| Edited by Fumio Tajima. Nobuyuki Inomata: Corresponding author. E-mail: ninomscb@kyudai.jp |
Mangrove is a general term for plant species distributed in tropical and subtropical coastal regions. Mangrove forests are important for preventing erosion and are habitat of a vast number of species, and thus constitute a unique ecosystem. They are also used for various economically important products such as e.g., charcoal (Lanly and Lindquist, 1985). In addition, some mangrove tree species have important morphological and physiological traits such as e.g., viviparous propagules, aerial roots and salt tolerance. These traits are believed to be adaptation to severe coastal environments. Revealing genetic structure of mangrove species provides useful information not only for the management of the mangrove forests but also for the understanding of evolutionary forces leading to the present biodiversity and adaptation.
In plants, reproductive system is one of the important factors affecting species distribution and genetic structure of populations. Dispersal of pollen and seeds, corresponding to gene flow, could determine the range of species distribution and the level of population differentiation. In Southeast Asia, two mangrove species of the Rhizophoraceae, Rhizophora apiculata and R. mucronata are predominant and sympatric. Distribution of R. mucronata reaches further east than that of R. apiculata (Tomlinson, 1986). Both species are restricted to regions with wet climate (Tomlinson, 1986). In the Indo-Malayan region R. mucronata grows with R. apiculata but becomes less conspicuous as one moves eastward (Tomlinson, 1986). R. apiculata is often dominant component of mangrove forest in the Malaysian region (Tomlinson, 1986). As far as we know, R. apiculata and R. mucronata are diploid species. At least, their close relatives, R. mangle and R. stylosa, are diploid species (Tyagi, 2002). R. apiculata and R. mucronata have similar reproductive systems and appear to be self-compatible (Tomlinson, 1986). Pollination of Rhizophora species is somewhat controversial. According to Tomlinson (1986) they are wind-pollinated. However, some other authors concluded that they may also be pollinated by insects (e.g., Coupland et al., 2006). The viviparous propagules of R. mucronata are much longer (up to 70 cm) than those of R. apiculata (20–30 cm) (Tomlinson, 1986). The propagules of both species spread by ocean currents (Tomlinson, 1986), but the efficiency of dispersal is poor (Aksornkoae et al., 1992).
In the Southeast Asia land connections and disconnections of the Malay Peninsula, Sumatra, Java and Borneo islands have repeatedly occurred by exposure and sinking of their continental shelf due to sea level changes in association with climatic changes (Voris, 2000). These past geomorphic changes in the coastal regions could have affected distribution of R. apiculata and R. mucronata and genetic structure of their populations, because their propagules spread by ocean currents.
Patterns of haplotype distribution of chloroplast DNA (cpDNA) regions in Ceriops tagal (Huang et al., 2008; Liao et al., 2007), a species of the Rhizophoraceae family, is consistent with the so-called land barrier hypothesis of the Malay Peninsula, in which the past and/or present land barrier of the Malay Peninsula is expected to prevent gene flow between mangrove species occurring along the coasts of the Pacific and Indian Oceans leading to population differentiation between western and eastern populations of the Malay Peninsula (Duke et al., 2002). The levels of polymorphism of cpDNA regions in C. tagal were unexpectedly high: however, it is not clear whether those cpDNA regions are also highly variable in other mangrove species. Although genetic markers such as AFLP (e.g., Giang et al., 2003), microsatellites (e.g., Maguire et al., 2000), cpDNA (e.g., Chiang et al., 2001) and mitochondrial DNA (e.g., Chiang et al., 2001) were developed for surveys of genetic structure, phylogeography and molecular phylogeny of some mangrove species, the amount and pattern of genetic variation of functional nuclear genes in mangrove species are largely unknown. Investigation of such regions is necessary for understanding of evolutionary forces shaping genetic structure and distribution of extant populations, finding genetic factors responsible for adaptation, and providing guidelines for conservation of mangrove forests.
