Limits...
Geographical range and local abundance of tree species in China.

Ren H, Condit R, Chen B, Mi X, Cao M, Ye W, Hao Z, Ma K - PLoS ONE (2013)

Bottom Line: There was no relationship between local abundance and range size, and no evidence for species being more abundant towards their range-centers.Finally, species' abundances were positively correlated between sites.The widespread nature of most tree species in China suggests few are vulnerable to global extinction, and there is no indication of the double-peril that would result if rare species also had narrow ranges.

View Article: PubMed Central - PubMed

Affiliation: Key Laboratory of Vegetation and Environmental Change, Institute of Botany, Chinese Academy of Sciences, Beijing, P.R. China.

ABSTRACT
Most studies on the geographical distribution of species have utilized a few well-known taxa in Europe and North America, with little research in China and its wide range of climate and forest types. We assembled large datasets to quantify the geographic ranges of tree species in China and to test several biogeographic hypotheses: 1) whether locally abundant species tend to be geographically widespread; 2) whether species are more abundant towards their range-centers; and 3) how abundances are correlated between sites. Local abundances of 651 species were derived from four tree plots of 20-25 ha where all individuals ≥1 cm in stem diameter were mapped and identified taxonomically. Range sizes of these species across China were then estimated from over 460,000 geo-referenced records; a Bayesian approach was used, allowing careful measures of error of each range estimate. The log-transformed range sizes had a bell-shaped distribution with a median of 703,000 km(2), and >90% of 651 species had ranges >10(5) km(2). There was no relationship between local abundance and range size, and no evidence for species being more abundant towards their range-centers. Finally, species' abundances were positively correlated between sites. The widespread nature of most tree species in China suggests few are vulnerable to global extinction, and there is no indication of the double-peril that would result if rare species also had narrow ranges.

Show MeSH
Relationship between estimated range size and sample size of geo-referenced records.Panel A is for real species, and Panel B for random draws from the entire pool of geo-referenced records. In panel A, solid points indicate species from Gutianshan; open circles are for species endemic to China (any of the plots). The green solid line (lower) is a trend curve through the Gutianshan points, and the solid red (upper) line through the endemics. The dotted curve is the lower 95% confidence limit of random ranges (from Panel B). Panel B is a test for bias in estimated range size caused by sample size. The solid, central curve is the mean range size from 1000 draws at each value of N (sample number of geo-referenced records), and two dotted curves are 95% confidence intervals (the central 95 percentiles of 1000 draws). Samples were done at N = 3, 5, 10, 15, 20, 25, 30, 40, 50, 60, 70, 80, 90, 100, 150, 200, 400, 600, 800, 1000, 1500, 2000, 2500, 3000, 3500, 4000, 4500, 5000, 5500, 6000, 6500, 7000.
© Copyright Policy
Related In: Results  -  Collection


getmorefigures.php?uid=PMC3794993&req=5

pone-0076374-g003: Relationship between estimated range size and sample size of geo-referenced records.Panel A is for real species, and Panel B for random draws from the entire pool of geo-referenced records. In panel A, solid points indicate species from Gutianshan; open circles are for species endemic to China (any of the plots). The green solid line (lower) is a trend curve through the Gutianshan points, and the solid red (upper) line through the endemics. The dotted curve is the lower 95% confidence limit of random ranges (from Panel B). Panel B is a test for bias in estimated range size caused by sample size. The solid, central curve is the mean range size from 1000 draws at each value of N (sample number of geo-referenced records), and two dotted curves are 95% confidence intervals (the central 95 percentiles of 1000 draws). Samples were done at N = 3, 5, 10, 15, 20, 25, 30, 40, 50, 60, 70, 80, 90, 100, 150, 200, 400, 600, 800, 1000, 1500, 2000, 2500, 3000, 3500, 4000, 4500, 5000, 5500, 6000, 6500, 7000.

Mentions: This method was designed to describe the area over which a species was found. It does not address whether all habitats within the area are appropriate. It has the substantial advantage of discounting rare outliers in a quantitative way (there are several examples in Fig. 2 and Fig. 3), and use of the Bayesian fitting precisely described confidence in the estimated ranges. Traditional techniques to estimate range size of species, such as Polygon methods surrounding the observed points, cannot offer precise confidence limits and moreover are biased with respect to abundance. For comparison, we also estimated range sizes for each species using the Minimum Convex Polygon (MCP) estimator in R-package ‘adehabitatHR’ Version: 0.4.7 [46]. The R2 between range sizes estimated by the two techniques was very high up to 0.90, and the ranges of two species with few records were greatly underestimated by MCP (Figure S1).


