Insect invasions track a tree invasion: Global distribution of black locust herbivores

Many invasive plant species benefit from enemy release resulting from the absence of insect herbivores in their invaded range. However, over time, specialized herbivores may ‘catch up’ with such invasive plants. Black locust is a tree species with a relatively limited native range in North America but has invaded large areas in virtually every temperate continent including North America. We hypothesize that both intra‐ and intercontinental spread of black locust leads to a parallel, though delayed pattern of intra‐ and intercontinental spread of insect herbivores.


| INTRODUC TI ON
Biological invasions are a major source of global change, altering a variety of ecosystem properties, functions, and processes (Ehrenfeld, 2010;Simberloff et al., 2013). Though influenced by a variety of socio-economic factors, increasing globalization in the form of international trade and travel is considered the main driver of biological invasions (Hulme, 2009;Hulme et al., 2008). There is a long history of moving plant species around the world for purposes of horticulture and agriculture; many of these species have escaped cultivation and become naturalized in non-native regions, displacing native plants and plant communities, with cascading effects on other types of organisms (Rai et al., 2022;van Kleunen et al., 2018).
Resulting from this accumulation of non-native species, there is a trend of homogenization of the global flora and fauna (Seebens et al., 2015(Seebens et al., , 2017. The enemy release hypothesis (ERH) has been advanced to explain in part why non-native plants are sometimes able to establish and become invasive in non-native regions (Blumenthal, 2005;Keane & Crawley, 2002). Under ERH, plants established outside the distributional range of their co-evolved herbivores experience lower rates of herbivory, which may result in exceptionally high growth and reproduction. For example, a cross-continental comparison of herbivore and fungal damage to Acer platanoides in North America vs. in its native Europe ) and a comparison of herbivory on introduced A. platanoides vs. native Acer saccharum in North America (Cincotta et al., 2009) showed that Acer species experienced higher growth and lower herbivory levels in their alien habitats. Hawkes (2007) conducted a meta-analysis of published literature to demonstrate a general tendency for enemy release in non-native plants. Accelerated growth rates, due in part to the absence of co-evolved insects and plant pathogens, have been broadly exploited in agriculture and commercial forestry. For example, trees in the genera Eucalyptus and Pinus are widely planted in the southern hemisphere where they are subject to relatively low levels of herbivory, exhibiting exceptional growth rates that exceed levels seen in native tree species and in the introduced trees' native range (Hurley et al., 2016;Lombardero et al., 2008). While such enemy release may lead to higher yields of wood products, it also may contribute to the invasiveness of these species when they escape from forest cultivation (Brundu et al., 2020).
The benefits of enemy release may be short-lived as over time the probability of invasions by insect specialists increases in areas where their hosts are widely distributed (Hawkes, 2007). Hurley et al. (2016) found that there has been an increased rate of invasions by Eucalyptus-feeding insect species (entirely native to Australia) in areas with expansive Eucalyptus plantations. Herbivorous insect invasions are not limited to co-evolved specialized herbivores; the introduction of alien plants may facilitate insect invasions in general, due to increased host plant richness in the established areas.
High levels of plant diversity across large spatial scales may create more ecological niches, thereby increasing the probability that an introduced insect herbivore will locate a host and establish (Guo et al., 2019;Liebhold et al., 2018). For example, invasions into North America by the spotted lanternfly Lycorma delicatula and the brown marmorated stink bug Halyomorpha halys, both native to East Asia and both considered serious pests, may have been facilitated by the presence of their preferred host Ailanthus altissima (Haye et al., 2015;Lee et al., 2019).
The success of invasive plants in their non-native ranges is also strongly affected by their interactions with native herbivore communities. Invasive plants affect native insects both directly, through introduction of a novel resource into ecosystems, and indirectly by altering the abundance and overall fitness of native hosts (Bezemer et al., 2014). Multiple studies have reported shifts and adaptations of native generalist and even specialist herbivores to non-native plants (Branco et al., 2015;Graves & Shapiro, 2003). However, invasive plants may sometimes possess chemical or morphological traits, which prevent successful utilization by local herbivore communities (Bezemer et al., 2014).
Black locust, Robinia pseudoacacia, provides a good example of a tree species with a limited native range that has been successfully introduced and spread through many parts of the Northern and Southern Hemispheres (Li et al., 2014;Martin, 2019). Before its introduction to Europe as an ornamental tree in the early 17th century (Wein, 1930), black locust was limited to its native range in the central Appalachian Mountains and the Ozark Plateau in the Eastern and Central United States (Huntley, 1990). Subsequently, its varied potential as a fast-growing source of high-quality timber and firewood, its resistance to air pollution in urban settings, and its production of nectar for use in apiculture led to its widespread planting around the world (Li et al., 2014;Martin, 2019). Robinia pseudoacacia performs well on disturbed sites and is considered invasive in many parts of its non-native range, with a strong negative impact on native plant species and associated communities (Martin, 2019;Vítková et al., 2017). Even in North America where it is native, R. pseudoacacia has greatly expanded its range, especially to the west (DeGomez & Wagner, 2001).
Though it is a member of the globally distributed Fabaceae family, the native range of the genus Robinia is limited to North America.
Four species of Robinia are currently recognized: R. pseudoacacia, the shrubs R. hispida and R. viscosa, and R. neomexicana (DeGomez & Wagner, 2001). While the native range of the former three species is the eastern United States, that of the small tree-like growing R. neomexicana is restricted to a patchy area in the southwestern United States (Isely & Peabody, 1984; Figure 1b).
While the invasion success of R. pseudoacacia might be attributed in part to its formerly mentioned abilities to colonize disturbed sites, its escape from natural enemies in its non-native range may also have contributed. The current study set out to use the multiple independent invasions of R. pseudoacacia into different continents across the globe to test whether invasions of insect herbivores track the tree's invasion. Through quantitative analysis of the composition of global insect communities associated with R. pseudoacacia in different world regions that each have unique historical timings of the tree's invasion, we test the following hypotheses: (1) the invasion of R. pseudoacacia in seven biogeographic regions is followed by the arrival and establishment of species from the assemblage of specialist herbivores associated with this tree species in its native range; (2) the species richness of these non-native herbivore assemblages increases through time, but decreases with distance from the native range.
Only limited information is available on the community of insect herbivores of R. pseudoacacia in both its native (e.g., Baker, 1972;Hargrove, 1986) and non-native range (e.g., Branco et al., 2019;Kulfan, 2012). A comprehensive summary of global patterns of insect communities utilizing R. pseudoacacia worldwide is lacking. Here, we strive to document and quantitatively analyse (i) the composition of global insect communities associated with R. pseudoacacia in different world regions, (ii) the establishment and spread of Robinia insect specialists in their non-native ranges, and to (iii) identify the environmental and socio-economic factors that influence the geographical distribution of herbivore specialists and their spread in North America. Our intention is to focus on this model system in order to better understand the factors driving invasions of specialist herbivores from the native ranges of invasive plants. We hypothesize that both intra-and intercontinental spread of invasive plants leads to a parallel, though delayed pattern of intra and intercontinental spread of insect herbivores. Illumination of how insect invasions track intra-and intercontinental spread of invasive plants would improve our understanding of the role of plant invasions as drivers of insect invasions worldwide (Liebhold et al., 2018). The system of Robinia herbivores studied here represents an excellent model system due to the extensive invasion success of R. pseudoacacia both in North America and other temperate regions worldwide.

