1. Introduction

Forests are the most important global repositories of terrestrial biodiversity (Shvidenko et al., 2005; Morales-Hidalgo et al., 2015), with total land forest coverage as high as 31% (Aerts and Honnay, 2011; FAO and UNEP, 2020). Forests hold 80% of terrestrial biomass and > 50% of terrestrial species of animals and plants (Shvidenko et al., 2005; Aerts and Honnay, 2011). In some cases, plant richness (trees and shrubs) exceeds 650 species/ha (Valencia et al., 2004). An important question in plant community ecology, therefore, is how such high levels of diversity are maintained. Forest ecologists have made significant advances in understanding the mechanisms that promote the coexistence of plant species and contribute to the maintenance of diversity in forests (Nakashizuka, 2001; Wright, 2002), but have focused almost entirely on woody plants. This bias leaves our knowledge of herbaceous plant communities far behind that of woody plants, and prohibits a fuller understanding of plant diversity throughout forests.

Forest ecology has focused on trees because they dominate carbon, biomass, and financial evaluations (Braun, 1950; Richards, 1952; Spurr and Barnes, 1973; Condit, 1995; Linares-Palomino et al., 2009; Lewis et al., 2009; Muller, 2014). However, other plant growth forms contribute significantly to forest composition, structure, biodiversity, function, and faunal habitat (Johansson, 1974; Gentry and Dodson, 1987; Schnitzer and Carson, 2000; Gilliam, 2014; Thrippleton et al., 2016). Herbaceous plants (herbs), for example, can contribute over 80% of species richness in temperate forests (Gilliam, 2007; Spicer et al., 2020). In addition, forest herbs often have important roles in water and soil conservation, regulating regional microclimates, and resources for animals (Hart and Chen, 2008), as well as community nutrient cycles and energy flows (Standish et al., 2004; Nilsson and Wardle, 2005; Gilliam, 2006), affecting seed germination, seedling development, and the survival and growth of woody plants (Wu et al., 2014). Ecological studies that include herbs will likely challenge our understanding of fundamental ecological processes and further promote the integrity of forest ecology research.

The distribution of herbaceous plant species diversity is influenced by biotic (e.g., woody plant competition, herbivore and insect predation, pathogens, etc.) and abiotic factors (e.g., habitat factors) (Warren and Bradford, 2011; Jia et al., 2022; Spicer et al., 2022). The most important abiotic factor that affects community species diversity is topography. Previous studies have examined how different aspects of topography (i.e., elevation, slope, slope direction and convexity) affect species composition and diversity (Cielo-Filho et al., 2007). Species diversity of the underlying vegetation of a forest can also be determined by stand structure, a comprehensive reflection of the development of a forest stand (Zhang et al., 2021). In addition, plant community structure, composition, and diversity characteristics may be influenced by soil factors, which play a key role in plant growth and development (Eisenhauer et al., 2011). However, these studies have mainly focused on the effects of one factor on the diversity of understory herbaceous plants, e.g., elevation (Jiang et al., 2018), slope (Wiharto et al., 2021), soil (Mao et al., 2021), or stand structure (Ali et al., 2016). Moreover, few studies have investigated the effects of both abiotic and biotic factors on understory herbaceous diversity patterns in forests.

Most studies of the drivers of species coexistence and the maintenance of diversity in the plants of forests have focused on woody plants (Lutz et al., 2013, 2018; Radhamoni et al., 2023). For example, only a handful of studies have examined dispersal limitation, resource niche partitioning, or other mechanisms in the herbaceous plant communities of forests (but see Burton et al., 2011; Reich et al., 2012). The question remains as to whether the diversity of herb and woody plant communities is maintained by the same mechanisms. In particular, we lack an understanding of how gradients of environmental resources (such as topography, forest stand structure, soil physicochemical, etc.) drive herbaceous plant distributions.

In this study, we hypothesize that herb and woody plant communities have similar patterns and drivers of alpha- and beta-diversity. To test this hypothesis, we surveyed herbs and woody plants in the forest dynamics plot of 20-ha warm-temperate deciduous broad-leaved secondary forest in Donglingshan, China. We specifically determined whether (1) alpha- and beta-diversity differ for herbs and woody plants; (2) spatial patterns of alpha-and beta-diversity differ for herbs and woody plants; and (3) the drivers of these spatial patterns of herb and woody plant alpha- and beta-diversity differ.

