1. Introduction

Habitat fragmentation is global threat to biodiversity (IPBES, 2019). Fragmentation often results in reduced species richness, altered species and community compositions, disrupted ecological processes, and degraded ecosystem functioning (Wilson et al., 2016). The drivers of biodiversity loss also affect species coexistence and community assembly processes (Thorn et al., 2016; Bovendorp et al., 2019). Understanding β diversity, which describes variation in species composition among communities within a region (Whittaker, 1972), is crucial for regional biodiversity conservation and can directly inform conservation planning. By evaluating β diversity, conservation managers can implement more effective decision-making and restoration efforts in degraded ecosystems (Socolar et al., 2016; Liu et al., 2025).

β-diversity has been widely employed to explore how habitat fragmentation shapes species coexistence and community assembly (Crabot et al., 2020). A growing body of work distinguishes between incidence-based and abundance-weighted approaches in β-diversity assessments. Incidence-based metrics, which consider only species presence/absence, are prevalent for their simplicity and broad applicability across taxa and spatial scales (Baselga, 2010; Cardoso et al., 2014); while abundance-weighted measures incorporate species dominance and rarity, thereby providing a more nuanced perspective on community composition (Baselga, 2013; Podani et al., 2013; Zhao et al., 2021). Critically, partitioning β-diversity into turnover and nestedness components helps elucidate whether species compositional differences across islands (i.e., patches and fragments) arise from species replacement or from species loss and nested subset patterns, with implications for conservation prioritization based on island area (i.e., patch area) and isolation (Si et al., 2015).

Phylogenetic and functional β-diversity reveal underlying ecological and evolutionary assembly processes by integrating evolutionary relationships and species trait, respectively (Carvalho et al., 2020; Yu et al., 2024). These dimensions can be measured based on either community species composition or pairwise distances between species in phylogenetic or functional trait space (Swenson, 2011; Luo et al., 2023). Notably, incidence-based and abundance-weighted phylogenetic or functional β-diversity metrics may diverge in their sensitivity to environmental gradients, as species abundances modulate community responses to fragmentation (Schweiger et al., 2018; Tucker et al., 2017). Traditional analyses often assume fixed, species-level average trait values, neglecting intraspecific trait variation (ITV), the phenotypic variability among individuals within a species, which can substantially influence functional diversity patterns and community responses to environmental heterogeneity (Agrawal, 2020; Zheng et al., 2022). Incorporating ITV can alter functional β-diversity estimates, sometimes complicating comparisons with species-mean trait approaches (Myers et al., 2021; Wong and Carmona, 2021). However, incorporating ITV into β-diversity frameworks enhances detection of environmental filtering and niche differentiation (He et al., 2021; Westerband et al., 2021).

The effects of habitat fragmentation on β-diversity are generally interpreted through two main variables: island area (i.e., patch/fragment size) and isolation (Si et al., 2016; Fletcher et al., 2018). Partitioning β-diversity into turnover and nestedness components relative to these fragmentation metrics reveals complex community assembly dynamics in fragmented ecosystems (Si et al., 2016). Building on the Theory of Island Biogeography (MacArthur and Wilson 1967), we hypothesize that (1) overall pairwise β-diversity (β_sor) increases as differences in island area increase, reflecting the small-island effect (Matthews et al., 2014; Wang et al., 2018), and with differences in isolation due to distance-decay patterns (Soininen et al., 2007; Baselga and Gómez-Rodríguez, 2021) (Fig. 1a); (2) the turnover component (β_sim) increases with differences in isolation due to dispersal limitation (Fig. 1b), whereas the nestedness component (β_sne) increases with differences in island area driven by environmental filtering (Nekola and White, 1999; HilleRisLambers et al., 2012) (Fig. 1c); (3) incidence-based and abundance-weighted approaches may reveal divergent β-diversity patterns in fragmented landscapes (Zhao et al., 2021). Specifically, we assume that abundance-weighted metrics strengthens the relationship between nestedness (β_sne) and difference in island area while weakening the turnover-isolation relationship (Fig. 1de); (4) by capturing the within-species trait variability that mediates dispersal and niche differentiation, ITV is expected to further enhance the relationship between nestedness (β_sne) and difference in island area while weakening the turnover-isolation relationship (Fig. 1de) (Wong and Carmona, 2021).

