Factors that shape the elevational patterns of plant diversity in the Yatsugatake Mountains, Japan

Abstract Elevation is involved in determining plant diversity in montane ecosystems. This study examined whether the distribution of plants in the Yatsugatake Mountains, central Japan, substantiated hypotheses associated with an elevational diversity gradient. Species richness of trees, shrubs, herbs, ferns, and bryophytes was investigated in study plots established at 200‐m elevational intervals from 1,800 to 2,800 m. The changes in plant diversity (alpha and beta diversities, plant functional types, and elevational ranges) with elevation were analyzed in relation to climatic factors and elevational diversity gradient hypotheses, that is, mass effect, mid‐domain effect, and Rapoport's elevational rule. In addition, the elevational patterns of dominance of plant functional types were also analyzed. A comparison of alpha and beta diversities revealed that different plant groups responded variably to elevation; the alpha diversity of trees and ferns decreased, that of herbs increased, whereas the alpha diversity of shrubs and bryophytes showed a U‐shaped relationship and a hump‐shaped pattern. The beta diversity of shrubs, herbs, and bryophytes increased above the subalpine–alpine ecotone. In accordance with these changes, the dominance of evergreen shrubs and graminoids increased above this ecotone, whereas that of evergreen trees and liverworts decreased. None of the plant groups showed a wide elevational range at higher elevations. These elevational patterns of plant groups were explained by climatic factors, and not by elevational diversity gradient hypotheses. Of note, the changes in the dominance of plant groups with elevation can be attributed to plant–plant interactions via competition for light and the changes in physical habitat. These interactions could alter the elevational diversity gradient shaped by climatic factors.


| INTRODUC TI ON
Elevation is a key factor that determines the patterns of plant diversity in mountainous areas (Jiang et al., 2018;Körner, 2007;Lee et al., 2013;Lee & La Roi, 1979). The patterns of plant diversity are often explained by several hypotheses for elevational diversity gradient such as the mass effect, mid-domain effect, and Rapoport's elevational rule.
According to the mass effect hypothesis, a species disperses from source areas occupied by larger populations to the adjacent sink areas; however, the dispersed species cannot maintain viable populations in the sink areas Shmida & Wilson, 1985). These temporal populations increase species richness around ecotones where two different communities meet. Unlike this hypothesis, the mid-domain effect hypothesis assumes a random distribution of species within a geometrically constrained domain such as that between the summit and base of a mountain. Consequently, species richness peaks in the middle of the domain because of the increasing overlap of species distribution toward the center (Colwell & Lees, 2000;Colwell et al., 2004). The Rapoport's elevational rule postulates that species at higher elevations can withstand a broad range of climatic conditions across a high range of elevations (Stevens, 1992).
However, explanation of plant elevational patterns using these hypotheses is often debatable. This is because plant groups show disparate responses to elevation, and even plants within a group present dissimilar elevational distribution in different areas (Bhattarai & Vetaas, 2003;Bruun et al., 2006;Grytnes et al., 2006;Miyajima et al., 2007;Sánchez-González & López-Mata, 2005). Hence, the applicability and limitations of these hypotheses for elevational diversity gradient should be further explored through careful examination of relationships between plant distribution and elevation.
Such an examination in mountainous areas in Japan is of interest because of the characteristic vegetation at higher elevations. These areas experience strong winds and heavy snowfall owing to the influence of jet streams and winter monsoons (Manabe, 1957;Riehl, 1962;Ueda et al., 2015). The severe climatic conditions enable a single Japanese stone pine tree species (Pinus pumila Regel) to become the dominant vegetation above the subalpine-alpine ecotone owing to its resistance to wind and snow (Okitsu, 1984). The dominance of P. pumila could alter the elevational patterns of plant diversity from those predicted based on the above-mentioned hypotheses (mass effect, mid-domain effect, and Rapoport's elevational rule). This is because the subalpine-alpine ecotone is otherwise dominated by a mixture of alpine and subalpine plants according to the mass effect. From other perspectives, the dominance of P. pumila could reduce the randomness of species distribution as postulated by the mid-domain effect and hinder the dispersal of alpine species in lower elevations against the assumption based on the Rapoport's elevational rule. Hence, testing the elevational hypotheses in this region can contribute to the understanding of the applicability and limitations of these hypotheses.
In this study, the elevational distribution of multiple plant groups in the Yatsugatake Mountains, Japan, was evaluated to determine whether the hypotheses for elevational diversity gradient (i.e., the mass effect, mid-domain effect, and Rapoport's elevational rule)

