Lizard species richness patterns in China and its ... - Springer Link

Biodivers Conserv (2011) 20:1399–1414 DOI 10.1007/s10531-011-0033-0 ORIGINAL PAPER

Lizard species richness patterns in China and its environmental associations Yong Huang • Qiang Dai • Yueying Chen • Hongfu Wan Jiatang Li • Yuezhao Wang

•

Received: 11 October 2010 / Accepted: 7 March 2011 / Published online: 26 April 2011 Ó Springer Science+Business Media B.V. 2011

Abstract Many ecological hypotheses have been widely used to explain species richness variation across the globe. We investigated lizard species richness patterns in China, and identified areas of high species richness. Furthermore, we tested hypotheses concerning the relationships between lizard richness and environmental variables. A large data including 30,902 records of point locality data for 151 lizard species occurring in China were retrieved from Herpetology museums of CIB/CAS and other museums through HerpNET, and published sources, and then predicted distributions maps were generated using ecological niche modeling. We overlaid all species prediction maps into a composite map to describe species richness patterns. A multiple regression analysis using eigenvector-based spatial filtering (SEVM) was performed to examine the best environmental predictors of species richness. Richness peaked mainly in southern China located in the Oriental realm. Our best multiple regression models explained a total of 80.1% variance of lizard richness (r2 = 0.801; F = 203.47; P \ 0.001). Among related factors in shaping species richness distribution, the best environmental predictors of species richness were: frost-day frequency, elevation, vegetation, and wet-day frequency. Based on models selection, our results revealed that underlying mechanisms related to different ecological hypotheses might work together and best explain lizard richness in China. We are in an initial step to develop a large data set on species richness, and provide the necessary conservation implications from habitat loss. Additional studies that test species richness at different geographical scale are required to better understand the factors that may influence the species richness distribution in East Asia. Keywords

Biogeography China Climate Lizard Spatial statistics Species richness

Y. Huang Q. Dai Y. Chen H. Wan J. Li Y. Wang (&) Department of Herpetology, Chengdu Institute of Biology, Chinese Academy of Sciences, Chengdu 610041, Sichuan, China e-mail: [email protected] Y. Huang H. Wan Graduate University of Chinese Academy of Sciences, Beijing 100049, China

123

1400

Biodivers Conserv (2011) 20:1399–1414

Abbreviation CIB/CAS Chengdu Institute of Biology, Chinese Academy of Sciences

Introduction Determining the causes of the great biodiversity variation across Earth has long been a major challenge for ecologists and biogeographers (Gaston 2000), ever since biotic diversity contrast between equatorial and polar latitudes was discovered two centuries ago (von Humboldt 1808). Among the considerable number of hypotheses that aim to explain species richness patterns (reviewed by Gaston 2000; Hawkins et al. 2003a), many ecological (environmental) hypotheses have been widely discussed and accepted (Terribile et al. 2009). Three alternative variants of ecological hypothesis, the species-energy, contemporary climate and habitat heterogeneity hypotheses, have received a great deal of attention as the primary determinants of species richness (Currie 1991; Kerr and Packer 1997; Kerr et al. 2001; Rahbek and Graves 2001; Terribile et al. 2009). The species-energy hypothesis includes at least two versions, the ambient energy and productive energy hypotheses (Hawkins et al. 2003a). The ambient energy hypothesis, widely indicated by temperature or allied measures, argues that species richness was influenced by energy inputs into an area that affects the physiological tolerance of organisms (e.g., Currie 1991; Turner et al. 1987). The productive energy hypothesis claims that animal species richness is limited by energy via food webs rather than by physiological requirements. The energy and water availability (i.e., energy–water dynamics) limits the total available plant productivity, which ultimately moves up the food chains (e.g., Hawkins et al. 2003a; Hawkins and Porter 2003; Mittelbach et al. 2001; Wright 1983). The contemporary climate hypothesis states that species richness correlates with contemporary climate conditions, and putative causal mechanisms are in terms of environmental stability, variability, favorability and harshness (e.g., Currie 1991; Costa et al. 2007; Tognelli and Kelt 2004). The habitat heterogeneity hypothesis is measured either as the number of habitat types or the topographic relief (range in elevation) presented within an area (Kerr et al. 2001; Hortal et al. 2009). It assumes that high species richness is found in physically or biologically complex habitats, through higher speciation rates and providing more ecological niches (e.g., Currie 1991; Kerr and Packer 1997; Kerr et al. 2001). Most studies have investigated the competing hypotheses or multiple environmental mechanisms to explain the relationships between species richness and environmental factors at local or even larger spatial scales, for example, fishes (e.g., Oberdoff et al. 1995; Tedesco et al. 2005), amphibians (e.g., Rodrı´guez et al. 2005; Buckley and Jetz 2007), birds (e.g., van Rensburg et al. 2002; Hawkins et al. 2007), and mammals (e.g., Badgley and Fox 2000; Ceballos and Ehrlich 2006). However, the knowledge of the determinants of reptile richness remains insufficiently documented among terrestrial vertebrates (Qian et al. 2007; Terribile et al. 2009; McCain 2010). It is urgency to understanding the drivers of reptile richness patterns due to global warming impact on species distribution and abundance (Pounds et al. 1999; Gibbons et al. 2000; Qian et al. 2007). Lizards belong to Reptila and are good model system to test these alternative hypotheses. Because their taxonomy is well resolved and distributional data are quite thorough. They are ectothermic and sensitive to environmental variables. Several previous studies attempted to test these competing

