While human activity, resource opportunity, and biotic resistance are recognized as key drivers of plant invasions, their relative importance in agricultural landscapes, particularly arid and semi-arid regions, remains poorly understood. This study assessed how the relative richness and relative coverage of invasive plants respond to these factors in the Ili River Basin. Based on plant community surveys conducted across 78 plots, we used beta regression to examine the relative richness and relative coverage of invasive plants in relation to human activities (including time to city, population density, and cropland proportion), abiotic variables (including macroclimate and soil nutrients), and community structures (e.g., native herbaceous plant richness and total basal area TBA of trees, etc). Variance partitioning and random forest analyses were used to evaluate the relative importance of different predictors.  Multiple regressions showed that both relative richness and relative cover of invasive plants significantly decreased with distance to urban area (time to city) and increased with population density. Both metrics also decreased significantly with native herbaceous plant richness and TBA. The effects of climate and soil variables related to resource availability were quite weak on both relative richness and relative cover. Relative richness was mainly influenced by biotic resistance factors, such as native herbaceous plant richness and TBA, whereas relative cover was primarily driven by variables associated with human activity, time to city. Our study suggests that human activities have facilitated plant invasion in the arid and semi-arid regions, but complex community structures can lessen their impact. Our results highlight that effective prevention and control of invasive plants require targeted monitoring in areas of intense human activity, especially around transportation hubs and corridors. Concurrently, enhancing native community resistance is crucial to limiting invasions.

(a) Study region.

The Ili River Basin, located in central Asia, extends across both China and Kazakhstan. This study focused on the section within China (80.16°-91.03°E, 40.24°-49.18°N), characterized by a continental temperate arid climate. The mean annual precipitation (MAP) ranges from approximately 200 to 800 mm and generally increases with elevation. The mean annual temperature (MAT) varies between 4°C and 18°C and generally decreases with elevation (Pueppke et al., 2018).

(b) Vegetation community surveys

Plant community surveys were conducted in July 2023, during the peak growth season, focusing on agricultural areas of the Ili River Basin where invasive plants are concentrated. To effectively examine the distribution patterns of invasive plant species and their influencing factors, this study established 78 plots, each containing at least one invasive species. These plots were established covering all county-level administrative regions within the Ili River Basin, spanning approximately 130 km north to south and 250 km east to west (Fig. 1). The plots covered three habitat types: farmland boundary, ditch side, and roadside. To ensure the independence of samples, a minimum distance of 30 m was maintained between plots of the same habitat type. These plots represent various environmental gradients with elevations ranging from 507 to 1846 m, MAP ranging from 270 to 538 mm, and MAT ranging from 4.07 to 11.08 °C.

As all three habitats, the farmland boundary, ditch side, and roadside, are narrow linear habitats, three 1m × 1m quadrats were established along a straight line, approximately 7 m apart for each quadrat. Referring to previous vegetation surveys in the Ili River Basin, three 1m × 1m quadrats are sufficient to capture herb community composition variability (Tian et al., 2013; Xu et al., 2011), especially given the simpler community structure in agricultural areas. If trees were present within these habitats, we surveyed for all trees with a diameter at breast height (DBH) greater than 3 cm in a 3m × 16m plot. We recorded the abundance, coverage, and average height of each herbaceous species within each plot. For all species with three or fewer individuals, all were measured for height. For those with more than three, three representative plants were selected for height measurement. When shrubs are present within plots, their species name, crown width, height, and basal diameter are recorded. Specimens that could not be identified in the field were collected for further identification using local flora references or expert consultation.

(c) Definition of invasive plants and the invasive performances

This study cleans up species names according to Flora of China (http://www.iplant.cn/foc). Invasive plants are defined as alien species capable of establishing in the natural or semi-natural habitats, negatively impacting the ecosystem or human health (Lin et al., 2022; Ma & Li, 2018). The origin of Cannabis sativa, one of the earliest domesticated plants, is controversial. Given that the current wild C. sativa populations are genetically distinct entities compared to the extinct wild populations; they can be considered invasive plants whenever they cause harm (Canavan et al., 2022). C. sativa not only serves as a weed in the agricultural fields of the Ili River Basin but also grows densely in habitats such as ditches, stream banks, and field margins, restricting the growth and development of other plant seedlings. Therefore, in this study, we classified C. sativa as an invasive plant.