Geographical range and local abundance of tree species in China.

Ren H, Condit R, Chen B, Mi X, Cao M, Ye W, Hao Z, Ma K - PLoS ONE (2013)

Relationship between estimated range size and sample size of geo-referenced records.Panel A is for real species, and Panel B for random draws from the entire pool of geo-referenced records. In panel A, solid points indicate species from Gutianshan; open circles are for species endemic to China (any of the plots). The green solid line (lower) is a trend curve through the Gutianshan points, and the solid red (upper) line through the endemics. The dotted curve is the lower 95% confidence limit of random ranges (from Panel B). Panel B is a test for bias in estimated range size caused by sample size. The solid, central curve is the mean range size from 1000 draws at each value of N (sample number of geo-referenced records), and two dotted curves are 95% confidence intervals (the central 95 percentiles of 1000 draws). Samples were done at N = 3, 5, 10, 15, 20, 25, 30, 40, 50, 60, 70, 80, 90, 100, 150, 200, 400, 600, 800, 1000, 1500, 2000, 2500, 3000, 3500, 4000, 4500, 5000, 5500, 6000, 6500, 7000.
© Copyright Policy
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC3794993&req=5

pone-0076374-g003: Relationship between estimated range size and sample size of geo-referenced records.Panel A is for real species, and Panel B for random draws from the entire pool of geo-referenced records. In panel A, solid points indicate species from Gutianshan; open circles are for species endemic to China (any of the plots). The green solid line (lower) is a trend curve through the Gutianshan points, and the solid red (upper) line through the endemics. The dotted curve is the lower 95% confidence limit of random ranges (from Panel B). Panel B is a test for bias in estimated range size caused by sample size. The solid, central curve is the mean range size from 1000 draws at each value of N (sample number of geo-referenced records), and two dotted curves are 95% confidence intervals (the central 95 percentiles of 1000 draws). Samples were done at N = 3, 5, 10, 15, 20, 25, 30, 40, 50, 60, 70, 80, 90, 100, 150, 200, 400, 600, 800, 1000, 1500, 2000, 2500, 3000, 3500, 4000, 4500, 5000, 5500, 6000, 6500, 7000.
Mentions: This method was designed to describe the area over which a species was found. It does not address whether all habitats within the area are appropriate. It has the substantial advantage of discounting rare outliers in a quantitative way (there are several examples in Fig. 2 and Fig. 3), and use of the Bayesian fitting precisely described confidence in the estimated ranges. Traditional techniques to estimate range size of species, such as Polygon methods surrounding the observed points, cannot offer precise confidence limits and moreover are biased with respect to abundance. For comparison, we also estimated range sizes for each species using the Minimum Convex Polygon (MCP) estimator in R-package ‘adehabitatHR’ Version: 0.4.7 [46]. The R2 between range sizes estimated by the two techniques was very high up to 0.90, and the ranges of two species with few records were greatly underestimated by MCP (Figure S1).

Bottom Line: There was no relationship between local abundance and range size, and no evidence for species being more abundant towards their range-centers.Finally, species' abundances were positively correlated between sites.The widespread nature of most tree species in China suggests few are vulnerable to global extinction, and there is no indication of the double-peril that would result if rare species also had narrow ranges.

View Article: PubMed Central - PubMed

Affiliation: Key Laboratory of Vegetation and Environmental Change, Institute of Botany, Chinese Academy of Sciences, Beijing, P.R. China.

ABSTRACT
Most studies on the geographical distribution of species have utilized a few well-known taxa in Europe and North America, with little research in China and its wide range of climate and forest types. We assembled large datasets to quantify the geographic ranges of tree species in China and to test several biogeographic hypotheses: 1) whether locally abundant species tend to be geographically widespread; 2) whether species are more abundant towards their range-centers; and 3) how abundances are correlated between sites. Local abundances of 651 species were derived from four tree plots of 20-25 ha where all individuals ≥1 cm in stem diameter were mapped and identified taxonomically. Range sizes of these species across China were then estimated from over 460,000 geo-referenced records; a Bayesian approach was used, allowing careful measures of error of each range estimate. The log-transformed range sizes had a bell-shaped distribution with a median of 703,000 km(2), and >90% of 651 species had ranges >10(5) km(2). There was no relationship between local abundance and range size, and no evidence for species being more abundant towards their range-centers. Finally, species' abundances were positively correlated between sites. The widespread nature of most tree species in China suggests few are vulnerable to global extinction, and there is no indication of the double-peril that would result if rare species also had narrow ranges.

Show MeSH