123


1401

hypotheses at different spatial scales using lizard species as model systems. Temperature (Pianka 1967; Scheibe 1987; Fu et al. 2007; Powney et al. 2010), precipitation (Scheibe 1987) and the number of hours of sunshine (Schall and Pianka1978) have been correlated with lizard species richness, whereas topographic heterogeneity (Owen 1989), actual evapotranspiration (AET) (Powney et al. 2010) and potential evapotranspiration (PET) (Fu et al. 2007) have been suggested to explain lizard species richness. The driving factors of these results may vary across geographical scale (Powney et al. 2010), but increasing studies on other taxa suggest that a series of competing hypotheses or predictors may play different roles to explain species richness variation over broad spatial scales (Costa et al. 2007; Gaston 2000; Kerr et al. 2001). China is one of the few countries in the world where the Palearctic and Oriental realms meet. It is a land of a varied topography, from deserts and mountains in the west to lower lands in the east, and of dramatic topographic relief. It has Earth’s highest point, the Himalayas, and the Tibetan Plateau, with an average elevation of over 4,000 m, and the Chinese lowest depression, Turpan Basin in Xinjiang, with 154 m below sea level. It possesses a highly astounding heterogeneous landscape, composed of a mosaic of different vegetation types, and is a place where dense forests give way to open grasslands from south to north. Tremendous differences in latitude, longitude, and altitude create the conditions for extremely diverse climate: tropic in the south to subarctic in the north, and result in regionally varied precipitation and temperature. Monsoon winds, caused by the seasonal changes in temperature between the continent and the ocean, dominate the climate. The Mountains of Southwest China is one of the global 34 biodiversity hotspots, as defined by diversity, endemism, and human threats (Myers et al. 2000; Mittermeier et al. 2005). Thus these unique and distinctive environments could provide a good opportunity for testing hypotheses concerning the relationship between environmental variables and lizard species richness. In this study, we examine the correlation between lizard species richness and various environmental factors across China. Our objectives include (1) mapping distributions of Chinese lizards and describing any patterns, and (2) testing various ecological factors in determining species richness patterns.

Materials and methods Data collection We collected locality data for lizard species which occur in China from a variety of sources. All date from the Chengdu Institute of Biology, Chinese Academy of Sciences (CIB/CAS) were directed collected from museum catalogue and field notes. Information from other museums was obtained through HerpNET (HerpNET data portal http://www. herpnet.org) on 31 May 2010, including University of Kansas Biodiversity Institute— Herpetology Collection, Arctos—MVZ Herp Catalog, Yale University Peabody Museum—Peabody Herpetology DiGIR Service, Staatliches Museum fu¨r Naturkunde Stuttgart—Staatliches Museum fu¨r Naturkunde Stuttgart, University of Nebraska State Museum—UNSM Vertebrate Specimens. Additional information was also obtained from a variety of published sources. The raw data set consisted of 30,902 presence-only records (individuals) in total, of these, a total of 23,475 records (76%) were retrieved from CIB, 1,892 records (6%) were retrieved from HerpNET, and the remaining sources provided 5,535 records (18%). The accurate taxonomy and the most precise locality information of

123

1402


Fig. 1 The raw sample point of species per grid cell based on museum collections, published literatures. Each grid cell is 100 9 100 km. Blank cells have no specimen based on our database. The map used an Albers Equal-Area Conic projection (at the standard parallels of 25° and 47° of N, central meridian of 105°). The solid lines identify the biogeographical limits of Zhang (1999) used in our analyses

all the museum specimen records were verified. To reduce the sources of potential errors records from HerpNET, we included latitude and longitude records in our analyses. Isolated distribution records and questionable identifications were checked again, if possible, we will borrow the material for examination. For the taxonomic revisions or updated information recently, we follow the taxonomy of Zhao et al. (1999) and the reptile database (http://www.reptile-database.org/). Finally, we built a database of 151 lizard with species names (Appendix 1) represented by a total of 3,391 records for unique point localities, with a range of 2–288 (mean = 22.5, standard deviation = 38.3) (Fig. 1). When available, we used geographical coordinates from museum databases or literature, and otherwise, we compiled coordinates from Google Earth. Ecological niche modeling and species richness For each of the 151 species, we used the Genetic Algorithm for Rule-set Prediction (GARP) (for free download see: http://www.nhm.ku.edu/desktopgarp/) (Stockwell and Peters 1999) for reconstructing species distribution maps. GARP uses an evolutionary computing genetic algorithm to searches iteratively for non-random correlations between species presence and environmental variables for localities using several different types of rules (i.e., atomic rules, range rules, negated range rules, logistic regression rules), and then creates ecological niche models for each species’ predicted distribution, as contrasted with environmental characteristics across the overall study area (Stockwell and Peters 1999). GARP was found that it did not tend to be more sensitive to sampling bias than Maxent,

123


1403

and GARP is a very useful technique to estimate richness and composition of unsampled areas and have been tested to correctly predict the most of the species’ distributional potential (Pearson et al. 2007, Peterson et al. 2007; Costa et al. 2010), for example in applications to invasive species (e.g., Peterson et al. 2003; Peterson and Robins 2003), tree species (e.g., Ferreira de Siqueira et al. 2009; Menon et al. 2010), squamate species (e.g., Raxworthy et al. 2003; Costa et al. 2007; Peterson et al. 2007), and so on. We included a total of eighteen environmental variables in the model. Variables for details, descriptions, and files for download are described in the following text. We set several optimization parameters while running the software following Costa et al. (2007). The parameters included: 20 runs, 0.001 convergence limit, and 1,000 maximum interactions; rule types: atomic, range, negated range, and logistic regression; best subset active, 5% omission error, 40% commission error, and 67% of points for training; omission measure = extrinsic, and omission threshold = hard; 10 models under hard omission threshold. The estimation output of DesktopGarp produced in Arc/Info grid maps with ‘zeros’, where the species were not predicted to occur, and ‘ones’, where the species were predicted to occur. The area covered by the coincidence of at least seven out of the 10 models in the best subset selection (optimum models considering omission/commission relationships; Anderson et al. 2003) were used as the predicted distribution of each species. By doing so and by setting the commission error to 40%, this approach added a component of conservatism in predicting distribution by GARP, which might otherwise extrapolate too much and predict areas that are too far from where the species have previously been collected (Costa et al. 2007). After generating such maps using the same criteria for all 151 species, we used ARCGIS software to overlay all species prediction maps into a composite map. This final map was used to create a girded of species richness map at a resolutions of 100 km (approximately equivalent to 1° at the equator) on an Albers Equal-Area Conic projection (at the standard parallels of 25° and 47° of N, central meridian of 105°). We excluded coastal grid cells with less than 96% land cover and all islands from the analysis in order to remove the effects of insularity, leaving 827 cells for analysis. Consequently, we used the occurrences of 151 lizard species within the 827 grid cells to calculate species richness, summing the value of overlaid corresponding grid cells. Environmental data We used eighteen environmental variables (Table 1). We selected these variables based on previous studies and the four associated hypotheses (Currie 1991; Kerr and Packer 1997; Kerr et al. 2001; Rahbek and Graves 2001; Tognelli and Kelt 2004). All environmental variables for assessing hypothesized explanations of species richness were re-projected and re-sampled to the same equal-area cell as the species richness data in ARCGIS. The hypotheses and their related variables are: (1)