2. Materials and methods 2.1. Study area

The study site is located in the center of Xiaolongmen National Forest Park, Donglingshan, Mentougou District, Beijing China, in a typical warm-temperate deciduous broad-leaved forest. This region has a warm-temperate and continental-monsoon climate, with an average annual temperature of 4.3 ℃ and a mean annual precipitation of 589 mm, with most rain falling in July (Su and Li, 2012). The forest is about 60 years old and is composed largely of Quercus wutaishanica, Populus davidiana, Juglans mandshurica, and Betula dahurica. The dominant understory herbaceous plant species are Deyeuxia pyramidalis, Saussurea mongolica, and Aster trinervius subsp. Ageratoide (Liu et al., 2014; Xu et al., 2023).

Deciduous broad-leaved forests are one of the most typical zonal vegetation types in warm-temperate regions of China, as a result of adaptations to climate and soil. However, human activities such as deforestation have led to a decrease in the extent of the native forest vegetation and now very few natural forests remain in the region. The Donglingshan Forest is currently one of the best preserved natural secondary forests in China's warm-temperate zone, conserving the rich species resources of this area (Xie and Chen, 1994).

2.2. Woody seedling and herb census

The Donglingshan forest dynamics plot (Donglingshan FDP, 39.95°N, 115.42°E) is a permanent 20-ha (400 m × 500 m) plot that was established in 2010 with the aim to monitor long-term dynamics at this site. All woody stems ≥ 1 cm diameter at breast height (DBH, measured at 1.3 m) within the plot are mapped, tagged, and measured every five years (Liu et al., 2014; Xu et al., 2023).

In 2015, we set up 150 seed-rain stations, distributed evenly throughout the whole plot about every 40 m (Fig. S1). At each station, we established three seedling plots (1 m × 1 m), distributed in three directions from the central seed rain trap, for a total of 450 seedling plots. Within each plot, we surveyed all tree and shrub (excluding liana) seedlings with DBH < 1 cm every year, from July to September, the main growing season. For each tagged woody seedling, we recorded species, status, height, and the number of leaves.

Starting in 2015, we also surveyed herbaceous plants (excluding herbaceous climbers) in each seedling plot. We recorded species name, height, percent cover, and the number of plants (clusters). In this study, we analyzed data on herbs and woody seedlings from 2022. Because some individual seedling plots did not contain either woody or herbaceous plants, we pooled data for the three plots at each station. The total sample size for analyses is therefore 150.

2.3. Environmental data

The 20-ha plot was surveyed. Specifically, it was divided into 500 consecutive 20 m × 20 m quadrats using a total station. The available topographic variables were elevation, slope, convexity, and aspect of each quadrat (20 m × 20 m). Elevation of each quadrat was calculated as the mean of the elevation of the four corners. Convexity is the quadrat elevation minus the mean elevation of the eight adjacent quadrats (for edge quadrats, convexity was the elevation of the center point minus the mean of the four corners). Slope is the mean angular deviation from horizontal of each of the four triangular planes formed by connecting three of its corners. Aspect was oriented such that any three vertices of the sample form four different triangular planes, and the average of the angles of deviation of these four planes from the due north direction was taken as the starting point of the due north direction (0°). Using the seed-rain collector as the center and a radius of 10 m, data from trees with diameter at breast height (DBH) ≥ 1 cm within 10 m radius were used to calculate the stand density inside each survey sample site.

We assigned a topographical habitat category to each 20 m × 20 m quadrat based on multivariate regression trees (MRT; De'ath, 2002), grouping the quadrats according to their topographical characteristics (i.e., elevation, slope). In this way, we defined four habitat categories: high-elevation slope (elevation > 1408 m; 46 seed-rain stations), middle-elevation slope (elevation < 1408 m and ≥ 1358 m; 65), low-elevation steep slope (slope > 40.31° and elevation < 1358 m; 37), and low-elevation gentle slope (slope < 40.31° and elevation < 1358 m; 2). For analysis we pooled low-elevation steep slope and low-elevation gentle slope (analyses with all four habitats are presented in Fig. S2 and are qualitatively similar).