In this study, we examined nucleotide variation of five nuclear genes and two cpDNA regions in R. apiculata and R. mucronata. Plant materials were collected in natural populations from three coast areas (Bangkok, Surat Thani and Trang) in Thailand. We addressed the following questions: (1) are DNA sequence data for chloroplast and nuclear DNA regions useful for population genetic studies on the two investigated species? (2) are populations genetically structured? (3) do sympatric species with similar reproductive systems have similar patterns of genetic variation?
Leave samples of R. apiculata and R. mucronata were collected from two eastern coastal populations in the gulf of Thailand (Bangkok, BK: 13° 44'N, 100°34'E, Surat Thani, ST: 09°07'N, 099°21'E) and one western coastal population in the Andaman Sea site of Thailand (Trang, TR: 07°31'N, 099°37'E). In R. apiculata, 11, 13 and 7 individuals from population BK, ST and TR were used. In R. mucronata, 12, 13 and 14 individuals from population BK, ST and TR were used.
Genomic DNA was isolated from leaves with a modified CTAB method (Murray and Thompson, 1980) and further purified with a GenElute Mammalian Genomic DNA Kit (Sigma). Five nuclear genes PAL1 (phenylalanine ammonia-lyase, EC 4.3.1.24), SBE2 (starch branching enzyme II, EC 2.4.1.18), DLDH (dihydrolipoamide dehydrogenase, EC 1.8.1.4), mang-1 (mangrin) and LAS (lipoic acid synthase), and two cpDNA regions of atpB-rbcL spacer and trnL-trnF spacer regions were amplified in R. apiculata and R. mucronata.
In Bruguiera gymnorrhiza the DLDH and LAS genes were identified as genes involved in response to salt stress (Banzai et al., 2002). The mang-1 gene of B. sexangula was a homolog of AOC (allene oxide cyclase, EC5.3.99.6) and was identified as a salt tolerance gene (Yamada et al., 2002). PCR primers for the investigated nuclear gene regions were designed based on comparisons with corresponding sequences from other plant species: PAL (PAL1 and PAL2 genes of R. mangle, PAL gene of Populus kitakamiensis), SBE2 (Shorea species by Ishiyama, personal communication), DLDH (DLDH genes of B. gymnorrhiza, Lycopersicon esculentum, Pisum sativum and Arabidopsis thaliana), mang-1 (mang-1 gene of B. sexangula, AOC genes of Nicotiana tabacum and A. thaliana), LAS (LAS genes of B. gymnorhiza, P. sativum and A. thaliana). The efficiency of all these primers was poor. We therefore redesigned them based on sequences obtained from R. apiculata and R. mucronata. The investigated cpDNA regions are identical to those used in the previous study on C. tagal (Liao et al., 2007). The primers specific to each DNA region are listed in Table 1.
 View Details | Table 1 PCR primers and annealing temperature |
PCR amplifications were performed in a reaction volume of 20 μl with an initial denaturation at 95°C for 3 min, followed by 35 cycles of denaturation at 95°C for 30 sec, annealing at 45°C, 50°C or 55°C for 30 sec and polymerization at 72°C for 2 min and a final extension at 72°C for 7 min. The purified PCR products were directly sequenced. The sequences of both strands were determined using ABI Prism 3100 automatic sequencer (Applied Biosystems) and a DNA sequencing kit (BigDye terminator v. 3.1/1.1 cycle sequencing kit, ABI) with the PCR primers and the internal primers for sequencing. Of the five nuclear genes and two cpDNA regions, PCR products from some regions could not be amplified for some individuals. The number of sequences obtained for each gene is summarized in Table 2-1 and Table 2-2.
