Early detection of hitchhiking pests requires the identification of strategic introduction points via transport. We propose a framework for achieving this in Europe using the Japanese beetle (Popillia japonica) as a case study. Human-mediated spread has been responsible for its introduction into several continents over the last century, including a recent introduction in continental Europe, where it is now listed as a priority pest. Furthermore, recent interceptions far from the infested area confirm the risk of unintentional transport within continental Europe. Here, we analysed how three modes of transport - air, rail and road - connect the infested area to the rest of Europe. We ranked all European regions from most to least reachable from the infested area. We identified border regions and distant major cities that are readily reachable and observed differences between modes. We propose a composite reachability index combining the three transport modes, which provides a valuable tool for designing a continental surveillance strategy and prioritising highly reachable regions, as demonstrated by recent interceptions.

Biological invasion, hitchhiking pest, likelihood of introduction, pest risk assessment, Popillia japonica, surveillance, transport network

The increasing global movement of goods and people provides countless opportunities for species to move around the world outside of their natural range, increasing the rate of biological invasion (Hulme 2009; Blackburn et al. 2011). Insects in particular can hitchhike on a variety of modes of transport, including planes, ships, trains and trucks (Saccaggi et al. 2016; Turner et al. 2021). Hitchhiking, which is part of unintentional human-mediated transport, facilitates the introduction of insects into new regions and, where conditions are suitable, their subsequent establishment (Early et al. 2016; Rosace et al. 2023). Predicting the risk of introduction of invasive insect pests through transport is therefore crucial for developing effective surveillance strategies (Essl et al. 2011).

Risk assessments of Invasive Alien Species (IAS) are becoming increasingly quantitative, particularly with the advent of environmental distribution models used to estimate suitability and hence establishment risk (Venette 2015). However, IAS introduction risk analyses are still often qualitative and expert-based, especially when considering human-mediated transport rather than active spread (accounted for by diffusion models) (Hulme 2009). Human-mediated introduction can occur by transport of traded goods and hitchhiking on cargo, on passengers or in vehicles. To be relevant to surveillance strategies for hitchhiking invasive insects, the analysis of risk of introduction should include directional (from infested regions to other regions) and weighted (volume measure) transport data, especially passenger travel (Hulme 2021). Most studies that use directional and weighted data focus on the trade of specific commodities that are pathways of entry for IAS (Piel et al. 2008; Yemshanov et al. 2012; Meurisse et al. 2019; Jamieson et al. 2022). Few studies include passenger movement data, and these studies usually focus on a specific region (Perry and Vice 2009; Szyniszewska et al. 2016) and rarely extend at a continental scale (Frem et al. 2020). As there is no framework for predicting the risk of hitchhiker IAS spreading across Europe through anthropogenic dispersal, we propose a method using directional and weighted transport data, including passenger movements, on a continental scale and at the finest possible spatial resolution. We use the Japanese beetle as a case study.

The Japanese beetle (Popillia japonica) is a prime example of a hitchhiker pest. Native to Japan, it was accidentally introduced into the United States of America at the beginning of the last century, causing a major invasion that still persists today (Frank 2016). From there, it was introduced to the Azores archipelago in the 1970s, and more recently to continental Europe. After its first detection in Italy in 2014, the beetle has spread to an area of more than 16,500 km² covering parts of northern Italy and southern Switzerland (Gotta et al. 2023). Due to its potential impact on the environment, food safety and economic balances, it has been listed as a priority pest by EU authorities (Commission Delegated Regulation (EU) 2019). Furthermore, several interceptions of the beetle far from the infested area have raised concerns about possible introductions to other parts of Europe. Although the source of the beetle’s introduction into Europe has been phylogenetically reconstructed (Strangi et al. 2023), its potential spread in Europe and the risk of introduction through human movement have not been investigated.

In this paper, we present a novel approach to map the potential human-mediated spread of the beetle from the infested area to the rest of Europe. We considered three transport networks - air, rail and road - that are relevant to the beetle’s pathways of entry from the infested area. We examined how reachability, i.e. the likelihood of introduction from the infested area, varied according to mode of transport. Finally, we combined transport modes to identify the most likely points of introduction and used interception sites to assess our reachability map.

Data processing and analyses were performed using R version 4.2.1 (R Core Team 2021).

