The distribution of the invasive herb Lupinus polyphyllus correlates with climate at large scales, but with human presence at local scales

INTRODUCTIONInvasive plants vary in their physiological and ecological traits (Van Kleunen et al., 2010) and may respond differently to management efforts (Ramula et al., 2008). Therefore, effectively managing an invasive plant requires knowledge of its biology, e.g. its habitat requirements (Jiménez-Valverde et al., 2011). Because the habitat requirements, or niche, of a species will be reflected in where it does or does not occur, the former can to some extent be inferred from the latter. This is the idea behind habitat suitability modeling (Elith & Leathwick, 2009). Such models can thus guide management by, for example, identifying potentially suitable but hitherto uncolonized areas (Elith et al., 2010; Formoso-Freire et al., 2023; Roura-Pascual et al., 2011).However, the distribution of a species does not only reflect environmental variation; rather, it results from the interplay between the abiotic and biotic environment, the species’ niche, and the dispersal rate (Pulliam, 2000; Soberón, 2007). In the case of an exotic species in a novel geographic region, the current distribution will additionally depend on the location(s) of the initial introduction, and the amount of time elapsed since then (Jiménez-Valverde et al., 2011). Gauging the amount of dispersal limitation is important for predicting the future course of an invasion, because if suitable but as-yet uncolonized environments are naively treated as unsuitable, projections in time and space become unreliable (Hattab et al., 2017; Jiménez-Valverde et al., 2011). Environmental data from the native range, where the species can be assumed to have dispersed to most suitable areas, can be used to improve niche estimates and predict performance in the invasive range (Formoso-Freire et al., 2023; Hui, 2023). However, the native range is likely to be constrained by biotic interactions (e.g. predation) that may not exist in the invaded region, which may also bias estimates of the potential invasive distribution (Early & Sax, 2014). Ideally, then, information from both the native and invasive ranges should be combined to get a more complete picture of an invasive species’ requirements and tolerances (Early & Sax, 2014; Guisan et al., 2014).A long-standing question in ecology is how and to what extent the effects of environmental variation on species’ distributions depend on spatial scale — from a species occurring or not occurring within a given region, to the number of populations in each region, to the density of individual populations (Crisfield et al., 2024; McGill, 2010; VanDerWal et al., 2009). The spatial scale of an analysis can drastically change variable effects in distribution models (Kotowska et al., 2022; Mod et al., 2020; Nyström Sandman et al., 2013), and models built to predict species presence/absence at larger scales transfer poorly to predicting local abundance (Lee-Yaw et al., 2022). Therefore, to get a fuller understanding of the factors that drive ongoing biological invasions, there is a need for investigations that model distributions at multiple spatial scales, and complement presence-absence data with more direct measurements of abundance.Here we address these issues with an extensive survey of the large-leaved lupine, Lupinus polyphyllus , in Sweden. Originally native to the Pacific coast of North America, L. polyphyllus has established populations on nearly all continents (Hejda, 2013; Meier et al., 2013; reviewed by Eckstein et al., 2023), and has been ranked among the twenty highest-impact invasive plants in Europe (Rumlerová et al., 2016). In its introduced range, it tends to outcompete native plants, especially smaller species (Thiele et al., 2010; Valtonen et al., 2006). As a result, plant communities heavily invaded by lupines tend to drop in diversity (Prass et al., 2022; Ramula & Pihlaja, 2012; Valtonen et al., 2006), and may become homogenized across habitat types (Hansen et al., 2021). Arthropod abundance has also been shown to be lower in lupine-invaded plots (Ramula & Sorvari, 2017; Valtonen et al., 2006).L. polyphyllus can inhabit a broad range of habitat types, and appears to have wide physiological tolerances (Eckstein et al., 2023; Vetter et al., 2019). However, its niche characteristics have yet to be evaluated quantitatively using high-resolution spatial data, and analyses linking its expansion to environmental variation are still lacking, especially in the context of ongoing climate change (Eckstein et al., 2023). Furthermore, to better understand the current distribution and potential for further spread, it is important to consider dispersal limitation in such analyses.We set out to investigate two central questions:What environmental factors best define the niche of L. polyphyllus and shape its distribution, and does the relative importance of these factor differ across spatial scales?To what extent is the species’ current invasive distribution constrained by patterns of dispersal?To answer these questions, we produced two parallel datasets of invasiveL. polyphyllus across Sweden (Fig. 1): i) a high-resolution presence/absence dataset based on a survey of 73 roadside transects (2100 km in total), and ii) in-depth habitat and population characteristics for 152 point-sampled lupine populations. We also complemented these two datasets with citizen-science observations of the species from both its native and global invasive ranges.We addressed Q1 by analyzing the effects of climate, soil, and other environmental predictors on lupine occurrence at four spatial scales: occupancy across transects, occupancy within transects, and patch filling within transects (using the survey data), and ground cover (using the point-sampled data). Furthermore, we tested for differences in environmental responses between road types (highways versus minor roads), and for effects of the environment on lupine size.We addressed Q2 in three ways: i) by measuring the association of lupine occurrence with human activity, ii) by modeling habitat suitability with putatively dispersal-limited absences removed from the input data, and iii) by using global citizen-science data on lupine occurrence to compare the positions in climate space of Swedish populations and other parts of the species’ range.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.