In this study, we did not determine sequences of both haplotypes of an individual with two or more heterozygous sites. When the sequences obtained by direct sequencing had no or only one heterozygous site, we inferred sequences of both haplotypes of an individual. Although the sequences of both haplotypes for each individual with two or more heterozygous sites cannot be determined, we can determine the genotype of each heterozygous site for such individuals. Using haplotype and genotype information, we can still estimate some measures of population genetic parameters such as the number of segregating sites and heterozygosity without determination of sequences of individual haplotypes. Therefore, in the following analyses we used all sequence information including both haplotypic and genotypic data.
DNA sequences were aligned using the CLUSTAL X program (Thompson et al., 1997) and further edited by hand. Molecular population genetic parameters were estimated, and Tajima’s test (Tajima, 1989) and MK test (McDonald and Kreitman, 1991) were performed using the DNAsp program, version 4.20 (Rozas et al., 2003). The population structure was examined using the Arlequin software ver. 3.11 (Excoffier et al., 2005). F-statistics, FIS and FST, for each nuclear gene were estimated by the locus-by-locus analysis of molecular variance (AMOVA) approach (e.g., Weir and Cockerham, 1984). In addition, FST for each cpDNA region was estimated by the locus-by-locus AMOVA approach. Overall values of F-statistics for the five nuclear genes and two cpDNA regions were obtained by summing variance components over the genes. In the AMOVA analyses, genotypic data, where gametic phase of genotypes is unknown, were used for nuclear genes and haplotypic data were used for the cpDNA regions. Sequences of individuals that failed to amplify were treated as missing data. A continuous alignment gap was counted as a single indel. For nuclear genes, the significance of FIS values was tested by 10,000 permutations of haplotype sequences among individuals within populations and the significance of FST values was tested by 10,000 permutations of individual genotypes between populations. For cpDNA regions, the significance of FST values was tested by 10,000 permutations of haplotype sequences between populations. The multilocus HKA test was performed using the HKA program obtained from Jody Hey’s web site (http://lifesci.rutgers.edu/~heylab/). DNA sequences obtained in this study were deposited in the DDBJ. Their accession numbers are AB446550–AB447246.
We determined sequences of partial regions of the five nuclear genes, PAL1, SBE2, DLDH, mang-1 and LAS, in R. apiculata and R. mucronata. We also determined sequences of two cpDNA regions, atpB-rbcL spacer and trnL-trnF spacer, in both species. The levels of DNA polymorphism in R. apiculata and R. mucronata were estimated and the results are summarized in Table 2-1 and Table 2-2.
Sequence alignment was first performed species-wise. Sequence alignment length in three regions PAL1, DLDH and trnL-trnF spacer was identical between the two species. In the other four regions, length variations between the two species were found (Table 2-1 and Table 2-2). The level of nucleotide variation was surprisingly low in both species. For example, the number of nucleotide differences per silent site with the Jukes and Cantor (1969) correction (πs) ranged from 0 to 0.00121 for the nuclear genes in R. apiculata (Table 2-1), 0 to 0.00014 in R. mucronata (Table 2-2). Among the five nuclear genes, no shared polymorphism was found between R. apiculata and R. mucronata. The cpDNA regions were monomorphic when indels were excluded in both R. apiculata and R. mucronata (Table 2-1 and Table 2-2). In the atpB-rbcL region a single indel was shared polymorphism between R. apiculata and R. mucronata. In both nuclear and cpDNA regions the number of indels was small, at most one in the two species. Extremely low polymorphism observed in the cpDNA regions of both R. apiculata and R. mucronata contrasts with the high polymorphism reported in C. tagal (Liao et al., 2007).