To test these hypotheses, this study integrates species abundance and ITV into the partitioning of β-diversity (taxonomic, functional, phylogenetic), as well as functional and phylogenetic distances, in plant communities across 10 fragmented islands in China's Thousand Island Lake region. Specifically, we examine how β-diversity components relate to island area and isolation differences to determine (1) whether incorporation of both abundance-weighted metrics and ITV strengthen nestedness-area and weaken turnover-isolation relationships; and (2) whether abundance-weighting modifies the relationships between functional/phylogenetic distances and difference in island area and isolation.

2. Material and methods 2.1. Study area

This study was conducted in a land-bridge island system, located in the Thousand Island Lake (TIL) region of Zhejiang Province, eastern China (29°22′–29°50′N, 118°34′–119°15′E). These land-bridge islands were formed following the construction of a dam in 1959 (Zhang et al., 2025). Before dam construction, the forests in this area were clear-cut. Today, the islands are predominantly covered by secondary successional forest dominated by Masson pine (Pinus massoniana) and other mixed broad-leaved species (Zhang et al., 2021). Following dam construction, the islands were legally protected from further human disturbance. The TIL region experiences a typical subtropical monsoon climate with a mean annual temperature of 17.0 ℃ and average annual precipitation of 1430 mm (Zhang et al., 2025).

2.2. Data collection

Forest dynamic plots were established on 29 islands in the TIL region between 2009 and 2010 (Fig. S1). These islands were selected because of the range of areas, degree of isolation, and shape, all with minimal low human disturbance (Jin et al., 2020; Zhang et al., 2021). On smaller islands (area ≤ 1 ha), the entire island was surveyed using contiguous 5m × 5m subplots (Fig. S1). On larger islands (area > 1 ha), two or three transects were established (40–210 m in length, consisting of 5m × 5m subplots) from the forest edge to the interior (Fig. S1). A complete census of woody plants (with stem diameter at breast height, DBH ≥ 1 cm) was conducted on the 29 islands in 2019. Here, our aim was to study the role of intraspecific trait variation (ITV) in shaping community assembly and species coexistence in the TIL region. However, exhaustive trait sampling across all 29 islands was logistically infeasible, as this included over 126,000 individual trees. Thus, in this study, we examined ITV of plants on 10 of the total 29 islands. The 10 islands selected for our study provide representative coverage of key fragmentation gradients, spanning a wide range of island areas (0.08–9.79 ha) and degrees of isolation (Table S1 and Fig. S1). Woody plant assemblages on each island were treated as distinct ecological communities. Sampling completeness was assessed using both rarefaction and species accumulation curves, generated with the iNEXT function from the iNEXT R package (Hsieh et al., 2022) and the specaccum function from the vegan R package (Oksanen et al., 2022), respectively (Fig. S2).

The surveyed vegetation across the 10 study islands comprises secondary successional forest communities, with a total of 48,958 individuals from 58 woody plant species. We measured five key functional traits related to plant resource-use strategies for all 58 woody species recorded during the field surveys conducted from 2020 to 2022. Specifically, we measured leaf area (LA, cm²), leaf dry matter content (LDMC), leaf thickness (LT, mm), specific leaf area (SLA, m²·kg⁻¹) and wood density (WD, g·cm⁻³), following standardized protocols proposed by Pérez-Harguindeguy et al. (2013). Due to the large population size, it was impractical to measure traits for every individual. To account for intraspecific trait variation (ITV) while maintaining sampling feasibility, we sampled individuals according to their abundance. For species with fewer than 10 individuals on an island, we sampled leaves and biennial branches from all individuals. For more abundant species, we randomly sampled 10 individuals per species per island, ensuring spatial representation by sampling from both forest edges and interiors. This approach yielded trait measurements for 1745 individual plants.

