This study was conducted from May to November 2022 on Doriseom Island (37°13'34.6"N, 127°43'17.9"E), which is located in Yeoju-si, Gyeonggi-do, South Korea. Doriseom Island is a fluvial island at the confluence of the Namhangang River and the Cheongmicheon Stream (Fig. 1a, b). The island is approximately 906 m in length at its longest point, approximately 520 m wide, and has an area of approximately 292,370 m². During the nearly 30 years from 1991 to 2020, the average annual temperature on Doriseom Island was 11.7 °C, and the average annual precipitation was 1,315.9 mm (KMA 2023). Doriseom Island has an entirely flat terrain formed by the accumulation of alluvial soil approximately 5 m above the surface of the Namhangang River and the Cheongmicheon Stream (National Geographic Information Institute 2021). There is no river running through the interior of the island, and during major floods in the summer, the entire island can be affected by flooding. There are no inhabitants and houses, although some researchers visit the island to study the plants and animals.

Doriseom Island is a valuable and unique habitat, particularly for the resident reptile Eremias argus and flowering plant Aster danyangensis, both of which are classified as endangered wildlife class II by the Ministry of Environment (Macias et al. 2021). However, anthropogenic activities such as a river maintenance project have led to habitat reduction, which has been identified as a major threat to E. argus (Macias et al. 2021).

We first recorded the distribution of E. argus, which largely occurs in the southwest of Doriseom Island (86 sites) and is rarer in the northeast (10 sites; Fig. 1c). We then used the frequency of E. argus occurrences to divide Doriseom Island into two zones: a zone in which E. argus frequently occurs in the southwest (F zone) and a zone in which E. argus rarely occurs in the northeast (R zone; Fig. 1d).

The vegetation survey to identify plant community structure was conducted by randomly installing 1 m × 1 m quadrats (n = 138; 78 in the F zone and 60 in the R zone) in Doriseom Island from May to November 2022. In each quadrat, the plant cover (%) was surveyed and recorded in 5% increments, and plant species were identified using various references (Lee 2003; Park 2009). In addition, we classified the observed plants into exotic and native plants. The nomenclature followed the “World Flora Online” (https://wfoplantlist.org/plant-list), the “Korea Biodiversity Information System Database” (http://www.nature.go.kr/main/Main.do) and Korea National Arboretum (2022).

Three environmental factors were measured: soil hardness (kg‧cm^-2), cobble cover (%), and plant litter cover (%). Soil hardness was measured with five randomly selected replicates in each quadrat using a pocket penetrometer (Forestry Suppliers, Inc., 77114, Jackson, United States), and an average value of soil hardness was calculated from the five replicates. If the soil was covered with cobbles and it was impossible to measure soil hardness, soil hardness was recorded as 4.5 kg‧cm^-2, which was the maximum value of the soil hardness gauge (measurement range 0.01–4.5 kg‧cm^-2). Like the plant cover, the cobble and plant litter covers (%) were recorded in 5% increments within each quadrat.

All statistical analyses were performed using R statistical software ver. 4.2.1 (R Core Team 2022) and the vegan (Oksanen et al. 2022), lawstat (Gastwirth et al. 2020), and ggplot2 (Wickham 2016) packages.

First, to identify significant differences in vegetation composition between the F zone and the R zone, we performed a permutational multivariate analysis of variance (PMAV) with the “adonis” function from the vegan package using Bray–Curtis distance (Oksanen et al. 2022). Standard deviational ellipses (“ordiellipse” function) were plotted in the NMDS ordination to describe zone site compositional differences. Species cover values were log-transformed (log(cover + 1)) to minimize the influence of rare plant species (Májeková et al. 2016). To examine how the dominant plant species differed between the two zones, we performed a rank-sum analysis. The percent cover of plant species in each quadrat was summed and each species’ total sum of cover was sorted in descending order and visualized in a rank sum plot. In addition, native and exotic species richness and the percent cover of Coreopsis lanceolata, which was the dominant plant species in Doriseom Island, were compared using a t-test. Prior to the t-test, homogeneity of variance was tested through the “levene.test” function in the lawstat package (Gastwirth et al. 2020). In addition, the environmental factors (soil hardness, cobble cover, and plant litter cover) were compared between F and R zones using a t-test analysis for equal variances and Welch’s test for heteroscedasticity (Zimmerman and Zumbo 1993; Burns 2006).

Second, non-metric multidimensional scaling (NMDS) was performed to elucidate the relationship between vegetation composition and the environmental factors (soil hardness, cobble cover, and plant litter cover) using the “metaMDS” and “envfit” functions from the vegan package (Oksanen et al. 2022). In addition, vegan’s “ordisurf” function was used to visualize the gradient of environmental factors (soil hardness, cobble cover, and plant litter cover) in the NMDS ordinations.

