Hong Zhu·Xiangui Yi·Yongfu Li·Yifan Duan·Xianrong Wang·Libing Zhang

Abstract Subtropical forest in China has received much attention due to its complex geologic environment and bioclimatic heterogeneity.There have been very few studies addressing which climatic factors have shaped both distribution patterns and niche differentiation of species from this region.It also remains unclear whether phylogenetic niche conservatism retains in plant species from this biodiversityrich subtropical region in China.In this study,we used geographic occurrence records and bioclimatic factors of Prunus dielsiana (Rosaceae),a wild cherry species,combined with the classical ENM-based DIVA-GIS software to access contemporary distribution and richness patterns of its natural populations.The current distribution of P.dielsiana occupied a relatively wide range but exhibited an uneven pattern eastward in general,and the core distribution zone of its populations are projected to concentrate in the Wushan and Wuling Mountain ranges of western China.Hydrothermic variables,particularly the Temperature Seasonality (bio4)are screened out quantitatively to be the most influential factors that have shaped the current geographical patterns of P.dielsiana.By comparison with other sympatric families,climatic niche at regional scale showed a pattern of phylogenetic niche conservatism within cherry species of Rosaceae.The effect of habitat filtering from altitude is more significant than those of longitude and latitude.We conclude that habitat filtering dominated by limiting hydrothermic factors is the primary driving process of the diversity pattern of P.dielsiana in subtropical China.

Keywords BIOCLIM·Climatic adaptation·Most limiting factors·Phylogenetic niche conservatism·Species distribution modeling

Introduction

Understanding the geographical distribution patterns for target species and the relationships they form with the environment across space is crucial in biodiversity conservation and habitat restoration (Fariasa et al.2017).Climate is an important determinant that can significantly influence the extent and distribution pattern of the plant species at a landscape level (Mantyka-pringle et al.2003;Retuerto and Carballeira 2004).However,a majority of these bioclimatic factors are difficult to be measured or to be quantified,and many previous studies mainly focused on single factors or executed with community-based approaches at a small scale (Yang et al.2018).Thus,the associations between them and plant distribution were largely uncertain (Pulliam 2000).Traditional methods can no longer meet the niche requirements of qualitative and quantitative predictions of distribution patterns of plant species.

In recent years,simulating the potential distribution of species has increasingly become a trend in the study of plant ecology and biogeography (Byeon et al.2018).Species distribution models (SDMs),integrated the application of geographic information systems,statistics,and ecology,have emerged as a reliable and widely used tool.Once combined with the increasing accumulation of global biodiversity databases in digital format,they will achieve considerable development.Accordingly,several algorithms of SDMs,e.g.,DOMAIN (Carpenter et al.1993),GARP (Stockwell 1999),MaxEnt (Phillips 2006),and CLIMAX (Sutherst et al.2007) have launched and been updated through time.Among them,the BIOCLIM (Hijmans et al.2001) has been for many years one of the pioneering SDMs packages worldwide owing to its open access and sample operation.It has been widely used in many fields,such as monitoring pest or biological invasion (Ganeshaiah et al.2003;Ning et al.2019),evaluating threatened or endangered species (Raina et al.2014;Xie et al.2020),and forecasting the biodiversity responses to global climate changes (Bellard et al.2012;Huang et al.2019).

As one of the biodiversity hotspots in the world,the subtropical forest in China has received much attention due to the high species diversity and long evolutionary history,related to its complex geologic environment and bioclimatic heterogeneity (López-Pujol et al.2013).Investigation of the relationships between geographical patterns in species diversity and contemporary environment may help to uncover the influences of evolutionary history on species diversity (Ricklefs 2006).Earlier studies partially confirmed the phylogenetic niche conservatism (PNC) of plant species in the subtropical forest in China–closely related species have similar niche requirements (Peterson et al.1999).Recent studies also demonstrated that precipitation plays the most important role affecting both distribution patterns and niche differentiation at large or regional scale (Wang et al.2014;Zou et al.2019).However,it remains unclear whether these findings will hold with additional and more diverse species investigated from this region.We are particularly interested in which hydrothermic factor(s),e.g.,annual precipitation,precipitation of warmest quarter,or annual mean temperature played the crucial role in limiting the distribution pattern and niche differentiation of plant species in this biodiversity-rich region in the subtropical forest in China.