2022 European infested area

We first assessed the extent of the European infested area, taking into account both the municipalities where the presence of the beetle was confirmed and the neighbouring municipalities included in the buffer zone, according to 2022 official reports (Poggi et al. 2022a). This area will be considered as the origin for human-mediated spread of the Japanese beetle in Europe. The 2022 infested area covers approximately 16,500 km², spanning over five regions and more than 1900 municipalities in northern Italy and southern Switzerland (in black on Fig. 1). Six interceptions of Popillia japonica, i.e. captures of an isolated adult without establishment of a population, were made in Europe since 2018, far away from the infested area: one in the Netherlands (Amsterdam, EPPO 2019), one in Germany (Freiburg, EPPO 2022a), two in Italy (Udine, Bassi et al. (2022) & Cagliari, EPPO (2022b)) and two in Switzerland (Basel, NPPO of Switzerland (2021) & Zurich, EPPO (2023)).

Figure 1.

Reachability of Europe for Popillia japonica from the infested area (in black), by air, rail, road transport, and a combination of modes (composite index). Quantile-classified reachability maps showing: A the number of passengers arriving at airports B the number of trains arriving at stations, and C the number of trucks per square kilometre reaching NUTS 3 regions, departing from the infested area. Darker colours correspond to higher reachability D composite reachability map, i.e. risk of introduction for NUTS 3 regions ordered by Pareto fronts from most to least reachable. Warmer colours correspond to higher reachability.

Among the 1675 European NUTS3 regions, twenty were considered infested in 2022 because they contained at least one infested municipality. Within this infested area, there are 6 airports and 540 railway stations. Outside of that area, a total of 160 airports (from 30 different countries), 422 railway stations (located in 5 different countries), and 1446 NUTS 3 regions (from 33 countries) can be reached by planes, trains, and trucks, respectively. Reachability from the infested area varies between modes of transport (Fig. 1A–C). With the exception of a few distant major European cities, the rail transport network mainly connects areas adjacent to the infested area (7 NUTS 3 in Germany, 6 in Austria and 4 in France), most of Italy (65 NUTS 3 out of 107) and Switzerland (13 NUTS 3 out of 26). The road freight network is both local, with Switzerland, northern and central Italy highly reachable from the infested area; and international, with many major European cities also highly reachable. Finally, the air transport network is mostly international, with direct access to all major cities in Europe. Our analysis shows that of all NUTS 3 regions directly reachable by at least one transport mode from the infested area, 10% and 7% are reachable by air and rail respectively. On the other hand, almost all of these regions (99.8%) are reachable by road freight.

Interestingly, the distribution of planes, trains and trucks that reach NUTS 3 regions in Europe is far from uniform, with very few NUTS 3 concentrating most of the traffic from the infested area. Indeed, the 1% of NUTS 3 most reachable by rail (14 NUTS 3) account for over 60% of all trains leaving the infested area. Similarly, the top 1% of NUTS 3 reachable by air account for 52% of all flights, and the same is true for road freight, with the top 1% of NUTS 3 reachable by trucks accounting for 46% of all trucks leaving the infested area.

The composite reachability index, which combines air, rail and road transport, ranks NUTS 3 regions into ordered groups, from most to least reachable (Fig. 1D). The five groups of most reachable regions contain 62 NUTS 3 (from 16 different countries), of which 13 are reachable by all three modes of transport. Furthermore, these 62 NUTS 3 account for 70%, 72%, and 47% of the total number of planes, trains and trucks leaving the infested area, respectively.

The distribution of composite reachability by number of NUTS 3 per country is shown in Fig. 2: Italy (IT), Germany (DE), United Kingdom (UK), France (FR), Switzerland (CH), and Spain (ES) contribute with both a high number of reachable NUTS 3 and their relative importance in terms of the composite reachability index. On the other hand, eastern countries like Hungary (HU), Romania (RO) and the Czech Republic (CZ) have fewer, but highly reachable NUTS 3. The large number of small NUTS 3 regions in Germany explains the over-representation of this country in Fig. 2.

Figure 2.

Distribution of reachable NUTS 3 regions per country when combining air, rail and road transport (composite reachability). Warmer colours correspond to higher reachability. Countries are shown using alpha-2 country ISO codes as described in the ISO 3166 international standard.

Finally, reachability correlates negatively with distance from the infested area for train, trucks and the composite index (Kendall correlation of -0.25***, -0.44*** and -0.32***, respectively), which means that more distant destinations are less reachable than closer destinations (Fig. 3). On the contrary, reachability by flight correlates positively with distance (Kendall correlation of 0.11***). Interestingly, reachability is anisotropic, meaning that it does not distribute uniformly in all direction (see the circular bar plots within the four panels of Fig. 3).

Figure 3.

