The study area is located in riparian forests of the Shennongjia Mountains (31°21′20″–31°36′20″N, 110°03′05″–110°33′50″E) in central China (Fig. 1). The Shennongjia Mountains (3105.4 m asl) are an important part of the south-central China biodiversity hot-spot and are rich in Tertiary-relict and endemic plants (Myers et al., 2000; Ying, 2001). Chinese endemic genera are limited to land below 2000 m (Shen et al., 2004). Altitude is the main factor determining the distribution pattern of forest tree communities (Jiang et al., 2002a; Zhao et al., 2005; Wei et al., 2010b), and the altitudinal pattern of plant diversity has a unimodal pattern with a peak (c. 1500 m) in the mixed evergreen and deciduous broad-leaved forest zone (Jiang et al., 2002b; Shen et al., 2004; Zhao et al., 2005).
Figure 1. Map of the study region with the locations of the 20 study sites along the upper reaches of the Nan River (eight sites) and Yandu River (12 sites) in central China. G209 (bold black line) is the No. 209 national highway of China.
Download figure to PowerPoint
Mountain riparian zones are crucial refuges for Tertiary-relict plants in this area (Jiang et al., 2002b; Wei et al., 2010b). There are four main river systems in the Shennongjia Mountains: the Yandu River and Xiangxi River on the south-facing slope, which are tributaries of the Yangtze River, and the Nan River and Du River on the north-facing slope, which are tributaries of the Han River (Fig. 1). Euptelea pleiospermum is one of the dominant relict tree species of the riparian plant communities on this subtropical mountain (Wei et al., 2008, 2010b).
The study was undertaken in riparian forests along the upper reaches of two river systems (Fig. 1). The Yandu River is relatively unaffected by human disturbance because of its inaccessibility. The Nan River, along which there is a national highway below 1700 m and a vehicle road for tourism above 1700 m, is severely disturbed by human activities. Study sites were established at roughly 100-m elevational intervals along the two rivers. The sampling range depended on the altitudinal range of E. pleiospermum along each river (Yandu River, 900–2000 m; Nan River, 1200–1900 m). As a result, a total of 12 and eight sites were established along the Yandu River and Nan River, respectively (Fig. 1, Table 1).
Table 1. Site characteristics, plant species diversity (S, species richness; D, Simpson’s diversity index), genetic diversity of Euptelea pleiospermum (AR, allelic richness; HE, expected heterozygosity) and effective population size (Ne–ONeSAMP and Ne–ldne) for the 20 sites along the Nan River (disturbed; N1–N8) and Yandu River (undisturbed; Y1–Y12)
|Site||Altitude (m)||S||D||N||AR||HE||Ne– ONeSAMP (prior = 2–500)||Ne–ldne (Pcrit = 0.05)|
|N1||1904||13||0.815||36||3.71||0.622||36.4 (28.3–58.4)||32.6 (17.7–80.8)|
|N2||1875||12||0.878||31||4.28||0.671||44.3 (32.2–79.8)||117.7 (36.5–∞)|
|N3||1720||12||0.835||25||4.27||0.656||26.7 (19.8–48.1)||33.3 (17.4–108.5)|
|N4||1675||14||0.859||30||4.03||0.613||35.8 (26.5–63.9)||137.0 (40–∞)|
|N6||1410||16||0.869||32||4.24||0.635||56.7 (41.2–108.0)||43.6 (22.6–143.1)|
|Y1||2010||15||0.821||33||3.65||0.558||25.5 (18.2–46.0)||7.5 (3.8–12.0)|
|Y2||1956||14||0.822||24||4.09||0.635||28.5 (20.4–52.8)||26.1 (13.1–89.3)|
|Y4||1750||18||0.909||28||3.76||0.621||22.6 (17.0–39.0)||26.2 (14.0–69.8)|
|Y5||1640||27||0.920||40||4.38||0.686||41.4 (30.0–81.4)||59.4 (33.4–158.7)|
|Y6||1575||22||0.888||61||4.33||0.657||54.6 (39.5–102.5)||115.3 (59.2–498.1)|
|Y7||1471||25||0.936||41||4.45||0.687||35.1 (24.4–76.3)||21.3 (14.8–31.9)|
|Y8||1320||20||0.899||28||4.09||0.643||26.7 (20.0–52.1)||63.9 (27.3–∞)|
|Y11||1005||17||0.847||37||3.79||0.555||57.6 (38.5–152.7)||22.5 (13.5–41.7)|
|Y12||980||14||0.859||32||4.18||0.613||30.3 (20.9–60.7)||28.1 (17.0–55.0)|
Community survey and sampling
A community survey was conducted between late July and early September of 2006. At each site, one plot (20 × 30 m) was established in a forest tree community dominated by E. pleiospermum. Within each plot, we recorded the species and diameter at breast height (DBH) of all living trees (DBH ≥ 2.5 cm). The nomenclature follows that of the Flora of China (Wu & Raven, 1994).