| Study site
This study was conducted in the Yatsugatake Mountains ( Figure 1).
The highest peak in these mountains is the 2,899 m summit of Mount Akadake (35°58′15″N, 138°22′12″E). The vegetation in the Yatsugatake Mountains is largely classified into the following four types: temperate deciduous forest dominated by Quercus crispula and Larix kaempferi below ~1,800 m; subalpine conifer forest dominated by Tsuga diversifolia, Abies veitchii, and Abies mariesii f. hayachinensis at ~1,800-2,600 m; alpine dwarf pine scrub dominated by P. pumila at ~2,600-2,800 m; and alpine meadow with turf-rock vegetation at ~2,800-2,900 m. The highest and lowest mean temperatures in the nearest weather station at 1,350 m are 19.2°C in August and −5.3°C in January, and the mean precipitation is 1,439.9 mm/ year (Japan Meteorological Agency, 2020).

| Field survey
Twelve 10 × 10 m study plots were established at 200-m elevational intervals from 1,800 to 2,800 m (i.e., 1,800, 2,000, 2,200, To examine the influence of climate on plant diversity, temperature and relative humidity (RH) were measured in each plot at ~5 cm above the ground at 4 hr intervals throughout the year, using HOBO U23 Pro v2 temperature/relative humidity data loggers (Onset Computer Corporation). To reduce the influence of sunlight on the measurements, the data loggers were covered with a CO-RS1 solar radiation shield (Onset Computer Corporation). Using the temperature and RH data, the following climatic variables were calculated: the mean annual temperature (Temp ann ), mean temperature during the growing season (Temp grow ), mean temperature during the snow season (Temp snow ), mean annual RH (RH ann ), mean RH during the growing season (RH grow ), mean RH during the snow season (RH snow ), and duration of the snow season (snow cover). A series of snowcovered days with over 90% RH was defined as the snow season.
This definition was based on a primary experiment in which data loggers were buried under a snow pack. The remaining days were defined as the growing season. RH snow was not used in the following analysis because this variable is nearly uniform according to its definition.

