Jie Gao?Xing Zhang?Zhifeng Luo?Junjie Lan?Yanhong Liu

Introduction

Understanding the mechanisms shaping geographic patterns of taxon richness has been a long-standing task for ecologists and biogeographers(Brown and Lomolino 1998;Rahbek and Graves 2001;Francis and Currie 2003;Wang et al.2007;Lomolino et al.2010).The number of hypotheses that have been proposed to explain the geographic distribution of species richness has increased sharply over the past half century(Chen et al.2014;Willig et al.2003;Field et al.2009)and new hypotheses are still being put forward(Hubbell 2001;Colwell et al.2004).Among these hypotheses,energy availability,seasonality and environmental heterogeneity are frequently proposed to explain the spatial distribution of plant and animal richness(Kreft and Jetz 2007;Tello and Stevens 2010;Chen et al.2014).Results suggest that no single factor can explain the formation and maintenance of diversity gradients within and among taxa and therefore,in most cases,more than one hypothesis or mechanism must be considered in order to fully understand patterns in species richness.

Taxocenes,(taxonomically related sets of species within a community),are fundamental units of investigation in macroecology,having many properties of interest to ecologists,including their abundance and diversity.Currently,there is little empirical or theoretical knowledge of the processes generating richness gradients across taxonomic levels(Kaspari 2001),as few studies have examined the sensitivity of ecological distribution patterns to taxonomic level(O’Brien 1993).However,O’Brien(1993)found that energy and water availability were correlated with family,genus,and species richness for seed plants in South Africa.Geographical patterns of family richness tend to be similar to those of species richness in seed plants(Francis and Currie 2003).Significant vertical variation in climate,the superposition of multiple environmental factors,and the single species pool that results from the close proximity of sample sites,makes montane systems ideally suited for exploring the mechanisms driving species richness gradients(Korner 2000,2007).Many studies have successfully selected and evaluated species diversity hot spots using data for higher taxonomic levels;the use of higher taxonomic data can save time and reduce costs,and therefore has high applied value(Balmford et al.2000).Testing whether richness-environment relationships are similar across taxonomic levels is of great potential significance for the protection of biodiversity and the prediction and selection of hot spots in seed plant diversity(Kaspari 2001).Hence,the aim of this study was to:(1)identify altitudinal patterns in richness and density for seed plants at the species,genus and family level;(2)evaluate the effects of contemporary environmental variables on spatial patterns in seed plant richness for different taxonomic levels;and,(3)explore how the effects of energy availability,seasonality and environmental heterogeneity on richness compare across taxonomic levels in seed plants.

Materials and methods

Study area

China’s Lancang(Mekong)River Nature Reserve in Yunnan Province covers an area of 75,186 ha(Fig.1).The highest elevation is 3430 meters above sea level(m.a.s.l.).Lancang River Nature Reserve has high biological diversity,including many endemic taxa.Numerous regional plant and animal species are present in the Reserve:197 plant families,806 genera and 1920 species,including 21 nationally-protected plant species.There are 60 families and 181 species(including subspecies)of terrestrial animals(Table S1).

Species data

Species distribution data from along an elevational gradient within the reserve were taken from Scientific Investigation and Research on the Nature Reserve of Lancang River in Yunnan,China(Wang et al.2010).This is the most reliable source for Lancang River plant species,describing 1051 wild seed plant species in 428 genera and 136 families.Because most species were distributed between 1400 and 3500 m.a.s.l.,the study area was divided into 21 sections along the elevation gradient,with each section covering a 100-m elevation interval.Family,genus and species richness were determined for each of these 100-m intervals.