Ambient energy—five variables are associated this hypothesis within each cell, including: mean annual potential evapotranspiration (PET) (Ahn and Tateishi 1994. 300 resolution, available at http://www.grid.unep.ch/data/grid/gnv183.html); mean annual highest temperature (HT), and mean annual lowest temperature (LT) (data from 1961 to 1990 with 1 km2 resolution, available at http://www.data.ac.cn/ index.asp); mean annual sum of effective temperature (C0°C) (SET0) and mean annual sum of effective temperature (C10°C) (SET10) (data from 1981 to 1996 with 500 m2 resolution, available at http://www.geodata.cn/Portal).

123

1404


Table 1 Each explanatory variable under test associated with the four hypotheses Associated hypothesis

Explanatory variables

Abbreviation

Ambient energy

Mean annual potential evapotranspiration (mm year-1)

PET

Ambient energy

Mean annual highest temperature (°C)

HT

Ambient energy

Mean annual lowest temperature (°C)

LT

Ambient energy

Mean annual sum of effective temperature (C0°C) (°C)

SET0

Ambient energy

Mean annual sum of effective temperature (C10°C) (°C)

SET10

Productive energy

Normalized difference vegetation index

NDVI

Productive energy

Mean annual actual evapotranspiration (mm year-1)

AET

Productive energy

Mean annual solar radiation (W m-2)

RAD

Climate hypothesis

Mean annual temperature (°C)

AT

Climate hypothesis

Mean annual sunshine (percent of day length)

SUN

Climate hypothesis

Mean annual diurnal temperature range (°C)

DTR

Climate hypothesis

Mean annual frost-day frequency (days)

FF

Climate hypothesis

Mean annual wind speed (ms-1)

WIND

Climate hypothesis

Mean annual precipitation (mm year-1)

PRE

Climate hypothesis

Mean annual wet-day frequency (mm year-1)

WET

Climate hypothesis

Mean annual relative humidity (percent)

REH

Habitat heterogeneity

Vegetation (number of vegetation classes/quadrat)

VEG


Elevation range (count of 300 m/quadrat)

ELE

(2)

(3)

(4)

Productive energy—three variables are used to account for productive energy hypothesis, including: mean annual remotely sensed Normalized Difference Vegetation Index (NDVI), obtained from Advanced Very High-Resolution Radiometer (AVHRR) record of monthly changes in the photosynthetic activity of terrestrial vegetation (data from 1998 to 2008 with 1 km2 resolution, Data source: Environment and Ecology Scientific Data Center of western China, National Natural Science Foundation of China, available at http://westdc.westgis.ac.cn), mean annual actual evapotranspiration (AET) (Ahn and Tateishi 1994. 300 resolution, available at http://www.grid.unep.ch/ data/grid/gnv183.html), and mean annual solar radiation (RAD) (data from 1950 to 1980 with 1 km2 resolution, available at http://www.geodata.cn/Portal). Contemporary climate hypothesis—eight variables are associated with this hypothesis within each cell, including: mean annual temperature (AT) (data from 1961 to 1990 with 1 km2 resolution, available at http://www.data.ac.cn/index.asp); mean annual sunshine (SUN) (percent of daylength), mean annual diurnal temperature range (DTR) and mean annual frost-day frequency (FF) (data from 1961 to 1990 with 100 resolution, New et al. 2002); and mean annual wind speed (WIND) (data from 1981 to 1996 with 500 m2 resolution, available at http://www.geodata.cn/Portal); mean annual precipitation (PRE) (data from 1961 to 1990 with 1 km2 resolution, available at http://www.data.ac.cn/index.asp), mean annual wet-day frequency (WET) (number days with [0.1 mm precipitation per month) and mean annual relative humidity (REH) (data from 1961 to 1990 with 100 resolution, New et al. 2002). Habitat heterogeneity—the count of 300 m elevation range within each quadrat (ELE) (HYDRO1 k data set for Asia, 300 resolution, available at http://eros.usgs.gov/) and the number of vegetation classes (VEG) (1 km2 resolution, Data source: Environment and Ecology Scientific Data Center of western China, National Natural Science Foundation

123


1405

of China, available at http://westdc.westgis.ac.cn) as indicators of habitat heterogeneity. Statistical analyses In order to examine the potential predictors of lizard richness patterns in China, we first tested the relationship between lizard richness and environmental variables using a multiple regression analysis. We did not use all environmental variables employed to run GARP, because including many highly correlated variables in a multiple regression creates several theoretical and statistical problems, especially estimating partial regression coefficients (Tabachnick and Fidell 2000). We selected variables previously identified as affecting species richness and were not highly correlated (r \ 0.80) and there were one variable represented each hypothesis at least (Appendix 2). We used the eigenvector-based filtering, or spatial eigenvector mapping (SEVM) obtained by Principal Coordinates Neighbour Matrices (PCNM) to account for spatial autocorrelation (Diniz-Filho and Bini 2005). Spatial autocorrelation is a potential problem when work with large-scale ecological data and explanatory variables (Legendre 1993; Legendre et al. 2002). Failure to account for spatial autocorrelation could result in inflating Type I error because model fitting may generate artificially narrow standard errors due to the lack of independence among residuals (Legendre 1993; Legendre et al. 2002). A truncation distance of 102.33 km, calculated in SAM—Spatial Analysis in Macroecology (Rangel et al. 2006, 2010), was used to create the spatial filters. Eigenvector filters were chosen when their influence on species richness was both statistically significant (P \ 0.05) and had sufficient explanatory power (r2 [ 0.02). We selected eigenvector filters in an iterative process, by minimizing both the spatial autocorrelation among residuals and the number of filters used in regression. Moran’s I coefficient were used to examine the model residuals of the spatial autocorrelation in reducing spatial autocorrelation (Diniz-Filho et al. 2003). These filters were then used as candidate predictor variables, together with other environmental predictors formed in the full model. In this way, the effects of environmental predictors are evaluated as partial effects, taking spatial factors into account explicitly (Rangel et al. 2006, 2010). The total explanatory power, r2 values, was divided into three parts: a part explained by space, a part explained by environmental variables, and a part of shared explained variance. To test which hypothesis best explains variation in lizard richness in China, we conducted separate regressions to fit each of the hypothesis presented in Table 1, with an addition of mixed models using all variables associated with each hypothesis. The sample-size-corrected Akaike information criterion (AICc) was used to evaluate the goodness of model fit. The model with the lowest AICc score was considered the most parsimonious, therefore optimizing the tradeoff between bias and precision in model construction (Burnham and Anderson 1998). The difference between any candidate models and the best model (DAICc) was used to evaluate the relative model fit when their AICc scores were close. The larger the DAICc, the less possible is the fitted model as being the best approximating model in the given models set. In general, Models having DAICc B 2 have substantial support (evidence), those in which 4 B DAICc B 7 have considerably less support, and models having DAICc [ 10 have essentially no support (Burnham and Anderson 2004). Model-averaging of estimates using Akaike weights (wi) was used to confront model selection uncertainty (Burnham and Anderson 1998; Anderson et al. 2000).