A phylogenetic tree encompassing the 58 woody plant species was constructed for further analyses, utilizing the phylo.maker function and the ‘GBOTB.extended.TPL.tre’ mega-tree provided by the U.PhyloMaker R package (Jin and Qian, 2023).

2.3. β-diversity partitioning

We partitioned taxonomic, functional, and phylogenetic β-diversity using species distribution data (both incidence and abundance), functional trait values (with and without intraspecific trait variation), and the phylogenetic tree. Functional and phylogenetic β-diversity metrics were interpreted as incidence-based dissimilarities weighted by functional composition and branch lengths, respectively (Cardoso et al., 2014). Adding abundance to functional and phylogenetic βdiversity metrics introduces a second weighting factor, which complicates interpretation because the patterns of species distribution and abundance may cancel each other out. Therefore, we performed four levels of β-diversity partitioning: taxonomic (incidence-based and abundance-weighted), functional (incidence-based), and phylogenetic (incidence-based).

We used the BAS framework (Baselga, 2010) to partition pairwise β-diversity into its underlying components. Specifically, we calculated pairwise Sørensen dissimilarity (β_sor) and its turnover and nestedness components (β_sim and β_sne, respectively), using the beta.pair function from the betapart R package (Baselga et al., 2023). To quantify abundance-weighted pairwise taxonomic β diversity, we applied Bray–Curtis dissimilarity (β_bray), calculated by the beta.pair.abund function from the same package. Following Baselga (2013), β_bray was further partitioned into components reflecting balanced variation in abundance (β_bal) and abundance gradients (β_gra).

In the TIL region, the mean of pairwise β-diversity cannot effectively estimate the multiple-site β-diversity because heterogeneity among islands leads to significant variation in pairwise dissimilarity (Yu et al., 2012; Si et al., 2017; Zhao et al., 2021). Therefore, we applied multiple-site β-diversity partitioning to assess the relative contributions of the turnover and nestedness components. Multiple-site β-diversity metrics were additionally calculated using the beta.multi and beta.multi.abund functions from the betapart R package (Baselga et al., 2023). To distinguish them from pairwise β-diversity metrics (β_sor, β_sim, β_sne, β_bray, β_bal, β_gra), the multiple-site metrics are denoted using capital letters (β_SOR, β_SIM, β_SNE, β_BRAY, β_BAL, β_GRA).

We reduced the dimensionality of the original functional trait matrix by calculating a trait distance matrix using the Gower distance metric. We then conducted a principal coordinates analysis (PCoA) using the pcoa function from the ape R package (Paradis and Schliep, 2019) to generate a transformed trait matrix based on the resulting coordinate axes. For interspecific variation (i.e., without intraspecific trait variation), the first three PCoA axes accounted for 91% of the variance and thus were used to calculate functional β-diversity (Table S2). To assess pairwise functional β-diversity incorporating intraspecific trait variation, we computed mean trait values for species on two islands, subsequently calculating functional β-diversity between these islands. Both pairwise functional β-diversity with and without intraspecific trait variation were measured using the BAS method.

2.4. Different metrics of β diversity: functional and phylogenetic distance

We calculated pairwise phylogenetic and functional distances to complement our β-diversity analyses. The mean pairwise phylogenetic distance (D_pw) measures the average phylogenetic distance between all species pairs across two assemblages, while the mean nearest phylogenetic distance (D_nn) calculates the distance between the most similar species in two assemblages (Swenson, 2011). D_pw reflects basal phylogenetic β diversity, indicating differentiation in habitat preferences among species, while D_nn reflects terminal phylogenetic β diversity, indicating differences in resource utilization (Swenson, 2011; Luo et al., 2023). Standardized phylogenetic β-diversity metrics, β_NRI (Net Relatedness Index) and β_NTI (Nearest Taxon Index), were calculated as the standardized negative effect size of D_pw and D_nn, respectively. Positive β_NRI and β_NTI values indicate that species across the two assemblages are more closely related than randomized assemblages, while negative values indicate these species are more distantly related than randomized assemblages (Webb et al., 2002).