Environmental data

Climate data were downloaded from the WorldClim-Global Climate Database,www.worldclim.org/current.htm.WorldClim is a set of global climate layers(climate grids)with a spatial resolution of 1 km2.The data can be used for mapping and spatial modeling in GIS or with other computer software.Either GeoTiff or ArcGIS BIL can be used to download data.The surface area of each 100 m elevation interval was obtained using interpolation tools in ArcGIS.The mid-domain effect was calculated with Range-Model 5(http://viceroy.eeb.uconn.edu/rangemodel)(Colwell 2008).With this model,the species richness of each 100-m interval was calculated,removing boundary effects.The thirteen environmental variables were assigned into groups as energy variables:(1)mean annual temperature(TEM,K),(2)mean temperature of the coldest month(TEMmin,K),(3)mean temperature of the warmest month(TEMmax,K),(4)annual precipitation(PREC,mm),(5)precipitation of wettest month(PRECwet,mm),and,(6)precipitation seasonality(PRECseason,mm),and seasonality variables:(7)the annual temperature range(TEMvar,K;the difference between TEMmax and TEMmin);(8)the standard deviation of mean monthly temperature(TEMsd);and,(9)the annual precipitation range(PRECvar,mm;the difference in precipitation between the wettest and driest months).Heterogeneity variables included:(10)the range of mean annual temperature within a province(TEMran,K,Chen et al.2015);(11)the range of annual precipitation within a province(PRECran,mm,Chen et al.2015);(12)the area of a given elevation interval(AREA,km2);and,(13)the mid-domain effect(the simulated number of species).

Statistical analyses

The normality of all environmental variables first was tested,and then correlations between each putative predictor and the environmental variables were assessed usingSpearman’s rank correlation coefficients(Table S2).Richness at the family,genus and species level,(as determined following the APG IV system),was regressed against elevation,choosing the best-fitting regression model.To eliminate autocorrelation along each of the 21 elevational gradients,the SAM (Spatial Analysis in Macroecology)software package(www.ecoevol.ufg.br/sam)was used.SAM correlograms for each variable in the dataset were used,showing Moran’s I and the correlogram.The explanatory power of the energy hypothesis,seasonality hypothesis,and heterogeneity hypothesis were evaluated using multiple regression.The independent variables were those included in the best models as selected using the Akaike information criterion(AICc;Montoya et al.2007)from different combinations of water and energy variables,seasonality variables and heterogeneity variables.The

model with the lowest AICc value(Heikkinen et al.2005)was selected as the best one.

Fig.1 Location of the study area

Variance and hierarchical partitioning were used to address the strong multicollinearity among the environmental variables.Strong multicollinearity can result in the exclusion of causal ecological variables from examined models(Heikkinen et al.2005).Variance partitioning was used to decompose the variation in richness across different taxonomic levels among the three groups of predictors,i.e.,the combinations of energy,seasonality and heterogeneity variables as selected by AICc,and was carried out in the R statistical package(vegan).Partitioning led to eight fractions:(1)pure effects of energy;(2)pure effects of seasonality;(3)pure effects of heterogeneity;and combined variation due to the joint effects of(4)energy and seasonality;(5)energy and heterogeneity;(6)seasonality and heterogeneity;(7)all three groups of explanatory variables;and,finally,(8)unexplained variation.The total amount of variation in species richness was explained by regressing species richness on the statistically significant variables selected from the three groups of explanatory variables(Heikkinen et al.2005;Chen et al.2015).

Hierarchical partitioning provides,for each explanatory variable separately,an estimate of its unique and its joint contribution when combined with all other variables(Walsh and Mac-Nally 2003;Heikkinen et al.2005).Hierarchical partitioning was carried out in the R statistical package(hier.part).Partitioning used linear regression with R2taken as a measure of the goodness-of-fit.Statistical significance of the independent contributions of variables were tested with a randomization routine which yielded Z-scores for a generated distribution of randomized independent contributions,and a measure of statistical signif icance based on an upper 0.95 confidence limit(Chen et al.2015).Statistical analyses were carried out using SAM or R.

Results

Families,genera,and species showed strong elevational trends in richness.With increasing elevation,richness across different taxonomic levels first rose and then declined.The amount of variation explained by the regression decreased with increasing taxonomic resolution(family richness:R2=0.907,p＜0.0001;genus richness:R2=0.828,p＜0.0001;species richness:R2=0.709,p＜0.0001)(Fig.2;Fig.S1).

