Relationships between Plant and Animal Species Richness at a ... - FAU

Relationships between Plant and Animal Species Richness at a Regional Scale in China HONG QIAN Research and Collections Center, Illinois State Museum, 1011 East Ash Street, Springfield, IL 62703, U.S.A., email [email protected]

Abstract: Important questions in conservation biology and ecology include whether species diversities of different groups of organisms are correlated and, in particular, whether plant diversity influences animal diversity. I used correlation and partial regression analyses to examine the relationships between species richness of vascular plants and four major groups of terrestrial vertebrates (mammals, amphibians, reptiles, and birds) in 28 provinces in China. Species richness data were obtained from the literature. Environmental variables included normalized difference vegetation index, mean January temperature, mean annual temperature, annual precipitation, May through August precipitation, actual evapotranspiration, potential evapotranspiration, and elevation range. Species richness was strongly and positively correlated among the five groups of organisms. Plant richness was correlated with animal richness more strongly than the richness of different animal groups correlated with each other except for reptile richness, which had a slightly higher correlation with amphibian richness than with plant richness. Plant richness uniquely explained 41 times more variance in the species richness of the four vertebrate groups combined than environmental variables uniquely did, suggesting that plant richness influences terrestrial vertebrate richness at the regional scale examined. Because of strong correlations between the diversity of vascular plants and vertebrates, the diversity of vascular plants may be used as a surrogate for the diversity of terrestrial animals in China. My results have implications for selection of areas to be protected at both regional and local scales.

Keywords: cross-taxon congruence, species diversity, species richness, species richness correlation, terrestrial vertebrates, vascular plants Relaciones de Riqueza de Especies entre Taxa de Plantas y Animales a una Escala Regional en China

Resumen: Algunas de las preguntas más importantes en biolog´ıa y ecolog´ıa de la conservación incluyen si las diversidades de especies de diferentes grupos de organismos est´ an correlacionadas y, en particular, si la diversidad de plantas influye en la diversidad de animales. Utilicé an´ alisis de correlaci´ on y regresi´ on parcial para examinar las relaciones entre la riqueza de especies de plantas vasculares y de cuatro grupos principales de vertebrados terrestres (mam´ıferos, anfibios, reptiles y aves) en 28 provincias de China. Los datos de riqueza de especies fueron obtenidos de la literatura. Las variables ambientales incluyeron el ´ındice de diferencia en la vegetaci´ on normalizado, la temperatura media de enero, la temperatura media anual, la precipitaci´ on anual, la precipitaci´ on mayo – agosto, la evapotranspiraci´ on real y potencial y el rango de elevaci´ on. La riqueza de especies se correlacion´ o fuerte y positivamente entre los cinco grupos de organismos. La riqueza de especies de plantas se correlacion´ o con la de animales m´ as fuertemente que la riqueza de diferentes grupos de animales entre s´ı, excepto la riqueza de reptiles, que tuvo una correlaci´ on ligeramente mayor con la riqueza de anfibios que con la de plantas. La riqueza de plantas explic´ o 41 veces m´ as que las variables ambientales la varianza en la riqueza de especies de los cuatro grupos de vertebrados combinados, lo que sugiere que la riqueza de plantas influye en la riqueza de vertebrados terrestres en la escala regional examinada. Debido a las fuertes correlaciones entre la diversidad de plantas vasculares y de vertebrados, la diversidad de plantas vasculares puede ser utilizada como un sustituto para la diversidad de animales terrestres en China. Nuestros resultados tienen implicaciones para la selecci´ on de a ´ reas a proteger, tanto a escala regional como local.

Paper submitted June 20, 2006; revised manuscript accepted December 21, 2006.

937 Conservation Biology Volume 21, No. 4, 937–944 C 2007 Society for Conservation Biology DOI: 10.1111/j.1523-1739.2007.00692.x

938

Cross-Taxon Congruence of Species Richness

Qian

Palabras Clave: congruencia trans–taxón, correlación de riqueza de especies, diversidad de especies, plantas vasculares, riqueza de especies, vertebrados terrestres

