1. Introduction

Beta-diversity reflects changes in species composition among sites (McFadden et al., 2019; Zhong et al., 2021; Jaeger et al., 2022). Theories to explain the beta-diversity fall into two classes: neutral theory and niche theory (Lu et al., 2021; Godsoe et al., 2022). The neutral theory proposes that the species composition of a community is affected by dispersal, and thus the similarity of species composition decreases with geographic distance (Hubbell, 2001; Niu et al., 2009). Niche theory posits that species composition is affected by both environmental factors and species interactions, and that species composition changes with environmental gradients (e.g., climatic distance, elevational distance) (Keddy, 1992; Chase and Leibold, 2003; Godsoe et al., 2022). However, environmental conditions change with geography, entangling the effects of dispersal limitation and niche processes on species composition (Lu, 2021). Thus, to better understand how species composition is regulated, the relative importance of dispersal limitation and niche processes must be disentangled (Legendre et al., 2005; Chen et al., 2010; Wang et al., 2018).

Total beta-diversity has been used to measure changes in community species composition (Soininen et al., 2018; Lu, 2021; Godsoe et al., 2022). Previous studies more focused on total beta-diversity while few studies have examined processes (e.g., species turnover or replacement, species loss or increase) that shape changes in species composition between communities (Si et al., 2017; Soininen et al., 2018). Baselga (2010, 2012) has proposed dividing beta-diversity into two components: the species turnover component and the nestedness-resultant component. The species turnover component represents the replacement of species among different communities, partly due to spatially structured environmental changes and historical processes (Lennon et al., 2001; Qian et al., 2005; Baselga, 2010). The nestedness-resultant component represents the communities with low richness treating as a subset of those with higher richness, partly due to non-random loss or increase of species (Gaston and Blackburn, 2000; Ulrich and Gotelli, 2007; Baselga, 2010; Wang et al., 2010).

Previous studies have demonstrated that beta-diversity of different plant communities are not consistent (Morlon et al., 2008; Chen et al., 2010; Tang et al., 2012a, 2012b). Furthermore, many studies on beta-diversity of forests have neglected coniferous forests, which are important vegetations for maintaining the environment and species diversity of mountains (Wang et al., 2018; Yao et al., 2020; Jaeger et al., 2022).

Larch (Larix Mill.) forests are widely distributed in the mountainous and plateau areas in North and Southwest China. These forests are critical for maintaining the environment and species diversity in these areas. Larch plantations in China are also common as larches are important timber trees. Research on the diversity of larch forests have focused on α-diversity (Wang et al., 2009; Fang et al., 2019, 2022), whereas few studies have explored spatial changes in species composition in these forests. Studies of beta-diversity have helped us understand the mechanisms of community assembly (i.e., mechanisms of forming and sustaining biodiversity) and develop the diversity protection policy (Niu et al., 2009; Rao et al., 2013; Chai and Yue, 2016; Socolar et al., 2016). In this study, we determined the beta-diversity of larch forests based on 483 plots across their whole distribution area in China. Specifically, we asked three questions: (1) how do beta-diversity components of larch forests vary in China? (2) do beta-diversity patterns of larch forests change with life forms (i.e., tree, shrub, and herb)? and (3) what factors regulate the beta-diversity patterns of larch forests?

2. Material and methods 2.1. Field investigation

We defined the larch forest as a plot with at least three larch trees that occupy at least 25% of the relative basal area (Wang et al., 2020; Fang et al., 2022). We investigated 483 larch forest plots, which covered almost all distribution areas of larch forests in China, ranging from 85.14 to 130.84°E longitude and 27.1–52.86°N latitude, and from elevations between 125 and 4399 m (Fig. 1). These plots belong to 13 larch forest types, namely, Larix gmelinii, L. gmelinii var. principis-rupprechtii, L. griffithii, L. himalaica, L. kaempferi, L. kongboensis, L. mastersiana, L. olgensis, L. potaninii, L. potaninii var. australis, L. potaninii var. chinensis, L. sibirica, and L. speciosa forests (Guo et al., 2020; Fang et al., 2022). Except for the L. kaempferi forest plots, all plots were natural forests without apparent human disturbance. Our analysis also included some near-natural L. kaempferi -forest plots without apparent forest management, as L. kaempferi is planted extensively in China.