We estimated soil physicochemical values for each seedling plot. We divided the whole plot into a 30 m × 30 m grid. We collected one soil sample at each grid point, and then selected two additional sampling points at random directions 2 m and 5 m, 2 m and 15 m, or 5 m and 15 m from the initial sampling point. Each soil sample was analyzed for the following physicochemical properties: pH, soil water (SW), soil organic matter (SOM), total nitrogen (TN), total potassium (TK), total phosphorus (TP), Ca, Mg, Al, Fe, available potassium (AK) and available phosphorus (AP). We calculated Kriged values at the 5 m × 5 m scale and assigned values to the appropriate seedling plot. Because there were multiple co-varying soil variables, we calculated a principal components analysis on these plot-level data and used the first two axes to describe overall soil fertility.

Understory light availability was estimated from light detection and ranging (LiDAR) data collected in 2018. LiDAR data of the ~4 km² study area within the Donglingshan district were collected in August 2018 using a Greenvalley Technology LiAir 1000 system with a GV1610 laser scanner from 200 m above the ground, with a wavelength of 1550 nm and an average point density of 19.7 points per m². To remove the effect of topography, we filtered the LiDAR data for outliers, classified returns into ground and non-ground (Guo et al., 2017). As a measure of understory light availability, we extracted the effective leaf area index (LAI) from the preprocessed laser point cloud, and output at a resolution of 20 m × 20 m, using the LiDAR360 software.

2.4. Alpha-diversity

To determine whether alpha- and beta-diversity differ for herbs and woody plants, we calculated raw species richness, the exponentiated Shannon–Wiener diversity index (H′), and the Chao-1 richness estimator, which corrects for undersampling (Chao, 1984), for each of the four habitats and two growth forms. To compare the spatial pattern of alpha-diversity, we tested for correlations in the three metrics of species richness between woody seedlings and herbs at each station.

2.5. Beta-diversity

To further determine whether alpha- and beta-diversity differ for herbs and woody plants, we quantified beta-diversity using three methods: the mean Morisita-Horn distance (β_MH), the Jaccard dissimilarity index (β_jac), and the total variance of the site-by-species matrix [SS(Y)]. In this paper, the larger the mean Morisita-Horn distance (β_MH) and the Jaccard dissimilarity index (β_jac), the greater the difference in species composition between the two communities. We calculated SS(Y) from the Hellinger-transformed site-by-species matrix (Legendre and Gallagher, 2001). We partitioned beta-diversity into contributions from individual site and species (Legendre et al., 2005; Legendre and De Cáceres, 2013). To compare the spatial pattern of beta-diversity, we tested for a correlation between the local-site contributions to beta-diversity (LCBD) of woody seedlings and herbs at each station.

To determine whether the local-sites that contributed most to beta-diversity were different between herbs and woody seedlings, we used LCBD, using a permutation test for significance (n = 9999). The null hypothesis of this test was that species were randomly distributed in the plot. We kept the total abundance of species unchanged, but the abundance distribution of species in each plot was randomly configured.

2.6. Habitat association analysis

We examined the species composition of herbs and woody seedlings across three topographic habitats using non-metric multidimensional scaling ordination (NMDS) with Bray–Curtis dissimilarity. We tested for significant differences between herb and woody seedling communities using permutational multivariate ANOVA (PERMANOVA, Anderson et al., 2011).

We then used variance partitioning to calculate the relative effects of environment and space on the herb and woody seedling communities. We partitioned variation in the richness and composition of each community into fractions, explained either by environmental (topography, forest stand structure and the soil physicochemical variables) or spatial factors. We used principal coordinates of neighbor matrices (PCNM; Peres-Neto et al., 2006) generated from the coordinates of the quadrats as the spatial factor. All the environmental factors were standardized to z-scores. We used forward selection to identify the factors (environmental and spatial) that were correlated significantly with richness and composition (Monte Carlo permutation, n = 999). We used adjusted R-squared values to measure the fractions of variation explained (Peres-Neto et al., 2006).

First, the herbaceous plant species composition data-sample was analyzed by detrended correspondence analysis (DCA), and the maximum value of the gradient length of the four ranking axes was 1.5, so the ranking method in this study used a linear model, i.e., redundancy analysis (RDA). The variance inflation factor (VIF) of each environmental variable was calculated, and it was found that its value was less than 10, indicating that there was no obvious covariance problem (Table S1). The species composition matrix of herbaceous plants was ranked against the matrix of environmental factors to analyze the relationship between species composition differences and environmental factors. The lengths of the arrows of the factors in the ranked plots indicate the strength of their relationships with the species composition differences.