In a vegetation survey, 69 species belonging to 47 genera within 18 families were observed: Poaceae (25 species, 36%), Asteraceae (9 species, 13%), Cyperaceae (6 species, 8%), and other families (Appendix 1). Exotic plant species, including ecosystem-disturbing species designated by the Korean Ministry of Environment, accounted for approximately 42% of the total flora and were represented by 29 species. There were 27 species of plants common to both F and R zones, as well as 23 species found only in the F zone and 19 species found only in the R zone. For example, the dune sedge Carex pumila was found only in the F zone, while Aster danyangensis, which thrives in cobble environments, was found only in the R zone.

Coreopsis lanceolata showed an overwhelmingly high total cover in both zones (F zone: 2,771%, R zone: 1,571%; Fig. 2). Of the 20 plant species with the highest total percent cover, C. lanceolata, Carex pumila, and Artemisia indica had the highest percent cover in the F zone, while C. lanceolata, Themeda triandra, and A. indica had the highest percent cover in the R zone.

Native plant richness in the F zone (2.5 ± 0.2, mean ± standard error) was significantly higher than that in the R zone (2.1 ± 0.2; Table 1). On the other hand, exotic plant richness was 2.2 ± 0.1 in the F zone and 2.0 ± 0.2 in the R zone, representing no significant difference. C. lanceolata average cover was 36 ± 3% in the F zone and 26 ± 3% in the R zone. Coreopsis lanceolata was the dominant species in the two zones, but its percent cover was significantly higher in the F zone (p < 0.05).

In the comparison of communities using Bray–Curtis distance, the vegetation composition in F and R zones differed significantly according to the frequency of E. argus occurrence (PMAV; F = 4.28; p = 0.001; r² = 0.03), although the r² value is very low. An NMDS was performed to elucidate the relationships between vegetation and soil environmental factors in Doriseom Island (stress value = 0.186). As a result, all three environmental factors (soil hardness, cobble cover, and plant litter cover) had a significant effect on the vegetation composition (p < 0.05; Fig. 3). In addition, as soil hardness increased, the amount of plant litter (such as dead plant leaves and twigs) on the surface tended to decrease.

Soil hardness showed a positive relationship with C. lanceolata (C.la) and Metaplexis japonica (M.ja), but was negatively correlated with A. indica (A.in) and Fallopia dumetorum (F.du), which were related with low plant litter cover. Cobble cover was positively correlated with Galium verum subsp. asiaticum (G.ve) and Populus nigra (P.de), but showed a negative correlation with Ambrosia artemisiifolia (A.ar) and Miscanthus sinensis var. purpurascens (M.si).

We compared soil environmental factors according to the frequency of E. argus occurrence and found significant differences in soil hardness and cobble cover. Soil hardness was 0.9 ± 0.1 kg‧cm^-2 in the F zone and significantly higher at 1.6 ± 0.2 kg‧cm^-2 in the R zone (p < 0.05; Fig. 4a). Particularly, a significant difference was observed in cobble cover, which was 18 ± 3% in the F zone and 40 ± 5% in the R zone (p < 0.01; Fig. 4b). On the other hand, plant litter cover was 59 ± 3% in the F zone and 49 ± 5% in the R zone, being statistically similar regardless of the frequency of E. argus occurrence (p > 0.05; Fig. 4c).

Soil hardness ranged between 0.5–2 kg‧cm^-2 in most quadrats regardless of the frequency of E. argus occurrence, but high hardness values (2 kg‧cm^-2 or more) were more common in the R zone (p < 0.05; Fig. 5a). Cobble cover ranged from 0 to 40% in most areas, but values above 70% appeared only in the R zone (p < 0.001; Fig. 5b). Plant litter cover was similar in both zones (p > 0.05; Fig. 5c).

According to the vegetation survey on Doriseom Island, Coreopsis lanceolata, Artemisia indica, and Erigeron annuus were dominant in both F and R zones, but there were slightly significant differences in species composition. In the F zone, the percent cover of Carex pumila and Ambrosia artemisiifolia was high, while in the R zone, grasses such as Themeda triandra, Phalaris arundinacea, and Miscanthus sinensis var. purpurascens, as well as a Korean endemic plant, Aster danyangensis, showed high percent cover (Fig. 2).

The most dominant herbaceous species on Doriseom Island, C. lanceolata, was positively correlated with soil hardness (Fig. 3). However, C. lanceolata was more abundant in the F zone, where the soil hardness was lower than that in the R zone, indicating that this species was widely distributed throughout the island. Compared to the R zone, which was mostly covered with cobbles and had high soil hardness, the F zone, in which E. argus was more common, had some sandy soil with low soil hardness. In addition, plant richness and native plant richness was higher in the F zone than in the R zone, where the soil hardness was especially high. High soil hardness can obstruct plant growth for several reasons (Unger and Kaspar 1994), including hindrance of root penetration (Singh et al. 2015), low soil water retention (Ngo-Cong et al. 2021), and reduced soil aeration (Watson and Kelsey 2006).