Introduction Geographic variation in numbers of species per unit area (species richness) is one of the most conspicuous patterns in biodiversity (Lennon et al. 2004). Understanding relationships of species richness among different groups of organisms is a major task of conservation biologists and ecologists. Cross-taxon relationships in species richness have been studied in a number of areas and for a variety of taxa at a range of spatial scales, and positive relationships have been found in most studies (Wolters et al. 2006). Conservation biologists have begun to use species diversity of certain taxa as indicators of species diversity of other taxa and to use these data to identify areas to be protected (e.g., Pearson & Carroll 1999; Myers et al. 2000; Moore et al. 2003). The cross-taxon relationship in species richness is scale dependent. Strong and positive cross-taxon relationships of species richness have been found frequently in broadscale analyses, whereas weak or no relationships between species richness of different taxa are more often found at finer scales (Wolters et al. 2006). The majority (∼90%) of the reported cross-taxon relationships have been examined for floras and faunas at scales of 1000 km2 involved plants (Wolters et al. 2006). Understanding the broad-scale relationships of plant diversity with animal diversity in terrestrial systems is crucial because plants are essential to virtually all forms of animal life (Myers et al. 2000), plant diversity influences animal diversity (Hutchinson 1959), and biodiversity hotspots (usually > 1000 km2 ) have been identified for conservation based primarily on plants (Myers et al. 2000). The provincial floras and faunas of China are ideal subjects to examine cross-taxon relationships in species richness for three reasons. First, these floras and faunas cover a wide range of latitudes and longitudes, and environment and species richness vary greatly from south to north (a striking warm-to-cold gradient) and from east to west (a striking wet-to-dry gradient) (Qian & Ricklefs 1999; Qian 2002). Second, China is one of the richest nations in the world in terms of species diversity of both plants and animals—2 of the 25 world biodiversity hotspots (Myers et al. 2000) are located in China. Third, species richness in each province has been well documented through efforts of multiple generations of botanists and zoologists. I examined correlations in species richness between vascular plants and four major groups (classes) of terrestrial vertebrates (mammals, birds, amphibians, and

Conservation Biology Volume 21, No. 4, 2007

reptiles) in provincial biotas in China and investigated whether plant richness has the power to explain variance in animal richness when environmental effects on cross-taxon richness relationships are controlled for.

Methods I examined the floras and faunas in all of China’s 23 provinces and five autonomous regions (hereafter referred to as provinces), which encompass 9.6 million km2 . I combined the cities of Beijing and Tianjin with Hebei Province; Shanghai with Zhejiang Province; Chongqing with Sichuan Province; and Hong Kong, Macau, and Shenzhen with Guangdong Province. The numbers of species of vascular plants, mammals, amphibians, reptiles, and birds in each province were documented based on complete species lists for the provinces published in Wu and Ding (1999), Cheng (2000), Wang (2002), Ji and Wen (2002), Fei et al. (2005), and other sources. I examined the following six environmental variables for each province: normalized difference vegetation index (NDVI, ratio); mean January temperature (TEM1, ◦ C); mean annual temperature (TEM, ◦ C); annual precipitation (PRE, mm); May through August (“summer”) precipitation (SRAIN, mm); annual actual evapotranspiration (AET, mm); and annual potential evapotranspiration (PET, mm). These variables were good representations of the major environmental variables used in previous studies (e.g., Currie 1991; Hawkins & Pausas 2004) that examined the relationships between species richness and the environment at broad spatial scales. NDVI was calculated as a normalized ratio between red (ρ red ) and near infrared (ρ nir ) bands with the following formula (Tucker 1979): NDVI = (ρ red – ρ nir )/(ρ red + ρ nir ). Data for the NDVI were originally derived from a new generation of advanced optical sensors: the VEGETATION (VGT) sensor onboard the SPOT-4 satellite launched in 1998 and the Moderate Resolution Imaging Spectroradiometer (MODIS) sensor onboard the Terra satellite launched in 1999 (Xiao et al. 2004). The NDVI is a measure of the greenness of the plant canopy and standing crop. Because the NDVI is strongly and positively correlated with primary productivity and plant mass, it is widely used as a measure of plant productivity (e.g., Hurlbert & Haskell 2003). Temperature data (TEM1, TEM7, and TEM) have been used widely as a measure of ambient energy input (Schall