123

1406


Fig. 2 Number of species per grid cell based on, A the raw data of museum collections, and B the ecological niche modeling of 151 species. The grid corresponds to the approximate area of China and the area of each cell is 100 km2. Blank cells have no specimen based on the major collections

Finally, to make a comparison between the actual available data and the lizard distributions predicted by ecological niche modeling, we mapped species locality points’ data and calculated species richness at 100 km resolution (Fig. 2A). This method allowed us to check whether a spatial sampling bias was shown in the final modeling map (i.e., areas that have more species collected coincide with the areas the model indicated as higher species richness) (Costa et al. 2007).

Results Patterns of species richness Figure 2B shows the distribution map which summed all 151 lizard species richness in China. Lizard species richness varied between 1 and 38 species per cell (mean: 13 ± 8 SD) and displayed a consistent pattern that species number increased from higher latitudes to lower ones, and from west to east. The Highest species richness occurs in the Oriental Realm tropics, around the border between southern China and southwestern China, and around the Nanling Mountains, at the border between southern China and central China. Other areas with relative high richness included southwestern China, southern China, northwestern and eastern central China of the Oriental realm (Fig. 2B). The raw data map shows a slight sampling bias to the northwest central China and northeast southwestern China (Figs. 1, 2A), where the largest herpetological museum CIB/ CAS in China are located. However, the niche modeling results are not highly influenced

123


1407

Fig. 3 Moran’s I correlogram for lizard species richness and the residuals of multiple regression with environmental predictors

by this bias, since areas corresponding to the highest species richness in China do not overlay completely with the pattern. Furthermore, high richness areas were found by the niche modeling in the relatively poor sampling regions, such as the northwest MongoliaXinjiang China, west-south Qinghai-Tibet China areas (Fig. 2B). Species richness and environmental variables The environmental models for the multiple regression analysis using SEVM with adding eigenvector spatial filters (PCNM), were sufficient to reduce autocorrelation in the residuals (filters data not shown). A spatial correlogram based on Moran’s I index was used to evaluate the pattern of spatial autocorrelation in the residuals of the regression (Fig. 3). The multiple regression model explained a total of 80.1% variance of lizard richness in China (r2 = 0.801; F = 203.47; P \ 0.001). The total explanatory power explained by predictors alone was 17.0%, and explained by space alone was 5.4%, and the shared explained variance was 57.7%. Based on the multiple regression analysis taking PCNM spatial filters into account, annual frost-day frequency (FF), elevation range (ELE), the number of vegetation classes (VEG), and wet-day frequency (WET) were the best predictors of species richness (Table 2). FF was negatively correlated with lizard richness, while ELE, VEG, and WET, respectively, were positively correlated with lizard species richness (Table 2). Based on model selection approach, the model with the lowest AICc value was the mixed model, which contained all variables related to the different hypotheses. It had an Akaike weight of 1.00 (Table 3). Other models had high DAICc values ([10, Burnham and Anderson 2004) and low values of Akaike weights.

Discussion Our results indicate that mechanisms related to different ecological hypotheses might work together to account for lizard richness in China (Table 3). It is important to consider the influence that environmental factors may have on shaping richness patterns (Powney et al. 2010). The frost-day frequency, elevation range, vegetation and wet-day frequency were the most important environmental variable predicting lizard species richness in China. Frost-day frequency (FF) was very important in predicting lizard species richness in a way that it is negatively related to richness (t = -13.584, P \ 0.01). This suggested that

123

1408


Table 2 Standardized regression coefficients of the multiple regression model (b), t statistics, and associated P-values for lizard species richness regressed against environmental variables Environmental variables

b

t

P

FRS

-0.826

-13.584

\0.01

ELE

0.099

4.567

\0.01

VEG

0.125

4.385

\0.01

WET

0.24

4.284

\0.01

RAD

0.053

1.921

0.06

REH

-0.091

-1.49

0.14

PET

-0.045

-1.292

0.20

WIND

-0.014

-0.587

0.56

NDVI

-0.007

-0.201

0.84

Spatial autocorrelation was accounted for in the multiple regressions using the eigenvector-based filtering (SEVM). The eigenvector filters obtained by the method of Principal Coordinates of Neighbour Matrices (PCNM). Filters were not shown in table