The plot size varied in the tree layer (600 m², 20 m × 30 m), shrub layer (2 × 100 m²) and herb layer (6 × 1 m²); in addition, whole forests plot size was 600 m² (Fang et al., 2009). For each plot, geographic information was recorded; for the tree layer, all individuals (DBH ≥ 3 cm) were measured and recorded; for shrub and herb layers, along with the investigation of shrubs and herbs for little cases (2 × 100 m² and 6 × 1 m²), all shrub and herb species of the whole plot (600 m²) were recorded (Fang et al., 2009). To avoid the influence of different plot size, in the following analysis we used the 0/1 data of species that belong to the same plot size for trees, shrubs and herbs (600 m²). For more details on the methods employed in our field investigation, please refer to Fang et al. (2009) and Wang et al. (2020).

2.2. Climate data

We extracted the monthly mean temperature and precipitation (MMT and MMP) for the longitude and latitude of plots from the World Climate Database with a spatial resolution of 30 arc seconds (v.2.1, http://www.worldclim.org). The annual potential evapotranspiration (APE, mm) was extracted from the Global Potential Evapo-Transpiration Climate Database with a spatial resolution of 30 arc seconds (http://www.csi.cgiar.org). Based on MMT and MMP, we calculated the mean annual temperature (MAT, ℃), mean temperature of the coldest quarter (MTCQ, ℃), mean temperature of the warmest quarter (MTWQ, ℃), annual precipitation (AP, mm), precipitation of the wettest quarter (PWQ, mm) and precipitation of the driest quarter (PDQ, mm). We used MAT, MTCQ, MTWQ, AP, PWQ, PDQ, and APE to indicate contemporary climatic conditions. The extraction of climatic data was performed in R (v.4.0.0, R Development Core Team, 2020) using the raster package (Hijmans, 2020).

Paleoclimatic conditions were reflected by the anomaly of MAT from the Last Glacial Maximum (about 22, 000 years ago, LGM) to the present (Anomaly.MAT) and the anomaly of AP from LGM to the present (Anomaly.AP) (Araújo et al., 2008). Based on the longitude and latitude of plots, we extracted MAT and AP of LGM under the MPI-ESM-P scenario (Watanabe et al., 2011) from the World Climate Database with a spatial resolution of 2.5 arc minutes (http://www.worldclim.com/paleo-climate1). We calculated Anomaly.MAT and Anomaly.AP by subtracting LGM annual values from contemporary annual values (Araújo et al., 2008).

2.3. Calculation of beta-diversity

The calculation of beta-diversity for larch forests was based on the 0/1 data of species. For each life form (i.e., tree, shrub, and herb), we used the beta.pair function of the betapart package (Baselga, 2012; Baselga et al., 2021) to calculate different components of the beta-diversity: the total beta-diversity (Jaccard dissimilarity index, β_jac, Eq. (1)), the species turnover component (β_jtu, Eq. (2)), and the nestedness-resultant component (β_jne, Eq. (3)).

(1)

(2)

(3)

where a indicates species numbers common to the two communities, while b and c indicated species numbers specific to each of the two communities, respectively.

2.4. Statistical analyses

To assess the redundancy of environmental factors (i.e., Elevation, MAT, MTCQ, MTWQ, AP, PWQ, PDQ, APE, Anomaly.MAT, and Anomaly.AP) and remove factors with higher correlation (Spearman's ρ2 > 0.5) (Fig. S1), we used the varclus function (in the Hmisc package) (Wang et al., 2017; Harrell and Dupont, 2020). For reserved environmental factors, we calculated the total environmental distance and the distances of different environmental factors using the vegdist function of the vegan package, respectively (Fig. S1c) (Oksanen et al., 2019). The current climatic and paleoclimatic distances (see Fig. S1c) were calculated using the vegdist function (Oksanen et al., 2019). Based on the longitude and latitude of plots, we calculated geographic distance using the distm function of the geosphere package (Hijmans, 2019). To disentangle geographic and environmental influences on the beta-diversity of larch forests, we used the partial Mantel test with 999 permutations in the vegan package (Oksanen et al., 2019).