To assess functional β diversity, we computed functional analogues of the corresponding phylogenetic metrics. The mean pairwise functional distance (DF_pw) quantifies the average functional dissimilarity between all species pairs across two communities, while the mean nearest functional distance (DF_nn) captures the functional dissimilarity between the most similar species pairs across the two communities. The standardized effect sizes associated with these metrics are β_NFRI (Net Functional Relatedness Index) and β_NFTI (Nearest Functional Taxon Index), respectively. These indices represent the standardized negative effect sizes of DF_pw and DF_nn, indicating deviations from a null expectation. We calculated D_pw and DF_pw using the comdist function in the picante R package (Kembel et al., 2010), and D_nn, and DF_nn using the comdistnt function. To derive standardized effect sizes, we employed the phylocomr R package (Ooms et al., 2023), implementing the independent swap null model. This model randomly reshuffles species identities in the community matrix while strictly preserving two key ecological constraints: the species richness of each local community and the frequency of each species across all communities (Gotelli and Entsminger, 2003). We performed 9999 randomizations to ensure robust estimation. The ph_comdist function was used to derive β_NRI and β_NFRI (which quantify overall deviation in mean pairwise distances between communities), and the ph_comdistn function was applied to obtain β_NTI and β_NFTI (which reflect standardized deviations in nearest-neighbor distances between communities). Positive values of these standardized effect sizes (β_NRI, β_NFRI, β_NTI and β_NFTI) indicate greater phylogenetic or functional similarity between communities than expected under the null model, whereas negative values suggest increased dissimilarity (Webb et al., 2008).

2.5. Data analyses

We analyzed the relationships between β-diversity and island geographical variables using (partial) Mantel tests based on Spearman's rank correlation, implemented by the mantel and mantel.partial functions in the vegan R package (Oksanen et al., 2022). Here, β-diversity metrics included the pairwise β-diversity (β_sor, β_jac and β_bray), turnover components (β_sim, β_jtu and β_bal), and nestedness components (β_sne, β_jne and β_gra), in addition to functional distances (DF_pw, DF_nn, β_NFRI and β_NFTI) and phylogenetic distances (D_pw, D_nn, β_NRI and β_NTI). Island geographical variables included differences in island area and differences in isolation (measured as the shortest edge-to-edge distance to the nearest mainland). To assess the potential influence of spatial autocorrelation, we tested for spatial dependence in the residuals of our Mantel models. The results indicated that the correlation between residuals and geographical predictors (island area and isolation) was near zero, suggesting minimal spatial autocorrelation and supporting the robustness of the Mantel test outcomes.

All statistical analyses were performed in R 4.5.0 (R core Team, 2025).

3. Results

A total of 58 plant species (belonging to 46 genera and 27 families) were recorded on the 10 study islands. Species richness ranged from 8 to 44 on islands, with the abundance of individuals ranging from 122 to 15,043 (Table S1).

3.1. The role of abundance

Incidence-based data indicated that the species turnover and nestedness components contributed equally to the multiple-site taxonomic β-diversity and the multiple-site phylogenetic β-diversity (β_RATIO ≈ 0.50, Fig. 2). However, for functional β diversity, the nestedness component contributed slightly more than turnover (β_RATIO = 0.68, Fig. 2). When abundance data were considered, the nestedness component contributed more than turnover to multiple-site taxonomic β-diversity (β_RATIO = 0.70, Fig. 2), indicating that the nestedness component played a more prominent role when abundance was considered.