Generally, C. lanceolata readily adapts to diverse ecosystems such as cut slopes, roads, and riverbanks (KIE 2018). It is also known to thrive in a wide range of climates, from temperate to subtropical (Batianoff and Halford 2002). Coreopsis lanceolata is native to the eastern United States and is considered a potentially serious weed in temperate to subtropical regions of Australia (Batianoff and Halford 2002). In addition, it is listed as one of Japan’s 100 most noxious invasive plants (Arifin and Okamoto 2023). This plant species limits the light available to native vegetation (Arifin and Okamoto 2023) and spreads from the approximately 12,000 seeds produced each flowering season (Zeng et al. 2014) or by vegetative growth (Batianoff and Halford 2002). In the future, it will become necessary to monitor how quickly C. lanceolata will spread in Doriseom Island.

Other species that demonstrated high cover in both the F and R zones were A. indica and E. annuus, which were generally distributed across Doriseom Island. Artemisia indica and E. annuus are typical pioneer species, dominating areas shortly after a disturbance (Pacanoski 2017). This pattern suggests that Doriseom Island frequently experiences disturbances such as periodic flooding and human visits.

Unlike C. lanceolata, A. indica, and E. annuus, a few species, like the sedge C. pumila, exhibited a particularly high percent cover in areas with relatively low soil hardness in the range of 1.0–1.5 kg‧cm^-2 (Fig. 5a). Carex pumila is a typical dune plant that thrives in sandy soil with low hardness (Burgess 1984). This plant can be seen in the coastal sand dunes of Sinduri in South Korea, which is the habitat for the Mongolian racerunner, E. argus, (Lee et al. 2020), mirroring our findings. Carex pumila is also found in the coastal sand dunes of southern South Korea and Jeju Island (Oh and Kim 2008). Therefore, E. argus and C. pumila appear to inhabit similar environments.

On the other hand, certain species were rarely found in the F zone but were common in the R zone. This pattern was especially true of A. danyangensis. This plant is endemic to South Korea, growing in limited areas of the Namhangang River basin, and is listed as a class II endangered species by the Ministry of Environment (Kim et al. 2022). A. danyangensis was especially abundant in the R zone and exhibited a strong positive correlation with cobble cover. More specifically, A. danyangensis only thrived in areas where cobble cover exceeded 80% (Fig. 5b). A. danyangensis typically inhabits riverside gravel sandbars, creating a shaded environment when the gravel and cobble remain unembedded and exposed on the surface (Lee et al. 2017). This microenvironment is known to be an important factor in the germination and establishment of A. danyangensis seedlings (Kagaya et al. 2008). In fact, an alternative habitat was created to restore the population of A. danyangensis in the eastern part of the R zone. Since the surface of the alternative habitat was covered with cobbles with a diameter of about 7 cm or more, the frequency and cover of A. danyangensis is judged to have a large positive relationship with cobble.

In general, herbaceous species dominate Doriseom Island, but woody and shrub species such as black poplar (Populus nigra), black locust (Robinia pseudoacacia), and indigo bush (Amorpha fruticosa) are becoming more common. Menke (2003) reported that a number of lizard species that occupy grassland habitats are threatened by the invasion of woody bushes. Therefore, the environmental condition of E. argus living on Doriseom Island may also be threatened by the expanding tree population. In addition to trees, river flooding and human visits represent disturbances on Doriseom Island. Garden et al. (2007) found a positive correlation between the presence of native reptiles and a moderate level of weed cover, ranging from 25% to 50%, indicating that the structure offered by vegetation (e.g., refuge) may be more important than vegetation composition for lizard habitat. These findings support those of Pinto et al. (2018) who noted that on a micro-habitat level, the likelihood of reptile presence, as well as their abundance and species diversity, grew as plant diversity increased. On the other hand, Jellinek et al. (2004) reported that overall lizard species richness was negatively associated with increased plant cover. The relationship between reptiles and vegetation can vary greatly depending on the species (Scott et al. 2006; Mizsei et al. 2020). Some reptiles prefer habitats with dense vegetation for a variety of reasons, including the vegetation’s role in offering shelter and protection from adverse conditions (Attum et al. 2006), refuge from predators (Sato et al. 2014), and reproduction sites (Martin and Murray 2011). Moreover, unlike in our study, E. argus have been found to be more abundant in areas with dense vegetation, preferring the warm thermal environment of denser habitats (Zeng et al. 2016). Consequently, generalizing about the significance of plant cover for all reptiles is challenging.