 View Details | Table 2-1 Nucleotide variation in R. apiculata |
 View Details | Table 2-2 Nucleotide variation in R. mucronata |
In contrast to low polymorphism, nucleotide divergence between R. apiculata and R. mucronata was considerable (Table 3). Nucleotide divergences were higher in nuclear genes than cpDNA regions. The number of nucleotide substitutions per silent site between the species (Ks, Nei and Gojobori, 1986) with the Jukes and Cantor (1969) correction ranged from 0.01821 to 0.04561 for the nuclear genes, and 0.00135 or 0.00445 for the cpDNA regions (Table 3), indicating that the nuclear genes evolve faster than the cpDNA regions. Roughly, the nuclear genes evolved 10 times faster than the cpDNA regions. The number of nucleotide substitutions per replacement site between the species (Ka, Nei and Gojobori, 1986) with the Jukes and Cantor (1969) correction ranged from 0.00205 to 0.00601 for the nuclear genes (Table 3). Thus, Ka/Ks value ranged from 0.045 for the LAS to 0.297 for the SBE2, indicating these nuclear sequences are not pseudogenes. In most nuclear genes included in this study the number of fixed nucleotide differences between the species was much larger than the number of polymorphic nucleotide differences within species (Table 2-1 and Table 2-2 and Table 3).
 View Details | Table 3 Nucleotide divergence between R. apiculata and R. mucronata |
The standard neutral model was tested by the Tajima’s test. In R. apiculata Tajima’s D values of the five nuclear genes, PAL1, SBE2, DLDH, mang-1 and LAS, were 0.269, 1.557, –0.675, 1.002 and 0.963, respectively. In R. mucronata, Tajima’s D values were –0.372 and –0.752 in PAL1 and mang-1, respectively, and those of other three genes were not obtained simply because of no variation. All Tajima’s D values in the two species were not significant. The number of polymorphic silent/replacement differences and that of fixed ones is shown in Table 2-1 and Table 2-2 and Table 3, respectively. In the MK test the result was significant for only one gene (PAL1, P = 0.047, Fisher’s exact test), indicating an excess of replacement polymorphism. We also performed the multilocus HKA test for the five nuclear genes. The result was significant (P = 0.043), indicating the ratio of the levels of polymorphism to divergence is heterogeneous among the gene regions. Probably the heterogeneous ratio is due to low and uniform polymorphism among the gene regions.
At first, F-statistics, FIS and FST, among the three populations (BK, ST and TR) were estimated by the AMOVA approach. In R. mucronata, three of the five nuclear genes (SBE2, DLDH and LAS) were monomorphic and thus they did not give any information on population structure. In both R. apiculata and R. mucronata, overall values of the five nuclear genes were significant for both FIS and FST (Table 4), although the significance level of FIS in R. apiculata was weak. Significant overall FIS value indicates the deviation from Hardy-Weinberg proportions within subpopulations. Positive FIS value indicate a deficiency of heterozygotes. Significant overall FST value indicates genetic differentiation among the three populations in both species.
 View Details | Table 4 F-statistics, FIS and FST, among three populations (BK, ST and TR) |
Actually, the number of heterozygotes appeared to be small. For example, in the PAL1 of R. apiculata, five individuals were heterozygous of 30 individuals in total. We also tested the deviation from Hardy-Weinberg proportions, where the alternative hypothesis is heterozygote deficiency, using the GenePop software ver. 4.0 (Rousset, 2008). We found highly significant heterozygote deficiency in the BK population of R. apiculata (P = 0.0078) and significant heterozygote deficiency in the ST population of R. mucronata (P = 0.0336).
For further examinations of population structure, we estimated F-statistics for pairs of populations. In R. apiculata, overall values for the five nuclear genes were significant for all F-statistics except for FIS values in population pairs: ST-BK and ST-TR (Table 5). Overall FST value for the five nuclear genes was lowest between ST and BK populations (FST = 0.140, P = 0.018, Table 5), and highest between BK and TR populations (FST = 0.875, P << 0.001, Table 5). This tendency in the nuclear genes was consistent with overall FST values of the two cpDNA regions. Overall FST value was lowest between ST and BK populations (FST = 0.012, Table 5), and highest between BK and TR populations (FST = 0.688, Table 5), although they were not significant. These results indicate high population differentiation in R. apiculata, suggesting strong differentiation between populations from the western and eastern coasts of Thailand.
