Non-native insects that become invasive in their introduced range are amongst the leading drivers of forest and urban-tree decline worldwide (Brockerhoff and Liebhold 2017). Wood-boring beetles (Coleoptera) are particularly damaging because some species can attack living trees, disrupt timber and fruit production and generate large management costs due to severe ecological and economic impacts when established outside their native range (Lovett et al. 2016; Kenis et al. 2017; Linnakoski and Forbes 2019; Haubrock et al. 2021). The increase in container shipping and movement of wood-packaging material since 1980 have resulted in a corresponding increase in detection of wood-boring insects in recent decades (Aukema et al. 2010). Cryptic invaders that include several longhorn beetles (Cerambycidae), jewel beetles (Buprestidae) and bark and ambrosia beetles (Curculionidae, Scolytinae), have repeatedly been transported outside their native ranges through global trade, often in wood or wood-packaging material (Haack 2006; Lantschner et al. 2020; Ruzzier et al. 2023a). Such repeated introductions facilitate successful establishments (Brockerhoff et al. 2014), underscoring the central role of international trade in facilitating biological invasions and the importance of stringent phytosanitary measures, early detection and rapid response strategies to mitigate ecological and economic impacts (Brockerhoff and Liebhold 2017; Epanchin-Niell et al. 2021; Wray et al. 2024). Preventing the arrival of non-native insects and their early detection are significantly more cost-effective than post-establishment control, making risk-based tools that predict where an organism could survive and reproduce outside its native range essential to plant health planning (Epanchin-Niell et al. 2014; Blaalid et al. 2021).

Species distribution models (SDMs) are used to identify the relationship between species and environmental conditions (Guisan et al. 2017). They can serve as useful tools to produce habitat- and climate-suitability maps for invasive taxa, including Cerambycidae (e.g. Shatz et al. (2016); Krishnankutty et al. (2020); Zhao et al. (2023a, 2023b); Kim et al. (2025)). SDMs can, therefore, provide information for surveillance prioritisation, pathway risk assessments and scenario testing under climate change. When combined with information on trade pathways, host availability and dispersal, they support targeted inspections, quarantine zoning and contingency planning (e.g. Soubeyrand et al. (2024)). However, SDMs must be applied with caution when modelling invasive species. Sampling bias and georeferencing accuracy in occurrence data, the allocation of pseudo-absence/background points, covariate selection, the omission of biotic interactions and dispersal limitations and uncertainty in climate scenarios can all hinder model transferability, leading to an over- or underestimation of areas potentially susceptible to invasion (e.g. Yates et al. (2018); Rousseau and Betts (2022); Valavi et al. (2022); Adde et al. (2023)).

The spatial heterogeneity of climate, host availability and the distribution of trade hubs across Europe creates a complex mosaic of sites with variable invasion risks for non-native cerambycid beetles (e.g. Roques et al. (2020); Rosace et al. (2025)). As a consequence, multiple non-native cerambycids potentially threaten European forestry, urban trees and fruit production (e.g. Malumphy et al. (2020); Lupi et al. (2023); Horrocks et al. (2024); Sarto i Monteys et al. (2024)). Moreover, Europe continues to experience repeated incursions of longhorn beetles, representing a significant risk for further establishment and invasion (e.g. Cocquempot and Lindelöw (2010); Keszthelyi et al. (2019); Ruzzier et al. (2020); Seidel et al. (2021); Ruzzier et al. (2023b)). These realities make integrative risk assessment, relying on SDMs, pathway analysis and surveillance data, critical to international and national plant-health strategies.

Recent detections in Europe of two North American cerambycid beetles, the round-headed apple-tree borer, Saperda candida Fabricius, 1787 and the elm borer, Saperda tridentata Olivier, 1795 (Lamiinae, Saperdini), have placed these species under European surveillance due to their potential threat to plant health on the continent (EFSA PLH Panel et al. 2020; EPPO 2021). Saperda candida is native to much of eastern North America, ranging from south-eastern Canada west to central Alberta and south through many U.S. States (Bousquet et al. 2013; Monné and Nearns 2023). Its primary hosts are fruit and ornamental trees in the family Rosaceae, including Malus, Pyrus, Crataegus and Amelanchier (Solomon 1995). It has also been reported from Betula sp. (Betulaceae) once (Vlasak and Vlasakova 2002), but reports of Cotoneaster (Hess, 1940), Prunus (Becker, 1918) and Pyracantha (Linsley & Chemsak, 1995) as larval hosts have not been substantiated (Maier 2020). Infestations cause extensive tunnelling in trunks and roots, leading to branch dieback and even tree mortality in orchards and nurseries (EPPO 2021). In Europe, S. candida has already been detected in the wild, with the first confirmed record in Germany in 2008 (DEFRA 2025; Julius Kühn-Institut 2025). Since then, active eradication measures have been implemented to prevent further establishment and spread (Julius Kühn-Institut 2025). The detection of the species in natural environments highlights its ability to survive outside its native range and indicates a potential risk to fruit crops, ornamental trees and natural vegetation (EPPO 2021).