Table 3 Summary of the model selection approach Hypothesis

Adjusted R2

AICc

DAICc

K

Wi


0.66

4980.12

419.18

3

0.00

Productivity energy

0.69

4926.31

365.36

3

0.00

Ambient energy

0.71

4854.22

293.28

2

0.00

Climate hypothesis

0.79

4611.93

50.99

5

0.00

Mixed model*

0.80

4560.94

0.00

10

1.00

2

In each hypothesis, the corrected Akaike Information Criterion (AICc) value and adjusted r were given. The model with the lowest AICc value was the most parsimonious one among the fitted models and was selected (marked in bold). DAICc values were compared to the best fitting model. Wi was the Akaike weight and it indicated the relative support a given model had when compared with the other models. K was the number of variables of the model, including intercept *All nine variables used in the multiple regressions

physiological tolerances of lizard species may favor benign conditions and cannot tolerate highly variable environments (Currie 1991; Costa et al. 2007). It is known that ground frost generates at very low temperatures. In China, climate varies dramatically from place to place. Most areas have two distinct seasons, a dry and cold winter and a wet and warm summer. Such FF formation due to variation and seasonality in temperature may affect species richness because some species may not be able to physiologically tolerate the harsh conditions. Many may become inactive in times of very low temperature especially in Palearctic realms, where winter temperature is below the freezing point of body fluids. Elevation range (ELE) and vegetation (VEG), the count of 300 m elevation range within each quadrat and the number of habitats within a cell, respectively, was positively correlated with lizard species richness, supporting the habitat heterogeneity hypothesis. The habitat heterogeneity-species richness relationship has an intuitive mechanistic basis: in general, few species are habitat generalists, and thus the addition of new habitats to a sample should lead to an increase in regional species in total diversity, and probably relates to regional diversity through species turnover (beta diversity) (Kerr et al. 2001). Our conclusions agree with previous studies of other species, in which the species richness positively correlates

123


1409

with habitat heterogeneity (e.g., Kerr and Packer 1997; Kerr et al. 2001; Diniz-Filho and Bini 2005; Hortal et al. 2009; Powney et al. 2010). There are conservation implications for the relationships between habitat diversity and species richness across very large areas. This suggests that the decline species number should be anticipated after substantial loss in habitat diversity, and that preservation of habitats is a general requirement for successful conservation, however, climate-based models of species richness are neutral on this issue and have an equivocal mechanistic basis (Kerr et al. 2001). Based on the available evidence, the distribution of rainfall has a stronger influence on diversity gradients than temperature over most of the globe (Hawkins et al. 2003a). It appears that energy is a strong predictor of animal diversity gradients in a small part of the planet. In our results, WET has a positive influence on lizard richness (t = 4.28, P \ 0.01). Our analysis thus provides evidence that water availability places strong constraints on diversity in the temperate zone (Hawkins et al. 2003a). Several studies have suggested that available water is associated with lizard richness and other animal species (Scheibe 1987; Rahbek and Graves 2001; Hawkins et al. 2003b; Costa et al. 2007). It appears contradictory to the physiological independence of most lizard species from aquatic or moist situations. However, there are a few plausible explanations. One probability is that ambient energy is limited in the far north where it is likely to be in short supply, but in the subtropics and tropics, a lack of energy inputs is almost certainly not an issue (rather, too much energy may be a problem), thus water becomes the primary limiting factor (Hawkins et al. 2003b). An alternative explanation is in accord with this situation: the range of lizard species richness values occurs in group of grid cells oriented roughly north to south and intermediate in precipitation and productivity (from the short grass plains of Mongolia-Xinjiang China to the tree association of Southern China and Southwestern China). Thus this does not only dependent upon the extreme segments of a moisture-productivity gradient and it may relate to changes in vegetation (Owen 1989). We found that VEG diversity was one of best predictors for lizard richness and increasing VEG diversity results in increasing lizard richness. Besides just understanding species richness in relation to a great variety of environmental variables, our results can also help to improve conservation strategies. The land cover and land use change in China have been dramatically shrinking forest and expanding cropland and urban areas during the recently years (Liu and Tian 2010). Although the implementation of ‘‘returning arable land into woodland or grassland’’ policies has initially succeeded in some areas, it is too early to say that the trend of deforestation has been effectively reversed through China (Liu et al. 2003). The areas corresponding to the Southern China, the Southwestern China (a part of Mountains of Southwest China hotspot, Myers et al. 2000; Mittermeier et al. 2005) and Central China harbor large patches of high lizard richness (Fig. 2B), but they are of special attention because these areas have common characteristics of the serious trend of wood land destroy, the decrease of grassland, built-up and residential expansion (Liu et al. 2003). Beyond the current level of vegetation deterioration or deforestation, the climate ongoing changes (Wang and Gong 2000; Xu and Ren 2004) in these areas will also threat to lizard richness. Therefore, to conserve species richness in these regions, it needs take action in future to maintain and protect existing reserves, establish ecological corridors among existing reserves, and build new nature reserves (Costa et al. 2007). The identification of conservation priority helps concentrate conservation efforts to maximize human and financial support (Myers et al. 2000; SechrestW et al. 2000). The northwest Mongolia-Xinjiang China, west-south Qinghai-Tibet China areas (Fig. 2B) show high levels of lizard species richness. Fortunately, there preserved most of the remaining pristine native areas in this region, however, recent infrastructure development,