Wild cherry species,Prunus dielsiana(Schneid.) Yu and Li,a member ofPrunussubgenusCerasus in Rosaceae (Li and Bartholomew 2003),offers a particularly promising study system.It is a deciduous species endemic to the subtropical forest of China (Schneid 1905).As an early flowering tree,it usually grows to 5–10 m tall,characterized by having a beautiful tree shape,graceful leaves,and gorgeous inflorescence (Fig.S1).Thus,this species is thought to be a promising material for the future genetic breeding industry with ornamental and economic values (Zhao et al.2019).Like many other flowering trees,P.dielsianaalso is known as an excellent nectar plant resource in the early spring season by attracting beneficial insects.Notably,its fruits are readily eaten by numerous kinds of mammals and birds in the harsh winter season,contributing to maintaining the forest community’s stability and diversity (Zhou et al.2008).Besides,this species may serve as an alternative recommended plant material for restoration and reconstruction of urban forest vegetation or environmental restoration in mining areas (Zhao et al.2003).What’s more,a recent study based on partial natural populations further indicated a gradient variation pattern dominated by longitude geographically in terms of its leaf phenotypic variation (Zhu et al.2018).Despite these preliminary reports,the scientific evaluation ofP.dielsiana’s distributional patterns and ecological characteristics across the whole ranges remains limited.Accordingly,this species has been categorized as Data Deficient (DD) plants on the IUCN Red List (Rhodes et al.2016).

In this study,the classical ENM-based DIVA-GIS software was used to map the geographic distribution ofPrunus dielsianato quantify the relationships between vegetation distribution and bioclimatic factors at a national scale.We focused on the following objectives:(a) to uncover the spatial patterns ofP.dielsianaand the current richness center of this species;(b) to screen the critical bioclimatic variables that shaped this above patterns;(c) to detect the niche signal responding to the geographic aspect;and (d) to test whether or not the PNC hypothesis applies in typical species of subtropical China at region scale.

Materials and methods

Collection of species occurrence data

The occurrence data ofPrunus dielsianawere obtained from herbarium records and field surveys collectively.Among them,62 records were retrieved from the platform of national specimen information infrastructure,namely the Chinese Virtual Herbarium (CVH;http://www.cvh.ac.cn/)and the Global Biodiversity Information Facility (GBIF;http://www.gbif.org/).The remaining 25 individuals were based on voucher specimens with GPS coordinates collected from field explorations in recent decades by members of our project team.To our knowledge,though,there are other public records,but none of them has provided detailed collection information.After the removal of ambiguous or repetitive information and other cultivated records from botanical gardens,parks or orchards,our final dataset consisted of 87 occurrence records in total for subsequent analysis.The analyses were confined to the study region of 102.048°−121.544° E,24.725°−31.648° N,from 1932 to 2017 (Table S1,Fig.1).

Fig.1 Research region and total occurrence points (n=87) of P.dielsiana in subtropical China.The blue circles represent the historical specimen records from herbaria points and literature (n1) and the red circles one represent our own field survey points (n2).Main mountain ranges (Mts.) are indicated in orange mentioned in this paper were also labeled.Provinces are shown in black

Bioclimatic data sources

We downloaded the current bioclimatic data (1970–2000)from the WorldClim version 2 (http://www.world clim.org/versi on2) at 2.5 arc-minutes spatial resolution,which was equivalent to approximately 5 km2at the equator.We considered a set of 19 standard bioclimatic variables derived from globally interpolated datasets as the potential predictors extracted through the DIVA-GIS v.7.5 software(Babasaheb et al.2014).The detailed variables were listed as follows:Annual Mean Temperature (bio1),Mean Diurnal Temperature Range (bio2),Isothermality [(bio2/bio7,STD × 100) (bio3)],Temperature Seasonality [(STD × 100)(bio4)],Maximum Temperature of Warmest Month (bio5),Minimum Temperature of Coldest Month (bio6),Temperature Annual Range (bio7),Mean Temperature of Wettest Quarter (bio8),Mean Temperature of Driest Quarter (bio9),Mean Temperature of Warmest Quarter (bio10),Mean Temperature of Coldest Quarter (bio11),Annual Precipitation(bio12),Precipitation of Wettest Month (bio13),Precipitation of Driest Month (bio14),Precipitation Seasonality[(CV) (bio15)],Precipitation of Wettest Quarter (bio16),Precipitation of Driest Quarter (bio17),Precipitation of Warmest Quarter (bio18),and Precipitation of Coldest Quarter (bio19).