Qian

& Pianka 1978; Currie 1991; Rodr´ıguez et al. 2005). The PET is also considered a measure of ambient energy (Hawkins & Porter 2003). The AET is considered a suitable productive energy metric (Evans et al. 2005). Data for these environmental variables were obtained from three global data sets for pixels of 0.5◦ of latitude and longitude. Information on sources of the data sets is available in Xiao et al. (2004) for the NDVI, New et al. (1999) for temperature and precipitation, and Tateishi and Ahn (1996) for the AET and PET. I extracted all 0.5◦ pixels for all of China from these data sets, assigned each pixel to a province according to its midpoint location, and calculated averages of the six variables for each province. These averages were used to represent average environmental conditions of the provinces. The six variables were strongly correlated (Pearson correlation coefficients ranging from 0.81 to 0.95). Because colinearity between explanatory variables may lead to biased parameter estimates in multivariate statistic analyses, I transformed the environmental variables into mutually unrelated variables through a principal component analysis (PCA) based on the correlation matrix of the variables (Legendre & Legendre 1998). I also documented elevation range and area for each province. Elevation range is considered as a measure of habitat heterogeneity (e.g., Rahbek & Graves 2001). It is well known that area influences species richness, and the size of the provinces I examined varied. To remove the area effect on the relationships between species richness of different taxa and the relationships between species richness and environment, I log 10 transformed area to normalize the distribution of variance and then regressed species richness of each taxonomic group against log 10 transformed area and used residual richness as the dependent variable in further analyses. The use of this statistical approach to control for sample area is common in ecological studies, including those on cross-taxon congruence of species richness (e.g., Howard et al. 1998; Lamoreux et al. 2006; Tushabe et al. 2006). I carried out a series of correlation analyses to evaluate relationships between pairs of different taxonomic groups and conducted a series of partial regressions (Legendre & Legendre 1998) to determine the degree to which environment and plant richness each can uniquely explain the variance of animal richness. Partial regressions partitioned the variance in the richness of animal groups explained by plant richness and environmental variables into three fractions: explained uniquely by plant richness, explained uniquely by environmental variables, and explained jointly by plant richness and environmental variables. Following Lamoreux et al. (2006), I used the following standards to interpret the strength of correlation coefficients: high correlations were nearly 0.50 and higher, moderate correlations were around 0.30, and low correlations were about 0.10. I used PC-ORD (version 4; McCune & Mefford 1999) for PCA, and SYSTAT (version


939

7; Wilkinson et al. 1992) for correlation and regression analyses. Spatial autocorrelation occurs in most, if not all, large-scale ecological data and may inflate the rate of Type I error in a significance test (Diniz-Filho et al. 2003). To remedy this problem, I used MODTTEST (Legendre 2000) to recalculate p values for significance tests based on reduced degrees of freedom corrected for spatial autocorrelation, which was determined according to Dutilleul’s (1993) approach.

Results China has approximately 29,200 species of vascular plants, 607 species of mammals, 325 species of amphibians, 384 species of reptiles, and 1244 species of birds. Each province of China has, on average, 3861.5, 113.8, 35.4, 65.9, and 435.3 species of vascular plants, mammals, amphibians, reptiles, and birds, respectively (Table 1). Low-to-moderate correlation coefficients (the highest r = 0.25) between log area and species richness of the five groups of organisms indicated that the variation in area among provinces in China did not have a strong effect on species richness. This is likely because the large provinces in China, such as Qinghai, Xingjiang, and Neimenggu, are located in the western and northwestern China, where climates are dry and cold and thus have low species richness. Correlations between species richness of plants and terrestrial vertebrates were high regardless of whether raw richness or residual richness was used (Table 2). Table 1. Descriptive statistics of geographic and environmental variables and species richness in the provincial floras and faunas of China (n = 28).

Min

Max

Mean

SD

Geography area (km2 ) 33,900 1,646,797 345,516 392,297 elevation range (m) 642 8765 3270 2191 Environment∗ NDVI (ratio) 0.1 0.6 0.4 0.1 SRAIN (mm) 64.5 1119.8 548.7 267.0 TEM1 (◦ C) −22.1 18.6 −1.4 10.5 TEM (◦ C) −2.4 24.1 11.5 7.0 AET (mm) 101.2 1095.7 694.9 283.4 PET (mm) 541.4 1436.5 893.4 209.2 Species richness plants 887 14,307 3861.5 2826.4 mammals 37 296 113.8 53.9 amphibians 6 117 35.4 28.7 reptiles 11 156 65.9 42.3 birds 285 814 435.3 114.7 ∗ Key:

NDVI, normalized difference vegetation index; SRAIN, May through August (“summer”) precipitation; TEM1, mean January temperature; TEM, mean annual temperature; PRE, annual precipitation; AET, actual evapotranspiration; and PET, potential evapotranspiration.