Additional correlations between pairwise β diversity's two components and island geographical variables (difference in island area and difference in isolation) were not significant (Table S3). In incidence-based community data, the overall pairwise β-diversity (taxonomic, phylogenetic, and functional without ITV) and the nestedness component increased with island area differences (Fig. 3), while turnover increased with isolation differences (Fig. 4g, h and i). When abundance was considered, the overall pairwise taxonomic β-diversity and nestedness increased with difference in island area (Fig. 3a and e), while turnover was less responsive to isolation differences (Fig. 4g). These results suggest that considering abundance reduces the importance of isolation in driving species turnover.

When we compared the (partial) Mantel test correlations between differences in island area and the overall pairwise taxonomic β-diversity alongside nestedness calculated from incidence-based community data, we found that correlations derived from abundance-weighted community data were stronger, indicated by steeper slopes (Fig. 3a–e and Table S3). This suggests that incorporating abundance into pairwise β-diversity partitioning enhances the relevance of the nestedness component in relation to differences in island area.

Neither the raw values nor the standardized negative effect sizes of functional and phylogenetic distances exhibited a significant relationship with differences in isolation (Table S5). However, functional distances showed significant positive correlations with differences in island area (Fig. 5), across both incidence-based DF_pw (incidence-based) and DF_nn (incidence-based, abundance-weighted and incidence-based with ITV), and β_NFTI (abundance-weighted with and without ITV). Phylogenetic distancing similarly responded to island area: only D_nn (both incidence-based and abundance-weighted) and β_NTI increased with area differences (Fig. 5d and h). Notably, incidence-based DF_nn and D_nn both demonstrated a stronger association, evidenced by steeper slopes, than their abundance-weighted counterpart (Fig. 5c and d). Incidence-based indices (β_NFRI, β_NRI) were lower than their abundance-weighted versions, while β_NFTI and β_NTI were elevated when weighted by abundance (Fig. S3), sometimes even reversing sign. These patterns underscore the pivotal influence of species abundance on both functional and phylogenetic differentiation.

3.2. The role of intraspecific trait variation

Intraspecific trait variation (ITV) played a critical role in shaping β-diversity patterns. Both the overall pairwise functional β-diversity and its nestedness component were positively correlated with differences in island area, regardless of whether ITV was included (Fig. 3c and f). Notably, the overall functional β-diversity was positively correlated with differences in isolation only when ITV was incorporated (Fig. 4c), whereas the turnover component was positively correlated with differences in isolation only when ITV was excluded (Fig. 4g). When assessing functional distance, only the incidence-based values of DF_nn that incorporated ITV showed a significantly positive relationship with differences in island area (Fig. 5c).

4. Discussion 4.1. Abundance strongly influences β-diversity partitioning and metrics

Our study demonstrates that incorporating abundance data substantially alters both multiple-site and pairwise taxonomic β-diversity patterns. Abundance-weighted metrics consistently increased the prominence of the nestedness component. Similar patterns have been reported in ant communities within the same Thousand Island Lake (TIL) system (Zhao et al., 2021), underscoring the broader relevance of abundance-weighted analyses across taxa. These findings suggest that abundance is a critical yet often underutilized dimension in partitioning β diversity.

The consistent increase in the relative contribution of nestedness across abundance-weighted metrics indicates that species assemblages on smaller islands tend to be nested subsets of those on larger islands, not just in species composition but also in relative abundance distributions. This means that dominant species are consistently represented across communities, while rarer species accumulate disproportionately on larger islands. Because the nestedness component increased with differences in island area (Fig. 3), this implies that area-driven differences in community size and habitat heterogeneity underlie observed abundance patterns. Indeed, both species richness (linear regression, p < 0.001, R_adj² = 0.883) and total abundance (linear regression, p < 0.001, R_adj² = 0.873) increased with island area, primarily due to the addition of rarer species on larger islands, whereas dominant taxa remained consistently abundant regardless of area (Fig. S4). This pattern also supports the hypothesis that species richness increases with island area (the small island effect; Matthews et al., 2014; Wang et al., 2018; Li et al., 2024).