The selected explanatory variables for family,genera and species richness are presented in Table 1 for each of the three groups of variables.Energy and heterogeneity variables generally explained more variation in richness across different taxonomic levels than did seasonality(Table 1).TEMsd and PRECvar were selected as the best seasonality variables across different taxonomic levels.For family richness,the explanatory power of the selected heterogeneity variables(PRECran and MDE,92.1%)and energy variables(TEMmax and TEMmin,82.3%)was much higher than that of the seasonality variables(TEMsd and PRECvar,66.5%).For genus richness,the explanatory power of energy variables(TEMmax,TEMmin and PREC,79.5%)was similar to that of the heterogeneity variables(PRECran and MDE,88.0%)and higher than that of the seasonality variables(TEMsd and PRECvar,54.6%).For species richness,the energy and heterogeneity variables were comparable(energy:72.6%;heterogeneity:71.0%),while the effect of seasonality was smaller(Table 1).

Variance partitioning revealed that the largest amount of variation in richness was accounted for by the joint effects of energy and heterogeneity variables,explaining 74.7%of variation in family richness,77.4%in genus richness and 57.6%in species richness(Fig.3).For family richness,the independent effects of energy and heterogeneity variables(7.3%,10.9%)were higher than that of the seasonality variables(3.7%)(Fig.3a).For genus richness,heterogeneity variables had the greatest independent effect(15.3%)followed by energy(2.8%)and seasonality variables(1.6%)(Fig.3b).For species richness,the independent effects of energy(12.7%)and heterogeneity(12.5%)were comparable,and both higher than that of seasonality(5.1%)(Fig.3c).

In hierarchical partitioning,the independent effects of all included variables were statistically significant,although some contributions were small.For family,genus and species richness,MDE and TEMsd had the greatest effects(Fig.4).Meanwhile,PRECvar,TEMmax,and TEMmin had comparable independent effects on family richness(Fig.4a).

Discussion

Do seed plant families,genera,and species show the elevational diversity gradient in richness?

Significant vertical variation in climate and the superposition of multiple environmental factors at different scales combine to make mountainous regions ideal for the exploration of patterns in biological diversity and the mechanisms that drive them(Korner 2000,2007).Studies have found that a number of species show different distributions along elevational gradients(Lomolino 2001;Sanders et al.2003;Wang et al.2007).Several studies have found that the highest species richness appears at midaltitudinal zones(Sanders 2002;McCain 2004),while some have found species richness to decrease monotonically with increasing elevation(Ohsawa 1995;Stevens 1996).

Fig.2 Elevational gradients in family,genus and species richness.The R2 values represent adjustment coefficients

Table 1 Summary of the multiple regression models for energy,seasonality,and heterogeneity variables on family,genus and species richness.The model with the lowest AICc was selected as the best model

Fig.3 Results of variance partitioning for family richness(a),genus richness(b)and species richness(c)with numbers representing the proportion of variation explained(%).Variation in family,genus and species richness was partitioned into that explained by three sets of variables:energy,seasonality and heterogeneity variables,as well as a proportion of unexplained variation.The letters a,b and c indicate the unique effects of energy,seasonality and heterogeneity variables,respectively;while d,e,f and g indicate their joint effects

Fig.4 Results of hierarchical variance partitioning for family richness(a),genus richness(b)and species richness(c)showing the independent and joint effects of the predictors

In this study,seed plant richness,whether evaluated at the species,genus or family level,showed a significant‘hump-shaped’pattern with elevation,supporting the theories of Rahbek(2005),McCain(2004)and Grytnes and Beaman(2006)(Fig.2)on the altitudinal patterns of species richness.Variation in area may be a causal factor for this diversity pattern.It is widely accepted that species richness increases as a function of area(Rahbek 2005).With increasing area,habitats become more complex and can provide a wider variety of ecological niches and environmental resources,thereby potentially increasing species diversity(Bazzaz 1975).However,whether humpshaped diversity curves are caused by area-heterogeneity trade-offs,i.e.,reductions in the area suitable for particular species as environmental heterogeneity increases,has recently been debated(Allouche et al.2012;Carnicer et al.2013).