123

1410


deforestation and demand for rapid economic development are threating this region. Therefore it needs more intensive fieldwork to confirm whether there exist undescribed species in these areas, though these areas, especially west-south Qinghai-Tibet China areas, are characterized by high relief and steep mountainous terrain, which may depict difficultly species’ distribution range. If necessary, it may need to create protected area in future. It will be important to establish conservation planning on biodiversity, and thus it is crucial to exploit as much as possible of these available data. However, based on the results of species niche modeling, studies cannot be considered unequivocally true, and it is necessary to permit identification via detailed fieldwork to ensure that proper decisions are being made (Costa et al. 2007). This validation approach has demonstrated considerable predictive power of models generated with scarce locality records, for example in targeting field surveys to accelerate the discovery of unknown populations of known or unknown species and species (Raxworthy et al. 2003; Pearson et al. 2007), and can be highly informative to conservation efforts in identifying suitable sites (Martıńez-Meyer et al. 2006). More detailed field surveys, such as in areas where only indirect evidence of the species’ presence was recorded, will likely maximize utility of ecological niche modeling as a tool for locating new populations and individuals of Gymnocladus assamicus species, which is a critically rare and endangered tree species endemic to northeastern India (Menon et al. 2010). In addition, ecological niche modeling usually emphasizes the role of abiotic factors and ignores biotic interactions (e.g., competition) (Costa and Schlupp 2010), and cannot account for historical factors, such as geographical barriers, resulting in speciation events (Costa et al. 2007). To conclude, the current alternative hypotheses are not mutually exclusive and may work together and best explain patterns of lizard species richness in China. Based on results of the model selection, our conclusion is in concordant with several previous studies that multiple hypotheses may best account for species richness patterns (Andrews and O’Brien 2000; Bohning-Gaese 1997; Diniz-Filho and Bini 2005; Costa et al. 2007). Clearly, a variety of factors works synergistically to determine species richness patterns. Our results indicate significant conservation implications, and habitat heterogeneity would be taken into account as an assessment of the threat to endemism from habitat loss in the future investigation. Lizards in China might have experienced large radiations and adapted to dramatic climatic fluctuations after the uplifting of the Tibetan Plateau in Pleistocene. For example, the diversification of the toad headed agamas Phrynocephalus (e.g., oviparous and viviparous group) and Gekkonid lizards of the Genus Teratoscincus resulted from the Tibetan Plateau (Qinghai-Xizang) uplifting after the India–Eurasian continental collision in the middle Eocene, since climatic and topographical changes such as gradual aridization occurred (Macey et al. 1999; Guo and Wang 2007). For future studies, it is important to test species richness distribution in East Asia at different spatial extent and sample resolution (Willis and Whittaker 2002), as well as to explore other factors known to affect species richness, such as historical factors and biotic interactions (e.g., competition, predation, and parasitism). Acknowledgments We thank S. Meiri, J. Fu, P. Guo and anonymous reviewers for comments and suggestions on earlier versions of the manuscript. We are grateful to J.A.F. Diniz-Filho, F. L.V.B. Rangel, and L.M. Bini for advice on how to use the PCNM spatial filters in SAM, and A.T. Peterson for advice on DesktopGarp. We thank L. Buckley, L. Anselin, B. Hawkins, K. Gaston, R. Davies, A. Lobo, K. Matsuura, F. Hall, Z.H. Li, H.X. Li, and our group members P. Yang for their great helps and providing suggestions for this study. This work was supported by the National Natural Science Foundation of China (grant no. 30470252) and Basic Research Program of the National Science and Technology (grant no. 2006FY110500-3).

123


1411

Appendix 1 List of the 151 species used in the analysis. Ablepharus deserti, Acanthosaura lepidogaster, Alsophylax pipiens, A. przewalskii, Asymblepharus alaicus, Ateuchosaurus chinensis, Calotes emma, C. jerdoni, C. kakhienensis, C. medogensis, C. mystaceus, C. versicolor, Cyrtopodion elongatus, C. brevipes, C. microlepis, Cyrtopodion khasiensis, C. medogensis, C. russowi, C. tibetanus, C. yarkandensis, Dibamus bourreti, Draco blanfordii, D. maculatus, Eremias argus, E. arguta, E. brenchleyi, E. grammica, E. multiocellata, E. przewalskii, E. quadrifrons, E. velox, E. vermiculata, Eumeces capito, E. chinensis, E. elegans, E. liui, E. popei, E. quadrilineatus, E. tunganus, Gehyra mutilata, Gekko auriverrucosus, G. chinensis, G. gecko, G. hokouensis, G. japonicus, G. liboensis, G. scabridus, G. subpalmatus, G. swinhonis, G. taibaiensis, Goniurosaurus hainanensis, G. lichtenfelderi, Hemidactylus bowringii, H. brooki, H. frenatus, H. garnotii, Hemiphyllodactylus typus, H. yunnanensis, Japalura andersoniana, J. bapoensis, J. batangensis, J. dymondi, J. flaviceps, J. grahami, J. kumaonensis, J. micangshanensis, J. splendida, J. szechwanensis, J. varcoae, J. yunnanensis, J. zhaoermii, Lacerta agilis, L. vivipara, Laudakia himalayana, L. papenfussi, L. sacra, L. stoliczkana, L. tarimensis, L. tuberculata, L. wui, Leiolepis reevesii, Mabuya longicaudata, M. multifasciata, Ophisaurus gracilis, O. hainanensis, O. harti, Oriocalotes paulus, Phrynocephalus acutirostris, P. affins, P. albolineatus, P. axillaris, P. erythrurus, P. forsythii, P. frontalis, P. grumgrzimailoi, P. guttaus, P. haeckeli, P. helioscopus, P. hongyuanensis, P. immaculatus, P. koslowi, P. ludovici, P. mystaceus, P. nasatus, P. przewalskii, P. putjatae, P. theobaldi, P. versicolor, P. vlangalii, P. zetangensis, Physignathus cocincinus, Platyurus platyurus, Platyplacopus intermedius, P. kuehnei, P. sylvaticus, Ptyctolaemus gularis, Scincella septentrionalis, S. barbouri, S. doriae, S. himalayana, S. huanrenensis, S. ladacensis, S. modesta, S. monticola, S. potanini, S. przewalskii, S. reevesii, S. schmidti, S. sikimmensis, S. tsinlingensis, Shinisaurus crocodilurus, Sphenomorphus courcyanus, S. incognitus, S. indicus, S. maculatus, Takydromus amurensis, T. septentrionalis, T. sexlineatus, T. wolteri, Teratoscincus przewalskii, T. robrowskii, T. scincus, T. toksunicus, Trapelus sanguinolenta, Tropidophorus berdmorei, T. guangxiensis, T. hainanus, T. sinicus, Varanus bengalensis, V. irrawadicus, V. salvator.

Appendix 2 See Table 4.

Table 4 Correlation matrix of all environmental variables PET PET NDVI RAD ELE VEG WET

–

NDVI

RAD

0.60**

-0.41**

–

-0.63** –

ELE 0.25** -0.02 0.12** –

VEG

WET

FF

REH

WIND

0.47**

0.62**

-0.70**

0.59**

-0.51**

0.72**

0.78**

-0.65**

0.80**

-0.51**

-0.60**

-0.50**

0.66**

-0.69**

0.29**

0.19**

0.13**

0.15**

-0.19**

-0.26**

–

0.65**

-0.54**

0.64**

-0.49**

–

-0.46**

0.74**

-0.53**

123

1412


Table 4 continued PET

NDVI

RAD

ELE

VEG

FF REH WIND

WET

FF –

REH -0.78** –

WIND 0.44** -0.34** –

The variables were selected in a way to minimize the correlation among each other and to relate to different hypothesis to explain species richness patterns ** Correlation is significant at the 0.01 level (2-tailed) * Correlation is significant at the 0.05 level (2-tailed)