| Statistical analyses
Alpha diversity, beta diversity, dominance of functional types, and elevational range of each plant group were calculated to examine the validity of the hypotheses for elevational diversity gradient. Alpha diversity was defined as the species richness of each plant group within a plot, and its correlation with elevation was fitted using generalized linear models (GLMs) with log-link function and Poisson distribution. Models were selected as follows (model selection 1). Both linear and second-order polynomial models were constructed, and the final model with significant coefficients and the lowest secondorder Akaike information criterion (AICc) value was selected. If a model did not have significant coefficients, then the model with the lowest AICc value was adopted.
The influence of climatic factors on alpha diversity was also analyzed using GLMs with log-link function and Poisson distribution.
The final model was selected as follows (model selection 2). Climatic factors that strongly correlated with alpha diversity (r > 0.5) were selected as explanatory variables. The models with different combinations of these explanatory variables were then constructed. Among these models, the final model had significant coefficients and the lowest AICc value.
Beta diversity was evaluated using the β sim index, considering that the performance of this metric was excellent under various criteria (Koleff et al., 2003). One of the advantages of this index is that it focuses more on compositional differences between the focal and neighboring quadrats than on the differences in species richness (Koleff et al., 2003). The β sim index was calculated between the adjacent plots as follows: F I G U R E 1 Location of the study site and plots, and the major vegetation types in the Yatsugatake Mountains, central Japan: (a) location of the Yatsugatake Mountains, (b) location of the study plots, (c) alpine meadow, (d) alpine dwarf pine scrub (Pinus pumila), (e) subalpine conifer forest, and (f) temperate deciduous forest. A part of this figure was adapted from where a is the total number of species found in both plots, b is the number of species that occur in the first plot but not in the second plot, and c is the number of species that occur in the second plot but not in the first plot. The changes in beta diversity with elevation were fitted by linear models ( To test the applicability of the mid-domain effect to elevational distribution, the discrepancy between the observed results and those predicted based on the mid-domain effect was determined using the null model (discrete mid-domain effect model) (Colwell & Hurtt, 1994). The predicted mean species richness was calculated by 9,999 Monte Carlo simulations using RangeModel software version 5.0 (Colwell, 2006). The predicted values were plotted against elevation and correlated with the observed alpha diversity using Pearson's correlation.
Finally, the applicability of the Rapoport's elevational rule to elevational distribution was examined based on the elevational range of each plant group. The elevational range was defined as the mean difference between the highest and lowest elevations of all the recorded species in each plot and was calculated as follows: where n is the total number of species recorded in a plot, H i is the highest elevation where the ith species was recorded, and L i is the lowest elevation where the ith species was recorded. The highest and lowest elevations were based on elevation records in the study site, herbarium specimens (Nagoya University Museum; NUM-Bt) collected in central Japan (Chubu, Kinki, and Kanto regions), plant diversity databases (Biodiversity Centre of Japan, 2020), and plant survey reports from the studied region (Akiyama, 1983(Akiyama, , 1984Hattori, 1958;Inoue, 1981;Kodama, 1971Kodama, , 1972Masuzaki & Katagiri, 2010;Takaki et al., 1970).
The changes in elevational ranges with elevation were examined using LMs. This modeling was performed with the same procedure for LMs between beta diversity and elevation.

| Species richness, elevational distribution, and climatic factors
Two hundred and sixty plant species (20 tree species, 17 shrubs, 49 herbs, 11 ferns, and 163 bryophytes) were recorded in the study plots.

| Alpha diversity
The alpha diversity of the plant groups varied with elevational distribution (Figure 2). All coefficients, except those for ferns, in

| Mid-domain effect
Species richness (alpha diversity) was predicted using a discrete middomain effect model to examine the applicability of this hypothesis to the sampling site (Figure 2, Appendix S4). In all plant groups, the To illustrate the alpha diversity, generalized linear models were used to explore the pertinent changes with elevation. When the coefficient was significant, a 95% confidence interval was calculated and colored Abbreviations: Res. var = response variable, Exp. var = explanatory variable, SE = standard error, R 2 = Nagelkerke's R 2 , Temp ann = mean annual temperature, Temp snow = mean temperature during the snow season, RH ann = mean annual relative humidity, RH grow = mean relative humidity during the growing season, and Snow cover = duration of snow season.

TA B L E 1 Generalized linear model results between alpha diversity and climate variables
correlations between the observed species richness and those predicted based on the mid-domain effect were not positively significant (Table 2).

| Beta diversity
When beta diversity (based on the β sim index) was plotted against elevation ( Figure 3, Appendix S5), the β sim index of shrubs and ferns was not calculated in several plots because of the absence of these plant groups (Figure 3c,e). The β sim of shrubs and herbs substantially increased and that of bryophytes slightly increased above the subalpine-alpine ecotone (Figure 3c,d,f).
The results of LMs between the β sim index and environmental factors are presented in Table 3. Similar to the results of alpha diversity, snow cover and RH grow exerted a significant influence on the beta diversity of herbs and bryophytes. LMs with significant coefficients were not constructed for the beta diversity of trees and ferns.