The effects of energy,seasonality and heterogeneity

The results of variance and hierarchical partitioning indicated that contemporary environmental variables,representing energy availability,seasonality and environmental heterogeneity,were the primary factors explaining variation in richness at the family,genus and species level(Tello and Stevens 2010;Stein et al.2014).Energy and heterogeneity variables explained the spatial variation in richness across taxonomic levels in the Lancang River Nature Reserve.The contributions of energy and habitat heterogeneity variables to family and species richness were similar.The correlation between species richness and water-and energy-availability variables was similar to that found in American plant species at a higher taxonomic level(Bogonovich et al.2014).Several previous studies have also found that heterogeneity variables were frequently selected in best-fitting environmental models of species richness(Kerr et al.2001;White and Kerr 2007).The strong effect of heterogeneity in shaping the geographic patterns of seed plant richness may be due to strong niche separation along elevational gradients.The mid-domain effect also had a strong independent effect on seed plant richness across the family,genus and species levels.Seasonality accounted for a much smaller proportion of variation in family,genus and species richness.

Conclusions

Family,genus and species richness data collected along an elevational gradient in the Lancang River Nature Reserve were used to investigate spatial patterns of richness.Findings were that:(1)seed plant richness at the species,genus and family level showed a ‘hump-shaped’relationship with elevation across a broad elevational range.Model fit improved with taxonomic level.Furthermore;(2)spatial patterns in family,genus and species richness were well-explained by contemporary environmental variables.The joint effects of energy availability and environmental heterogeneity variables explained the greatest proportion of variation in richness across taxonomic levels;(3),with these two categories being highly redundant,the effect of seasonality was mostly complementary.The mid-domain effect independently contributed to patterns in seed plant richness across at all taxonomic levels.

AcknowledgementsWe would like to thank Emily Drummond at the University of British Columbia and J.W.Ferry Slik at the Institute for Biodiversity and Environmental Research(IBER)for their assistance with English language and grammatical editing of the manuscript.

Allouche O,Kalyuzhny M,Moreno-Rueda G,Pizarro M,Kadmon R(2012)Area-heterogeneity tradeoff and the diversity of ecological communities.Proc Natl Acad Sci USA 109:17495–17500

Balmford A,Lyon AJE,Lang RM(2000)Testing the higher-taxon approach to conservation planning in a megadiverse group:the macrofungi.Biol Convers 93:209–217

Bazzaz FA(1975)Plant species diversity in old-field successional ecosystems in southern Illinois.Ecology 56:485–488

Bogonovich M,Robeson SM,Watson MA(2014)Patterns of north American fern and lycophyte richness at three taxonomic levels.Am Fern J 103:193–214

Brown JH,Lomolino MV(1998)Biogeography.Sinauer Associates Inc,Sunderland

Carnicer J,Brotons L,Herrando S,Sol D(2013)Improved empirical tests of area-heterogeneity tradeoffs.Proc Natl Acad Sci USA 110:E2858–E2860

Chen SB,Mao LF,Zhang JL,Zhou KX,Gao JX(2014)Environmental determinants of geographic butterfly richness pattern in eastern China.Biodivers Conserv 23:1453–1467

Chen SB,Ferry Slik JW,Mao LF,Bi MJ,Shen MW,Zhou KX(2015)Latitudinal diversity gradients in bryophytes and woody plants:roles of temperature and water availability.J Syst Evol 53:535–545

Colwell RK(2008)RangeModel:tools for exploring and assessing geometric constraints on species richness(the mid-domain effect)along transects.Ecography 31:4–7

Colwell RK,Rahbek C,Gotelli NJ(2004)The mid-domain effect and species richness patterns:what have we learned so far?Am Nat 163:E1–E23

Field R,Hawkins BA,Cornell HV,Currie DJ,Diniz-Filho JAF,Gue′gan JF,Kaufman DM,Kerr JT,Mittelbach GG,Oberdorff T,O’Brien EM,Turner JRG (2009)Spatial species-richness gradients across scales:a meta-analysis.J Biogeogr 36:132–147