Model manipulation and statistical analysis

We modeled the results on the base of China’s provincial administrative map sourced from the National Geomatics Center of China (NGCC;http://ngcc.sbsm.gov.cn/),with topographic data at a 1:4,000,000 scale.Coordinates ofPrunus dielsianarecords were uploaded into DIVA-GIS software and then coupled with 19 bioclimatic variables and the BIOCLIM model.The prediction of potentialP.dielsianadistribution patterns and richness pattern determination,as well as the forecast of the most limiting factors,were all simulated and presented using DIVA-GIS software.

Bioclimatic variables evaluation and comparison

The standard 19 bioclimatic variables in determining the geographic distribution ofPrunus dielsianawere all extracted by DIVA-GIS.Then,the entire dataset was analyzed to reduce the multicollinearity by principal component analysis (PCA) using the PAST3 v.3.15 software package(http://folk.uio.no/ohamm er/past/).Those components with a score ＞ 0.4 in PC1 and PC2 were identified as critical bioclimatic factors.The cumulative frequency curves were calculated and drawn by DIVA-GIS.Those continuous regions from 10% to 90% of cumulative frequencies were regarded as the proper growth ranges forP.dielsiana,following the guidelines applied by Xie et al.(2020).To explore the relationships of major geographical factors,Pearson’s correlation analysis of the dominant bioclimatic factors with elevation,latitude,and longitude was conducted using the SPSS v.16.0 statistical software.Furthermore,to compare the climatic similarity with ecologically adjacent species reported from subtropical China,the non-metric multidimensional scaling (NMDS) analysis and minimum spanning tree (MST) were simultaneously conducted using the Euclidean similarity algorithm in PAST3 v.3.15 software package.

Model evaluation

To assess the accuracy of the BIOCLIM model with the native range,a stack data set was created to build the area under the ROC (receiver operating characteristic) curve,(AUC).The AUC value is a graphical method and independent of threshold effects,that represents the relationships between the false-positive fraction (1-specificity) and the sensitivity for the arrangement of threshold effects calculated within the program (Fielding and Bell 1997).When running the model,random data partitioning 75% of the training data and 25% of the testing data were simulated and repeated three times.Meanwhile,the Kappa statistic(K),another most frequently used prediction accuracy index(Robertson et al.2003),was also applied.Both AUC and Kappa have values range from 0 to 1;the higher the values,the closer the model agrees with the data.Generally,if the AUC is higher than 0.5,the result is better than a random performance event (Yang et al.2018).

Results

Potential distribution pattern modeling

The occurrence data revealed thatPrunus dielsiananaturally occurred mainly in mountainous forests of the subtropical zone in China and occurred in 12 provinces of China,namely Anhui,Chongqing,Fujian,Guangdong,Guangxi,Guizhou,Hubei,Hunan,Jiangxi,Sichuan,Yunnan,and Zhejiang.Its suitable area was along with the Yangtze River from west to east,across the subtropical evergreen broad-leaved forest(EBLF) region.The southernmost known populations were located in Nanling Mountains (Fig.1).As shown in Fig.2,the output layers of BIOCLIM model with grid cells in different colors represented six classes as a whole,associated with the threshold values ranked as follows,not suitable:white (0–5.5%),low:green (5.6–11.0%),medium:dark green (11.1–16.5%),high:yellow (16.6–20.0%),extremely high:orange (20.1–27.5%),and optimum:red (27.6–33.0%).In general,the populations ofP.dielsianawere geographically unevenly distributed,and there was a vast majority of these potential areas dominating the west region and its surrounding areas,mainly on the border among four Provinces:Chongqing,Guizhou,Hubei,and Hunan.They were followed by the Yunnan-Guizhou Plateau and its adjacent area.By contrast,the remaining potentially suitable areas were scattered eastward (shown in green).The grid map of richness pattern determination was further applied in a 9 × 9 neighborhood grid size,with the richness values ranging from one to six.Both simulations projected the occurrence of the greatest species richness value (red grid cells) ofP.dielsianain the regions of Bashan Mountains and Wuling Mountains (Fig.3).