To determine the relative importance of geographic and environmental factors in the beta-diversity of larch forests, we conducted multiple regression analysis with the multiple regression on distance matrices (MRM) approach using the MRM function of the ecodist package (Goslee and Urban, 2007). Before MRM analysis, we standardized geographic and environmental factors using the stdize function included in the MuMIn package (Bartoń, 2020). We conducted permutational multivariate analysis of variance (PERMANOVA) to assess the influence of larch forest types on the beta-diversity using the adonis function of the vegan package (Oksanen et al., 2019).

Geographic, elevational, current climatic, and paleoclimatic distances were transformed to ln(x+1). The Mantel test was used to perform correlation analysis between the matrix of beta-diversity and different distance matrices (ln(x+1) form) of geographic, elevational, current climatic, and paleoclimatic factors by the mantel function, which can be found in the vegan package (Oksanen et al., 2019). To determine the relationships between different components of beta-diversity and geographic, elevational, current climatic, and paleoclimatic distances, respectively, we used the sma function of the smatr package for standard major axis regression (SMA) (Warton et al., 2012; Talbot et al., 2014). All statistical analyses were performed in R (v.4.0.0, R Development Core Team, 2020).

3. Results 3.1. Geographical and climatic patterns of beta-diversity components

For shrubs and herbs, the main component of beta-diversity across all latitudes and elevations within the geographical ranges of the study was species turnover (Fig. 2). Species turnover was also a main component of tree beta-diversity, but only at latitudes between 25 and 50°N and elevations less than 4000 m (Fig. 2). At higher latitudes (50–55°N) and elevations (4000–4500 m), the main component of tree beta-diversity was nestedness-resultant (Fig. 2).

Species turnover was also the main component of beta-diversity in all climatic conditions for different life forms; for shrubs and herbs, the nestedness-resultant component was negligible (Fig. 3). Even under severe climatic conditions (MAT < 0 ℃, AP < 500 mm), the nestedness-resultant component was less than 1/5 of total beta-diversity of trees (Fig. 3).

3.2. Changes in beta-diversity along geographic, elevational, and climatic distances

Mantel tests indicated that the beta-diversity matrix was significantly correlated with the geographic, elevational, and climatic distance matrices (Table S1). SMA indicated that total beta-diversity in different life forms increased with increasing geographic, elevational, current climatic, and paleoclimatic distances (Fig. 4). The rate at which total beta-diversity changed with geographic, elevational, current climatic, and paleoclimatic distances was largest for the tree, followed by the shrub and herb (Fig. 4 and Table S2). These analyses also indicated that total beta-diversity was largest for trees, followed by shrubs and herbs (see intercepts of Fig. 4 and Table S2).

For different life forms, the species turnover component increased whereas the nestedness-resultant component decreased with increasing geographic, elevational, current climatic and paleoclimatic distances (Fig. 5). The rate at which beta-diversity components changed with geographic, elevational, current climatic, and paleoclimatic distances was also the largest for the trees, followed by shrubs and herbs (Fig. 5 and Table S2).

3.3. Relative explanatory strengths of different factors on the beta-diversity

Our analysis indicates that for all life forms total beta-diversity and two of its components were influenced more by geographic distance than by environmental distance (Tables 1 and S3). MRM analysis indicated that the geographic and environmental factors explained 25% of variance in total beta-diversity of trees, 27% of the species turnover component, and 16% of the nestedness-resultant component (Table 1). For shrubs, these factors explained 20%, 18%, and 8% of the variances (Table 1). For herbs, these factors explained 23%, 18%, and 4% of the variances (Table 1). The factor with the best partial regression coefficient of MRM was the geographic distance for total beta-diversity and the two components. Furthermore, partial regression coefficients of current climatic factors were generally larger than those of paleoclimatic factors (Table 1). Correlations between measures of diversity (i.e., total beta-diversity, species turnover, and nestedness-resultant) in different life forms (i.e., trees, shrubs, and herbs) and different larch forest types were significantly different from those between these components and overall larch forests, except for total beta-diversity of trees (Fig. 6 and Table S4). PERMANOVA indicated that larch forest types significantly affected the total beta-diversity and species turnover component of trees (R²: 0.45 and 0.67), shrubs (R²: 0.18 and 0.23), and herbs (R²: 0.15 and 0.19) (Table 2).