Abundance-weighted nestedness thus reflects a hierarchical structuring of communities, likely driven by deterministic processes such as environmental filtering, competitive exclusion, and source–sink dynamics (Baselga, 2013; Pitta et al., 2019). This metric provides insight into the stability and redundancy of community structure: dominant taxa may provide consistent functional contributions across islands, while the accumulation of rare species contributes to ecological complementarity on larger islands. By integrating species abundance, this nestedness component offers a more nuanced perspective on how species composition and population structure interact to shape β-diversity in fragmented systems.

Moreover, we found that abundance-weighted data modulated the relationship between isolation and species turnover. Specifically, turnover showed a weaker association with isolation when abundance was considered, suggesting that dominant species (more influential in abundance-weighted metrics) are less sensitive to isolation-driven stochasticity. This attenuation may result from environmental filtering maintaining similar dominant species across islands or from limited dispersal affecting only rare species. Alternatively, the interaction between environmental filtering and competitive exclusion under resource constraints (e.g., nutrient limitation; Zhang et al., 2023) may reinforce trait convergence and dominance of a few species, diminishing turnover signals. Thus, after species selection by environmental filtering and competitive exclusion caused by resource limitation, different species may have occupied suitable species-specific habitats, causing species turnover (Gutiérrez-Cánovas et al., 2013).

Patterns in functional and phylogenetic β-diversity offer additional insights. Although incidence-based phylogenetic and functional distances increased with island area differences, these associations diminished or disappeared in abundance-weighted metrics. In contrast, only abundance-weighted standardized effect sizes (β_NFTI and β_NTI) increased with differences in island area, indicating greater phylogenetic and functional clustering among larger islands. This pattern suggests that stronger environmental filtering favors species with similar traits or evolutionary histories. Moreover, the results based on nearest distances (abundance-weighted β_NFTI, abundance-weighted β_NFTI including ITV, and abundance-weighted β_NTI) indicate that differentiation in the utilization of specific resources may be lower among communities on larger island (Swenson, 2011; Luo et al., 2023). In such cases, competitive exclusion could become more important when resource limitation occurs on larger islands. When resources are limited, differences in competitive ability are more important than differences in environmental niche. Thus, more closely related species with similar competitive ability survive, leading to a clustering pattern among larger islands (Mayfield and Levine 2010; Zhang et al., 2023). These findings are also consistent with the observed effects of isolation on species turnover, emphasizing that resource limitation is a key driver of community assembly processes in this fragmented island system.

Overall, these findings reinforce the importance of incorporating abundance into β-diversity frameworks. Abundance-weighted nestedness not only enhances our understanding of compositional gradients but also serves as an indicator of underlying ecological processes that structure fragmented ecosystems, such as environmental filtering, dispersal limitation, and competitive hierarchies.

4.2. Intraspecific trait variation: a buffer and modulator of functional diversity

Intraspecific trait variation also significantly influenced functional β-diversity and its components. Our study revealed that including intraspecific variation increased the nestedness component of functional β-diversity while weakening the association between isolation and species turnover. This pattern suggests that intraspecific variation may buffer the effects of dispersal limitation, possibly by allowing species to maintain functional presence across a broader range of environments. Although we cannot directly identify the underlying mechanisms, such buffering could arise from phenotypic plasticity, local adaptation, or both (Westerband et al., 2021).