| Global database of Robinia pseudoacacia herbivore communities
We compiled a comprehensive database of known records of insect herbivore species feeding on R. pseudoacacia drawing from the scientific literature, online databases, and personal observations (see Appendix S1 for a complete list of sources). To examine the interregional patterns of herbivore communities, species occurrence data were summarized for each biogeographic region sensu Wallace (1876). The Palearctic region was further subdivided into its Asian and European portions with the Ural Mountains, the Caspian Sea, and the Zagros Mountains marking the division line. Each recorded herbivore species was classified into one of six feeding guilds: folivores (feeding on leaf tissue), gall-makers, sap-feeders, wood-borers, root-feeders, and frugivores (feeding on fruit). Furthermore, herbivores were classified according to levels of host specificity consisting of monophages (feeding only on the genus Robinia), oligophages (feeding on Robinia and other Fabaceae genera), and polyphages (hosts spanning multiple plant families). Differences in proportions of species in these categories among biogeographical regions were tested for statistical significance based on Chi-squared statistics using the chisq.test function in the R statistical software v4.0.0 (R Core Team, 2020).
Species identified as monophagous (referred to as 'Robinia specialists' below) were chosen for further analysis of spatial and temporal patterns in their distributions. For this purpose, geo-referenced occurrence records including the year of observation of these species were compiled from the scientific literature as well as from the online databases GBIF (https://www.gbif.org), iNaturalist (https://www. inatu ralist.org) and BAMONA (https://www.butte rflie sandm oths.org).
Appendix S2, Table S2.1 provides DOI links to the GBIF occurrence dataset for each specialist species. Furthermore, the occurrence records of Robinia specialist species that we assembled from the literature and databases other than GBIF were uploaded to GBIF as occurrence datasets for each species (Medzihorský et al., 2023). For records where source publications did not include specific geographical locations, the occurrence of species was defined as a geometric centroid of the region (county, municipal district, city, etc.) listed in the source material using GeoHack (https://www.media wiki.org/wiki/GeoHack).