Distribution of air, rail, road and composite reachability of European regions for Popillia japonica as a function of distance from the infested area (in km), with the corresponding value of the Kendall correlation test. Within each panel, a circular bar graph shows the main directions in which the four reachability indices spread with respect to the infested area. Air, rail and road reachability are expressed as the number of passengers arriving at airports, the number of trains arriving at stations, and the number of trucks per square kilometre reaching NUTS 3 regions, departing from the infested area, respectively. Composite reachability of NUTS 3 regions is displayed from most reachable (group 1) to least reachable (groups 128-1225).

In this study, we have mapped the risk of introduction of the Japanese beetle in continental Europe by air, rail and road transport from the infested area as defined in 2022. We found that reachability of regions varies by mode, and detected topological features of transport networks, ranging from a local and national predominance (rail and road transport) to an almost exclusively international dimension (air transport) (Banks et al. 2015; Tatem 2017). The proposed composite reachability index, which combines the three transport modes, highlights a few scattered highly-reachable major cities across Europe, as well as a cluster of high reachability comprising many regions of Italy, Germany, Switzerland and France surrounding the infested area.

As this is the first analysis of the risk of Japanese beetle spread through human-mediated transport across continental Europe, our identification of likely introduction points cannot be compared with previous results. Nevertheless, the BeNeLux countries and northern Italy have also been identified as presenting a high risk of IAS introduction into Europe by a previous study that examined risk as a function of climate, soil, water, and anthropogenic factors (Schneider et al. 2021). Our results highlight transport network characteristics that have also been observed in previous studies, such as the international nature of air transport and the predominantly regional and national nature of rail and road freight transport (Hulme 2021). In addition, we have identified a disproportionate distribution of connections between the infested area and a small number of distinguished NUTS 3: indeed, for each of the three modes of transport, the 1% of the most reachable NUTS 3 accounts for more than 45% of the total outbound flow from the infested area. This demonstrates an assortative type of mixing (Newman 2003), a property of complex networks where nodes (here NUTS 3 regions) that are in some sense similar tend to be more connected. In our case, the infested area encompasses a major European hub (MXP-Milan Malpensa Airport), which is naturally well connected to other hubs on the continent. On the other hand, Figs 1, 2 also show that the infested area is simultaneously connected to scattered, highly reachable hubs and is capillary-linked to surrounding, less reachable regions. This feature is typical of core-periphery graphs, where a subset of nodes in the network (the core) is connected to a few nodes of very high degree, as well as to many peripheric nodes (Pittel et al. 1996; Malliaros et al. 2019).

The highly-reachable hubs identified by combining air, rail and road transport, have already been shown to have particular potential for the spread of IAS (Banks et al. 2015). Interestingly, all Japanese beetle interceptions that have occurred in Europe over the last 5 years have been reported in regions that our analysis identified as highly reachable (Fig. 4), providing preliminary evidence of the robustness of our approach. Furthermore, an outbreak was detected in July 2023 in one of the nine regions we identified as being the most reachable from the infested area (Zurich region, Switzerland). Of all the interceptions known to date, only those in Zurich and Amsterdam could have originated from outside the continental European infested area, as they occurred at or near airports with direct connections to North America and/or Japan. Molecular methods could be used to further assess the most likely origin of these introductions (Strangi et al. 2023).

Figure 4.

Interceptions of P. japonica made in Europe since 2018 and their position in relation to the distribution of reachability indices (air, rail, road, and composite). The number in brackets to the right of the interception site name indicates the group number assigned to the NUTS 3 region by the Pareto ranking method, from most reachable (1) to least reachable (16).

Although our results appear to be relevant based on interceptions and the published literature on transport networks, this analysis could be improved by considering hitchhiking on air freight, rail freight and private cars (domestic travel). Call detail record (CDR) could be a useful source for domestic travel, which could play an important role in facilitating the spread of the Japanese beetle, especially around the infested area (Tatem 2017).

The proposed framework provides a rapid response tool for decision-makers and phytosanitary services to anticipate the likelihood of hitchhiking pest introduction on a continental scale. Informing risk-based surveillance strategies with likelihood of introduction can significantly reduce surveillance efforts and promote early detection of invasive species (Parnell et al. 2014). As data become available, further improvements may be achieved, for example by targeting commodity movements specifically identified as pest carriers (Fenn-Moltu et al. 2023), or by including other transport modes. Highly reachable regions could also be surveyed for the presence of susceptible host plants or favourable environmental conditions (Tatem et al. 2006; Borner et al. 2023). Our framework highlights the need for local surveillance combined with a transboundary strategy, involving official authorities and stakeholders, and adapted to the scale and means of spread of the pest under surveillance (Radici et al. 2023).