Leaf sampling was carried out in April and May 2008. Within and around each surveyed plot, we randomly (without replacement) collected several young leaves of E. pleiospermum individuals that were at least 30 m apart. The sampled leaves were immediately dried in a 10 × 5 cm plastic bag containing silica gel. Because of the sporadic distribution at some altitudes, sample sizes were sometimes relatively low, ranging from six to 61 per population. As a result, a total of 351 and 230 individuals were sampled along the Yandu River and Nan River, respectively (Table 1).
All leaf samples were stored at 4°C before DNA extraction. Total DNA was extracted from leaves using a modified CTAB (Cetyl Trimethyl Ammonium Bromide) method (Doyle & Doyle, 1987). We tested 14 nuclear microsatellite loci developed for E. pleiospermum (Zhang et al., 2008), from which eight (EP021, EP036, EP059, EP081, EP087, EP091, EP278 and EP294) were selected as they had suitable levels of polymorphism. PCR amplification and the allele resolution procedure were performed as described by Zhang et al. (2008).
Two measures of SD were used: species richness (S) and Simpson’s diversity index (D). The number of species per plot was taken as a measure of species richness. If the relative frequency of species i was fi (∑fi = 1), Simpson’s diversity index was estimated as . Both measures of SD were calculated using pc-ord 4.0 (McCune & Mefford, 1999). The measures of GD used were allelic richness (AR) and expected heterozygosity (HE), as they are analogous to S and D, respectively (Etienne, 2005; Evanno et al., 2009). AR was calculated using rarefaction analysis with fstat 18.104.22.168 (Goudet, 2001) and HE was estimated using genetix 4.05 (Belkhir et al., 1996–2004). fstat 22.214.171.124 was also used to test for deviation from Hardy–Weinberg equilibrium (HWE) and linkage disequilibrium (LD). The significance of LD was Bonferroni-corrected for multiple comparisons (Rice, 1989). Relationships between SD (S and D) and GD (AR and HE) were tested using Pearson correlation.
To reveal the altitudinal patterns of SD and GD along the Yandu River, we divided sampled sites into three groups (L, low altitudes, 900–1200 m; M, middle altitudes, 1300–1600 m; and H, high altitudes, 1700–2000 m; each with four sites) and then compared mean values of diversity parameters (S, D, AR, and HE) among the groups. Statistical significance was determined by one-way analysis of variance (ANOVA) followed by Tukey’s post hoc test.
To investigate whether human disturbances reduced SD and GD, we compared mean values of diversity parameters between the Yandu River and the Nan River. Statistical significance was determined by performing a paired-samples t-test. We removed sites Y1, Y10, Y11 and Y12 from the natural habitat for pairwise comparisons of species and genetic data (SD, GD, FST, FST-C, Ne, and recent migration rate) between the two rivers.
To investigate whether human disturbances increased genetic divergence and community divergence, we compared mean values (paired-samples t-test) of divergence parameters between the two rivers. Pairwise population genetic differentiation (FST) was calculated using fstat 126.96.36.199 (Goudet, 2001). FST was estimated as (HT – HS)/HT, where HT is the total expected heterozygosity, and HS is average HE across populations (Nei, 1977). FST-C, a direct analogue of FST, was used to measure community divergence (Vellend, 2004). By treating community as ‘locus’ and species as ‘allele’ (Vellend, 2004), we estimated the analogues of HT and HS at community level and then used the same formula to calculate FST-C.
To investigate whether effective population size (Ne) was reduced in human-disturbed populations, we calculated Ne for each population and then compared the mean values between the two rivers; significance was also determined via a paired-samples t-test. The contemporary Ne was calculated with ONeSAMP 1.2 (Tallmon et al., 2008) and ldne 1.31 (Waples & Do, 2008). ONeSAMP uses summary statistics and approximate Bayesian computation to calculate Ne estimates. The lower and upper limits of the prior distribution for Ne were 2 and 500, respectively. ldne uses a linkage disequilibrium method to calculate Ne estimates and incorporates the bias correction from Waples (2006). We assumed a random mating model and calculated separate estimates using three threshold allele frequencies (Pcrit: 0.05, 0.02, and 0.01) for excluding rare alleles.
To investigate whether road margins along rivers act as corridors of dispersal, we compared mean values (paired-samples t-test) of recent migration rates along the two river valleys. Recent migration rate is a measure that can be used as an indirect estimate of gene flow among populations over the last several generations. We estimated recent migration rates between populations along each river by using the program BayesAss version 1.3 (Wilson & Rannala, 2003). The program calculates unidirectional estimates of migrant (m) for each population pair (Fraser et al., 2007); this method is based on the model in which individuals are exchanged between populations over generations (Goossens et al., 2005). The program was run using a Markov chain Monte Carlo (MCMC) length of 3 000 000 with a burn-in period of 1 000 000 (initial conditions of Δp = allele frequency, Δm = migration, and ΔF = inbreeding coefficient, all equal to 0.15).