| Spatio-temporal analysis of Robinia specialists
The geographic coordinates of all occurrence records were aggregated into 50 × 50 km cells prior to analysing the spatial and temporal distribution of Robinia specialists. Each grid cell contained a presence-absence value for each species and the total number of specialist species (based on occurrence records from 1800 to 2022) was summed for each cell.
In cases of multiple records of the same species inside one grid cell, only the first (oldest) record was considered to analyse the temporal pattern of spread of Robinia monophages established outside the native range of R. pseudoacacia, using the cumulative number of invaded grid cells through time. We acknowledge that the lack of records for a species from a cell does not prove its absence, but sample species richness is statistically associated with underlying true species richness.
In addition, a detailed evaluation of the geographical variation in the richness of monophagous species was conducted for the Nearctic region. For this analysis, occurrence records were limited to those located inside the Nearctic region between latitudes 20° N (Central Mexico) and 50° N (Southern Canada), broadly corresponding with the current distribution of R. pseudoacacia in North America (see Appendix S3, Figure S3.1). As in the global analysis, counts of species richness were made for each 50 × 50 km cell.
To explore possible mechanisms explaining current specialist species richness, we evaluated associations with a set of candidate socioeconomic and environmental variables (Appendix S3, Figure  (CEC, 2010b), and (x) distance [km] from the native range of Robinia calculated as the nearest planar distance between the geometrical centroid of a given grid cell and the border of the native range. We performed two versions of our analysis. In the first analysis, we fit a mixedeffects model using the above predictor variables (i-x), and including only numbers of R. pseudoacacia GBIF records and distance from its native range, that is, without GBIF records of R. neomexicana. In the second analysis, we fit the same model but using numbers of GBIF records for both R. pseudoacacia and R. neomexicana and the distance from the native range calculated from the merged native ranges of both Robinia species. Our logic here was that R. neomexicana may be an ancestral native host for these Robinia specialists and therefore partly explain their successful establishment in new areas. Native ranges of both Robinia species were based on digitized versions of maps published by Little (1971) (Thompson et al., 1999).
All variables were first checked for collinearity by calculating pairwise Pearson's correlation coefficients (Appendix S3, Table S3.2) and any collinear variables with |r| > 0.7 were eliminated from the analysis, based on their relative performance in the model (comparing Z-values).
Given the over-dispersion of source data, with the conditional variance greatly exceeding the conditional mean, a negative binomial regression model (McCullagh & Nelder, 1989) was used to estimate the effect of the candidate socio-economic and environmental variables on specialist richness (No. of species) in the Nearctic region.
Data were analysed using the R statistical software 4.0.0 (R Core Team, 2020) using the glm.nb function from the MASS package (Venables & Ripley, 2002).

| Herbivore community composition
The current distribution of R. pseudoacacia includes seven biogeographic regions: the Nearctic (where the native range is located), the European and Asian Palearctic, the Neotropic, Indo-Malaya, the Afrotropic and Australasia (Figure 1a). Worldwide, 454 insect herbivorous species are documented feeding on R. pseudoacacia.
The Nearctic has the highest number of species recorded feeding on Robinia (225 species The proportion of species in each order also differs significantly among biogeographic regions (X 2 = 184.2, df = 42, p < 0.01). Except for Indo-Malaya, Lepidoptera (moths and butterflies) are the dominant group of Robinia-feeding insects in all biogeographic regions, contributing 35%-100% to the total species pool, though Hemiptera (true bugs, aphids, scale insects, cicadas) have the highest fraction of species recorded feeding on R. pseudoacacia in the native Nearctic region (Figure 2b; Appendix S4). Coleoptera (beetles) are dominant in Indo-Malaya, accounting for almost 50% of the species in this region ( Figure 2b; Appendix S4).
Most of the herbivore community can be considered polyphagous, ranging from 70% to 90% of species recorded in the various F I G U R E 2 Summary of globally recorded insect herbivores of Robinia pseudoacacia (see Figure 1a for locations of biogeographical regions) (a) Total numbers of species in biogeographic regions where R. pseudoacacia is present; (b) proportion of classes of host specificity in each biogeographic region; (c) proportion of feeding guilds in each biogeographic region; (d) proportion of insect orders in each biogeographic region. biogeographic regions; however, there is no difference among biogeographic regions (X 2 = 17.7, df = 12, p = 0.13; Figure 2d). In total, 23 of the 454 globally recorded species feeding on black locust can be considered monophages.