For trees, total beta-diversity and two of its components were significantly affected by the distances of geography, elevation, MAT, AP, Anomaly.MAT, and Anomaly.AP (Table 1). The absolute value of the partial regression coefficient of MRM indicated that the most important factors was geographic distance (partial regression coefficient: 0.041, 0.076, and −0.036) for the tree beta-diversity of larch forests, followed by the distance of MAT (0.017, 0.029, and −0.012) (Table 1).

For the shrubs, the factors that significantly affected the total beta-diversity and two components were also the geographical distance and differences in elevation, MAT, AP, Anomaly.MAT, and Anomaly.AP (Table 1). The distance of PDQ also significantly affected the total beta-diversity, but its partial regression coefficient was very small (0.001) (Table 1). Also, the dominant factors for shrub beta-diversity in larch forests was geographic distance (0.013, 0.02, and −0.007), followed by the distance of MAT (0.01, 0.013, and −0.004) (Table 1).

For the herbs, all factors significantly affected the total beta-diversity and species turnover component except that PDQ was not significant for the species turnover component (Table 1). The distances of geography, MAT, Anomaly.MAT, and Anomaly.AP were significantly affected by the nestedness-resultant component (Table 1). The distances of geography and MAT were important for the total beta-diversity and two components of the herb beta-diversity in larch forests; elevational distance was also important for the herb species turnover component (Table 1).

4. Discussion 4.1. Changes in beta-diversity with geographic and climatic distances

Wang et al. (2018) revealed that species replacement (i.e., species turnover) is the predominant component of spatial community dissimilarity for the temperate forests in Northern China and Northern USA. Similarly, we found that species turnover was the major component of beta-diversity for larch forests in China (Figs. 2 and 3). This result is consistent with that found for the global lake aquatic macrophytes (Alahuhta et al., 2017) and that of recent studies on bird beta-diversity (Zeng et al., 2022).

Several studies have found that the total beta-diversity of different plant communities significantly increases with increasing geographic and environmental distances (Li et al., 2016; Wang et al., 2018; He et al., 2022). Similarly, our results showed that the total beta-diversity and the species turnover component of larch forests significantly increased with geographic and climatic distances; however, the nestedness-resultant component significantly decreased across these distances (Figs. 4 and 5). These findings are reasonable given that species turnover is the major component of beta-diversity in China's larch forests (Figs. 2 and 3). Furthermore, the nestedness-resultant component has also been shown to decease along geographic and climatic distances in large-scale angiosperm assemblages (Qian et al., 2020), amphibian communities (Wang et al., 2022) and stream bacteria communities (Li et al., 2020).

Beta-diversity decreases in communities with high dispersal ability (Nekola and White, 1999; Qian, 2009; Chen et al., 2010). Thus, beta-diversity changes faster in communities composed of species with low dispersal ability than in communities composed of species with high dispersal ability (Nekola and White, 1999; Qian, 2009; Chen et al., 2010). In larch forests, we found that beta-diversity of trees changes faster than that of shrubs and herbs in response to geographic distance (Fig. 4). This finding indicates that dispersal ability at lower levels in trees than in shrubs and herbs. Similarly, beta-diversity changes faster in trees than in shrubs and herbs in response to climatic factors (Fig. 4). This finding suggests that the water-energy range is smaller in trees than in shrubs and herbs (Qian et al., 2005, 2009; Liu et al., 2015). Larch forests are typical cold temperate coniferous forests. The tree layer of larch forests largely consists of coniferous species with less species richness. The understory species of these forests are relatively abundant because the transmittance of larch forests is relatively high and provides favorable light conditions for understory plants (Fang et al., 2019, 2020). We found beta-diversity was lower in trees than in shrubs and herbs growing at similar geographic or climatic distances (Fig. 4). This finding is consistent with the differences in species richness between the life forms of larch forests (Fang et al., 2019, 2022).