Francis AP,Currie DJ(2003)A globally consistent richness climate relationship for angiosperms.Am Nat 161:523–536

Grytnes JA,Beaman JH(2006)Elevational species richness patterns for vascular plants on Mount Kinbalu,Borneo.J Biogeogr 33:1838–1849

Heikkinen RK,Luoto M,Kuussaari M,Poyry J(2005)New insights into butterfly—environment relationships using partitioning methods.Proc R Soc B 272:2203–2210

Hubbell SP(2001)The unified neutral theory of biodiversity and biogeography.Princeton University Press,Princeton

Kaspari M (2001)Taxonomic level,trophic biology and the regulation of local abundance.Glob Ecol Biogeogr 10:229–244

Kerr JT,Southwood TRE,Cihlar J(2001)Remotely sensed habitat diversity predicts butterfly species richness and community similarity in Canada.Proc Nat Acad Sci USA 98:11365–11370

Korner C(2000)Why are there global gradients in species richness?Mountains might hold the answer.Trends Ecol Evol 15:513–514

Korner C(2007)The use of ‘a(chǎn)ltitude’in ecological research.Trends Ecol Evol 22:569–574

Kreft H,Jetz W(2007)Global patterns and determinants of vascular plant diversity.PNAS 104:5925–5930

Lomolino MV(2001)Elevation gradients of species-density:historical and prospective views.Glob Ecol Biogeogr 10:3–13

Lomolino MV,Riddle BR,Whittaker RJ,Brown JH (2010)Biogeography.Sinauer Associates,Inc.,Sunderland

McCain CM(2004)The mid-domain effect applied to elevational gradients:species richness of small mammals in Costa Rica.J Biogeogr 2004(31):19–31

Montoya D,Rodr′?guez MA,Zavala MA,Hawkins BA(2007)Contemporary richness of holarctic trees and the historical pattern of glacial retreat.Ecography 30:173–182

O’Brien EM(1993)Climatic gradients in woody plant species richness:towards an explanation based on an analysis of southern Africa’s woody flora.J Biogeogr 20:181–198

Ohsawa M(1995)Latitudinal comparison of altitudinal changes in forest structure,leaf-type,and species richness in humid monsoon Asia.Vegetatio 121:3–10

Rahbek C(2005)The role of spatial scale and the perception of largescale species-richness patterns.Ecol Lett 8:224–239

Rahbek C,Graves GR(2001)Multiscale assessment of patterns of avian species richness.Proc Natl Acad Sci USA 98:4534–4539

Sanders NJ(2002)Elevational gradients in ant species richness:area,geometry,and Rapoport’s rule.Ecography 25:25–32

Sanders NJ,Moss J,Wagner D(2003)Patterns of ant species richness along elevational gradients in an arid ecosystem.Glob Ecol Biogeogr 12:93–102

Stein A,Gerstner K,Kreft H(2014)Environmental heterogeneity as a universal driver of species richness across taxa,biomes and spatial scales.Ecol Lett 17:70–81

Stevens GC(1996)Extending Rapoport’s rule to Pacific marine fishes.J Biogeogr 23:149–154

Tello JS,Stevens RD(2010)Multiple environmental determinants of regional species richness and effects of geographic range size.Ecography 33:796–808

Walsh CJ,Mac-Nally R(2003)Hierarchical Partitioning,R Project for Statistical Computing.http://cran.r-project.org/

Wang ZH,Tang ZY,Fang JY(2007)Altitudinal patterns of seed plant richness in the Gaoligong Mountains,south-east Tibet,China.Divers Distrib 13:845–854

Wang J,Du F,Yang YM,Tian K,Wang YX(2010)Research on the nature reserve of Lancang River in Yunnan,China.Science Press,Beijing

White PJT,Kerr JT(2007)Human impacts on environment-diversity relationships:evidence for biotic homogenization from butterfly species richness patterns.Glob Ecol Biogeogr 16:290–299

Willig MR,Kaufman DM,Stevens RD(2003)Latitudinal gradients of biodiversity:pattern,process,scale,and synthesis.Annu Rev Ecol Evol Syst 34:273–309