References Ahn CH, Tateishi R (1994) Development of a global 30-minute grid potential evapotranspiration data set. J Jpn Soc Photogramm Remote Sens 33:12–21 Anderson DR, Burnham KP, Thompson WL (2000) Null hypothesis testing: problems, prevalence, and an alternative. J Wildlife Manage 64:912–923 Anderson RP, Lew D, Peterson AT (2003) Evaluating predictive models of species’ distributions: criteria for selecting optimal models. Ecol Model 162:211–232 Andrews P, O’Brien EM (2000) Climate, vegetation, and predictable gradients in mammal species richness in southern Africa. J Zool 251:205–231 Badgley C, Fox DL (2000) Ecological biogeography of North American mammals: species density and ecological structure in relation to environmental gradients. J Biogeogr 27:1437–1467 Bohning-Gaese K (1997) Determinants of avian species richness at different spatial scales. J Biogeogr 24:49–60 Buckley LB, Jetz W (2007) Environmental and historical constraints on global patterns of amphibian richness. Proc Biol Sci 274(1614):1167–1173 Burnham KP, Anderson DR (1998) Model selection and inference: a practical information theoretic approach. Springer-Verlag, New York Burnham KP, Anderson DR (2004) Multimodel inference—understanding AIC and BIC in model selection. Sociol Methods Res 33:261–304 Ceballos G, Ehrlich PR (2006) Global mammal distributions, biodiversity hotspots, and conservation. Proc Natl Acad Sci USA 103:19374–19379 Costa GC, Schlupp I (2010) Biogeography of the Amazon molly: ecological niche and range limits of an asexual hybrid species. Glob Ecol Biogeogr 19:442–451 Costa GC, Nogueira C, Machado RB, Colli GR (2007) Squamate richness in the Brazilian Cerrado and its environmental-climatic associations. Divers Distrib 13(6):714–724 Costa GC, Nogueira C, Machado RB, Colli GR (2010) Sampling bias and the use of ecological niche modeling in conservation planning: a field evaluation in a biodiversity hotspot. Biodivers Conserv 19:883–899 Currie DJ (1991) Energy and large-scale patterns of animal- and plant-species richness. Am Nat 137:27–49 Diniz-Filho JAF, Bini LM (2005) Modelling geographical patterns in species richness using eigenvectorbased spatial filters. Glob Ecol Biogeogr 14:177–185 Diniz-Filho JAF, Bini LM, Hawkins BA (2003) Spatial autocorrelation and red herrings in geographical ecology. Glob Ecol Biogeogr 12:53–64 Ferreira de Siqueira M, Durigan G, de Marco Juńior P, Peterson AT (2009) Something from nothing: using landscape similarity and ecological niche modeling to find rare plant species. J Nat Conserv 17:25–32 Fu C, Wang J, Pu Z, Zhang S, Chen H, Zhao B, Chen J, Wu J (2007) Elevational gradients of diversity for lizards and snakes in the Hengduan Mountains, China. Biodivers Conserv 16(3):707–726 Gaston KJ (2000) Global patterns in biodiversity. Nature 405:220–227 Gibbons JW, Scott DE, Ryan TJ, Buhlmann KA, Tuberville TD, Metts BS, Greene JL, Mills T, Leiden Y, Poppy S, Winne CT (2000) The global decline of reptiles, de´ja`‘ vu amphibians. BioScience 50: 653–666 Guo X, Wang Y (2007) Partitioned Bayesian analyses, dispersal-vicariance analysis, and the biogeography of Chinese toad-headed lizards (Agamidae: Phrynocephalus): a re-evaluation. Mol Phylogenet Evol 45(2):643–662

123


1413

Hawkins BA, Porter EE (2003) Water-energy balance and the geographic pattern of species richness of western Palearctic butterflies. Ecol Entomol 28:678–686 Hawkins BA, Field R, Cornell HV, Currie DJ, Guegan JF, Kaufman DM, Kerr JT, Mittelbach GG, Oberdorff T, O’Brien EM, Porter EE, Turner JRG (2003a) Energy, water, and broad-scale geographic patterns of species richness. Ecology 84(12):3105–3117 Hawkins BA, Porter EE, Diniz JAF (2003b) Productivity and history as predictors of the latitudinal diversity gradient of terrestrial birds. Ecology 84(6):1608–1623 Hawkins BA, Diniz-Filho JAF, Jaramillo CA, Soeller SA (2007) Climate, niche conservatism, and the global bird diversity gradient. Am Nat 170:S16–S27 Hortal J, Triantis KA, Meiri S, The´bault E, Sfenthourakis S (2009) Island species richness increases with habitat diversity. Am Nat 174(6):E205–E217 Kerr JT, Packer L (1997) Habitat heterogeneity as a determinant of mammal species richness in high-energy regions. Nature 385(6613):252–254 Kerr JT, Southwood TRE, Cihlar J (2001) Remotely sensed habitat diversity predicts butterfly species richness and community similarity in Canada. Proc Natl Acad Sci USA 98:11365–11370 Legendre P (1993) Spatial autocorrelation—trouble or new paradigm. Ecology 74(6):1659–1673 Legendre P, Dale MRT, Fortin MJ, Gurevitch J, Hohn M, Myers D (2002) The consequences of spatial structure for the design and analysis of ecological field surveys. Ecography 25(5):601–615 Liu M, Tian H (2010) China’s land cover and land use change from 1700 to 2005: estimations from highresolution satellite data and historical archives. Global Biogeochem Cycles 24: GB3003. doi: 10.1029/2009GB003687 Liu JY, Liu ML, Zhuang DF, Zhang ZX, Deng XZ (2003) Study on spatial pattern of land-use change in China during 1995–2000. Sci China Ser D 46(4):373–384. doi:10.1360/03yd9033 Macey JR, Wang Y, Ananjeva NB, Larson A, Papenfuss TJ (1999) Vicariant patterns of fragmentation among Gekkonid Lizards of the Genus Teratoscincus produced by the Indian Collision: a molecular phylogenetic perspective and an area cladogram for Central Asia. Mol Phylogenet Evol 12(3):320–332 Martıńez-Meyer E, Peterson AT, Servıń J, Kiff LF (2006) Ecological niche modelling and prioritizing areas for species reintroductions. Oryx 40:411–418 McCain CM (2010) Global analysis of reptile elevational diversity. Glob Ecol Biogeogr 19:541–553 Menon S, Choudhury BI, Khan LM, Peterson AT (2010) Ecological niche modeling and local knowledge predict new populations of Gymnocladus assamicus a critically endangered tree species. Endanger Species Res 11:175–181 Mittelbach GG, Steiner CF, Scheiner SM, Gross KL, Reynolds HL, Waide RB, Willig MR, Dodson SI, Gough L (2001) What is the observed relationship between species richness and productivity? Ecology 82:2381–2396 Mittermeier RA, Gil PR, Hoffman M, Pilgrim J, Brooks T, Mittermeier CJ, Lamoreux J, da Fonseca GAB (2005) Hotspots revisited: earth’s biologically richest and most endangered terrestrial ecoregions, 2nd edn. Conservation International, Arlington, VA Myers N, Mittermeier RA, Mittermeier CG, da Fonseca GAB, Kent J (2000) Biodiversity hotspots for conservation priorities. Nature 403:853–858 New M, Lister D, Hulme M, Makin I (2002) A high-resolution data set of surface climate over global land areas. Clim Res 21:1–25 Oberdoff T, Gue´gan JF, Hugueny B (1995) Global scale patterns of fish species richness in rivers. Ecography 18:345–352 Owen JG (1989) Patterns of herpetofaunal species-relation to temperature, precipitation, and variance in elevation. J Biogeogr 16(2):141–150 Pearson RG, Raxworthy C, Nakamura M, Peterson AT (2007) Predicting species’ distributions from small numbers of occurrence records: a test case using cryptic geckos in Madagascar. J Biogeogr 34:102–117 Peterson AT, Robins CR (2003) Using ecological-niche modeling to predict barred owl invasions with implications for spotted owl conservation. Conserv Biol 17:1161–1165 Peterson AT, Papeß M, Kluza DA (2003) Predicting the potential invasive distributions of four alien plant species in North America. Weed Sci 51:863–868 Peterson AT, Papefl M, Eaton M (2007) Transferability and model evaluation in ecological niche modeling: a comparison of GARP and Maxent. Ecography 30:550–560 Pianka ER (1967) On lizard species diversity—north American flatland deserts. Ecology 48(3):333–351 Pounds JA, Fogden MPL, Campbell JH (1999) Biological response to climate change on a tropical mountain. Nature 398:611–615 Powney GD, Grenyer R, Orme CDL, Owens IPF, Meiri S (2010) Hot, dry and different Australian lizard richness is unlike that of mammals, amphibians, and birds. Glob Ecol Biogeogr 19:386–396