Saperda tridentata occurs across eastern North America, with records from the United States and southern Canada (Bousquet et al. 2013; Monné and Nearns 2023). It is primarily associated with elm species (e.g. Ulmus americana, U. rubra, U. crassifolia) (Solomon 1995). Early isolated records reporting Acer (Harrington 1883) and Populus (Washburn 1910) as larval host plants are considered doubtful and their potential role as substrates for larval feeding and development is currently under investigation (Santoiemma et al., personal communication). Damage results from larval feeding beneath the bark, causing bark splitting, frass exudation, canopy thinning and progressive branch decline (EFSA PLH Panel et al. 2020). In contrast to S. candida, S. tridentata does not have known adventive populations in Europe, despite repeated interceptions in imported Ulmus logs from North America (Binazzi et al. 2019). These interceptions underscore the ongoing risk of accidental introduction and establishment through international trade in wood and plant material. The repeated detection of S. tridentata in imported materials demonstrates the species’ potential for entry and emphasises the need for continued surveillance and early warning systems to prevent establishment and mitigate future impacts. Both species pose a potential invasion risk to Europe through trade in infested wood or live plants, prompting their inclusion in European surveillance and pest risk analyses (EFSA 2020; EPPO 2021). Given their biology with cryptic larval stages inside trunks/roots, multiple suitable hosts, the likelihood of detection is low until damage becomes noticeable.

The present study aimed to assess the invasion risk of S. candida and S. tridentata in Europe. We hypothesised that Europe provides climatic and habitat conditions similar to those of the native ranges of both species, potentially allowing their establishment. However, invasion risk was expected to increase with the intensity of commercial trade and transport, for which maritime and inland connectivity provide suitable proxies. We tested these hypotheses by: (1) developing, for both species, a data-driven joint bioclimatic–habitat species distribution model (JBH-SDM), based on two complementary SDMs: a bioclimatic suitability model (BSM) and a habitat suitability model (HSM), both developed using an ensemble framework that explicitly accounted for sampling bias and the georeferencing accuracy of occurrence records; and (2) integrating JBH-SDM predictions with maritime and inland connectivity linking the species’ native ranges to potentially invadable (i.e. environmentally suitable) areas in Europe.

The modelling framework we employed in this study showed strong overall performance, delivering consistent and robust predictions for both Saperda species across their native ranges as well as for projections in Europe. Forecasting invasion risk is key to effectively managing invasive species (Thomas et al. 2024), although limitations and challenges of SDMs have been questioned (Barbet-Massin et al. 2018; Lee-Yaw et al. 2022; Fenollosa et al. 2025). Our modelling framework allowed us to evaluate separately the effect of scenopoetic variables and local variables on the ecological niche of the investigated species, thereby avoiding potential confounding effects in reconstructing the two niche components (bioclimatic and habitat-related) that may arise when jointly considering variables describing processes operating at different spatial scales, as previously highlighted by Adde et al. (2023) and Ruzzier et al. (2024, 2025a, 2025b). This modelling approach is particularly important when estimating the actual climatic and habitat suitability in areas that are potentially invadable and differ from those used for model calibration (i.e. the native range).

Moreover, a particularly important aspect concerns modelling, based exclusively on presence data within the species’ native range; such occurrences can be assumed to reflect a species distribution of a population that is in equilibrium with the environment as a whole (Hui 2023). Consequently, this allows for the estimation of the realized niche occupied by the species within its native range and the identification of areas outside this range where analogous environmental conditions occur. In addition, for species whose biology is closely associated with a specific set of host plants, including these variables in habitat modelling further improved the ability to identify and refine suitability at a local scale.