940


Qian

Table 2. Pearson correlation coefficientsa of species richness among vascular plants, mammals, amphibians, reptiles, and birds in China.b

Taxona Mammals Amphibians Reptiles Birds

Plants

Mammals

Amphibians

Reptiles

∗∗∗

0.823 0.837∗∗∗ 0.909∗∗∗ 0.924∗∗∗ 0.744∗∗ 0.829∗∗ 0.840∗∗∗ 0.829∗∗∗

Axis ∗∗

0.733 0.820∗∗∗ 0.516∗ 0.725∗∗ 0.781∗∗∗ 0.768∗∗∗

0.818∗∗ 0.845∗∗ 0.770∗∗∗ 0.789∗∗∗

0.675∗∗ 0.778∗∗∗

a For

each pair of taxa, the correlation coefficient in the upper row is for raw richness and that in the lower is row for residual richness (corrected for area). b The p values are calculated with spatial autocorrelation accounted for: ∗∗∗ p < 0.001; ∗∗ p < 0.01; ∗ p < 0.08.

Correlation coefficients were higher in 8 of the 10 crosstaxon comparisons (Table 2) when residual richness was used. The average correlation coefficient of the 10 comparisons was 0.76 for raw richness and 0.81 for residual richness. Plant richness was correlated with animal richness more strongly than the richness of different animal groups correlated with each other except for reptile richness, which had a slightly higher correlation with amphibian richness than with plant richness (0.845 vs. 0.829). The average correlation coefficient for comparisons between plants and animal groups was higher than that between different animal groups (0.85 ± 0.05 SD vs. 0.79 ± 0.04 SD). Of all the 10 comparisons, plants and amphibians had the highest correlation coefficient (Table 2). The combination of the four vertebrate groups had a correlation coefficient of 0.92 with plants when raw richness was used and 0.91 when residual richness was used. The df corrected for spatial autocorrelation was on average 13.5 (ranging from 7.7 to 19.2) for the 11 correlations when raw richness was used and 12.3 (8.1 to 17.1) when residual richness was used. After the effect of spatial autocorrelation being removed, all correlations reported in Table 2 and the correlations between plant richness and the richness of the four vertebrate groups combined were significant at p < 0.01 (Table 2), except that the correlation between the raw richness of mammals and the raw richness of reptiles was marginally significant (p = 0.08). The first three axes produced by the PCA accounted for 98% of the variance in the six environmental variables (Table 3). The first axis (PC1) alone explained 89% of the variance and had relatively even loadings for all six variables, suggesting that PC1 can be considered a surrogate for the six environmental variables. The PC2 represented primarily gradients for NDVI, TEM1, and AET; and the PC3 primarily represented TEM and PET. When both plant richness and the three PC axes plus elevation range were included as independent variables in a partial regression, these variables together explained


Table 3. Results of the first three axes (PC1 through PC3) of a principal component analysis of environmental variables considered in the examination of the relationships between species richness of vascular plants and terrestrial vertebrates in China.

Eigenvalue Variance (%) Cumulative variance (%) Component loading∗ NDVI SRAIN TEM1 TEM AET PET

PC1

PC2

PC3

5.35 89.18 89.18

0.32 5.31 94.48

0.19 3.15 97.63

−0.406 −0.412 −0.412 −0.406 −0.406 −0.408

−0.484 −0.229 0.507 0.346 −0.477 0.330

0.339 −0.405 0.077 0.611 −0.029 −0.584

∗ Key: NDVI, normalized difference vegetation index; SRAIN, May through August (“summer”) precipitation; TEM1, mean January temperature; TEM, mean annual temperature; PRE, annual precipitation; AET, actual evapotranspiration; and PET, potential evapotranspiration.

73%, 89%, 77%, 69%, and 84% of the variance in the richness of mammals, amphibians, reptiles, birds, and all the vertebrates combined, respectively. With the effect of spatial autocorrelation removed, all five regressions remained significant at p < 0.001 (Table 4). The variance in animal richness that was explained uniquely by plant richness and uniquely by environmental variables was, respectively, 23.1% and 3.0% for mammals, 17.0% and 3.3% for amphibians, 7.6% and 8.4% for reptiles, and 20.3% and 0.4% for birds. Thus, plant richness explained uniquely 7.7, 5.2, and 51 times more variance in the richness of mammals, amphibians, and birds, respectively, than environmental variables did uniquely and explained uniquely approximately as much variance in reptile richness as environmental variables uniquely did (Fig. 1). For all four groups of vertebrates combined,

Table 4. Coefficient of determination (r2 ), F ratio (F), original p value (p original ), and p value corrected for spatial autocorrelation (p) of partial regressions determined in the analysis of relationships between species richness of vascular plants and terrestrial vertebrates in China.∗

Animal group Mammals Amphibians Reptiles Birds All groups

r2

F

p original

p

0.731 0.887 0.772 0.692 0.840

11.9 34.4 14.9 9.9 23.1

1.1 × 10−5 2.0 × 10−9 2.0 × 10−6 4.7 × 10−5 4.4 × 10−8