Robinia specialists
To date, six species of Robinia specialists have invaded biogeographic regions outside of their native Nearctic region: the aphid Appendiseta robiniae, the sawfly Euura tibialis, the gall midge tibialis was the first species to be recorded outside of the Nearctic, and it is documented from the European Palearctic region since the early 19th century (Figure 3b) and recently in the Asian Palearctic region. The aphid A. robiniae is the only specialist that is recorded from the Neotropic region (Figure 3a).

| Robinia specialist richness in the Nearctic region
Robinia pseudoacacia has widely expanded its native range in North America (Appendix S3, Figure S3.1a). While the data shown in Figure 4b are limited to forest (non-urban) settings in the conterminous USA, GBIF records indicate that it has a rather limited distribution in the neighbouring countries of Canada and Mexico (Appendix S3, Figure   S3.1a). Overall, 22 insect species in North America were identified as Robinia specialists (Figure 4c; Appendix S2, Table S2.1; Appendix S7, Table S7.4); one species (

| Effects of socio-economic and environmental variables on Robinia specialist richness
Results of the negative binomial regression including only R.
pseudoacacia variables are shown in Appendix S3, Table S3.4, and Appendix S3, Table S3.3 shows the results of the full model.
Population density, total length of railroads, and temperature seasonality were removed due to collinearity with other variables (Appendix S3, Table S3.2). The final model with significant effects includes total length of roads, basal area of R. pseudoacacia, number of GBIF records of R. pseudoacacia, distance from the native range of R. pseudoacacia, annual precipitation total, precipitation seasonality, and annual mean temperature. The total length of roads has the highest predictive power and most positive effect (Z = 27.0; Appendix S3, Table S3.4a) on the number of recorded specialists.
The second highest scoring predictor is basal area of R. pseudoacacia, followed by distance from the native range of R. pseudoacacia, where a shorter distance is associated with a higher number of re-