In the constrained ordering analysis of redundancy analysis, the explanatory rate of the whole model can be obtained, while the relative contribution of individual explanatory variables can be obtained by introducing hierarchical partitioning theory into redundancy analysis, which further reveals the relative importance of each variable in analyzing community species diversity. Hierarchical partitioning allows for a better comparison of the relative importance of each environmental factor. All analyses were carried out using the software R v.4.1.1 (R Development Core Team, 2021) and the packages "vegan" (Oksanen et al., 2020) and "adespatial" (Dray et al., 2021).

3. Results 3.1. Alpha- and beta-diversity of the herb and woody seedling community

Within the 450 seedling plots, we recorded a total of 81 species of herbaceous plants from 57 genera and 30 families. The most speciose family was Asteraceae (19 species), followed by Ranunculaceae (7), and Violaceae (6). These three families accounted for 39.5% of herb species. The dominant species were Deyeuxia pyramidalis (Importance value = 10.68), Saussurea mongolica (10.44), and Aster trinervius subsp. ageratoides (7.94) (Table S2).

In these same plots, we recorded 31 species of woody seedlings, belonging to 23 genera and 19 families. The most speciose families were Rosaceae, Betulaceae, and Ulmaceae, with 3 species each, accounting for 29% of all woody seedling species. The woody seedling layer was dominated by a few common species, such as Deutzia baroniana (Importance value = 30.09), Deutzia parviflora (26.75), and Spiraea pubescens (12.56) (Table S3; Fig. S5).

Overall and in all three habitats, herb species outnumbered woody seedlings 2:1 (Table 1). Both indices of alpha-diversity were higher (often considerably) for herbaceous plants than for woody seedlings in all three habitats. Similarly, all three indices of beta-diversity were higher for herbaceous plants than for woody seedlings.

3.2. Correlations in local composition and diversity among sites

Spatial patterns of alpha-diversity for herbaceous plants differed from those of woody seedlings. Herbaceous plants had the highest species richness in low-elevation habitats, whereas woody seedlings had the lowest (Table 1). There was no correlation between the species richness of herbs and woody seedlings per seed-rain station across the whole plot (r = −0.014, P = 0.87; Fig. 1a). There were similar non-significant results for correlations between woody seedlings and herbs using the exponentiated Shannon–Wiener diversity index (r = −0.039, P = 0.64; Fig. S3a) and the Chao-1 richness estimator (r = −0.023, P = 0.69; Fig. S3b).

In contrast, spatial patterns of beta-diversity for herbaceous plants were correlated to those of woody seedlings. Specifically, the local-site contributions to beta-diversity (LCBD) values of herbs and woody seedlings were significantly negatively correlated (r = −0.2, P < 0.05, Fig. 1b). Sites with higher woody seedling LCBD had lower herb LCBD.

3.3. Habitat associations and resource gradients

Categorical habitat type explained only 3% of the spatial variation in community diversity, and was significant for both herb and woody seedling communities (Fig. 2, left column). Further, habitat type explained a slightly larger amount of the variation in herb community composition than in woody seedling composition (PERMANOVA: F = 2.42, R² = 0.033 for herbs vs. F = 2.33, R² = 0.031 for woody seedlings). However, continuous environmental variables explained more of the variation in each community and were significant for both herbs and woody seedlings (F = 2.61, R_adj² = 0.14 for herbs vs. F = 2.29, R_adj² = 0.12 for woody seedlings; Fig. 2, right column).

Variance partitioning models explained more than three times the total variation in the species richness of herbaceous plants than of woody seedlings (Fig. 3a). Space accounted for most of the explained variation in herb richness (Fig. 3a). In contrast, space accounted for less than half of the explained variation in the composition of each community (Fig. 3b). The fraction of variation in species composition explained by environment, space and shared factors was greater for herbaceous plant than for woody seedlings (Fig. 3b).

Hierarchical partitioning showed that herb and woody seedling communities differed greatly in their associations with individual environmental variables, and most factors explained < 6% of variation (Fig. 4). Important factors for herb species richness included stand density and soil PCA1, whereas for woody seedlings these included leaf area index. Factors important for herb species composition included stand density and soil water, whereas those for woody seedlings included soil water. RDA ordination plots showed that soil water and stand density were negatively correlated with herb species composition variability (Fig. 2a). However, for woody seedlings, soil water was positively correlated with variability in species composition (Fig. 2b).