Fig.2 Projected distribution of Prunus dielsiana in subtropical China obtained using the BIOCLIM model implemented in DIVA-GIS software.Different colors from white to red refer to different values of probalility of occurrence from low to high

Fig.3 Current richness determination of Prunus dielsiana with six gradients obtained using analysis/point to grid implemented in DIVAGIS software.Different colors from white to red refer to the values of

Key bioclimatic factors and suitable ranges

The results of PCA analysis quantified the contribution rate of the first two principal components,which accounted for 83.75% (PC1) and 11.83% (PC2) of the variance,respectively.The accumulated contribution rates reached up to 95.58%.Hence,it was able to explain the majority of the variation information of the 19 bioclimatic factors.In the first principal component (PC1),the Annual Precipitation (bio12)and the Precipitation of Wettest Quarter (bio16) exhibited the highest coefficients,0.8293 and 0.4146,respectively,and both factors showed significant positive correlations with PC1,indicating that the precipitation was the main driving factor.The second principal component (PC2) exhibited a negative correlation with the Precipitation of Warmest Quarter (bio18),and a positive correlation with the Temperature Seasonality (bio4),both reflecting the role of heat variation as the secondary driving factor (Table 1,Fig.S2).Presentation of the cumulative frequency curve showed that the suitable range of four key bioclimatic factors approximated to a normal distribution as follows:bio12 (1022–1729 mm),bio16 (480–808 mm),bio8 (460–641 mm),and bio4 (647-866 STD × 100) (Fig.S3).

The most limiting factors and the correlation with main geographical factors

The distribution pattern ofPrunus dielsianain the area outside the blue-dotted line on the map (Fig.4) of the BIOCLIM most-limiting factors showed that the Temperature Seasonality (bio4) was one of the most limiting factors,which determined the northern and southern boundaries of the distribution ofP.dielsiana.By contrast,the distribution pattern ofP.dielsianain the regions inside the blue-dotted line was much more complicated,even though the bioclimatic variables applied reduced from 19 to 9.Some precipitation variables,such as the Precipitation Seasonality(bio15),the Precipitation of Wettest Quarter (bio16),and the Precipitation of Coldest Quarter (bio19),became the factors which most limited the distribution ofP.dielsianainside the blue-dotted line on the map.The Pearson correlation analysis of dominant climatic factors with elevation,latitude,and longitude further showed that bio4 was positively correlated with latitude and longitude (P＜ 0.01),while negatively correlated with elevation (P＜ 0.01),and their correlation coeffi-cients were 0.485,0.519 and −0.706,respectively (Table 2).

Comparison of niche-related climatic requirements of subtropical species in China

Results of NMDS and MST analysis (Fig.5) detected two groups in terms of bioclimatic differentiation.Group I only consisted of species in the family Rosaceae.Five species of Prunus and one species of Sorbus in Rosaceae were included.Group II consisted of species of Lauraceae and Magnoliaceae.Five species of Lindera in Lauraceae were included.In Magnoliaceae onlyYulania liliiflorawas included.Climatic requirements were listed in Table S2.richness gradients from the squares from white to red indicates values from low to high

Table 1 Coefficients of the former two principle components among the 19 bioclimatic variables

Model performance

The degrees of agreement between the observed and predicted values are shown in Fig.S4 A,B.The calculated results of the ROC curve test (AUC=0.975) and Kappa values (k=0.870) indicated a high performance for the BIOCLIM models ofPrunus dielsiana.

Discussion

Model evaluation and explanation

In this study,we simulated and visualized the potential distribution pattern ofPrunus dielsianausing the classical SDMs software package (BIOCLIM) and characterized its bioclimatic features for the first time.A quantitative assessment of model performance is vital to determining the suitability of the model for this application (Pearson 2010),and the accuracy of prediction varies,depending on those species’ ecological characteristics and different algorithms(Babar et al.2012).Thus,we integrated two classical statistical methods of model performance in ecology,via the ROC curve as a threshold-independent and the kappa as a threshold-dependent measure of predictive accuracy,respectively.The values of AUC and Kappa are both high,indicating that the BIOCLIM model achieved excellent performance in simulating the distribution ofP.cerasoides.When compared with other cases of cherry species such asPrunus cerasoides(AUC=0.799) vs.P.campanulata(AUC=0.816) andP.xueluoensis(AUC=0.751)(Table S2),our modeling achieved higher accuracy running the same algorithm.The explanation of these results might require both methodological and biological interpretations.Firstly,the performance of the BIOCLIM model might be affected by both the sample size and equilibrium.It is highly possible that the occurrence records were uneven,especially when the spatial resolution was rough or topography was complex,making the data a small number of mismatches between the occurrence data and the model output.The amount of our data available is relatively abundant and continuous,avoiding being sensitive to the sample size effect on a small scale.Secondly,the PCA for BIOCLIM variables provided a suit means to explore the climatic limits of poorly-known species,which may help avoid the occurrence of overfitting phenomena (Kriticos et al.2014).Thirdly,P.dielsianahas a long history and excellent adaptation to its habitat,and its growth region remains stable,which contributes to determining the suitability of the ecological niche(Zhu et al.2019).