| DISCUSS ION
Based on the global analysis reported here, we confirm our hypothesis that both intra-and intercontinental spread of invasive R. pseudoacacia leads to a parallel, though delayed pattern of intra and intercontinental spread of insect herbivores that feed on it.
Specifically, we find that the richness of established specialist herbivores feeding on R. pseudoacacia in seven biogeographic regions is positively related to the length of time since the tree species has been introduced (Appendix S5,  Figure 4).
Herbivory levels and herbivore community composition associated with alien plants such as Eucalyptus or Prunus serotina have been shown to increase over time after the introduction of a nonnative plant species (Hawkes, 2007;Hurley et al., 2016;Schilthuizen et al., 2016), and patterns of global diversity of Robinia herbivores are consistent with these observations. A meta-analysis of 62 studies showed that herbivory on invasive plants tends to increase over time since naturalization, supporting the concept of declining enemy release (Hawkes, 2007 ). The Nearctic region, in and around the native black locust range, supports the most diverse herbivore community, followed by the European Palearctic (Figure 2a).
The latter region also harbours most of the globally recorded nonnative Robinia herbivores, likely due to the long-established trans-Atlantic trade and respective pathways for insect spread (Wilson et al., 2009). Furthermore, black locust has been established in Europe longer than in any other alien region, with its introduction in the first half of the 17th century (Wein, 1930), whereas it was introduced in other world regions around the turn of the 19th century (e.g., Guo et al., 2022;Martin, 2019;Owen, 1996) century, and its distribution on the continent is largely restricted to this country (Figure 1a; Martin, 2019); in China, it is reported naturalized since 1878 (Guo et al., 2022); in Australia, the first known trees were planted before 1850 (Martin, 2019), and in New Zealand, it was naturalized in 1870 (Owen, 1996). generations per year (Shang et al., 2015), a high average spread rate of 128 km per year in Europe, and its (potentially human-mediated) ability to quickly spread over long distances (Mally et al., 2021).
The sawfly Euura tibialis has been established for the longest time in a biogeographic region other than the Nearctic (Figure 3b).
It is present in Europe at least since 1825 (Rasplus et al., 2010), and has recently also been reported from Japan (Ichikawa, 2015;Shinohara & Hara, 2020). In Europe, this parthenogenetic species feeds on R. pseudoacacia and R. viscosa. In its native North American range, it was also reported to feed on R. hispida and the honey locust, Gleditsia triacanthos (Darling & Smith, 1985;Raizenne, 1957).
The report of Raizenne (1957) of honey locust as a food plant is suspicious, and we suppose that the author might have misidentified R.
pseudoacacia as G. triacanthos. Euura tibialis might be confused with a related species, E. hispidae, which also feeds on Robinia; so far, the latter species is only known from the USA (Darling & Smith, 1985;Taeger et al., 2018).
Appendiseta robiniae is the only one of the six Robinia specialists that has established in South America (Pagnone et al., 1993). In favourable conditions, the species produces up to 11 generations per year; development is however strongly temperature-dependent, with an optimum below 30°C (Borowiak-Sobkowiak & Durak, 2012).
Being such a strongly reproducing species and adapted to temperate climates, A. robiniae is likely to spread to other regions where its host is found in abundance, such as New Zealand and South Africa. ostensackenella from Central Italy as a first record for Europe, and they suspect the species to be already established in that region but confirmation of this is needed.
In addition to the six North American Robinia-feeding insect species that have invaded other biogeographic regions (Figure 3), there appears to be a seventh species. The phytophagous chalcidoid wasp Bruchophagus robiniae, feeding on the seeds of R. pseudoacacia and R. viscosa, was described in 1970 from Ukraine (Zerova, 1970). As the species has only been recorded from Robinia and not from other Fabaceae native to the Palearctic (Zerova, 1970;Zerova et al., 2017), it appears unlikely that it is native to that region. However, to our knowledge, B. robiniae has not been recorded from its presumably native North American range (see Appendix S6, Figure S6.3b). It may be that the fauna associated with R. pseudoacacia in its rather limited native range is incompletely characterized. Interestingly, B. robiniae is not the only Robinia specialist that was first scientifically described from its (presumably) non-native range: Appendiseta robiniae was described from Denver, Colorado (Gillette, 1907), and Euura tibialis from the English Isle of Wight (Newman, 1837), but both have subsequently been recorded from the native range of R. pseudoacacia.
North America supports the highest diversity of Robinia specialists. The range of most specialists is found at least 500 km be-  (Ford & Cavey, 1985). Despite its extreme abundance, this species appears to be a poor invader. It has never invaded another biogeographic region and only has a limited area in North America where black locust is abundant (Appendix S7, Figure S7.4n).
F I G U R E 4 Spatial distribution of Robinia pseudoacacia specialist insect species in North America. (a) Map of the Nearctic region, showing numbers of species recorded in 50 × 50 km grid cells. Map created using a Behrmann cylindrical equal-area map projection; (b) sum of R. pseudoacacia basal area (log x + 1 m 2 ) in the native range ('native' on x-axis) and in 100 km buffer zones around the tree's native range; (c) spatial records of Robinia specialists in 100 km buffer zones around the native range of R. pseudoacacia. Figure 4b,c share a common x-axis. Liebhold et al. (2021) found Chrysomelidae to be slightly underrepresented among non-native assemblages around the world, suggesting that there may be some unknown characteristics of beetles in this family that lessen their likelihood of successful establishment in non-native regions.