4. Discussion

Through our study of a warm-temperate broad-leaved deciduous forest in China, we found that herbaceous plants contribute greatly to the diversity of understory vegetation. First, species richness of herbaceous plants was much higher than that of woody seedlings alone, with as many as 81 species in 450 1 m × 1 m plots. Second, herb species tended to be more clumped than woody seedling species, leading to higher overall beta-diversity. This pattern is caused by greater dispersal limitation and habitat associations. We also found no correlation between herb and woody seedling community richness across sites. However, local-site contributions to beta-diversity (LCBD) of herbs and woody seedlings were negatively correlated, suggesting that herb and woody seedling communities may respond differently to gradients of environmental resource availability. Our interpretation of these results is that the mechanisms that drive community composition, and/or their relative effect, differ between herbs and woody plants.

4.1. Alpha- and beta-diversity of understory herbs and woody seedlings

In the 20-ha forest plot, herbs (81 species, 30 families, and 57 genera) contributed far more to total species richness (72% of all understory species sampled) than woody seedlings. Herbaceous plants can coexist at high abundance, not only with each other but also with woody seedling species in the community. In many temperate forests, herbaceous plants comprise the large majority of the alpha-diversity of the ecosystem (Spicer et al., 2020).

Our findings indicate that species richness of forest herbaceous plants is distinct in different climatic zones. For example, species richness of warm-temperate deciduous broad-leaved forest understory herbaceous plants in Donglingshan was lower than that of a Parashorea chinensis forest in Xishuangbanna (115 species, 41 families, and 76 genera) (Chen, 2008), a tropical forest. However, species richness in Donglinshan was similar to that of an evergreen broad-leaved forest in Gutianshan (81 species, 31 families, and 55 genera) (Tian et al., 2023), which is a subtropical forest.

Beta-diversity was also higher for herbs than for woody seedlings. These findings suggest that herbaceous plants may be more sensitive to differences in habitat or gradients of environmental resources than are woody seedlings. For example, in the only similar study to ours, Murphy et al. (2016) documented much stronger habitat filtering for herbaceous plants than for tree and liana seedlings. In their tropical forest site in Panama, herb species had the highest beta-diversity, and habitat explained twice the variance in the composition of the herb community than in the woody seedling community. Further, understory herbs tend to experience higher dispersal limitation than woody plants.

4.2. Correlations in local diversity among communities

We found that spatial patterns of diversity differed for herbs and woody plants. Previous studies showed that in tropical and subalpine forests local richness and compositional uniqueness were similarly not correlated with herbs and woody plants across sites (Both et al., 2011; Murphy et al., 2016), indicating that different processes control the spatial patterns of woody and herbaceous diversity and composition. Across the entire 20-ha plot, quadrats with high or low richness levels differed between herb and woody plant communities. Likewise, for herb and woody plant communities, quadrats with unique species composition varied, suggesting that herbs may respond differently to resources compared to woody plants. Further, the two growth forms may be limited in their ability to survive together by competitive interactions. In contrast, some previous studies (e.g., Gentry and Dodson, 1987; Wright, 1992; Gilliam, 2007) found that herb and woody plant communities were positively correlated in terms of richness/diversity, albeit at larger spatial scales. Variation in climate or soil fertility may similarly affect herb and woody plant communities at the landscape scale.

4.3. Response to environmental resources and spatial gradients

Herb community (diversity and composition) was strongly correlated with environmental variables (topography, soil fertility, and forest stand structure) and spatial gradients. Documenting these relationships improves our understanding of the mechanisms that drive herb diversity in this warm-temperate forest. Our results are similar to those of a previous study on a tropical forest that found herb and tree seedling composition differ in their responses to environmental variation, but at a slightly larger spatial scale (50-ha) (Murphy et al., 2016). At the 20-ha scale analyzed here, we found that herbs were more sensitive than woody plants to variation in the habitat type and environment. Specifically, low-elevation habitat had a greater role in driving herb beta-diversity than woody plant beta-diversity. This difference is mainly due to the suitable temperature and water availability of these sites, with deep and well-developed soils, better water retention, available resources, and superior hydrothermal conditions that allow more herb species to coexist, thus forming a peak area of species diversity. In addition, 19 herb species were only found in these low-elevation habitats, suggesting that many of them may not have survived in other areas of the plot. Finally, herbs in low-elevation habitats may benefit from less obvious effects of dominant species and higher uniformity.