In this study, all invasive plants are herbaceous. For each plot with three 1 m × 1 m quadrats, we calculated two invasive-performance indices, i.e., relative richness and relative coverage of invasive plants, reflecting the invasion severity in terms of invasive plant richness and coverage, respectively (Delavaux et al., 2023; Guo Q. F. et al., 2015). Relative richness was defined as invasive plant richness divided by total herbaceous plant richness, and relative coverage was defined as the absolute sum of invasive plant coverages divided by the absolute sum of all herbaceous plant coverages in tree 1m × 1 m quadrat in each plot, respectively. 

(d) human activity, macroclimate, and soil nutrient data.

To assess the impact of human activities, we obtained an interpolated dataset of travel times to cities (in minutes) with a resolution of 1 km from the Global Urban Travel Time Map website (https://www.map.ox.ac.uk/accessibility_to_cities/) (Weiss et al., 2018). This dataset measures the time required to reach the nearest city center, providing an assessment of urban resource accessibility and the intensity of human activities. We extracted population density data at a 1 km × 1 km resolution from the Chinese Academy of Sciences' Resource and Environmental Science Data Center (https://www.resdc.cn/). We also extracted land use type data for 2022 from the Annual China Land Cover Dataset (CLCD) (https://zenodo.org/records/8176941) to calculate the proportion of cropland area (P_cropland) within a 2 km radius buffer around the sampling sites. These variables were selected because urban areas are considered hotspots for invasive plant introductions (Borden & Flory, 2021; Klotz & Kühn, 2010), while population density and cropland coverage serve as integrated proxies for the intensity of overall human disturbance and agricultural disturbance (Boscutti et al., 2017; Wagner et al., 2021).

To examine the effects of macroclimate on invasive performances, we extracted MAP and MAT data from 2010 to 2020 at a 1 km × 1 km resolution from the Chinese Academy of Sciences' Resource and Environmental Science Data Center (https://www.resdc.cn/). Soil nutrient variables, including soil organic carbon content, total nitrogen content, total phosphorus content, and pH for the top 0-30 cm of soil, were obtained from the National Earth System Science Data Center (https://www.geodata.cn/main/). We calculated soil C:N, N:P, and C:P to characterize the availability of soil nitrogen and phosphorus. We performed a principal component analysis (PCA) on soil variables, with the first two axes accounting for 86.5% of the variation. Specifically, the first principal component axis (PCA1) accounted for 56.7% of the variation, positively correlated with soil C:P and organic carbon content. The second principal component axis (PCA2) explained 29.8% of the variation, reflecting negative correlations of soil total nitrogen, total phosphorus, and N:P (Supplementary Materials Table S1).

(e) Community structure data.

To investigate the influence of community structure on invasive performances, we calculated native herbaceous plant richness and total basal area (TBA) of woody plants at each site. Shrub variables were excluded, as shrubs were present at only a few sites, which could bias the results. We also included two functional traits of herbaceous plants, plant height and life history type. The plant height is the mean height of each species in the field. We grouped herbaceous plants into two types of life history, annuals and perennials (including biennials). If a species is either annual or biennial, we classify it as annual. We calculated community-weighted mean plant height and life history by species importance value of native herbaceous plants. Species importance value is defined as the mean of the relative coverage and relative abundance here.

To explore the influence of phylogenetic relationships among invasive and native species on their richness and coverage, we calculated the standardized mean phylogenetic distance (SES MPD) for invasive and native plants within the same plot. We constructed a phylogenetic tree for all species observed at the 78 sites using the V.PhyloMaker2 function from the "V.PhyloMaker2" package (Jin Y. & Qian, 2022). We calculated the mean pairwise phylogenetic distance (MPD) between invasive and native species for each site. The "Frequency" null model was used for standardization, randomizing evolutionary distances of the invasive plants while retaining observed frequencies. For each simulated community, we calculated the MPD and repeated the randomization process 1,000 times to obtain the average null distribution. The standardized effect size (SES MPD) was calculated using the formula (observed value - expected value) / standard deviation of the expected value. A larger SES MPD indicates that invasive species are more distantly related to local species, while a smaller value suggests closer phylogenetic relationships. This process was conducted in R using the "Picante" package (Kembel et al., 2010).