 View Details | Table 5 Pairwise F-statistics, FIS and FST |
In R. mucronata, overall values for the five nuclear genes were significant for all F-statistics except for pairs of BK-TR populations (Table 5). In contrast to R. apiculata, overall FST value of the five nuclear genes was highest between ST and BK populations (FST = 0.213, P = 0.002, Table 5), and lowest between BK and TR populations (FST = 0.027, Table 5). Overall FST value of two cpDNA regions in the pair of populations ST-TR (FST = 0.170, Table 5) was similar to that in pair BK-TR (FST = 0.144, Table 5). In R. mucronata, the pattern of population differentiation did not appear to be consistent between the nuclear genes and cpDNA regions.
We examined nucleotide variation of the cpDNA region that was identical to the one studied in C. tagal (Liao et al., 2007). In C. tagal, π values of cpDNA regions excluding alignment gaps were 0.00307 for the atpB-rbcL and 0.02515 for the trnL-trnF (Liao et al., 2007). In contrast to that study, π values of cpDNA regions were zero in both R. apiculata and R. mucronata, and even when indels were included, both species still showed low polymorphism (Table 2-1 and Table 2-2). The cpDNA regions, in particular the trnL-trnF region, investigated in this study are frequently used to reconstruct phylogenetic trees across a wide range of plant taxa, because they are conservative. Therefore, we can expect that the regions show low polymorphism. In this context, variation found in C. tagal seems to be unexpectedly high.
In some pairs of populations, we found some degree of population differentiations, but FST values of the two cpDNA regions were not significant. Only when the three populations were considered together, FST value was significant in R. mucronata (Table 4). Unlike the previous study on C. tagal, our results indicate that in both R. apiculata and R. mucronata the cpDNA regions are not very useful for surveys of population structure. On the other hand, although variation of the five nuclear genes was low in both R. apiculata and R. mucronata, specifically in R. mucronata three of the five genes were monomorphic, significant F-statistics values were found between populations in both species, indicating that these nuclear genes are more useful for such purpose than the cpDNA regions.
In contrast to most previous studies on nucleotide variation in plant populations the level of polymorphism in all nuclear gene regions used in our study was very low. The average πs values (0.00059 and 0.00003 in R. apiculata and R. mucronata, respectively) were one or two orders lower than those in other plant species (e.g., Cryptomeria japonica: πs = 0.0038 (Kado et al., 2003); A. thaliana: πs = 0.0083 (Schmid et al., 2005); Pinus tabuliformis, P. densata and P. yunnanensis: πs = 0.0087–0.0128 (Ma et al., 2006). Taking into account that five different nuclear loci totaling more than 5500 bp including more than 2900 bp of introns were investigated, the observed low levels of polymorphism appear to be a general feature of both R. apiculata and R. mucronata. Some previous studies on other mangrove tree species based on genetic markers also suggested that they harbor particularly little genetic variation (Arnaud-Haond et al., 2006).
Low variation found in this study could be caused by failures of PCR amplification. Both low quality of template DNA and mismatches of PCR primers at priming sites in distinct sequences could result in the missing data (failure of amplification). However, in this study, most individuals with the missing data failed to amplify in the multiple loci (only a single individual failed to amplify in a single locus out of 7 loci). In addition, sequences of the distinct haplotypes or heterozygotes found here showed very low number of segregating sites. These results are more likely to support the possibility of low quality of template DNAs rather than the presence of undetected very distinct haplotypes.
In contrast to low nucleotide polymorphism within both R. apiculata and R. mucronata, the nucleotide divergence between the two species was considerable (Table 3). R. apiculata and R. mucronata are not likely to be the most closely related species (Lakshmi et al., 2002). Our results (low polymorphism and high divergence) indicate that the investigated gene regions are not conservative or do not evolve at extremely low evolutionary rate.
