al. 2020) is a large and taxonomically difficult section of dandelions consisting of cosmopolitan weeds and many apomictic microspecies. In mainland Norway alone, more than 150–200 microspecies have been named, and there have not been any attempts to identify the Svalbard plants to microspecies. Occurrence records were therefore retrieved as T. ruderalia (GBIF synonym for T. sect. Ruderalia) to include data from the accepted section according to the Norwegian Nomenclature Database, to avoid including irrelevant occurrences, but with an unavoidable caveat of missing some relevant occurrences. Our results suggest low current climate suitability for Taraxacum sect. Ruderalia, but this will increase in the future.

In summary, the use of large datasets of species observations must accept the accompanying taxonomic uncertainty for individual data points. The precautionary principle suggests that leaning towards inclusivity of records is best, particularly where the focus is placed on aggregated results across species, such as this study. In many of the species, the limited number of current occurrences indicates that only a few lineages have established and thus their potential for adaptation to the high-Arctic can be limited by their genetic diversity. Hence our models could be over-predicting the potential distribution beyond the existing occurrences (Maclean and Early 2023). However, there is a high risk for repeated introductions in the future, thus the models provide a precautionary estimate of potential distributions. Even so, processes of adaptation to novel conditions with niche shifts are also documented for non-native plants (Bates and Bertelsmeier 2021; Häkkinen et al. 2022). Therefore, our models might in some cases even under-predict potential distributions.

Risk assessments of non-native species often focus on aspects of invasion potential and ecological impacts. This study estimates the potential area of occupancy of non-native species based on their realised niche, complementing quantitative ecological impact assessment (Sandvik et al. 2019; Artsdatabanken 2023). It also complements systematic, ongoing field monitoring of non-native vascular plant species (Ravolainen et al. 2019; Bartlett et al. 2021). Our results identify species that could be prioritised for early detection of local dispersal and rapid management response. However, it is clear that mechanistic studies, considering source pools of species, introduction pathways, establishment, and the ecological impact of single and multiple non-native species, are required to address the potential ecological impact of non-native plant species on Svalbard.

The terrestrial Arctic has been fortunate to escape widespread impacts from non-native vascular plants. However, climate change will exacerbate potential establishment whilst increased human activity will favour dispersal and wider establishment opportunities at disturbed regimes. Our study maps the potential climate niche of non-native vascular plant species across Svalbard, finding that climate does not prevent the distribution of many species today, and will prevent the distribution of even fewer in the future. Environmental management on Svalbard has a window of opportunity to take proactive steps to minimise new introductions and establishment of non-native plants and avoid further dispersal of already established non-native species.

We are grateful to the reviewers whose constructive comments on an earlier draft helped to improve this work, and to all those who have contributed the species occurrence data that was used in this study.