Hierarchical partitioning revealed that stand density more greatly impacted the diversity and composition of understory herbaceous plants than that of woody seedlings. Furthermore, differences in the species composition of understory herbaceous plants were negatively correlated with stand density. Stand density affects the survival of understory herbaceous plants by controlling light transmission through the forest canopy and by trees competing with the herbs (Cook, 2015; Machado and De Almeida, 2019). Understory herb abundance may be positively correlated with light availability (Radhamoni et al., 2023), and high forest density may inhibit the light absorption efficiency of herbs, making it difficult for some sun-loving herbs to survive. In other cases, competitive exclusion may lead to a few herbs dominating, resulting in a lower diversity of species in the community (Grime, 1973). Competition for nutrients in the high-density understory is intense, increasing competition for understory herbaceous ecological niches and thus affecting the growth of some disadvantaged herbaceous plants (Zhang et al., 2021).

For herb and woody seedling communities, spatial descriptors included the contributions of spatially structured unobserved variables, plus stochastic aggregative processes such as dispersal limitation of propagules (Legendre et al., 2009) and light availability (Chazdon and Pearcy, 1991). Similarly, we expect the greatest dispersal limitation in the herbaceous layer: most forest and woodland herbs do not have wings, hairs, or hooks for long-distance dispersal. Even herbaceous plants with these morphological features tend to have short seed-dispersal distances (Matlack, 1994). Clearly, these niche-defining processes shape the assemblages of herbaceous plants at our site.

5. Conclusions

We examined patterns of herbaceous plant diversity and their environmental drivers in the understory of a warm-temperate deciduous broad-leaf forest in China. We found three times the number of herb species compared to species of woody seedling. Contrary to some of our original expectations, we did not find that herb and woody plant communities had similar patterns and drivers of alpha- and beta-diversity. The local-level variation in the diversity of herbs suggest environmental filtering strongly shapes patterns of plant diversity, with stand density being the key limiting factor for understory herbaceous plants, followed by soil moisture. Herbs exhibited stronger dispersal limitation and/or niche-partitioning of the environment than woody seedlings. Thus, our study highlights that even within the same community, different plant growth forms may require different management and conservation strategies to maintain stable populations and levels of diversity. Furthermore, our study suggests that patterns of tree diversity and abundance cannot be extrapolated simplistically to other plant life-forms in these warm-temperate forests. Similar studies on the diversity and distribution of all warm-temperate forest plant lifeforms, including epiphytes and non-flowering plants, would be a highly desirable step.

In warm-temperate forests, where seasonal changes are evident and constraints are abundant, herbaceous plant surveys take longer than woody plant diversity surveys. Even though the sample plots surveyed in this study evenly covered the entire forest plot and involved the months of July–September, it is still possible that some early spring herbaceous plants were not surveyed. Even so, the results of this study provide a new perspective on patterns of herbaceous plant diversity and their maintenance in warm-temperate deciduous broad-leaf forests in the Donglingshan. We recommend that the study of forest herb communities be strengthened in plant ecology research (Spicer et al., 2022). This goal can be achieved by incorporating herbs into existing long-term study sites (e.g., the ForestGEO and other similar networks), and standardizing survey methods (Crisci et al., 2020). Such a comprehensive understanding of the patterns of diversity and distribution of warm-temperate forest plants would not only add to ecological knowledge but also aid efforts to predict shifts in plant communities under global change and design better conservation and restoration strategies for diverse warm-temperate forest ecosystems.

Acknowledgements

We thank editors and three reviewers for valuable comments that improved the paper. We also thank Dr. Naili Zhang and her group for helping collect soil samples. Yan Zhu was financially supported by the NSF of China (32271614; 31870408); Biological Resources Programme, Chinese Academy of Sciences; State Key Laboratory of Vegetation and Environmental Change of China (Y7206F1016); the Strategic Priority Research Program of the Chinese Academy of Sciences (XDB31030000); National key basic R & D program of China (2017YFA0605100).

CRediT authorship contribution statement

Tingting Deng: Writing – original draft, Methodology, Investigation, Data curation. Qingqing Du: Writing – review & editing. Yan Zhu: Writing – review & editing, Writing – original draft, Supervision, Methodology, Investigation, Funding acquisition, Data curation, Conceptualization. Simon A. Queenborough: Writing – review & editing, Supervision, Methodology, Formal analysis.

Consent to participate

Not applicable.

Ethical approval

Ethics approval was not required for this study.

Availability of data and materials

The datasets generated during the study are available from the corresponding author on reasonable request.

Code availability

The R code used to analyze data in the current study is available from the corresponding author on reasonable request.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that may have influenced the work reported in this paper.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.pld.2025.01.003.