Climatic evaluation for the distribution patterns

Bioclimatic factors are shaping the distribution pattern and niche differentiation of plant species on regional scales.Among all the climatic factors,the hydrothermic variables,particularly precipitation and temperature,have long been considered as the driving roles for the distribution pattern of plant species (Clarke and Gaston 2006).For instance,humidity and temperature are essential factors for the blooming time and the chilling hours in Rosaceae (Shi et al.2019).After the first step of macro prediction for scientific introduction and cultivation ofP.dielsiana,it is necessary to identify those bioclimatic factors that most constrain its survival under local site conditions as well.To realize this target,we evaluated the maximum areas of subtropical forest in southern China.The results of PCA for the 19 bioclimatic factors also confirmed that precipitation was the decisive factor that has shaped the richness pattern ofP.dielsianaacross its whole distributional area (Fig.S2).This result is consistent with a recent study based on more extensive data as well,which believed that among all environmental factors,the diversity of Chinese Rosaceae species had the strongest correlation with humidity factors,and far beyond the temperature factor (Zou et al.2019).Meanwhile,to some extent,species may likely be more sensitive to climatic changes at the geographical margins of their distribution (Carey 1999).Hence,we performed the simulation of the most-limiting climatic factors analysis for this species accordingly.It turned out that the bio4 acted as the role of most limiting climatic factors in shaping the northern and southern boundaries across the whole distribution area(Fig.4).According to our field survey,P.dielsianaoccupies a relatively broad altitudinal amplitude,ranging from 123 to 1448 m of elevations across the whole area.The various hydrothermic conditions along the elevation gradient in the Chinese subtropical mountain ecosystems can lead to the change of plasticity of population characteristics through the adjustments of plant physiology,ecology,and growth rate,and ultimately,will affect the diversity pattern in plant populations (Abdusalam and Li 2018).In addition,compared to the Pearson correlation coefficient of latitude and longitude,the bio4 was significantly negatively influenced by the elevation (r=−0.706,P＜ 0.01) (Table 2),which also indicates that the effect of habitat filtering from altitude is more significant than that of longitude and latitude at the regional scale.

Table 2 Pearson correlation analysis of dominant bioclimatic factors with elevation,latitude and longitude

Fig.4 Spatial distribution of limiting factors for Prunus dielsiana showing how the bioclimatic variables with the most effect on prediction vary across the subtropical forest of China.Blue dotted line marks the potential distribution range of P.dielsiana

In conclusion,our result above is indicating thatPrunus dielsiana’s biological character tended to restrict its distribution to mountainous regions with a warm and humid climate,but well-drained circumstances.Our study region may be the only continental area located within the moist subtropical zone and acts as a monsoon path,which connects the vapor source (the South China Sea) with the mainland (Wu et al.2010).This region has played a vital role in maintaining the unbroken sequence of subtropical forest formation,resulting in most rainfall at marginal areas.Thus,we speculated that the bioclimatic distribution pattern ofP.dielsianamight have been influenced by the Asian monsoon climate systems(i.e.,winter and summer monsoon seasons).During every summer monsoon season,from June to August,the East Asia monsoon,accompanied by intense tropical cyclones,brings a large amount of precipitation and high level of groundwater,which could result in waterlogging (detrimental toP.dielsianasurvival) and possibly root decay.It would,therefore,be highly unlikely thatP.dielsianawould live in the eastern basin and lowland areas with poorly aerated soils.