While the joint bioclimatic–habitat models identified extensive areas of environmental suitability across Europe, the invasion-risk maps revealed a more spatially structured pattern. Rather than being uniformly distributed across all environmentally suitable regions, the highest invasion risk became concentrated within a limited number of geographically coherent hotspots where suitable environmental conditions overlapped with high values of transport-based proxies of propagule pressure. This framework highlights the added value of integrating environmental suitability with these proxies, allowing environmentally suitable areas to be distinguished from those that also exhibit greater potential for introduction and subsequent spread.

The modelled native range of S. candida primarily encompasses eastern North America, extending from the south-eastern and central-eastern United States to the Midwest and north-eastern regions, with partial inclusion of adjacent areas in southern Canada. Bioclimatic models indicate a strong association with humid temperate climates, characterised by low precipitation seasonality and relatively stable conditions throughout the year. In particular, S. candida is closely associated with stable precipitation regimes and high moisture availability during the warmest quarter. These results are consistent with suitability assessments reported by EPPO (2021). From a habitat perspective, the species is strongly linked to the availability of host plants, which represents the primary constraint on its distribution. It is narrowly associated with host-dominated forest habitats, avoiding cultivated and managed areas lacking host plants, as well as coniferous forests. Overall, its native range corresponds predominantly to mesic temperate forests of eastern North America, where suitable climatic conditions and host availability co-occur. When projected to Europe, S. candida shows a wide extent of suitable and relatively continuous areas, largely driven by the distribution of host plants and associated land-cover types. However, bioclimatic projections alone indicate a more restricted suitability, suggesting that climate may act as a filtering factor. In particular, the species appears more strongly associated with the moister climates of central Europe, characterised by precipitation distributed throughout the year.

The modelled native range of S. tridentata largely overlaps with that of S. candida, encompassing eastern North America from the south-eastern and central-eastern United States to the Midwest and north-eastern regions, with some extension into southern Canada. Similarly, bioclimatic models indicate an association with humid temperate climates and low precipitation seasonality. However, S. tridentata shows a greater association with temperature conditions, occurring across temperate to warm environments during the warmest quarter, suggesting a stronger influence of temperature gradients in shaping its climatic niche. These findings are also consistent with EFSA PLH Panel et al. (2020). As for habitat, the species remains dependent on host plant availability, but exhibits a broader ecological tolerance compared to S. candida, occurring in mixed forests and, to a limited extent, in cultivated and managed areas, with a preference for low-elevation environments. As with S. candida, its native distribution reflects the overlap between suitable climate and host availability within mesic temperate forests. When projected to Europe, S. tridentata also shows extensive areas of potential suitability driven by host plant distribution. However, bioclimatic projections indicate a more constrained pattern, highlighting the role of climate as a limiting factor. In Europe, the species appears associated with milder and more temperate conditions, such as those found in central Italy. Nevertheless, the potential host plant range still requires further investigation, as precise information on host use is currently lacking. Indeed, Ulmus is currently the only confirmed host genus, whereas the inclusion of Acer and Populus is based on sporadic occurrence records rather than confirmed host associations (Washburn 1910; EFSA 2020).

Overall, for both species, habitat availability emerges as a primary environmental filter within the native range, while climate plays a stronger constraining role in shaping potential areas of establishment under a European invasion scenario. This pattern is consistent with findings for other invasive longhorn beetles, such as Aromia bungii (Faldermann, 1835) (Cerambycinae, Callichromatini) and Psacothea hilaris (Pascoe, 1857) (Lamiinae, Monochamini), for which host plant availability represents a key determinant of distribution (Ruzzier et al. 2024, 2025a). Moreover, habitat may be even more limiting; indeed, habitat suitability may be slightly overestimated due to the inclusion in the HSM of host plants whose use by the two Saperda species remains to be verified.

Considering the results presented in this study, the current presence of S. candida on the island of Fehmarn (northern Germany) represents a clear phytosanitary concern. This situation is consistent with previous assessments by competent authorities and fully supports the inclusion of the species in the European Union A1 Quarantine Pest list (Annex II A). Our modelling results provide quantitative and spatially explicit evidence that reinforces the high-risk status attributed to this species.