Our result establishing that the areas where E. argus frequently appeared had a low cobble cover and low soil hardness is consistent with the findings of previous studies, which have shown that reptiles are absent from environments with high soil compaction (Garden et al. 2007). Individuals in one coastal dune region consistently utilized the grass dune habitat throughout the year, using it even for hibernation (Kim et al. 2012b). Eremias argus is most commonly found in areas with low soil hardness that are dominated by dune vegetation, which is similar to E. argus habitat in South Korea (Song et al. 2013; Lee et al. 2020). Since sand dunes are typically arid environments known for their very low solidity (Zaady and Bouskila 2002), they are preferred by E. argus, which lay eggs on the ground. Female E. argus lay around six eggs in a minimum of two batches, typically from late May to early July in South Korea (Kim et al. 2012a). Soil surface resistance, measured in this study as soil hardness, has been understood as important in burrow site selection for some burrowing lizards (Souter et al. 2007). Although lizards’ preferred habitats may vary slightly, generally, habitats must provide essential areas for foraging, hiding, hibernation, egg-laying, and mating (Kim et al. 2012b).

Eremias argus is known as a specialist in dune areas (Park et al. 2014). However, habitat specialists are most likely to become locally extinct because they are rarely able to adapt to drastic habitat changes (Cordier et al. 2021). Furthermore, Doriseom Island is likely to be affected by habitat changes due to human activities and riverside development, resulting in changes in animal movement patterns (Doherty et al. 2021).

Vegetation composition and soil environment were found to differ according to the frequency of E. argus occurrence. However, several predominant plant species were ubiquitously distributed throughout Doriseom Island, and soil environment, especially soil hardness and cobble cover, may affect vegetation structure. In previous research conducted by Souter et al. (2007), no specific vegetation structure was found to be correlated with lizard occupancy within grassland habitats, but potential for soil type to influence the plant community was found. Carpio et al. (2017) also indicated that a more diverse herbaceous ground cover, containing a greater variety of species and resembling natural vegetation, would offer a better solution for promoting diversity in reptile assemblages within agricultural landscapes. Because specific plant species are less important for reptile habitats than the physical conditions that they create, a broader array of habitats, such as a combination of open spaces in close proximity to vegetation cover, can fulfill the requirements of a habitat with diverse structure (Edgar et al. 2010).

Coreopsis lanceolata, an exotic plant that is expanding and forming uniform patches across Doriseom Island, poses a potential threat to the habitat of E. argus. As the climate crisis intensifies, it facilitates the spread of invasive exotic species, which in turn changes the composition of native communities (Vilà et al. 2011). Some have also reported that when the average annual temperature rises by 1 °C, invasive exotic plants become better competitors; in addition, it is known that when invasive exotic species dominate a specific ecosystem, species diversity decreases and a highly homogenized community is created (Kortz and Magurran 2019; Mathakutha et al. 2019). Moreover, negative relationships have been observed between exotic plant cover and lizard species abundance and diversity (Jellinek et al. 2004). Hence, it is crucial to persistently monitor and conduct follow-up research regarding the proliferation of C. lanceolata on Doriseom Island.

Ecotones—the transitional zones between habitats such as sunny woodland edges, grassland-scrub interfaces, and interfaces within grasslands of varying vegetation heights—are especially important for reptiles (Urbina-Cardona et al. 2006; Edgar et al. 2010). These zones are characterized by having a diverse range of plant species and habitat structures, providing various microhabitats and microclimates that are particularly favored by reptiles (Edgar et al. 2010). Pinto et al. (2018) found that a Mediterranean lizard, Psammodromus algirus, showed a positive reaction to habitat diversity, with its population growing as habitat complexity increased. Furthermore, to ensure the viability of E. argus populations, it is crucial to preserve habitat connectivity and natural dispersal dynamics (Park et al. 2014).

The population size and distribution of E. argus may be related to factors shaped by surrounding vegetation, such as food availability, preferences for specific microhabitats, or interspecific interactions. Some preys of E. argus are known (grasshoppers, beetles, ants, leafhoppers, moths, bees, and spiders), which belong to Orthoptera, Lepidoptera, Araneae, Dermaptera, and Amphipoda, among others (Jeong and Song 2010; Huang et al. 2016). In our study, Poaceae family plants accounted for the highest proportion, comprising 25 species among 69 total plant species, providing E. argus populations protection from predators and allowing light to enter between the plants and help E. argus populations to bask properly (Kim et al. 2011; Chang et al. 2021). Although this study did not unveil the intricate connections among vegetation, soil, and food sources, future studies investigating habitat characteristics (vegetation structure and soil) and prey sources should yield additional knowledge to help inform a conservation strategy for E. argus.