Model Maps How Fast the Japanese Beetle Could Sweep Across Europe

An invasive beetle that slipped into northern Italy in 2014 and has since overrun more than 15,000 square kilometers of northern Italy and Switzerland now has a detailed mathematical portrait of its own advance. Researchers have built a process-based model of the Japanese beetle, Popillia japonica, that reproduces the pest’s spread from its first years of detection through December 2023 and projects how fast the invasion front could move if new outbreaks establish elsewhere in Europe. The work, published in Ecological Processes, was designed to give phytosanitary services a quantitative basis for drawing the boundaries of delimitation and containment zones.

Popillia japonica is native to Japan and reached North America in 1916, where it now occupies most of the eastern United States and south-eastern Canada. After establishing in the Azores in the 1970s, it was detected in continental Europe for the first time in 2014 in northern Italy. European authorities classify it as a quarantine priority pest and rank it as the second most dangerous pest for crops in the European Union. Scenario analyses cited by the authors estimate that, if the beetle continues to spread unhindered without unified intervention, continental crop damage could cost between 30 million and 7.8 billion euros per year, with viticulture and arboriculture absorbing the largest share because of their high market value and the beetle’s feeding preferences.

The new model divides the beetle’s life cycle into two stages with very different roles in the invasion. Adults, which emerge once a year at temperate latitudes, drive the advance of the invasion front during the summer flight season through active flight and short-distance human-mediated transport. The buried ovo-larval stages, treated as a single overwintering phase, do not move the front but determine whether the population persists, because eggs and larvae are highly sensitive to environmental conditions such as moisture and temperature. The researchers coupled a continuous reaction-diffusion equation describing adult dispersal and mortality with a discrete Beverton-Holt model capturing the survival and competition of overwintering immatures, then linked the simulated densities to binary presence or absence records through an observation model.

Two environmental variables anchor the biology in real geography. Adult flight, which the literature establishes as occurring only in sunny, warm conditions above 21 degrees Celsius and increasing in speed with temperature up to about 35 degrees, is represented through a spatially variable diffusion coefficient scaled by a summer days index: the average number of days per year on which peak temperature exceeds 25 degrees Celsius, computed from climate data for 1979 to 2018 and normalized against an upper-bound value of 150 days for Italy and Switzerland. Immature survival, in turn, is tied to an environmental suitability map produced by a species distribution model calibrated with presence data from long-invaded areas, projected onto a 4-kilometer grid. In the study area, suitability reaches a maximum of only 0.55, spanning four suitability classes.

The simulation starts in 2016, the first year for which observational data are available, initialized from trap captures in the Piedmont region, where intense trapping was deployed after the first discovery in an attempt to eradicate beetles from the initially hit areas. Because no information existed for the Lombardy portion of the outbreak, initial densities were mirrored across the Ticino River using symmetric Piedmont data, restricted to municipalities declared infested in 2015. The model then runs through eight annual adult-larva cycles up to 2023, and an optimization algorithm searches the parameter space by maximizing the likelihood of the observed municipal infestation statuses, which regional Phytosanitary Services classify each year as infested, buffer, or pest-free.

Validation against the official municipal records yields an accuracy of 0.926, a sensitivity of 0.787, a specificity of 0.974, and an F-score of 0.846. The model tends to underestimate presence along the invasion front in the early years from 2016 to 2020 and to overestimate it along the northern and north-eastern front in the final years from 2021 to 2023. Simulated densities grow quickly after first detection, reaching roughly 75 percent of the asymptotic reference value within about three to four years, with the highest average densities at the core of the outbreak and declining values toward the front.

The fitted relationship between suitability and immature survival takes a clean sigmoid shape. The curve projects essentially no survival of eggs and larvae below a suitability of 0.1, rises steeply across the interval from 0.1 to 0.3, and reaches complete survival near a suitability of 0.3. Notably, this curve tracks the suitability classification proposed in earlier work without being told to: survival is lowest, at or near zero, in the S0 class predicted to be extremely unlikely to host the beetle, highest in classes S2 and S3, and highly variable across the intermediate S1 class, the zone of greatest uncertainty. In S1 areas, establishment remains possible but not guaranteed, leaving populations vulnerable to adverse weather or management measures in any given year.

Geographically, the results confirm that the ongoing outbreak is unfolding in one of the most favorable zones for the beetle, the Po Valley, whose warm summers and high suitability sustain rapid spread and dense populations. Mountain ranges act as barriers of very different strength. The Alps, apart from their valleys, emerge as an almost insurmountable defense, with no spread expected except in a few valleys at the Swiss border, while the Apennines only slow the southward advance without stopping it. Within the outbreak area, the estimated invasion speed ranges from zero to approximately 17.2 kilometers per year, consistent with earlier estimates of 4.5 to 16 kilometers per year for this species, and the observed wavefront has advanced at a roughly constant speed of about 10 kilometers per year.

Extended to the continental scale, the model projects a maximum invasion speed of roughly 20 kilometers per year, with high-speed clusters in the Rhone Valley and Nouvelle-Aquitaine in southern France and in the Balkans. Most of Central and Western Europe falls below 10 kilometers per year, with limited spread projected for southern England. Invasion appears viable across northern and central Italy but is largely precluded in Spain and Portugal by low suitability, attributed in part to dry soils, except for low speeds along their northern coasts. Southeastern Europe is generally suitable except for Greece, while Scandinavia’s cold summers make it unsuitable. The authors stress that these projections carry uncertainty from unmodeled factors, including natural predation, human-assisted dispersal along dense transport networks, and ongoing control efforts, and that the model is suited to predicting spread during the early stages of new outbreaks rather than long-term continental trajectories.

The study also documents the fingerprints of human help in the invasion pattern. An isolated 2022 outbreak in four municipalities in the Aosta Valley, spatially disconnected from the main infested area, can only be explained by a long-distance dispersal event, and several streaks of infested municipalities follow nearly linear paths aligned with major highways and heavily trafficked roads, patterns too elongated to arise from natural movement alone. The authors conclude that the estimated survival probabilities and diffusion speeds offer practical tools for biosecurity agencies: survival estimates can help approximate when a new outbreak was introduced from captured populations, and predicted dispersal speeds can define search radii for delimitation surveys around new detections, sharpening surveillance where it matters most.

Subject of Research: Spatial population dynamics modeling of the Japanese beetle invasion in Europe

Article Title: A model for the spatial population dynamics of the European invasion of the Japanese beetle (Popillia japonica)

Article References: Martinetti, D., Erbetta, L., Bevacqua, D., Bonnefon, O., Gatto, M., Mari, L., & Casagrandi, R. (2026). A model for the spatial population dynamics of the European invasion of the Japanese beetle (Popillia japonica). Ecological Processes, 15(1), Article 99. https://doi.org/10.1186/s13717-026-00754-w

Image Credits: AI Generated

DOI: 10.1186/s13717-026-00754-w

Keywords: Popillia japonica, Japanese beetle, invasive species, reaction-diffusion model, Po Valley, environmental suitability, invasion speed, pest surveillance, quarantine pest, northern Italy, Switzerland, crop damage