| Functional types
Elevational patterns of dominance of plant functional types are shown in Figure 4 and Appendix S6. The changes in the dominance of plant functional types were notable around 2,600-2,800 m ( Figure 4). The dominance of evergreen trees and liverworts decreased in this elevation range (Figure 4a,d), whereas that of evergreen shrubs and graminoids increased (Figure 4b,c).
Regarding the relationships between functional types, the dominance of evergreen trees had a negative influence on that of evergreen shrubs and graminoids whereas it positively affected the dominance of liverworts (Table 4).

| Elevational range
The mean elevational range of each plant group was calculated and plotted as a function of elevation ( Figure 5, Appendix S7). None of the plant groups presented a wider elevational range at higher elevations. Liverworts groups showed varied responses to the elevational gradients and climatic factors in the study site. Regarding the elevational changes in alpha diversity, only shrubs presented higher alpha diversity in the subalpine-alpine ecotone, as predicted based on the mass effect.

| D ISCUSS I ON
However, none of the plant groups exhibited elevational distribution patterns corresponding to those predicted based on the mid-domain effect or Rapoport's elevational rule. In contrast, the alpha diversity of all plant groups was explained by the climatic factors. The dominance of plant functional types correlated with each other.

| Elevational patterns of plants
The alpha diversity of trees decreased with elevation (Figure 2b), and this elevational pattern was explained by Temp ann . These results are supported by the findings that the elevational range of trees can be determined by temperature through changes in their physiochemical activities (Körner, 2003). This obvious influence of Temp ann on tree diversity might have led to the relatively constant changes in the beta diversity of trees compared with that of other plant groups ( Figure 3b) owing to a linear decrease in Temp ann with elevation.
Similar to the results for trees, the alpha diversity of ferns decreased with elevation (Figure 2e), which can be attributed to Temp snow in the GLM. These results are consistent with a decrease in fern richness at higher elevation and lower temperature in central Japan (Tanaka & Sato, 2013, 2014. In contrast to the results for the trees and ferns, the alpha diversity of herbs positively correlated with elevation ( Figure 2d). Given that snow cover positively affected herb diversity in the GLM, the species richness of herbs increased with the environmental changes related to snow cover, such as soil moisture and canopy closure. Long periods of snow cover augment soil moisture (Hardy et al., 2001) essential for the proliferation of herbs in arctic-alpine habitats (Litaor et al., 2008;Nabe-Nielsen et al., 2017;Roux et al., 2013 (Song et al., 2017), which improves light intensity and reduces litter accumulation on the forest floor. These changes in soil moisture and canopy closure could enhance herb richness by guaranteeing the photosynthesis and germination of herbs.
The alpha diversity of shrubs rapidly increased from the subalpine-alpine to the alpine zone ( Figure 2c). This elevational pattern was explained by the RH grow in the GLM. Considering that canopy closure reduces shrub diversity (Speziale et al., 2010) and intensifies air humidity (Jung et al., 2017), RH grow may be selected in the GLM as a surrogate to represent the negative influence of canopy closure on shrub richness. Accordingly, this change in canopy closure from dense to open could have caused the rapid increase in the beta diversity of shrubs above the subalpine-alpine ecotone (Figure 3c).
Only the bryophytes showed a hump-shaped change in alpha diversity with an increase in elevation (Figure 2f). The GLM results demonstrated that this pattern could be mainly attributed to RH grow .
This relationship is consistent with the results of a previous study, which suggested that water stress is a major determinant of bryophyte diversity (Grau et al., 2007). The negative correlation between RH grow and bryophyte beta diversity (

