Non-native plants frequently escape herbivory in their introduced ranges because local herbivores lack the evolutionary history needed to recognize or tolerate unfamiliar defenses. Whether this escape reflects phylogenetic novelty, biogeographic origin, or both remains unresolved. I used the Tallamy host-use dataset to test whether Lepidoptera richness on alien plant genera declines with increasing phylogenetic and geographic distance from the native flora. Nearest-native phylogenetic distance and nearest-native geographic distance were uncorrelated, indicating that they captured independent dimensions of novelty. Both predictors had negative associations with total Lepidoptera richness. Standardized coefficients were of similar magnitude for both predictors, and both effects were directionally consistent across model specifications, indicating that evolutionary proximity and biogeographic origin each contribute to herbivore associations with introduced plants. Growth form was the strongest predictor overall: woody genera supported approximately twice the Lepidoptera richness of herbaceous genera after accounting for both distance metrics and sampling effort. These results indicate that both phylogenetic and geographic novelty are associated with lower herbivore richness on introduced plants, whereas plant growth form accounts for more variation than either distance dimension alone.

The Lepidoptera host-use dataset compiled by Tallamy and Shropshire (2009) was used in this analysis because it provides standardized records of herbivorous Lepidoptera species associated with a wide array of native and non-native plant genera in North America. The dataset reports the total number of Lepidoptera species known to feed on each plant genus, based on rearing records and published host associations, offering a consistent genus-level metric of herbivore richness. Although the dataset lacks species-level resolution and direct measures of herbivore abundance, its genus-level consistency facilitates broad inference across taxonomic and biogeographic gradients. It serves as a proxy for relative herbivore load across plant lineages in a standardized framework. All 1,333 plant genera included in the dataset were analyzed. Because the host-use data provide only genus-level associations between Lepidoptera and plant taxa without functional trait measurements, phylogenetic and geographic distances were used as indirect indicators of potential trait similarity or ecological familiarity.

Phylogenetic distance

Phylogenetic relationships among plant genera were reconstructed using the V.PhyloMaker package in R (Jin and Qian 2019), which integrates taxa into a large-scale angiosperm phylogeny using the GBOTB.extended backbone under Scenario 3 of the phylo.maker() function, inserting missing taxa as polytomies at the genus or family level to maximize taxon inclusion while preserving higher-order relationships. Because the dataset was genus-based, whereas V.PhyloMaker requires species-level input, each genus was represented by a placeholder species name and tree tips were subsequently relabeled to genus names after phylogeny construction. Pairwise phylogenetic distances were extracted as cophenetic distances — sums of branch lengths between terminal nodes (Fig. 1a) — using the ape package (Paradis et al. 2004). Alien and native genera were identified from the Tallamy dataset and restricted to genera present in the phylogenetic distance matrix. For each alien genus, the nearest-native phylogenetic distance was defined as the minimum cophenetic distance to any native genus.

Geographic centroids

Geographic centroids for plant genera were estimated from occurrence records obtained from GBIF using the rgbif package (Chamberlain et al. 2023). Up to 1,000 occurrence records were retrieved per genus, and records lacking valid latitude or longitude were excluded before centroid calculation. For each genus, centroid latitude and longitude were calculated as the mean of all retained occurrence coordinates. These centroid values were then incorporated into the analysis dataset used for distance calculations. For the present analysis, duplicate genus–centroid rows were removed separately for native and alien genera, and genera lacking centroid coordinates were excluded from the geographic-distance analysis. Nearest-native geographic distance for each alien genus was then calculated as the minimum great-circle distance, in kilometers, between the alien genus centroid and the centroid of any native genus (Fig. 1b), using the Haversine formula implemented in geosphere (Hijmans 2026). This nearest-neighbor metric was used as the geographic analogue of novelty relative to the native flora.

Because geographic coordinates represented genus centroids rather than plant occurrences, some centroids fell in water bodies. As a sensitivity analysis, offshore centroids were snapped to the nearest land point, and nearest-native geographic distances were recalculated; this correction did not materially alter model results.