Niche plasticity of subtropical species in China and research shortages

When comparing the dominant climatic factors ofPrunus dielsianawith those of other recently published plants from subtropical China,we found thatP.dielsianais most similar toP.cerasodisin overall niche requirements,a species in the same genus.The bio4 value ofP.dielsiana(780.0 STD × 100) is ranked the first among all the four species in the same genus investigated,but are much smaller than that ofSorbus alnifolia(935.8 STD × 100) (Table S2),a species from a different genus in the same family Rosaceae.According to Fig.5,the climatic requirements within cherry species are more similar to one another than to that ofS.alnifolia.These phenomena might be closely related to the northernmost and broadest distributional range ofS.alnifoliaamong the Rosaceae species compared,resulting in the largest thermal niche variations in the habitable zone.The concept of PNC is the tendency of lineages to retain their niche-related traits through speciation events (Crisp and Cook 2012).Thus,our data,along with those published earlier,within Rosaceae,support the initial PNC hypothesis and indicate that hydrothermic factors may dominate as environmental filters in cherry species’ survival,reproduction,growth,and dispersal.Contrastingly,climatic requirements within another family,Lauraceae,(five species ofLindera),were mostly different from one another.Lindera aggregataandL.megaphyllaare similar to each other and they together are more similar toYulania liliiflorain climatic requirements,a member of Magnoliaceae.The above result is inconsistent with the PNC hypothesis of subtropical species in China,which might be explained by quite diversified habitat preferences and niche selections under heterogeneity and complexity background of different distributions across different families.Previous research has also shown that niche plasticity may increase diversity in plant communities through complementarity (Niklaus et al.2017).Thus,our comparison across different familial and generic levelson niche plasticity can be useful for a better understanding of the forming mechanism of climatic filtering in subtropical China.

Fig.5 Non-metric multidimensional scaling (NMDS)and Minimum spanning tree(MST) showing climate climatic among different species reported in subtropical China.The two-dimensional plot was constructed based on the average values of dominant hydrothermic factors (Shown in Table S2).Black dotted line marks the niche differentiation of Cerasus in the Rosaceae(Group I) from the rest of sympatric species (Group II)

It is worth mentioning that our results only represent an ideal niche case,because a species’ ecology niche not only depends on the species’ adaptation to its present climate,but also influenced by many other biotic and abiotic factors comprehensively (Prinzing et al.2001),such as species interactions (Feroz et al.2014),seed dispersals (Seidler and Plotkin 2006),biological invasion (Less et al.2009) soil preferences(Zhang et al.2011),natural fire frequency (Silva et al.2017),etc.Besides,the historical process,such as geohistorical effects and speciation,also make significant contributions to forming the current patterns of geographical distribution and population dynamics.Thus,as the first step of a macro investigation,our findings must be interpreted with caution,and we advocate for the importance of integrating both the ecological processes and evolutionary aspects into niche modeling simultaneously for better understanding of their mechanisms in further research.

Conservation and introducing strategies for Prunus dielsiana

Though the previous occurrence records seem to suggest the natural populations ofPrunus dielsianawith a broad range(Fig.1,Table S1),however,based on our simulation map,the natural population ofP.dielsiananot only occupied a relatively wide range but also exhibited an uneven distribution pattern in general.Our field surveys in recent years also suggest thatP.dielsianais facing the severe challenge of habitat fragmentation and heterogeneity due to forest harvesting,human disturbance,inter-species competition and other threats,resulting in remarkably decreased distribution to only restricted mountain regions in reality.Some individuals have even been squeezed out to the edge of plant communities.The most suitable areas ofP.dielsianamainly concentrated in the subtropical EBLF of central China (Fig.2).In the meantime,the results of the richness pattern determination map yielded a similar diversity pattern (Fig.3).The region of Wushan Mountains and Wuling Mountains are the two most famous mountain ranges in central China and support a wide variety of rare plants and wildlife,thought to be the current biodiversity hotspot,as well as a vital glacial refugium (Xu et al.2017).Interestingly,according to the recent studies on the genetic structure ofP.dielsianapopulations,high genetic differentiation between different natural groups was detected (Zhu et al.2019).Accordingly,these two regions should be given priority for conservation,and the in situ conservation will be the most efficient approach.

Moreover,the cumulative frequency curves of critical factors forPrunus dielsianaare outputted vividly (Fig.S3),which will also contribute to forest management to identify the quantitative indexes for the occurrence of a particular species in a given area,specifically,P.dielsianacan be planted in sunny areas or on a hillside as a fast-growing afforestation species,as well as introduced as an excellent urban landscaping material into other subtropical monsoon areas of eastern China.

Conclusion

This study analyzed the climatic-based current distribution patterns ofPrunus dielsianaqualitatively and quantitatively as well as compared its niche differentiation with their sympatric species for the first time.We found that the highly suitable regions ofP.dielsianaare mainly located in central-western China.At a regional scale,the habitat filtration dominated by hydrothermic factors was the primary driving process of the distribution pattern of the diversity ofP.dielsianain the subtropical region.This cherry species,along with other cherry species of Rosaceae showed a pattern relatively consistent with PNC.Besides,the effect of habitat filtering from altitude was more significant than those of longitude and latitude.Our research can be used for updating its new evaluation in the IUCN Red List,and contribute to the improvement of scientific research,conservation,and general utilization of this typical cherry species.