123

1414


Qian H, Wang XH, Wang SL, Li YL (2007) Environmental determinants of amphibian and reptile species richness in China. Ecography 30:471–482 Rahbek C, Graves GR (2001) Multiscale assessment of patterns of avian species richness. PNAS 98(8): 4534–4539 Rangel TFLVB, Diniz-Filho JAF, Bini LM (2006) Towards an integrated computational tool for spatial analysis in macroecology and biogeography. Glob Ecol Biogeogr 15:321–327 Rangel TF, Diniz-Filho JAF, Bini LM (2010) SAM: a comprehensive application for spatial analysis in macroecology. Ecography 33:46–50 Raxworthy C, Martıńez-Meyer E, Horning N, Nussbaum R, Schneider G, Ortega-Huerta M, Peterson AT (2003) Predicting distributions of known and unknown reptile species in Madagascar. Nature 426:837–841 ´ , Belmontes JA, Hawkins BA (2005) Energy, water and large-scale patterns of reptile and Rodrı´guez MA amphibian species richness in Europe. Acta Oecol 28:65–70 Schall JJ, Pianka ER (1978) Geographical trends in numbers of species. Science 201(4357):679–686 Scheibe JS (1987) Climate, competition, and the structure of temperate zone lizard communities. Ecol 68:1424–1436 Sechrest W, Brooks TM, da Fonseca GAB, Konstant WR, Mittermier RA, Purvis A, Rylands AB, Gittleman JL (2000) Hotspots and the conservation of evolutionary history. Proc Natl Acad Sci USA 99: 2067–2071 Stockwell D, Peters D (1999) The GARP modelling system: problems and solutions to automated spatial prediction. Int J Geogr Inform Sci 13:143–158 Tabachnick BG, Fidell LS (2000) Using multivariate statistics, 4th edn. Allyn and Bacon, Boston, MA Tedesco PA, Oberdorff T, Lasso CA, Zapata M, Hugueny B (2005) Evidence of history in explaining diversity patterns in tropical riverine fish. J Biogeogr 32(11):1899–1907 ´ , Morales-Castilla I, Rueda M, Vidanes RM, Rodrı´guez MA ´ , Diniz-Filho Terribile LC, Olalla-Ta´rraga MA JAF (2009) Global richness patterns of venomous snakes reveal contrasting influences of ecology and history in two different clades. Oecologia 159:617–626 Tognelli MF, Kelt DA (2004) Analysis of determinants of mammalian species richness in South America using spatial autoregressive models. Ecography 27:427–436 Turner JRG, Gatehouse CM, Corey CA (1987) Does solar energy control organic diversity? Butterflies, moths and British climate. Oikos 48:195–205 van Rensburg BJ, Chown SL, Gaston KJ (2002) Species richness, environmental correlates, and spatial scale: a test using South African birds. Am Nat 159:566–577 von Humboldt A (1808) Ansichten der Natur mit wissenschaftlichen Erlauterungen. J. G. Cotta, Tu¨bingen, Germany Wang SW, Gong DY (2000) Enhancement of the warming trend in China. Geophys Res Lett 27(16): 2581–2584 Willis KJ, Whittaker RJ (2002) Ecology: enhanced: species diversity-scale matters. Science 295(5558): 1245–1248 Wright DH (1983) Species–energy theory: an extension of species–area theory. Oikos 41:496–506 Xu MZ, Ren GY (2004) Change in growing season over China: 1961–2000. J Appl Meteorol Sci 15(3): 306–312 Zhao EM, Zhao KT, Zhou KY (1999) Fauna sinica, Reptilia, vol 2. Science press, (Squamata, Lacertilia). Beijing

123