In the Thousand Island Lake (TIL) system, isolation did not strongly affect functional trait composition or functional diversity when based solely interspecific trait variation, i.e., without intraspecific trait variation (Jin et al., 2020; Zhang et al., 2021). In contrast, studies that used interspecific trait data gathered from other ecosystems have reported isolation significantly affects community functional structure in fragmented habitats, e.g., in European forests (Lindborg et al., 2012), semi-evergreen tropical forests (Arellano-Rivas et al., 2016), and grasslands (Janečková et al., 2017). As a functional analogue to D_pw, the metric DF_pw also reflects species-level habitat differentiation and niche divergence (Swenson, 2011; Luo et al., 2023). Increasing evidence indicates that intraspecific trait variation can substantially influence the responses of species to environmental changes (Midolo et al., 2019; Westerband et al., 2021). Our results provide further empirical support for the view that incorporating intraspecific trait variation enhances the accuracy and ecological relevance of functional diversity assessments, particularly in fragmented landscapes. By explicitly accounting for intraspecific trait variation, we gain a more mechanistic understanding of how trait-mediated processes govern community assembly in response to habitat fragmentation, offering insights into both deterministic and stochastic components of functional β diversity.

4.3. Conservation implications

Our findings carry significant implications for biodiversity conservation in fragmented landscapes, including archipelagos, terrestrial habitat patches, and patchy grasslands. First, our findings emphasize the necessity of integrating both species abundance and intraspecific trait variation into biodiversity assessments. These components offer a more ecologically realistic portrayal of community structure and β-diversity patterns. Traditional incidence-based metrics, while widely used, may underestimate the influence of environmental filtering and dispersal limitation, potentially resulting in conservation strategies that are incomplete or even misleading. Second, the observed positive relationship between island area and both phylogenetic and functional clustering suggests that larger habitat patches disproportionately support lineages and functional groups with narrower ecological niches. This pattern underscores the critical role of large habitat fragments as refugia for specialists and functionally unique taxa. However, our results also indicate that smaller, more isolated islands may contribute meaningfully to regional biodiversity, particularly when species persistence is mediated by high intraspecific trait variation. Such buffering effects, potentially rooted in phenotypic plasticity or local adaptation, have been demonstrated to enhance community resilience under environmental stress (Westerband et al., 2021). Third, discrepancies between incidence-based and abundance-weighted β-diversity metrics highlight the limitations of relying solely on presence–absence data for conservation planning. Abundance-weighted approaches uncover stronger nestedness components, revealing hierarchical structure in community composition that incidence-based measures may obscure. This insight aligns with recent calls for multicomponent assessments of biodiversity that incorporate species' relative abundances and trait variability (Wong and Carmona, 2021). While we acknowledge that collecting abundance data is often resource-intensive, especially in large-scale or applied contexts, targeted monitoring of dominant or ecologically influential species, paired with trait-based approaches, may offer a practical and ecologically informative compromise. Such strategies can improve the sensitivity of biodiversity assessments and enhance the detection of functional shifts in community structure. Importantly, this integrative framework supports conservation planning that is both empirically grounded and responsive to the multi-dimensional nature of biodiversity in fragmented systems.

Acknowledgements

We are grateful to the anonymous students from the Community Ecology and Conservation Biology Group at Zhejiang University for their assistance with field surveys and laboratory measurements. We also thank Dr. Yi Jin and Dr. Yuhao Zhao for their support with data analyses. We sincerely thank the anonymous editors and reviewers for their constructive comments and valuable suggestions, which greatly improved the quality of this manuscript. This work was supported by the National Natural Science Foundation of China (grant no. 32371603, 31901211 and 31930073).

Data availability statement

The data supporting the findings of this study are openly available in Wang et al. (2024) at https://doi.org/10.1002/ecy.4300 and FigShare at https://doi.org/10.6084/m9.figshare.25231430.v3.

CRediT authorship contribution statement

Aiying Zhang: Conceptualization, Writing – original draft, Writing – review & editing, Funding acquisition, Investigation, Data curation. Xiaofei Wei: Writing – original draft, Writing – review & editing, Data curation. Donghao Wu: Conceptualization, Writing – review & editing, Visualization. Zhonghan Wang: Investigation, Writing – review & editing. Mingjian Yu: Conceptualization, Writing – review & editing. Lingfeng Mao: Conceptualization, Writing – review & editing.

Declaration of competing interest

The authors declare no conflict of interest.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.pld.2025.08.003.