4.2. Effects of dispersal limitation and niche differentiation on beta-diversity

Previous studies found that geographical distance and environmental divergence significantly affect the beta-diversity of plant communities (Liu et al., 2015; Tang et al., 2021b; He et al., 2022). Furthermore, the explanatory strength of geographic and environmental distances on beta-diversity have been shown to reflect the roles of dispersal limitation and niche differentiation on community composition (Qian, 2005, 2009; Chen et al., 2010). Many studies have shown that the relative strengths of dispersal limitation and niche differentiation depend largely on taxonomic groups (Tang et al., 2012b; Rao et al., 2013; Liu et al., 2015; Li et al., 2016; Viana et al., 2016; He et al., 2022). In this study, partial Mantel tests showed that beta-diversity of larch forests depends most on geographic distance (Table S3). This finding indicates that dispersal limitation best explains the beta-diversity of larch forests compared with niche differentiation. However, the total explanatory strengths of geographic and environmental distances were less than 30 % (Table 1), indicating there are other important factors responsible for beta-diversity in larch forests (Wang et al., 2018; Yao et al., 2020; Jaeger et al., 2022).

Previous studies have indicated that other factors influence the beta-diversity of forests, including assembly processes, spatial structure, and the intensity of regeneration treatments (Wang et al., 2018; Yao et al., 2020; Jaeger et al., 2022). We found that the R² of MRM analysis on total beta-diversity and two components of different larch forest types were significantly different from that of overall larch forests, except for the total beta-diversity of the trees (Fig. 6 and Table S4). In this analysis, in addition to environmental distance, larch forest type (identified by dominant larch species in the forest) also reflected the results of species niche differentiation since the distribution of different larch forest types in China has distinct zonality formed by a lengthy historical process (Liu et al., 2002; Sundaram et al., 2019, 2021; Fang et al., 2020, 2022). Our results showed that larch forest types significantly affect total beta-diversity and the species turnover component of larch forests (Table 2). In other words, niche differentiation also greatly explains the total beta-diversity and species turnover component of larch forests.

MRM analysis has been applied extensively to study the explanatory strength of independent factors in beta-diversity studies (Li et al., 2016; Wang et al., 2017). Here, MRM analysis showed that current climatic factors (especially MAT) were more important than paleoclimatic factors for the beta-diversity of larch forests (Table 1). This finding is consistent with that of research on species richness of larch forests (Fang et al., 2022) and with a study of beta-diversity of plant communities in the Americas (McFadden et al., 2019). Tang et al. (2012a) found that MAT was the most important climatic factor for tree beta-diversity in the mountain forests of China, whereas the annual actual evapotranspiration and AP were the main climatic predictors for shrub beta-diversity. These findings are partly consistent with our finding that MAT is the most important climatic factor for the beta-diversity of trees and shrubs.

5. Conclusions

Here, we used 483 larch forest plots to determine beta-diversity of larch forests in China. Our results showed that the main component of beta-diversity for larch forests is the species turnover component. The change of species turnover component with the geographic and environmental distances was consistent with the total beta-diversity, but the change of nestedness-resultant component was the opposite. These geographic and environmental distance patterns of beta-diversity differed with different life forms of larch forests. Although the influences of geographic distance were bigger than those of environmental distance, larch forest types also influenced beta-diversity. The beta-diversity of larch forests was affected by both of niche differentiation and dispersal limitation.

Acknowledgements

This study was supported by the Major Program for Basic Research Project of Yunnan Province (No. 202101BC070002), the National Natural Science Foundation of China (No. 32201426, No. 31988102), and the National Science and Technology Basic Project of China (No. 2015FY210200). Thanks for the work of Chunyun Liu and many others in the field investigation, and the assistance of the local forestry bureau.

Author contributions

Jiangling Zhu: Project administration. Chengjun Ji: Investigation. Qiong Cai: Investigation. Wenjing Fang: Writing – original draft, Writing – review & editing. Jingyun Fang: Conceptualization, Funding acquisition. Zhiyao Tang: Conceptualization, Writing – review & editing

Data statement

Data 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 could have appeared to influence 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.2023.10.003.