Variation among herbivore species in the extent of both their intra-and intercontinental invasion success most likely coincides with the phenomenon of 'invasion disharmony', where certain groups of insects are over-represented among non-natives, while others are under-represented. This phenomenon has previously been demonstrated for worldwide Coleoptera  and Lepidoptera (Mally et al., 2022)  Within North America, there was considerable variation in the species richness of Robinia specialists (Figure 4a). Some of the areas with the greatest richness fell slightly outside of the native range of R. pseudoacacia. This reflects not only the spread of these species within North America, but also a likely greater search/reporting effort in the mid-Atlantic region. Road density has the highest predictive power on specialist richness, and significantly contributed to explaining the species richness in our final model (Appendix S3, Table S3.4a). Historically, R. pseudoacacia has frequently been used for soil erosion control (Keresztesi, 1980;Vítková et al., 2017) and in afforestation efforts in heavily disturbed areas (Zeleznik & Skousen, 1996) such as mining lands and along transport corridors (Bettosini, 2014;Kolbek et al., 2004). Such large-scale changes in landscape elements promote the movement of herbivorous species by altering the abundance of host plant species, by improving the structural connectivity of favourable habitats (Bezemer et al., 2014), and by increasing the overall risk of invasion in urban-influenced areas (Branco et al., 2019;Paap et al., 2017;Ward et al., 2019).
Robinia density (both basal area and number of GBIF records) was also a significant predictor of species richness, most likely reflecting the positive effect of host abundance on population growth and spread (Liebhold & Tobin, 2008). Similarly, Mally et al. (2021) found that Robinia density facilitated the spread of three non-native Robinia specialists in Europe. Moving farther from the black locust native range, specialist herbivore richness generally declined in areas beyond 500 km from the native range (Figure 4c), and distance was the third most significant predictor of richness in our model (Appendix S3, Table S3.4a). For most insects, climatic suitability is a critical factor that influences their invasiveness and potential spread rate (Bacon et al., 2014;Ward & Masters, 2007). The negative effect of annual mean temperature and precipitation seasonality on species richness observed in our results may initially seem perplexing; however, it probably reflects the mostly northward expansion of R.
Given its extensive distribution in North America and elsewhere in the world, black locust can be considered a massively successful invasive species (Martin, 2019;Vítková et al., 2017). Successful plant invasions may be attributable to different factors, such as changes in the plant's genetic composition following its introduction, the plant's phenotypic plasticity, as well as escape from herbivores and pathogens (Hawkes, 2007). Numerous studies have investigated these different aspects, but it remains difficult to clearly distinguish between the effects of these factors on the success of this particular plant. Black locust exhibits a high genetic diversity in its native range (Mebrahtu & Hanover, 1989;Surles et al., 1989) as well as in China (Guo et al., 2022), but studies directly comparing the genetic composition between native and non-native populations are still lacking. Bouteiller et al. (2021) found differences in the phenotypic plasticity between native and non-native (European) populations of black locust, with an increased germination rate in the European populations. Furthermore, numerous other studies investigated black locust in its non-native range and found a generally high degree of phenotypic plasticity, also in comparison with other trees (e.g., Granata et al., 2020;Guo et al., 2018;Luo et al., 2016;Mantovani et al., 2014;Su et al., 2021;Xu et al., 2009).
Within North America and elsewhere, the success of black locust can be attributed to its ability to thrive in disturbed habitats (Boring & Swank, 1984). Its traits of nitrogen fixation and indeterminant growth allow the species to grow quickly in sites with poor soil conditions.
Human activities such as agricultural clearing, road building, etc. have no doubt promoted the spread of this species. However, the role of enemy release in the success of black locust remains uncertain. The species often suffers from very high levels of herbivory in its native range (Hargrove et al., 1984), but similar levels of herbivory are not reported from other biogeographic regions. Here we document markedly lower diversity of Robinia-feeding herbivores in its invaded range.
To clarify the role and impact of insect herbivory in the success of black locust as an invasive plant, it would be useful to carry out studies in North America (both in the native and invaded ranges) as well as in invaded areas in other continents. In North America, the complete native herbivore assemblage appears to be tracking the spread of black locust beyond its native range (Figure 4a,c) while herbivore assemblages in other continents remain incomplete; this pattern may facilitate future studies to distinguish between the effects of 'enemy release' and those of 'competition release' (Keane & Crawley, 2002) in black locust.
Results presented here demonstrate a more general pattern in which the global spread and local dominance of invasive plants promote invasions of specialist herbivores. Widely distributed invasive plants provide ecological niches for non-native herbivores, and their presence increases the likelihood of successful herbivore establishment following arrival. The global proliferation of invasive plants thus explains the global association between plant and insect invasions (Liebhold et al., 2018). Furthermore, the steady accumulation of specialist herbivores over time contributes to a decline in the beneficial release of invasive plants from natural enemies.

ACK N O WLE D G E M ENTS
No permits were required. We thank Etsuko Shoda-Kagaya (Forest

CO N FLI C T O F I NTE R E S T S TATE M E NT
The authors declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are openly available in GBIF as occurrence records dataset at https://doi.org/10.15468/ zt8g2r.