The invasion-risk map identifies a broad high-risk corridor extending across western and central Europe, including much of Germany and neighbouring countries. Several of these high-risk areas occur in close proximity to the current outbreak on Fehmarn, potentially increasing the likelihood of further spread if containment and eradication measures prove unsuccessful. The continuity of environmentally suitable habitats, together with the high connectivity characterising these regions, suggests that S. candida could establish and persist over extensive areas of continental Europe if propagules disperse from the current infestation site, either naturally or through human-mediated movement of infested materials. Propagule pressure is, in fact, a key factor influencing the establishment and spread of non-native species (Brockerhoff et al. 2014).

However, the comparatively moderate suitability predicted for the island of Fehmarn suggests that containment and eradication measures could be successfully implemented. The predominance of agricultural land use reduces habitat continuity and the availability of host plants required to sustain populations capable of generating high propagule pressure. Moreover, the local bioclimatic conditions, although not entirely unfavourable, are less suitable than those predicted for the neighbouring continental high-risk areas. Together, these factors indicate that the currently established population likely occurs under ecologically marginal conditions, where both long-term persistence and spatial expansion are constrained. Under such circumstances, the timely implementation of targeted phytosanitary measures could substantially increase the likelihood of complete eradication.

The elevated invasion risk predicted for S. tridentata across western and central Europe, together with the sustained and well-documented propagule pressure resulting from repeated interceptions (Binazzi et al. 2019), indicates that continued vigilance is necessary to prevent its establishment in Europe. In particular, the predicted high-risk corridor extending from France through Belgium, the Netherlands and Luxembourg to Germany, Switzerland and northern Italy, together with secondary hotspots in eastern England and along the northern Adriatic coast, identifies priority areas where surveillance and early detection efforts should be concentrated. For both species, the comparison between overall suitability and invasion-risk maps demonstrates that environmental suitability alone does not fully explain the geography of invasion risk. Although environmentally suitable conditions occur across extensive portions of Europe, invasion risk is highest in hotspots where favourable environmental conditions coincide with intense transport connectivity. Both species share a broad high-risk core across western and central Europe, but differ in the extent and distribution of secondary hotspots. In particular, S. candida exhibits more extensive high-risk areas across Great Britain, southern Sweden and the northern Adriatic Region, whereas S. tridentata remains more strongly concentrated within the central European corridor. These differences indicate that species-specific environmental responses continue to shape the spatial pattern of invasion risk despite both species sharing similar introduction pathways.

We acknowledge that the outputs outlined here represent a potential invasion scenario that does not account for possible limiting factors within European ecosystems that could counteract the invasion process. Amongst these, biotic resistance may play a key role. For instance, the presence of native congeners in invaded areas may limit the establishment and spread of non-native species, as natural enemies associated with native congeners could shift hosts and regulate populations of introduced relatives, preventing rapid spread or high population densities (Cavender-Bares et al. 2009; Dearborn et al. 2016). In Europe, nine species and three subspecies of the genus Saperda are currently recorded. If S. candida and S. tridentata were to become sympatric with native Saperda spp., interactions with the resident community could potentially reduce their invasiveness. Similar dynamics have been observed following the establishment of the European brown spruce longhorned beetle, Tetropium fuscum (Fabricius, 1787), in eastern Canada, where native natural enemies contributed to limiting its population growth (Dearborn et al. 2016; Heustis et al. 2018).

The risk maps produced in this study indicate a substantial potential for both Saperda species to establish and spread across Europe, although invasion risk is far from uniform. The highest-risk areas are concentrated within a broad western and central European corridor, where favourable environmental conditions, widespread host availability and intense maritime and terrestrial connectivity converge. Additional regional hotspots, particularly in Great Britain, the northern Adriatic Region and, for S. candida, southern Sweden, further demonstrate that invasion risk extends beyond the main continental core. Conversely, large portions of northern, eastern and southern Europe remain characterised by comparatively low invasion risk, reflecting lower environmental suitability, weaker transport connectivity or a combination of both.

Taken together, these results demonstrate that, although Europe offers extensive environmentally suitable areas for both S. candida and S. tridentata, environmental suitability alone does not fully explain the geography of invasion risk. The integration of transport connectivity concentrates the highest-risk areas into a limited number of geographically coherent hotspots, providing a more realistic representation of where introduction and establishment are most likely to occur. This finding underscores the limitations of suitability-based projections alone and highlights the added value of integrated risk frameworks that explicitly incorporate dispersal pathways. Such approaches provide a more robust and operational basis for prioritising surveillance, early detection and preventative phytosanitary measures, ultimately supporting more effective management of invasive wood-boring insects.