R. apiculata and R. mucronata appear to be self-compatible (Tomlinson, 1986). Our results showed positive FIS values indicating a deficiency of heterozygotes within subpopulations in both species. In addition, when we tested the deviation from Hardy-Weinberg proportions, where the alternative hypothesis is heterozygote deficiency, we found highly significant heterozygote deficiency in the BK population of R. apiculata (P = 0.0004) and the ST population of R. mucronata (P = 0.0006). The observed heterozygote deficiency across the loci could be explained by inbreeding, the Wahlund effect, or both. In plants, albino individuals are usually recessive homozygotes and their frequent occurrence may be associated with increased levels of inbreeding. High rates of albinism and selfing were reported for a closely related species R. mangle from Central America (Lowenfeld and Klekowski, 1992). Albino mutants were also observed in R. apiculata and R. mucronata (Szmidt, personal observation). In both R. apiculata and R. mucronata pollen release is very limited (Kusmana, personal communication), which could cause increased levels of inbreeding. Therefore, our results suggest the effect of inbreeding within populations of both species included in our study.
In R. apiculata, the patterns of population differentiation of the nuclear genes were consistent with those of the cpDNA regions. That is, population differentiation was smallest in the population pair of ST-BK and largest in the population pair of BK-TR (Table 5). This is likely to be consistent with the land barrier hypothesis of the Malay Peninsula, in which the Malay Peninsula prevents gene flow between mangrove species occurring along the coasts of the Pacific and Indian Oceans leading to population differentiation between western and eastern populations of the Malay Peninsula (Duke et al., 2002). Results reported for the Ceriops species are also consistent with this hypothesis (Huang et al., 2008; Liao et al., 2007). In contrast, in R. mucronata the pattern of population differentiation of the nuclear genes was inconsistent with that of the cpDNA regions. In addition, the geographic pattern of population differentiation was different from that of R. apiculata, and the degree of differentiation was not strong. This pattern of population differentiation of the nuclear genes did not give much support for the land barrier hypothesis of the Malay Peninsula, because the FST value between BK and TR populations was lowest and not significant. Probably genetic variation detected in our study was not sufficient to estimate F-statistics of R. mucronata. Indeed, three of the five nuclear genes were monomorphic in R. mucronata. Unfortunately, in both species only three populations from Thailand were examined in this study. To test the land barrier hypothesis, it is necessary to survey additional nuclear gene regions and populations collected from western and eastern coasts of the Malay Peninsula.
The levels of seas surrounding Malay Peninsula and its neighboring islands have experienced frequent and substantial fluctuations in the past 20,000 years (Voris, 2000). As a result, Malay Peninsula was repeatedly separated and reconnected with the surrounding islands. These changes of the coastal lines may have caused frequent and repeated extinctions and re-colonization events of mangrove species. Frequent and repeated extinctions and re-colonization events are known to greatly reduce effective population size (Maruyama and Kimura, 1980). Unfortunately, probably because of low variation, we did not obtain significant Tajima’s D values. However, a significant result of the multilocus HKA test, which indicates extremely low and uniform level of intra-specific variation relative to inter-specific variation, is consistent with such scenario.
In this study we revealed unexpectedly low level of polymorphism in contrast to considerable divergence between R. apiculata and R. mucronata and the usefulness of the developed nuclear gene markers for population surveys. In addition, our results suggest that the demographic history of the investigated species is not similar even though they are sympatric and have similar reproductive systems.
We thank Y. Nakano for technical assistance. We thank Dr. H. Ishiyama for testing PCR amplification. We thank Drs. M. Heuertz and M. Lascoux for their advice and help with Arlequin files. This work was supported by the research grants from Ministry of Education, Culture, Sports, Science, and Technology to A. E. S. and N. I. and from SIDA to X-R. W.
In both R. apiculata and R. mucronata pollen release is very limited (Kusmana, personal communication), which could cause increased levels of inbreeding.
|
Index