| Functional types and plant-plant interactions
Here, the elevational patterns of plant diversity are further discussed in relation to plant functional types. Evergreen shrub species (P. pumila) have started to dominate the subalpine-alpine ecotone instead of evergreen conifer trees (Figure 4a,b). This replacement has caused a reduction in canopy closure above the ecotone, as indicated by the negative correlation between the dominance of evergreen trees and evergreen shrubs (Table 4).
In accordance with the decrease in evergreen trees, the dominance of graminoids increased but that of liverworts decreased (Figure 4c,d).
These correlations can be related to improved light availability and increased drought stress owing to the change from a closed to an open canopy. Graminoids dominate open forests (Roberts & Zhu, 2002;Thomas et al., 1999) and are more resistant to drought stress than forbs (Rosbakh et al., 2017), whereas liverworts are more shade adapted (Marschall & Proctor, 2004) and are more vulnerable to drought stress than mosses (Grau et al., 2007;Jan & Wolf, 1993). These plant-plant interactions resulting from the change in canopy closure would affect the species composition, altering elevational patterns of plants influenced by climatic factors. These interactions might then facilitate species turnover around subalpine-alpine ecotone, and increase the beta diversity of shrubs, herbs, and bryophytes (Figure 3c,d,f).

| Mass effect on elevational patterns
The validity of Hypothesis 1 is assessed based on an analysis of elevational patterns of the alpha diversity of plant groups. Among the evaluated plant groups, only shrubs showed a high alpha diversity in the vicinity of the subalpine-alpine ecotone (Figure 2), and all elevational patterns, including those of shrubs, could be explained by climatic factors and plant-plant interactions (Sections 4.1 and 4.2).
Thus, the diversity of plants did not correspond to that predicted based on the mass effect. The plant-plant interactions might limit the predicted increase in species richness in the subalpine-alpine ecotone, given that the dominance of P. pumila inhibits the establishment of small plants by shading.

| Mid-domain effect and the Rapoport's elevational rule on elevational patterns
The validity of Hypotheses 2 and 3 is discussed through a comparison between the predicted and actual plant diversity. The alpha diversity of all plant groups differed from that predicted based on the mid-domain effect hypothesis ( Figure 2, Table 2). On the contrary, a higher alpha diversity of trees and bryophytes at lower elevations (Figure 2b,f) seems to validate the Rapoport's elevational rule.
However, considering that none of the plant groups showed a wider elevational range in the alpine area ( Figure 5), the Rapoport's elevational rule cannot be applied to the elevational patterns. Thus, both Hypotheses 2 and 3 were denied according to these results.
The lack of applicability of the mid-domain effect and Rapoport's rule to elevational patterns can be related to the simple assumptions of these hypotheses; the former excludes any effects of gradients on the expected patterns (Colwell et al., 2004), whereas the latter fails when species distribution is affected by the interaction of several factors (Bhattarai & Vetaas, 2006;Grau et al., 2007).
Specific to the study site, the climate and plant-plant interactions at higher elevations are characterized by severe winter climates and the resultant dominance of P. pumila above the subalpine-alpine ecotones. These characteristics can strengthen the relationships of species distribution with climate and vegetation, and thereby increase discrepancies between the actual and predicted elevational patterns by the mid-domain effect hypothesis and Rapoport's elevational rule.

| CON CLUS IONS
In summary, the elevational patterns of plant diversity in the Yatsugatake Mountains were explained by climate and plant-plant interactions, and not by hypotheses for elevational diversity distribution (Hypothesis 1-3). Therefore, these results correspond to Hypothesis 4, indicating that the applicability of hypotheses for elevational diversity distribution largely depends on the environment and biota in the targeted study site.
Furthermore, this study emphasizes the importance of studies on elevational patterns using multiple plant groups and multiple indices of plant diversity (e.g., alpha and beta diversities, and functional types). These groups and indices are necessary to understand the characteristics of plants' responses to elevation and their interactions that influence elevational diversity distribution.

ACK N OWLED G M ENTS
The author is grateful to Taichi Watanabe (Chiba Prefectural Makuhari Sohgoh High School) for his support in the field survey of vascular plants. The author acknowledges the financial support for this research received from the Japan Society for the Promotion of Science-Grant-in-Aid for Young Scientists (B) (grant number 24710029).

CO N FLI C T O F I NTE R E S T
The author declares no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The datasets used in this study are available in the Supporting information.